A Survey of Word Accentual Patterns in the Languages of the World 9783110198966, 9783110196313

Table of contents :
Frontmatter
Table of contents
1. Word accent: Terms, typologies and theories
2. An overview of word stress in Australian Aboriginal languages
3. Stress types in Austronesian languages
4. Word stress and pitch accent in Papuan languages
5. Accent in the native languages of North America
6. The Languages of Middle America
7. A Survey of South American stress systems
8. Accent in African languages
9. Word accent systems in the languages of Europe
10. Word accent systems in the languages of Asia
11. Word accent systems in the Middle East
12. A typology of stress patterns
Australian languages
Austronesian languages
Papua New Guinea and Irian Jaya languages
North American languages
Middle American languages
South American languages
Eurasian languages
Asian languages
Middle Eastern and African languages
Backmatter

Citation preview

A Survey of Word Accentual Patterns in the Languages of the World

A Survey of Word Accentual Patterns in the Languages of the World Edited by Harry van der Hulst Rob Goedemans Ellen van Zanten

De Gruyter Mouton

ISBN 978-3-11-019631-3 e-ISBN 978-3-11-019896-6 Library of Congress Cataloging-in-Publication Data A survey of word accentual patterns in the languages of the world / edited by Harry van der Hulst, Rob Goedemans, Ellen van Zanten. p. cm. Includes bibliographical references and index. ISBN 978-3-11-019631-3 (hardcover : alk. paper) 1. Accents and accentuation. 2. Emphasis (Linguistics) I. Hulst, Harry van der. II. Goedemans, Rob. III. Zanten, Ellen van. P231.S87 2008 4141.6⫺dc22 2010046145

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. 쑔 2010 Walter de Gruyter GmbH & Co. KG, Berlin/New York Cover image: Creatas/Thinkstock Typesetting: RoyalStandard, Hong Kong Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany www.degruyter.com

Preface Over a decade ago, in 1996, Stress patterns of the world: part 1 was published. This book provided background to the structure, organization and content of StressTyp1, a database for word stress/accent systems on which a variety of people had been working since 1992. The 1996 volume contained theoretical and practical information on the database, as well as possible applications and extensions of StressTyp. As is clear from the ‘part 1’ of the title, a sequel volume was conceived which would focus on presenting data from StressTyp along with typological studies in the domain of word stress/accent. Around the start of the new millennium the first steps towards this typological and data-oriented volume were taken. Experts on various language areas were asked to contribute chapters on the stress/accent systems of as many languages in ‘their’ part of the world as they could lay their hands on. No preconditions (theoretical or otherwise) were set, but the authors were encouraged to use the StressTyp data in their chapters. Some did so more than others. As we had hoped, all authors went beyond the data that are already present in StressTyp so that, as a consequence, the material collected in this volume will provide an excellent base for a future extension of StressTyp. In this preface we will describe the content of this volume, showing how the various language families of the world are covered in its chapters. This book is largely organized in a geographical fashion. A list of chapters is presented below. We indicate which languages are covered in each chapter (for the actual chapter titles see the Table of Contents): Part I: Surveys Chapter 1:

Word accent: Terms, typologies and theories (Harry van der Hulst)

Chapter 2:

Australian Languages (Rob Goedemans)

Chapter 3:

Austronesian Languages (Ellen van Zanten, Ruben Stoel and Bert Remijsen)

1. A version of StressTyp is available on the web, which comes with a simple search engine. The full database (implemented in ACCESS) can be obtained by writing to Rob Goedemans. See Goedemans and van der Hulst (2009) for a detailed and up-to-date discussion of StressTyp, and relevant references.

vi

Preface

Chapter 4:

Papuan Languages (Ellen van Zanten and Philomena Dol)

Chapter 5: Chapter 6:

North American Languages (Keren Rice) Middle American Languages (Harry van der Hulst, Keren Rice and Leo Wetzels)

Chapter 7: Chapter 8:

South American Languages (Leo Wetzels and Sergio Meira) African Languages (Laura Downing)

Chapter 9: European Languages (Harry van der Hulst) Chapter 10: Asian Languages (Rene´ Schiering and Harry van der Hulst) Chapter 11: Middle Eastern Languages (Harry van der Hulst and Sam Hellmuth) Chapter 12: Stress Typology (Rob Goedemans) Chapters 2–11 each deal with a geographical area which covers one or more language families, although, as might be expected, the match between area and family, or families is never exact. Moreover, the approximate match refers to the distribution of languages in the pre-colonial period, not taking into account the migration of speakers and languages that has occurred since, roughly, the 17th century. These ten chapters contain elaborate typological and/or areal overviews, including languages that are in StressTyp and languages that are not present in StressTyp at this time. They also contain theoretical excursions into interesting phenomena encountered in the region or mention striking examples of geographical spreading of accentual phenomena. Together these chapters provide the reader with a thorough idea of the diversity of stress/accent systems that occur in the language families of the world. Chapter 2 has a double function. It presents an overview of stress patterns in Australian languages, but at the same time it is intended to (re-)familiarize readers with the coding, terminology and theoretical ideas of the StressTyp database. We have provided an introductory chapter (Chapter 1) that will provide the reader with elementary terminology and theoretical tools to organize the variety of accentual systems that will be discussed in the subsequent chapters of this book. This chapter serves as an introduction to the kinds of phenomena that this book is about, including some of the commonly used distinctions and terminology. The phenomena that are studied here are rich and varied, and not always fully understood. This has inevitably led to a proliferation of terms in the linguistic and phonetic literature, which is also reflected in this volume. We did not ‘require’ all authors to use the same terminology. Although chapter 1 advocates a particular usage of terms, it also relates this to other practices, some of which

Preface

vii

are used in other chapters. Chapter 12 is another general chapter which o¤ers a quantitative study regarding some of the most important stress phenomena, such as location, quantity sensitivity, rhythm etc., based on the StressTyp database.2 Part II:

Language profiles

Part II of this volume contains ‘language profiles’ which are, for each of the 511 languages contained in StressTyp (in 2009), extracts from the information that is contained in the database. Each profile contains: – – – – – – –

The language name A Stress Type Code A genetic a‰liation A geographical area A prose description of the primary stress location Some examples in IPA with glosses Some references which have been used.

Except for some minor corrections, these profiles are exactly as the information occurs in StressTyp even in those cases in which the relevant chapter reports conflicting information. As shown, in StressTyp information about stress is encoded in three di¤erent ways. The so-called Stress Type Code o¤ers a compressed formula which captures the primary stress location; these codes are listed in the appendix to this book. The prose description states the location of both primary and, if present (reported) non-primary stresses in the manner it would be described in a descriptive or reference grammar. Not included in the language profiles in the book is the third and most important aspect of StressTyp which is a formal encoding of both primary and non-primary stress in terms of several explicit parameters, which are discussed in chapter 2. Below we place the languages that are surveyed in this volume within the context of the language families of the world. There are many classifications of the languages of the world that di¤er in many details. Here we will follow Ruhlen (1987/1991: 290) who distinguishes 17 (super-)families (the numbers in parentheses indicate the numbers of languages according to the Ethnologue website (2004 edition) and Ruhlen 1991, respectively):

2. Another important application of the database has led to the contributions by Goedemans and van der Hulst to the World Atlas of Linguistic Structures (WALS) (Haspelmath et al. 2005), while van Zanten and Goedemans (2007) also can serve as an introduction to StressTyp and its possible uses.

viii

Preface

number of languages (super)family

Ethnologue

Ruhlen (1991)

1.

Khoisan

(18

31)

2.

(1489

1064)

3.

Niger-Kordofanian (Niger-Congo) Nilo-Saharan

(199

138)

4.

Afro-Asiatic

(372

241)

5.

Caucasian

(North: 34, South: 5

38)

6.

Indo-Hittite

(443

144)

7.

Uralic-Yukaghir

(38 þ 1 for Yukaghir

24)

8.

Altaic

(65 þ 12 for Japanese

63)

9.

Chukotko-Kamchatkan

(5

5)

10.

Eskimo-Aleut

(11

9)

11. 12.

Elamo-Dravidian Sino-Tibetan

(75 (365

28) 258)

13.

Austric

(1532

1175)

I II III

(32 (168 (1332 (70 (1262 (821 þ 15 Andamanese

4) 155) 1016) 57) 959) 731)

(258

170)

Miao-Yao Austroasiatic Austro-Tai A Daic (Tai) B Austronesian

14. 15.

Indo-Pacific (Papuan)3 Australian

16.

Na Dene

(47

34)

17.

Amerind

(N: 307 þ C: 114 þ S: 315 ¼ 736

583)

Taking Ruhlen’s classification merely as a reference point, the following table indicates which language families are covered by the chapters in this volume: Chapter 2: Chapter 3: Chapter 4:

Australian Languages: 15 Austronesian Languages: 13IIIB Papuan Languages: 14

3. Indo-Pacific has become a label for all Pacific languages that are not Austronesian or Australian. As such it includes the languages of New Guinea plus surrounding islands, the Andamanese languages spoken on the Andaman Islands in the Bay of Bengal and the now extinct Tasmanian language family.

Preface

Chapter 5:

North American Languages: 10, 16, 17

Chapter 6: Chapter 7:

Middle American Languages: 17 South American Languages: 17

Chapter 8:

African Languages: 1–4

ix

Chapter 9: European Languages: 4 (non-African), 5–9 Chapter 10: Asian Languages: 11–13 (exc. 13IIIB) Chapter 11: Middle Eastern Languages: part of 4 There are two remaining categories in Ruhlen’s system: 18.

Language isolates [well-studied]

(30

19.

Unclassified

(96 16)

[new discoveries, undocumented]

5)

No attempt has been made to include all of these in either this or other chapters. However, a number of these languages are discussed nonetheless. There is no account of two other categories that the Ethnologue and Ruhlen mention: 20. 21.

Pidgins and Creoles (and ‘mixed’ languages) (17 þ 81 þ 8 37) Invented (or artificial) (3 2)

Also excluded from this volume are sign languages. We believe that this volume will be of interest to people in the field of theoretical phonology and language typology. It will function as a reference work for these groups of researchers, but also, more generally, for people working on syntax and other fields of linguistics, who might wish to know certain basic facts about the distribution of word accent systems. We hope that its availability ‘as is’ will spark new research and new interest in the phenomenon of word stress/accent and that it will contribute to the growing body of data and theories in this domain.

Acknowledgment The editors would like to thank Kristine Hildebrandt for her helpful advice during the preparation of this book. We would also like to thank an anonymous reviewer for valuable comments on all chapters. This reviewer did not comment on chapter 6, but did express the wish that we add it.

x

Preface

References Goedemans, Rob and Harry van der Hulst 2009 StressTyp: A Database for Word Accentual Patterns in the World’s Languages. In: Martin Everaert, Simon Musgrave and Alexis Dimitriadis (eds.). The use of databases in cross-linguistics research, 235–282. New York/Berlin: Mouton de Gruyter. Gordon, Raymond (ed.) 2005 Ethnologue: Languages of the world, 15th edition. Dallas, Texas: SIL international. Haspelmath, Martin Matthew Dryer, David Gil and Bernard Comrie (eds.) 2005 The World Atlas of Linguistic Structures, Oxford: Oxford University Press. Ruhlen, Merritt 1987 A guide to the world’s languages. Volume 1: Classification. Stanford: Stanford University Press. Ruhlen, Merritt 1991 A guide to the world’s languages. Volume 1: Classification. Stanford, CA: Stanford University Press. Zanten, Ellen van and Rob Goedemans 2007 A functional typology of Australian and Papuan stress systems. In: Vincent J. van Heuven & Ellen van Zanten (eds.), Prosody in Indonesian Languages (LOT Occasional series, 9) (pp. 63–88). Utrecht: LOT.

A‰liation of authors Philomena Dol Leiden University Centre for Linguistics PO Box 9515 2300 RA Leiden The Netherlands [email protected]

Sergio Meira de Santa Cruz Oliveira Leiden University Centre for Linguistics Leiden University PO Box 9515 2300 RA Leiden The Netherlands [email protected]

Laura J. Downing ZAS, Berlin Schuetzenstr. 18, D-10117 Berlin Germany [email protected]

Bert Remijsen Linguistics & English Language University of Edinburgh Dugald Stewart Building 3 Charles Street Edinburgh EH8 9AD United Kingdom [email protected]

Rob Goedemans Leiden University Centre for Linguistics Leiden University PO Box 9515 2300 RA Leiden The Netherlands [email protected] Sam Hellmuth Department of Language and Linguistic Science University of York Heslington, York YO10 5DD United Kingdom [email protected] Harry van der Hulst Department of Linguistics University of Connecticut 337 Mansfield Road Unit 1145 Storrs, Connecticut 06269-1145 USA [email protected]

Keren Rice Department of Linguistics University of Toronto 130 St. George Street Toronto, Ontario Canada M5S 3H1 [email protected] Rene´ Schiering Institut fu¨r Allgemeine Sprachwissenschaft Westfa¨lische Wilhelms-Universita¨t Mu¨nster Aegidiistr. 5, 48143 Mu¨nster Germany [email protected] Ruben Stoel Medical Library Academic Medical Centre University of Amsterdam PO Box 22660 1100 DD Amsterdam The Netherlands [email protected]

xii

A‰liation of authors

Leo Wetzels Vrije Universiteit Amsterdam & Laboratoire de Phone´tique et Phonologie (LPP) CNRS/Paris III, Sorbonne Nouvelle Paris France [email protected]

Ellen van Zanten Leiden University Centre for Linguistics Leiden University PO Box 9515 2300 RA Leiden The Netherlands [email protected]

Table of contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

A‰liation of authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

Part I. Typological Surveys 1. Word accent: Terms, typologies and theories. . . . . . . . . . . . . . Harry van der Hulst

3

2. An overview of word stress in Australian Aboriginal languages . Rob Goedemans

55

3. Stress types in Austronesian languages . . . . . . . . . . . . . . . . . . Ellen van Zanten, Ruben Stoel and Bert Remijsen

87

4. Word stress and pitch accent in Papuan languages . . . . . . . . . Ellen van Zanten and Philomena Dol

113

5. Accent in the native languages of North America . . . . . . . . . . Keren Rice

155

6. The languages of Middle America . . . . . . . . . . . . . . . . . . . . . Harry van der Hulst, Keren Rice and Leo Wetzels

249

7. A Survey of South American stress systems. . . . . . . . . . . . . . . Leo Wetzels and Se´rgio Meira

313

8. Accent in African languages . . . . . . . . . . . . . . . . . . . . . . . . . . Laura J. Downing

381

9. Word accent systems in the languages of Europe. . . . . . . . . . . Harry van der Hulst

429

10. Word accent systems in the languages of Asia. . . . . . . . . . . . . Rene´ Schiering and Harry van der Hulst

509

11. Word accent systems in the Middle East . . . . . . . . . . . . . . . . . Harry van der Hulst and Sam Hellmuth

615

12. A typology of stress patterns. . . . . . . . . . . . . . . . . . . . . . . . . . Rob Goedemans

647

xiv

Table of contents

Part II. StressTyp Data Australian languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

669

Austronesian languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

691

Papua New Guinea and Irian Jaya languages . . . . . . . . . . . . . . . .

731

North American languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

745

Middle American languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

771

South American languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

775

Eurasian languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

788

Asian languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

825

Middle Eastern and languages . . . . . . . . . . . . . . . . . . . . . . . . . . . .

833

Appendix: The StressTyp Codes . . . . . . . . . . . . . . . . . . . . . . . . . . Subject index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Language index to Part I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Language index to Part II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

847 851 857 876

Part I.

Typological Surveys

1.

Word accent: Terms, typologies and theories

Harry van der Hulst 1. Introduction This chapter provides a discussion of terminology, typology and theories in the domain of word accent. Section 2 discusses the use of terms like ‘accent’ and ‘stress’ and promotes the use of the former term as designating a unique abstract property of domains (such as words) which serves as a reference point for a number of phonetic, phonological and other grammatical properties. Section 3 discusses the idea that we can distinguish several levels of accentuation. Section 4 asks whether words can be without accent (unaccented), while section 5 briefly addresses variability in the location of accents. Section 6 reviews typologies and theories of accent locations. In section 7 we deal with the notion of syllable weight. Finally, section 8 o¤ers a summary and some conclusions. Where required, we will use illustrative examples that clarify central properties of our subject, even though an extensive analysis of these examples may ultimately raise problems of various kinds that cannot be dealt with here.

2. Terminological practices 2.1. Accent and cues In dictionary entries lexicographers often use a graphic symbol, adjacent to or on top of one of the letters, to indicate what is called the location of ‘accent’ or ‘stress’. We will use these two terms interchangeably until, below, we will explicitly propose to di¤erentiate them in a particular way. If a phonetic transcription is added to the spelling form, the accent symbol is often a small superscripted vertical line which is placed before the syllable that is accented. Sometimes the accented syllable is capitalized. One of these practices is illustrated with a few random examples from an English dictionary: (1) escalade [¨EskleId] . . . escalate [¨EskleIt] . . . escallop [¨EskloUp] . . .

4

Harry van der Hulst

The graphic indication of accent is meant to provide information regarding the correct pronunciation of the entries. In the examples at hand, the idea is that the syllable following the symbol is pronounced in a manner that makes it perceptually more ‘‘salient’’ than the other syllables. Let us consider what this might mean by discussing a specific example. Consider the word hippopotamus. Almost every speaker of English will pronounce the medial syllable of this word with more ‘force’ than all other syllables. Phonetically, articulatory force involves a number of properties which the accented syllable has to a greater degree relative to the unaccented syllables. For example, the vowel of an accented syllable is more precisely articulated (less centralized) and/or longer (cf. Gussenhoven 2004: 14–15).1 In many other places one also finds reference to accented syllables having higher amplitude levels and higher pitch. In this respect, one also finds the term ‘‘intensity’’ which refers to greater energy but is sometimes used as a cover term for the cumulative e¤ect of duration, amplitude and pitch. Preciseness of articulation, spectral tilt, duration, amplitude, energy and pitch are all ‘stretchable’ properties that all vowels (indeed all segments2) have to some degree, but it would seem that the accented vowels have all of these to the greatest degree. In addition, the consonants of the accented syllable may have special phonetic properties such as articulatory precision, longer duration, aspiration, a¤rication, etc. Unaccented syllables have these properties to a lesser degree, which sometimes a¤ects their articulation, causing vowels to reduce or consonants to be realized more ‘weakly’. Many of these phenomena can be observed in our example which might be transcribed phonetically as follows: [hIpphaRms]. In this word, the second, fourth and fifth vowel is [], the schwa, a vowel that never occurs in an accented syllable, where we always find a full vowel quality. The occurrence of the vowel [I] and [a] in the first and third syllable seems to indicate that these syllables are not stressless or without accent, which is an indication of the fact that even though the third syllable stands out over the rest, the rest is not undi¤erentiated, but instead displays a rhythmic organization with ‘beats’ and ‘non-beats’; the first syllable apparently has a ‘beat’, which is very much like an accent, although it does not provide its syllable with the same force that the third 1. Gussenhoven also refers to a more technical notion ‘spectral tilt’, discussed in Sluijter and van Heuven (1996) which involves a more even intensity distribution across the frequency spectrum for accented vowels. 2. Of course, voiceless consonants technically have no pitch level.

Word accent: Terms, typologies and theories

5

syllable has. We will return to this phenomenon of multiple accents, with only one being ‘primary’, in section 3; for the moment we will focus on the primary accent. Notice that the /p/ of the medial syllable is aspirated, unlike the /p/ of the second syllable. Traditionally, [ph ] and [p] are called allophones (‘realizations’) of one phoneme /p/. The lexical representation of the word only has a segment /p/. The aspirated segment, as the stronger allophone, is derived by an allophonic rule. Since [ph] only occurs in accented syllables (if not preceded by /s/) its presence is a cue of accent. Thus, one might say that English has a process ‘add aspiration’ which ensures that the lexical form /pin/ is rendered as [phIn]. The flap [R] is also a cue of accent, since this ‘weaker’ sound can only occur following an accented and preceding an unaccented vowel, as an allophone of, in this case, /t/. Besides these phonetic allophonic cues, accent also has phonotactic cues. Phonotactic cues for accent lie in the regularities that govern the distribution of phonemes in the word (i.e. phonotactics). Regularities in the phonotactic structure of words involve two classes of statements, i.e. statements about the inventory of phonemes (segment structure constraints) and statements about the possible combinations of these segments (sequence structure constraints). An important subset of the latter is the class of syllable structure constraints. It is not unusual to make general statements about the segmental inventory and the set of possible syllable structures of a given language without reference to the position in the word that the syllables occur in. However, it is important to realize that often certain syllables allow more segment types and more complex combinations than others, depending on their position in the word. At this point it will not come as a surprise to learn that the syllables that allow ‘more’ are the ones that can be called accented. Compared to accented syllables, non-accented syllables display a reduction in segmental and sequential options, i.e. a neutralization of contrastive options. In fact, the occurrence of schwa in unaccented syllables can be interpreted in this way. We could say that, in English, accented syllables allow a contrast between all the (full) vowels that the language has. Whereas, in unaccented syllables, we only find a very small set of vowels, notably a vowel called ‘schwa’.3 In fact, this schwa does not occur in accented syllables so that the sets of vowels in accented and unaccented syllables are, in fact, complementary. By all standards, the schwa is less complex than the 3. Bolinger (1981) identifies three reduced vowels in English, forming a triangular system: happy, hollow, and father).

6

Harry van der Hulst

other, full vowels, and this is a typical result: accented syllables allow more complexity in terms of the feature combinations that make up segments. This account is based on the idea that English schwa is a lexical segment (as suggested in Bolinger 1981). A di¤erent, and perhaps more common interpretation (following Chomsky and Halle 1968), of the occurrence of schwa, which was implied in the preceding paragraph, is that the occurrence of schwa is ‘merely’ an allophonic e¤ect. In fact, both approaches (allophonic and phonotactic) are not mutually exclusive and indeed, languages like Dutch and German are commonly analyzed as having both lexical and allophonic schwas.4 Another example of phonotactic, distributional asymmetries concerning vowels is found in various Romance languages. In those cases, the accented syllable allows all vowels that the language has. Unaccented syllables di¤er depending on whether they follow or precede the accented syllable. In post-accent syllables only one vowel is found (a schwa), whereas in syllables that precede the accented syllable a subset of the vowels is found (typically /a/, /i/ or /u/) (Cristo´faro-Silva 1992 for Brazilian Portuguese). The di¤erences between syllables following and preceding the accent will not be addressed here; they may involve di¤erent degrees of accent (see section 3). The point that is important now is that accented syllables allow a larger set of vowel phonemes than unaccented ones (possibly with a greater degree of accent correlation with a greater degree of contrastive options). A set of specific cases of a neutralization of contrast in unaccented syllables should be mentioned here because they involve features whose phonetic correlates are duration and pitch, which can be mere phonetic cues of accent. A language that has a length contrast among vowels might have that contrast only in accented syllables, unaccented syllables allowing only short vowels. The Australian language Yidiå (Hayes 1982) dis4. In general, an allophonic cue of accent may, over time, phonologize and thus become a phonotactic cue. As might be expected, linguists may di¤er in their analysis of a given language with respect to judging the cue as being ‘phonetic’ or ‘phonological’. In the phonological analysis, instead of adopting an allophonic rule of vowel reduction, the analysis would postulate a phonological repair rule that replaces a full vowel by a schwa phoneme when an accented vowel alternates with an unaccented occurrence in a morphologically related form. This may be the proper analysis for English where ‘vowel reduction’ is complete and obligatory. In Dutch, on the other hand, even though this language has lexical schwas, vowel reduction is clearly incomplete and allophonic.

Word accent: Terms, typologies and theories

7

plays this property. Do we say that in those languages duration is a cue of accent? No! Rather, we should say that the manifestation of a length contrast is a cue of accent. There is a subtle, yet crucial di¤erence between a language that has no length contrast, but adds duration to all vowels in accented syllables (sometimes only in open syllables as in Mohawk) and a language that limits a length contrast to the accented syllable. In a similar vein, we can find languages that show a tonal contrast only in accented syllables (such as Swedish).5 Many languages that are called tonal, in fact, show evidence for accent in that tonal reduction rules are operative that eliminate or reduce contrast in certain syllables (which, in my view, would be unaccented); cf. van der Hulst and Smith (1988). Such reduction rules are just like the reduction rules that lead to vowel reduction (full to schwa) in English. The only di¤erence is that di¤erent distinctive features are involved.6 Another distributional phonotactic di¤erence between accented and unaccented syllables involves sequential complexity or complexity in syllable structure. A syllable can contain consonants that precede the vowel (collectively called the onset) or consonants that follow the vowel (the coda). All languages allow monoconsonantal onsets, which in some cases is the only possibility. Other languages allow monosegmental onsets and ‘empty’ onsets (i.e. no consonant at all). Still other languages allow the onset to contain more than one consonant. As for the coda, some languages do not allow them, others do and may apparently even tolerate more than one consonant in the coda.7 Linguists agree that a simple CV syllable (C for consonant, V for vowel) is the most common (or least ‘marked’) syllable form (more common than just V), which illustrates that what is common is not necessarily the most simple in terms of the number of segments. They also agree that increasing the options (two consonantal or ‘empty’ onsets, presence of coda, and so on) leads to more marked syllable types. With this background, we return to asymmetries 5. The property of amplitude is not used distinctively in any language, so this cue can only be an ‘allophonic’ manifestation of accent, and not be involved in a neutralization of contrast situation. The same must be said of energy. This leaves duration, pitch and vowel quality (i.e. full version schwa, or perhaps peripheral versus central, i.e. precision of articulation) as properties that could be phonetic cues or phonotactic cues. 6. Again it could be that such phenomena are phonotactic or allophonic. 7. Whether prevocalic and postvocalic consonants clusters are technically tautosyllabic with the vowel is of course a matter of analysis; cf. van der Hulst and Ritter (1999).

8

Harry van der Hulst

between accented and unaccented syllables. In languages that allow more than just CV, it is not untypical to find that only accented syllables display the full array of syllable types that is allowed by the language. Unaccented syllables may be restricted to CV, for example. In Dutch, for instance, a syllable that contains a lexical schwa (which is always unaccented or indeed unaccentable) cannot be preceded by a complex onset (Kager and Zonneveld 1986). Downing (this volume) also draws attention to phonotactic cues of accent, with special reference to African languages. She refers to the fact that ‘‘in a number of West African Niger-Congo languages one finds that the full range of consonant contrasts can typically only be found in C1 position of roots. . . . In C2 position, contrasts are often reduced drastically.’’ (p. 384). She also remarks, on the basis of this and other examples: ‘‘Even though African languages with contrast distribution asymmetries do not have the usual stress-accent motivating the segmental asymmetries, it is plausible to propose that, universally, these kinds of distributional restrictions are to be accounted for in terms of accent, as they create culminative prominence asymmetries within a bimoraic/bisyllabic constituent akin to a stress foot. These segmental asymmetries are, moreover, demarcative, typically highlighting a major morpheme edge, like root- or stem-initial syllable.’’ (p. 385). The segmental di¤erences between the contrastive options that can occur in accented and unaccented syllables, as well as the di¤erences in syllable types that are allowed in these two circumstances are clear examples of phonotactic cues for accent.8 Summarizing, accentual cues may be of a wide variety of types including phonetic, allophonic phenomena, as well as distributional asymmetries in phonotactic patterns. Given the separation of accent and its cues, and given the variety of cues, we should perhaps also allow for the possibility that accents (in certain languages or certain word positions) correlate with cues that have not yet been established or gone unnoted because certain potential cues, especially the phonotactic one were not reckoned with. It 8. Certain positions, here called accented, have been referred to as displaying ‘positional faithfulness’ in the Optimality Theory literature. See for example Beckman (1998). In this literature, positions that promote contrast are not necessarily called accented or stressed. Rather, faithfulness is attributed to salient positions which may be accented/stressed or, for example, close to an edge without necessarily being accented. The spirit of what is proposed here is to identify ‘salient’ with ‘accented’ in all cases.

Word accent: Terms, typologies and theories

9

is, in addition, possible that accents are merely signaled by playing a role in morphological processes involving the selection of specific allomorphs or the site of infixation.9 Finally, accents may be exclusively cued by the fact that certain syllables form anchors for intonational tones; see section 2.2. That in English accented syllables, which indeed form anchors for intonational tones, are also cued by other factors does not imply that such other factors must be present in all languages. When phonotactic, morphological and intonational factors are all taken into account, the number of languages with noticeable stress/accent will go up. This, of course, then raises the important question whether we should entertain the hypothesis that perhaps all languages are accentual. Given that a total of approximately 90% of the languages that are covered by the StressTyp/WALS samples discussed in Goedemans (this volume) have been described as having stress or accent, based on the ‘usual phonetic cues’ (rather than on the basis of phonotactic, morphological or intonational criteria), it is perhaps not outrageous to hypothesize that indeed all languages have stress/accent.10 The list in (2) summarizes the cues for stress/accent: (2) Phonetic and phonological properties of accented syllables a. The syllable has greater duration b. The syllable has a balanced spectral tilt c. The syllable has greater amplitude d. The syllable has a higher fundamental frequency (Ppitch) e. The segments are pronounced with greater precision or extra phonetic traits (such as full vowel quality, aspiration) f. Extra phonotactic possibilities, i.e. greater complexity g. The syllable marks sites for morphological processes h. The syllable is an anchor for intonational tones This list is not meant to be finite. Whatever the details, an accented syllable will di¤er from unaccented syllables in having ‘more’ of whatever ‘stretchable’ property any syllable may have (such as duration, spectral tilt, amplitude, fundamental frequency, precision of articulation; cf. Lehiste

9. An area that could be studied is ‘prosodic morphology’ which may provide pointers to a word accentual structure that is not otherwise cued. 10. If we allow for the logical possibility that there could be no cue for accent at all one might characterize the claim that all languages are accentual as an unfalsifiable hypothesis (Hyman 2008).

10

Harry van der Hulst

1970), allow greater phonotactic complexity or is somehow singled out by other components of the grammar (morphology, intonation). In English, it would seem that all the cues in (2) together signal the location of the accent. Referring specifically to the cues in (2a–e), phoneticians have asked for the longest time whether it is the case that there is a di¤erence in importance between the various cues of ‘stress’. By working with digital speech recordings one can, in principle, eliminate or reduce one or more factors and see whether speakers are still able to hear where the accent is. Or one can test the various cues separately under conditions of interfering noise, and establish which factor is the most robust one. These are interesting and important matters, but they will not be discussed here (see Lehiste 1970, Fox 2000, 120 ¤.). We will simply take it as a fact that accent in English is manifested as a collection of several kinds of cues which need not be equally important under all circumstances. While many languages have been recognized as displaying the phenomenon of word stress/accent as just described, many other languages display a partly similar phenomenon in that a particular syllable in the word is singled out, the di¤erence being that the cues for its special status are not the same set of phonetic, allophonic and phonotactic cues that we find in English. Here, let us take a look at a language that, apparently, manifests its accent in a di¤erent way, Safwa (Bantu) as described in Odden (1988: 227). Consider the following words or word combinations: (3) ¨ami-ino ¨gami-ino mi-i¨no in¨koombe ¨iim-bisi in¨koombe m-bi¨si

‘teeth’ ‘the very teeth’ ‘it is teeth’ ‘uncooked beans’ ‘the beans are uncooked’

Again, we have provided certain vowels with what is often (and appropriately) called an ‘accent mark’. As in the case of English, speakers of Safwa will presumably perceive the syllables that contain these accented vowels as more prominent than the surrounding syllables. When we now look at the articulatory and acoustic properties of the vowels in question, it appears that what distinguishes them from other vowels in the word is just their relative higher pitch. Thus, the relevant vowels are singled out by only one of the properties that can cue the presence of accent in English, namely pitch. Another similar case that is often mentioned, is Tokyo Japanese. The syllable that stands out in this language is again mainly or even exclusively

Word accent: Terms, typologies and theories

11

associated with a high(er) pitch. In this case there is additional leftward spreading of this high pitch up to but not including the first syllable;11 McCawley (1968), Haraguchi (1977). Consider the four following nouns in (4a): (4) a. b.

inoti ‘life’ ¨inoti

kokoro atama ‘heart’ ‘head’ ko¨koro ata¨ma

Japanese words have a specific ‘tonal pattern’ (L) H (L). There is always an H (possibly spread out over more than one syllable) and sometimes a preceding L and/or a following L. Instead, however, of literally assigning tones to words, it may be su‰cient to mark the syllable after which there is a drop to low, as in (4b). Then we assign high pitch to the marked syllable and the syllable preceding it, except for the first mora. The ‘mark’ in (4b) can be regarded as an accent and thus Tokyo Japanese is quite like Safwa in that the accented syllable is realized with high pitch, the di¤erence being that in Japanese this high pitch ‘spreads’ leftwards.12 Let us now introduce a specific use of terms like ‘accent’ and ‘stress’. Hyman (1977) referred to systems such as Safwa and Tokyo Japanese as pitch-accent systems, while proposing to use the term stress-accent systems for cases like English. In this terminological proposal, the term ‘accent’ refers to the property of ‘standing out’, while the terms ‘stress’ and ‘pitch’ refer to the cues that signal the location of the accented syllable.13 Fox (2002: chapter 3) provides a general discussion of the notion ‘accent’ and shows how this term is used di¤erently in di¤erent traditions. The use proposed in Hyman (1977) (accent as an abstract property of words or morphemes without any specific content) represents a tradition of usage that can also be found in Abercrombie (1976), Laver (1994) and Beckman (1986), and is adopted by Fox. In this practice, then, we adopt ‘compound’ terms of the form ‘X-accent where ‘X’ characterizes the cue that correlates with accent. Hence ‘pitch-accent’ means: the accent is cued by

11. This should actually be the first half of the first syllable, or rather the first mora; cf. section 7. 12. There are words without a lexically marked accent which will be discussed in section 4. 13. The traditional distinction between musical accent and dynamic accent refers to the di¤erence between pitch-accent and stress–accent languages.

12

Harry van der Hulst

pitch (typically, but not necessarily high pitch).14 It would, furthermore, seem that in this practice, the term ‘stress’ refers to a rather heterogeneous collection of phonetic properties (such as found in English). It may be that such a collection of properties represents a true type, but it may also be that, on closer investigation so-called pitch-accent languages also involve additional phonetic cues such as duration, but to a much lesser extent than English. In fact, this is the view of Beckman (1986) who divides accent systems in stress-accent and non stress-accent systems, pitch-accent systems falling in the latter category. Whereas stress-accent systems make use of a variety of cues, pitch-accent systems, while perhaps using the same variety, single out pitch as the primary cue. (English, as we will see below, is perhaps less varied in terms of its phonetic cues than we have made it sound thus far, duration perhaps being the primary cue in this case.) But, one might ask, why not refer to Japanese and Safwa as a toneaccent (X ¼ tone) languages, meaning that the accent location is manifested by attracting a tone, H in this case. There is a reason to avoid this term for these languages. Generally, linguists reserve the term tone for cases in which di¤erent pitch levels are used distinctively, just like we use the term length when di¤erent degrees of duration (usually just two) are used distinctively. So when we talk about tone languages, we talk about those cases in which the languages in question allow (all or some) vowels in a word to have di¤erent distinctive pitch levels, typically high and low, sometimes including one or two mid options, or even falling or rising tone contours. These tonal properties can be seen as forming part of the vowels, just like their other properties (open vs. closed jaw, front vs. back raising of the tongue, rounded lips vs. spread, and others). It is crucial, then, that in those cases, the pitch property does not function to make one syllable more prominent than the other(s). Rather, pitch is used distinctively as part of the inventory of distinctive features for vowels, or segments in the rhyme of the syllable (which may sometimes include consonants as well).15 It is crucial to see that tone and accent are not mutually exclusive notions. It is possible, as we have remarked earlier, that tonal distinctions may be limited to the accented syllables (as in Swedish) which would be an instance of phonotactic cues to accent, or that tonal distinctions are neutralized in unaccented syllables. Following the terminological practice 14. We are not aware of a system in which the accent is cued by low pitch. 15. Here we ignore the idea that tones may link to ‘structural’ units such as moras or syllables.

Word accent: Terms, typologies and theories

13

promoted here, we should adopt the term tone-accent language for languages in which the tone contrast is limited to the accented syllable.16 Accordingly, we do not wish to call Safwa a tone (or tone-accent) language, because there are no contrastive tones in this language. Rather pitch is used as a cue to accent. Hence we refer to this language as a pitch-accent language. Given that there are, as we have seen, many cues for accent, it is likely that we must reckon with, for example, ‘duration-accent systems’ (in which duration is the primary or only cue). In short, it is possible, if not likely, that accent may be signaled in terms of many combinations of cues of various sorts and that among these cues one may be the strongest, most salient and thus primary cue. If this is pitch, we speak of a pitchaccent language, if it is duration we speak of a duration-accent language, etc. The cross-linguistic array of accentual cues is at present far from understood and much detailed phonetic and phonological analysis is needed in this respect. The important point here is that it seems useful to separate the notion accent (a ‘formal’ property of syllables without any link to specific cues) and accentual cues, so that we can develop a theory about the former without specific regard for the latter, and vice versa. At this point it may be useful to illustrate the various ways in which a specific phonetic property may correlate with accent. We will use the property ‘pitch’ for this illustration: (5)

16. In fact Hyman (1987) uses the term ‘tonal accent language’ for this case.

14

Harry van der Hulst

In (a) and (b), tone-accent languages, the distribution (a) or realization (b) of tones is dependent on accent. Type (c) represents the case in which accent location is sensitive to tone (e.g. accent is on the last high tone and if there is no high tone on the last syllable, as in Golin; cf. Hayes 1995: 278–9). Case (d) has tone and accent, but there is no correlation. Moving to non-tonal systems, case (e) is a pitch-accent language. Case (f ) is a language in which words have non-distinctive pitch properties (such as inherent pitch correlating with vowel height, or phonation of consonants due to which native speaker experience the sensation of ‘accent’. This case may be di¤erent from the previous types in that the notion ‘accent’ in this case may be ‘non-structural’, and thus epiphenomenal; Finally, in case (g) we have accent and phonetic pitch properties that are not correlated. A similar schema of possible relations can be drawn for other phonetic properties such as duration and ‘articulation precision’ and we refer to van der Hulst (in prep.) for further discussion and exemplification. The schema in (5) strengthens the important point that we must separate the notion of accent from the cues that may or may not correlate with its location. We do not, however, wish to suggest that cues that correlate with accent are motivated by the need to signal the accent. Rather it would seem that accents, being there (for whatever reason) in some cases merely lend themselves as ‘reference points’ for these other grammatical and phonetic properties. The separation of accent and cue, and the recognition that there is a diversity of cues, explains why attempts to find a unique phonetic basis for accent (whether in articulation, acoustics or perception) never led to a generally accepted proposal (Fox 2000: 120– 127). Below, we will see that the raison d’eˆtre of accents may be the fact that words and large utterances have an organizational structure of some kind of which accents are an inherent part (cf. Fox 2000: 149 ¤.). Having separated the notion ‘accent’ from whatever its cues are, it will now be possible to focus on the notion of accent alone and discuss a theory of accent that simply ignores cues. Likewise, one could focus on developing a typology of accentual cues (cf. 2 above). The crucial issue for a theory of accent locations will be, as we will see, the regularities (if any) that govern the location of accents within the word, i.e. what these locations are and how they can be formally represented and explained. It is an empirical (and interesting) issue as to whether the typology of accent locations and the typology of phonetic cues are totally independent, or whether there are dependencies and constraints that limit certain cues to certain accent locations, or vice versa (see Dogil and Williams 1999).

Word accent: Terms, typologies and theories

15

2.2. Intonation Meanwhile, we have to be aware of other terminological practices. In one (represented by Bolinger 1958, and later work), the term ‘stress’ is used for the abstract property of words, while the term ‘accent’ is used for the observable manifestation that stressed syllables (may) have. This usage is apparently opposite to the use of these terms promoted here. Bolinger is specifically referring to the fact that ‘stressed’ syllables function as anchors for tonal events that make up the intonational melody of sentences. In fact, and this adds to the confusion, the intonational events are called ‘pitch-accents’. Before we can establish whether, these two terminological views are irreconcilable, we must review a specific approach to intonation that follows Bolinger’s term use, although not all other aspects of his theory of intonation. Intonational melodies can be understood as unanalyzable ‘holistic events’ or as being composed of smaller building blocks which are either seen as pitch movements (one approach), or as pitch levels (another approach). The pitch-level approach was proposed in Liberman (1975) and worked out in Pierrehumbert (1980).17 Since there are distinctive (combinations of ) pitch levels (expressing di¤erent intonational meanings), the units that make them up are phonological tones, namely H and L. The manner in which the tones that make up these pitch accents are lined up with the words in the utterance expresses information regarding which parts of the utterance are ‘‘important’’. Additional tones (called boundary tones) add further structure to the intonation melody. Thus, intonation contours provide cues bearing on the overall syntactic and semantic structure of utterances, i.e. the grouping of words into meaningful ‘‘chunks’’ and the informational status of these chunks. Consider the following simple example (which ignores boundary tones): (6) a.

H | Larry bought [a small hippopotamus]

b.

H | Larry bought a [small] hippopotamus.

Let us assume that the utterance in (6a) is an answer to the question: What did Larry buy? The important part of the utterance is in that case a small 17. See Ladd (2009), Gussenhoven (2004) for discussions of this approach.

16

Harry van der Hulst

hippopotamus. We say that the phrase in question is in focus. In (6a) we only indicated the pitch-accent (a simple H tone) which signals the phrase that is informationally foregrounded (or is ‘in focus’). The point of interest to us is that the pitch accent, which designates this part as important, is lined up with a particular syllable in the relevant phrase, more specifically with a particular syllable of the word hippopotamus. This is precisely the syllable that the lexicographer would represent as being accented (or, as others would say ‘stressed’). In English, it would be inconceivable to line up the pitch peak with the syllable mus. The reason is that this syllable does not bear accent. Note that if we line up the H tone with (the accented syllable of ) the word small as in (6b), the relevant utterance would more likely be an answer to the question: What kind of hippopotamus did Larry buy? In the answer to this question, the phrase in focus is small. The examples in (6) show that an intonational tone that signals focus associates with a specific word in the relevant phrase. Speakers of English, then, must know which word in a focused phrase will make its accented syllable available for this function. Looking at (6) one might suggest that it is the last word in a focused phrase that does this, but matters are not that simple. We refer to Fuchs (1976), Gussenhoven (1984), Baart (1987) and Selkirk (1984, 1995) for extensive discussion of these issues.18 The association locus of intonational pitch accents is often referred to as the phrasal accent which, then, is the (primary) accent of the word in the focused phrase that is selected for this purpose. Thus, a particular syllable that carries the word accent can at the same time carry a phrasal accent. This fact shows that syllables can be accented with reference to several inclusive domains, i.e. words and phrases.19 Word accents, then, can reveal themselves by functioning as anchoring points for some of the tones that make up the intonation melody. 18. Here we cannot explain why a pitch accent on the noun hippopotamus can cause the whole noun phrase including the adjective to be in focus. In fact, the intonation structure in (6) is ambiguous in that it could also be an answer to the question ‘What did Larry buy that was small?’. In that case the ‘in focus’ part of the answer is only the noun hippopotamus; cf. Selkirk (1995) for an introduction to this phenomenon. 19. If the same phrase is uttered without being focused, we will assume the phrasal accent is still there, possibly having phonetic cues, even though it is not associated to an intonational tone. This would follow if we take the notion ‘phrasal accent’ to also be abstract, like word accent, and not intrinsically linked to cues.

Word accent: Terms, typologies and theories

17

Let us now return to Bolinger’s terminological schema. Bolinger uses the term pitch accent for intonational pitch events which are properties of phrasal units. Earlier we used the term pitch-accent to characterize a system like that found in Japanese. For this language we could in fact refer to the pitch events as pitch accents, which, in this case would be properties of words. Pitch-accents, then, are pitch events that can be associated with units of di¤erent sizes, i.e. words or phrases. But why are pitch accents in Japanese associated with ‘accents’, while at the phrasal level they are associated with ‘stresses’. It is only a little step to say that Bolinger’s phrasal pitch accents are in fact also linked to accents, namely phrasal accents. Accepting this point, we can reconcile the two apparently contradictory terminological usages in the following manner: (7)

(In (7) we leave undecided whether phrasal accents can be promoted further, e.g. to sentence accents.) Thus, intonational pitch-accents are analogous to the pitch-accents in Japanese, the di¤erence being the domain that these pitch-accents belong to. The only quarrel one could have with Bolinger’s use of the term ‘pitch accent’ is that, in English, there are at least two phonological distinct pitch levels, namely H and L, and various combinations of these, which implies that the term ‘tone-accent’ would be more appropriate. Intonational pitch-accents, in our view, would exist in languages that realize phrasal accents in terms of a non-distinctive pitch event. This discussion of intonation, meant to clarify a particular tradition in the uses of terms like stress and accent, allows us to make an important point about the set of cues for accent in a stress-accent language like English. The role that word accents play in attracting intonational tones

18

Harry van der Hulst

has, in the past, given rise to the idea that the word level stress-accented syllables in English are primarily cued by ‘pitch’. However, if the relevant words form part of phrases that are not in focus, implying that their accent does not attract an intonational tone, this ‘cue’ disappears, while others remain (although there may be a modest pitch elevation still which is independent from intonation). The most notable cue that remains appears to be duration (see e.g. Van Heuven en Sluijter 1996). 2.3. Other terminological practices The term ‘pitch-accent’ system is used di¤erently in at least two of the chapters in this volume. Van Zanten and Dol (this volume) reserve this term for systems in which ‘‘one syllable is more prominent than the other syllables in the same word, a prominence that is achieved by pitch.’’ (p. 120) This prominence must be noticeable by native-speakers. This definition thus excludes many cases which would be pitch-accent systems according to the preceding section where perceptible prominence is not a criterion. Downing (this volume), on the other hand, uses a definition that is broader than what we have proposed. She refers to any language that has some kind of interaction between tone/pitch and accent as a pitch accent language. This would include cases in which accent assignment is dependent on tone.

3. Levels of stress/accent We have so far focused on the notion of primary accent (at the word or phrase level). However, when we consider the ‘prominence pattern’ of such units they appear to have two distinguishable general characteristics. Firstly, there is one syllable that prevails over all others in the domain, the so-called primary accent. Secondly, there tends to be a regular alternation of ‘beats’ and ‘non-beats’; this is their rhythmical aspect. Considering the rhythmic profile of words or phrases, we perceive their primary accents as the strongest beat, but other syllables may have beats too, that are subordinated to the primary accent beat. We will call these beats rhythmic (or non-primary) accents, although we will question below whether the term ‘accent’ should, in fact, be used for syllables that are rhythmically ‘strong’.

Word accent: Terms, typologies and theories

19

Even though the primary word20 accent is introduced as a property of a particular syllable, it is quite clearly a property of the whole word, a point that is also strongly suggested by the fact that each dictionary entry is normally provided with at most one such symbol. This property of accent is often called culminativity. Primary accents are ‘‘maxima’’ of some kind, which implies that each such accent ‘‘signals’’ the presence of one accentual ‘‘domain’’. One might say that primary word accents function to signal the number of words in a sentence. Moreover, we can say that if two primary word accents are detected, a word boundary must be somewhere in between. Thus, primary accents may play a role in parsing sentences into their constituting words. In fact, in languages where the location of accent is on a fixed syllable in the word (e.g. the first one, as in Icelandic, Hungarian and Czech), the exact boundary between words can be uniquely determined. But also if the accent location is not exactly at the edge and instead is separated from the edge by one syllable (as in Polish where the location is the penultimate syllable), accents still provide clear information about the parsing of sentences into words. This is what is called the (potential) demarcative function of accent. The culminative and demarcative functions of primary accent were the focus of study within the Prague School (see Fox 2002: 134–138).21 This school did not deny that accent can sometimes be contrastive, namely in languages in which its distribution is ‘free’, but this function of accent was clearly seen as marginal. This being said, rhythmic beats can also have a demarcative role at the word level if, for example, one edge is always or typically marked by a rhythmic beat; this, for example, is the case in Dutch or English where initial syllables (if not primary accented), tend to have a ‘‘secondary accent’’. The culminative property of accent implies that accent is a syntagmatic property, i.e. a property of the linear structure of units that form the accent domain. Syntagmatic properties contrast with paradigmatic properties, i.e. properties that can be present or absent on more and possibly all linearly arranged units that form a domain. Thus, vowel frontness is paradigmatic if any vowel in a word can be front (or back) in principle. On the other hand, some languages (such as most of the Finno-Ugric and Turkic languages) show a phenomenon of vowel harmony which involves (roughly 20. We will focus here on the word domain. Below we return to the phrase domain. In general, we believe that both domains display entirely analogous properties. 21. They largely neglected rhythmic structure (cf. Fox 2002).

20

Harry van der Hulst

speaking) the situation that all vowels in the word must be front or back. In such cases, frontness is in fact a syntagmatic property, rather than a paradigmatic property. Therefore, Garde (1968) proposes to refer to vowel harmony as accentual. From a functional point of view, harmony probably indeed helps to parse sentences, since a shift from front to back vowels (or vice versa) in principle marks the vicinity of a word boundary. Thus, harmony may be said to have an identifying and demarcative function, like accent. The analysis of accent within the tradition of American structuralism can be characterized as an attempt to analyze accent as a paradigmatic property (Fox 2000: 127–134). Hence, within this tradition there are multiple ‘values’ of accent, just like other paradigmatic properties can have multiple (although typically two) values. This brought the abovementioned rhythmic accents (which the Prague school did not pay much attention to) to the forefront. As Fox (2000) shows, various proposals did not limit the values of stress to two (primary, non-primary), but instead distinguished sometimes up to 8 degrees, although more commonly 4 degrees: (8) a. b. c. d.

Primary accent Secondary accent Tertiary accent Weak or no accent

¨father ¨door¨¨bell ¨ele˙vator all vowels not marked for stress

As can be seen, non-primary accents appear to have di¤erent sources. The tertiary accents are what we will call rhythmic beats, whereas the secondary accent on bell is a ‘demoted’ primary accent which will be called a cyclic accent. Within the SPE tradition (Chomsky and Halle 1968), in principle, an infinite number of accentual levels is recognized, because in this model, accent is assigned cyclically to morphologically or syntactically complex domains. Each domain has a higher degree of stress than less deeply embedded domains. However, it has been widely recognized that there is no requirement that all these levels correspond to perceptible levels of prominence and, moreover, it has also been claimed that these cyclic accents often ‘make way’ for rhythmically assigned beats; cf. Giegerich (1985), Hayes (1995, ch. 9) and Visch (1999). We return to cyclic accents in section 5 and to the apparent conflict between cyclic accents and rhythmic beats below. Some linguists propose a more modest inventory of accents. Fox (2000) proposes to recognize one level of accent within the word domain. He calls

Word accent: Terms, typologies and theories

21

this level 1 accent. All accent types in (8) would be level 1 accents in his view which implies that words, both simplex and complex, can have more than one such accent. The di¤erentiation of level 1 accents is a matter of superimposing an intonational peak on certain level 1 accents which operates at the phrasal level. In another view (one that we promote), the assignment of intonational peaks is ‘fed’ by rules that assign phrasal accents, some of which may then be anchors for intonational tones. Fox’s suggestion is to skip the assignment of phrasal accents and directly assign intonational tones to certain level 1 accents. A language that has level 1 accents is called a stress-timed language by Fox. He also suggests that languages may do without level 1 accent, mentioning French as a case in point, which would be a syllable-timed language. He then refers to languages such as Spanish and Italian that, while syllable-timed in his view, do also appear to display rhythmic level 1 accents; he considers these cases ‘unclear’. Gussenhoven (2004: 19–22) advocates a similar view. In his terminology words can have one or more, what he calls, ‘stressed’ syllables (corresponding to Fox’s level 1 accents). He also introduces the notion ‘accented’, which means that the stressed vowels have an intonational pitch configuration (i.e. a pitch accent); this seems to correspond to Fox’s level 2 accent. However, in Gussenhoven’s model, the accent mark (‘‘*’’) is formally distinct from an ‘intonational pitch accent’. All stressed vowels will bear the accent, but then there are rules that will delete accents in certain positions; the resulting vowels remain stressed, but are no longer possible anchors for intonational pitch accents. He proposes three constraints that govern the distribution of accents in English: (9)

a.

All lexical words have an accent

b. c.

In compounds the right hand member’s accent is deleted In complex words, all accents except the rightmost one are deleted

(10) a.

* (*) [door][bell]

c.

(*) * Rembrandtesque

d.

* (*) elevator

b.

(*) * (*) [[kitchen][[towel][rack]]

e.

* * ballerina

22

Harry van der Hulst

In kitchen towel rack (a compound with a branching right hand member) the accent on towel remains because towel by itself is not a righthand compound member. Then rule (c) will delete the accent on kitchen in (10b). In (10c), a lexically stressed/accented syllable prevails over the stress/accent of the stem due to rule (9c). What remains unaccounted for is how it is decided which stressed/accented syllable in a long simplex words gets to keep the accent which is, as shown, not always the second one (cf. 10d and 10e); Gussenhoven seems to assume that in ballerina both accents are preserved, while in elevator the second one has to be lexically specified as being unaccented. Whereas Fox is quite explicit in having only one level of accent within the word domain (hence no notion of primary versus secondary word accent, but simply allowing multiple word accents), Gussenhoven, with his distinctions in (10) between stressed/accented and stressed/unaccented syllables, reconstructs the distinction between primary and non-primary ‘stress’. After all, his notion of accent means potential for pitch accent. The syllables that loose their accentability are still considered to be stressed and indeed, even though these syllables can not be anchors for intonational pitch accents, they still display stress-cues such as extra duration. Gussenhoven acknowledges that ‘‘. . . the distinction between secondary and primary stress must be phonologically represented independently of the presence of an accent.’’ (Gussenhoven 2004: 21). Also, syllables that remain accented do not necessarily end up being linked to a pitch-accent. In (6a) repeated here as (11), all lexical words have accents, but only the accent of hippopotamus is linked to a pitchaccent: (11)

H | * * * * * Larry bought [a small hippopotamus]

In the British tradition (cf. Fox 2000: chapter 5), the pitch accent in (11) would be called the nuclear accent. Preceding accented syllables can have intonational pitch properties too, which may or may not be called pitchaccents. It is sometimes proposed that in such a case, i.e. if a focus phrase has more than one pitch accent, we have to assume that the focus domain is split. However, this approach cannot explain that a single word can

Word accent: Terms, typologies and theories

23

have more than one ‘pitch accent’ as remarked in Gussenhoven (2004: 22): ‘‘. . . English words frequently have two pitch accents, one on the main stress, and one on a preceding secondary stress, like . . . in sardine.’’ It would be odd to suggest that a single word can be split into two focus domains. A solution is to recognize the earlier pitch accents as ‘copies’ of the final ‘nuclear’ pitch accent. (cf. Fox 2000: 301 ¤.). Here we will not consider these issues in any further depth. We summarize Gussenhoven’s distinctions in (12):

(12)

The terminology that we propose would be di¤erent, although the same distinctions are made:

(13)

Terminology aside, we must note another di¤erence in view which has to do with the question whether domains have just one accent or whether

24

Harry van der Hulst

they can have several. Both Fox and Gussenhoven assume that words can have several accented syllables, whereas we seem to imply that domains have only one accent, somewhat redundantly called the primary accent. We can illustrate the di¤erence by considering hippopotamus. We would say that there is an accent on the antepenultimate syllable, whereas the first syllable has a rhythmic beat. Where, then, in our view do rhythmic beats enter the scene? Here, following van der Hulst (1984, 1999, 2009a among others), we adopt a model in which accents (as cumulative properties of domains) are assigned ‘first’, whereas rhythmic beats are assigned ‘later’. Returning to the example in (11), it has been claimed that, quite independent of pitch accents, primary accented syllables may display a rhythmic organization and this has suggested to researchers that, at the phrasal level, we need to distinguish between primary (phrasal) accent and nonprimary (i.e. rhythmic) beats. (14)

* * * * * * * * * * * Larry bought [a small hippopotamus] | H

In this display (called a metrical grid; cf. Liberman and Prince 1977) primary word and phrase accents are represented by underlined asterisks and it is assumed that the rhythmic structure is also represented by asterisks. The column height represents rhythmic strength. Note that the rhythmic organization respects the localization of accents in two ways. Accents are rhythmically strong and, secondly, they are rhythmically stronger than rhythmic beats that do not correspond to accents. At this point the question arises whether word-internal rhythmic beats are any di¤erent from phrasal rhythmic beats. Often, phonologists speak of the rhythmic structure of words without realizing, or making explicit, that these words are considered in isolation which e¤ectively provides them with the realization of a phrase. Also, it has often been remarked that the rhythmic structure of words can be dependent on the ‘phrasal context’ (Tı`condero´ga versus Fo´rt Tico`ndero´ga; Prince 1983). This suggests that we do not need to distinguish between modules for word and phrasal rhythm, but that rhythm is phrasal, or rather belongs to a level of repre-

Word accent: Terms, typologies and theories

25

sentation at which we consider the way in which linguistic expressions are actually realized. Following this model, in languages like English there are rules for the assignment of word accent and phrasal accent, each domain having exactly one such accent. Phrasal accents are assigned to word accents according to certain principles. The phrasal accent is the locus of focal pitch accents which mark the fact that a phrase is in focus. We assume that all these aspects belong to ‘grammar proper’, i.e. to the mental module that accounts for the grammatical properties of linguistic expressions. Then, as a separate procedure, utterances are provided with a rhythmic structure which (a) fills in the spaces in between word accents by adding beats on syllables and (b) creates rhythmic alternation of word accents, again by adding rhythmic beats to certain word accents. Hayes (1984) provides clear demonstration of the rhythmic principles which are responsible for the creation of structures which are optimally eurhythmic. At the lowest level, rhythmic beats may be sensitive to syllable weight, just like primary accent can be. The separation of accentuation and rhythm can be argued for on several grounds. Perhaps the most powerful argument is that whereas accentuation rules can be governed by lexical and grammatical information (as it is in English, both at the word and phrasal level), rhythmic rules are blind to such information. The separation of accent (as a property of linguistic expressions) and rhythm (as a property of utterances) also suggests a solution to the ‘problem’ that the non-primary accents that are predicted by the cyclic application of accent rules (both at the word and phrase level) do not always ‘show up’, but rather can be overwritten by rhythmic beats. There is no conflict here if we maintain that accents are abstract structural entities, belonging to cognitive linguistic expressions, whereas rhythmic beats are properties of utterances (i.e. instances or specific implementations of linguistic expressions). As abstract entities, accents, whether primary or demoted (i.e. cyclic), do not necessarily correspond to any phonetic cue, although they can. Rhythmic beats on the other hand belong to the realm of utterances and as such they are typically associated with phonetic cues. Whereas rhythmic beats typically ‘respect’ the location of primary accents, they are less faithful to non-primary cyclic accents. When cues of non-primary accents and cues of rhythmic beats together produce an a-rhythmical e¤ect, either cyclic accents give up their cues, or rhythmic principles are ‘stretched’ (from binary to ternary patterns) to achieve a match between cyclic and rhythmic accents. Let us now return to the proposal made by Fox (2000) that some lan-

26

Harry van der Hulst

guages, such as French, lack word accentuation. All phrases are accented on the final syllable which he attributes (as we have seen earlier) to the assignment of an intonational unit (although we could also say, as argued above, that a phrase accent is assigned, which subsequently forms an anchor for intonation units). The question is whether in addition to phrasal accents, we can think of any reason for first assigning a final accent to each individual lexical French word. If primary word accents are understood as inherent properties of a structural organization that all words, in all languages, necessarily have, we would say that there are word accents which happen to not correspond with any specific cues. However, this remains a speculation at this point. We could also say that in French there is only a phrasal accent which is assigned to the last syllable in the phrase. However, we could push this further. Given its fully automatic character, it isn’t even obvious that French has a phrasal accentuation rule. The ‘phrasal accent’ could be an ‘utterance e¤ect’ and as such be regarded as part of the rhythmic component. As a result, French would have no accentuation rules at all, neither at the word level, nor at the phrasal level. The entire prominence pattern of French phrases could be due to the utterance level rhythmic component. We refer to Gussenhoven (2004) for an analysis of French accent and intonation and Dell (1984) for careful analysis of the rhythmic properties of French. In this perspective we would assign accent rules only to languages in which the location of accents (at the word and/or phrase level) shows signs of sensitivity to grammatical information (i.e. lexical diacritics, morphological information, syntactic structure). On the other hand all languages would have an utterance-level rhythmic component which would show respect to primary accents and would be less faithful to (cyclic) nonprimary accents: (15)

English

French

Grammar (linguistic expression)

word accent

–

phrase accent

?

Implementation (utterance)

rhythm (prosodic structure)

rhythm (prosodic structure)

In (14) we represented the rhythmic structure as a pure grid, i.e. as having no grouping. This does not need to be so. The implementational component can be understood as building not only a rhythmic structure but also imposing a grouping which sets domains to which various segmental

Word accent: Terms, typologies and theories

27

sandhi processes (cf. Nespor and Vogel 1986) and ‘tune’ processes respond.22 If this view is on the right track, it would follow that the so-called prosodic hierarchy, or rather some version of it, belongs to the implementational level together with all processes that are dependent on it. This would include the fast-speech, or more generally, the ‘P2’ rules in Kaisse’s model (Kaisse 1985, 1990). E¤ectively, then, there are two phonologies (Kaisse 1990, van der Hulst 2009b). The first phonological system interacts with the morpho-syntactic structure (‘direct reference’), whereas the second system is e¤ectively the implementational system. The structure that this system creates is dependent on the morpho-syntactic structure perhaps only to the extent that it reckons with the accents assigned in the first system and the division of sentences into focus domains which can be understood as grammatical directives for constructing the prosodic organization of utterances.

4. Unaccented words? Returning to common dictionary experience, we might note that some entries (or words), particularly those consisting of one syllable, are not provided with the accent symbol. To the user of a dictionary this causes no problems since he uses the tacit rule that in such cases the accent falls on the only syllable there is. That a monosyllable can bear accent suggests that ‘‘being accented’’ is not a purely relative notion as has sometimes been suggested. Among the monosyllabic words in English there is a majority which must always be pronounced with a full vowel quality, i.e. not a schwa, but for a small category of words (like articles), a pronunciation with schwa is perfectly possible. This appears to indicate that not all monosyllables are accented. Indeed, many languages have ‘little words’ that do not belong to such categories as Noun, Verb, Adjective and Adverb, but rather to the classes of conjunctions, pronouns, articles etc. which typically do not appear to have a primary accent, thus also lacking the kinds of cues that are associated with accent in the relevant language. Such words are often called (phonological) clitics, and in the sentence these clitic words ‘attach’ them-

22. The combination of rhythm and grouping could be formally represented in terms of a bracketed grid (Halle and Vergnaud 1987) or in two di¤erent planes (Nespor 1990).

28

Harry van der Hulst

selves to accented words, forming ‘extended words’ which are often called prosodic words or clitic groups.23 These ‘little words’ may have two variants, one accented and one unaccented. This is rather typical for pronouns. In this case, the accented variant will be used if the pronoun stands in the right place in a focused phrase that it is properly contained in or if the pronoun itself forms the focused phrase: (16) a.

b.

H | I saw him (reduced) H | I’m going with him (full)

Some clitic words can only bear intonation tones if they are themselves placed in focus, as in the following utterances: (17)

H H | | I didn’t say [A] long introduction, I said [THE] long introduction

In this case, the unaccented word is not properly contained in a focused phrase, but rather forms a focused phrase by itself. In such cases, it would seem that an accent is forced onto the word, which is then typically pronounced with a full vowel. A natural question at this point is whether polysyllabic words must have an accent, or, put di¤erently, whether phonological clitics must be monosyllabic. We observe that in English, and in many other languages, there are no polysyllabic words that contain only syllables with a schwa, which seems to indicate that unaccented words must be monosyllabic. However, in other languages, clitic words that are polysyllabic do occur. Also, clitics di¤er in whether or not they ‘influence’ the accent location in the base in the sense that accent seems to be re-applied to the extended word (host plus clitic(s)). Here we refer to Peperkamp (1997) and Vogel (2009) for a broader discussion of the accentual properties of clitics. Another issue that we need to investigate further is the fact that in so-called pitch accent languages, we often find reference to a class of unaccented words, words that belong to major categories and therefore 23. Syntactic clitics are words that occupy special syntactic positions without necessarily being phonologically enclitic.

Word accent: Terms, typologies and theories

29

are not clitics. Thus, in Tokyo Japanese there is a class of nouns that is not lexically accented (see 18b). When used without a su‰x, the final syllable acquires an pitch-accent ‘by default’, whereas if a su‰x is added, the default pitch-accent will fall on that su‰x. Thus there is a subtle di¤erence between nouns with final lexical accent and nouns without lexical accent: (18) a.

ata¨ma-ga

b.

sakura-ga

Here we will leave undecided whether the default ‘pitch-accent’ is indeed an accent or just pitch (see van der Hulst, in press).

5. Variability or uncertainty in accent locations? We have seen that in English longer words have non-primary ‘accents’ (i.e. rhythmic beats) in addition to primary accents: (19) hu´rrica`ne te´lepho`ne pa´radı`se a`pala`chico´la The desire to mark non-primary ‘accents’ stems from the fact that not all syllables lacking the primary accent are felt to be equal in salience, and this salience, as we have seen, a¤ects the pronunciations of the words in sometimes significant ways. In English, for example, syllables marked with a non-primary accent symbol cannot have a pronunciation with a schwa-like vowel. They have a full-vowel quality, a property they share with primary accented vowels. Still, such syllables are felt to be less salient than the primary accented syllable and furthermore they normally fail to function as anchor points for intonational tones. At the end of section 3, we have suggested that whereas the primary accents are due to lexical accentuation rules (which take into account grammatical and lexical properties of morphemes to account for the fact that the location of primary accent is far from ‘automatic’), rhythmic beats could be attributed to an utterance-level rhythmic component.24

24. By relegating rhythmic beats to the implementational level, we seem to be committed to the view that the English schwas in all rhythmically weak syllables are due to phonetic implementation, i.e. due to allophonic reduction rules. However in all cases in which schwas do not alternate with full vowels, they can regarded as being lexical(ized).

30

Harry van der Hulst

Opinions sometimes di¤er with respect to the location of syllables that bear rhythmic beats. This is especially so if these beats do not manifest clearly detectable phonetic cues and one therefore has to rely on impressionistic judgments or ‘‘intuitions’’. We must also acknowledge the fact that linguists may claim to perceive rhythmic beats that objectively are simply not present in the signal. If such perceptual beats are, or can be, biased by a linguist’s own language, this may result in attributing a rhythmic signature to the language being described that native speakers of that language would not perceive. Di¤erences in opinion with respect to the location of rhythmic beats may of course also be due to the fact that the location of these beats are dependent on the phrasal context in which a word occurs, or performance factors such as speech style, rate of speech and so on. In general, rhythmic accents are more di‰cult to study for reasons of this sort, and many descriptions simply do not mention them at all. If non-primary ‘accents’ are indeed purely rhythmic then we would not expect that their location is dependent on anything else than the syllabic make-up of words and phrases. However, there are cases in which the location of non-primary accents appears to be dependent on the location of the primary accents of the words that are embedded in complex words. This is seen most clearly in the case of compounds, in which the primary accent of the right-hand member (in English and Dutch) surfaces as a nonprimary accent; the examples are from Dutch: (20) a.

* * almanak

* committee

b. * * almanak

* commissie

c.

* * * almanak omnibus

Here the location of the secondary accent follows the pattern of the primary accent location in the right-hand member. Clearly, these nonprimary accents are not rhythmic, but rather cyclic. We have suggested that cyclic accents belong to the realm of grammar proper. The question is whether non-primary accent locations of complex

Word accent: Terms, typologies and theories

31

words that do not involve compounding can also be dependent on the accentual patterns of their parts. It has been argued that this is indeed possible. Consider the following ‘classical’ examples: (21) co´mpensa`te co`nde`ns

co`mpensa´tion co`nde`nsa´tion

The second syllable in the words on the right are di¤erent in that in ‘compensation’ the vowel is reduced (schwa), whereas in ‘condensation’ it is (or can be) pronounced with full vowel quality. The explanation is that the words on the right are derived from words that already have a lexical accent and this lexical accent ‘persists’ in the derived words by showing up as a cyclic accent.25 Whereas cyclic accents in compound words are not controversial (but see below), the reality of cyclic accents in derived words has been called into question. Firstly, condensation can easily be pronounced with a reduced vowel in the second syllable. In this case it could be said that the rhythmic component overwrites the cyclic accents. A further possibility is that the cues of the cyclic accents are simply incorporated in the lexical representation of the phonological form of the morphemes as ‘weight’ (i.e. as actual heavy syllables). In this case the two words in (21) on the right have lexical representations in which vowels may be schwas: co`mp«nsa´tion, co`ndnsa´tion. Bolinger (1981) adopts this perspective. The issue of whether or not accents are assigned cyclically repeats itself at the phrasal level, and proponents of cyclic application (plus rhythmic adjustment) and non-cyclic application can be found; cf. Hayes 1995, chapter 9; Gussenhoven 2004: 282 ¤.). Earlier, we suggested that cyclic application, i.e. application of accent rules with reference to morphosyntactic structure is not necessarily at odds with rhythmic patterns that do not reflect the abstract cyclic accents. We need to accept that accents (both primary and cyclic) are abstract entities. Whereas primary accents are respected by the rhythmic component, demoted, cyclic accents can

25. Fox (2000, 129 fn 16) refers to Daniel Jones’ claim that non-primary accents can play a ‘linguistic role’. He cites the word certification which may have secondary accent on the first or second syllable with a di¤erence in meaning. The form with initial accent means ‘act of certifying’, while the form with secondary accent on the second syllable means ‘granting a certificate’. Clearly, Jones was observing cyclic accents since the location of the secondary accent in this case can be predicted in this way if we assume that the two forms are derived from certify and certificate, respectively.

32

Harry van der Hulst

easily be overwritten by the rhythmic component if their distribution is not compatible with the ‘laws of rhythm’. Disagreement with respect to primary accent location is untypical. If there is disagreement about primary accent location this usually means that there are two possible primary accentuations of the word. Consider the following examples from Dutch: (22) he´lsinki chı´mpansee

helsı´nki ‘Helsinki’ chimpanse´e ‘chimpanzee’

Usually in such cases one of the accent locations is exceptional whereas the other is a regularized form. The initial accentuations in (22), for example, violate the rule in Dutch that primary accent cannot lie to the left of a penultimate closed syllable. But even for primary accent location systematic disagreement sometimes occurs, especially if the language lacks clearly detectable phonetic cues (cf. the studies in Ode´ & van Heuven 1994 on ‘stress’ in Indonesian). 6. Possible locations for accent26 Let us now consider what kind of theories can account for accentual structure. Two broad approaches can be found which di¤er in terms of the dependency between primary accent and rhythmic accent: (23) Metrical theory Rhythm % Primary accent

Alternative Primary accent % Rhythm

Before we discuss these alternatives and their merits, let us (re)familiarize ourselves with the basic generalization concerning accentual systems. The diversity in accentual systems is huge. A search in StressTyp reveals no less than 132 di¤erent ways in which languages can encode the location of primary accent. In an insightful global typological survey, we therefore need to limit ourselves to a presentation of the main trends only. Such a presentation is, however, impossible without an introduction of some of the basic parameters that underlie the surface accentual patterns.

26. This section is in part based on section 2. in van Zanten and Goedemans (2007).

Word accent: Terms, typologies and theories

33

We will leave the statistics for these surface patterns, and a quantitative analysis of the accent parameters themselves for more theoretically oriented work, such as Goedemans (this volume, chapter 11). We divide the languages into two groups: (i) a group of languages that always have primary accent on a particular syllable in the word, the so-called fixed accent languages, and (ii) a group of variable accent languages in which the location of accent is not the same for every word but depends on one or more word-internal factors. The location is fully determined for each given word, but for the lexicon as a whole we see that various accent locations occur. To keep the proliferation of surface patterns in check, we dispense with all language-internal exceptions. Languages may easily have a rule that places primary accent on the last syllable in all words, except for a small group of words, in which accent is located on the penultimate syllable. For our purposes, we will deal with such languages as if they were purely final-accent languages. The reduction thus achieved leaves us with a manageable set of possible surface patterns for both groups of languages. Let us look at these more closely. 6.1. Fixed primary accents Limiting the attention to primary accent, let us first note that the following six positions in the word have been attested as possible accent locations: (24)

Left Initial Czech Finnish

Right Second Dakota

Third Winnebago

Ultimate Turkish French

Penultimate Polish

Antepenultimate Macedonian

We find three terminological systems to refer to one of three right edge locations: (25) Terms for right edge accent: Ultimate Penult(imate) Antepenult(imate)

Final Prefinal (Pre-prefinal)

Oxytone 27 Paroxytone Preparoxytone

27. The . . . tone terminology, which prevails in the study of Romance languages, seems to be based on the idea that the principal cue for accent is pitch.

34

Harry van der Hulst

For left edge locations there is no commonly used set of terms, but we do find expressions like: (26) Terms for left edge accent: Initial/First syllable accent Second syllable accent (‘post-initial’/‘pen-initial’) Third syllable accent (‘post-post-initial’/‘post-pen-initial’) Any student of word accent must deal with this kind of terminological proliferation which, in any event, does not seem to be based on any theoretical di¤erences. As shown in (24), fixed accent languages come in six flavors: (i) A fairly large number of languages have initial accent (I). An example from this group is Ono (Trans New Guinea Phylum; Papua New Guinea); ¨ and ˙ denote main and secondary accent, respectively): (27) ¨lolot˙ne

‘many’

¨ari˙mage˙ake

‘he always goes’.

(ii) A few languages have accent on the second syllable (S). Siroi (Trans New Guinea Phylum; Papua New Guinea) exemplifies the pattern: (28) ku¨mah ‘dead’

ku¨bele ‘yesterday’.

(iii) Only one language in our sample, Winnebago (Siouan; Illinois), exhibits accent on the third syllable (T) (see also Hayes 1995): (29) hochi¨chinik

‘boy’

waghi¨ghi

‘ball’.

(iv) The third syllable from the right we call the antepenultimate (A). An example of a language with predominantly antepenultimate accent is Pa’disua (Austronesian; Halmahera): (30) ¨igono ‘coconut’

be¨le/asa ‘shoulder’

(v) The second syllable from the right is the penultimate (P). Penultimate accent is fairly common. A language with this pattern is Lenakel (dialect of Tanna; Austronesian; Tanna Island): (31) kay˙law¨law ˙tIna˙gamyasI¨NvIn

‘kind of dance’ ‘you will be copying it’.

(vi) Languages with accent on the final, or ultimate (U), syllable are exemplified by Weri (Trans-New Guinea; Morobe, Papua New Guinea): (32) u˙lua¨mit

‘mist’

˙aku˙nete¨pal ‘times’.

Word accent: Terms, typologies and theories

35

In one of the first typological surveys Hyman (1977) counts more penultimate than initial cases, final accent coming in third place. Postinitial and antepenultimate are rare. Hardly any cases of accent falling on the antepenultimate syllable have been reported and third syllable accent is not mentioned at all by him. The StressTyp collection confirms this distribution. Of the 506 usable languages in StressTyp, 283, or 56%, exhibit a fixed accent pattern (see Goedemans, this volume [Chapter 11]). Initial and ultimate accent could be accounted for by primary accent rules that simply seek out edges of the accentual domain.28 But what about postinitial, penultimate and antepenultimate accent? Let us first note the observed asymmetry between left edge accent and right edge accent. Whereas the latter seems to be able to ‘‘reach’’ the third syllable from the edge (as in Macedonian), third syllable accent is hardly ever attested, if at all. Even though only few examples of fixed antepenultimate accent occur, this location is frequently found in the exceptional vocabulary of languages that have penultimate accent. A theory of accent placement must perhaps account for this asymmetry, but it certainly must also account for the fact that (fixed) patterns other than those in (24) are never found. If primary accent placement was unrestricted, in the sense that any syllable that is at a fixed distance from the word edge could be reached, we would expect to find languages having accent on the fourth syllable from either the left or right edge. Another pattern that we never find is that accent is placed on the medial syllable in odd-syllable words and in even-syllabled words to the left or right of the word middle. We therefore need a mechanism for primary accent placement that will not allow us to construct such cases. We will first consider what would not be an appropriate mechanism. Suppose we formulate primary accent rules that literally place an accent mark on a particular syllable. We will call this the lexicographic practice. A first drawback of this theory is that it fails to account for the fact that words can have only one primary accent, i.e. it does not account for the culminative property of accent. After all if accent is thus treated as a paradigmatic property (i.e. a local property that can be present or absent on each vowel, or syllable), nothing will stop us from assigning accents to multiple or all vowels in the word. The theory of accent placement proposed in Chomsky & Halle (1968) has this drawback.

28. This domain is ‘the word’ in languages in which a‰xation does not seem to make a di¤erence, and ‘simplex word’ in languages in which complex words call for special attention.

36

Harry van der Hulst

This is the lexicographic practice in a formal disguise, which acknowledges a segmental feature [estress], formally identical to other segmental features such as [eround] and [esonorant].29 The lexicographic practice, then, does not explain the culminative character of accent because no aspect of that theory prevents us from assigning an accent mark to the first and last syllable, or indeed to every syllable in the word.30 The lexicographic practice also does not account for the ways in which accent can exhibit its preference for ‘edge-closeness’. If the rules in (33a–f ) are necessary for the attested cases, we can formulate the rules in (33g–h) just as easily. (33) a.

s

!

x s

/

–

]

(Turkish)

b.

s

!

x s

/

[

–

(Hungarian)

!

x s

/

–

(s)]

(Polish)

/

[(s)

c.

s

d.

s

!

x s

e.

s

!

x s

/

–

f.

s

!

x s

/

[((s) s)

g.

s

!

x s

/

–

h.

s

!

x s

/

[(((s) s)s)

–

(Dakota)

((s) s)]

(Macedonian)

–

(Winnebago)

(((s) s) s)] –

(Unattested) (Unattested)

29. In all fairness, Chomsky and Halle (1968) did propose a convention which lowers the degree of stress of all vowels that are already marked as [þ stress] by a previous application of the same or another stress rule. In this way (excluding simultaneous rule application), all vowels will bear a di¤erent degree of stress. 30. Whether indeed accent is culminative in all cases is discussed in van der Hulst (in press).

Word accent: Terms, typologies and theories

37

Now (33f ) may be a highly improbable option but (33g–h) seem to go beyond what appears to be possible. The lexicographic practice is clearly inadequate as a theory of primary accent placement. To more properly express the restricted edge-based nature of accents, phonologists have, following original ideas of Liberman (1975), Prince (1976) and Liberman & Prince (1977), constructed a framework of parameters leading to what is called metrical phonology. The parameters that are relevant to the discussion here are summed up in (34).31 (34) a.

the location the bisyllabic accent window (the stretch of the word in which main accent can be located). Left or Right edge of the word. the location of accent within the accent window. Left or right, or in more common terms, use a Trochaic (x .) or an Iambic (. x) foot.

b.

c.

non-peripherality or extrametricality. One element, usually a syllable, at one of the word’s edges is extrametrical; it is skipped in the computation of accent locations.

To derive, for instance, the primary accent pattern of Ono in (27), we need to place the accent window at the left side of the word, starting at the first syllable, and construct a trochaic foot there, as in (35). (35) (x .) ¨lolot˙ne Antepenultimate accent in Pa’disua can be derived by locating a window on the right edge of the word which, due to extrametricality skips the final syllable: (36)

(x .) be¨le/asa

By making the bisyllabic accent window the basic tool of the theory, while allowing skipping of one syllable on the edge, this theory accounts for the restricted edge-location of fixed accents.

31. Note that (34) only accounts for the location of primary accent. In most versions of metrical theory feet are the all important building blocks in the derivation of rhythm as well. For the separation of primary accent and rhythm see van der Hulst (1996, 2009a) and section 6.5.

38

Harry van der Hulst

The theory does not answer the question why penultimate accent shows up more often than final accent, nor does it explain the sharp di¤erence between third syllable accent and antepenultimate accent, which are, formally, equally complex. 6.2. Variable primary accents In so-called quantity-sensitive languages, the accent is not fixed on a particular syllable in the word, but neither does the accent rule indiscriminately target just any syllable. It is sensitive to internal properties of the target syllable, or, to use a common term for this phenomenon, syllable weight. Which properties may exactly count towards syllabic weight is something we will discuss in section 7. It su‰ces here to say that syllables are either heavy, or light in a quantity-sensitive accent system. If there is a heavy syllable in the main accent window, it attracts accent, while one of the light ones only receives accent if there are no heavies in the accent window. Naturally, a choice must also be made in case the accent window is filled with two heavy syllables. With this in mind it is not di‰cult to envisage the enormous growth in possible accent types that the introduction of quantity sensitivity entails. Consider the pairs in (37) which show H ¼ heavy and L ¼ light syllables at the right side of the word in a hypothetical language. (37) a. (H L)]

b. (L H)]

c. (H H)]

d. (L L)]

The universal property of a quantity-sensitive system is that in cases (37a) and (37b) accent will always be located on the heavy syllable (shown in bold face). Languages with right-edge windows and weight-sensitivity may di¤er from each other with respect to (37c) and (37d). In case (37c) the accent falls on the final (i.e. rightmost) heavy syllable in most languages. In (37d), the most usual case is to have accent on the left-hand light syllable (i.e. the familiar trochaic pattern). Overall then, the most common right-edge weight-sensitive system would be described in (38). (38) a. (H L)] b. (L H)] c. (H H)] d. (L L)] e.g. Epena Pedee (Choco, Southern Embera, Colombia; only long vowels form heavy syllables): ¨taama ‘snake’ war¨raa ‘flavourful’ tee¨soo

‘long’

¨warra

‘son’

The other logical options for the (H H) and (L L) cases do also occur in natural languages. The Austronesian languages Yapese, Sunda and Aklan

Word accent: Terms, typologies and theories

39

only di¤er from Epena Pedee, and from each other, in their choice for which syllable is accented in these cases. (39) i.

a. (H L)]

b. (L H)] c. (H H)]

d. (L L)]

e.g. Epena Pedee

ii. a. (H L)]

b. (L H)] c. (H H)]

d. (L L)] e.g. Yapese

iii. a. (H L)] iv. a. (H L)]

b. (L H)] c. (H H)] b. (L H)] c. (H H)]

d. (L L)] e.g. Sunda d. (L L)] e.g. Aklan

If we then add to these four logical possibilities the four that may occur at the left edge of the word, we come to eight di¤erent quantity-sensitive accent systems. All eight occur in the languages of the world. The Malayalam (Dravidian; southern India) examples in (40) illustrate the most common left-edge pattern ([(H L), [(L H), [(H H), [(L L)). Ossetic, Archi and Capanahua are examples of languages that use the other three logical left-edge options. (40) i. a. [(H L) ii. a. [(H L)

b. [(L H) c. [(H H) b. [(L H) c. [(H H)

d. [(L L) e.g. Capanahua d. [(L L) e.g. Archi

iii. a. [(H L) iv. a. [(H L)

b. [(L H) c. [(H H) b. [(L H) c. [(H H)

d. [(L L) e.g. Malayalam d. [(L L) e.g. Ossetic

(41) Malayalam (long vowels make syllables heavy) a. ¨kuuam ‘crowd’ b. pa¨aaak˙kaaran ‘soldier’ c. ¨aakaaCam ‘sky’ d. ¨kutira ‘horse’ In both Epena Pedee and Malayalam, accent falls on the heavy syllable that is closest to the word edge, with the trochaic pattern being the default option in case the two candidate syllables are both light. Now remember that initial or final syllables can be made invisible to the accent rules, a phenomenon we called extrametricality. Suppose each of the eight patterns revealed here could occur with or without extrametricality, then we are left with 16 possibilities. From a typological point of view the number of classes now quickly becomes unmanageable, while the aforementioned types do not even exhaust the possible variable accent systems we find in the world’s languages. 6.3. Unbounded systems and their theoretical consequences We must now consider yet another parameter in the discussion of quantitysensitive accent systems: boundedness. So far, we have assumed that the windows in which accent is assigned are always bisyllabic, or bounded.

40

Harry van der Hulst

Unbounded windows also exist, however. In accent systems with unbounded windows, primary accent may occur anywhere in the word. The rules typically favor either the first or the last heavy syllable in the word, placing primary accent at either the left or right edge in the absence of heavy syllables. Thus, we derive the four possible unbounded accent types in (42). (42) a.

Accent the first heavy, or else the first light syllable; Amele, Trans New Guinea phylum b. Accent the first heavy, or else the last light syllable; Tahitian, Austronesian c. Accent the last heavy, or else the last light syllable; Puluwatese, Austronesian d. Accent the last heavy, or else the first light syllable; Sikaritai, Geelvink Bay

All four patterns are attested in the languages of the world. Below we give some examples from Amele (Trans New-Guinea phylum; Madang, Papua New Guinea). (43) Amele (codas render heavy) ja¨walti ‘wind from north’ iti¨tom

‘righteous’

¨nifula

‘species of beetle’

On top of this, extrametricality may come into play once more to move the unbounded window one syllable away from one of the edges and derive patterns that, for instance, place accent on the first heavy syllable, or else on the penult, but never on the ultimate (not even if it is the only heavy syllable). Hence, we can add twelve (four basic ones, and all these with either left- or right-edge extrametricality) unbounded systems to the sixteen bounded ones, and add up a subtotal of 28 possible quantitysensitive accent systems. Unbounded systems can be recognized as such when the location of accent is variable, i.e. dependent on syllable weight. Logically, we could ask what happens when the accent is fixed within the unbounded window. The result of this mix of parameter settings, taking into account extrametricality is yet another 4 possible systems which, however, are descriptively equivalent to fixed bounded systems: (44) Unbounded [(ssssssss)] Unbounded þ EM [s(sssssss)]

[(ssssssss)] [(sssssss)s]

Word accent: Terms, typologies and theories

41

This undesirable ambiguity could be eliminated if it is assumed that the set of parameters is organized in such a way that the unbounded domain choice is only available if weight plays a role: (45)

From a functional point of view, unbounded systems are curious because the location of accents provides no information about word edges. It must be concluded that in systems of this sort the ‘greed’ of heavy syllables in snatching the word accent has overtaken the edge-based preference of the accents that have fallen victim to their attraction. Unbounded systems have always been problematic for metrical theories since, as is evident, the location of accent does not seem to be determined by the choice of feet, which are inherently bounded. In early varieties of this theory (such as Vergnaud and Halle 1978, Hayes 1990) a special category of unbounded feet was proposed which, however, cannot be equated with what we have here called the unbounded domain. Working with such unbounded feet created various ambiguities (cf. van der Hulst, in prep.) and in the end Hayes (1995) presents the majority view in rejecting such unbounded foot types, thus restricting the scope of metrical theory to bounded systems. However, such a strict separation of bounded and unbounded systems is not necessary if, as we did here, we simply adopt the choice of domain (bounded or unbounded) as a basic parameter, without insisting that we construe these domains as ‘feet’. In fact, if we do this, as is proposed in van der Hulst (1996, 2009a), we reveal a clear generalization about the variety of bounded and unbounded systems which is that, once the domain (and extrametricality) has been set, the location of accent is determined by two accent rules: (46) Weight-sensitive Bounded (right/left edge) Unbounded LAST/LAST FIRST/LAST FIRST/FIRST LAST/FIRST

(ss) (ss) (ss) (ss)

(ss) (ss) (ss) (ss)

(ss) (ss) (ss) (ss) (ss) (ss) (ss) (ss)

(ssssssss) (ssssss) (ssssssss) (ssssss) (ssssssss) (ssssss) (ssssssss) (ssssss)

42

Harry van der Hulst

We have four accentual possibilities in both bounded and unbounded domains when weight enters into the location of primary accent. The parameter choice before the slash determines where the accent lies in case the domain contains heavy syllables, whereas the choice after the slash settles the case when there are no heavy syllables. Hence, for example, ‘LAST/LAST’ reads as ‘place an accent on the last (rightmost) heavy syllable and, if there are no heavy syllables, on the last syllable’. The two choices are independent from each other, which leads to four possibilities. Clearly, if weight does not play a role in a system only one choice needs to be made; one could say that, in that case, there is no evidence for setting the weight-choice; as stated earlier, we assume that this option does not apply for unbounded domains: (47) Weight-insensitive X/LAST X/FIRST

Bounded (right/left edge) (ss) (ss)

This domain-based theory, although still bearing a resemblance to standard metrical theory in its use of a foot-like domain for bounded systems, and to the lexicographic theory in assigning accent marks, is, in fact, very di¤erent from both. Within both kinds of domains we distinguish two rules for the selection of what might be called the head of the domain. Translated into standard metrical theory this would mean that for bounded domains we need to choose two foot-types in weight-sensitive systems rather than one, which is the common practice in this theory. That this method fails is evident from the fact that the four-way varieties in (39) and (40) cannot be accommodated in this way without postulating additional ‘retraction rules’ (cf. van der Hulst, in prep.).

6.4. Lexical accent systems To add to the 28 variable-accent languages, there is a final type in which accent is completely unpredictable and thus lexically marked. In so-called lexical accent languages, it is usually assumed that the location of accent may be anywhere in the domain, so that it needs to be specified in the lexicon. Such languages are by some considered to be as ‘weight-sensitive’ as the others mentioned in this section, since in a way the accent rule that

Word accent: Terms, typologies and theories

43

places main accent can be said to be sensitive to syllables that are lexically marked for ‘diacritic’ weight. Note that even a slight degree of lexicality32 may introduce minimal pairs that are di¤erent only in their accent locations, as for instance, in the Meah (Bird’s Head, Papua Province, Indonesia) examples in (48) in which accent is a phonemic (distinctive) property. (48) ¨EREsa ‘go visit’

ERE¨sa ‘child’

Tokyo Japanese, discussed earlier, is a lexical accent system, and so is Russian, in which accent falls on the rightmost accented syllable (of which there can be more than one in morphologically complex words) or on the first syllable if there is no lexical accent. Lexical accent systems are not necessarily unbounded. They can be, as in Japanese and Russian and this implies that the accent location is not restricted to a two- or three syllable window on the right or left edge. However, in bounded systems too we can encounter the case in which accent is variable, let us say ultimate or penultimate, without any way of predicting which choice obtains for any given word.33 6.5. Count systems The following pattern is found in Creek in words that contain only light syllables: (49) In even-syllable words, primary accent final In odd-syllable words, primary accent is prefinal In words that contain heavy syllables, we need to locate the rightmost heavy syllable and then apply the procedure in (49) to the stretch of light syllables following this heavy syllable (Hayes 1995: 64–67). In (50) we only illustrate the case with words that have only light syllables:

32. Languages may use lexical specification of stress for only a part of their vocabulary. In Dutch, for instance, 85% of the vocabulary has stress in a predictable location, leaving 15% of lexically specified exceptions (cf. Langeweg 1988). Many languages in StressTyp act like Dutch. This fact is not considered here, however, because we only look at dominant patterns in order to keep the number of possible stress types in check. 33. In StressTyp such systems are coded as ‘irregular’ rather than lexical, the latter code being applied only to unbounded varieties.

44

Harry van der Hulst

(50) coko´ osa´na apataka´ anocı´ta isimahicita´ itiwanayipı´ta

‘house’ ‘otter’ ‘pancake’ ‘to love’ ‘one to sight at one’ ‘to tie each other’

Here accent appears to be right-edge bounded but the location, while being rule-based, cannot be locally determined on that edge. To know the location one has to group the syllables from left to right into bounded domains in which the head is located on the right: (51) ( . x) isi

(. x) mahi

(. x) cita´

( . x) (. x) (. x) iti wana yipı´ ta

Strictly speaking, we only need to locate the head in the rightmost domain. This iterative way of assigning bounded domains provides the only evidence for the idea, characteristic of metrical theory, that words are parsed into binary domains (‘feet’) from right to left or from left to right before the primary accent location can be selected. In this approach, indeed, primary accent location is determined in two steps: (52) a. ( x) ( . x) (. x) (. x) isi mahi cita´

( x ) Word Tree RH 34 ( . x) (. x) (. x) Feet (L % R) iti wana yipı´ ta

b.

The notation in (52a) is now commonly used after Halle and Vergnaud (1987), while that in (52b) is closer to the original ‘arboreal’ notation in Liberman and Prince (1977). None of the bounded systems considered earlier require such an elaborate procedure for primary accent location which is why we simply adopted the practice to locate one bounded domain on the edge where primary accent falls. This deviates from the standard metrical approach

34. ‘RH’ stands for ‘right-headed’.

Word accent: Terms, typologies and theories

45

which assumes that primary accent is always located by first parsing the entire word into feet and then selecting the leftmost or rightmost foot as the head foot of the word. However, as will be clear, only in systems in which the edge at which footing starts di¤ers from the edge location of the head foot (as is the case in Creek) we crucially must apply iterative footing first. In all other cases we can simply locate one foot-like domain, select the head and leave it to that. At this point it would seem that a powerful argument in favor of the standard procedure could run like this. All this time we have focused our attention on primary accent. However, as discussed earlier, words frequently have rhythmic beats in addition to their primary accent. It would seem that iterative foot structure perfectly captures the distribution of rhythmic beats. Now, in non-count systems, iterative feet that account for rhythmic pattern could be assigned after we have located the primary accent or, and this might be considered more elegant, we might simply say (as metrical theory does) that iterative foot assignment applies first and that primary accent selection is done by selecting a peripheral foot as the head foot of the word. Count systems, then, simply represent the logical possibility where head foot selection chooses the foot that was assigned last, rather than the foot that was assigned first. A second argument could be that this way of doing things explains the bounded nature of primary accents, given that feet are bounded units. In van der Hulst (1984) it was suggested that the metrical ‘bottom-up’ procedure of standard metrical theory is at odds with certain di¤erences that may exist between the rhythmic beats and the beat that is promoted to primary accent status. The most notable di¤erence is that in many languages primary accents, although regular, have lexical exceptions, whereas rhythmic beats always appear to be completely regular. We noted these properties already in section 3. This suggests that primary accent is ‘lexical’, while rhythm is ‘post-lexical’ (or perhaps: post-grammatical, belonging to the implementation system). Roca (1986) also proposes this for Spanish. But if rhythm is post-lexical, primary accent, if lexical, cannot be fed by it and thus has to be assigned independently first. We refer to van der Hulst (2009a, in prep.) for an elaboration of these points which lead to a strict separation of primary accent and rhythm. This separation, which was also suggested at the end of section 3 above, is also reflected in the organization of StressTyp. Thus, whereas Hayes (1995) and most others take the bottom-up procedure as the default case, allowing topdown construction where needed, we have chosen to adopt the top-down

46

Harry van der Hulst

view (primary accent first) as the default, which leaves count systems as the special case. This leaves the question how count systems are accounted for? We refer to van der Hulst (1997, in prep.) for a further discussion of count systems and their theoretical consequences. One proposal is to assign a bounded accent window and suppress the accentual rules that normally would assign a head. This then allows rhythmic beats to ‘invade’ this domain so that a rhythmic beat ends up being the head of the accentual domain. 7. Weight factors To make the quantity-sensitivity picture complete, we should devote some attention to the phenomenon of weight itself. In the previous section, we have seen that languages may accent certain syllables on the basis of their weight. But what is weight?35 The syllabic properties that determine weight can di¤er from one language to the next. An important determinant of weight is vowel length, as we have seen in (38) and (41), closely followed by syllable closure. In principle, both factors are independent, but often both long vowels and closed syllables will cause syllables to be heavy in one and the same language, leaving the category of light syllables for open syllables with short vowels.36 Thus, we distinguish three regular types of weight: (53) a. Long vowels make syllables heavy. b. Closing consonants make syllables heavy. c. Both long vowels and closing consonants make syllables heavy.37 An example of (52c) is Cebuano (Austronesian; Philippines). If the final syllable is light (CV or CVC, i.e. word-final consonants are extrametrical), accent falls on the penult if it has a long vowel or coda: ¨tinda ‘sell’ ˙tagman¨saanas ‘fond of apples’, ˙ka/u¨sa ‘once’.

35. See Goedemans (1998) for a study of phonological weight. 36. Note that, phonologically, short vowels are represented by one segment slot, while long vowels have two such slots. Apparently, the presence of a segment after the vowel, be it another vowel (identical, or di¤erent, as in diphthongs) or a consonant, can make syllables heavy. 37. Current metrical theory excludes prevocalic consonants as a weight factor. However, weightful geminate onsets are reported for Pattani Malay by Hajek & Goedemans (2003).

Word accent: Terms, typologies and theories

47

Considering the enormous diversity we already have encountered in the world of accent, one might guess this is not the complete story. And indeed it is not. To these three weight factors, we must add another one, labeled prominence (Kenstowicz 1994). In systems that use prominence to determine whether syllables are heavy or light, certain properties of the segments in the syllable count towards weight, not their mere presence. There are various properties that come into play here. One of the most important ones is tone. In languages that have contrastively pitched syllables (i.e. tone languages), accent may be sensitive to such distinctions and, for example, be located on the leftmost or rightmost high-pitched syllable in the accent window. Look at the Sikaritai (Trans New Guinea Phylum; Papua Province, Indonesia) examples in (54) in which the last high-toned syllable is accented, or else the first syllable (acute accent indicates high tone). (54) sE@bE@¨k"@ ‘narrow’ ¨aparE

hu@¨Ra@RE ‘male’

‘handle’

Another prominence factor concerns vowel aperture, or more generally vowel quality. If overall vowel quality is relevant, the opposition typically is reduced (light) as opposed to full (heavy vowels). We will see such vowel quality systems abundantly in the Austronesian languages (chapter 3). If aperture is relevant, more open (low) vowels will count as heavy, as opposed to closed (high) vowels. The Yindjibarndi (Pama-Nyungan; Western Australia) examples in (55) show initial accent unless the second syllable contains a low long vowel. (55) ¨martuur˙raa

‘twilight’

nyi¨laarti

‘native mead’

This behavior reflects a general tendency among prominence factors. Many of these divide syllables such that the more sonorous ones are heavy while the others are light. The next prominence factor directly relates to consonant sonority. In Inga (Quechuan; Colombia) only sonorant codas make syllables heavy, while syllables ending in obstruents are light. The final syllable is accented if it is heavy, otherwise accent is penultimate. Some examples are given in (56). (56) ya¨war apa¨muy

‘blood’ ‘to bring’

¨kancˇis ‘seven’ kam¨kuna ‘you.PL’

48

Harry van der Hulst

Languages with other heavy-light divisions among the set of possible codas exist as well. Quite often these involve the glottal stop. In Mam (Mayan; Guatemala), for instance, weight is assigned according to a scale (a phenomenon we find more often) in which syllables with long vowels are the heaviest, followed by syllables that have a glottal stop in the coda. Syllables closed by any other consonant than the glottal stop are at the bottom of the scale. To the four weight categories described above we add a final one. We have already noted in section 6.4. that we view syllables that are lexically marked for accent as heavy. Therefore we must now add lexical marking to the weight factors. 8. Summary In the beginning of this chapter we have promoted the use of the term ‘accent’, making the point that accents can be manifested in a variety of ways. We proposed to follow Hyman (1977) in making a distinction between stress-accent languages and pitch-accent languages. The generalizing notion for him was accent which does not have an inherent cue or exponent. In a pitch-accent language, the accent is cued by a pitch property (an elevated pitch or a pitch rise, typically). In a stress-accent language, the manifestation is ‘stress’ which he took to be the kind of properties that are typically associated with ‘stress’ in languages such as English (extra duration, extra loudness, hyper-articulation etc.). However, we then saw that there is no reason to limit the manifestation possibilities of accent to these two cases. For example, accent could be manifested by duration alone (a duration-accent language), or by full vowel quality (accentless vowels being reduced), etc. In addition, the accented syllable may distinguish itself from non-accented syllables by a greater array of phonotactic possibilities, or by being the locus of tonal distinctions, or by being the anchor point for intonational tones. By making a distinction between accent and cue we allowed ourselves to then focus on the factors that determine the location of the accented syllable in the word. We have identified the role of domains and syllable weight as critical factors. We also promoted the idea to separate the treatment of primary and rhythmic accents which together make up the prosodic structure of words. We hope that this introduction will provide the reader with elementary terminology and theoretical tools to organize the variety of accentual systems that will be discussed in the subsequent chapters of this book.

Word accent: Terms, typologies and theories

49

References Abercrombie, David 1976 ‘‘Stress’’ and some other terms. Work in Progress. Department of Linguistics, University of Edinburgh 9: 51–3. Baart, Joan 1987 Focus, Syntax, and Accent Placement. Towards a Rule System for the Derivation of Pitch Accent Patterns in Dutch as Spoken by Humans and Machines. [Unpublished Ph.D. dissertation, University of Leiden.] Beckman, M. 1986 Stress and non-stress accent. Dordrecht: Foris publications. Beckman, Jill 1998 Positional Faithfulness. Doctoral dissertation, University of Massachusetts, Amherst. Bolinger, D. L. 1958 A Theory of Pitch Accent in English. Word 14: 109–49. Bolinger, D. L. 1981 Two Kinds of Vowels, Two Kinds of Rhythm (distributed by the Indiana University Linguistics Club, Bloomington, Indiana). Chomsky, N. and M. Halle 1968 The sound pattern of English. New York: Harper and Row. Cristo´faro-Silva, Thaı¨s 1992 Nuclear phenomena in Brazilian Portuguese. Doctoral dissertation, School of Oriental and African Studies, University of London. Dell, F. 1984 L’accentuation des phrases en Francais. In: F. Dell, D. Hirst and J. R. Vergnaud (eds.), Forme sonore du langage, 65–122. Paris: Hermann. Dogil G. and B. Williams 1999 The phonetic manifestation of word stress. In: H. van der Hulst (ed.), Word Prosodic Systems in the Languages of Europe, 273– 334. Berlin, New York: Mouton de Gruyter. Downing, Laura This volume Accent in African languages. Fox, Anthony 2000 Prosodic features and prosodic structure. Oxford: Oxford University press. Fuchs, Anna 1976 ‘Normaler’ und ‘kontrastiver’ Akzent. Lingua 38: 293–312. Garde, P. 1968 L’Accent. Presses universitaires de France, Paris. Giegerich, H. J. 1985 Metrical Phonology and Phonological Structure. Cambridge University Press, Cambridge.

50

Harry van der Hulst

Goedemans, Rob 1996 Syllable weight and prominence. In: R. Goedemans, Harry van der Hulst & E. Visch. Stress Patterns of the World: Part 1 Background, 115–163. HIL Publications 2. The Hague: HAG. Goedemans, Rob 1998 Weightless Segments: A phonetic and phonological study concerning the metrical irrelevance of syllable onsets. Ph.D. dissertation, University of Leiden. LOT Dissertations 9. This volume A Typology of stress patterns. Gussenhoven, Carlos 1984 On the Grammar and Semantics of Sentence Accents. Dordrecht: Foris Publications. Gussenhoven, Carlos 2004 The Phonology of Tone and Intonation. Cambridge: Cambridge University. Hajek, John & Rob Goedemans 2003 Word-initial geminates and stress in Pattani-Malay. The Linguistic Review 20: 1, 79–94. Halle, Morris and Vergnaud, Jean-Roger 1987 An essay on stress. Cambridge, MA: MIT press. Haraguchi, Shoshuke 1977 The tone pattern of Japanese: an autosegmental analysis. Tokyo: Kaitakusha. Haspelmath, Martin Matthew Dryer, David Gil and Bernard Comrie 2005 The World Atlas of Linguistic Structures, (eds.). Oxford: Oxford University Press. Hayes, Bruce 1982 Metrical structure as the organizing principle of Yidiny phonology. In: Harry van der Hulst and Norval Smith (eds.). The structure of phonological representations. Part I, 97–110. Dordrecht: Foris Publications. Hayes, Bruce 1984 The phonology of rhythm in English. Linguistic Inquiry 15: 33–74. Hayes, Bruce 1990 Precompiled phrasal phonology. In: Sharon Inkelas and Draga Zec (eds.), The Syntax-Phonology Connection, 85–108. Center for the Study of Language and Information, Stanford, CA, and University of Chicago Press, Chicago. Hayes, Bruce 1995 Metrical stress theory: Principles and case studies. Chicago: University of Chicago Press. Hulst, Harry, van der 1984 Syllable Structure and Stress in Dutch. Dordrecht: Foris. Hulst, Harry, van der 1996 Separating primary accent and secondary accent. In: R. Goedemans, H. van der Hulst and E. Visch (eds.), Stress Patterns

Word accent: Terms, typologies and theories

51

of the World, 1–26. HIL Publications 2. The Hague: Holland Academic Graphics. Hulst, Harry, van der 1997 Primary accent is non-metrical. Rivista di Linguistica Linguistica 9/1: 99–127. Hulst, Harry, van der 1999 Word accent. In: H. van der Hulst (ed.), Word Prosodic Systems in the Languages of Europe, 3–116. Berlin and New York: Mouton de Gruyter. Hulst, Harry, van der 2002 Stress and accent. In: Nadel, L. (ed.), Encyclopedia of Cognitive Science. Vol. 4, 4: 246–254. London: Nature Publishing Group. Hulst, Harry, van der 2006 Word stress. In: Keith Brown (ed.), The encyclopedia of language and linguistics. 2nd edition. Volume 13, 655–665. Oxford: Elsevier. Hulst, Harry, van der 2009a Brackets and grid marks or theories of primary accent and rhythm. In: E. Raimy and C. Cairns (eds.), Contemporary views on architecture and representations in phonological theory, 225– 245. MIT press. Hulst, Harry, van der 2009b Two phonologies. In: Janet Grijzenhout and Baris Kabak (eds.), Phonological domains: universals and deviations, 315–352. Berlin: mouton de Gruyter. Hulst, Harry van der (in press) Pitch accent systems. In: Marc van Oostendorp et al (eds.). The Blackwell Companion to Phonology. Wiley-Blackwell. Hulst, Harry, van der (in prep) Word accent. Ms. University of Connecticut. Hulst, Harry van der and Nancy Ritter 1999 Theories of the syllable. In: Hulst, Harry van der & N. Ritter (eds.). The syllable: views & facts, 13–52. Berlin: Mouton de Gruyter. Hulst, Harry van der and Norval Smith 1988 The variety of pitch-accent systems. In: H. van der Hulst & N. Smith (eds.), Autosegmental studies on pitch accent, i–xiv. Dordrecht: Foris. Hyman, Larry 1977 On the nature of linguistic stress. In: L. Hyman (ed.), Studies in stress and accent, 37–82. Scopil 4. Hyman, Larry 1987 Prosodic domains in Kukuya. Natural Language and Linguistic Theory 5, 311–334.

52

Harry van der Hulst

Hyman, Larry 2008

Universals in phonology. In: Harry van der Hulst (ed.), Examples of linguistic universals. Special theme issue of The Linguistic review, volume 25, 1/2, 35–82. Kager, Rene´ & Wim Zonneveld 1986 Schwa, syllables and extrametricality in Dutch. The Linguistic Review 5, 197–221. Kaisse, Ellen 1985 Connected speech. The interaction of syntax and phonology. Orlando: Academic Press. Kaisse, Ellen 1990 A typology of post-lexical rules. In: Sharon Inkelas & Draga Zec, The phonology-syntax connection, 127–143. Chicago: Chicago University Press. Kenstowicz, Michael 1994 Sonority-driven stress. [Unpublished Ms., MIT, Cambridge, Massachusetts]. Ladd, Robert 2009 Intonational Phonology. Cambridge: Cambridge University Press. Laver, John 1994 Principles of Phonetics. Cambridge: Cambridge University Press. Lehiste, Ilse 1970 Suprasegmentals. Cambridge, MA.: MIT Press. Liberman, Mark 1975 The Intonational System of English. Ph.D. dissertation, MIT, Cambridge, Massachusetts. [published in 1979 by Garland Press, New York]. Liberman, Mark & Alan Prince 1977 On stress and linguistic rhythm. Linguistic Inquiry 8: 249–336. McCawley, James 1968 The phonological component of a grammar of Japanese. The Hague: Mouton. Nespor, Marina 1990 Vowel deletion in Italian. The organization of the phonological component. The Linguistic Review 7: 375–398. Nespor, Marina and Vogel, Irene 1986 Prosodic phonology. Foris Publications, Dordrecht. Odden, David 1988 Predictable tone systems in Bantu. In: H. van der Hulst and N. Smith (eds.), Autosegment studies on pitch accent, 225–52. Dordrecht: Foris publications. Ode´, Cecilia and Vincent van Heuven (eds.) 1994 Phonetic studies of Indonesian prosody, (Semaian 9). University of Leiden: Vakgroep Talen en Culturen van Zuidoost-Azie¨ en Oceanie¨.

Word accent: Terms, typologies and theories

53

Peperkamp, Sharon 1997 Prosodic words. LOT/HIL dissertation #34. The Hague: Holland Academic Graphics. Pierrehumbert, Janet 1980 The phonology and phonetics of English intonation. [Unpublished Ph.D. dissertation, MIT, Cambridge, Massachusetts.] Prince, Alan 1976 Applying Stress. [Unpublished Ms., University of Massachusetts, Amherst]. Prince, Alan 1983 Relating to the grid. Linguistic Inquiry 14: 19–100. Roca, I. 1986 Secondary stress and metrical rhythm. Phonology Yearbook 3: 341–70. Selkirk, Elisabeth 1984 Phonology and Syntax. The Relation between Sound and Structure. MIT Press, Cambridge, Massachusetts. Selkirk, Elisabeth 1995 Sentence Prosody: Intonation, Stress, and Phrasing. In: John Goldsmith (ed.), The handbook of phonological theory, 550–569. Cambridge, MA and Oxford, UK: Blackwell. Sluijter, A. M. C. and van Heuven, V. J. 1996 Spectral balance as an acoustic correlate of linguistic stress. Journal of the Acoustical Society of America, Vol. 100 (4, Pt 1): 2471–2485. Visch, E. 1999 The rhythmic organization of compounds and phrases. In: H. van der Hulst (ed.), Word prosodic systems in the languages of Europe, 161–232. Berlin and New York: Mouton de Gruyter. Vogel, Irene 2009 The status of the clitic group. In: Janet Grijzenhout and Baris Kabak (eds.). Phonological domains: universals and deviations, 15–46. Berlin: Mouton de Gruyter. Zanten, Ellen van, & P. Dol This volume Word stress and pitch accent in Papuan languages. Zanten, Ellen van and Rob Goedemans 2007 A functional typology of Australian and Papuan stress systems. In: Vincent J. van Heuven & Ellen van Zanten (eds.), Prosody in Indonesian Languages (LOT Occasional series, 9) (pp. 63–88). Utrecht: LOT.

2.

An overview of word stress in Australian Aboriginal languages1

Rob Goedemans 1. Introduction In this chapter I survey the various types of stress systems that we may find in the Aboriginal languages of Australia. From whatever point of view we look at these metrical patterns, we observe that there is not much diversity. In an overwhelming majority of Aboriginal languages main stress appears somewhere at the beginning of the word. The prototypical stress pattern for these languages places main stress on the first syllable and secondary stress on alternate syllables thereafter, but we do find some variations on this theme. In the first part of this chapter I will discuss the prototypical pattern and its variants. The only other basic stress pattern is found in Arnhem Land, in the north of the Northern Territory. In this area, a group of languages is spoken that is genetically distinct from the other languages in Australia, which has given rise to the division into Pama-Nyungan and non-PamaNyungan languages (attributed to Ken Hale by Dixon 1980: 21, see also Hale 1964 and figure 1 below), pama and nyungan being the words for ‘man’ in the languages in the north-east and south-west of the PamaNyungan group, respectively. Some of the non-Pama-Nyungan languages spoken in Arnhem Land have a stress pattern that is the (trochaic) directional inverse of the most common pattern in the majority of PamaNyungan languages. In these non-Pama-Nyungan languages, main stress

1. I am greatly indebted to Bob Dixon, Luise Hercus, Grace Koch, Harold Koch and Geraldine Tri‰th for their help in gathering the information contained in this chapter. Furthermore, I thank the Australian National University, the Australian Institute for Aboriginal and Torres Strait Islander Studies and the Netherlands Organisation for Scientific Research (NWO) for giving me the opportunity to work in Australia, doing the research that led to this chapter. This research was also funded by NWO through the Foundation for Speech, Language and Logic under project #300-173-026, and the Royal Dutch Academy of Sciences (KNAW).

56

Rob Goedemans

is placed on the penultimate syllable, while secondary stress occurs on alternate syllables before the main stress. In the contact area between the initial stress languages and the ones that have penultimate stress, we find some interesting stress patterns that seem to be transitional between these two directional opposites. In the second part of this chapter I will present an areal overview of the stress patterns in Arnhem Land, revealing the full range of transitional patterns and the relevance of their geographical location with respect to the two basic patterns. This is where the lack of diversity in Australian stress patterns comes to our aid. It is precisely the simple and straightforward nature of the two basic patterns that allows us to get into the nitty gritty of the transitional ones. The theoretical framework in which the relevant patterns will be presented is that which is also used to encode stress patterns in the StressTyp database, and which was outlined in Goedemans, van der Hulst and Visch (1996). It will be shown that the derivation of the transitional, or ‘‘intermediate’’, patterns from the basic stress types proceeds more naturally in this framework than in other metrical theories, in which main stress and rhythm are not separate entities. In that respect, this chapter may serve as a reintroduction to the views on metrical principles that underly StressTyp, and the structure we use to express these views. In conclusion, I will shed some light on a question that immediately arises in relation to stress patterns that are labelled ‘‘intermediate’’ or ‘‘transitional’’: Are the stress patterns in the languages in question intermediate, but stable, between the two opposites of the scale, or were they recorded while still evolving from one end of the scale to the other, and, if so, in which direction? Some theoretical and di¤usional evidence will be brought to bear in an attempt to answer this question.

2. Left oriented stress Most Australian Aboriginal languages have a stress rule that favours the left word edge as the location of main stress. Of the 66 Australian languages in StressTyp, a staggering 54 (or 81%) employ a stress rule that is left oriented.2 Needless to say that this group is not homogeneous. There 2. A larger survey of stress rules in Australian languages, which will be incorporated in StressTyp in the future, can be found in Goedemans (1998). The percentage of left oriented stress languages is even bigger in that set. 92% of the 180 recorded cases place stress somewhere at the left word edge.

An overview of word stress in Australian Aboriginal languages

57

is more than one way to derive stress on one of the syllables near the left word edge, given the metrical machinery that languages have at their disposal. In Goedemans, van der Hulst and Visch (1996), we have introduced the set of rules we believe to constitute this metrical machinery. In this chapter, I will briefly recapitulate some of those rules, while exploring the diversity of Aboriginal stress patterns. By far the most straightforward way to derive stress at the left word edge in our, and in fact any other, metrical theory is to place a Quantity-Insensitive (QI) trochee at the left word edge and make sure that foot receives the main stress. Doing so, we arrive at a language in which all the words have main stress on the initial syllable. Mantjiltjara, a dialect of Martu Wangka from the Gibson desert in Western Australia (Marsh 1969), is such a language, as we can see in (1).3 (1) ¨yapu ‘stone’ ¨waæ pu˙kur&u ‘tall’

¨kakar&a ¨kapu˙kur&i˙åanta

‘east’ ‘I dreamed of you’

Nearly half of the Australian languages in StressTyp (29 out of 66) share the Mantjiltjara rule for main stress. Some of these also have secondary stresses, like Mantjiltjara, while in others, like Burarra (Glasgow 1981), such stresses seem absent, or information on secondary stress is simply lacking. Let us now renew our acquaintance with the parameters we use to encode stress patterns like that of Mantjiltjara in StressTyp. For theoretical and practical reasons we have decided to separate the parameters for main stress from those for secondary stress. In bounded languages (those in which main stress is restricted to either the first, or the last three syllables of the word) we assign main stress in a disyllabic domain that is placed at one of the word edges. In all 29 languages that have initial main stress, we must place that domain at the left word edge and place a trochaic foot inside it. We set the Stress Domain parameter to L(eft). Since a trochee is assigned irrespective of the internal composition of the syllables (it is QI) we set the Stress Weight parameter to N(o), and Stress if Both Light to tr(ochee). The latter parameter specifies the location of main stress in case both syllables in the domain are light. Strictly speaking we cannot say that in a QI language, but we still use the parameter here because the standard foot in QI languages is comparable to the default 3. Mantjiltjara has an entry in StressTyp. Not all languages discussed in this chapter have such an entry. Sometimes some other language from the survey in Goedemans (1998) is better suited to make a point, while in section 3, languages from outside StressTyp are mentioned to complete the areal picture. To avoid worsening the bias towards Australian languages in StressTyp, we refrain from entering all those languages at this point.

58

Rob Goedemans

foot we get in QS languages when there are no heavy syllables in the domain. For four languages the derivation stops there: Rhythm is set to N(o) and we leave it at that. The other 25 have secondary stress, which we derive through the assignment of QI trochees starting at the left word edge. Hence, for those languages, Rhythm is set to Y(es), Rhythm Direction is set to L(eft), Rhythm Weight to N(o) and Rhythm Type to tr(ochee). Since more than one secondary stress is assigned, Rhythm Iterative is set to Y(es). Syllables that cannot be parsed into binary feet, like the final syllable of kakar&a, remain unstressed. To be able to stress those syllables, we would have needed a foot spanning only one syllable; a so-called degenerate foot. To avoid the stressing of singleton syllables that are left over after rhythmic parsing into binary feet, the parameter governing the use of Degenerate Feet is set to N(o). In (2) the metrical analysis of the last word in (1) is given. For expositional purposes, some parameters that are irrelevant for Mantjiltjara are also represented. The meaning of these parameters will become clear during the following discussion. (2) | * | ¨kapu˙kur&i˙åanta ( * .)( * .)( * .) Stress Domain: Stress EM: EM Unit: Stress Weight: Stress if Both Heavy: Stress if Both Light: Rhythm:

main stress plane rhythm plane L – – N – tr Y

Rhythm Direction: L Rhythm EM: – EM Unit: – Rhythm Weight: N Rhythm Type: tr Rhythm Iterative: Y Degenerate Feet: N

In some of the Mantjiltjara-type languages stress is not assigned to words, but to morphemes. For the derivation of the stress patterns in such languages we need not change the parameters in (2). Application of the rule is repeated for each non-monosyllabic morpheme, and the first morpheme is then selected as the head at word level. Some words from ArabanaWangkangurru (Hercus 1994) exemplify this pattern. (3) ¨warpa ¨wanparda ¨yurku ˙rangku ¨katha rnda ˙nangku ˙liparna

‘storm’ ‘carry’ ‘the ancestral black snake’ ‘they used to travel continually and in a hurry long ago’

An overview of word stress in Australian Aboriginal languages

59

The alternative stress rules we found in Burarra and Arabana-Wangkangurru were both quite close to the basic Australian pattern found in Mantjiltjara. To derive them we only have to either leave out the secondary stress rules, or apply the rules several times. To find the first real variant of the basic left oriented stress rule we need to play a little with the parameters shown in (2). Suppose we were to introduce Extrametricality, by which a prosodic unit at one of the word edges becomes invisible to the metrical rules. We would have to set the Stress EM parameter, and since we are dealing with left oriented stress, we set it to L(eft). We also need to specify the unit that is extrametrical. Let us take the standard unit for EM: the syllable. We set these EM parameters for both stress and rhythm. The rest of the parameters remain the same. Thus, we have derived from the basic initial stress pattern, a pattern in which main stress always occurs on the second syllable, while secondary stresses appear on all other non-final even syllables.4 This pattern can be found among the Aboriginal languages. First of all, there is a group of four languages from South-Eastern Queensland (not represented in StressTyp) that are reported to have main stress on the second syllable. Information on these languages is very scanty though. For the first three, the main stress rule is briefly reported on in a survey by Holmer (1983), and for the fourth only old word lists are available on which stress is sometimes poorly indicated. All this indicates that we must be very careful in attributing any theoretical importance to this pattern. In (4) I present some examples from a representative of the first three (Gureng Gureng, Holmer 1983) that illustrate the pattern in question. (4) ba¨laN ‘creek, river’ gu¨naNal ‘frightened’ gi¨lamanmin ‘turned around’ Basically, languages of the Gureng Gureng type have the same parameter settings as the ‘‘stress the first postconsonantal vowel’’ languages that I discussed in chapter 4 of the first StressTyp book (Goedemans 1996: 129–163). In these languages, main stress is initial in words that begin with a consonant, while main stress appears on the second syllable of words that begin with a vowel. I refer to Goedemans (1998) for an overview of the possible analyses for this apparently curious stress pattern, and 4. See Goedemans (1998) for reasons why an iambic analysis of these secondsyllable stress languages, though intuitively most plausible, is not feasible.

60

Rob Goedemans

will su‰ce here by saying that the gist of the most plausible analyses is that initial onsetless syllables are ‘‘deficient’’ and have to be skipped somehow by the metrical rules. The common tool to place syllables outside the metrical domain is extrametricality. (Note that we need EM to be selective in this case: onsetless syllables are extrametrical, while those with an onset are not. This probably means that syllables need to be specified for EM in the lexicon. See Goedemans 1998 for a more elegant solution in Optimality Theory.) In (5) I present an example from Alyawarra (Yallop 1977) with its metrical structure. Apart from the secondary stresses, the structure is as it would be in Gureng Gureng. (5)

|* | ¨nada˙wara (* . )(* .)

place name

When we count Gureng Gureng and Alyawarra as one type, 10% of the Australian languages in StressTyp have this pattern. Confirmation for this percentage can be found in Goedemans (1998), who, in a di¤erent sample of 180 languages, finds 18 languages showing the pattern above. Variants of the Mantjiltjara pattern may, of course, just concern the settings of the rhythm parameters. Maranunggu (Daly River, Tryon 1970), and four other languages in StressTyp have main stress on the initial syllable and secondary stress on alternate syllables thereafter, even on the last syllable of words with an odd number of syllables. This means that the syllable that is left over after the word is parsed into binary feet is parsed into a degenerate foot. In these languages, the Degenerate Feet parameter is set to (Y)es. An example word from Maranunggu is parsed as such in (6). (6) | * | ¨langka˙rate˙ti (* .)(* .)(*)

‘prawn’

This discussion of secondary stress on final syllables leads us to cases like Gaanay, a language from Victoria, for which Hercus (1986) gives the following rule for secondary stress assignment: a secondary stress occurs on the final syllable if it is closed. In (7) I present two pairs of words that illustrate the crucial di¤erence. (7) ¨NaraNda ‘to bury’ ¨naNera ‘saltwater mussel’

¨bindju˙laN ‘cat’ ¨jala˙man ‘several’

An overview of word stress in Australian Aboriginal languages

61

Quantity-sensitive (QS) stress assignment occurs in many languages world wide. QS languages look at syllable weight to determine the positions of main and secondary stress. The weight of a syllable is usually determined by the length of the vowel and/or the presence of a coda consonant. In Gaanay, only the assignment of secondary stress is QS. Hence, the only modification we need to make to the Mantjiltjara parameters in (2) to derive the Gaanay pattern is to set the Rhythm Weight parameter to Y(es). This automatically means that the rhythmic trochees assigned are of the QS type; they are allowed to span only one syllable, as long as that syllable is heavy (the foot contains two moras). This explains the di¤erence between the left-hand words and the right-hand words in (7). The left-over syllables in the former are light and degenerate feet (spanning only one mora) would be needed to parse them. Such feet are prohibited in Gaanay. The final syllables in the words on the right in (7) are heavy and can be parsed into normal QS trochees. It is clear that the Stress Weight parameter must remain as it is. Should we respect quantity when assigning main stress, we would wrongly derive main stress on the second syllable of NaraNda, since it is heavy while the first syllable is light. Obviously, main stress assignment must be QI, since Gaanay main stress is always on the initial syllable. Gaanay is not unique among the world’s languages. We find many instances in which main stress is assigned in a QS fashion, while secondary stress is QI, or, albeit less frequently, the other way around. Such di¤erences form strong support for separation of the algorithms used for the assignment of main and secondary stress, as we do in StressTyp. I will come back to this issue in section 3 where I will present more evidence to support this approach. 2.1. Variability on the left With the previous discussion on Gaanay, we have entered quantitysensitivity into the equation. We have seen that QS parsing may lead to variable stress patterns. Languages that make use of the QS option for main stress typically allow it to occur on either one of the two syllables in the main stress domain. The choice is governed by syllable weight. In StressTyp, we find fourteen Australian languages with left-oriented QS main stress. Before we discuss these, let us look at a language in which we find variable stress that is not governed by syllable weight. Kitja, a language from the East Kimberleys in Western Australia (Taylor and Taylor 1971), represents a group of languages that have main stress either on the first or the second syllable, but for which there seems to

62

Rob Goedemans

be no rule with which we might be able to predict its location; it is truly random. In (8) I present some examples to illustrate the point. (8) ¨kukan ‘nothing’ ¨palawam ‘ground’ ¨tjapi˙yapim ‘top branch’

pi¨ni ‘you do’ na¨nina ‘you went’ ku¨lawala ‘flower’

Unless there is some unknown morphological rule at work here, we have no other option but to mark the stressed syllables in the lexicon and make the stress rule sensitive to these lexical markings. A di¤erent picture emerges when we regard languages like Yindjibarndi, spoken on the Northwestern coast of Western Australia (Wordick 1982). In this language, main stress is reported to fall on the second syllable if it contains a long vowel and the first does not, otherwise the initial syllable is stressed. The pattern is illustrated in (9) (unfortunately there are no examples in which both the first and the second syllable contain a long vowel). (9) ¨wilar˙ra ‘moon’ ¨kaayu.˙urru ‘southward’ nyi¨laarti ‘native mead’ To derive this pattern, we must switch the Stress Weight parameter in (2) to Y(es). That means we also have to enter a value for the Stress if Both Heavy parameter, which specifies the location of main stress in case both syllables in the domain are heavy. Since we have no examples with two heavy syllables in the main stress domain, we rely on Wordick’s statement and set it to L(eft), indicating that the first of these heavy syllables will be stressed. In case both syllables are light, we assign a trochee (as in Mantjiltjara). The cases in which only one syllable in the main stress domain is heavy do not need coding. The heavy syllable will automatically receive stress (see Goedemans, van der Hulst and Visch 1996). A variant of this pattern can be found in the south Australian language Yukulta (Keen 1983). Main stress in Yukulta falls on the first long vowel or, if there are no heavy syllables, on the first syllable. Secondary stress is on the penult if possible. Examples are given in (10). (10) ¨Nita ‘fire, firewood’ ¨makuwa ‘woman’ ¨kulu˙runa ‘bushfire’

¨Nuørpa˙lut5a ‘to finish’ pi¨t5iønta ‘boy toddler’ puNkalan‘¨t‘iøtya ‘to kneel’

An overview of word stress in Australian Aboriginal languages

63

It is not di‰cult to change the parameter settings for Yindjibarndi in such a way that they yield the Yukulta pattern. In this language, the domain for main stress assignment is the whole word. Hence, we set the Stress Domain parameter to U(nbounded). Primary stress is assigned to the first heavy syllable in that domain, so Stress if Both Heavy remains L(eft). If no long vowel is present in the word, stress is initial. Therefore, Stress if No Heavy remains tr(ochaic). In cases where it would not appear adjacent to the main stress, secondary stress always occurs on the penultimate syllable. If we can do it without causing a clash with the main stress, we build a single trochee at the right side of the word (for parameter changes see (11)). Hence, main stress is QS and unbounded while secondary stress is QI and bounded. Like Gaanay, Yukulta supports our separation of the main stress and rhythm parameters. In (11) the parameter settings for Yukulta are given, together with analyses of some of the words in (10). In the second example, the secondary stress that could be assigned would occur next to the main stress; it clashes and is deleted. In the third and fourth example the secondary stress rule is void because secondary stress would end up in the same location as the main stress. (11)

| * | ¨kulu˙runa (* .)

| * | ¨makuwa clash

| * | pi¨t5iønta void

| * | puNkalan‘¨t‘iøtya void

Stress Domain: Stress EM: EM Unit: Stress Weight: Stress if Both Heavy: Stress if Both Light: Rhythm:

U – – Y L tr Y

Rhythm Direction: Rhythm EM: EM Unit: Rhythm Weight: Rhythm Type: Rhythm Iterative: Degenerate Feet:

R – – N tr N N

The Yukulta stress pattern concludes this overview of the left oriented stress rules in the Australian Aboriginal languages. Not all patterns that occur have been discussed. Some of the stress rules we encounter are true exceptions, like that of Guugu Ya’u (Thompson 1976) in which the last heavy syllable is stressed, while the first syllable receives stress in case there are no heavies. For reasons of space, a discussion of these exceptions cannot be included here, but they can be found in an annotated list in Goedemans (1998). Other variants of the patterns discussed above appear to be intermediate stages between left and right oriented stress. These will be discussed in the next section, in which I will show what happens when the predominantly initial stress patterns introduced above come in contact with another basic stress pattern that can be found on the Australian continent; that in which main stress is penultimate.

64

Rob Goedemans

3. Right oriented stress and transitional patterns in Arnhem Land What most of the languages discussed in section 2 have in common is that their default location for main stress is the initial syllable. Such uniformity in default patterns results from rigorous application of the unmarked foot type in stress assignment: the syllabic (or QI) trochee. However, the assignment of syllabic trochees is not necessarily left oriented. In many languages, quantity-insensitive trochees are assigned starting at the right side of the word. Thus, a pattern emerges in which main stress occurs on the penultimate syllable and secondary stress on all alternate syllables to its left. This pattern is the directional opposite of the Mantjiltjara pattern, only deviating from it in the location of the main stress domain and the direction of rhythmic footing. On the Australian continent, this penultimate stress pattern can be found among the non-Pama-Nyungan languages in Arnhem Land. As an example, let us look at stress in Tiwi, a language spoken on two islands o¤ the north coast of Arnhem Land (Lee 1987): (12) ¨poja ‘seed’ tu¨wara ‘tail’

˙kalam¨para ‘sweat’

We observe that primary stress is on the penultimate syllable. Secondary stress appears on the first syllable in four-syllable words.5 As was noted above, the parameter settings for Tiwi do not di¤er greatly from those for Mantjiltjara. I give them in (13), together with the structure for kalampara. (13)

|* | ˙kalam¨para ( * .) (* .) Stress Domain: Stress EM: EM Unit: Stress Weight: Stress if Both Heavy: Stress if Both Light: Rhythm:

R – – N – tr Y

Rhythm Direction: R Rhythm EM: – EM Unit: – Rhythm Weight: N Rhythm Type: tr Rhythm Iterative: Y Degenerate Feet: N

5. Unfortunately we do not have words long enough to determine whether more than one secondary stress is possible, but for the sake of the argument we assume that this is the case.

An overview of word stress in Australian Aboriginal languages

65

Several other languages from Arnhem Land share the Tiwi stress rule to a greater or lesser extent. Those that also have main stress on the penultimate syllable as their predominant stress pattern are Alawa (Sharpe 1972), (Gunbarlang reportedly, no grammar or examples), Jingulu (Chadwick 1975, Pensalifini 2003), Limilngan (Harvey 2002), Mangarayi (Merlan 1989), Maung (Capell and Hinch 1970), Ngalakan (Merlan 1983), Ngalkbun (Sandefur and Jenhan 1977), Ngarndji (Chadwick 1971), Nunggubuyu (Hore 1981), Umbugarla (Davies 1989), Wardaman (Merlan 1994) and Yanyuwa (Kirton 1977), altogether about 7% of the Australian languages in Goedemans (1998). In figure 1 we can see that these penultimate stress languages drive a wedge between the initial stress languages in Arnhem Land. Given the relatively high level of di¤usional activity that is charactersitic of this area, mainly as the result of a high rate of linguistic intermarriage (Heath 1978; Dixon 1980), we might expect that the situation depicted in figure 1 is unstable. The languages that have directionally opposite stress patterns are bound to influence one another in this respect. If the stress patterns of these languages change under the influence of neighbouring patterns, this change will most likely be gradual. No language will simply flip-flop from initial stress to penultimate stress or vice versa. It is likely, therefore, that the wedge area in figure 1 is riddled with stress patterns that are neither truly trochaic and left-to-right, like Mantjiltjara, nor truly trochaic and right-to-left, like Tiwi. Such contact areas are of huge theoretical importance. From the variety of intermediate patterns that occur, we can infer certain properties of our metrical theory. Any theory that must go through great pains to analyse one of these intermediate patterns must be flawed. We must assume that the patterns in question are logical steps from A to B that are easily made by the speakers. Our theory should reflect that. While delving into the nitty gritty of the Arnhem Land stress patterns I will use the possible settings of the StressTyp parameters as a guideline. Let us, for the sake of the argument, start out with the pattern of Waray (number 24 in figure 1), which is identical to that of Mantjiltjara discussed in section 2. The relevant parameter settings are given in (14). (14) Stress Domain: L Stress if Both Light: tr Rhythm: Y

Rhythm Direction: Rhythm Type: Rhythm Iterative:

L tr Y

If such a language comes under the influence of a neighbouring language with penultimate main stress, it might alter the parameter settings in (14) to reflect the preference for right oriented stress in its neighbour. In a

66

Rob Goedemans

Figure 1. Arnhem Land Languages by stress type (details ignored).6 Compiled with data from, among others, Wurm and Hattori (1981) and Dixon (1980). Cue to numbers: Alawa 40 Anindilyakwa 31 Burarra 8 Dhuwal 11 Djamindjung 39 Djapu 12 Djinang 9 Djinba 10 Gaagudju 19 Garawa 42 Gunbarlang 5

Gungarakany 23 Kunwinjku 14 Gupapuyngu 7 Gurgoni 17 Gurindji 50 Iwaidja 2 Jingulu 45 Larikiya 22 Limilngan 51 Malak-malak 25 Mangarayi 32

Maranunggu 26 Margu 3 Marithiel 37 Maung 4 Miriwung 51 Mudbura 48 Murrinh-patha 36 Nakara 6 Ndje´bbana 16 Ngalakan 29 Ngalkbun 28

Ngankikurrungkurr 38 Ngarinman 49 Ngarndji 46 Nungali 35 Nunggubuyu 30 Nyanga 44 Patjamalh 27 Rembarrnga 15 Ritharngu 13 Tiwi 1

Umbugarla 20 Urningang 18 Waanji 43 Wageman 34 Waray 24 Wardaman 33 Warlmanpa 47 Wulna 21 Yanyuwa 41

6. In languages like Larikiya, Limilngan and Anindilyakwa, for instance, we find traces of quantity-sensitive stress at the left side of the word. This property has no bearing on the issue at hand and we will ignore it here. We also abstract away from competing patterns. Most of the languages mentioned here have scores of exceptions with, sometimes even regularised, antepenultimate or initial stress. For the moment, we will focus only on the broad division between right-oriented (predominantly penultimate) and left-oriented (predominantly initial) stress systems.

An overview of word stress in Australian Aboriginal languages

67

marginal e¤ort for change, it might choose to make sure a secondary stress always occurs on the penultimate syllable. Hence, Rhythm Direction is set to R(ight). A trochee is still built at the left edge, so main stress remains initial, but secondary stresses are assigned to the penult and every alternate syllable to its left, excepting the syllable adjacent to the main stress. This is exactly the pattern we find in Nakara (Eather 1990). In the word diøarabaga in (15) we observe that the avoidance of secondary stresses adjacent to the main stress may yield words in which two unstressed syllabes occur between the initial main stress and the first secondary stress. We call this the initial dactyl e¤ect. The final example shows that polysyllabic prefixes receive their own secondary stress. (15) ¨budøa ‘yabby (fish)’ ¨diøara˙baga ‘he emerges’ ¨gabøula ‘old man’ ˙gabarna-¨marbøa˙wabøa˙Neba ‘my (two) wives’ In StressTyp, we can account for the initial dactyl e¤ect in two ways. We can either disallow the building of secondary stress feet on syllables that are already covered by the main stress domain or we can build rhythmic feet exhaustively, and delete any secondary stress that clashes with the main stress. I choose the first option, since it is theoretically the least demanding one.7 In (16) I repeat the parameters from (14), adapted for Nakara, and provide the metrical structure of diøarabaga. (16) | * | Stress Domain: L ¨diøara˙baga Stress if Both Light: tr (* .) Rhythm: Y

Rhythm Direction: R Rhythm Type: tr Rhythm Iterative: Y

Nakara is spoken in the vicinity of some penultimate stress languages in the north of Arnhem Land, such as Gunbarlang and Maung. In the southeast, near Yanyuwa, Jingulu and Ngarndji, a similar picture emerges in a well-known case from the metrical literature: Garawa (Furby 1974). Apart 7. In most languages the decision whether to build rhythmic feet throughout the entire word including the main stress domain or not is irrelevant to begin with. In the vast majority of cases, a secondary stress would appear where the main stress is already located, so that it remains invisible (see examples 2, 5, 6, and 13). We could opt to refrain from application of rhythm rules in the main stress domain in all cases, were it not for languages like Mathimathi and Aklan, in which we find secondary stresses in the main stress domain that are not removed in clash. In the future we might express the rarity of such cases by adopting the rule that skipping the main stress domain in the assigment of rhythm is the default case in StressTyp.

68

Rob Goedemans

from the stress occurring on polysyllabic prefixes, the stress pattern of Garawa is identical to that of Nakara. In traditional metrical theories that build secondary stress feet from one word edge to the other, and then assign main stress to the head of either the leftmost or the rightmost foot by an End Rule, the Garawa case is problematic. When trying to tackle Garawa-type languages in such a framework, one must take a step similar to the one McCarthy and Prince (1994) and Hayes (1995) take in their analyses of Garawa: ‘‘Form an initial syllabic trochee; syllabic trochees from right to left; end rule left’’ (Hayes 1995: 203). With this move they introduce two directions of footing, and e¤ectively split the algorithms for the assignment of main and secondary stress, though such a split is not an explicit feature of their theory. In a sense, they move closer to the StressTyp way of looking at things in this analysis of Garawa, and other languages like it. That move is only half-hearted, however, and in the majority of the analyses, primary and secondary stress remain entangled. Waray-type stress, for instance, is derived by Hayes through one set of rules that builds trochees from left to right and promotes the leftmost foot-head to main stress. Thus, in Hayes’ framework, the step from Waray to Garawa involves a huge theoretical leap, while in StressTyp, where primary stress and rhythm have already been separated for independent reasons (see van der Hulst 1996a, and the discussions on Gaanay and Yukulta above), we go from one to the other by switching one single parameter. We do not have to look far to find another language that only slightly deviates from the basic left oriented rules for primary and secondary stress. Waanji, a dialect of Garawa that is spoken on the western fringe of the Garawa territory, closest to Yanyuwa and Jingulu, exhibits a pattern that is only minimally di¤erent from that found in Garawa itself. Osbourne (1966) claims that main stress is initial and that a single secondary stress occurs on the penultimate syllable: (17) ¨gulu ‘axe’ ¨wulu˙lugu ‘old man’

¨jarala ¨wabinbara˙ulu

‘root’ ‘for turtles’

The analysis of Waanji proceeds along the same lines as that of Nakara and Garawa. The only di¤erence is the setting of the Rhythm Iterative parameter to N(o). It is as if Waanji is influenced enough to enforce a secondary stress on the penult, but is not yet right oriented enough to parse from right to left. The next step in our move towards right oriented stress could involve a shift of the main stress domain to the right word edge. That would mean the penultimate secondary stress is promoted to main stress, and the initial

An overview of word stress in Australian Aboriginal languages

69

main stress is degraded to secondary stress. However, there are two intermediate stages we can envisage that may occur before a shift of the main stress domain to the right word edge. Suppose we were to leave the domain at the left word edge, but that we would refrain from assigning a default trochee to that domain. Should we then assign right-to-left trochees to the entire word, prohibiting degenerate feet, we would arrive at a language in which an initial stress occurs in words that have an even number of syllables (the whole word is parsed into binary trochees) and stress is on the second syllable in words that have an odd number of syllables (the initial syllable is left over, and the head of the leftmost trochee is on the second syllable). Since the main stress domain covers the first two syllables, we could assume that the first or the second syllable automatically receives main stress, depending on which one ends up under the head of a rhythmic trochee. Such systems were extensively discussed in Goedemans, van der Hulst and Visch (1996), and were labeled ‘‘count systems’’ there. Two such systems occur in Arnhem Land: Malak-malak (Birk 1976) and Ngankikurrungkurr (Hoddinott and Kofod 1988). In (18) I present some Ngankikurrungkurr examples with their metrical structure. (18) | * | ¨efe˙kimi ‘rabbit’ (* .)(* .) Stress Domain: L Stress if Both Light: – Rhythm: Y

| *| a¨nimpirr˙mire (* .) (* .)

‘firefly’

Rhythm Direction: R Rhythm Type: tr Rhythm Iterative: Y

Previously we analysed these languages as unbounded systems in which the assignment of main stress was sensitive to secondary stresses. The first of these stresses would receive main stress. In the light of the evidence presented here I would like to propose that count systems are bounded after all, an option we also considered in the past (see Visch 1996, footnote 8).8 Much more can be said about these count systems, but I will not enter that discussion here. The only important thing to note with respect to the present discussion is that count systems are truly intermediate. No definitive choice for either initial or penultimate stress is made. Instead, stress is on one of the syllables on the left edge depending on right-to-left rhythmic parsing. 8. It could also be the case that both types exist, and that the unbounded versions form intermediate stages between unbounded and bounded stress types, while the bounded ones are intermediate stages between left and right oriented bounded systems.

70

Rob Goedemans

It was noted above that we could envisage two types of stress systems in which main stress cannot yet be said to be penultimate, but in which it is not initial either. A logical way in which languages can fulfill these requirements is to have no main stress at all. Languages in which di¤erences between main and secondary stress are reported to be lacking do occur in the border areas between the patterns in figure 1. In Wardaman (Merlan 1994) on the edge of the penultimate area, close to Ngankikurrunungkurr and the initial stress languages Djamindjung and Marithiel, stress is reported to occur on the penult and alternate syllables to the left. The analysis of this language di¤ers only in one respect from that of Ngangkikurrungkurr: the main stress domain shifts to the right (or is absent, depending on the exact analysis, see below). The same holds true for Yanyuwa (Kirton 1977), located between Alawa and Garawa. In this language, however, stresses are always assigned to the penultimate and the initial syllable in words of four or five syllables (no information on stresses in longer words is available, so we cannot determine the direction of rhythmic footing). Some Yanyuwa and Wardaman examples can be found in (19). (19) Yanyuwa ni˙wanji ‘animal flesh’ ˙alanj˙djilu ‘to the camp’ ˙mar&uwa˙r&ala ‘with the cousin’

Wardaman ja˙warrga ‘liver’ ˙gala˙warra ‘plumed pigeon’ ˙gura˙mana ‘gums’

Note that the stresses in all the examples in (19) are all equal; the sources report that there is no di¤erence between primary and secondary stress in both Wardaman and Yanyuwa. In StressTyp, however, the main stress parameters for these languages are being used anyway. For a language like Yanyuwa, which is much like Waanji as far as stress locations are concerned, we need two sets of parameters to derive stress at both edges, and we need to specify that primary and secondary stress are equally prominent. Though it seems arbitrary, in Yanyuwa-like cases, which edge is covered by the primary, and which by the secondary stress parameters, there are often good reasons to choose one or the other. In some cases, the location of the one stress in three-syllable words indicates whether initial or penultimate stress is more important (primary) in the language. If stress patterns in three syllable words are too irregular (which is often the case), we need to resort to other diagnostics. Had there been longer Yanyuwa examples, for instance, we could have decided at which edge the ‘main’ stress was located on the basis of the location of a third stress. The ‘main’

An overview of word stress in Australian Aboriginal languages

71

stress in such cases is the one singleton foot-head at one of the edges, while the ‘secondary’ stresses are assigned iteratively, starting at the other edge. Other, more language specific, diagnostics are likely to exist. Often it is made quite clear in the grammar which of the stresses is the most ‘basic’ one, and hence, can be considered primary. For Wardaman we suspect that the penult fulfils this role but we cannot be sure. In an alternative analysis the primary stress parameters in (18) can be left unspecified. In principle, the assignment of right-to-left rhythmic trochees is enough to derive the pattern. With Wardaman and Yanyuwa, we have arrived at the border between left and right oriented stress systems – in fact, we have even crossed it slightly, since both languages should be characterised as right oriented rather than left oriented. One might suggest, then, that we have skipped a logical possibility. Indeed, just on the other side of the border, we would expect to find languages that assign rhythmic trochees from left to right, and do not di¤erentiate between main and secondary stress. Again the di¤usional melting pot of Arnhem Land contains an example that shows the predicted pattern. West of Ngankikurrungkurr we find Murrinh-patha (Walsh 1976), which stresses the initial syllables and all alternates thereafter, even the final. Note, then, that this language needs degenerate feet. There is no di¤erence between primary stress and rhythm, as can be seen in (20). (20) ˙we¤e ‘dog’ ˙nigu˙nu ‘she’ ˙kana˙NandaN ‘emu’ The fact that Murrinh-patha and Wardaman exist as opposites in this area suggests to us that we might also find opposites to the right-to-left count systems there. In these opposites main stress would occur on the right side of the word, the exact location being determined by the number of syllables in the word. Hence, the main stress domain is at the right word edge, while parsing proceeds from left to right. Such languages exist, but they do not occur in Australia.9 This conspicuous absence of left-to-right count systems is significant. I will come back to it in section 3.2. 9. Harvey (2002) postulates that this exact pattern occurred in the history of Gaagudju, but this is highly speculative. There is no clear evidence that Gaagudju was ever a count system, since Harvey now describes Gaagudju as having synchronically unpredictable penultimate or antepenultimate stress governed by alignment to tonal phenomena in its more recent history.

72

Rob Goedemans

What we do find in Arnhem Land are some directional opposites of Nakara and Garawa. The most striking example is Nunggubuyu (Hore 1981), a language spoken on the east coast, close to the strictly left oriented Yolngu languages. Look at the examples in (21). (21) wu¨rugu ‘billabong’ ˙ngalaa¨ligi ‘turtle’

˙ambala¨lari ˙rawu˙rrumugu¨rrumu

‘poor’ ‘plant species’

Main stress is located on the penultimate syllable, and secondary stress comes in from the left. The directional parameters for Nunggubuyu are obviously the exact opposite of those for Nakara in (16). In (22) they are presented together with the structure of rawurrumugurrumu. (22)

| * | ˙rawu˙rrumugu¨rrumu (* .)( * .) Stress Domain: R Stress if Both Light: tr Rhythm: Y

Rhythm Direction: L Rhythm Type: tr Rhythm Iterative: Y

The non-iterative rhythm variant of this pattern, that was embodied in Waanji in the other camp, can also be found among the penultimate stress languages. Davies (1989: 29) describes the stress pattern of Umbugarla, spoken on the northwestern fringe of the penultimate stress area, as: ‘‘The basic pattern in Umbugarla is for primary stress to be located on the penultimate syllable of the word. Secondary stress is located on the first syllable.’’10 As in Waanji, we only have to set the Rhythm Iterative parameter in (22) to N(o) to derive this pattern. A fact that Davies also mentions with respect to Umbugarla is that there are many exceptions that have initial main stress. In an area that is as di¤usionally active as Arnhem Land, such exceptions can be highly instrumental in our search to answer the question in which direction the stress patterns are evolving, especially if all the exceptions in a particular language have one single stress type. In the next section, therefore, I complete the overview of stress types in Arnhem Land with a brief look at the languages for which competing stress patterns have been reported in the form of regular exceptions.

10. To make absolutely sure there is only one secondary stress, one should check whether the third syllable of six syllable words is stressed or not. Unfortunately, I did not find su‰ciently long examples in Davies (1989).

An overview of word stress in Australian Aboriginal languages

73

3.1. Exceptional patterns A tendency we often find in languages that have initial or penultimate main stress concerns three syllable words. There are only two ways in which we can place a trochee over a three syllable word, and these seem to be in competition quite often. In these cases no grammatical choice has been made between an initial trochee that leads to stress on the first syllable, or a final trochee that leads to stress on the second (or penultimate) syllable. Trisyllabic words have unpredictable stress, which is nonetheless consistent for a given lexical item, while in larger words stress is often invariably initial or penultimate. Two Arnhem Land languages that behave in this way are Alawa and Wulna, as we can see in (23). The languages are opposites in larger words, where Alawa has penultimate stress while Wulna stresses the initial syllable. (23) Alawa a¨lawal ‘properly’ ¨par&akal‘ ‘spear’

Wulna Na¨mada ‘shoulder’ ¨miy˙gil ‘shin’

At first sight it may be unclear whether the inability of these languages to choose between initial or penultimate stress in three syllable words points in the direction of a possible intermediate position they might occupy between strictly initial and strictly penultimate languages, or to a more theoretical principle that governs the behaviour of QI stress in three syllable words in general. However, I would suggest that at least the Arnhem Land cases are the result of di¤usion, and thus represent intermediate stages between the two patterns mentioned above. Support for this suggestion can be found in neighbouring languages that show the same variability between initial and penultimate stress in words larger than three syllables. In those cases, a theoretical principle cannot be at play; there is no obvious theory internal reason for variability between initial and penultimate main stress in, for instance, four syllable words. Before we discuss those languages, let us look at an example of a language that resolves the issue of variability in three syllable words in a more drastic way than Alawa and Wulna. Consider the examples from Maung in (24). (24) ¨ba¨ladji ‘bag’

¨ma¨miNa ‘clam shell’

The claim made by Capell and Hinch (1970) is that words of up to three syllables in length have equal stress on the first and second syllable. In longer words, primary stress is penultimate, or antepenultimate if that syllable is closed, while secondary stresses occur on alternate syllables

74

Rob Goedemans

before the main stress. The equality of the two stresses in the words in (24) and the fact that disyllabic words reportedly also have such stresses could spawn lengthy theoretical debates. This chapter, however, is not intended to discuss such theoretical intricacies in great detail. What we must remember from Maung is that the exceptional behaviour of the three syllable words may serve as an argument to place the language in an intermediate position between left and right oriented stress, albeit closer to the latter than to the former. Two languages in which exceptions occur in words of more than three syllables are Ngalakan and Mangarayi. Merlan (1983, 1989) describes these languages as having stress patterns that are predominantly penultimate. However, exceptions in which primary stress is initial, and sometimes antepenultimate, are easy to find, as we can see in (25). (25) Ngalakan ¨baragal/ bur¨kaji ¨dakba˙rara ˙mili¨balkin) gan‘¨d‘alpuru

Mangarayi ‘spear’ ¨gad‘ugu ‘genuine, real’ gu¨rajn)in ‘green tree frog’ ¨wuru˙mumu ‘salt water’ ˙warin¨jalan ‘female plain kangaroo’ yi¨riniwun

‘woman’ ‘blood’ ‘hornet’ ‘Exocarpus Latifolius’ ‘hard tree wax’

We must investigate the languages closer to find out whether a generalisation concerning the choice for penultimate or antepenultimate stress can be made. In Mangarayi, for instance, Merlan notes a tendency towards quantity sensitivity by which stress sometimes moves to the antepenult if the penult is open. The cases with initial stress, however, clearly belong to a set of regular exceptions. I do not yet know whether these exceptions reflect an old pattern that has given way under pressure of the penultimate stress languages to the south of Ngalakan, or whether they signify the influence of the initial stress languages to the north. The discussion in section 3.2 may shed more light on this matter. Another language that may fall in this category is Rembarrnga (McKay 1975), spoken to the north of Ngalakan. Most words reportedly have initial stress, but many exceptions with penultimate stress occur. Noteworthy with respect to the Rembarrnga pattern is the fact that McKay claims there is no di¤erence between primary and secondary stress. Perhaps the most intriguing case belonging to the set of regularexception languages is Ngalkbun. There are two descriptive sources for this language, Capell (1962), cited in Hayes (1995), and Sandefur and Jenhan (1977). Capell calls this language Dalabon (which, according to

An overview of word stress in Australian Aboriginal languages

75

Grimes and Grimes 1996 is just an alternative name for Ngalkbun) and states that primary stress is initial, while secondary stress is located on every odd syllable after the main stress, excepting the final syllable. Sandefur and Jenhan, on the other hand, claim that, in words of three syllables stress is on either the initial or the penult, and in words of more than three syllables primary stress is penult and one secondary stress occurs on the initial syllable. Some of the examples they give can be found in (26). Note that there is even variation within words. (26) na¨komtutj ~ ¨nakomtutj Na/¨tjor‘mu ˙tar‘par&a¨pa/No

‘little boy’ ‘I cough’ ‘afternoon’

It would be extremely far-fetched to claim that the two sources, which describe the language with a 15-year interval, have captured the evolution of its stress pattern. I would expect a 15-year period to be too short for such a change. In my view, the discrepancy between the two sources neatly reflects the relative chaos the stress patterns of contact languages like Ngalkbun are in. It is not at all strange to assume Ngalkbun is in transition, but I rather believe that Capell and Sandefur and Jenhan have captured the essence of transition, i.e. variable (or variably interpretable) stress patterns, than that they have recorded two consecutive stages in the metrical evolution of the language. A similar case is Ndje´banna. Both sources I consulted were written by McKay, but with a 25-year interval between the two (1975, 2000). It is quite di‰cult to infer a clear stress pattern from McKay’s statements and examples (a tell tale sign in itself ), but still the older source pointed in the direction of left rather than right orie¨nted stress, whereas the newer source suggests a right-orie¨nted pattern. In this section, and the previous one, it has become clear that the stress patterns in Arnhem Land languages are in turmoil. In the next section, I will investigate whether we are dealing with random chaos or whether the patterns we have come across indicate coherent di¤usional movement towards left- or right-edge oriented stress rules. 3.2. Stress and the direction of di¤usion With respect to the status of the stress patterns in Arnhem Land it is worth noting that none of the languages that we categorised in the penultimate stress group have a stress rule that strictly assigns primary stress to the penult and secondary stress to alternate syllables to the left. Some languages show variants of this rule, others simply have a host of exceptions.

76

Rob Goedemans

This fact, combined with the observation that many of the patterns we do find can be viewed as intermediate stages between left and right oriented stress, conveys to us a certain sense of development. Many of the patterns we have seen seem to be unstable. From a theoretical point of view, double edged systems like Garawa, and count systems like Malak-malak, may be too complex to last. The latter have previously been identified as intermediate between more stable stress systems (see van der Hulst 1996b, pc). Other languages simply have stress rules that are virtually indescribable because the boundary between predominant pattern and regular exceptions has disappeared, or because distinctions between primary and secondary stress are unclear. Given the fact that di¤usion of linguistic patterning is extremely commonplace in the Arnhem Land area (see Heath 1978), there is no doubt in my mind that we have captured some of the languages described above in their evolution from left to right oriented stress systems, or vice versa. The alternative would be that the stress rules in the languages are stable derivatives from either the initial or the penultimate basic pattern, and that their possible placement along a theoretically motivated continuum is a sheer coincidence. Surely, exceptions can occur without immediately implying that the language occupies an intermediate position between two stress types. However, I submit that all those forms, together covering almost the full range of possible intermediate systems, all occuring in a small contiguous area, is too much of a coincidence. This must have a deeper meaning. In the following discussion, therefore, I will assume that the stress systems of the Arnhem Land languages are subject to di¤usional forces, and that these forces do not flip over a certain language from one basic pattern to the other, but that it goes through one or more intermediate stages instead. The assumption above triggers a logical follow-up question. If the stress systems of the languages in Arnhem Land are really on the move, then in which direction are they going? Are the languages with initial stress simply mopping up an old penultimate pattern, or is the penultimate pattern gradually gaining headway among the surrounding languages with initial stress patterns. This question is extremely di‰cult to answer. As in most language contact situations, it is probably neither exclusively the one nor strictly the other. A pattern might gain ground in one area while losing it elsewhere. However, one of the two patterns might be dominant even in this situation. If that is the case, it would still be di‰cult to determine which one that is. We cannot decide in favour of one or the other, but we can speculate a bit, based on what little evidence we have.

An overview of word stress in Australian Aboriginal languages

77

First of all, it must be noted that the white area in the northeast corner of Arnhem Land in figure 1 is probably not indicative of ‘‘a remnant pocket’’ of initial stress languages that are gradually being overtaken by the penultimate stress rule. It has been proven (see Dixon 1980), that the Yolngu tribes in this area have only recently moved to their present location. Their languages (no’s 7, 9, 10, 11, 12 and 13 in figure 1) are related to Western Desert, rather than to any non-Pama-Nyungan language. However, Dixon notes that after the migration, there has been considerable di¤usional pressure in the direction of the Yolngu languages (Dixon 1980: 257). In this light it seems unlikely that the intermediate patterns in languages like Rembarrnga and Nakara have developed from penultimate patterns under pressure from the Yolngu languages. Though we cannot be sure, the other option, in which they evolved from the basic initial stress pattern to their present state under pressure from the penultimate stress languages to the east seems more likely. In the southeast, we find language internal evidence for the same direction of di¤usion. In her description of Yanyuwa stress, Kirton (1977) notes that the pattern is as exemplified in (19), but that in longer words a tendency is developing to replace the secondary stress on the initial syllable of a word internal morpheme by a stress on the pre-antepenultimate syllable. Compare the alternatives in (27). (27) ˙gumba˙ramanda˙ninjdja ˙gumbara˙manda˙ninjdja

‘he was hitting himself’

The top pattern is the traditional one, the bottom word represents the tendency. Clearly, this tendency reflects a movement to a more rigid right-to-left way of assigning secondary stresses. The initial secondary stress that is still there is the last remnant of the left oriented pattern. So far, this is the only language internal piece of evidence I have in favour of movement towards penultimate stress. In the future, a thorough search through other language material from Arnhem Land will hopefully reveal more evidence. The final piece of evidence I will present is theoretical in nature. In section 3, it was noted that no left-to-right count systems were found in Arnhem Land. This absence was deemed to be significant. Suppose we were to start out with a traditional left-to-right stress system like Waray. If this language started to favour the penultimate syllable for secondary stressing through di¤usional pressure, we would have to represent that tendency by placing one single foot over the last two syllables. The main

78

Rob Goedemans

stress domain would remain on the left. This is the first and smallest step a language can take towards right oriented stress. However, after this step, a left-to-right count system can no longer evolve. Consider (28a), the resulting type schematically. (28) a.

|* | xx xxxxx (* .)(* .) (* .)

c.

|* | xxxxxxx (* .) (* .)(* .)

b. | * | xxxxxxx (* .)(* .)(* .)

d.

|* | xxxxx xx (* .)(* .) (* .)

Starting from (28a), we can change the direction of iterative secondary footing and start at the right edge, while leaving the main stress domain on the left, to derive a right-to-left count system (28b), or we can leave one main stress foot on the left to derive the Garawa type (28c). We cannot, however, move the main stress domain to the right to derive a count system. That movement would lead to straightforward penultimate stress, whatever we do with the other secondary stresses (28d). It is true that movement of the main stress domain to the right, as a first step before placing the secondary stress foot over the last two syllables (which results in (28a)), leads to a left-to-right count system, but I consider it to be extremely unlikely that moving the main stress domain constitutes the first reaction to di¤usional pressure. If we start out with a traditional penultimate stress system, the first reaction to di¤usional pressure would likely be to put a secondary stress foot over the first two syllables, deriving initial secondary stress. The reasoning above applies again, only inversely. This time right-to-left count systems can no longer develop. Since such right-to-left count systems are the only ones we find in Arnhem Land, this theoretical evidence shows that it is more likely that we started out with initial stress systems that moved towards right oriented systems under di¤usional pressure than the other way around. I agree that the above does not constitute enough evidence to decide for or against di¤usion in the direction of penultimate stress systems. Much more evidence needs to be gathered from the sources that describe the languages in the region. However, it would seem there is at least enough evidence to allow further speculation with respect to one final riddle. If the penultimate stress languages are the innovation, as I suggest, this

An overview of word stress in Australian Aboriginal languages

79

pattern must somehow have developed spontaneously in one or more Arnhem Land languages, and we need an explanation for that. Such an explanation is easy to find. First note that the languages of Australia have been isolated for too long to attribute the development of the penultimate stress pattern to borrowing from languages outside Australia. We must look for an internal reason. From Dixon (1980) we learn that Proto-Australian most likely had initial stress.11 The language had an agglutinative morphology, allowing su‰xes but no prefixes to be attached to stems, like most Aboriginal languages today. This means there was no circumstance in which stress would not be word initial. If no QS factors were at play there would be a rock solid stress located on the first syllable. Since we have just assumed that penultimate stress is a trait that developed in the Arnhem Land languages, we must also assume that these languages once showed regular initial stress, like their Proto-Australian ancestor. The key to the development of non-initial stress patterns could be the fact that these languages started to develop prefixes in a later stage (Heath 1978, Dixon 1980). With regard to prefixes and stress, there are two possible courses of action. Firstly, the stress rules can be applied to the entire derived form (post-lexically) and thus derive stress on the initial syllable of all words, including those with prefixes. Secondly, the primary stress can be maintained on the first syllable of the stem, while secondary stresses might occur on the first syllable of disyllabic prefixes. I propose that the latter pattern followed in the wake of prefixing in at least one of the Arnhem Land languages. In this language, the prototypical Australian stem, which is disyllabic, could be augmented with a monosyllabic prefix to form a trisyllabic word with penultimate stress. Should the prefix be disyllabic, main stress would be penultimate and secondary stress initial. This pattern, which occurred in abundance since a fair number of the stems in Australian languages are disyllabic, may have regularised later, such 11. Note that this does not necessarily give us any information about the direction of di¤usion. It makes it more likely that the penultimate stress pattern arose spontaneously (which it had to do if Dixon is right and penultimate stressing did not exist in Proto-Australian) and is now gradually expanding its influence among the traditional initial stress languages (in that respect Dixon’s reconstructive evidence can be added to the evidence for penultimate stress as the dominant di¤usional pattern above). However, given just this evidence concerning stress in Proto-Australian, there is no reason to reject the hypothesis that penultimate stressing did indeed arise spontaneously, had its moment of glory among the Arnhem Land languages, but is now back on the retreat under the influence of the neighbouring intial stress languages.

80

Rob Goedemans

that larger stems and monomorphemic words were also stressed on the penultimate syllable. We thus end up with two competing alternatives for the stress pattern of monomorphemic words that are larger than two syllables. They can either stress the first syllable of the stem, or follow the new rule and stress the penult. This would neatly explain the numerous exceptions we find in the prefixing Arnhem Land languages (see section 3.1). In Evans’ (2003) work on the Bininj Gun-Wok languages (one of which is Kunwinjku), we find a tale-telling paragraph on stress. Evans claims that stress in Bininj Gun-Wok is right-orie¨nted, and indeed, many verbs with bisyllabic and trisyllabic roots string so many prefixes to their left that many of the resulting patterns look like penultimate and antepenultimate stress patterns (respectively) at a superficial level. There are, however, sure signs of initial stress in Bininj Gun-Wok. Four syllable nouns, for instance, have initial stress (and, as Evans notes, a weaker stress on the third syllable). In fact, stress is nearly always morpheme initial. We can easily derive the pattern by assigning standard trochees, left-to-right, to every morpheme individually, where stress on the root is usually the strongest (allowing for special rules when monosyllabic morphemes come into play). Thus, Kunwinjku is an initial-stress language, as indicated in figure 1. Evans, however, analyzes stress by assigning a ‘‘foot’’ to every morpheme (however long it is, and thus incorporating the primary and secondary stress in four-syllable nouns in one ‘‘foot’’) and promoting the leftmost syllable in the foot to its head (thus allowing for primary stress to occur far away from the right word edge, as there can be more than two syllables in the ‘‘foot’’). A second rule states that stress must be as close to the right edge as possible but never on the final syllable. Hence, stress usually ends up on the head of the rightmost ‘‘foot’’ unless that foot is monosyllabic. Thus, Evans claims that Bininj Gun-Wok languages, and therefore Kunwinjku, belong in the right-edge (penultimate) camp. As such, Evans’ paragraph on stress in Bininj Gun-Wok is a striking example of how an initial stress language can be reinterpreted as a rightorie¨nted stress language in the presence of confounding prefixes. The innovative penultimate pattern may have spread to neighbouring prefixing languages in a later period, through the usual means of intermarriage, borrowing lexical items, etc. (see Heath 1978), skipping the Daly River family (left white area in figure 1) at first, since internal morphological composition is unclear in these languages, but eventually even conquering some non-prefixing languages, and a¤ecting the formerly rigid stress rules of languages like Garawa.

An overview of word stress in Australian Aboriginal languages

81

It is important to realise, however, that if the direction of change described here is in fact accurate, it is nothing more than a general tendency. Individual languages may behave di¤erently. In his grammar of Jingulu, Pensalfini (2003) for instance, hints at the possibility that stress in this language is changing slightly under the influence of neighbouring initial-stress languages. All in all, the evidence I have found for such movement towards initial stress is limited. Hence the proposal for movement towards penultimate stress above. Finally, I would note that the logical possibility of penultimate stress is open to all prefixing languages, not just those that are in the spreading area in figure 1. The area in which prefixing languages are spoken roughly coincides with the (non-)Pama-Nyungan border, which runs on to the left of figure 1 up to the west coast. It will not come as a surprise that two of the languages on the prefixing side of the border I looked at, Nyigina (Stokes 1982) and Ngarinyin (Coate and Oates 1970), show traces of right oriented stress. More research in this area is needed to determine whether Nyigina and Ngarinyin are isolated cases or not.

4. Conclusion In this chapter I have given a fairly complete overview of the stress types we can find on the Australian continent. This overview served to reintroduce the StressTyp framework to the reader. After introducing the basic initial type and its variants, I zoomed in on the languages spoken in Arnhem Land, an area in which a second basic pattern can be found. Some of the Arnhem Land languages place primary stress on the penultimate instead of the initial syllable. It appeared that many stress systems of Arnhem Land languages could be characterised as intermediate between the two basic patterns. A detailed overview of these systems was presented in sections 3 and 3.1. In the final section an attempt was made to determine the direction in which the stress patterns of these languages are evolving. A provisional conclusion we could draw from the scarce evidence we have is that they are moving towards penultimate stress. It was proposed that the penultimate stress pattern arose in one or more languages as the result of the development of prefixing in the non-PamaNyungan languages and spread through Arnhem Land to produce the division into initial, penultimate and intermediate stress systems we find today.

82

Rob Goedemans

References Birk, David B.W. 1976 The Malakmalak language: Daly River (Western Arnhem Land). Pacific Linguistics, Series B-45. Canberra. Capell, Arthur C. 1962 Some Linguistic Types in Australia. Oceania Linguistic Monographs 7. University of Sydney. Capell, Arthur C. and H.E. Hinch 1970 Maung grammar: Texts and vocabulary. The Hague: Mouton. Chadwick, Neil 1971 Ngarndji word list and phonological key. Papers on the Languages of Australian Aborigines, 34–45. Australian Aboriginal Studies 38. Canberra: AIATSIS. Chadwick, Neil 1975 A descriptive study of the Djingili language. Research and Regional Studies 2. Canberra: AIATSIS. Coate, Howard H.J. and Lynette Oates 1970 A grammar of Ngarinjin: Western Australia. Australian Aboriginal Studies 25. Linguistic Series 10. Canberra: AIATSIS. Davies, Jennifer 1989 Umbugarla: A sketch grammar. Honours Thesis, ANU. Dixon, Robert M.W. 1980 The Languages of Australia. Cambridge: Cambridge University Press. Eather, Bronwyn 1990 A grammar of Nakkara. PhD Dissertation, ANU. Evans, Nicholas 2003 A grammar of Bininj Gun-Wok: A pan-dialectal grammar of Mayali, Kunwinjku and Kune. Pacific Linguistics 541. Canberra. Furby, Christine E. 1974 Garawa phonology. Pacific Linguistics, Series A-7, 1–11. Canberra. Glasgow, Kathleen 1981 Burarra. Bruce Waters (ed.), AAB SIL Working Papers. Series A-5, 63–90. Goedemans, Rob W.N. 1996 Syllable Weight and Prominence. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 115–163. Goedemans, Rob W.N. 1998 Weightless Segments: A Phonetic and Phonological Study Concerning the Metrical Irrelevance of Syllable Onsets. LOT Dissertations 9. The Hague: Holland Academic Graphics.

An overview of word stress in Australian Aboriginal languages

83

Goedemans, Rob W.N., Harry G. van der Hulst and Ellis A.M. Visch 1996 Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics. Grimes, Joseph E. and Barbara F. Grimes 1996 SIL Ethnologue, 13th Edition. Dallas: Summer Institute of Linguistics. Hale, Kenneth 1964 Classification of the Northern Paman Languages, Cape York Peninsula, Australia: a research report. Oceanic Linguistics 3: 248–265. Harvey, Mark 2002 A grammar of Limilngan. Pacific Linguistics 516. Canberra. Harvey, Mark 2002 A grammar of Gaagudju. Berlin, New York: Mouton. Hayes, Bruce 1995 Metrical Stress Theory. Chicago: University of Chicago Press. Heath, Je¤rey 1978 Linguistic Di¤usion in Arnhem Land. Canberra: AIATSIS. Hercus, Luise 1986 Victorian languages: A late survey. Pacific Linguistics, Series B-77. Canberra. Hercus, Luise 1994 A grammar of the Arabana-Wangkangurru language: Lake Eyre Basin, South Australia. Pacific Linguistics, Series C-128. Canberra. Hoddinott, W.G. and F.M. Kofod 1988 The Ngankikurungkurr language (Daly River area, Northern Territory). Pacific Linguistics, Series D-77. Canberra. Holmer, Nils M. 1983 Linguistic survey of South-Eastern Queensland. Pacific Linguistics, Series D-54. Canberra. Hore, Michael 1981 Syllable length and stress in Nunggubuyu. Bruce Waters (ed.), AAB SIL Work Papers. Series A-5, 1–62. Hulst, Harry G. van der 1996 Separating Primary and Secondary Accent. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 1–25. Hulst, Harry G. van der 1996b Primary Accent is Non-metrical. Revista di Linguistica 9/1: 1–119. Keen, Sandra 1983 Yukulta. Robert M.W. Dixon and Barry Blake (eds.) Handbook of Australian Languages 3. Amsterdam: John Benjamins, 191– 305.

84

Rob Goedemans

Kirton, Jean F. 1977 Anyula phonology. Pacific Linguistics, Series A-10. Canberra. Lee, Jennifer 1987 Tiwi today: A study of language change in a contact situation. Pacific Linguistics, Series C-96. Canberra. Marsh, James 1969 Mantjiltjara phonology. Oceanic Linguistics 8: 2, 131–152. McCarthy, John J. and Alan Prince 1994 Generalized alignment. Yearbook of Morphology. 1993: 79–153. McKay, Graham R. 1975 Rembarnga: A language of central Arnhem Land. PhD Dissertation, Australian National University. McKay, Graham R. 2000 Ndje´banna. Robert M.W. Dixon and Barry Blake (eds.). The Handbook of Australian Languages 5. Oxford: Oxfod University Press, 155–356. Merlan, Francesca 1983 Ngalakan grammar, Texts and vocabulary. Pacific Linguistics, Series B-89. Canberra. Merlan, Francesca 1989 Mangarayi. New York: Routledge. Merlan, Francesca 1994 A grammar of Wardaman: A language of the Nortern Territory of Australia. Berlin-New York: Mouton. Osbourne, C.R. 1966 A tentative description of the Waanji language. Ms, ANU. Osbourne, C.R. 1974 The Tiwi language. Australian Aboriginal Studies 55. Linguistic Series 21. Caberra: AIATSIS. Pensalfini, Rob 2003 A Grammar of Jingulu: an aboriginal language of the Northern Territory. Pacific Linguistics 536. Canberra. Sandefur, John and David Nangan:golod Jenhan 1977 A tentative description of the phonemes of the Ngalkbun language. Workpapers of SIL AAB. Series A-1, 57–96. Sharpe, Margaret C. 1972 Alawa phonology and grammar. Canberra: AIATSIS. Stokes, Bronwyn 1982 A description of Nyigina: A language of the West Kimberley, western Australia. PhD Dissertation, Australian National University. Taylor, Peter and John Taylor 1971 Kitja. Papers on the languages of the Australian aborigines. Australian Aboriginal Studies 38, 100–109. Canberra: AIATSIS.

An overview of word stress in Australian Aboriginal languages

85

Thompson, David A. 1976 A phonology of Kuuku-Ya/u. Peter Sutton (ed.) The Languages of Cape York. Canberra: AIATSIS, 213–235. Tryon, Darrell T. 1970 An introduction to Maranungku. Pacific Linguistics, Series B-14. Canberra. Visch, Ellis A.M. 1996 The Linguistic Structure of StressTyp. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 69–114. Walsh, Michael J. 1976 The Murinypata language of North-West Australia. PhD Dissertation, Australian National University. Wordick, F.J.F. 1982 The Yindjibarndi language. Pacific Linguistics, Series C-71. Canberra. Wurm, Stephen A. and Shiroˆ Hattori 1981 Language atlas of the Pacific area. Canberra: The Australian Academy of the Humanities in collaboration with the Japan Academy. Yallop, Colin 1977 Alyawarra: An aboriginal language of Central Australia. Canberra: AIATSIS.

3.

Stress types in Austronesian languages

Ellen van Zanten, Ruben Stoel and Bert Remijsen1 1. Introduction The Austronesian language family is estimated to comprise around 1,200 languages, or approximately 20% of all the languages of the world. The total number of speakers is around 270 million (Tryon 1995: 6). Austronesian languages extend from Madagascar in the west to Easter Island in the east, and from Taiwan in the north to New Zealand in the south. Austronesian languages are usually grouped as in Figure 1 (after Blust 1978). On account of prehistoric evidence, the first Austronesians are thought to have originated in the South China area.2 Around 3500 B.C., some of them moved to the island of Taiwan (Formosa), which is considered to be the Austronesian homeland, i.e. the centre of the greatest linguistic diversity (Dyen 1956, 1965). The Austronesian languages spoken on Taiwan (sometimes called Formosan languages) are said to fall into three highest order subgroups: Atayalic, Rukai-Tsouic and Paiwanic. The fourth first order subgroup, Malayo-Polynesian, comprises all of the Austronesian languages spoken outside Taiwan (Blust 1978). Members of the Malayo-Polynesian subgroup moved south from Taiwan to the Philippines. A next migration moved south-west (roughly the Western Malayo-Polynesian subgroup). Part of this group subsequently settled on the Southeast Asian mainland. Another migration moved south, eventually splitting south (corresponding to the Central Malayo-Polynesian subgroup) and east (corresponding to the South Halmahera-West New Guinea subgroup). During the second millennium B.C. migrations took 1. The authors are indebted to Rob Goedemans, Harry van der Hulst and Hein Steinhauer and an anonymous reviewer for valuable comments on earlier versions of this paper. Rob Goedemans also kindly drew Figure 2. 2. The following contains some scant information on the Austronesian language family. For an extensive overview the reader is referred to Tryon (1995) and references cited there. Recently, the migration account was corroborated by a computational linguistic analysis (Gray, Drummond and Greenhill 2009) and by analysis of the distribution of the bacterial parasite of humans Helicobacter pylori (Moodley et al. 2009).

88

Ellen van Zanten, Ruben Stoel and Bert Remijsen

Figure 1. Subgrouping of Austronesian languages (after Blust 1978).

place from the Halmahera-West New Guinea area to the many islands in the Pacific Ocean (the Oceanic subgroup; see figure 2). Some Malayo-Polynesian languages have very few speakers. Together the 493 Oceanic languages (Grimes 2000) comprise approximately two million speakers only. The vast majority of speakers of Austronesian languages belong to the Western Malayo-Polynesian subgroup. In 1990, the twenty-five largest Western Malayo-Polynesian languages together comprised 240 of the total of 270 million Austronesian speakers. With 75 million speakers (Tryon 1995: 17), Javanese was by far the largest Austronesian language in 1990. Second in number of speakers in 1990 was Malay/Malaysian/Indonesian, which also belongs to the Western Malayo-Polynesian subgroup. Since the thirteenth century, Malay has served as a lingua franca in the region, spreading over many coastal areas. Gradually, it took on an ever increasing role in o‰cial institutions as well (Teeuw 1959; Steinhauer 1980; Adelaar and Prentice 1996). In 1928, Malay (which, at that time, was spoken by less than 10% of the Indonesian population; Steinhauer 2000) was chosen by the Indonesian nationalists as the national language, and renamed Bahasa Indonesia, the Indonesian language. Nowadays, varieties of Malay serve as national languages of Malaysia, Brunei, Singapore and Indonesia. In 1995, these were estimated to number 100 million speakers, which is almost double the number of speakers in 1990, according to Tryon (1995: 17). The Indonesian 1990 census revealed that knowledge of Indonesian had reached almost 100% in the urban areas of Indonesia and that there were just over 24 million first-language speakers

Stress types in Austronesian languages

89

Figure 2. The Austronesian subgroups; geography and migrations (after Tryon 1995). F ¼ Formosan (comprising the Atayalic, Rukai-Tsouic and Paiwanic subgroups), W ¼ Western Malayo-Polynesian, C ¼ Central Malayo-Polynesian, S ¼ South Halmahera-West New Guinea,3 O ¼ Oceanic.

of Indonesian (0–4 year olds not included; Steinhauer 1994: 784). Many of the smaller Austronesian languages, on the other hand, are in danger of extinction. In the following sections we will discuss the StressTyp data on Austronesian languages. Section 2 summarizes the data we found in the database and Section 3 o¤ers some background information on these data. In Section 4, we discuss one of the di‰culties which researchers may encounter in establishing stress patterns. Section 5, finally, summarizes our findings on stress in Austronesian languages.

2. Stress types The StressTyp database contains 117 of the 1236 Austronesian languages (Grimes 2000), amounting to almost 10%. Remark, however, that the sizes of samples for the various subgroups di¤er considerably. The Central 3. Or: South Halmahera-Irian Jaya (Lynch, Ross and Crowley 2002).

90

Ellen van Zanten, Ruben Stoel and Bert Remijsen

Figure 3. Stress types and numbers of Austronesian languages in StressTyp. Each symbol F, W, etc. represents one language (F ¼ Formosan, W ¼ Western Malayo-Polynesian, C ¼ Central Malayo-Polynesian, S ¼ South Halmahera-West New Guinea, O ¼ Oceanic, U ¼ unclassified).

Malayo-Polynesian subgroup is well represented: almost 15% of the languages belonging to this subgroup are included in the database. The Oceanic subgroup (493 languages), on the other hand, is represented by a sample of 31 languages only, i.e. just over 6%. A list of subgroup and sample sizes can be found in the Appendix. Figure 3 summarizes the stress types of the 117 languages in the database and the subgroups to which they belong. Each letter (F, W, etc.) represents one language. The label ‘‘Unique’’ collapses types that occur only once; they are listed in the Appendix. In the great majority of Austronesian languages the stress domain is located at the right edge of the word (104 languages, or 89%). In 88 languages (75%) stress is restricted to the penultimate or ultimate syllable. These include 45 languages with (fixed) penultimate stress (P), 6 languages with (fixed) ultimate stress (U), and 37 languages in which stress falls on the penultimate syllable in some words, and on the ultimate syllable in others. The group of 37 languages in which the stress can be either on the penultimate or on the final syllable can be divided into two subsets. The first subset contains 24 languages for which the stress position is

Stress types in Austronesian languages

91

predictable by comparison of the weight of the syllables in the stress domain. This group includes the P/P, P/U, U/P, and U/U languages. Table 1 presents the location where the stress will end up given the weight divisions in the stress domain for these four codes. Table 1. Position of the stressed syllable in four quantity-sensitive bounded stress types (H ¼ heavy syllable, L ¼ light syllable). Stress type

Syllables in PU window HH

HL

LH

LL

P/P

P

P

U

P

P/U

P

P

U

U

U/P

U

P

U

P

U/U

U

P

U

U

The second subset includes thirteen languages for which the position of the stressed syllable is not predictable. Of these thirteen languages, nine, with the code P;U, have penultimate stress in most words, but words with ultimate stress also occur.4 Stress assignment in the two P/U;P languages basically follows the rule for the P/U languages, but a few words have penultimate stress where the rule would predict ultimate stress. Two languages with ‘‘unique’’ stress type codes also belong to this subset: Mongondow of Northeast Sulawesi with the code P;P/U, which indicates that the language concerned is similar to P/U;P languages except for the size of the exception vs. regular groups;5 secondly, Southeast Ambrym, an Oceanic language, with the code U þ P/P which refers to a truly unique stress pattern. Besides the 88 languages in which stress position is restricted to the penultimate or ultimate syllable, StressTyp contains 29 Austronesian languages (25%) where stress is not limited to these two syllables. These include two languages that have penultimate stress in most words, but unpredictable stress in a few others (P;LEX), and two languages with

4. This is also true for some languages which are coded P. In Taba, for example, some disyllabic words have final stress. For a language to be coded P, the exceptions have to be few, however. 5. According to the second author’s observations Mongondow does not have word-based stress.

92

Ellen van Zanten, Ruben Stoel and Bert Remijsen

antepenultimate stress (A). The only stress type with a stress domain that is not located at the right edge of the word and which is found in more than one language is initial stress (I). It is attested in four languages, all of which belong to the Oceanic subgroup. The remaining 21 languages for which the stress position is not limited to the penultimate or final syllable all have a unique stress type code. Most of these are Oceanic languages. For a full list of unique types the reader is referred to the Appendix. Figure 3 shows that penultimate stress (P) is the most widespread stress type, but P/P and U/P, which are closely related to P, are also common. In U/P languages stress falls on the ultimate syllable if this syllable is heavy, and on the penultimate syllable in all other cases (cf. Table 1). Typically, a syllable with a long vowel is counted as heavy. Note that these languages would have belonged to the P type, if long vowels had been analyzed as two syllables. This analysis is indeed found in some of the descriptive grammars. In StressTyp it is only followed in the few cases where further evidence for bisyllabicity of long vowels is provided. In P/P languages, stress falls on the ultimate syllable if it is heavy and the penultimate is not, and on the penultimate syllable in all other instances6 (cf. Table 1). Remarkably, not one of the seven P/P languages in StressTyp has long vowels; typically, all syllables with a vowel other than schwa are counted as heavy. Also, long vowels are found in only five of the 45 P languages (11%). Eight of the twelve U/P languages, on the other hand, do have long vowels (67%). Somewhat simplifying, one could say that Austronesian languages that have only full vowels tend to be coded P, languages that also have a schwa tend to be P/P, and languages that make a distinction between long and short vowels tend to be U/P.

3. Past and present In Section 2 we pointed out that almost 90% of the Austronesian languages in the StressTyp database have the stress located at the right edge of the word. Note that the Austronesian languages in the database belong almost exclusively to one first order subgroup, viz. Malayo-Polynesian. Of 6. In P/P languages with an extrametrical final syllable, stress is on the antepenultimate syllable if this syllable is heavy and the penultimate light. This does not occur in Austronesian languages.

Stress types in Austronesian languages

93

the 23 languages which belong to the other three highest order subgroups (Atayalic, Rukai-Tsouic and Paiwanic, all spoken on Formosa; cf. Figure 1) only one, Paiwan (StressTyp code P), is included in the database. Within the Malayo-Polynesian subgroup three main groups of stress patterns deserve a closer look, viz. contrastive stress, ultimate stress and penultimate stress. 3.1. Contrastive stress languages StressTyp includes 25 Malayo-Polynesian languages for which the stress position is not predictable although it occurs in a limited range of positions. They are coded P;U (9 languages), P/U;P (2), P;LEX (2) and 12 ‘‘unique’’ types (cf. Figure 3 and Appendix). In these languages, words may be segmentally identical but di¤er in stress position (minimal stress pairs). As the stress position is the contrastive element, we will call those languages contrastive stress languages.7 Contrastive stress languages are primarily found in the Central and Southern Philippine subgroups of Western Malayo-Polynesian. Five out of the seven Philippine languages in StressTyp have contrastive stress. Apart from these Philippine languages, there are nine Western Malayo-Polynesian languages with contrastive stress in the database. Of these, Toba Batak on Sumatra has a stress system which is very similar to that of the Philippine subgroup.8 Both have cognate forms with the same stress patterns, as well as minimal stress pairs with penultimate stress for nouns, and ultimate stress for (stative) adjectives. This similarity is seen as evidence that the phenomenon is inherited and that the Proto-Western Malayo-Polynesian subgroup had contrastive stress (Wol¤ 1993; Zorc 1978, 1993). Taking the argument a step further, Wol¤ sees the historical weakening and loss of vowels in

7. The terms lexical stress and free stress are also used to describe unpredictable stress position. However, the term lexical suggests that for each word the stress position has to be stored in the lexicon, whereas in the Austronesian languages under scrutiny most words have penultimate stress, and for only a minority of words the stress position has to be stored in the lexicon. Free stress is an ambiguous term as it is not only applied to languages in which the stress position varies from word to word, but also to languages where the stress can vary in di¤erent realizations of one word. By our definition (cf. Section 4), the latter type of languages do not have a word-based stress, but rather a phrase-based accent (van Heuven 1994). 8. Actually, all (four) languages belonging to the Southern Batak group have distinctive stress according to Adelaar (1981).

94

Ellen van Zanten, Ruben Stoel and Bert Remijsen

unstressed (penultimate or final) syllables in two of the Formosan first order subgroups (Atayalic and Rukai-Tsouic) as evidence of contrastive stress in Proto-Austronesian (Wol¤ 1993). This viewpoint is shared by Ross (1992) and Zorc (1993). Zorc (1997) implicitly adheres to the contrastive-stress position by indicating stress position in Proto-Austronesian forms. Blust (1997), however, considers the Formosan evidence for ProtoAustronesian stress unconvincing: the comparisons between Rukai and Proto-Philippine contrastive stress forms by Ross (1992) do not exclude an independent development of contrastive stress in Rukai and ProtoPhilippines (Blust 1997). The question whether Proto-Austronesian had contrastive stress or penultimate stress was already raised by Brandstetter (1915: 91), who suggested that, parallel to developments in Indo-European languages, penultimate stress may have developed from the original contrastive stress. If Proto-Austronesian had contrastive stress, then the loss of the phenomenon in many Philippine and other Austronesian languages must be explained. This is done by Zorc (1977: 217–218; 1978: 84) by pointing out the frequent bilingualism and substratum influence from other languages of the region. For instance, Tausug maintained the inherited contrastive stress patterns on one island, but lost it on another, where it was in close contact with a non-contrastive stress language, Samal. In a number of Austronesian languages contrastive stress developed from earlier penultimate stress. On the island of Sulawesi, penultimate stress is the prevailing pattern. However, in two Northern Sulawesi languages, Bantik and Ratahan, stress is no longer strictly penultimate. In the common ancestor Proto-Sangiric, stress fell always on the penultimate syllable (Sneddon 1984: 53). In Bantik (code P;A;U) stress shifted from the penultimate schwa to the following syllable. The penultimate schwa subsequently adopted the features of the final vowel, while the stress remained on the final syllable. Synchronically, the stress position is thus no longer predictable (Sneddon 1984: 53): (1) sa¨haN pu¨ku pa¨higi

‘ant’ ‘to bend’ ‘well’

¨sahaN ¨buku pahi¨gi

‘tree’ ‘to kneel’ ‘knife’ (Sneddon 1984: 23)

A similar process took place in Ratahan (code P;U). In nine Sulawesi languages in the StressTyp database there is a di¤erent reason why the position of the stressed syllable is not fixed, although stress position in these languages is again often predictable. The shift in stress

Stress types in Austronesian languages

95

position is mainly due to the so-called paragogic vowel (support vowel, echo vowel), which occurs in these languages. The paragogic vowel is added after certain word-final consonants (Sneddon 1989: 89; Sneddon 1993; Sneddon and Usup 1986: 411). The stress position does not shift when the paragoge is added, and words with a paragoge thus have antepenultimate stress. The quality of the paragoge vowel may di¤er from language to language. Consider again two languages from the Northern Sulawesi Sangiric group. In Sangir (code P), /«// is added after all roots with a final consonant other than /// or /N/. Consequently, all words ending in /«// have antepenultimate stress (Sneddon 1984: 20). In Bantik (code P;A;U), on the other hand, the added /V// contains a copy of the last vowel of the root. Here it is not true that words ending in this vowel necessarily have antepenultimate stress, since many roots end in this vowel too. Bantik, then, is one of the few languages in Sulawesi where stress has become contrastive, and some minimal stress pairs do indeed occur, as shown in (1). In a number of Malay languages, like Ambonese Malay (van Minde 1997), Kupang Malay (Steinhauer 1983), Manado Malay (Prentice 1994), and North Moluccan Malay (Taylor 1983), the inherited unstressed schwa in the penultimate syllable was, like in Bantik, replaced by a full vowel, which remained unstressed. This resulted in some minimal stress pairs in these Malay languages. (2a) Kupang Malay: ba¨rat ‘heavy’ ba¨tu ‘correct’

¨barat ‘west’ ¨batu ‘stone’ (Steinhauer 1983: 44)

(2b) Manado Malay: se¨pi la¨la

‘silent, lonely’ ¨sepi ‘tired’ ¨lala

‘edge, brink’ (Prentice 1994: 418) ‘girl’s name’ (Stoel 2007: 118)

(2c) N. Moluccan Malay: ba¨rat ‘heavy’

¨barat ‘west’

ju¨mat ‘congregation’ ¨jumat ‘Friday’ (Taylor 1983: 17) Note that variants occur in Manado Malay: ba¨sar and ¨basar ‘large’ ( > >

flı´ga`no` ‘Filigano’ kha´vu`da´ ‘tree bark type’ flu´flu´ ‘quiet, calm’

Ford’s vowel deletion reminds us of Renck’s reduced syllables, in which the vowel is said to be hardly audible. Such reduction of unstressed syllables is well known in stress languages like Dutch (Nooteboom 1972) and Russian (Kenman 1975); in tone systems, on the other hand, vowel deletion in low syllables versus high syllables would seem unlikely (cf. Pike 1974).16 In the examples that Ford provides, vowel deletion is only found in initial low-tone syllables which are followed by high syllables, and not in other low-tone syllables (with the exception of fu`lu´fu`lu´ > flu´flu´ ‘quiet, calm’, which may be a reduplicated form).17 Renck’s statements that stressed syllables are usually high (‘‘mid’’ as opposed to ‘‘low’’ in Renck’s terminology) and that unstressed syllables may be reduced, in combination with Ford’s examples of deletion of low vowels preceding a high vowel, suggest to us that Move may well be a stress language. The most likely default stress position would then be the 16. Low-tone vowel deletion is attested in Ga˜, a Kwa language spoken in Ghana, Africa. In Ga˜, the non-high subject prefix -e is deleted when it is preceded by a high-tone noun phrase subject and followed by a high-initial verb stem (Paster 2003: 31–33). We thank the anonymous reviewer for drawing our attention to this. 17. However, the non-initial low syllables in the examples do not meet Ford’s other prerequisites for reduction (i.e. low vowel and following high vowel are identical or separated by /h/). Therefore, it is impossible to decide from the examples whether the reduction is indeed restricted to (low) initial syllables.

Word stress and pitch accent in Papuan languages

129

first high syllable. On account of the examples provided by Renck (1975) and Ford (1993), it is obviously impossible to decide whether this analysis is correct. Finally, both Renck and Ford quote a few minimal word pairs in Move, be they stress or tone oppositions.18 In our tentative analysis this would suggest contrastive stress. We therefore propose the code F/LEX (tone) for Move: stress the first high tone (with some exceptions). This analysis is also proposed by Hendriks (1995: 89) – at least for Move nouns – who cites the example ¨ha´lo´te´qna´ ‘light’ with stress on the first high-pitched syllable from Renck (1967: 41). c. Tone is conditioned by stress position In only one language in our sample, Mairasi, tone is reportedly conditioned by the position of the stress. Peckham (1991: 112–113) writes that primary stress is on the penultimate syllable. There are two contrastive pitch levels in Mairasi, which contrast only on stressed syllables. These normally take high pitch. But ‘‘Only twelve minimal pairs of contrasting pitch have been found to date.’’ (Peckham 1991: 142, note 3). Unfortunately, Peckham’s examples do not quite bear out his statements. Although stress is penultimate in the majority of Peckham’s examples, apparently the stress may be located on the final syllable if this syllable contains a long vowel (as in /a¨bo´: ‘close’ and /a¨bo`: ‘soon’; Peckham 1991: 113), or if it ends in a consonant (as in na/¨gwar ‘alone’ and fa¨En ‘split’; Peckham 1991: 142, note 4). Moreover, the only examples with pitch as well as stress indications provided by Peckham (1991: 113) show a tonal opposition in unstressed syllables as well as in stressed syllables: (6) ¨/anda´ ‘branch’ /a¨bo´: ‘close’

¨/anda` ‘pants’ /a¨bo`: ‘soon’

If these examples are correct, Mairasi may have either a four-way contrast, with no relation between stress position and tone, or the tonal contrast may be restricted to the final syllable. Evidently, more information is needed to decide on this issue. For lack of a more elaborate description we

18. For instance, Ford (1993: 201) gives ha`lı´ (low–extra-high) ‘firewood’, versus ha`lı´ (low–high) ‘arrow’. Above, Renck analysed these examples as ha#¨lı # (mid–mid [þstress]) ‘arrow’, versus ha`¨lı # (low–mid [þstress]) ‘fire’.

130

Ellen van Zanten and Philomena Dol

make do with Peckham’s analysis that the pitch contrast only occurs on stressed syllables.19 It is remarkable that hybrid languages are only found in a few areas. As Table 3 shows, the largest number (five) is spoken in the Lakes Plains area of Papua Province, Indonesia. Three hybrids are spoken in the East New Guinea Highlands, and two in the Western Highland Province of Papua New Guinea. In each of these regions, we found a bewildering variety of word-prosodic systems. Hardly two languages seem to use the same system. Even within single families the stress placement rules vary. For instance, the three hybrid languages of the Central Lakes Plain family each have a di¤erent system: fixed initial stress (Doutai; McAllister & McAllister 1991), ‘‘stress the last high tone, or else the first syllable’’ (Sikaritai; Martin 1991: 94), and ‘‘stress the last high tone, or else the last syllable’’ (Obokuitai; Scott & Jenison 1991). The other two hybrids of the Lakes Plains area, Kaure and Iau, belong to di¤erent families, and have, yet again, di¤erent systems. Possibly accidentally, we did not come across any pure tone or pure stress languages in this area.20 The word-prosodic systems of Kaure and Wahgi resemble that of Ma¨ya (Remijsen 2001, 2002). Ma¨ya is an Austronesian language spoken on Salawati o¤ the west coast of the Bird’s Head peninsula, in which both tone and stress are used contrastively. Stress is on the penultimate or final syllable. The three-member tonal contrast in Ma¨ya is restricted to the final syllable. Remijsen (2001, 2002), who gives a thorough phonetic as well as phonological description of the Ma¨ya word-prosodic system, claims that this system, with both tone and stress used contrastively, is unique in the world. However, both Kaure and Wahgi are also said to have contrastive stress as well as contrastive tone. Wahgi mainly seems to di¤er from Ma¨ya in that its lexical tones (two according to Luzbetak 1956; two or three according to Phillips 1976) are not restricted to the final syllable. Kaure (Dommel & Dommel 1991) has three tones; the stress is usually, but not always, on the word-final high-mid fall. And in words with only (level) high tones stress position is unpredictable.21 19. Similar systems may exist in Kaure (Donohue 1997: 370) and in Fasu and Momuna (Donohue 1997: 371 and references there). 20. Unless one sees Iau not as a language with (marginal) stress but as a pure tone language. 21. As indicated before, Donohue (1997) analyses Kaure as a pitch-accent language, with the pitch contrast (realized as high-level or fall) only on the stressed syllable.

Word stress and pitch accent in Papuan languages

131

Hybrid systems are comparatively rare, and we would like to know what brings about such systems. Remijsen (2001: 102–104) suggests that the Ma¨ya word-prosodic system developed through contact between the Austronesian parent language, which featured stress, and Papuan tone languages. Similarly, Papiamento, a Caribbean Creole, inherited lexical stress from Portuguese and/or Spanish, and lexical tone from tonal West African languages, according to Remijsen (2001: 125–126). We might consider a similar explanation for Papuan hybrid languages. In the East New Guinea Highlands, hybrid languages are sometimes found side by side with lexical tone languages or lexical stress languages. We collected descriptions of four languages/dialects of the Gorokan family in this area (the stress systems Hua (I; Haiman 1980) and Kamano (LEX; Drew & Payne no date; James 1963, both cited in Hendriks 1995), and the hybrids Fore and Move). We also collected descriptions of four languages of the Kainantu family (the tone languages Awa, Binumarien and Wa¤a, and the hybrid Usarufa). In the Western Highlands the hybrids Wahgi and Golin, and the stress language Kaugel (LEX; Blowers 1970) are spoken (all of the Chimbu family). It is tempting to suppose some developmental relationship between the prosodic systems of languages within one family. In our search of the literature, we found, however, no evidence for such relationships. None of the hybrid languages seemed in close contact with both stress languages and tone languages. Also, we found no evidence that the hybrid languages existed as pure stress or pure tone systems before the hybrid systems developed. Moreover, we have to bear in mind, that the analysis of the word-prosodic systems of these languages is not unequivocal. For instance, Ford (1993) sees Hua, Kamano and Move not as stress systems, but as pure tone systems. Evidently, more comparative research is needed before anything can be said about the development of hybrid systems on mainland New Guinea.

5. Problems, and how to solve them Problems. For the current survey we used descriptions by other linguists. These descriptions vary from very short and unspecific to clear and reliable. An example of what seems to be a clear and reliable description is McAllister & McAllister (1991). McAllister & McAllister report that Doutai has four tones, and stress, realized as length, on the first syllable of a word. We trust that the authors have indeed found two separate word-prosodic phenomena in Doutai: lexical tone, realized by pitch, and

132

Ellen van Zanten and Philomena Dol

word stress, realized by duration. Other language descriptions also explicitly mention both tone and stress, provide stress rules, or mention stresslike features, like durational e¤ects. Such languages were, like Doutai, categorized as hybrid languages in this survey. Unfortunately, for other languages the situation is less clear. In the previous sections, we already stipulated that some languages were analyzed in di¤erent ways by linguists. A case in point is Kewa, which, as indicated in Section 3, was analyzed in three conflicting ways: as a pitchaccent language, as a tone language (with four tones), and as a stress language. To further illustrate the complicated situation we will now give a short account of the problems we encountered with the categorization of two other languages, Wa¤a and Ekari. In the Ekari case, we came across three contradictory primary descriptions. For Wa¤a, we consulted a primary source and its rather radical phonological re-interpretation. Wa¤a. The description of Wa¤a is by Stringer & Hotz (1973). They state that in Wa¤a contrastive stress may occur on one or more syllables of a word. Moreover, some words have no stressed syllables and some have all syllables stressed. (7) kama ‘round taro’

sı´pu ‘door’

qa´vu´

‘eye’22

Phonetic vowel length is interpreted as sequences of geminate vowels by Stringer & Hotz (1973). In such sequences ‘‘either vowel may be stressed, both vowels may be stressed, or both may be unstressed’’. Stringer & Hotz (1973: 525) provide the following examples: (8) haa´ya mma´ata

‘wing’ ‘bed’

kua´anu´u ‘spit’ mma´a´ra ‘revenge’

ha´aya mmaa´pu

‘fuzz’ ‘son’

na´amme´e ‘type of tree’ naa´noo ‘big sister’

Pike (1974: 169) interprets this as a ‘‘multiple stress system’’. Hendriks (1996) strongly refutes such a multiple stress analysis, arguing that Wa¤a words appear to have an inherent pitch pattern, and that therefore Wa¤a must be either a pitch accent or a tone language. She then analyses Wa¤a as a (spreading) pitch-accent system. The position of the accent is not fixed but contrastive. Hendriks basically gives three reasons 22. We copied Stringer and Hotz’ way of indicating stress here, i.e. an acute accent over the stressed vowel. mm symbolizes a nasalized voiced bilabial fricative in (8).

Word stress and pitch accent in Papuan languages

133

for her analysis. Firstly, Wa¤a has characteristics of both tone and stress systems. Secondly, most Wa¤a words have only one high syllable, or sequence of high-pitched syllables.23 Thirdly, a pitch-accent analysis would bring Wa¤a in line with other languages of the (Kainantu) family. Hendriks may well be right in interpreting Stringer & Hotz’ stressed syllables as high-pitched syllables, and spreading pitch accents could indeed explain the sequences of high-pitched syllables in Wa¤a words. She does not explain what ‘stress characteristics’ Wa¤a has, however, nor why one particular high syllable of a word should be seen as bearing the pitch accent. In our opinion, if the stressed syllables in Stringer & Hotz (1973) are interpreted as high-pitched syllables, there is no evidence of word-based prominence left, and Wa¤a should be regarded as a tone language. Hendriks’ third reason for analyzing Wa¤a as a pitch-accent language is that this would bring it in line with the word-prosody of other languages of the Kainantu family. This is not a very convincing argument, given the lack of agreement in word-prosodic descriptions of languages in the family and in the area as a whole. Hendriks (1996) discusses one other language of the Kainantu family, Usarufa, which she analyses as having both pitch accent and stress. We analyzed Usarufa as a hybrid language with lexical tone and stress (cf. Section 4). Awa and Binumarien, also of the Kainantu family, were categorized as pure tone languages in the current survey, but Awa may be a pitch-accent language according to Foley (1986). Foley specifically mentions the languages of the Kainantu family as languages ‘‘with the most elaborate accent systems, which, with more detailed analysis, might turn out to be true tone systems’’. Further research is, indeed, urgently needed for these languages. For the time being, it would seem rather hazardous to rely on any shared word prosody in the Kainantu family. Until further evidence to the contrary has been collected, we categorize Wa¤a as a tone language as it seems to lack a prominence system. Ekari. The classification of Ekari, our final example, proves particularly problematic. Ekari (also known as Ekagi or Kapauku) is a comparatively large Wissel-lakes language of the West Central Highlands of Papua province, Indonesia. We consulted three descriptions of this language. Drabbe (1952) describes Ekari as spoken at lake Tigi, one of the Wissel lakes. This grammar, which contains only little information on word prosody, uses the term emphasis, not stress, for Ekari, ‘‘omdat we de nadruk die op een lettergreep gelegd wordt niet willen gelijkstellen met de Nederl. 23. Morae in Hendriks’ analysis.

134

Ellen van Zanten and Philomena Dol

klemtoon’’ [as we do not wish to put the emphasis on a syllable on a par with Dutch stress] (Drabbe 1952: 4). Regrettably, Drabbe does not state in what respect it is di¤erent. Ekari ‘emphasis’ is rather strong, but its position shifts in phrases. This suggests phrasal accent rather than word stress. On the other hand, Drabbe (1952) also mentions some minimal stress pairs, for instance ¨ego ‘shy’ versus e¨go ‘tooth’. Lexical tone is not mentioned by Drabbe (1952). According to Doble (1987: 58–61) ‘‘There are two pitch-accent contrasts in Ekari’’. High pitch is marked, but of limited distribution. Apart from the term ‘pitch accent’, Doble (1987) seems to use ‘stress-accent’, and ‘stress’ in a synonymous way. A further di‰culty lies in the fact that the ‘pitch accent’ apparently also occurs in monosyllables: ii ‘yes’ (V:; neutral pitch) versus ii ‘sand’ (V:; high pitch). A pitch contrast in monosyllables suggests a lexical tone contrast rather than a pitch accent. The third source, Steltenpool & van der Stap (n.d.) states that Ekari has three lexical tones as well as (contrastive) stress [‘‘dynamisch accent’’]. This suggests a hybrid system. However, ‘‘Er zijn sommige woorden waarbij de verschillende lettergrepen een gelijk dynamisch accent krijgen’’ [In some words all syllables get the same level of stress] (Steltenpool & van der Stap n.d.: 6). Moreover, there are not many minimal tone pairs, as the tone contrast is usually accompanied by other phenomena (Steltenpool & van der Stap n.d.: 5); tone and stress are related in the sense that the stressed syllable is always the highest in the word, and words usually follow a mid-high-low pattern. These facts point at a pitch-accent system in the classical definition, with the melody spreading over any pre-stress syllables. On the other hand, monosyllables consisting of a long vowel can have contrastive lexical tone, as in ii ‘yes’ (high fall) versus ii ‘sand’ (mid tone).24 The StressTyp code [S/- (LEX;pitch)] for Ekari also points at a pitchaccent analysis. This may indeed be the most likely analysis. It leaves unexplained (at least) two phenomena, however: (i) the pitch contrast in monosyllables, which suggests a tonal system rather than pitch accent; and (ii) the limited distribution of the (stress or pitch accent) contrast. The fact remains that a comparatively large Papuan language lacks a clear description of its word-prosodic system. Native speakers’ judgements. In our opinion, the decision which wordprosodic system a language has should depend on the native speakers’ 24. Notice that this analysis is virtually the reverse of Doble’s (1987), who attributes neutral pitch to ii ‘yes’, and high pitch to ii ‘sand’.

Word stress and pitch accent in Papuan languages

135

intuition. Restricted tone languages and pitch-accent languages may have similar pitch patterns. It may well be that native speakers perceive highpitched syllables as more prominent than low or neutral syllables, but this is not necessarily so. Experimental phonetic research can enable us to analyse particular word-prosodic systems, like that of the Ekari; it may also help us to distinguish between pitch-accent languages and tone languages in general. Such research should include production studies as well as perceptual experiments. We have no knowledge of any comparative phonetic research on Papuan languages. The question on how to decide on phonetic grounds that high-pitched syllables are part of a tone system rather than part of a stress system, was posed by Pike (1974). According to Pike, in a tone system (as opposed to a stress system) high-tone syllables are frequently shorter than low-tone syllables in analogous environments. Pike suggests that this may be a universal (Pike 1974 and references there). Relatively short durations would make high syllables less prominent than their low counterparts. This would then fit in nicely with the idea that in tone languages high syllables are no more prominent than low syllables. In stress languages, on the other hand, stressed syllables are relatively long; and in pitch-accent languages, the high pitch may be accompanied by increased duration and intensity, which enhance the prominence of the accented syllable. This is the case in European pitch-accent languages like Swedish and Serbo-Croatian (Ladd 1996), but not in Japanese where prominence is achieved by a pitch movement only (Beckman 1986). Phonetic research on Papuan (restricted) tone languages and putative pitch-accent languages may shed some light on the exact way in which the tonal or accentual patterns in these languages are realized. Production studies that compare the realizations of the various wordprosodic systems should be complemented by perceptual research. According to Beckman (1986: 27–28), tone and stress di¤er in the attitudes that they characteristically evoke in native speakers. Whereas the speaker of a tone language finds it di‰cult to separate an utterance’s tonal specification from its segmental specification, the speaker of an accent language 25 separates them quite easily. Whereas the tones are viewed as part of the word’s phonemic makeup, the accent pattern seems to be something fastened onto the word after the segmental content is specified. Accent is, in this sense, more suprasegmental than is tone. 25. ‘Accent language’ here includes both stress languages and pitch-accent languages.

136

Ellen van Zanten and Philomena Dol

Accent is perceived as something extra added to the segmental specification of the words; it makes some parts of the utterance more prominent than others. Beckman explains this di¤erence in attitudes on the part of native speakers in terms of their di¤erent functions. ‘‘Tone is di‰cult to view as distinct from the segmental specification of a word because its salient function is identical to that of the segmental phonemes; both specify the morphemic content of an utterance’’ (Beckman 1986: 31). As ‘‘anecdotal evidence’’, Beckman (1986: 28, 30) compares the attitudes of a speaker of Burmese, a tone language, and a speaker of Japanese, a pitch-accent language. When the Burmese informant was asked how she categorized the tonal contrasts in a (Burmese) minimal pair she said that she perceived the words as having di¤erent vowels. The Japanese speaker, on the other hand, when asked to pronounce a (Japanese) word he had never heard before, applied a reasoning to find out the accent position. According to Beckman, it is di‰cult to imagine such reasoning for a native speaker of a tone language when trying to decipher the tones of an unfamiliar written word. If native speakers of (restricted) tone languages and pitch-accent languages indeed di¤er in their view on the pitch contrasts in their own language, it should be possible to determine to which category a language belongs by collecting native speakers’ judgments on this issue. Neurolinguistic approach. Finally, we briefly touch upon the subject of linguistic processing in the brain, to see if this may help clarify the issue. We take a quick look at three neurolinguistic reports involving speakers of the tone languages Chinese and Thai. Neuropsychological data from Standard Chinese suggest that segmental structure and tone are separate functions with separate locations in the brain (Liang & van Heuven 2004). If phonemes and tones are indeed processed in di¤erent parts of the brain, this might put doubt on Beckman’s view that segments and tones are closely connected in tone languages. Beckman’s view is, however, supported by Gandour et al. (2000) who found evidence that for Thai listeners, the same area of the brain is involved in processing segments and phonological tones alike. Gandour et al. (2000: 213) state that ‘‘tonal information conveyed on a vowel and the vowel information itself are unlikely to be processed fully independently.’’ Both Gandour et al. (2000) and Hsieh et al. (2001) suggest that linguistic processing may be language-specific. It is possible, then, that in some languages, like Thai, processing of segments and tones are connected more closely than in other languages, like Chinese. In Chinese, the so-called Neutral Tone and the Low Tone are subject to tone sandhi; each of these tones has three very

Word stress and pitch accent in Papuan languages

137

di¤erent context-dependent realizations (Liang & van Heuven 2004). So, part of the Chinese lexicon has no fixed relation between segments and tone. This may have caused the connection between segments and tones for Chinese speakers to be less close than Beckman hypothesizes for speakers of tone languages in general. If linguistic processing is indeed language-specific, it is as yet di‰cult to see how neurolinguistic research can help us distinguish between tone languages on the one hand, and pitch-accent languages on the other.

6. Conclusion This survey was based on descriptions of 74 (a mere 10%) of the approximately 750 Papuan languages. Of these 74 sample languages, fifty-one are stress languages (60%), and twelve more have word stress as well as lexical tone – the so-called hybrid languages (16%). This high proportion of stress languages may be partly due to a bias in our material. Our first aim was to trace and categorize Papuan word-stress languages. This search was, obviously, facilitated by the StressTyp database, which only contains languages that feature stress. On the other hand, a closer look at the Bird’s Head languages suggests that the bias towards stress languages may not be as strong as it seems to be. We have prosodic information on eight of the twenty languages spoken in the Bird’s Head. Of these eight, four are pure stress languages26 and one is a hybrid, whereas only two are pure tone languages, and one has neither a word-based stress nor lexical tone. So, at least in the Bird’s Head, stress languages seem to be the largest category. Also, our percentage for Papuan stress languages and hybrids taken together (60% þ 16% ¼ 76%) does not deviate much from the estimated percentage of stress languages in the world: 80% (Goedemans this volume). The twelve hybrid languages in this survey are mainly found in three areas: the Central Lakes Plains area, and the Central and Eastern Highlands. They show an enormous variety in tone-stress relations as well as 26. This includes Meah (Meyah). Based on Gravelle & Gravelle (1991) StressTyp classifies Meah as a stress language (code LEX). More recently, Meah is seen as a pitch-accent language by Gravelle (2000, 2002), which, incidentally, would produce the code LEX (pitch). The examples in Gravelle (2002) remind us of ‘multiple stress’ systems as discussed for Wa¤a in Section 5. Like Wa¤a, Meah may turn out to be a tone language according to our definition.

138

Ellen van Zanten and Philomena Dol

in stress rules, even within families. Obviously, more research may bring to light tendencies which are as yet undiscovered because of the limited amount of data on which this survey is based. Equally obvious are the gaps in our knowledge of Papuan wordprosodic systems in general. Many languages in the area have not yet been described, or, if they have been, the sections on (word) prosody are often short and rather vague. More in-the-field research is necessary to get a complete picture of the richness in Papuan word prosody. Such research should include acoustic measurements designed to reveal specific phonetic correlates for the various prosodic categories, as well as perceptual experiments, in which native speakers’ judgements on these categories are investigated. The vagueness in many prosodic descriptions is partly due to the existing terminological di¤useness. Bringing more clarity and consistency into the terminology should be one other priority of prosodic research in Papua as well as in other areas of the world. Clearly defined terms will undoubtedly enhance typological word-prosodic research. In the meantime, the reader will have to make do with this survey. We hope that, despite its shortcomings, the survey may be useful for future research on the prosody of Papuan languages.

References (*: not in text, only in statistics) Abbott, Stan 1985

A tentative multilevel multiunit phonological analysis of the Murik language. In Papers in New Guinea Linguistics 22 (Pacific Linguistics A-63): 339–373.

Bailey, David A. 1975 The phonology of the Abau language. In Workpapers in Papua New Guinea Languages 9: 5–58. Bateman, Janet 1986 Iau verb morphology. NUSA 26. Jakarta: Atma Jaya. Bateman, Janet 1990 Iau segmental and tone phonology. In B. Kaswanti Purwo (ed.) Miscellaneous Studies of Indonesian and other Languages in Indonesia X (NUSA 32): 29–42. Jakarta: Atma Jaya. Beckman, Mary E. 1986 Stress and non-stress accent. Dordrecht/Riverton: Foris Publications. Bee, Darlene and Kathleen Barker Glasgow 1973 Usarufa tone and segmental phonemes. In H. McKaughan (ed.) The languages of the eastern family of the East New Guinea High-

Word stress and pitch accent in Papuan languages

139

land Stock: 190–203. Washington: University of Washington Press. Berinstein, Ava E. 1979 A cross-linguistic study on the perception and production of stress. University of California at Los Angeles Working Papers in Phonetics 47. Berry, Keith and Christine Berry 1999 A description of Abun: a West Papuan language of Irian Jaya. Pacific Linguistics B-115. Blowers, Bruce L. 1970 Kaugel phonemic statement. Pacific Linguistics A-26: 1–12. *Boxwell, Helen and Maurice Boxwell 1966 Weri phonemes. In S.A. Wurm (ed.) Papers in New Guinea Linguistics 5: 77–93. *Bromley, H. Myron 1980 A Grammar of Lower Grand Valley Dani. Pacific Linguistics C-63. *Brown, William U. 1991 A quantitative phonology of Mai Brat. Pacific Linguistics A-73: 1–27. Bruce, Les 1984 The Alamblak language of Papua New Guinea (East Sepik). Pacific Linguistics C-81. Bunn, Gordon and Ruth Bunn 1970 Golin phonology. Pacific Linguistics A-23: 1–7. Collier, K. and M. Collier 1986 Tabla Phonology. In Workpapers in Indonesian Languages and Cultures 3: 1–16. Jayapura: Cenderawasih University & Summer Institute of Linguistics. Conrad, Robert J. and Kepas Wogiga 1991 An outline of Bukiyip grammar. Pacific Linguistics C-113. *Daniel, R. and K.A. Shaw 1977 Samo phonemes, description, interpretation and resulting orthography. In Workpapers in Papua New Guinea Languages 19: 5–28. *Davies, H.J. 1980 Kobon phonology. Pacific Linguistics B-68. Doble, M. 1987 A description of some features of Ekari language structure. Oceanic Linguistics 26: 55–113. *Dol, Philomena 1999 A grammar of Maybrat, a language of the Bird’s Head, Irian Jaya, Indonesia. PhD dissertation Leiden University. Dommel Peter R. and Gudrun E. Dommel 1991 Kaure phonology. In Workpapers in Indonesian Languages and Cultures 9: 1–68. Jayapura: Cenderawasih University & Summer Institute of Linguistics.

140

Ellen van Zanten and Philomena Dol

Donohue, Mark 1997 Tone systems in New Guinea. Linguistic Typology 1: 347–386. Drabbe, P. 1952 Spraakkunst van het Ekagi [Ekagi grammar]. ’s-Gravenhage: M. Nijho¤. *Drabbe, P. 1955 Spraakkunst van het Marind [Marind grammar]. Studia Instituti Anthropos 11. Drew, D. and A. Payne [n.d.] Kamano stress. [unpublished manuscript of the Summer Institute of Linguistics, Ukarumpa]. Eefting, W.Z.F. 1991 The e¤ect of information value and accentuation on the duration of Dutch words, syllables and segments. Journal of the Acoustic Society of America 89: 412–414. Elenbaas, Nine 1999 A unified account of binary and ternary stress; considerations from Sentani and Finnish. PhD dissertation Utrecht University. Utrecht: Netherlands Graduate School of Linguistics (LOT) 20. *Enk, Gerrit J. van and Lourens de Vries 1997 The Korowai of Irian Jaya; their language in its cultural context. Oxford/New York: Oxford University Press. *Farr, James and Cynthia Farr 1975 Some features of Korafe phonology. In T.E. Dutton (ed.) Studies in languages of Central and South-East Papua (Pacific Linguistics C-29): 731–769. Feldman, Harry 1986 A grammar of Awtuw. Pacific Linguistics B-94. *Flassy, Don A.L. and W.A.L. Stokhof 1979 A note on Tehit (Bird’s Head, Irian Jaya). NUSA 7: 35–83. Foley, William A. 1986 The Papuan languages of New Guinea. Cambridge: Cambridge University Press. Foley, William A. 1991 The Yimas language of New Guinea. Stanford: Stanford University Press. Ford, Kevin 1993 A preliminary comparison of Kamano-Yagaria. Language and Linguistics in Melanesia 24: 191–202. Fortgens, J. 1928 Grammatikale aanteekeningen van het Tabaroesch; Tabaroesche volksverhalen en raadsels [Grammatical notes on Tabaru; Tabaru stories and riddles]. Bijdragen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde [Contributions of the Royal Institute] 84: 300–544.

Word stress and pitch accent in Papuan languages

141

Franklin, Karl J. 1967 Kewa sentence structure. Papers in New Guinea Linguistics 7, 27–59. (Pacific Linguistics A-13). Franklin, Karl J. 1971 A grammar of Kewa, New Guinea. Pacific Linguistics C-16. Franklin, Karl J. and Joice Franklin 1962 Kewa I: phonological asymmetry. Anthropological Linguistics 4/7: 29–37. Franklin, Karl J. and Joice Franklin 1978 A Kewa dictionary, with supplementary grammatical and anthropological materials. Pacific Linguistics C-53. Gandour, J.T. 1978 The perception of tone. In Victoria A. Fromkin (ed.) Tone; a linguistic survey, 41–76. New York: Academic Press. Gandour, J., D. Wong, L. Hsieh, B. Weinzapfel, D. Van Lancker and G.D. Hutchins 2000 A crosslinguistic PET study of tone perception. Journal of Cognitive Neuroscience 12-1: 207–222. *Goddard, Jean 1980 Notes on Agarabi grammar. Pacific Linguistics A-56: 35–76. Gravelle, Gilles 2000 Meyah. In Ger P. Reesink (ed.) Studies in Irian languages Part II (NUSA 47), 71–104. Gravelle, Gilles 2002 Morphosyntactic properties of Meyah word classes. In G.P. Reesink (ed.) Languages of the Eastern Bird’s Head. Pacific Linguistics 524: 109–180. Gravelle, G. and G. Gravelle 1991 Meah Phonology. In Workpapers in Indonesian Languages and Cultures 10: 1–36. Jayapura: Cenderawasih University & Summer Institute of Linguistics. Haiman, John 1980 Hua, a Papuan language of the Eastern Highlands of New Guinea. Amsterdam: Benjamins. Healey, Alan 1964 Telefol phonology. Pacific Linguistics B-3. Henderson, James 1995 Phonology and grammar of Yele, Papua New Guinea. Pacific Linguistics B-112. Hendriks, Bernadet 1995 Word-prosodic systems in the languages of the East New Guinea Highlands Stock in the Highlands of Papua New Guinea. [unpublished MA thesis Leiden University]. Hendriks, Bernadet 1996 The accentual systems of some Papuan languages. In Rob Goedemans, Harry van der Hulst and Ellis Visch (eds) Stress

142

Ellen van Zanten and Philomena Dol patterns of the world. Part 1: Background. (HIL Publications 2): 201–232. The Hague: Holland Academic Graphics.

*Hesse, R.G. 1995

Syllable structure in Imyan Tehit. Language and Linguistics in Melanesia 26: 101–171. Hsieh, L., J. Gandour, D. Wong and G.D. Hutchins 2001 Functional heterogeneity of inferior frontal gyrus is shaped by linguistic experience. Brain and Language 76: 227–252. Hualde, J.I., G. Elordieta, I. Gaminde and R. Smiljanic 2002 From pitch accent to stress accent in Basque. In C. Gussenhoven and N. Warner (eds) Laboratory Phonology 7: 547–584. Berlin/ New York: Mouton de Gruyter. Hueting, A. 1936 Iets over de spraakkunst van de Tobeloreesche taal [A little on Tobelo grammar]. Bijdragen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde [Contributions of the Royal Institute] 94: 295–325. Hulst, Harry van der 1999 Word accent. In Harry van der Hulst (ed.) Word prosodic systems in the languages of Europe. (Empirical approaches to language typology; Eurotyp 20-4): 3–115. Berlin/New York: Mouton de Gruyter. Hulst, Harry van der and Norval Smith (eds) 1988 Autosegmental studies on pitch accent. Dordrecht: Foris publications. James, D.J. 1963 Kamano phonology. [unpublished manuscript at the Summer Institute of Linguistics, Ukarumpa]. Kenman, Leon F. 1975 The phonetics of Standard Russian unstressed vowels together with a critical survey of phonological interpretations of AkanjeIkanje. PhD Dissertation University of Texas at Austin. Ladd, D. Robert 1996 Intonational phonology. Cambridge: Cambridge University Press. *Larsen, R. and M. Larsen 1977 Orokaiva phonology and orthography. In Workpapers in Papua New Guinea Languages 19: 5–29. Laycock, D.C. 1965a Three Upper Sepik phonologies [Abau, Iwam and Wagamusin]. Oceanic Linguistics 4-I & II: 113–118. Laycock, D.C. 1965b The Ndu language family. Pacific Linguistics C-1. Liang, J. and V.J. van Heuven 2004 Evidence for separate tonal and segmental tiers in the lexical specification of words: A case study of a brain-damaged Chinese speaker. Brain and Language 91: 282–293.

Word stress and pitch accent in Papuan languages Lindstro¨m, E. 2002

Loving, R. 1973

143

Topics in the grammar of Kuot, a non-Austronesian language of New Ireland, Papua New Guinea. PhD dissertation Stockholm University.

Awa phonemes, tonemes, and tonally di¤erentiated allomorphs. In H. McKaughan (ed.) The languages of the eastern family of the East New Guinea Highland Stock: 10–18. Washington: University of Washington Press. Luzbetak, Louis J. 1956 Middle Wahgi phonology. Oceania Linguistic Monographs 2. Manning, M. and N. Jaggers 1977 A tentative phonemic analysis of Ningil. In R. Loving (ed.) Phonologies of Five Papua New Guinea languages: 49–72. Ukarumpa: Summer Institute of Linguistics. Martens, M. and S. Tuonimen 1977 A tentative phonemic statement of Yil in West Sepik district. In R. Loving (ed.) Phonologies of Five Papua New Guinea languages: 29–48. Ukarumpa: Summer Institute of Linguistics. Martin, D.L. 1991 Sikaritai phonology. In Workpapers in Indonesian Languages and Cultures 9: 91–120. Jayapura: Cenderawasih University & Summer Institute of Linguistics. McAllister, Lawrence and Kay McAllister 1991 The process of phonological change in Doutai. In Workpapers in Indonesian Languages and Cultures 9: 121–142. Jayapura: Cenderawasih University & Summer Institute of Linguistics. McCawley, James D. 1978 What is a tone language? In Victoria A. Fromkin (ed.) Tone: a linguistic survey: 113–131. New York: Academic Press. *McElhanon, K.A. 1970 Selepet phonology. Pacific Linguistics B-14. Nicholson, R. and R. Nicholson 1962 Fore phonemes and their interpretation. Oceania Linguistic Monographs 6. Nooteboom, S.G. 1972 Production and perception of vowel duration; a study of durational properties of vowels in Dutch. PhD dissertation Utrecht University. Oatridge, D. and J. Oatridge 1973 Phonemes of Binumarien. In H. McKaughan (ed.) The languages of the eastern family of the East New Guinea Highland Stock: 517–522. Washington: University of Washington Press. Ode´, Cecilia 2002 Mpur prosody: An experimental-phonetic analysis with examples from two versions of the Fentora myth. Osaka: Osaka Gakuin

144

Ellen van Zanten and Philomena Dol University. Endangered Languages of the Pacific Rim Publications Series A1-003.

Paster, Mary 2003 Floating tones in Ga˜. Studies in African Linguistics 32.1: 17–39. Pawley, Andrew 1998 The Trans New Guinea Phylum hypothesis: a reassessment. In J. Miedema, C. Ode´ and R. Dam (eds) Perspectives on the Bird’s Head of Irian Jaya, Indonesia. Proceedings of the conference at Leiden, 13–17 October 1997: 655–690. Amsterdam/Atlanta: Rodopi. Peckham, L. 1991 Mairasi phonology. In Workpapers in Indonesian Languages and Cultures 10: 111–145. Jayapura: Cenderawasih University & Summer Institute of Linguistics. Phillips, Donald J. 1976 Wahgi phonology and morphology. Pacific Linguistics B-36. *Phinnemore, Thomas R. 1985 Ono phonology and morphophonemics. Pacific Linguistics A-63: 173–214. Pike, Eunice V. 1974 A multiple stress system versus a tone system. International Journal of American Linguistics 40-3: 169–175. *Reesink, Ger P. 1987 Structures and their functions in Usan, a Papuan language of Papua New Guinea. Amsterdam/Philadelphia: John Benjamins. Reesink, Ger P. 1998 The Bird’s Head as Sprachbund. In Jelle Miedema, Cecilia Ode´ and Rien A.C. Dam (eds) Perspectives on the Bird’s Head of Irian Jaya, Indonesia. Proceedings of the conference at Leiden, 13–17 October 1997: 603–642. Amsterdam/Atlanta: Rodopi. Reesink, Ger P. 1999 A grammar of Hatam. Pacific Linguistics C-146. Reesink, Ger P. 2000 Western Papuan languages: roots and development. [unpublished paper presented at the Conference on Papuan Pasts, Canberra, November 27–30, 2000]. Reesink, Ger P. 2002 A grammar sketch of Sougb. In G.P. Reesink (ed.) Languages of the Eastern Bird’s Head. Pacific Linguistics 524: 181–275. Remijsen, Bert 2001 Word-prosodic systems of Raja Ampat languages. PhD dissertation Leiden University. Netherlands Graduate School of Linguistics (LOT) 49. Remijsen, Bert 2002 Lexically contrastive stress and lexical tone in Ma¨ya. In C.

Word stress and pitch accent in Papuan languages

145

Gussenhoven and N. Warner (eds) Laboratory Phonology 7: 585–614. Berlin/New York: Mouton de Gruyter. Renck, G.L. 1967

A tentative statement of the phonemes of Yagaria. In Papers in New Guinea Linguistics 6 (Pacific Linguistics A-12): 19–48.

Renck, G.L. 1975 A grammar of Yagaria. Pacific Linguistics B-40. Roberts, J.R. 1987 Amele. Guildford (England): Biddles. Ross, Malcolm 1980 Some elements of Vanimo, a New Guinea tone language. Pacific Linguistics A-56: 77–109. Ross, Malcolm 2005 Pronouns as a preliminary diagnostic for grouping Papuan languages. In A. Pawley, R. Attenborough, R. Hide and J. Golson (eds) Papuan pasts: Investigations into the cultural, linguistic and biological history of the Papuan speaking peoples. *Rule, M. 1964 Customs, alphabet and grammar of the Kaluli people of Bosavi, Papua. [unpublished manuscript]. Sanders, Arden G. and Joy Sanders 1980 Phonology of the Kamasau language. Pacific Linguistics A-56: 111–135. Schadeberg, Thilo C. 1973 Kinga: a restricted tone system. Studies in African Linguistics IV1: 23–47. *Schie¤elin, Bambi B. and Steven Feld 1998 Bosavi-English-Tok Pisin dictionary. Pacific Linguistics C-153. Scorza, David 1985 A sketch of Au morphology and syntax. In Papers in New Guinea Linguistics 22 (Pacific Linguistics A-63): 215–273. Scott, D. and P.B. Jenison 1991 Obokuitai phonology. In Workpapers in Indonesian Languages and Cultures 9: 69–90. Jayapura: Cenderawasih University & Summer Institute of Linguistics. Scott, Graham 1978 The Fore language of Papua New Guinea. Pacific Linguistics B-47. Shelden, H. 1989 Galela phonemes and stress. In W.D. Laidig (ed.) Workpapers in Indonesian Languages and Cultures 7: 77–88. Ambon: Pattimura University & Summer Institute of Linguistics. *Sims, Andrew and Anne Sims 1982 Ketengban phonology. In Workpapers in Indonesian linguistics 1: 50–74. Jayapura: Cenderawasih University & Summer Institute of Linguistics.

146

Ellen van Zanten and Philomena Dol

Sluijter, A.M.C. and V.J. van Heuven 1996 Spectral balance as an acoustic correlate of linguistic stress. Journal of the Acoustic Society of America 100: 2471–2485. Staalsen, Philip 1966 The phonemes of Iatmul. Pacific Linguistics A-7: 69–76. Staden, Miriam van 2000 Tidore, a linguistic description of a language of the North Moluccas. PhD dissertation Leiden University. Steltenpool, J. and P.A.M. van der Stap [n.d] Leerboek van het Kapauku [Kapauke textbook]. [n.p.] *Stokhof, W.A.L. 1979 Woisika II. Phonemics. Pacific Linguistics B-59. Stringer, Mary and Joyce Hotz 1973 Wa¤a phonemes. In H. McKaughan (ed.) The languages of the eastern family of the East New Guinea Highland Stock, 523–529. Washington: University of Washington Press. Terrill, Angela 1999 A grammar of Lavukaleve: a Papuan language of the Solomon Islands. PhD dissertation Australian National University. *Thompson, N.P. 1975 Magi phonology and grammar fifty years afterwards. In T.E. Dutton (ed.) Studies in languages of Central and South-East Papua (Pacific Linguistics C-29): 599–666. Visser, L.E. and C.L. Voorhoeve 1987 Sahu-Indonesian-English dictionary and Sahu grammar sketch. Verhandelingen van het Koninklijk Instituut voor Taal-, Landen Volkenkunde 126. Dordrecht: Foris. *Voorhoeve, C.L. 1965 The Flamingo Bay dialect of the Asmat Language. Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 46. The Hague: M. Nijho¤. Voorhoeve, C.L. 1981 Some comparative notes on the Non-Austronesian languages of North Halmahera [unpublished paper presented at the Seminar Halmahera dan Raja Ampat, Jakarta, 1–5 Juni 1981]. Voorhoeve, Jan 1973 Safwa as a restricted tone system. Studies in African Linguistics IV-1: 1–21. *Vries, Lourens de 2002 An introduction to the Inanwatan language of Irian Jaya. In K.A. Adelaar and R. Blust (eds) Between worlds; Linguistic papers in memory of David John Prentice. (Pacific Linguistics 529): 77–93. *Vries, Lourens de and Robinia de Vries-Wiersma 1992 The morphology of Wambon. Verhandelingen van het Koninklijk

Word stress and pitch accent in Papuan languages

147

Instituut voor Taal- Land- en Volkenkunde 151. Leiden: KITLV Press. *Wells, Margaret A. 1979 Siroi grammar. Pacific Linguistics B-51. Wimbish, Sandra G. 1992 Pagu phonology. In Donald A. Burquest and Wyn D. Laidig (eds) Descriptive studies in languages of Maluku (NUSA 34): 69–90. Jakarta: Atma Jaya University. Wurm, S.A. 1982 Papuan Languages of Oceania. Ars Linguistica 7. Tu¨bingen: Gu¨nter Narr.

U (tone)

LEX

P/A % L/F

F/F

U;P (IRR)

F/F

Abun

*Agarabi

Alamblak

Amele

*Asmat

Au

3

2

Nr of tones tone l. and hybrids F: pitch accent

Alternative analysis D: Donohue (1997) F: Foley (1986) H: Hendriks (1996)

Torricelli Family (Wurm: Wapei Family), Torricelli Phylum

Asmat Family, TNGP

Gum (or: Abaian) Family, TNGP

Alamblak Family, Sepik-Ramu Phylum

Eastern or: Kainantu Family, TNGP

Family-level isolate, West Papuan Phylum

Ndu Family, Sepik-Ramu Phylum

Family-level isolate, Sepik-Ramu Phylum

A‰liation27 TNPG: Trans New Guinea Phylum

Torricelli Ranges, eastward into Madang Province, PNG

South PP

Madang Province, PNG

Sepik Hill area, PNG

East New Guinea Highlands, PNG

North-West Bird’s Head, PP

Sepik River area, PNG

Upper Sepik area, PNG

Region PNG: Papua New Guinea PP: Papua Province, Indonesia

27. Phylum-level a‰liations are based on Wurm (1982). Family-level a‰liations are based on Foley (1986), Wurm (1982), and, for the languages of the West-Papuan Phylum, on Reesink (2000).

I/I

Code ‘pure’ stress languages and hybrids

Abelam

Abau

Language italics: not in StressTyp *: not in text, only in statistics

Appendix: Word-prosodic categories

148 Ellen van Zanten and Philomena Dol

P

U

I (tone)

S/- (LEX; pitch)

F/F (tone)

P;A

*Dani

Doutai

Ekari (or: Ekagi, Kapauku)

Fore

Galela

P

P;I

Bukiyip (or: Mountain Arapesh)

Brat see Maybrat

*Bosavi

Binumarien

Baburiwa see Obokutai

Awtuw (or: Autu)

Awa

2

2 or 3

4

2

4

F, H, Scott (1978): pitch accent

F: pitch accent or tone

North Halmahera Family, West Papuan Phylum

East-Central or: Gorokan Family, TNGP

Wissel Lakes Family, TNGP

Central Lakes Plain Family, TNGP

Dani Family, TNGP

Torricelli Family (Wurm: Arapesh Family), Torricelli Phylum

Bosavi Family, TNGP

Eastern or: Kainantu Family, TNGP

Ram Family, Sepik-Ramu Phylum

Eastern or: Kainantu Family, TNGP

North Moluccas (Halmahera Island), Indonesia

East New Guinea Highlands, PNG

W-C Highlands, around Wissel Lakes, PP

Lakes Plain area, PP

Baliem Valley, PP

Torricelli Ranges, eastward into Madang Province, PNG

S Highlands Province, PNG

East New Guinea Highlands, PNG

Sepik River area, PNG

East New Guinea Highlands, PNG

Word stress and pitch accent in Papuan languages

149

I/I

Kamasau

Kaugel

LEX

LEX

Kamano

Kapauku see Ekari

I

*Iwam

Imyan Tehit see Tehit

Iatmul

Ford (1993): tone

LEX

*Inanwatan

Central or: Chimbu Family, TNGP

Torricelli Family (Wurm: Marienberg Family), Torricelli Phylum

East-Central or: Gorokan Family, TNGP

Iwam Family, Sepik-Ramu Phylum

Ndu Family, Sepik-Ramu Phylum

Inanwatan Family, TNGP

Turu Family, TNGP

U (tone)

Iau

8

East-Central or: Gorokan Family, TNGP

Ford (1993): tone

I

Central or: Chimbu Family, TNGP

Hu(v)a dialect of Yagaria

2

Hatam-Mansim group, West Papuan Phylum

L/L (tone)

Hatam

Golin

Chimbu and Western Highlands Provinces, PNG

Torricelli Ranges, eastward into Madang Province, PNG

East New Guinea Highlands, PNG

Upper Sepik area, PNG

Sepik River area, PNG

South Bird’s Head, PP

Western Lakes Plains, PP

East New Guinea Highlands, PNG

Eastern Bird’s Head, PP

Chimbu and Western Highlands Provinces, PNG

150 Ellen van Zanten and Philomena Dol

LEX

LEX

P

LEX

LEX

LEX

I;LEX

*Ketengban

Kewa

*Kobon

*Korafe

*Korowai

Kuot

Lavukaleve

P

U/P (tone)

U;P

I

*Magi

Mairasi

*Marind

*Maybrat (or: Mai Brat)

Lower Grand Valley Dani see Dani

L/LEX (tone)

Kaure

2

3

D: pitch accent

F: pitch accent D: tone

D: pitch accent?

D: pitch accent and stress

Family-level isolate, West Papuan Phylum

Marind Family, TNGP

Mairasi Family, TNGP

Mailuan Family, TNGP

Central Solomons Family, East Papuan Phylum

Family-level isolate, East Papuan Phylum

Family-level isolate, TNGP

Binanderean Family, TNGP

Kalam Family, TNGP

West-Central or: Engan Family, TNGP

Mek (or: Goliath) Family, TNGP

Kaure Family, TNGP

Central Bird’s Head, PP

South-East PP

Bomberai Peninsula, PP

South-East PNG

Russell Island, PNG

New Ireland, PNG

South of Central Ranges, South-East PP

North coast of South-east PNG

Central Highlands, PNG

Enga Province and adjacent areas, PNG

Eastern Highlands, PP

Lakes Plain area, PP

Word stress and pitch accent in Papuan languages

151

F/LEX (tone)

Move dialect of Yagaria

I

I

P

A;IRR

LEX

*Ono

*Orokaiva

Pagu

Sahu (Palisua or: Pa’disua dialect)

*Samo

East Strickland Family, TNGP

North Halmahera Family, West Papuan Phylum

North Halmahera Family, West Papuan Phylum

Binanderean Family, TNGP

Huon Family, TNGP

Central Lakes Plain Family, TNGP

L/L (tone)

Obokuitai (or: Baburiwa)

2

Torricelli Family (Wurm: Wapei Family), Torricelli Phylum

I (NMS)

Lower Sepik Family (Wurm: Nor Family), Sepik Ramu Phylum

Family-level isolate, West Papuan Phylum

East-Central or: Gorokan Family, TNGP

Ningil

Ford (1993): tone

East Bird’s Head Family, West Papuan Phylum

F/F

4

3

Gravelle (2000, 2002): pitch accent

Murik (or: Nor)

Mpur

LEX

Meah (or: Meyah)

Western Province, PNG

North Moluccas (Halmahera Island), Indonesia

North Moluccas (Halmahera Island), Indonesia

North coast of SouthEast PNG

Huon Peninsula, PNG

Lakes Plain area, PP

Torricelli Ranges, eastward into Madang Province, PNG

Coast of East Sepik Province, PNG

North-Central Bird’s Head, PP

East New Guinea Highlands, PNG

East Bird’s Head, PP

152 Ellen van Zanten and Philomena Dol

U/P

L/F (tone)

S

Sentani

Sikaritai

*Siroi

P/A

LEX

Tabla (or: Tanah Merah)

*Tehit (or: Imyan Tehit dialect)

U/P

U/P

LEX

F/L/L (tone)

Tidore

Tobelo

*Usan

Usarufa

Telefol

P

Tabaru

Sougb

I

*Selepet

2 or 3

2

2

2

H: stress and pitch accent

F: pitch accent

Reesink (2002): pitch accent

Eastern or: Kainantu Family, TNGP

Numugenan Family, TNGP

North Halmahera Family, West Papuan Phylum

North Halmahera Family, West Papuan Phylum

Ok Family, TNGP

West Bird’s Head Family, West Papuan Phylum

Sentani family, TNGP

North Halmahera Family, West Papuan Phylum

East Bird’s Head Family, West Papuan Phylum

Kabenau Family, TNGP

Central Lakes Plain Family, TNGP

Sentani Family, TNGP

Huon Family, TNGP

East New Guinea Highlands, PNG

Madang Province, PNG

North Moluccas and Raja Ampat Islands, Indonesia

North Moluccas (Halmahera Island), Indonesia

Central mountainous New Guinea, PNG and PP

South-west Bird’s Head, PP

North-East coast of PP

North Moluccas (Halmahera Island), Indonesia

East Bird’s Head, PP

Madang Province, PNG

Lakes Plain area, PP

North-east coast of PP

Huon peninsula, PNG

Word stress and pitch accent in Papuan languages

153

U;P;LEX (tone)

P;U (IRR)

U

IRR

Wahgi

*Wambon

*Weri

*Woisika

I

I/I

I

Yele

Yil

Yimas

Yagaria see Hua and Move

U 2 or 3

2

Wa¤a

*Wagamusin

3

Vanimo H: pitch accent Stringer & Hotz: multiple stress

Lower Sepik Family (Wurm: Pondo F.), Sepik-Ramu Phylum

Torricelli Family (Wurm: Wapei Family), Torricelli Phylum

Family-level isolate, East Papuan Phylum

Makasai-Alor-Pantar Family, West Papuan Phylum

Goilalan Family, TNGP

Awyu Family, TNGP

Central or: Chimbu Family, TNGP

Wagamusin Family, Sepik-River Phylum

Eastern or: Kainantu Family, TNGP

Sko Family, Sko Phylum(-level Stock)

Coast of East Sepik Province, PNG

Torricelli Ranges, eastward into Madang Province, PNG

Rossel Island, PNG

Alor Island, Eastern Indonesia

Northern Central Province and adjacent areas, PNG

Digul river, Southern PP

Chimbu and Western Highlands Provinces, PNG

Upper Sepik area, PNG

East New Guinea Highlands, PNG

North West coast of PNG

154 Ellen van Zanten and Philomena Dol

5.

Accent in the native languages of North America

Keren Rice North America is a vast continent, and a large number of languages were once spoken across the continent. In a recent classification of native languages of North America, Goddard (1996) proposes sixty-two families and isolates in what he calls a consensus classification; see Campbell (1997) and Mithun (1999) for other recent classifications. In addition, it has been proposed that there are several linguistic areas in North America; see Campbell (1997: 330–344) and Mithun (1999: 314–321) for recent discussion. North American languages exhibit an array of accent systems, adopting the terminology of van der Hulst (this volume, chapter 1) where ‘accent’ represents an abstract property that can have various phonetic manifestations. Given the linguistic diversity of the continent, one can ask if accent, or its phonetic manifestations (pitch, duration, loudness), might be features that are useful in supporting proposed genetic relationships among languages and/or whether they might be areal features. For one, accent patterns might commonly be stable within a language family, and used to help establish genetic relationships. For another, accent patterns might be easily transmitted, making accent and its phonetic realizations potential features that identify linguistic areas (see, for instance, Thomason and Kaufman 1988: 16, 62, 75, 90, 122, 211). The survey in this article suggests the need for additional research to understand the answers to these questions, but accent patterns often vary across a family in systematic ways, and there is little immediate evidence that accent patterns can be regarded as an areal trait in North America. The article is organized as follows. After some brief background presentation, I survey the accent patterns of the native languages of North America. I then focus on the families and geographic linguistic areas in North America to investigate if accent patterns constitute a feature that can be used to define these groupings. In the following sections, I examine interfaces that involve accent, namely its interaction with morphology and its phonetic manifestations. A hallmark of many native languages of North America is morphological complexity, and I examine the role that morphology plays in accent systems of the languages, focusing on whether

156

Keren Rice

accent systems that have a morphological basis are more stable than those which are purely phonologically defined. Finally, the phonetic manifestations of accent in North American languages are varied, with it often realized as a combination of pitch, duration, and amplitude, and it can also be manifested as simply duration or, frequently, simply pitch. The major goal of the article is to survey accent patterns and their manifestations. I accept analyses given in StressTyp and use this framework for languages not included in the database; I do not propose new theoretical alternatives.

1. A hint of diversity Native languages of North America exhibit a range of accent patterns. Even within a family or linguistic area, variety exists. Before beginning the language survey, I present a table with a few of the types of accent patterns found across the continent, organized in terms of the position of primary accent, however it is realized, and focusing on di¤erences within a family. The designation under the column ‘primary’ follows StressTyp categories for the position of primary accent as discussed in chapter 1 of this volume. I summarize below the major abbreviations used in the current chapter for ease of the reader: I P T A P U I/I I/S U/U U/P

Primary stress on the initial syllable. Primary stress on the second syllable. Primary stress on the third syllable. Primary stress on the antepenultimate syllable. Primary stress on the penultimate syllable. Primary stress on the final syllable. Stress the initial syllable if heavy; otherwise stress the second syllable if heavy; if neither is heavy, stress the first syllable. Stress the initial syllable if heavy; otherwise stress the second syllable. Stress the ultimate syllable if heavy; otherwise stress the penultimate if heavy; if neither is heavy stress the ultimate. Stress the ultimate syllable if heavy; otherwise stress the penultimate.

Accent in the native languages of North America

157

P/U

Stress the penultimate if heavy; otherwise stress the ultimate.

F/F

Stress the first heavy syllable in the word; if no heavy syllable, stress the first syllable.

F/L

Stress the first heavy syllable in the word; if no heavy syllable, stress the last syllable. Stress the last heavy syllable in the word; if no heavy syllable, stress the last syllable. No main stress. All stresses are equally prominent. Primary stress is assigned to the head of the last foot; stress is assigned from left-to-right.

L/L NMS L(CNT)

(1) Some accent patterns in North American languages: primary accent primary

example

I

Cahuilla (Uto-Aztecan)

I/S

Hopi (Uto-Aztecan)

S

Chemehuevi (Uto-Aztecan)

P

Nahuatl (Uto-Aztecan)

I

Southeastern Pomo (Pomoan)

P

Southern Pomo (Pomoan)

I/S

Makah (Wakashan)

I/I

Nootka [Nuu-Chah-Nulth] (Wakashan)

F/F

Kwak’wala (Wakashan)

P

Mohawk (Iroquoian)

L (CNT)

Cayuga (Iroquoian)

As (1) illustrates, within a family accent can be initial (I), initial if heavy (I/S), or penultimate (P), as in Uto-Aztecan languages; it can be restricted to initial (I) or penultimate (P), as in Pomoan languages; it can on the first heavy syllable but di¤er in terms of foot size and default stress, as in the Wakashan examples; it may di¤er in terms of quantity sensitivity, as in the Iroquoian languages.

158

Keren Rice

2. Some caveats Before beginning the language survey, a few cautions are in order. First while I use the term language after Goddard (1996), many of the languages have several dialects, and these often di¤er in ways that are not always well reported in the literature. See, for instance, Valentine (1994) on Ojibwa (Algonquian family) dialects for the kind of diversity that exists within what is usually classified as a single language. Second, many North American languages are known under a variety of names. Even three recent classifications, Campbell (1997), Goddard (1996) and Mithun (1999), are not in total agreement as to names, spellings, and internal relationships within a family. Names must be approached with caution, and detective work is sometimes necessary to determine whether two names refer to a single language, or to more than one language. The sources are helpful in sorting this out; see Campbell (1997: 5–7) and Goddard (1996: 4–8, 9). Along similar lines, there are disagreements on spelling. I follow Goddard (1996: 4–8) in both classification and spelling throughout unless otherwise noted; alternative names are sometimes given in parentheses. Third, in many cases, the number of fluent speakers of a language is small, and the languages are undergoing rapid change; thus data checked today may be quite di¤erent from data gathered fifty or so years ago, or even more recently. Hayes (1995: 3–4), in his book on metrical stress theory, makes a remark that one would like to heed: ‘‘One final remark on data: readers conducting their own research in metrical theory should be warned that as a reporter of other people’s data, I am fallible, and they are therefore urged to consult the original secondary sources (obtainable in research libraries and by interlibrary loan), or if possible to find a native speaker consultant.’’ This advice sets out best practice; it is, unfortunately, all too di‰cult to follow the latter part of his advice in many cases in North America. Fourth, as cautioned in the StressTyp Manual and as the Hayes’ quote above indicates, the data are as trustworthy as the information in the sources. Sources are sometimes di‰cult to interpret, they do not always contain the full range of data required to make statements about the languages, not all sources agree on the descriptive facts of a language, and the analyses on which the classification is based may themselves be controversial. The latter is often hinted at in the remarks field of StressTyp. Many of the languages presented in this chapter are discussed in Hayes (1995), although his theoretical framework is di¤erent from that of

Accent in the native languages of North America

159

StressTyp. Hayes (1995) is an invaluable source as he often includes more detailed analysis of a particular language than I am able to in the confines of this article. With these thoughts in mind, I turn now to a study of the languages. Some discussions are longer than others, as more is known about some languages than others. Many language families and isolates are not represented in StressTyp. I have added languages that do not appear in the database, but which flesh out patterns in a family and provide the reader with a better overall picture of both diversity and uniformity within a family or which extend the breadth of coverage of families. In general, I have used the transcription of the source, except for marking stress with ‘¨’ and ‘˙’; I reserve ‘´’ and ‘`’ for high and low tone respectively.

3. Accent patterns in language families The goal of this section is to bring together the North American native languages for which something about stress is known, being as comprehensive as possible. The discussion of language families is ordered following Goddard (1996). (See the appendix for the overall classification from Goddard 1996.) My major focus is on where accent falls in the languages, using the StressTyp parameters to frame the discussion. The phonetic manifestation of accent is sometimes mentioned, and is discussed in section 4.4. 3.1. Eskimo-Aleut The Eskimo-Aleut family has two major branches, Eskimoan and Aleut. Several Eskimoan languages are included in StressTyp, including languages from the Inuit-Inupiaq branch and from the Sirenikski-Yupik branch. The Yupik languages have similar patterns of prominence, and are generally classified as I/NMS in StressTyp, with an initial quantity sensitive (V:) iamb and quantity sensitive iambs built thereafter (see Hayes 1995: 239), and without di¤erentiation in degrees of stress (NMS). Central Siberian Yupik (St. Lawrence Island dialect), discussed in Krauss (1975, 1985a, b) and Jacobson (1985), classified I/S (NMS), has accent on long vowels and non-final even-numbered members of a lightsyllable (V, VC) string, counting from the left. No di¤erences in degree of stress are noted. Examples are given in (2); parentheses indicate footing.

160

Keren Rice

(2) Central Siberian Yupik: St. Lawrence Island dialect (Jacobson 1985 unless otherwise noted) a. initial light syllable (V, VC) followed by light syllables qayani ! (qa¨ya:)ni

‘his own kayak’ 27

(aN¨yax. )(´ax. ¨´aN)yuxtuq ‘he wants to make a big boat’ 26 akisimaNisimakaNa ! (a¨ki:)(si¨ma:)(Ni¨si:)(ma¨ka:)Na ‘he didn’t answer him’ Krauss (1975: 54–55) b. initial light syllable (V, VC) with heavy syllable (V:) later in word sa'uya:ni ! (sa¨'u:)(¨ya:)ni

‘in his (another’s) drum’ 27

c. initial heavy syllable (V:) followed by light syllables (V, VC) (¨a:N)(qax. ¨´ax. )(´aN¨yux)tuq

‘he wants to make a big ball’ 26

d. initial heavy syllables (V:) followed by heavy syllable (V:) (¨tu:)(¨na:)'a: ‘he’ll give it’ Krauss (1975: 54) Accented short vowels in open syllables are lengthened (as in some forms in (2a, b)), and, for some speakers, long vowels are lengthened in open syllables as well (not shown). Jacobson (1985: 27) notes that a morphological component is involved in accent in Siberian Yupik languages: accent is on the stem of a word with su‰xes. In particular, an initial vowel of a word with su‰xes is lengthened if necessary to place accent on a stem. Central Alaskan Yupik has four major dialects, General Central Yupik, Norton Sound Yupik, Hooper Bay-Chevak, and Nunivak. The overall pattern in Central Alaskan Yupik is one of accent on a word-initial heavy syllable (V:, VC), and on alternating syllables thereafter, with lengthening of accented vowels; final syllables are never accented. (3) Central Alaskan Yupik: overall pattern (Miyaoka 1985) a. initial heavy syllable (V:, VC) aNyamini ! (¨aN)(ya¨mi:)ni ‘in his own boat’ 57 b.

initial light syllable (V) qayamini ! (qa¨ya:)mini

‘in his own kayak’ 57

In Norton Sound Yupik (Unaliq dialect; Central Alaskan Yupik) (I/S (NMS)), initial V: and VC syllables are heavy; non-initial VC is not. Long vowels and even numbered syllables in a string of light syllables are

Accent in the native languages of North America

161

accented; accented vowels in open syllables lengthen and conditions exist under which consonants geminate (not shown). Final syllables are never accented. A vowel before a heavy syllable has secondary accent (4c). (4) Central Alaskan Yupik: Norton Sound Yupik, Unaliq dialect a. initial heavy syllable (V:, VC), light syllables (V, VC) later in the word aNyax.paka ! (¨aN)(yax. ¨pa:)ka ‘my big boat’ Krauss (1985b: 21) b.

initial light syllable (V), light syllables (V, VC) later in word (qa¨yax. )(paN¨yux)tuq ‘he wants to get a big kayak’ Jacobson (1985: 30)

c.

initial light syllable (V), V: later in word qayapigka:ni ! (qa¨ya:)(˙pix¨ka:)ni ‘in his (another’s) future authentic kayak’ Jacobson (1985: 31)

Norton Sound Yupik di¤ers from Central Siberian Yupik in several ways; we see here that initial V: and VC syllables count as heavy in the former, but only V: in the latter. General Central Yupik is largely identical to Norton Sound Yupik except that CVC preceding a non-final CV is accented. In addition, there are conditions under which the vowel schwa (written e) deletes. (5) Central Alaskan Yupik: General Central Yupik dialect (Jacobson 1985 unless otherwise noted) a. initial light syllable (V) qayapixkani ! (qa¨ya:)(¨pix)kani ‘his own future authentic kayak’ 33 (cf. Norton Sound: (qa¨ya:)(pix¨ka:)ni) qayapixka:ni ! (qa¨ya:)(¨pix)(¨ka:)ni ‘in his (another’s) future authentic kayak’ 33 cf. Norton Sound: (qa¨ya:)(¨pix)(¨ka:)ni b.

initial heavy syllable (V:, VC) paqnaksaqunaku ! (¨paq)(¨nak)(sa¨qu:)naku ‘don’t be inquisitive about it’ Krauss (1985b: 21) aNyax. paka ! (¨aN)(¨yax. )paka ‘my big boat’ Krauss (1985b: 21)

The Chevak dialect of Central Alaskan Yupik (I/S NMS) has a similar accent pattern to General Central Yupik, di¤ering in the environment for

162

Keren Rice

schwa deletion, the lack of accent on post-tonic VC before an accentless syllable, and vowel shortening under certain conditions; see Jakobson (1985: 36) and Hayes (1995: 251). Syllables with long vowels are accented, and closed syllables are accented preceding a word-medial light syllable. (6) Central Alaskan Yupik (Chevak dialect) a. (¨qus)(Nir¨Nal)(Nur¨pag)(taN¨qerr)(sug¨nar)quq ‘there seems to be a big goat’ Miyaoka (1985: 57) (orthography except for the velar nasal) b. (i¨kam)‚i‚lutN ‘they having sleds’ (Hayes 1995: 251) One further Central Alaskan Yupik dialect, Nunivak, is similar to the Chevak dialect but shortens heavy syllables under a wider range of conditions. Pacific Yupik or Alutiiq (Chugach dialect), I/S (NMS), is similar to other Central Alaskan Yupik languages, with accent on long vowels and closed first syllables. There is alternating accent, with lengthening of accented vowels in open syllables and special treatment of syllables preceding V:. Unlike other languages, the Chugach dialect of Pacific Yupik allows final accent, as in several of the forms in (7). (7) Pacific Yupik, Chugach dialect (Leer 1985a) a. all light syllables ataka ! (a¨ta:)ka atunnirtuq ! (a¨tun)(nix.¨tuq)

‘my father’ 116 ‘he stopped singing’ 87

b. initial light; heavy syllables later in word alika: ! (a¨li:)(¨ka:) ‘she is afraid of it’ 84 c. initial heavy, followed by lights annaqa ! (¨an)(Na¨qa) ‘my older brother’ 110 taq´uni ! (¨taq)(´u¨ni) ‘getting done’ 89 d. initial heavy, heavies elsewhere in word agNuaqutartuaNa ! (¨ag)(¨Nua)(qu¨tar)(¨tua)Na ‘I am going to dance’ 170 The facts of Pacific Yupik are more complex; see Leer (1985a, 1985b) and Hayes (1995) for details and interpretation. The Yupik languages show additional complexities in their accent systems; see Jakobson (1985), Krauss (1975, 1985a, b), Leer (1985a, b), Miyaoka (1985), and others for foundational work and Hayes (1995) for one interpretation.

Accent in the native languages of North America

163

The Inuit-Inupiaq branch of Eskimoan is represented in StressTyp by West Greenlandic, classified as NMS. The final syllable is accented, as are heavy syllables (V:, VC). Rischel (1974: 96) notes that much is unclear about the accent system in West Greenlandic, and an analysis that syllables with branching rhymes and final syllables are accented is tentative. Krauss (1985b: 8) remarks that accent falls on the ultima, penult, or antepenult. Fortescue (1984: 340) points out that there is an impression of stress on heavy syllables that form the intonational nucleus. Aleut (Aleut branch) is discussed in Bergsland (1997). The accent system is complex, but basically the initial syllable of a word is accented. Placement of subsequent accents is based on a constellation of factors – syllable weight, sonority of a postvocalic consonant, and rhythm. The language is probably best treated as having accent on the first syllable with accent thereafter determined by a quantity-sensitive rhythm system. Krauss (1985b: 10) points out that prosody in Aleut is probably more a sentence- or discourse-level phenomenon than word-level. Various characteristics of the Eskimo-Aleut languages represented in StressTyp are summarized in (8). These include information about the language in terms of primary accent and rhythm; direction signifies the edge orientation for rhythm. Much information is omitted; see the discussion and the database for details. The chart nevertheless highlights the major similarities and di¤erences between the languages of the family. (8) Accent systems in Eskimo-Aleut languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Central Siberian Yupik (St. Lawrence Island)

I/S (NMS)

yes V:

yes

L, iamb

yes V:

Central Alaskan Yupik, Pacific Gulf Yupik (Alutiiq)

I/S (NMS)

yes V:, VC

yes

L, iamb

yes V:

These languages are similar overall in their accent systems, but di¤er in detail. 3.2. Algonquian languages The Algonquian family is a vast one, with languages found virtually from coast to coast in North America. A large number of Algonquian lan-

164

Keren Rice

guages are generally analyzed as having iambic systems – these include Algonquin (Piggott 1978, Valentine 1994), Malecite-Passamaquoddy (Hagstrom 1995, Hayes 1995, Stowell 1979, Teeter 1971, Teeter and LeSourd 1983), Menominee (Bloomfield 1939, 1962, 1975, Hayes 1995, Hockett 1981, Milligan 2005, Pesetsky 1979), Munsee Delaware and Unami Delaware (Goddard 1979, 1982, Hayes 1995), Eastern Ojibwa (Bloomfield 1957, Halle and Vergnaud 1987, Hayes 1995, Kaye 1973, Piggott 1980, 1983, Valentine 1994), North Western Ojibwa (Valentine 1994), Severn Ojibwa (Todd 1970, Valentine 1994), and Potawatomi (Hayes 1995, Hockett 1948) (from Brittain 2000: 181–182). See Milligan (2005: 28–37) for a detailed overview of accent in the Algonquian family. I begin the study of Algonquian languages with an Ojibwayan language in which the accent system is fairly transparent. In discussion of accent in the Nishnaabemwin dialect of Eastern Ojibwa (Valentine 2001: 54), Valentine proposes the following rules: (9) ‘‘A metrical foot consists of a pairing of a weak and a strong syllable, with the following restrictions: Only short vowels can be weak; long vowels are always strong. The last vowel in a word is always strong (or, alternatively, is exempt from the stress rules). Build weak-strong metrical feet starting from the beginning of the word. Once footing is completed, de-emphasize metrically weak vowels by reducing them to schwa or deleting them, and give stress to metrically strong vowels. Give main stress to the strong syllable in the third foot counting back from the end of the word.’’ Only vowel length contributes to syllable weight. This system is illustrated for Minnesota Ojibwa (Valentine 2001) in (10). (10) Minnesota Ojibwa (Valentine 2001 except as indicated) a. (gi¨da)(go˙shin) ‘you arrive’ 52 b. (da¨go)(˙shin) ‘animate singular arrives’ 53 c. (¨e:)(si˙ba)(˙nag) ‘raccoons’ 53, 55 d. (ni¨wa:)(ba˙mi)(go˙na:n) ‘s/he sees us (exclusive)’ 55 e. (in˙gi:)(˙na:)( ji¨mi:)( ji˙me:)(˙min) ‘we fetched food’ (Valentine p.c. 2000) Valentine notes that there is variation in the form in (10e); an alternative has primary accent on the penultimate syllable. This is a count system,

Accent in the native languages of North America

165

since rhythm precedes and determines the assignment of primary accent. The placement of primary stress is unusual, and worthy of additional study. Other Ojibwa dialects di¤er from Minnesota Ojibwa in one major way: unaccented vowels syncopate. For instance, in Nishnaabemwin, the Minnesota Ojibwa words in (10) have the forms in (11). The words are identical except for syncopation in Nishnaabemwin. The primary accent is arbitrarily placed before the consonant preceding the accented vowel; this should not be interpreted here or elsewhere as a reflection of syllabification. (11) Nishnaabemwin (Valentine 2001) a. g¨dag˙shin ‘you arrive’ 52 b. c.

d¨go˙shin ¨e:s˙ba˙nag

‘animate singular arrives’53 ‘raccoons’ 55

d.

n¨wa:b˙mig˙na:

‘s/he sees us (excl.)’ 55

(There are complications; see Valentine 1994, 2001 for detailed discussion of the variation in Ojibwa dialects; see also Hayes 1995.) Syncope is common in Algonquian languages, and Valentine (2001: 51) remarks that the Ojibwa accent rules are similar to those of Delaware, Potawatomi, and Menominee, and I turn to some of these languages now. Munsee Delaware and Unami Delaware, closely related Eastern Algonquian languages, are similar to Nishnaabemwin; they di¤er primarily in what counts as heavy (V: and VC) and in the placement of primary accent. Initial heavy syllables and alternating syllables thereafter are accented, with primary accent on the last non-final accented syllable. Like Nishnaabemwin, unaccented vowels are syncopated. In Unami Delaware, Goddard (1979: xvi) states that a short vowel is syncopated ‘‘before a cluster with first member x, s, or sˇ, except after a weak-vowel syllable’’ (Goddard 1979: xvi). In addition, consonant gemination applies after an accented vowel in Munsee (if the consonant is voiceless), and after a short stressed vowel in Unami (any consonant). See Hayes (1995: 211– 215) for detailed discussion. (12)

Unami Delaware (Goddard 1979) a. (pkw) [p¨ko]

‘(vegetable) gum’ xi

b.

(Saw)(sw)

[Sa¨wso]

‘he is weak’ xiv

c.

(max)(ksw)

[¨maxkso]

‘he is red’ xx

d.

(Sa:)(wala:)(mwi:)(twak)

[Sa:la:¨mwi:ttowak]

‘the little ones starved to death’ xiii

166

Keren Rice

(13) Munsee Delaware (Goddard 1982) a. (w˙la)(ma¨l)(sw) [w(lama¨lsw] ‘he is well’ 35 b. (n˙ka)(k¨ta)(kka) [nkak¨takka] ‘I do a fast dance’ 35 Maliseet (Malecite)-Passamaquoddy, closely related Eastern Algonquian languages, is also similar to Ojibwa in terms of accent pattern. See LeSourd (1993) for discussion of di¤erences between Maliseet and Passamaquoddy and for details of the system; forms here are from LeSourd (1993) unless otherwise noted and represent Passamaquoddy. Like Nishnaabemwin and unlike Delaware, full vowel syllables (and not schwa) count as heavy in Passamaquoddy. The first syllable is accented, as is the penult (main stress), and accent occurs on alternate syllables before it. (14) Passamaquoddy (LeSourd 1993) a. ¨wicoh¨ke-m-a-l ‘he helps the other’ 74 b. ¨wi¨cohke-¨kemo ‘he helps out’ 74 c. ¨was-¨is-k ‘children’ 75 d. ¨pem-s¨kot-e-k ‘field’ 75 e. ¨pem-sko¨t-e-k-il ‘fields’ 75 The vowel schwa sometimes counts towards accent, as in (15) – it, like other vowels, is skipped in a final syllable, and the preceding syllable, the penultimate, is accented. (15) Passamaquoddy (LeSourd 1993) a. ¨tohsan ‘shed’ 80 b. ¨toh¨san-k ‘shed, locative’ 80 c. h-¨pihi¨tin-ah¨kwem-l ‘his arms’ 80 However, schwa may be skipped for purposes of accent when it is not in the final syllable – it is not accented when it is penultimate. (However note that it can be accented, as in (16c); other conditions are involved as well.) (16) Passamaquoddy (LeSourd 1993) a. ¨sok-lan ‘it pours (rain)’ 81 b. ¨wicoh¨ke-t-m-n ‘he helps it’ 81 c. ˙mek˙nut¨spnik ‘those animate who must have been chosen’ [me:knu:tspni:k] Stowell 1979: 57, Hayes 1995: 216 2 0 3 1 0 3

Accent in the native languages of North America

167

The system di¤ers in minor ways from the Algonquian languages seen so far. Western Abenaki (Day 1994), another Eastern Algonquian language, diverges from the Algonquian languages examined so far. Day describes the accent pattern as follows: ‘‘. . . the primary accent is on the final syllable and lesser accents fall on every second syllable counting backwards without counting those syllables whose nucleus is e.’’ (xii) (the symbol is the orthographic representation for schwa). This is reminiscent of Passamaquoddy, where non-final schwas generally do not a¤ect the count. (17) Western Abenaki (Day 1994) a. (˙mo)(za¨gen) ‘moose hide’ xii b. (˙mi)(ko¨wa) ‘squirrel’ xii c. (˙gwi)(gwi¨gem) ‘black duck’ xii d. (mo˙za)ge(ni¨ya) ‘moose hide material, of moose hide’ xii Primary accent triggers vowel lengthening (not shown) and is accompanied by a pitch rise. Western Abenaki can be classified as U, without any complications in the primary accent system. The rhythm system creates alternating accent from the right, and is sensitive to vowel quality, treating schwa as if it were absent in non-final syllables. Menominee has a quantity-sensitive system, with V: counting as heavy; it is classified as L/. Primary accent is on penultimate long vowels and diphthongs and on long vowels followed by a syllable with a short vowel; secondary accent is on even numbered syllables in a series of short ones, but not on a final vowel. Menominee has vowel lengthening and shortening, with lengthening applying to monosyllabic light items and, in words with two light syllables in the first foot, to lengthen the head; shortening a¤ects vowels in weak syllables. Lengthening and shortening also apply in other environments; see Hayes (1995: 219–200) and references therein for discussion; see also Milligan (2005) for recent work. (18) Menominee a.

(ahsa)(ma:w) ! (ah¨sa:)(¨ma:w)

b.

(netah)(sama:w) ! (ne¨ta:h)(sa¨ma:w)

c.

me:wenEsehEwak ! (¨me:)(we˙nE)(se˙hE)wak (¨se:)(pe:w) (e¨nE:)(niw)

d. e.

‘he is fed’ Hayes (1995: 219), Bloomfield (1962: 57) ‘I feed him’ Hayes 219, B 57 ‘they drive him away’ Bloomfield (1962: 20) ‘river’ Bloomfield (1962: 19) ‘man’ Bloomfield (1962: 19)

168

Keren Rice

No primary accent is reported in words without long vowels (Bloomfield 1962: 20). Cree-Montagnais is a major group within Algonquian. In recent work on an East Cree subdialect, Southern East Cree, Brittain (2000) argues that it is similar to the languages discussed above; she identifies Southern East Cree as having iambic feet and weight distinctions (V:, VG, VN count as heavy). Mackenzie (1980: 48) describes the accent pattern of Southern East Cree as follows: ‘‘In di-syllables the long vowel . . . receives the stress, the penultimate one in the case of a word having two long vowels. . . . In words of three syllables or more, stress usually falls on the rightmost underlying long vowel.’’

Brittain (2000) refines this statement, noting that one of the last two syllables is always heavy. See Brittain (2000) for detailed discussion and analysis. In Montagnais, accent is word final (Brittan 2000: 186–187). Plains Cree accent is described as follows (Wolfart 1996: 431; see also Wolvengrey 2001: xxiv): ‘‘The stress pattern of words seems to depend primarily on the number of syllables rather than on vowel length. Disyllabic words are stressed on the last syllable: [Iske´:w] iskwe:w ‘woman’, [mIhtI´] mihti ‘piece of firewood.’ . . . In words of three or more syllables, primary stress falls on the third syllable from the end. Secondary stress then falls on alternative syllables in either direction, reckoned from the antepenult: [ne´:hiya`w] ne:hiyaw ‘Cree Indian’, [pˆskwa`:pIsı´mowı`n] pasakwa`:pisimowin ‘Shut-eye dance’. Note that the above rule holds for Plains Cree even where the penultima is long: [mI´ta:ta`ht] mita:taht ‘10’, [spwa:ga`n] ospwa:kan ‘pipe’.’’ This system is di¤erent from those of the Algonquian languages seen so far: weight is not relevant and the accent is determined from the right rather than the left edge. For many other Algonquian languages, available documentation is not adequate to determine accent systems; see Milligan (2005). Frantz (1991) and Frantz and Russell (1995) treat Blackfoot as having lexically marked pitch accent based on the existence of minimal pairs such as those in (19). Note that the acute accent marks pitch accent. (19) Blackfoot (Frantz and Russell 1995: 438) a. a´pssiwa ‘it’s an arrow’ apssı´wa ‘it’s a fig’

Accent in the native languages of North America

b.

169

a´kaohkiimiwa ‘he’s married’ a´ka´a´ ‘perfect, completive’ þ ohkiimi ‘wife’ þ wa ‘3 sg.’ aka´o´hkiimiwa ‘he has many wives’ aka´ ‘many’ þ ohkimmi ‘wife’ þ wa ‘3 sg.’ (accent spreading to an adjacent vowel is found here; Frantz 1991: 153)

The pair in (19a) suggests lexical accent; that in (19b), with morphological di¤erences, suggests that work on what is lexical and what is predictable might be of interest. The table in (20) summarizes the accent patterns found in the Algonquian languages examined here in terms of relevant parameters for accent and rhythm. (20) Accent systems in Algonquian languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Nishnaabemwin, Ojibwa, Minnesota, MaliseetPassamaquoddy

L(CNT)

yes rhythmically strong

yes

L, iamb

yes V:

Delaware, Munsee and Unami

L(CNT)

yes rhythmically strong

yes

L, iamb

yes V:, VC

Montagnais

L

no

uncertain

Menominee

L/

yes V:

yes

L, iamb

yes V:

Western Abenaki

U

no

yes

R, iamb

yes full V

Plains Cree

A

no

yes

C

no

We see here diversity in synchronic accent systems of Algonquian languages: in terms of primary accent, languages are often L(CNT), but U, L/, and A systems exist as do languages with no apparent predictable accent system, or at least without an apparent phonetic realization of a more abstract accent system. Hayes (1995: 222) concludes in his discussion

170

Keren Rice

of Algonquian languages that despite the widespread presence of quantity sensitive iambic systems, many Algonquian languages do not have such systems, and it is not now possible to reconstruct a stress system for Proto-Algonquian. Before leaving Algonquian languages, I examine briefly the higher-level classification, Algic. Algic includes the Algonquian languages as well as two California languages, Wiyot and Yurok. Wiyot has accent on what are reconstructed as Proto-Algic long vowels; in the absence of such a vowel, it has accent on the first vowel. It is classified in StressTyp as an F/F language, sensitive to reconstructed long vowels, treated synchronically as lexical accent. Proto-Algic forms are included in (21). (21) Wiyot (Proulx 1984) a b¨lid

*meli:ni

‘eye’ 181

b. ¨wakl c. ¨bsd

*wa:kel*meseni

‘peppernuts’ 191 ‘breast’ 172

d. pu¨mipl /¨pumip?l

*pegemi:pi

‘stone knife’ 195

In Yurok, Robins (1958) comments that accent, realized as high pitch and sometimes loudness, may vary on a single word with style, tempo, and rhythmic patterns of a sentence. According to Blevins (2003), in Yurok nouns long vowels attract accent, which is manifested as a steady high pitch. Closed syllables with short vowels may also bear primary stress. 3.3. Athabaskan languages The study of accent systems in Athabaskan languages is in its infancy. This family, like the Algonquian family, is geographically widespread, with languages spoken in three discontinuous geographic areas of North America. The Athabaskan family points to the importance of the morphological constituent of stem in the determination of accent. Languages of this family have a rich and complex prefixing system. In this section I deal only with accent on the stem. See Rice and Hargus (2005) for detailed discussion. Ahtna is reported by Kari (1990) to have primary accent on the rightmost heavy stem syllable in a word. I ignore secondary accent as Kari is less certain about its placement (Kari 1990: 17). In the following forms, a hyphen separates the stem from the rest of the word, both prefixes and su‰xes. This system has a short/long vowel contrast. Long vowels are

Accent in the native languages of North America

171

written with two symbols (e.g., ii, aa, ae) and short vowels with a single symbol (e.g., e, a), and the stems are underlined. The orthography from Kari (1990) is used; note that {gh} is a voiced velar fricative. (22) Ahtna (Kari 1990: 17) a. u-¨ghael-e’ ‘his pack’ ‘his (detached) horn’ b. uc’e-¨de’ ‘dipnet’ c. ¨ciis-i ‘cottonwood’ d. t’a¨ghes e. da’es-¨yaas ‘I enter’ ‘pity’ f. ¨tege The rightmost heavy syllable (V:, VC) of the stem is accented, with glottal stops ignored for the computation of accent. This system can be seen by comparing the stems with two syllables, (22d) and (22f ). In the former, with a final closed syllable, accent is on the righthand stem syllable; in the latter, where the right syllable is light, this syllable is not accented. The fact that it is necessary to refer to the stem can be seen in (22b), where the final vowel does not create a heavy syllable, being short rather than long, nor does the glottal stop contribute to weight, but the stem vowel is accented nevertheless. Long vowels elsewhere in the word also attract primary accent. In Salcha and Minto, Tanana dialects, stress is assigned to stems, following the same pattern as Ahtna, to heavy syllables (V:, VC), and to toned syllables (in the Minto dialect, which has lexical low tone); see Tuttle (1998). In Slave, too, the stem plays an important role in determining accent placement. In the South Slavey dialect, quantity distinctions have generally been neutralized, and accent is on the stem vowel. The stem vowel is underlined; an acute accent marks high tone, and the hyphen indicates a stem boundary. (23) South Slavey (dialect of Slave; data from Rice 1989 unless otherwise indicated) a. b. c. d. e.

Eh-¨ke ¨´u-E nE-¨do¸ dE-¨tsil-E aE-¨tsEl-ih

‘boy’ ‘fish’ ‘you sg. drink’ ‘it is red’ Howard 1990: 587 ‘it is small’ Howard 1990: 586

172

Keren Rice

The stem vowel is stressed, whether a su‰x is present or not (23b, d, e), and whether that su‰x ends in a consonant (23e) or a vowel (23b, d). In the Hare dialect of Slave the position of accent di¤ers in nouns and verbs. Nouns are accented on the stem (24a), as in South Slavey; verbs have prestem accent, as in (24b), whether the stem is final (24b i) or the final vowel is in a su‰x (24b ii, iii). (24) Hare (dialect of Slave) a.

nouns: primary accent on the stem (data from Rice 1989) Eh-¨ke ‘boy’ ¨lug-E ‘fish’

b. verbs: primary stress on the pre-stem syllable i. ¨nE-do¸ ‘you sg. drink’ ii. ¨dE-dEl-E ‘it is red’ iii. ¨hiƒ-sEl-E ‘it is small’ Vowel weight distinctions are not found. In Witsuwit’en, accent is attracted to stems, as shown by the examples in (25). Again, I ignore prefixes. The stems are underlined. (25) Witsuwit’en (Hargus 2005: 396) ‘above him’ a. b-¨dG-t (b- third person þ dG ‘above’ þ -t su‰x) b. (¨b-)d¨cts ‘(her) fist’ (b- third person þ dcts ‘fist’) In the case of a bisyllabic stem, the final vowel is accented (25b); su‰x vowels do not receive an accent (25a). Other factors come into play in determining accent on prefixes, the most important being weight, as in Ahtna and Tanana. Fort Ware Sekani (Hargus 2005) also shows the importance of morphology, with stem vowels (underlined) being accented. (26) Sekani (Fort Ware dialect) (Hargus 2005: 406) a. ¨dne ‘person, man’ (stem: dn) b. d-¨de ‘marmot’ (stem: de) Prefix accent is determined by factors such as vowel quality, position in the word, and, to a lesser degree, tone. In other Athabaskan languages in which accent has been studied, the morphological category of stem is also important (e.g., Tahltan, Alderete and Bob 2005; Navajo, McDonough 1999; Tanacross, Holton 2005; San

Accent in the native languages of North America

173

Carlos dialect of Western Apache, Tuttle 2005; Jicarilla (Eastern Apache), Tuttle (2005), and Leer (2005) proposes that accent has been an identifier of the stem since Proto-Athabaskan. It is only in the rare case of languages with weight contrasts and two syllable stems that one can sort out the role of phonology in determining the placement of accent within the stem domain. Hupa, however, appears to be di¤erent, with stress falling on a two syllable window at the left edge and, in some speech, on the stem in the absence of prefix long vowels (see Gordon and Luna 2004). The languages studied so far suggest overall that primary accent associated with the stem is a stable characteristic of this family. (27) Accent systems in Athabaskan languages (stem stress; su‰xes ignored) language

primary accent type

quantity sensitivity

Ahtna, Salcha

U/P

yes V:, VC

Sekani

U

no

Slave (most dialects)

U

no length distinction

Slave (Hare dialect)

P – verb U – noun

no length distinction

Witsuwit’en

U

no

Athabaskan languages are part of a larger subgroup, Athabaskan-Eyak. Tlingit is included in yet a larger subgroup, Na-Dene. Leer (1991: 15) notes that at the stage of what he terms Common Tlingit, just before the language split into its modern dialects, accent was predictably associated with the stem except for prefixed and proclitic stems. Thus, Tlingit historically shares with Athabaskan languages the property that accent placement is partially morphologically determined, with the stem being prominent. Leer (2005) notes that in Northern and Interior Tlingit, stem stress is phonetically detectable, but not as prominent as tone. 3.4. Haida Haida is an isolate (see Campbell 1997: 114–115 for discussion of the Na-Dene proposal, with Athabaskan languages, Tlingit, Eyak, and Haida forming a superfamily). It has two major dialects, Northern (Masset, Alaskan) and Southern (Skidegate, Kunghit). The two languages reported here are from the Northern group.

174

Keren Rice

Masset Haida is classified in StressTyp as L/L (tone), with primary accent on the last high tone syllable, or, in the absence of a high tone, on the final syllable. Secondary accent falls on the final syllable if it does not host the primary accent and on alternating syllables before the primary accent. In (28a, b), the words have level tone, and final accent with alternating accent preceding. The word in (28c) has high tone on the penultimate syllable, and that syllable attracts the primary accent; the final syllable and alternating syllables before the primary accent receive secondary accent. (28) Masset Haida (Enrico 1991) a

˙gu:diN.¨e:

b. ˙gudi˙sahlda-¨hid c. ˙ga dla˙dajan-¨da´:l-˙gaN

‘giant purple urchin’ 111, 112 ‘worry, inceptive’ 112 ‘jump up-iter.-along.pr’ 114

Alaskan Haida realizes accent as tone on the leftmost heavy syllable (V:, VR, and, surprisingly, Vp). (29) Alaskan Haida (Lawrence 1977) ‘rub with stick’ 51 a. da-sk’a-¨na:n ‘rub with heavy compact object’ 51 b. da-¨k’i:-na:n c. gı´: ¨kal-ya:nda:l ‘large person is walking briskly’ 52 No accent is assigned (i.e., no phonetic tone occurs) if no vowel is accentable. (30) ch’a-t’as

‘wear something made of cloth’ Lawrence: 53

Tone may also be intrinsic. 3.5. Wakashan Wakashan languages fall into two major branches, Kwakiutlan (Northern Wakashan) and Nootkan (Southern Wakashan). Kwakiutl (Kwak’wala), a Northern Wakashan language, is F/L, with accent on the leftmost heavy syllable (V:, VR, where ‘R’ is a non-glottalized resonant), and, in the absence of a heavy syllable, on the right edge (31e, f, h). The pattern is illustrated in (31), with (31h) showing that glottalized sonorants do not create heavy syllables. (31) Kwak’wala (Boas 1947 unless otherwise noted) a. ¨qa:sa ‘to walk’ 218 b. ¨dlxa ‘damp’ 218

Accent in the native languages of North America

c.

¨ha:dza:pa:ma:

‘yarrow’

d.

t’¨li:dzu

‘large board on which fish are cut’ Grubb 1977: 207

e.

m’kw¨la

‘moon’ 218

f.

gas¨xa

‘to carry on fingers’ 217

g.

m¨xnx. nd

‘to strike edge’ 219

h.

ml’¨qa

‘to repair canoe’ 219

175

Makah (Southern Wakashan) is classified in StressTyp as I/S, the bounded equivalent of F/L. This is an iambic system, with weight sensitivity (V:); accent is on the first syllable if the vowel is long, and otherwise on the second. No rhythm system is reported. Tsishaath Nootka (I/I) (Southern Wakashan) has primary accent on one of the first two syllables – on the first heavy syllable (V:, VR) (32a–d) or, in the absence of a heavy syllable in the first two syllables, on the first syllable (32e, f ) (Stonham 1999: 59). (32) Tsishaath Nootka (Stonham 1999: 59) a. ¨ku:h. sinqin´?ap ‘always cause hole to be in the side’ b.

¨?inksy’iqu:

‘make a fire’

c.

qah. ¨na:k’atl

‘someone now died’

d.

¨wa:ma:h. su:

‘I was saying so’

e.

¨t’an’anak

‘have a child’

f.

¨tanakmi:cˇi?atl

‘turned now into mosquitoes’

Stonham (1999: 61) remarks that secondary accent appears to occur on alternating syllables after primary accent, but that further investigation is required. In this small sample of Wakashan languages, the following properties are found: sensitivity to weight and attraction of primary accent to a heavy syllable near the left-edge of the word. The languages di¤er primarily in terms of what counts as heavy (V: only or V: and VR) and where accent defaults to in the absence of heavy syllables. See Fortescue (2007: 12) for brief discussion of accent patterns in some additional Wakashan languages.

176

Keren Rice

(33) Accent systems in Wakashan languages language

primary accent type

quantity sensitivity

rhythm

Kwak’wala

F/L

yes V, VR

none

Makah

I/S

yes V:

none

Tsishaath Nootka

I/I

yes V:, VR

yes

direction

quantity sensitivity

L, iamb

no?

3.6. Chimakuan Chimakuan consists of two languages, Chemakum and Quileute. Mithun (1999: 377) reports that accent is often, but not always, penultimate, realized as pitch. 3.7. Salishan The Goddard (1996: 5–6) classification of the Salishan languages proposes six major subgroupings for these twenty-three languages; see CzaykowskaHiggins and Kinkade (1997: 3) for a slightly di¤erent division and for recent names. Czaykowska-Higgins and Kinkade (1997: 15–16) include a brief overview of accent patterns within the family, and the following discussion is based on their work. Before summarizing the accent types proposed by Czaykowska-Higgins and Kinkade (1997), it is worthwhile to consider a comment made by Kinkade (1997: 208): ‘‘Stress is an extremely complicated issue in nearly all Salish languages.’’ One must keep in mind that this discussion is simplified, and that morphological factors are important in almost all Salish languages, often obscuring the phonological generalizations discussed below. Only three Salish languages are included in StressTyp, with two open to reanalysis. I discuss several Salish languages here, with all information about accent systems taken from Czaykowska-Higgins and Kinkade (1997) unless otherwise indicated. See also Shaw, Blake, Campbell, and Shepherd (1999) for a detailed analysis of Musqueam (Halkomelem; Central) as well as analyses of several other languages (Lillooet [St’a´t’imcets], Cowichan, Lushootseed, Mainland Comox, Squamish, Sliammon). Czaykowska-Higgins and Kinkade (1997) identify four basic types of accent systems within Salish. One occurs in the interior languages except Lillooet and perhaps Coeur d’Alene. Primary accent occurs as far right

Accent in the native languages of North America

177

in a word as possible given lexically specified accent properties of the morphemes involved. For instance, Czaykowska-Higgins (1997) argues that Columbian has a basic accent rule by which accent is assigned to the rightmost syllable in a sequence. Word-internal structure is indicated in the following forms, with ‘-’ marking a prefix boundary, ‘q’ the root and ‘¼’ another boundary type. (34) Columbian (Czaykowska-Higgins 1997) ‘turtle’ 176 a. ?ara¨sikw w q?arasik b.

¨naqs ‘one’ 176 nnaqs¨qin ‘one tipi’ 176 n-qnaqs ¼ qin locative-qone ¼ head

c.

naqsqn¨wil ‘one load’ 176 qnaqs ¼ qin ¼ wil qone ¼ head ¼ container

Morphological complications exist. For instance, with one class of roots, accent is maintained on the root when a single su‰x is present, but shifts to the final su‰x when two or more su‰xes are added. (35) Columbian (Czaykowska-Higgins 1997) ¨xwirkstm ‘reach out’ 176 a. qxwir ¼ akst-m qreach ¼ hand-middle verb b.

qxw ir ¼ akst ¼ atkw xwirks¨tatkw ‘reach into water’ 176 qreach ¼ hand ¼ water

Czaykowska-Higgins (1993) proposes five classes of su‰xes with di¤erent accent properties, and four classes of roots. Abstracting away from morphological properties, further facts suggest that Columbian has an unbounded system: if a word contains only schwas, accent falls on the leftmost vowel. (36) Columbian (Czaykowska-Higgins 1997) ¨k’mlqstxn ‘lower leg’ 179 qk’m ¼ lqst ¼ xn qsurface of ¼ shin ¼ leg

178

Keren Rice

Prefixes are never accented even if they contain the only full vowel of the word. (37) Columbian (Czaykowska-Higgins 1997) ?ac¨p’np’n ‘windfalls, logs’ 179 ?ac-qp’n þ p’n aspect-qlong objects þ CVRED Again ignoring morphological classes, the major phonological stress in Columbian requires a L/F system, with primary accent on the rightmost full vowel in a string (within the appropriate domain, which CzaykowskaHiggins identifies as the Root). If no full vowels are available in this domain, then accent falls on the leftmost non-prefix vowel. Spokane, another Interior Salish language, while considered to have a lexical accent system by Carlson (1989) (the analysis in StressTyp), is reanalyzed by Bates and Carlson (1997: 103–104) as much like Columbian in its accent properties. In the absence of lexical marking for accent, accent is on the final full vowel, with it defaulting to the left-most nonprefix syllable if no full vowel is present. Czaykowska-Higgins and Kinkade’s second type of accent system, represented by Lillooet (Interior) and Squamish (Central), has penultimate or final accent, with weight restrictions and some morphological properties playing a role. In Lillooet, stress falls on the first full vowel (38a, b) that is not in a prefix; in the absence of a full vowel, stress is on the first (non-prefix) vowel (38c); stress is iterative. See van Eijk (1997) and Shaw (2009) for discussion. (38) Lillooet (Shaw 2009) a. ¨pun-´kan b. c’Xm-¨a-m-l’us c. ¨p’l-qw

‘I found it/him/her’ ‘blinded (when facing the sun)’ ‘to turn around’

Czaykowska-Higgins and Kinkade’s third type of system, found in Saanich (Northern Straits dialect; Central), basically has penultimate accent but with morphological properties interfering. Tillamook seems to fall in this category as well. Egesdahl and Thompson (1997: 236) note that there is a strong tendency toward penultimate accent in Tillamook, with vowel reduction and consonant adjustment; they remark that morphological classes for accent, if they exist, have yet to be worked out. The final category proposed by Czaykowska-Higgins and Kinkade includes languages which tend to show predictable accent. In Sliammon

Accent in the native languages of North America

179

(Comox; Central) accent is on the initial syllable. In Northern Lushootseed (Central) primary accent is on the leftmost non-prefix full (i.e., nonschwa) vowel or the leftmost of a sequence of schwas. Morphological factors enter into the determination of accent placement in this group as well. In general, accent in Salishan languages tends to fall on prefixes only in exceptional circumstances, and schwas tend to surface as accented only when there are no full vowels available. In most languages (Upper Chehalis and Cowlitz are exceptions) unaccented full vowels reduce to schwa or delete. StressTyp includes three Salishan languages, Southern Puget Sound (Lushootseed) (Snyder 1968), Spokane (Carlson 1989), and Squamish (Demers and Horn 1978), with the first two classified as having lexical accent and the last as having penultimate accent, at least in morphologically simple forms. However, as seen in the discussion of Columbian and Spokane, if one abstracts away from the fact that roots and su‰xes may have lexical accent, there is often phonological regularity in these systems which suggests that they are characterized as L/F. See the remarks on Southern Puget Sound Salish (Lushootseed) in StressTyp; Snyder 1968 analyzes the language as having lexical accent but, with a reanalysis of the vowel system, it can be analyzed as having predictable accent, falling in the final category proposed by Czaykowska-Higgins and Kinkade (1997). Czaykowska-Higgins and Kinkade (1997: 15) remark that the diversity in accent systems within the Salish family is reminiscent of Indo-European generally, and suggest that the area requires further study. 3.8. Penutian Penutian is a controversial superfamily (see Campbell 1997: 295–297, for instance). I will refer to the better-established family groupings rather than to the superfamily. 3.8.1. Tsimshianic Coast Tsimshian (Sm’algyax), a Coast Tsimshianic language, is classified U; P in StressTyp, with final accent. The penultimate syllable is accented in the presence of a su‰x or connective, and in some exceptional forms (39b) (thus ‘P’). This language might be characterized as U, with su‰xes and connectives outside the domain of accent. Epenthetic vowels break

180

Keren Rice

up word-final consonant clusters, and these vowels are unaccented, rendering the word-final stress placement opaque. See Dunn (1979: 7, 1995). (39) Coast Tsimshian (Sm’algyax) (Dunn 1978) a. final accent ‘dream’ 50 ks¨wo:x. ?a´di¨g. aws ‘humming bird’ 4 b. penultimate accent (exceptions to generalizations) ¨?awta ‘porcupine’ 9 nip¨?a?la ‘button’ 80 c. epenthesis and accent: opacity (Dunn 1979) ?a:dt ! ¨?a:dit ‘fisherman’ 7 dalbksk ! ¨dalbksk ‘shorten’ 7 hoyx ! ¨hoyax ‘correct’ 8 Accented short vowels may lengthen. 3.8.2. Coosan Hanis Coos is characterized as LEX(P), with ‘‘each word having its own stress accent’’ (Frachtenberg 1922). (40) Hanis Coos (Frachtenberg 1922) a. ¨helaq b. he¨laq

‘to get, to arrive’ ‘to climb up’

Goedemans notes in StressTyp that accent generally appears to be penultimate. 3.8.3. Takelman Takelman has two branches, Takelma and Kalapuyan. Central Kalapuyan, a Takelman language, has a simple system, with accent on the first syllable of the stem (Berman 1990). In Takelma accent is realized as pitch (Sapir 1912); its position is not discussed. 3.8.4. Wintuan In Wintu (I/I), accent has the stem as its domain, falling on one of the first two syllables. It is on the first heavy syllable (V:), with first syllable default (41c, d). Secondary accent falls on following heavy syllables (d, e, f ).

Accent in the native languages of North America

181

(41) Wintu (Pitkin 1984) a. b.

¨be:le:s c’i¨yi:ya

‘it could be’ 20 ‘to be all squashed up’ 19

c. d.

¨c’iyel ¨werle˙bo:sken

‘one to be squashed’ 19 ‘you will have to come’ 20

e. f.

¨keneha˙le:s ?o¨le:l˙be:s

‘it might be so’ 20 ‘God’ 20

Prefixes take secondary accent, and placement follows the pattern of primary accent. (42) Wintu (Pitkin 1984) a. b.

˙net-¨nen ˙wayti-no¨me:l

‘my mother’ 22 ‘Dog Creek’ 22

c.

pu˙ba:n-¨p’urun

‘of those others’ 22

3.8.5. Maiduan Mountain Maidu (I/S;I) has primary accent on an initial heavy (VC) syllable (43a). Accent falls on the first syllable if in Proto-Maiduan the syllable was heavy (43b); it falls on the second syllable otherwise (43c). See (44) for Proto-Maiduan forms. (43) Mountain Maidu (Shipley 1964) a. b. c.

¨jukbom ¨hukitsa wi¨setpem

‘bear dance’ 12 ‘solitary, lonely by nature’ 8 ‘frightened’ 12

The language has secondary accent and lengthening, not shown here. Two other Maiduan languages, Konkow and Nisenan, have initial accent with certain automatic shifts (Mithun 1999: 45). Eatough (1999: 4) reports that in the Central Hill Nisenan dialect, accent is word initial, realized phonetically as loudness and falling pitch, especially on long vowels. Some comparative data is given in (44). (44) Comparative Maiduan (Mithun 1999: 456) Proto-Maiduan Maidu Nisenan, Konkow a. *polo ‘buckeye’ po¨lo ¨polo b. *c’awa: ‘jaw’ c’a¨wa ¨c’awa: c. *k’o:do ‘land’ ¨k’odo ¨k’o:do

182

Keren Rice

3.8.6. Utian Utian is divided into Miwokan and Costanoan, and Miwokan into Western Miwok and Eastern Miwok. The three Miwok languages in StressTyp (Northern Sierra Miwok, Central Sierra Miwok, and Southern Sierra Miwok) are from the Eastern Miwok branch. Northern Sierra Miwok and Central Sierra Miwok are I/S systems, with accent on heavy initial syllables (V:, VC), and on the second syllable otherwise. Examples from Northern Sierra Miwok are given in (45) and from Central Sierra Miwok in (46). (45) Northern Sierra Miwok (Callaghan 1987) a. initial heavy ¨to:no‘bull’ 220 ¨poku‘to burst, explode’ 183 b. initial light pa¨latta˙ta-

‘woodpecker’ 176

(46) Central Sierra Miwok (Freeland and Broadbent 1960) a. initial heavy ¨ko:kotSu:‘clover’ 8 ¨nopta‘to drop’ 13 b. initial light ta¨na:jahe¨na-

‘evening star’ 17 ‘breath, wind’ 3

No rhythm is reported in Central Sierra Miwok, while Northern Sierra Miwok has alternating accent from the left. Southern Sierra Miwok (Broadbent 1964: 16–17) is classified as I/, with the leftmost heavy syllable (V:, VC) of the first two syllables receiving accent. Morphological factors dictate that one of the first two syllables is heavy. Broadbent (1964) reports that secondary accent occurs on succeeding heavy syllables, and in a sequence of two heavy syllables, alternating secondary accent is also found. Secondary accent is not marked here. (47) Southern Sierra Miwok (Broadbent 1964) a. hi¨sa:k ‘to hiss’ 21 b. ¨hu:»u? ‘buzzard’ 13 c. ha¨ka:»a? ‘golden cup oak’ 13 d. ¨to:ko»u? ‘ear’ 13

Accent in the native languages of North America

183

(48) Accent systems in Utian (Miwok branch) language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Northern Sierra Miwok

I/S

yes V:, VC

yes

L, iamb

no

Central Sierra Miwok

I/S

yes V:, VC

no

Southern Sierra Miwok

I/–

yes V:, VC

yes

L, trochee

yes V:, VC

In the second branch of Utian, Costanoan, Mutsun has first syllable stress with final intonation (Okrand 1977). 3.8.7. Yokutsan The Yokutsan family includes a number of languages, one of which is represented in StressTyp, the Yawelmani dialect of Valley Yokuts. In Yokuts in general, words are regularly accented on the penultimate (Newman 1944: 38). There are two exceptions. In words with the durative su‰x, accent is antepenultimate; Newman remarks that this is because this su‰x was formerly an enclitic and outside the accent domain. In some threesyllable nouns, accent is antepenultimate; this class, A-induced plurals, uses accent for grammatical purposes (Newman 1944: 28). The language is classified as P; A. The regular penultimate pattern is illustrated in (49). (49) Yawelmani (Yokuts dialect) (Newman 1944) a. ?oh¨yo:hin ‘searched for’ b. ho:ye:¨hana? ‘messenger, subject’ The Wikchamni dialect of Tule-Kaweah, or Foothills Yokuts, has the same accent pattern, with penultimate accent except for the A-induced noun plurals, in which accent is antepenultimate (Gamble 1978). 3.8.8. Plateau Penutian Two Plateau Penutian languages are represented in StressTyp, Klamath and Nez Perce. Klamath, in its own branch, is classified in StressTyp

184

Keren Rice

as L/P/A. Accent occurs on the rightmost long vowel of the word (50a). In the absence of such a vowel, the word is accented on the penultimate if it is closed (50b) or the antepenultimate if the final two syllables are open (50c). (50) Klamath (Barker 1964) a. words with long vowels n’is¨q’a:k s?o:¨di:la ga¨ba:tambli ga¨w’i:napgabli

‘little girl’ 37 ‘puts a tray of food under’ 35 ‘goes back to shore’ 36 ‘is going among again’ 37

b. words with consonant-closed syllables, no long vowels gat¨bambli c.

‘returns home’ 36

words with light syllables ¨cˇ’aw’iga ‘is crazy’ 37 ¨gepgi ‘come!’ 37

Nez Perce falls in another branch of Plateau Penutian, Sahaptian. Nez Perce is classified as having lexically marked accent, realized as high pitch and loudness. (51) Nez Perce (Aoki 1970) a. ¨tehes ‘ice’ 12 b. taqa¨k’alkt ‘to close door’ 25 c.

qu¨quke?ykt

‘to gallop’ 25

Crook (1999) treats Nez Perce as having lexical accent, with predictable stress that is obscured when the word has a lexical accent. Primary stress is assigned to accented syllables. In the absence of lexical accent, in general primary accent falls on the penultimate syllable. Secondary accent is also found, with sensitivity to weight and rhythm playing roles. Hargus (2001) and Hargus and Beavert (2001, 2002, 2006) examine Yakima Sahaptin, a Sahaptian language of the Sahaptin branch. They conclude that Yakima Sahaptin has lexically specified accent in many cases, but predictable accent also exists, falling on the first or second syllable, whichever is heavy (V:, VC), with default accent on the initial syllable. Accent is resolved in words with multiple lexically accented morphemes as follows: the rightmost lexically accented su‰x receives the primary accent; in the absence of su‰xes and the presence of prefixes, the

Accent in the native languages of North America

185

leftmost lexically accented prefix receives primary accent; in the absence of lexically accented a‰xes, a lexically accented root has primary accent. Rigsby and Rude (1996: 671) discuss Sahaptin, in the same group as Yakima. They state: ‘‘Primary stress, which is distinctive and is indicated by the acute accent, occurs on one syllable of every word. Examples of stress contrast are: a´mapa ‘husband’ (objective case) and ama´pa ‘island’ (locative case); pa´q’inusˇana ‘he saw him’ and paq’ı´nusˇana ‘they saw (him)’. Nondistinctive secondary and lesser stresses occur phonetically . . .’’ This language might be subject to an analysis combining lexical and predictable accent. 3.9. Hokan Another proposed superfamily is Hokan. This superfamily, like Penutian, is controversial, and I organize the discussion by the better-established families. 3.9.1. Karok Karok (Karuk) is analyzed as a F/L system, with accent on a long vowel if present (52a, b), and otherwise on the final syllable (53c); it is not clear where stress falls if there is more than one long vowel. (52) Karok (Bright 1957) a. b. c.

?im¨ma:n kuk¨?u:mi ?iS¨puk

‘tomorrow’ 13 ‘go there’ 13 ‘money, dentalium shells’ 13

3.9.2. Chimariko Jany (2009) reports that the root is accented in Chimariko, with accent on the penultimate syllable of a polysyllabic root. 3.9.3. Shastan Silver (1966) analyzes Shasta as having stress on heavy syllables, with penultimate stress in a sequence of same syllable types. 3.9.4. Palaihnihan There are two Palaihnihan languages, Achumawi and Atsugewi. In Atsugewi, primary accent falls on the first heavy syllable (V:, VC) after the initial one, with alternating accent thereafter (not shown) (Mithun 1999: 470 based on Talmy 1972: 443, 444).

186

Keren Rice

(53) Atsugewi (Mithun 1999, from Talmy 1972: 443, 444) ‘I stepped into a deep mud-hole’ 470 a. sm’a:s¨p’aq’thpu:ma b. st’usp’a¨q’iw

‘I stuck my hand into the mud’ 470

In Achumawi, every syllable has high or low tone, and accent is not reported (Mithun 1999: 470). 3.9.5. Pomoan Pomoan languages show diversity in accent systems, as discussed in McLendon (1973). Eastern Pomo, classified as S in StressTyp, has primary accent on the second syllable of the root, although there are exceptions with primary accent on the first syllable. McLendon (1996: 514) treats accent as phonemic since in many cases words with first-syllable accent are not synchronically analyzable morphologically. Secondary accent appears on alternating syllables thereafter (McLendon 1996: 515). McLendon further notes that all roots carry primary accent, and that some su‰xes have lexical accent. Southeastern Pomo has initial accent, and Southern Pomo has penultimate accent (McLendon 1973: 33, Mithun 1999: 474). Mithun (1999: 474) reports that in Central, Northern, and Northeastern Pomo accent falls on roots, and McLendon (1973) treats accent in these languages as phonemic. Mithun (1999: 474) notes that Central Pomo has lengthening of accented vowels as well. Kashaya, or Southwestern Pomo, is classified as S-T/F. If the first syllable of the word is a monosyllabic root that is heavy (V:, VC), it receives primary accent (54a); if a monosyllabic initial root is light, accent is on the second syllable (54b). In longer roots, the first syllable is skipped for purposes of accent, and the second (54c) or third (54d) syllable receives accent, depending on weight (sensitive to V:, VC). The roots are bolded. Buckley (1991) argues that first syllables are extrametrical unless they are monosyllabic roots (54a); see Buckley (1991) for details. Assuming first syllable extrametricality, after Buckley (1991), Kashaya can be described as having accent on the first (non-extrametrical) heavy syllable with default on the second (I/S). The brackets in (54) assume this analysis. (54) Kashaya (Oswalt 1961 [O], Buckley 1994 [B]) ‘didn’t peer around’ B 171 a. (¨kel)-mul-th b. (mo-¨mac)-mela ‘I ran in’ O 247, B 171 ‘mountain lizard’ B 176 c. kon(¨hom)t h u:nu

Accent in the native languages of North America

d.

bu(Ta¨qa) pi§(ku¨du) caqham-ala-w-ıˆm’ic-? ! ca(qha¨ma:)(lawi:)(bi?) lim’ut-an’-i ! li(bu¨ta:)du

187

‘bear’ B 175 ‘bad, ugly’ B 176 ‘start to cut downward’ B 173 ‘keep whistling’ B 173

Kashaya has a rhythm pattern, manifested as lengthening of short vowels when they are heads of their foot (Buckley 1991). The root is bolded in (55). (55) a.

mo-mul-ic’-en’-icen’-i ! (mo¨mu:)(lic’e:)(duce:)du ‘keep running all the way around (sg)’ O 220, B 172

b.

mo-ht-mul-ic’-wac’-wac’-iyic’-? ! (¨moh)(timu:)(lic’)(wac’)(wac’i:)(yi?) ‘keep running all the way around (pl)’ O 221, B 173

Pomoan accent systems are summarized in (56). (56) Accent systems in Pomoan languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Eastern Pomo

S (root)

no

yes

L

no

Southeastern Pomo

I

Southern Pomo

P

Kashaya

S-T/F

yes V:, VC

yes

L

yes V:, VC

others

root

3.9.6. Yana Yana an F/F system, has accent on the first heavy (V:, VC) syllable (57a, b), and first syllable accent in the absence of a heavy syllable (57c). (57) Yana (Sapir and Swadesh 1960) a. i¨ta:l?pa ‘head scratcher’ 34 b. hap’a¨la:maubi:wi ‘mud’ 92 c. ¨galu ‘arm’ 83

188

Keren Rice

3.9.7. Cochimı´-Yuman The Cochimı´-Yuman family is divided into Yuman and Cochimı´. Yuman languages are represented in StressTyp. Langdon (1975: 219) remarks that in Yuman languages words contain one accent, and it is on the morphological root, which consists of a single syllable. Gordon echoes this (1986: 9), and (Miller 2001: 11) adds that accent tends to fall near the end of the word since there are more prefixes than su‰xes. The accented vowel thus identifies the root. Accent is sometimes said to occur on the stem, with some prefixes included in the stem. Dieguen˜o (Mesa Grande dialect of Ipai Dieguen˜o, or Northern Dieguen˜o), classified in StressTyp as U/P, has accent on a final heavy syllable of the stem (V:, VC). As all syllables in this position are heavy, accent is final in the stem (root plus inflection). In the presence of clitics, accent is on the penultimate syllable, which is always the stem-final syllable. Mesa Grande Dieguen˜o thus could be classified as a U language, with accent taking the stem as its domain. Prefixes are unaccented. (58) Mesa Grande Dieguen˜o (Langdon 1970) a. final syllable of stem is accented ?¨mat ‘land’ 18, 47 b. accent is penultimate in the presence of a postclitic ?¨mat-i ‘on the ground’ 19, 156 ?¨n-a:-vk ‘if we go’ (?¨n-a: ‘pl. go’; -a: ‘to go’ 114) 19 The analysis that accent is stem-final is reinforced in analyses of other Yuman languages. For instance, Jamul Tipay has accent on the root (Miller 2001), where, for purposes here, the terms ‘root’ and ‘stem’ are largely interchangeable. Quechan (U) has accent on the final syllable of the stem (59a), although some su‰xes are inherently accented (59b) (Halpern 1946). In the presence of such a su‰x, the su‰x accent is primary and the stem is also phonetically stressed. The position of accent interacts with epenthesis, rendering accent placement opaque (59c). (59) Quechan (Halpern 1946) a. accent on the final syllable of the stem i:¨Do wi:u:v¨so:yk aS¨ma aS¨ma-pa

‘face, eye’ 253 ‘he curses him’ 253 ‘to sleep’ 275 ‘to be sound asleep’ 275

Accent in the native languages of North America

b.

c.

189

accenting su‰xes ˙?a¨?a¨v-u cf. ?a¨?av

‘let me hear’ 285 ‘to hear’ 275

epenthesis a¨?av-k ! a¨?avk

‘he hears’ 253

Maricopa, also a U language, has accent on the final vowel of the root. The examples in (80b, c) show that su‰xes are not counted for the purposes of accent. (60) Maricopa (Gordon 1986) a.

?a¨ve i:¨Do

‘snake’ 8 ‘face, eye’ 8

b.

hot-k ! ¨xotik

‘good-real’ 9

c.

hwet-hot-m ! ¨xwetxotim

‘red-very-real’ 9

A summary of Yuman accent patterns is given in (61). (61) Accent systems in Yuman language

primary accent type

Dieguen˜o, Mesa Grande, Quechan, Maricopa, Jamul Tipay

U (in stem domain)

3.9.8. Seri Seri is classified as F/F. Accent generally occurs on the leftmost syllable of the root (Marlett 1988: 252). (62) Seri (Marlett 1988) ¨xpeyo ‘sailfish’ 253, 277 If the root contains what Marlett terms a complex nucleus (V:, VG) it bears main accent. (63) Seri (Marlett 1988) xa¨po: ‘sea lion’ 253, 277 There are some exceptions in which accent must be lexically marked (e.g., ko¨topis ‘shellfish (species)’, Marlett 1988: 253). In verbs, the exceptions

190

Keren Rice

usually result from lexicalization of a prefix-root combination, in nouns from lexicalization of a compound. A rhythmic pattern is present, but is not illustrated here. See Marlett (2008) for a revised analysis. 3.9.9. Washoe (Washo) In Washoe, accent is assigned to stems, generally falling on the penultimate syllable of the stem although stem-final stress is possible on stemfinal long vowels; some su‰xes carry accent (Mithun 1999: 557, Yu 2008). Yu (2005, 2008) notes that stressed syllables are phonologically heavy in Washo, with a stressed syllables either having a long vowel or being closed by the first half of a geminate. 3.10. Yukian There are two Yukian languages, Yuki and Wappo. In Yuki, accent fell on the first stem syllable, and was realized as very high pitch. Secondary accent was present and could be realized as lower pitch, although the placement of secondary stress is not discussed (Mithun 1999: 574, based on Schlichter 1985). In Wappo, accent occurs on a stem vowel, and does not shift with su‰xation according to Radin’s analysis (given below); Thompson, Park, and Li (2006) based on transcriptions by Sawyer, note that stress falls on the first syllable which is not synchronically or transparently diachronically a prefix. (64) Wappo (Radin 1929) a. a¨t’a-tasa b. a¨t’a-tatIsta

‘he leaches acorns’ (root: at’a) 12 ‘he leached acorns’ 12

3.11. Esselen Esselen, or Huelel (Shaul 1995: 191), an extinct language, had penultimate accent. Su‰xes are included in the domain of stress (Shaul 1995: 199). (65) Esselen (Shaul 1995) a. pu¨tu-ki b. ya¨ki-s-ki c. ¨m-atsi

‘large’ 199 ‘large’ 199 ‘someone’s mother’ 200

Accent in the native languages of North America

191

3.12. Chumashan Chumashan consists of several languages. Accent is usually penultimate in these languages (Mithun 1999: 390). 3.13. Uto-Aztecan Uto-Aztecan is a geographically widespread family that is divided into several groups. Miller (1996: 699) describes accent patterns in the Numic group of Uto-Aztecan. ‘‘Almost all of the Numic languages have an alternating stress pattern. It is absent in Kawaiisu, present only in attenuated form in Comanche, but it is strongly represented in all the other languages. Very roughly, every other mora is stressed, with long vowels and clusters counting as two moras, short vowels as one. The stressed mora is sometimes the even mora (in Southern Paiute, sometimes the odd mora (Shoshone); sometimes counting starts at the beginning (Shoshone), sometimes at the end (Mono).’’

The three closely related Central Numic languages, Comanche, Tu¨mpisa Panamint, and Shoshone, exhibit left edge oriented primary accent, with Comanche classified as I, Panamint as I/I, and Shoshone as I/I. Comanche has initial accent (66) without quantity sensitivity. Charney (1993) notes that there are some instances of non-initial primary accent; for instance possessive and object prefixes do not take accent, see (66b). Secondary accent is alternating, as in (66a), and final vowels are unaccented. The examples in (66) show regular accent; see Charney (1993) for discussion of some other patterns. (66) Comanche (Charney 1993) ‘twelve’ 42 a. ¨waha˙tima˙to?i˙katI b. ni-¨via ‘your mother’ 40 your-mother Tu¨mpisa Panamint (I/I) has primary accent on a heavy second syllable (V:) if the first syllable has a short vowel (67a); otherwise primary accent is initial (67b, c). Alternating accent occurs thereafter, and is sensitive to length. Dayley (1989: 438) and Miller (1996: 698) note some variation in accent placement. The following forms are in orthography. (67) Tu¨mpisa Panamint (Dayley 1989) a. initial light syllable, second syllable heavy (kuk¨kwii)ppu¨h ‘smoke’ 437 (pih¨naa)(˙witu¨n) ‘bee’ 437

192

Keren Rice

b. initial heavy syllable (¨paa)(˙kantu¨n) c.

first two syllables light (¨tukum)(˙mahan)(˙ningku¨n)˙na

‘having water’ 437 ‘cook for’ 436

Single final syllables show variation, sometimes being accented and sometimes devoiced (in ‘cook for,’ for instance, accent or devoicing are options); this is not illustrated here. Shoshone (Miller 1996: 698), a Central Numic language, has alternating accent, with every odd mora accented from the beginning of the word. If the second vowel is long and the first short, primary accent usually falls on the second vowel (Miller 1996: 698). Two Southern Numic languages are included in StressTyp, Ute, represented by Southern Paiute, and Kawaiisu. Southern Paiute, like the Numic languages discussed so far, has accent determined from the left edge. Primary accent is on the first syllable if it is heavy (V:) (68a); otherwise it is on the second syllable, and accent alternates thereafter (68b); final syllables are not accented and their vowels are voiceless (not shown). The S classification of Southern Paiute in StressTyp follows from primary accent falling on the second mora, with long vowels considered to be disyllabic. Bisyllabic words have initial accent since the final vowel is not footed (68c). See the sources for discussion of gemination and devoicing. (68) Southern Paiute (Sapir 1930 unless otherwise noted) a.

first syllable is heavy (¨pa:)(v?a:x)xi ‘over the water’ 39 (¨ka:)kiyç ‘comes to sing’ Harms 1966: 234

b. first syllable is light (ma¨va)?a:xxi ‘over that’ 39 (a¨tSak)kç ‘to fall’ Harms 1966: 233, Hayes 1995: 268 (o¨ta:)pi ‘pole’ Harms 1966: 233 (nam¨pa)ni ‘my foot’ 38 tç.¨Xw i:.˙na.tç.Bı`.tSu.˙Xwai.?ı`.Nwa ‘go and ask him to tell a story’ Sapir 1930: 81, Hayes 1995: 121 c.

(¨ka(:))ni

‘house’ 39

Accent in the native languages of North America

193

In another Ute dialect, Chemehuevi, primary accent is on the second vowel of the word, with secondary accent on even numbered vowels thereafter (Press 1979: 27). Chemehuevi has lost the final vowels that are voiceless in Southern Paiute. Kawaiisu is an example of the U/P pattern, with right-edge oriented accent on the final vowel if it is heavy (V:) (69a) and penultimate accent otherwise (69b, c). (69) Kawaiisu (Zigmond, Booth, and Munro 1990: 6) a. pa:(¨ya:) ‘surface, front, wall’ 251 b. ?a(¨ta:-vi) ‘tree worm’ 186 c.

?ikç(¨vadç)

‘finger’ 189

Mono, a language of the Western Numic branch, has final accent. Within these Numic languages, there are few generalizations regarding accent patterns beyond that made by Miller (1966) – generally an alternating pattern is found. Tu¨batulabal, in its own subgroup of Uto-Aztecan, is classed as U (NMS). Accent is right-edge oriented, with final vowels, long vowels, and alternating short vowels before an accented syllable accented. No prominence di¤erences are reported. (70) Tu¨batulabal (Voegelin 1935) a. ha(¨ni:)(¨lap)

‘in the house’ 79

b. (¨tciNi)(¨yal) ti(¨Niya)(¨la:p)

‘the red thistle’ 75 ‘on the red thistle’ 75

c. pç(¨tçtpç)(¨tç:di)(¨nat)

‘he is turning it over repeatedly’ 76

d. (¨yu:u)(¨du:)(¨yu:u)(¨dat)

‘the fruit is mashing’ 76

e. wi(¨taNha)(¨tal)

‘the Tejon Indians’ 75

f.

(¨witaN)(¨hata)(¨la:ba)(¨tsu) ‘away from the Tejon Indians’ 75

Cahuilla, Cupen˜o, and Luisen˜o are three Cupan languages in the Takic group of Uto-Aztecan. The development of accent in these languages is a topic of debate; see Hill and Hill (1968) and Munro (1990). Cahuilla is an I language with some lexical accent. Primary accent falls on the rootinitial vowel in the Desert and Mountain dialects; in the Wanikik dialect, accent is word-initial, an innovative pattern. See Munro (1990: 219). Note that in (71d), papen- is a prefix. The forms in (71c, e) show that there is

194

Keren Rice

quantity sensitivity, with V:, VG, and V? contributing weight. Secondary accent is present, on alternating vowels following the accented syllable (71a, b, d) unless the first syllable is heavy (V:, V?), in which case the syllable immediately following the accented syllable receives accent (71c), and accent occurs on alternating syllables after that. Prefixes receive secondary accent (71d) on alternating syllables counting to the left from the primary accented vowel. (71) Cahuilla (Seiler 1965 unless otherwise noted) a. (¨n˜a?a)(˙tSeh) ‘sit down’ 52 b. (¨taxmu)(˙?at)

‘song’ 57

c.

‘palo verde (pl.)’ Seiler 1977: 27

(¨qa:n)(˙kitSem)

d. (˙papen)-(¨tule)(˙qale)(˙veh)

‘where I was grinding it’ 52

e.

‘the deer (objective case)’ Seiler 1977: 28

(¨su)(˙ka?)(˙ti)

Cupen˜o is analyzed in StressTyp as I;LEX. Cupen˜o preserves an historical pattern – accent is on vowels that used to be long. This is shown in (72). (72) Cupen˜o (Munro 1990) a. ¨naql ‘ear’ (Proto-Cupan: *¨na:qa-la) 240 ¨maxi-sˇ ‘acorn flour’ (Proto-Cupan *¨ma:xi-sˇ) 237 b. k¨rt ?a¨mu-l

‘bird species’ (Proto-Cupan: *qa¨r:-t) 238 ‘agave’ (Proto-Cupan *?a¨mu:l) 237

The facts of Cupen˜o accent are complex; see Hill and Hill (1968), Hill (2005), and Munro (1990). Hill (2005: 22) notes that in the majority of cases, primary stress falls on the first syllable of the root, but roots may be inherently stressed on the second or later syllables. Luisen˜o is described by Munro (1990: 224) as having the following general properties: (i) if there is a long vowel in the first or second syllable, it is accented, (ii) if not, the second syllable in nouns and the first syllable in verbs is accented. This holds of roots rather than words, and thus there are surface counterexamples where a first syllable short vowel has accent, and a second syllable long vowel does not, but the second syllable is a su‰x. The examples in (73) illustrate properties of Luisen˜o accent in nouns.

Accent in the native languages of North America

195

(73) Luisen˜o (Munro 1990) a.

accent an initial long vowel ¨ma:xi-sˇ ‘acorn flour’ 237 ¨hu:na-l ‘badger’ 219

b.

accent the second vowel if the initial vowel is short and the second vowel is long qa¨ru:-t ‘sandhill crane’ 238

c.

accent the second vowel if both vowels are short tu¨va-t (¨to:va-t) ‘pinyon’ 239 wi¨?e-t ‘grasshopper’ 242

Munro notes that Luisen˜o accent cannot be fully accounted for without an extremely abstract analysis, but that the generalizations above are overall valid (1990: 224). Hill and Hill (1968: 240) suggest that Cupen˜o and Luisen˜o generally have the same accent patterns. They argue that in terms of accent, the overall pattern of Cupen˜o and Luisen˜o is conservative in the Cupan group, while that of Cahuilla is innovative. Kitanemuk, of a second branch of Takic, the Serrano-Gabrielino branch, has accent on the first syllable (Mithun 1999: 547). Hopi shows some of the initial accentual properties found in many other Uto-Aztecan languages. It is classified as I/S, with accent on an initial heavy (V:, VC) syllable (74a), and otherwise on the second syllable (74b). In disyllabic words, accent is initial (74d), suggesting that the rightmost syllable is ignored for purposes of accent. Although secondary accent is reported, its properties are not well understood. Accent is manifested as high tone. (74) Hopi (Hopi dictionary. Hopı`ikwa Lava`ytutuveni 1998) a. initial if heavy ¨lestavi ‘roof beam’ 207 ¨sipmasmi ‘silver bracelet’ 505, 864 ¨u:tungwini ‘your name’ 864 b.

second syllable if initial is light me¨lo:ni ‘musk melon, canteloupe’ 240 ka¨wayom ‘horses’ 136, 864

196

Keren Rice

c.

first syllable in disyllabic words ¨koho ‘wood’ 145 ¨mamant ‘maidens, adolescent girls, unmarried women’ 226, 864

In the Tepiman branch, an Upper Piman language, Tohono O’odham (Papago), with no long vowels, has primary accent on the initial syllable of the word. Secondary accent falls on alternating syllables, with wordfinal accent being blocked in non-derived forms. (75) Tohono O’odham (Fitzgerald 1997) a.

¨pako˙?ola

b. ¨cˇikpan-˙dam work-one who c.

¨pimi˙ando-˙mad pepper-adding

‘Pascola dancer’ 80 ‘worker’ 84 ‘adding pepper’ 85

d. ¨ha-hai˙waç-ga-˙kam ‘ones having cattle’ 90 plural-cow-possession-one who e.

¨wa-pai˙nomI-ga ‘to own knives’ 70 plural-metal-possession

A word-initial reduplicated syllable is accented (75d, e). Fitzgerald (1997) does not identify other prefixes. Instead, she assumes, after Saxton (1982) and Hill and Zepeda (1992), that person markers that precede the stem to mark possession in nouns and objects in verbs and a stative marker are clitics. She provides two reasons – these are not accented and there is generally a pause between these and the content word. In Northern Tepehuan (Tepiman group), Woo (1970) states that accent, manifested as high tone, is on the first or second vowel of the stem, with prefixes playing no role in accent assignment. Note that the morphological constituent of stem is referred to in Northern Tepehuan; in Tohono O’odham, with a similar accent pattern, Fitzgerald (1997) states that accent has the word as its domain. The analyses are likely based on di¤erent assumptions about whether there is prefixation in the language. Southeastern Tepehuan (Tepiman group) words are accented on the first or second syllable of the stem, including reduplicated stems but ex-

Accent in the native languages of North America

197

cluding prefixes. The heavier of the first two syllables is accented; if they are equally heavy, the first is accented. Prefixes are not in the accent domain. The metrical system interacts with vowel deletion; see Willett (1982) and Rice (1992) for discussion. Both V: and VC count towards weight, with VC lighter than V:, where VC syllables are final, resulting from final vowel dropping or su‰xing (Willett 1982: 176). Accent is a combination of loudness and higher pitch. Note that the raised comma indicates a glottal stop. (76) Southeastern Tepehuan (Willett 1982) a. first syllable heavy, second light ¨vç:pç’ ‘before’ 176 ¨ba:ban ‘coyotes’ 176 b.

first syllable light, second heavy ma¨tai ‘ashes’ 177 ga¨ga:t ‘bows’ 177 sa¨poc ‘story’ 177

c.

equal weight ¨ya:tui ‘potato’ 176 ¨cusupa-’a ‘the back of his neck’ 178

The Tepehuan languages have similar accent patterns, with a left-edge of stem quantity sensitive foot (I/I), while in Tohono O’odham, quantity sensitivity is absent, and accent falls on the first syllable. Yaqui, a Taracahitan language of the Cahitan subgroup, is classified as I; S in StressTyp; Demers, Escalante, and Jelinek (1999) analyze Yaqui as having initial accent (81a, b), with forms with second syllable primary accent resulting from lexically marked initial extrametricality (81c), and they present evidence from reduplication for this type of extrametricality. The language also has secondary accent, attracted to V: and VC; this is not indicated here. (77) Yaqui (Johnson 1962) a. ¨ba:?a ‘water’ b. ¨ho?e ‘to eat/drink’ c. ba¨ka?apo ‘kind of tree’ Mayo, closely related to Yaqui, also has accent on the first or second syllable of the word. The major di¤erence between these languages is per-

198

Keren Rice

haps analytical: Yaqui generally has first syllable accent, while in Mayo the regular accent is considered to be on the second syllable (Demers, Escalante, and Jelinek 1999: 51). Nahuatl (Tetelcingo dialect), of the Aztecan branch, has penultimate accent. (78) Nahuatl (Tetelcingo) (Tuggy 1979) a. ¨tlami b. ka¨laki

‘end’ ‘enter’

Penultimate accent is the general pattern in the Nahuatl group (see articles in Langacker 1979 – Bellor and Bellor, Brockway, Sischo, Tuggy – on di¤erent Nahuatl languages). Demers, Escalante, and Jelinek (1999: 50) suggest there is a tendency in Uto-Aztecan for primary accent to be attracted to the left edge. Munro (1977, 1990: 217–218) is more detailed, proposing that Proto-Uto-Aztecan had a general second-mora accent pattern. This left-edge pattern for main accent is certainly common in the Uto-Aztecan family, but other patterns are found, as discussed above. The position of main accent is summarized in (79). In some languages, ‘heaviness’ is related to lexical marking reflecting historical weight distinctions rather than synchronic syllable weight. (79) Accent systems in the Uto-Aztecan languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Comanche, Cahuilla

I

no

yes

L, trochee

no

Kitenemuk

I

no

not reported

Tohono O’odham

I

no

yes

L, trochee

Tu¨mpisa Panamint

I/I

yes V:

yes

L, trochee

yes V:

Shoshone

I/I

(yes V:)

yes

L, trochee

?

S. Tepehuan

I/I

yes V:, VC

yes

L, trochee

yes V:, VC

Accent in the native languages of North America Southern Paiute

I/S

yes V:

yes

L, iamb

no

Hopi

I/S

yes V:, VC

yes

L

no

N. Tepehuan

I; S

no

Yaqui

I; S

yes, lexical marking

yes

L, trochee

yes V:, VC

Cupen˜o

I; LEX

no

not reported

Luisen˜o

?/S nouns ?/I verbs

no

not reported

Chemehuevi

S

no

yes

L, iamb

no

Mayo

S; I

yes, lexical marking

Mono

U

no

Tu¨batulabal

U (NMS)

no

yes

R, trochee

yes V:

Kawaiisu

U/P

yes V:

not reported

Nahuatl

P

no

not reported

199

As can be seen in (79), many Uto-Aztecan languages have alternating accent patterns, sensitive to vowel length. But, as Miller (1996) says about the Numic subgroup, the systems cannot be captured in the same terms despite the overall surface similarities – some are quantity-sensitive, some are not; some have rhythm, some do not; the metrical count is oriented towards the left edge in some and the right edge in others. 3.14. Kiowa-Tanoan Kiowa has phonemic tone, with predictable stress falling on syllables with H and HL tones and syllables with long vowels (Watkins 1984). In Jemez, tone is phonemic, while stress falls on the initial syllable of nouns and verb themes (Hale 1967: 119). 3.15. Zuni Zuni, an isolate, has primary accent on the first syllable. In long words (four or more syllables); secondary accent is on the penultimate or antepenultimate syllable (Newman 1965). Only primary accent is illustrated here.

200

Keren Rice

(80) Zuni (Miner 1986) a. ¨tepowa b. ¨ta-tepowa-nne

‘be hitched’ 247 ‘wagon’ 247

3.16. Kutenai Kutenai (P: U), an isolate, generally has penultimate accent, but may have exceptional final accent. Secondary accent is rhythmic, occurring on alternating syllables before the primary accent. Only the penultimate accent pattern is illustrated here. (81) Kutenai (Garvin 1948) a. ¨sa?n pa´ b. ˙kqaqa¨na´kqa:c c. ˙nis a˙kı`k´u¨namis

‘but indeed’ 37 ‘automobile’ 37 ‘the town, obviative’ 37

3.17. Siouan-Catawban The Siouan-Catawban family has two major branches, Siouan and Catawban, with only Siouan represented in StressTyp. The Common Mississippi Valley Siouan accent pattern has accent on the second syllable if the initial syllable is light and on the first syllable otherwise, an I/I pattern (see Hayes 1995: 356). Consider first accent in the Dakotan languages. In all documented dialects (Shaw 1985a: 175), accent falls on the second syllable of a word, and on the only syllable of monosyllabic words. Dakota (Waxpe´tu`wa˜ Santee dialect) is illustrated in (82). (82) Waxpe´tu`wa˜ Santee Dakota (Shaw 1985a) a. wa-¨kte ‘I kill’ 175 I-kill b. ma-¨ya-kte I-you-kill

‘you kill me’ 175

c.

‘you kill them there’ 175

o-¨witSha-ya-kte locative-them-you-kill

Shaw (1985a: 175) notes that vowel deletion may obscure the underlying regularity by creating surface instances of word-initial accent. In addition, postclitics do not take accent, and thus monosyllabic roots with a postverbal clitic are accented, as in (83).

Accent in the native languages of North America

201

(83) Dakota (Shaw 1985a) ¨mni-ki˛ water-the

‘the water’ 178

Shaw further reports that there is no evidence for secondary accent (Shaw 1985: 176). In Lakhota (Teton Sioux) another Dakota (Sioux) dialect, the second vowel is usually accented, realized with higher pitch and greater loudness (Rood and Taylor 1996). In compounds, the first member has primary accent and the second secondary accent. The placement of accent in Stoney Dakota indicates a development from the basic Dakota pattern of second syllable accent to an emerging pattern of final or penultimate accent (Shaw 1985b: 1). In Stoney, primary accent is on the final syllable when it is heavy (VCC) and on the penult when the final is light, making it a U/P system. There is no phonemic vowel length in Dakota and the quantity-sensitivity is to final VC vs. VCC. In addition, Stoney retains the Dakota accent rule, with secondary accent on the second syllable. Thus it is classified as S;U/P. The examples in (88a) show words with penultimate accent and without edge clusters, and the examples in (84b) show that primary accent occurs on the final vowel in the presence of a final cluster. (84) Stoney (Shaw 1985a) a.

a¨kida aki¨da-kta-tS a˙kida-¨bi-kta-tS

‘look at it’ 189 ‘he will look at it’ 189 ‘they will look at it’ 189

b.

a˙kida-¨bin-tS a˙kida-bi-¨si˛-n-tS

‘they looked at it’ 189 ‘they didn’t look at it’ 189

Generally the two stress rules coexist, but in cases of competition, the Stoney stress rule takes precedence (Shaw 1985a). Moving beyond Dakotan within Mississippi Valley Siouan, with its basic second syllable accent pattern, other languages di¤er at first glance. Chiwere (Otoe dialect) is classified as I/S, but this pattern is lexicalized synchronically. This language has second syllable accent if the initial syllable was light historically, and first syllable accent if the initial was heavy historically. Weight distinctions neutralized, and an analysis of the language as having a left edge iamb (I/S) is possible if the historical length distinction is treated as lexical accent, as in StressTyp. Winnebago forms are also given to show historical vowel length.

202

Keren Rice

(85) Chiwere (Miner 1979) a. [wa¨sose] ‘brave’ 31 (cf. Winnebago wasˇo¨sˇe) b. [¨kiDa] ‘fight’ 31 (cf. Winnebago ki:¨za) Chiwere is similar to the Dakotan group in having a basic second syllable accent pattern historically. Altshuler (2009) provides an analysis of Osage, of the Dhegihan branch of Mississippi Valley Siouan, treating it as having quantity insensitive iambs. Winnebago (Hoca 7k), closely related to Chiwere, has a complex accent system that is much discussed in the literature; see Susman (1943), Miner (1979, 1989), Hale and White Eagle (1980), Hale (1985), Halle and Vergnaud (1987), and Hayes (1995). Its StressTyp classification is unusual, T (primary accent on third syllable), under the assumption that long vowels are disyllabic. I summarize an analysis that treats long vowels as monosyllabic but bimoraic. Under this assumption, accent is on the third syllable if the initial syllable is light (86a, b), the second syllable if the initial is heavy (CV:) (86b), and the second syllable in bisyllabic words (86c). This analysis makes Winnebago similar to other Mississippi Valley languages, with accent shifted a syllable to the right. Secondary accent alternates after primary accent; the facts are complex (see sources for discussion). (86) Winnebago (Hale and White Eagle 1980 unless otherwise noted) a. all lights: initial monomoraic syllable, third syllable accent wa'i˛¨'ı´˛ ‘ball’ 118 hotSi˛¨tSi˛ni˛k ‘boy’ 118 b. initial bimoraic syllable ha:¨kitu˙dZı`k ‘I pull it taut (plain)’ 118 ha:¨kitu˙dZı`ksha7˙na7 ‘I pull it taught (declarative)’ 118 ma7:¨tSaire ‘they cut a piece o¤ ’ Miner 1979: 29, Hayes 1995: 347 c.

non-initial heavy syllables (initial syllable mono- or bi-moraic) ki¨ri:na7 ‘returned’ Halle 1990: 149, Hayes 1995: 347 bo:¨ta:na7 ‘he hit him’ Halle 1990: 149, Hayes 1995: 347

d. monosyllabic and bisyllabic words, final syllable accent ¨tSi: ‘house’ 130 wa¨je ‘dress’ 118

Accent in the native languages of North America

203

Winnebago exhibits further complexities that are summarized in Hayes (1995: 346–365). Accent in Hidatsa, a Missouri River language, appears to fall on one of the first two syllables, although there are exceptions, and additional information is required (Matthews 1965, Robinette 1955). No rhythm system is reported. Graczyk (2007: 20–23) notes that accent in Crow is phonemic, although it is largely predictable in certain verb paradigms. He notes minimal pairs such as chı´a ‘white’ and chia´ ‘extinguish’ and hu´upa ‘handle’ and huupa´ ‘shoe.’ Accented vowels are realized with high pitch, with following vowels having low pitch, and the pitch of preceding vowels dependent on the length. Mandan has primary accent on the first syllable, making it an I language, and alternating accent thereafter (Mixco 1997: 12). In terms of primary accent, if Winnebago is treated as showing initial extrametricality, second syllable accent is common in Siouan. Mandan has developed di¤erently from the other languages, with the others maintaining the basic iamb based on two light syllables and Mandan having first syllable accent instead. (87) Accent systems in Siouan languages language

primary accent type

quantity sensitivity

rhythm

Dakota, Sioux

S

no

no

Stoney

S; U/P

U/P: yes VCC S: no

no

Chiwere

I/S (lexical)

yes, lexical accent

no

Winnebago

T

no

yes

Hidatsa

S/I

Crow

lexical

Mandan

I

direction

quantity sensitivity

L, iamb

no

no

no

yes

L, trochee

3.18. Tonkawan Hoijer (1946: 292) reports that Tonkawa syllables are more or less evenly stressed. Disyllabic forms generally have heavier final stress, with penulti-

204

Keren Rice

mate accent in longer words. It is possible that Tonkawa should be interpreted as having a rhythmic accent system with accent on oddnumbered syllables. Unaccented vowels (even-numbered) delete (88a) or shorten (88b). (88) Tonkawa (Hoijer 1946) a.

vowel deletion (yamaxa- ‘to paint someone’s face’) 294 (yama)xa-o ! yamx-o’ ‘he paints his face’ 294 (ke-ya)max-o’ ! ke-ymax-o’ ‘he paints my face’ 294 (yama)xa-no’ ! yamxe-no’ ‘he is painting his face’ 294

b. vowel shortening (ka:na ‘to throw away’ 294) (xaka:)na-o’ ! xa-kan-o’ ‘he throws it far away’ 3.19. Caddoan The Caddoan family has Northern and Southern subgroups, with the former represented in StressTyp by Kitsai and Wichita. In Wichita (all/L), high tone syllables generally receive primary accent; if there are no lexically high tone syllables in a word then long vowels receive accent; if neither are present, the last voiced vowel in the word receives accent; see Rood (1996: 582). Secondary accent alternates to the left and right of main accent. (89) Wichita (Rood 1996) a. ˙nahe:¨na´rih ‘creek’ (stressed vowel also has high tone) 582 b. ni¨ye:skic¨?i:s ‘baby’ 582 c. ˙tika¨?acs ‘he is eating it’ 582 In Kitsai, primary accent is simply noted to be irregular (Bucca and Lesser 1969). Parks (1976) states that in Pawnee accent generally falls on the first syllable of nouns and particles, but that it varies in verbs and is not understood. In more recent discussion of Skiri Pawnee, Parks and Pratt (2008: 15) note that stress is not phonemic, with general emphasis on the first syllable or the following syllable if the vowel of the first syllable is short and that of the second syllable long. However, they caution that ‘‘most patterns are ultimately variable.’’

Accent in the native languages of North America

205

3.20. Chitimacha Chitimacha has first syllable accent (Swadesh 1946, Mithun 1999: 387). 3.21. Tunica The general pattern in Tunica (Haas 1946) is the following. Primary accent occurs on the first syllable with alternating accent thereafter; the final syllable does not take accent. Cases of second syllable accent can arise. Overall, Tunica is classified as I, with rhythm. 3.22. Muskogean Muskogean has three subgroups, Western Muskogean, Central Muskogean, and Creek. Choctaw (Western) is classified in StressTyp as a pitch language, with a lexical accent system determining the most prominent syllable. In addition, Choctaw has rhythm, with iterative, quantity sensitive (to V:, VC) alternating accent from the left edge (with a number of prefixes excluded). The phonetic manifestation of rhythm comes through open syllable lengthening. Rhythmically accented vowels are underlined. (90) Choctaw (Munro and Ulrich 1984 unless otherwise noted) ‘see-I (subject)’ a. pisali (pisa:)li Nicklas 1975: 242 b.

litiha-tok

(liti:)(hatok)

‘it was dirty’ 192

c.

sa-litiha-tok

(sali:)(tiha:)(tok)

‘I was dirty’ 192

d.

okca-li-li-h

(ok)(cali:)(lih)

‘I woke him up’ Ulrich 1986: 52

Pitch is assigned by di¤erent principles. Choctaw could be classified as I/S with duration as the phonetic accent correlate; however, as noted in StressTyp, the selection of the most prominent syllable in the word, or primary accent, is determined by lexical pitch accent. Chickasaw is similar to Choctaw in its accent pattern; see Gordon (2004, 2005). It is sensitive to weight (V:, VC) and exhibits rhythmic lengthening; Gordon reports that both long and lengthened vowels attract accent, with these vowels counting as longer than closed syllables (VC). Primary accent falls on a long vowel; in the absence of such a vowel, primary accent is final. Chickasaw has a lexical pitch accent system as well.

206

Keren Rice

Koasati (Central) has lexical accent (Kimball 1991). It is classified as P/U (pitch) in StressTyp. In verbs accent is morphologically determined, but in underived nouns, the placement of the accent is largely predictable: it is on the penultimate syllable if it is long (91a), and the final syllable otherwise (91b, c), thus with accent related to position and realized as pitch in nouns (Kimball 1991: 27). (91) Koasati a. pa¨la:na b. taktalas¨ba c. salak¨la

‘plate’ Kimball 1991: 28 ‘handerchief ’ Kimball 1994: 310 ‘goose’ Kimball 1994: 307

The third Muskogean subgroup is represented by Creek-Seminole, listed as Muskogee; Creek in StressTyp. Creek is classified as L(CNT, pitch). In order to determine the position of primary accent, quantity sensitive (to V:, VC) feet are built from left to right, and accent is realized as high tone on the rightmost syllable of the rightmost foot. (92) Creek-Seminole (Hayes 1995; see also Haas 1977, Martin and Johnson 2002) a. b. c. d. e. f. g. h. i.

(i¨fa) ‘dog’ 64 (ko¨toc)ko ‘mint’ 65 (o¨sa)na ‘otter’ 65 (apa)(ta¨ka) ‘pancake’ 64 (ano)(ki¨ci)ta ‘to love’ 64 (iti)(wana)(yi¨pi)ta ‘to tie each other’ 64 (hok)(¨ti:) ‘woman’ 64 (ta:)(ski¨ta) ‘to jump (sg. subject)’ 65 (ti:)(ni:t)(¨ki:) ‘thunder’ 65

Martin and Johnson’s (2002) phonetic work further demonstrates that accent is realized as pitch, and interacts with a lexical pitch accent system. Further, there are tones marking left edges of words and downstep. Creek clearly characterizes an important property of accent systems: the organizational structure of a metrical system may be required without that structure necessarily showing phonetic realization in all cases. Pitch is clearly important in the Muskogean family, and is related to accent in two branches of the family. All three branches have some rhythmic structure.

Accent in the native languages of North America

207

(93) Accent systems in Muskogean languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Choctaw

I/S (NMS)

yes V:, VC

yes

L, iamb

yes V:, VC

Koasati nouns

P/U (pitch)

yes V:

no

Creek

L (CNT, pitch)

yes V:, VC

yes

L, iamb

yes V:, VC

3.23. Natchez Natchez had accent on the penultimate vowel if it is long and the antepenult vowel otherwise (Mithun 1999: 467). It can be classified as P/A. 3.24. Yuchi (Euchee) Yuchi, an isolate, has final accent (Wol¤ 1948: 241). 3.25. Timucuan Timucuan is divided into two groups, Timucua and Tawasa. A Timucua dialect, Mocama, is represented in StressTyp. It is type I, with initial primary accent and rhythmic accent on alternating syllables thereafter; the final syllable is unaccented (Granberry 1993). 3.26. Iroquoian The Iroquoian family covers a wide geographical range. The family is divided into two major branches, Southern and Northern. The former contains one language, Cherokee, and the latter consists of a number of languages. (Mithun 1999 provides more detailed subgrouping than does Goddard 1996.) The accent systems of several Northern Iroquoian languages have received detailed study. For the Lake-Iroquoian group (Seneca, Cayuga, Onondaga, Mohawk, Oneida), Michelson (1988: 52) argues that ProtoLakes Iroquoian accent was penultimate, with lengthening of open accented syllables unless the vowel was preceded by a laryngeal consonant. An epenthetic vowel was ignored for purposes of accent, leading to surface antepenultimate accent. From this system, two major metrical patterns developed.

208

Keren Rice

Mohawk has penultimate stress, P, as does Oneida. There are shortening, lengthening, and epenthesis processes in Mohawk that render surface accent placement opaque. Accent is realized phonetically as falling or level tone. Some Mohawk forms are given in (94). (94) Mohawk (Bonvillain 1973) a. o¨nu:ta? ‘hill’ 44 b. ¨kakhwa? ‘food’ 44 The e¤ect of epenthesis is shown in (95), where epenthetic vowels are in square brackets. Accent falls on the antepenultimate syllable. (95) Mohawk (Bonvillain 1973) a. ¨a:?al[e]? b. o¨jis[e]la

‘crinoline slip’ 44 ‘window glass, window pane’ 45

In Oneida, classified in StressTyp as P, accent falls on the penultimate syllable, with lengthening of an accented open penult and accent manifested as tone (Michelson 1988: 66). Michelson assumes that Oneida underwent an accent shift which shifted, or copied, the accent from an open penult to the following syllable (Lounsbury 1942: 58–62, 1953: 96, Chafe 1977: 172). An alternative analysis, pointed out in the remarks portion of StressTyp, is P/U, with accent on a closed penultimate (VC) and, if the penultimate syllable is open, on the final syllable. This analysis is supported by surface facts. (96) Oneida (Michelson 1988) a. closed penult teyutoh¨talhos ‘she tidies up’ 67 ‘skeleton’ 67 ¨yeskn b. closed final syllable, open penult (penult is lengthened when it is open) waki:¨kwats ‘I am well-o¤ ’ 68 wak-ikwat-s 1sgPatient-be well o¤-Habitual lanitsyato:¨lats ‘he is fishing’ 68 hl(a)-an-itsy-atolat-s 3MasculineSingular-Semireflexive-fish-hunt-Habitual kashe:¨tas ‘I count, add it’ 68 k-ashet-as 1sgAgent-add, count-Habitual

Accent in the native languages of North America

c.

209

open final syllable, open penult (penult is not lengthened) wakashe¨tu ‘I already added it up’ 68 wak-ashet-u 1sgPatient-add, count-Stative

In the western Lake-Iroquoian group, Cayuga, Onondaga, and Seneca, primary accent falls on the last even numbered non-final syllable counting from the left; if the penult contains a long vowel it has primary accent even if it is an odd-numbered syllable (97b); see Michelson (1988: 90) and references therein. The long vowel in penultimate position can be traced back to the accentual system of Proto-Northern-Iroquoian. Its length is derived by Tonic Lengthening (see comments under processes and remarks in StressTyp). Cayuga is analyzed L(CNT), with accent on the final foot. (97) Cayuga (Foster 1982) a even heavy penult (ot¨ho:)we? (e˜he˜)(na¨to:)wat (ak¨ye˜:)tho? (e˜ka)(tato˜)(kw?e¨to˜:)nye˜? b.

c. d.

‘it is cold’ 61 ‘they will hunt’ 61 ‘I planted it’ 62 ‘I will make some people for myself ’ 61

odd heavy penult (he˜na:)(¨to:)was (kano˜h)(¨so:)to˜h

‘they are hunting’ 63 ‘several houses’ 63

even light penult (ka¨nes)ta?

‘board’ 62

odd light penult (tewa)(ka¨ta)we˜nye?

‘I am moving about’ 63

In Seneca, an accent, realized as higher pitch (Chafe 1996: 557), is placed on the last even-numbered non-final closed syllable or on the last even-numbered non-final open syllable immediately followed by a closed syllable. The pattern is iambic, with the syllable count from the left edge of the word (Chafe 1996: 558). Accent on vowels in open penultimate syllables is manifested as length (a trace of Proto-Iroquoian penultimate accent and lengthening), and is shown as length in (102). See Melinger (2002) for a di¤erent analysis of the Seneca system.

210

Keren Rice

(98) Seneca (Foster 1982) a.

(a¨kas)(nowæ:)to˜h

‘I am going fast’ 68

b. (ho¨ti)(no˜hso˜:)ni:h ‘Iroquois league’ 68 If no accentable syllable is present, the word bears no accent (Chafe 1977: 178, Foster 1982: 68). (99) Seneca (Foster 1982) hato:wæ:s

‘he hunts’ 68

Seneca is classified as a L/_ (CNT:tone) language in StressTyp. See Prince (1983) and Hayes (1993: 198) also for analyses of Seneca accent patterns. The Cayuga-Seneca accent pattern is assumed to be innovative, having developed from the Mohawk-type pattern in this group (see Hayes 1995: 222, based on Chafe 1977, Foster 1982, Michelson 1983, 1988). In Tuscarora, as in Mohawk and Oneida, primary accent is commonly penultimate, and the accented syllable bears a high or rising tone. (100) Tuscarora (Williams 1976) a. o¨haheh ‘road’ 289 b. oha¨ha?ke

‘on the road’ 289

Epenthesis sometimes creates antepenultimate stress, and lexical accent on the final syllable may block penultimate accent. In Cherokee, the accent system is manifested tonally (Haag 2001: personal communication), as in other Iroquoian languages. Haag proposes that accent is realized as a superhigh tone followed by a fall, with the superhigh part falling on either the penultimate or ultimate syllable of the lexical word. In (101), ‘2’ represents a low tone. (101) Cherokee (Haag 2001: personal communication) ‘meat’ a. ha2qi2yaHL b. a2wiHnaL

‘young man’

Conditions governing accent placement are not clear.

Accent in the native languages of North America

211

(102) Accent systems in Iroquoian languages language

primary accent type

quantity sensitivity

rhythm

direction

quantity sensitivity

Mohawk

P

yes, rhythmically strong

no

Oneida

P or P/U

no

no

Cayuga

L/CNT

yes, rhythmically strong

yes

L, iamb

yes V: in penult

Seneca

L/_ CNT:tone

yes, rhythmically strong

yes

L, iamb

no

Tuscarora

P

no

no

3.27. Summary 3.27.1. Accent type In the native languages of North America, a primary accent system may be unbounded or bounded; it may be quantity sensitive or not; primary accent may be on the initial, second, third, antepenultimate, penultimate, or final syllable; the position of primary accent may depend on syllable count rather than on an edge; the position of accent may be lexical or irregular as well as predictable. The languages of the continent vary, with many of the accent types found cross-linguistically occurring. However, while there is variability, some tendencies emerge. Perhaps most obvious is that many native languages of North America determine the position of primary accent from the left. This left-oriented bias is the opposite of that found in the rest or the world, and is likely linked to the predominance of iambicity in the languages of North America (see Hayes 1995: 269, Goedemans 2001 p.c.). In (103), the location of primary accent is summarized, focusing on whether the language has a left-edge or a right-edge orientation. The information is highly simplified. Several points should be noted. First is the interpretation of L(CNT) systems. While primary accent in these systems may occur on the last foot, the foot count begins at the left edge, so I classify these systems with those with accent determined from the left. Second, the language types are set up without

212

Keren Rice

regard for domain; for instance, in some cases ‘left-edge’ might mean ‘beginning of word’ and in other cases ‘beginning of stem or root.’ StressTyp, the sources, and the text should be consulted to determine the appropriate domain. Third, some languages appear in more than one category as they have more than one foot type (e.g., Stoney (Siouan)). Fourth, some languages are included even though the position of primary accent is not synchronically predictable. For instance, in Mountain Maidu (Maiduan), knowledge of historical vowel length allows the position of accent to be determined, but with the loss of length, the position is not predictable. I follow StressTyp in this. The lefthand column shows the category; in the righthand column languages in that category are listed. (103) Primary stress category Type

Languages

left-edge orientation

Eskimo-Aleut (Pacific Gulf Yupik, Central Alaskan Yupik, Central Siberian Yupik), Algonquian (Ojibwa, Delaware, Maleseet-Passamaquoddy), Algic (Wiyot), Haida (Alaskan Haida), Salishan (Sliammon, Northern Lushootseed), Wakashan (Makah, Tsishaath Nootka, Kwak’wala), Wintuan (Wintu), Plateau Penutian (Yakima), Takelman (Kalapuyan), Maiduan (Mountain Maidu, Konkow, Nisenan), Pomoan (Eastern Pomo, Southeastern Pomo, Kashaya), Yukian (Yuki), Utian (Central Sierra Miwok, Northern Sierra Miwok, Southern Sierra Miwok), Yana, Seri, Palaihnihan (Atsugewi), Uto-Aztecan (Comanche, Cahuilla, Kitenemuk, Tohono O’odham Tu¨mpisa Panamint, Shoshone, Southern Paiute, Chemehuevi, Cupen˜o, Luisen˜o, Hopi, Southeastern Tepehuan Yaqui, Mayo, Northern Tepehuan), Zuni, Siouan (Dakota, Stoney, Winnebago, Chiwere, Hidatsa, Mandan), Tonkawa, Chitimacha, Tunica, Timucuan (Mocama Timucua), Muskogean (Choctaw, Creek-Seminole), Iroquoian (Cayuga)

right-edge orientation

Algonquian (Western Abenaki, Montagnais, Menominee, Plains Cree), Athabaskan (Sekani, Slave, Witsuwit’en, Ahtna, Salcha), Haida (Masset Haida), Salishan (Saanich, Lillooet, Squamish, Columbian, Spokane), Yokutsan (Yawelmani Yokuts), Chimakuan, Pomoan (Southern Pomo), Washoe, Esselen (Esselen), Plateau Penutian (Klamath-Modoc), Chumashan languages, Tsimshian (Coast Tsimshian), Cochimı´-Yuman (Quechan, Maricopa, Dieguen˜o), Kutenai, Uto-Aztecan (Mono, Tu¨batulabal, Kawaiisu, Nahuatl) Siouan (Stoney), Natchez, Muskogean (Koasati nouns), Iroquoian (Mohawk, Oneida, Tuscarora, Seneca)

Accent in the native languages of North America

213

3.27.2. Rhythm Many North American languages exhibit rhythm, some overtly, with a phonetic realization, and others covertly, where rhythm must be invoked to determine, for instance, the placement of primary accent, whatever its phonetic realization. A summary of rhythmic patterns in the languages discussed in this article, including direction, is given in (104). (Rhythm may be present in more languages, but not noted in the descriptions.) (104) Rhythm types Rhythm

Languages

left-edge orientation

Eskimo-Aleut (Pacific Yupik, General Central Yupik, Norton Sound Yupik, Central Siberian Yupik), Algonquian (Ojibwa, Menominee, Maleseet-Passamaquoddy, Delaware), Wakashan (Tsishaath Nootka), Wintuan (Wintuan), Maiduan (Mountain Maidu), Utian (Northern Sierra Miwok, Southern Sierra Miwok), Pomoan (Eastern Pomo, Kashaya), Uto-Aztecan (Comanche, Tu¨mpisa Panamint, Shoshone, Southern Paiute, Chemehuevi, Cahuilla, Hopi, Tohono O’odham, Southern Tepehuan, Yaqui), Siouan (Winnebago, Mandan), Tonkawa, Tunica, Muskogean (Choctaw, Creek), Iroquoian (Cayuga, Seneca), Timucuan (Mocama Timucua)

right-edge orientation

Algonquian (Western Abenaki), Plateau Penutian (KlamathModoc), Uto-Aztecan (Tu¨batulabal), Zuni, Kutenai

bidirectional

Algonquian (Plains Cree), Haida (Masset Haida), Uto-Aztecan (Cahuilla), Caddoan (Wichita)

This table shows that, of the languages with identified rhythm systems, left-to-right rhythm is the most common, with smaller numbers of languages having right-to-left or bidirectional rhythm. This is not surprising, given the left-edge bias for primary accent already noted.

4. Some areal and linguistic perspectives on accent I have focussed on descriptions of accent systems in native language families of North America. In this section, I examine accent from a genetic and areal perspective. I first investigate accent patterns within a language family, asking what kind of stability there is and what kinds of languageinternal reasons exist for why diversity might develop over the course of

214

Keren Rice

time (section 4.1). I then discuss whether accent patterns might help to define geographic linguistic areas. It is widely agreed that several linguistic areas with shared typological features can be identified in North America. There is no question that accent systems are subject to di¤usion (Thomason and Kaufman 1988). What is the situation in North America (section 4.2)? I then address two linguistic questions. One concerns the relationship between primary accent and morphological structure. In a system in which morphological factors enter into determining where primary accent appears, does this lead to greater stability than would exist in a system where only phonological factors are at play (section 4.3)? Finally, I examine the phonetic realization of accent, focusing on primary accent (section 4.4). 4.1. Genetic relationships and accent There appears to be relatively little literature on accent in native North American languages from a comparative perspective. This topic has been examined in detail in the Iroquoian family (e.g., Michelson 1988 and references therein), and family patterns are discussed in the literature on Cochimı´-Yuman (Langdon 1975, Gordon 1986), Uto-Aztecan (Munro 1977, Demers, Escalante, and Jelinek 1999), Siouan (summarized in Hayes 1995), Athabaskan (Hargus and Rice 2005), Algonquian (Milligan 2005) and Wakashan (Fortescue 2007: 12). One might ask why this topic is not more frequently discussed. I examine this question by considering the stability of accent systems based on comparison of synchronic patterns. My focus is not so much ‘can a proto-system be reconstructed?’ as ‘how can synchronic systems di¤er?’ In this section then, I return to the diversity in accent systems within established families. Consider first Algonquian languages (section 3.2). Many languages of this family have iambic systems with quantity sensitivity, but not all exhibit iambs, not all exhibit quantity sensitivity, and, of those that are quantity sensitive, di¤erent syllable types contribute to quantity. In addition, some do not appear to have accent systems (e.g., Blackfoot) in the sense under consideration in this chapter. As noted in section 3.2, Hayes (1995: 222) remarks that while many Algonquian languages exhibit quanitity sensitive iambic systems, not all do, and it is not possible to reconstruct the Proto-Algonquian accent system. There are also di¤erences between the Haida dialects (section 3.4). In Masset, accent placement is determined by lexical tone and has a rightedge orientation, while in Alaskan Haida, accent is realized as tone, and has a left-edge orientation.

Accent in the native languages of North America

215

Uto-Aztecan languages (section 3.13) also show diversity with respect to accent patterns, with primary accent sometimes being determined from the left edge and sometimes from the right, with accent systems sometimes being quantity sensitive and sometimes not, and with iambs or trochees. Many, but not all, have rhythm systems. The Wakashan languages (section 3.5) share certain characteristics, but also show variation. In terms of similarity, primary accent falls on a heavy syllable near the left edge of the word. However, in Kwak’wala, the foot is unbounded, while in Makah and Tsishaath Nootka accent must be on one of the first two syllables. Fortescue (2007: 12) suggests that ProtoWakashan accent was non-distinctive, probably coinciding with the first long vowel or short vowel pus sonorant in the first two syllables. The superfamilies exhibit extensive variation in accent patterns within them, and in general insu‰cient material is available on the betterestablished subgroupings to determine whether accent is a genetically shared characteristic in these groups. Overall, with exceptions, accent systems seem to be relatively vulnerable to change, and do not appear to serve as a good diagnostic in language classification. Why might diversity of accent patterns exist within a family? One can imagine several possibilities. For instance, consider a case where a language had vowel length distinctions and a quantity sensitive accent system. Vowel length might be lost in a daughter language, but the position of accent may be maintained. A few possibilities of how a language with quantity sensitive iambs might develop if vowel length ceases to be distinctive are laid out in (105). The symbol ‘V’ indicates an accented vowel and ‘v’ an unaccented vowel. (105) proto-state *CV:Cv

*CvCV

a. development 1: neutralization of length, first syllable accent CVCv CVCv b. development 2: neutralization of length, second syllable accent CvCV CvCV c. development 3: neutralization of length, maintenance of accent CVCv CvCV In all three cases vowel length has neutralized. However, the accent patterns di¤er. In (105a), the first vowel is accented, maintaining the fact that the first vowel was accented when it was long. In (105b), the second vowel is accented, maintaining the fact that the second vowel was

216

Keren Rice

accented when the first vowel was short. And in (105c), the length contrast is lost, but the original accent patterns are kept, with accent becoming lexicalized. While it may be possible to use these three languages to reconstruct the original accent system, the original accent system itself is not maintained in any of them. This is essentially the Siouan situation: Dakota is like (105b), Chiwere like (105c), and Mandan like (105a). Several of the languages discussed are noted to be quantity sensitive to reconstructed vowel length; these languages likely have lexical accent synchronically. To consider another way in which languages may di¤er, take the Numic subgroup of Uto-Aztecan languages where, as Miller (1996) points out, patterns of alternating accent are common; other factors vary within the group. One can imagine several scenarios that might allow a number of accent systems to arise from a common ancestor. For instance, suppose every other mora is accented, and both long and short vowels exist. If primary accent is at the left edge and the foot is iambic, then accent will occur on the first vowel if it is bimoraic and the second vowel otherwise. Such an accent system could remain iambic, but become syllabically rather than moraically based, leading to accent on alternating syllables. The length distinction could neutralize, creating a situation such as that described in the above paragraph. The loss of final short vowels (and shortening of long vowels) as in Chemehuevi (Press 1979) might lead to final vowels being accented, and the edge at which footing starts could shift. Some of these developments have occurred in the Numic group, creating languages that share an alternating accent pattern but not other properties. Consider a last example, the Muskogean languages Choctaw and Creek. These show something di¤erent, namely languages with similar accent systems but di¤erent realizations of accent. Choctaw has quantity sensitive binary feet, with accent realized as lengthening of a vowel in an open syllable. Creek too has quantity sensitive binary feet; but accent is realized as high tone on the final foot without lengthening. These languages are almost identical in terms of the abstract construction of footing, but the surface patterns are di¤erent because of di¤erent realizations of the abstract accent. Metrical correspondences could conceivably be involved in establishing genetic relationships, just as, say, vowel correspondences are. It seems that comparative metrics has not been a major focus of study in the languages of the Americas, and continuing work on reconstruction and possible paths of change is certainly in order.

Accent in the native languages of North America

217

4.2. Areal relationships and stress Hayes (1995: 269) comments that it is a ‘‘curious generalization that most iambic languages are (or were) spoken in the Americas. Given that Dryer 1989 has argued that typological similarities can extend over linguistic areas that comprise entire continents (including, for example, the Americas), this geographical symmetry may not be a coincidence. On the other hand, throughout the Americas the iambic languages are interspersed with great numbers of non-iambic languages, so the idea that stress is an areal phenomenon of the Americas should not be taken as a certainty’’ (see section 3.26). Since areal patterns and borrowing involving accent are found (e.g., Thomason and Kaufman 1988: 16, 62, 75, 90, 122, 241) and North America contains regions in which languages have come to share a substantial number of features due to longstanding contact among their speakers rather than for genetic reasons (e.g., Campbell 1997, Mithun 1999: 214), it is interesting to examine whether accent patterns have been subject to dispersion within North America. Some of the areas are Northwest Coast, parts of California, Southwest, and Southeast. Languages within a linguistic area often share phonological, morphological, and syntactic features due to external rather than internal factors. It must be cautioned that linguistic areas are not necessarily easily identifiable, as not all languages in an area share all defining features of the area. The Northwest Coast is a well-known linguistic area of North America, and includes Eyak, Tlingit, Haida, Tsimshianic languages, Wakashan languages, Chimakuan languages, Coast Salishan languages, Lower Chinook, Alsea, Siuslaw, Takelma, Kalapuya, Coos, and some Pacific Coast Athabaskan. Some features that characterize this area are large consonant inventories, ejectives, a frequent velar/uvular distinction, multiple lateral consonants, complex morphophonemics, general absence of labial consonants, and tone in many languages. Turning to accent, we see diversity in metrical patterns in the Northwest Coast. Masset Haida has primary accent on the last high tone vowel or else on the last syllable, with a right-edge orientation; Alaskan Haida has accent, realized as tone, on the leftmost heavy syllable, with a left-edge orientation. Other Northwest Coast languages di¤er – Coast Tsimshian has final or penultimate accent and no rhythm system is reported. The Wakashan languages are like Masset Haida in having a quantity sensitive foot for primary accent, but di¤er in that the foot is sensitive to length rather than to tone and accent is at the left edge rather than determined from the last syllable. Kalapuya has first syllable accent. Kutenai, an isolate, has penultimate accent

218

Keren Rice

overall. Hanis Coos may have penultimate accent, although Frachtenberg (1922) considers accent to be lexical. Despite the possible shared accent pattern among some of these languages (e.g., Hanis Coos, Kutenai), they are spoken at a distance and intervening languages have di¤erent accent patterns. The Nez Perce system appears to be lexical. Accent is varied within the Salish group itself, as discussed in section 3.7. At this gross level, it is di‰cult to find much in common about accent patterns across this linguistic area. To focus on a slightly more micro-level, the Wakashan languages of Kwak’wala and Nootka and some Salish languages such as Sliammon are spoken in proximity to each other. At this level, we perhaps begin to see similarities – all involve accent systems with a left edge orientation, although in Kwak’wala the foot is unbounded, in the Nootkan dialect examined here primary accent is on the first or second syllable depending on quantity, and in Sliammon (Salish), it is on the first syllable (with some morphological properties). In another group in geographical proximity to each other, Coosan, Kalapuyan, some Athabaskan languages, and Tillamook, the patterns di¤er – Hanis Coos has lexical accent which is often penultimate, Kalapuya has initial accent, the Athabaskan languages have accent on the stem, at or near the right edge of the word, and Tillamook has basically penultimate accent but with lexical factors entering in as well. While Northwest Coast languages share many features, accent patterns do not appear to contribute to the definition of the linguistic area, at least not in a major way. More careful work on languages with a history of contact may reveal patterns that cannot be found without additional data. In addition, a more careful study of cultural areas might help one to focus on the areas in which to look for commonality in accent patterns. California includes some linguistic areas. Northern California languages share several Northwest Coast traits, as well as others not found in the Northwest Coast. For instance, Klamath, Chimariko, Yana, the Athabaskan languages of the region, the Wintuan languages, the Pomoan languages, Lake Miwok, and the Chumashan languages have three series of stops, plain, aspirated, and ejective (Mithun 1999: 316); multiple laterals are found in the Athabaskan languages, the Algic languages, Wintu, Lake Miwok, and some other languages. Pomoan, Wintu, Lake Miwok, and Maidu have distinctive voiced stops. With respect to accent, Klamath is an L/P/A system, Yana a F/F system, and Maidu has lexical accent on one of the first two syllables, while the other languages in its family (Konkow, Nisenan) have initial accent. Eastern Pomo has accent on the

Accent in the native languages of North America

219

root, and Kashaya (Pomoan) has a complex system with accent determined from the left edge. Other Pomoan languages have di¤erent systems (section 3.9.5). Chumashan languages usually have penultimate accent (Mithun 1990: 390). Again, more study is required, but on the face of it there seems to be little in common areally about these patterns. Considering secondary accent, Eastern Pomo, Maidu, and Klamath have secondary accent systems, while Kashaya and Yana are not reported to have secondary accent. At this gross level, again accent does not appear to be an areal characteristic. Languages of the Pueblo southwest include Tanoan, Zuni, Keresan, Hopi, and some Athabaskan languages, considered to be intrusive (Campbell 1997: 339). Mithun 1999: 318 notes that this is a clear culture area, with linguistic di¤usion as well. Bereznak 1995 examines the Pueblo southwest as a linguistic area, and argues that shared features demarcate the area. These include glottalized consonants (except Hopi), tones (absent in Zuni, present in the Third Mesa dialect of Hopi), final devoicing of sonorants and vowels, dual number distinctions (see Campbell 1997: 339). What about accent? Zuni, an isolate, has primary accent on the first syllable. Hopi (Uto-Aztecan) accents the initial syllable if it is heavy, and the second syllable otherwise, with a manifestation of stress as high tone. Navajo (Athabaskan) has accent on the stem, which is generally the final syllable (Rice and Hargus 2005, McDonough 1999), as do some Apache Athabaskan languages (Tuttle, 2000, 2005). Languages of the KiowaTanoan and Keresan families are not included in StressTyp, nor have I found su‰cient information to sort out their accent systems beyond a mention of Tewa as having root-initial accent (Gordon 2002: 52), and Jemez having initial accent (Hale 1967) and this requires confirmation. Watkins 1984, in a discussion of Kiowa, treats accent as predictable, being determined by lexical tonal properties and length. Based on known properties, this group has little shared in terms of accent placement. While the number of languages represented and the level of knowledge of these languages is not enough to allow one to draw definite conclusions about the existence of areal accent patterns, this survey suggests that there are not obvious shared areal characteristics involving accent systems. This is not to say that these are not present, as accent systems can be borrowed. For instance, Mackenzie (personal communication, 2001) suggests that the final accent in Montagnais (Algonquian) is a result of borrowing from French. More detailed work is required to ascertain the extent to which accent systems might have been borrowed, and the linguistic factors that favour this happening.

220

Keren Rice

4.3. Morphological complexity A striking characteristic of North American languages is their strong propensity towards intricate morphology. Words in many of the languages are often long, containing inflectional and derivational morphology; see Mithun (1999) for recent discussion. One might expect in such languages that accent might reinforce morphology, for instance, aiding the listener in identifying constituents in a morphologically complex word. StressTyp divides accent into two types, primary and rhythm. Rhythm is phonological, while primary accent can serve more than one function – it can delimit words, but it can also give cues to morphology. The codes in StressTyp specify ‘within the domain’, and in this section I return to primary accent patterns in order to determine to what extent accent has morphological as well as phonological foundations. A uniform property of the Athabaskan language family is that the stem is a domain of accent. Leer (2005) uses accent as a cue to identify the stem in cases where two analyses of a possible string exist. For instance, consider the Fort Ware Sekani (Kwadacha) words discussed in section 3.3. (106) Sekani (Fort Ware) (Hargus 2005: 406) ‘person, man’ (stem: dn) a. ¨dne ‘marmot’ (stem: de) b. d-¨de The stem vowel is underlined. In this pair, the constant that allows the placement of accent to be predicted is morphological – the vowel of the stem is accented. The function of accent to highlight the stem is a conservative characteristic. Morphology likewise plays an important role in the accent systems of many Salishan languages, helping to identify the internal structure of words. Salishan languages have prefixes and su‰xes; the basic structure of morpheme order is shown in (111), from Czaykowska-Higgins and Kinkade (1998: 23). (107) possessive/subject clitics – aspect – locative – reduplicative – ROOT – reduplicative – primary a‰xes – lexical su‰xes – transitive/intransitive/control/applicative – object – subject/possessive – aspect Morphology enters into these languages in di¤erent ways. First, accent avoids prefixes in many Salishan languages, making it necessary to know whether a morpheme is a prefix or not. Second, accent has a morphological component in that stems and lexical su‰xes may have lexical

Accent in the native languages of North America

221

accent. Not all languages of the family share these characteristics, with some having phonologically predictable accent, and the links between morphological structure and position of accent are not a cross-family characteristic. Several Uto-Aztecan languages involve morphology in their primary accent systems. These languages have prefixes. The verb stem may be complex, involving verb-verb compounding and noun incorporation (Mithun 1999: 547). Verbal su‰xes are also found. In Cahuilla, primary accent is located on the first syllable of the root in some dialects and of the word in others. In Comanche, with a basic first syllable accent pattern, possessive and objective pronouns and the indefinite subject prefix are analyzed as enclitic to noun and verb stems, and do not take accent. Other elements that precede verb roots do attract accent (e.g., instrumental), and are treated as prefixes by Charney (1993). Thus, accent is sensitive to the type of prestem morphological element. Shoshone, like Comanche, has accentneutral prefixes. Southeastern Tepehuan has accent-neutral prefixes and su‰xes, with accent falling on the heavier of the first two syllables of the stem, and in Northern Tepehuan the left edge of the stem is relevant to the positioning of accent. There seems to be a reasonably consistent pattern: prefixes are generally outside the domain of accent. In Tu¨batulabal, on the other hand, Voegelin (1935) reports that the domain of stress is the word, or perhaps something even larger than the word. Turning to Yuman, the following general structure occurs: prefix – root – su‰x – clitic. Prefixes and su‰xes do not influence accent placement (although su‰xes may bear lexical accent). Accent picks out the root, aiding access to morphology. Pomoan languages have prefixes and su‰xes. Accent systems vary – accent is initial in Southeastern Pomo and penultimate in Southern Pomo. However, in Central, Eastern, Northern, and Northeastern Pomo, sensitivity to internal structure is found, with primary accent on a root. In Kashaya too, morphology enters into the determination of accent. Eskimo-Aleut languages are basically su‰xing. Accent is determined from the left edge in these languages, and one might think that only rhythmic factors were involved. However, Jacobson (1985) remarks that in Siberian Yupik, a principle of stem-stress is found by which the vowel of the initial syllable of a word with su‰xes is lengthened if necessary to put the first accent of the word on the stem rather than on a su‰x (Jacobson 1985: 27), showing that morphological as well as phonological factors are involved. Something similar occurs in at least Dakota of the Siouan family.

222

Keren Rice

While Dakota is classified as a second syllable accent language, cases exist in which this is not so. Siouan languages have prefixes and su‰xes. Su‰xes cannot take accent. If a monosyllabic stem has a su‰x, accent will fall on the stem, the first syllable, as in (108). (108) Dakota (Shaw 1985b) ¨pte-wa ‘bu¤alo’ 4 bu¤alo-a In addition, a set of consonant-final monosyllabic roots must be augmented by the addition of a vowel; this vowel is never accented. (109) Dakota (Shaw 1985b) /puz/

¨puza

‘be dry’ 5

Thus this system is not uniform in having second syllable accent; morphological factors intervene such that accent is attracted to the root. See Shaw (1985b) for details. Muskogean languages also are sensitive to morphological structure. Martin and Johnson (2002) remark that certain su‰xes are outside the domain of accent in Creek. In some language families, primary accent is phonologically determined, and serves to delimit words rather than to signal word-internal structure. (One must exercise caution in making claims about morphology not being involved in an accent system. For instance, in a recent article that surveys quantity-insensitive accent systems, Gordon 2002: 3 notes that many of the systems in his survey are sensitive to morphology to some degree, but he does not take this into account.) In the Algonquian languages discussed in section 3.2, for instance, accent is independent of morphological structure. Recall that in the Ojibwayan group, primary accent is located on the penultimate foot of the word. The final syllable may be part of the stem as in (110a), or it may be a su‰x, as in (110b). (110) Ojibwa (Minnesota) (Valentine 2001) a. (ma¨ki)zin ‘shoe’ 57 b. (¨e:)(si¨ban)-ag ‘raccoons’ 53, 55 In Abenaki, primary accent is computed from the end of the word, and it falls on final syllables, including su‰xes. Accent patterns thus are surfacetransparent, or at least phonologically conditioned. In the Iroquoian languages discussed in this article, accent is blind to word-internal structure. Recall that in Mohawk accent targets the penulti-

Accent in the native languages of North America

223

mate syllable. This syllable has accent regardless of word-internal structure, as in (111), where the root is bolded. (111) Mohawk (Michelson 1988) a. kha¨ra:tas k-haratat-s ‘I am lifting it up a little (with a lever)’ 53 1sgAgent-life-Habitual b. wakhara¨ta:tu wak-haratat-u ‘I am holding it up’ 53 1sgPatient-lift-Stative c. teyehyatuhseray¨tahkhwa? ‘playing cards’ te-ye-hyatu-hsr-a-yŠt-a-hkw-ha? Dual-3FeminineIndefiite-write-nominalizer-joiner-put-joinerinstrumental-habitual In (111a), accent is on the penultimate syllable of the root; in (111b) it is on the final syllable of the root; in (111c) it is on a su‰x. This is a clear case of phonological determination of the placement of primary accent. The Wakashan languages are polysynthetic, with extensive su‰xation. In Nootka, which has primary accent on the second syllable if it is heavy and otherwise on the first syllable, accent is based on phonological factors alone, and primary accent can occur on either a stem or a su‰x. Kalapuya (Takelman) is another language in which accent is surface-transparent: this language has prefixes, and accent is word initial. Klamath (Plateau Penutian) has accent on the final long vowel of the word, and, in the absence of such a vowel, on the penultimate if it is closed or the antepenultimate if the final two syllables are open. Klamath has prefix and su‰x systems, accent is not sensitive to this structure, but is phonologically determined. The Maiduan languages Konkow and Nisenan have initial accent. This falls on the stem since the languages are su‰xing, but the accent system is phonologically transparent to wordinitial position. It is not surprising to find that morphologically complex languages may utilize accent to identify word-internal structure. Alternatively, accent may function to demarcate word boundaries rather than internal boundaries, without regard to word-internal structure. Combining the observations of this and the previous two sections, the languages of North America suggest that accent is vulnerable to change, but that if it interacts with morphological properties (e.g., the stem is accented, prefixes are not accented), this might aid in stability within a

224

Keren Rice

family. However, morphological conditioning (e.g., accent attracted to a stem) need not be consistent within a family. More work is required on the interaction between morphology and accent to see whether sensitivity of an accent system to morphological structure might influence the stability of the system (and, in addition, it would be worthwhile to study more deeply whether it is the analyses of the languages or the languages themselves that really di¤er). 4.4. On the phonetic manifestation of accent Accent has several phonetic correlates, with pitch, duration, and loudness being the most important. In this section, I briefly survey the major manifestations of accent reported in North America. I assume that, in the absence of comments on the phonetic realization of accent, it involves a combination of the three factors noted above; instrumental work is needed to determine how these factors interact in the di¤erent languages. In addition, research is required to determine whether the phonetic realizations of accent might be a property shared across a linguistic area even if the more abstract accent systems are not. 4.4.1. Pitch Pitch is the major phonetic correlate of accent in several North American languages. The following languages are among those that are reported to have a pitch-type manifestation of primary accent: Alaskan Haida, Creek, Western Greenlandic, Hopi, Yaqui, Northern Tepehuan, Southeastern Tepehuan, Mayo (Uto-Aztecan; Demers, Escalante, and Jelinek 1999), Koasati, Northern Pomo (stressed syllables vary in pitch depending upon syllable structure; Mithun 1999: 474), Crow (Siouan; Mithun 1999: 25), Winnebago, Cayuga and other Iroquoian languages, Bella Coola (Salishan, Czaykowska-Higgins and Kinkade 1997: 15), Quechan (primary accent is realized as a high falling tone and secondary as a high tone; Yuman, Halpern 1946), Havasupai (Hinton 1984, Redden 1966), Kickapoo (Algonquian, Voorhis 1974), Yuki (Mithun 1999: 574). Barker 1964 reports that in Klamath pitch is dependent on stress and juncture, suggesting a pitch-type realization of accent. Pitch is one of the correlates of accent in other languages. In Eastern Pomo primary accent is realized as loudness and pitch but not lengthening; secondary accent is realized as loudness (McLendon 1966). In several Salishan languages (Nooksack, Chilliwack, Halkomelem), pitch and vowel length are reported to co-occur as the major phonetic correlate of accent;

Accent in the native languages of North America

225

see Czaykowska-Higgins and Kinkade (1997: 15). In Maidu primary accent is realized as high pitch, loudness, and tenseness, and secondary accent is lower or middle pitch and loudness and lengthening. In other languages, the interaction between accent and pitch is di¤erent – accent is attracted to syllables with marked pitch. These include Masset Haida, Wichita, Kiowa (Watkins 1984), Takelma (Mithun 1999: 515, based on Sapir 1912). I have not discussed accent in prefixes of Athabaskan languages, but it is attracted to marked tone prefixes in some languages (e.g., Sekani, Hargus 2005). Other languages have lexical pitch that does not interact with the predictable accent system. These include Kashaya, some Athabaskan languages, Choctaw, and Arapaho (Algonquian). Upriver Halkomelem is reported to have pitch that developed from glottal stops; this is independent of the accent system (Susan Blake, personal communication, 2001). 4.4.2. Duration Some languages realize accent primarily through lengthening. The following languages are in this category: Choctaw/Chickasaw, Yupik group, Kashaya. These are sometimes coded as NMS in StressTyp, with lengthening being the primary cue to accent. As is well known, the reverse is often found: accent is attracted to syllables with long vowels (and closed syllables generally). This is illustrated by a number of the languages discussed in this article and I do not enumerate these here. Unaccented vowels may be lost (e.g., some Ojibwa dialects, perhaps Tonkawa short vowels, some Central Alaskan Yupik) in addition to the common reduction of unaccented vowels (e.g., some Salishan languages, Tonkawa long vowels). They may also be devoiced, as in some UtoAztecan languages. 4.4.3. Loudness In the literature on North American languages, loudness is recorded as the only correlate of accent in at least one language, secondary accent in Eastern Pomo. I have not found languages reported to have loudness as the only correlate of primary accent. 4.4.4. Summary Given the range of possible manifestations of accent – pitch alone, duration alone, a combination of these and loudness – it becomes clear, as

226

Keren Rice

Hayes (1995) and van der Hulst (this volume, chapter 1), among others, discuss, that in talking about accent it is necessary to consider two levels, abstract structures (accents) and their concrete manifestations. Pitch is especially remarked on as a phonetic correlate of accent in North American languages. It is possible that some of the languages which are analyzed as having lexical pitch might better be analyzed as having a pitch realization of accent. The relationship between abstract accent and its concrete manifestations is clearly worthy of additional study.

5. Conclusion In this survey of accent systems in native languages of North American, I have noted variation both within a family and areally with respect to factors including the position of primary accent, the role of weight, what contributes weight, and whether rhythm plays a role. There is likewise diversity in whether primary accent is basically a word-level phenomenon, used as a cue to recognize words, or whether it also involves morphology, aiding in recognition of the internal structure of the word. Similarly, diversity exists in the phonetic realization of accent, with the possibility that it is manifested purely as pitch or purely as duration, without the complex interplay of factors that often characterizes the phonetic realization of accent. Despite the diversity, the variation is limited – a left-edge bias exists for both primary accent and rhythm, and a pitch manifestation of accent appears to be common. Studies of comparative accent within the language families would be of value, with a systematic set of assumptions regarding what a word is, as would more studies of the phonetic realizations of accent. This study thus serves as a precursor to a more detailed study of accent in the native languages of North America. References Alderete, John and Tanya Bob 2005 A corpus-based approach to Tahltan stress. In Sharon Hargus and Keren Rice (editors.) Athabaskan Prosody. Amsterdam/ Philadelphia: John Benjamins. Altshuler, Daniel 2009 Quantity-insensitive iambs in Osage. International Journal of American Linguistics 75.3. 365–398.

Accent in the native languages of North America Aoki, Haruo 1970

227

Nez Perce grammar. University of California Publications in Linguistics 62. Berkeley: University of California Press. Archangeli, Diana B. 1984 Extrametricality in Yawelmani. The Linguistic Review 4. 101– 120. Barker, M. A. R. 1964 Klamath grammar. Berkeley: University of California Press. Bates, Dawn and Barry Carlson 1997 Spokane (Npoqı´nisˇcn) syllable structure and reduplication. In Ewa Czaykowska-Higgins and M. Dale Kinkade (editors.). Salish languages and linguistics. Theoretical and descriptive perspectives. Berlin: Mouton de Gruyter. 99–123. Bellor, Richard and Patricia Beller 1979 Huasteca Nahuatl. In Ronald W. Langacker (editor). Studies in Uto-Aztecan grammar volume 2. Modern Aztec grammatical sketches. SIL and University of Texas at Arlington. 199–306. Bergsland, Knut 1997 Aleut grammar. Unangam tunuganaam achixaasixˆ. Fairbanks: Alaska Native Language Center, University of Alaska Fairbanks and Knut Bergsland. Berman, Howard 1990 An outline of Kalapuya historical phonology. International Journal of American Linguistics 56. 27–59. Blevins, Juliette 2003 Yurok syllable weight. International Journal of American Linguistics 69. 4–24. Bloomfield, Leonard 1939 Menomini morphophonemics. Etudes phonologiques de´die´es a` la me´moire de N.S. Trubetzkoy. Travaux du Cercle Linguistique de Prague 8. 105–115. Reprinted in Charles Hockett (editor). 1970. 351–362; 1987. 243–254. Bloomfield, Leonard 1946 Algonquian. In Harry Hoijer (ed.) Linguistic structures in Native North America. Viking Fund Publications in Anthropology 6. Facsimile reprint 1963, New York: Johnson. Bloomfield, Leonard 1962 The Menomini language. Charles F. Hockett (editor). New Haven: Yale University Press. Bloomfield, Leonard 1975 Menomini lexicon. Charles F. Hockett (editor). Milwaukee Public Museum in Anthropology and History 3. Milwaukee: Milwaukee Public Museum. Boas, Franz 1947 Kwakiutl grammar with glossary of the su‰xes. Transactions of the American Philosophical Society, New Series, volume 37, Part 3. Philadelphia.

228

Keren Rice

Bright, William 1957 The Karok language. Berkeley: University of California Publications in Linguistics 13. University of California Press. Brittain, Julie 2000 A metrical analysis of primary stress placement in Southern East Cree. International Journal of American Linguistics 66. 181–217. Broadbent, Sylvia M. 1964 The Southern Sierra Miwok language. University of California Publications in Linguistics 38. Berkeley: University of California Press. Brockway, Earl 1979 North Pueblo Nahuatl. In Ronald W. Langacker (ed.) Studies in Uto-Aztecan grammar. vol. 2. Modern Aztec grammatical sketches. SIL and University of Texas at Arlington. 141–198. Buckley, Eugene 1991 Kashaya closed-syllable shortening and prosodic syllabification. WCCFL 10. 65–74. Buckley, Eugene 1994 Theoretical aspects of Kashaya phonology and morphology. Stanford: CSLI Publications. Callaghan, Catherine A. 1987 Northern Sierra Miwok dictionary. Berkeley: University of California Press. Campbell, Lyle 1997 American Indian languages: the historical linguistics of Native America. Oxford: Oxford University Press. Carlson, Barry F. 1989 Reduplication and stress in Spokane. International Journal of American Linguistics 55. 204–213. Chafe, Wallace 1977 Accent and related phenomena in the Five Nations Iroquois languages. In Larry Hyman (editor). Studies in stress and accent. Southern California Occasional Papers in Linguistics 5. Department of Linguistics, University of Southern California. 169–181. Chafe, Wallace 1996 Sketch of Seneca, an Iroquoian language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 551–579. Charney, Jean Ormsbee 1993 A grammatical sketch of the Comanche language. Lincoln: University of Nebraska Press. Crook, Harold David 1999 The phonology and morphology of Nez Perce stress. Ph.D. dissertation, University of California, Los Angeles. Czaykowska-Higgins, Ewa 1993 Cyclicity and stress in Moses-Columbia Salish (Nxa?amxcı´n). Natural Language and Linguistic Theory 11. 197–278.

Accent in the native languages of North America

229

Czaykowska-Higgins, Ewa 1997 The morphological and phonological constituent structure of words in Moses-Columbia Salish (Nxa?amxcı´n). In Ewa Czaykowska-Higgins and M. Dale Kinkade (editors). Salish languages and linguistics. Theoretical and descriptive perspectives. Berlin: Mouton de Gruyter. 153–195. Czaykowska-Higgins, Ewa and Dale Kinkade 1997 Salish languages and linguistics. In Ewa Czaykowska-Higgins and M. Dale Kinkade (editors). Salish languages and linguistics. Theoretical and descriptive perspectives. Berlin: Mouton de Gruyter. 1–68. Day, Gordon M. 1994 Western Abenaki dictionary. Volume 1: Abenaki – English. Ottawa: Canadian Museum of Civilization. Dayley, John 1989a Tumpisa (Panamint) Shoshone grammar. University of California Publications in Linguistics 115. Berkeley: University of California Press. Demers, Richard A. and George M. Horn 1978 Stress assignment in Squamish. International Journal of American Linguistics 44. 180–191. Demers, Richard, Fernando Escalante, and Eloise Jelinek 1999 Prominence in Yaqui words. International Journal of American Linguistics 65. 40–55. Dryer, Matthew 1989 Large linguistic areas and language sampling. Studies in Language 13. 257–292. Dunn, John Asher 1978 A practical dictionary of the Boast Tsimshian Language. Mercury Series Canadian Ethnology Service Paper Number 42. Ottawa: National Museum of Man. Dunn, John Asher 1979 A reference grammar for the Coast Tsimshian language. Mercury Series Canadian Ethnology Service Paper Number 55. Ottawa: National Museum of Man. Dunn, John Asher 1995 Sm’algyax: a reference dictionary and grammar for the Coast Tsimshian language. Seattle: University of Washington/Sealaska Heritage Foundation. Eatough, Andrew 1999 Central Hill Nisenan texts with grammatical sketch. University of California Publications in Linguistics. Berkeley: University of California Press. Egesdahl, Steven M. and M. Terry Thompson 1997 A fresh look at Tillamook (Yutye´yu) inflectional morphology. In Ewa Czaykowska-Higgins and M. Dale Kinkade (editors).

230

Keren Rice Salish languages and linguistics. Theoretical and descriptive perspectives. Berlin: Mouton de Gruyter. 235–273.

Enrico, John 1991

The lexical phonology of Masset Haida. Alaska Native Language Center Research Papers 8. Fairbanks: Alaska Native Language Center, University of Alaska Fairbanks. Fitzgerald, Colleen M. 1997 O’odham rhythms. Ph.D. dissertation, University of Arizona. Fortescue, Michael 1984 West Greenlandic. London: Croom Helm. Fortescue, Michael 2007 Comparative Wakashan dictionary. Muenchen: LINCOM Europa. Foster, Michael 1982 Alternating weak and strong syllables in Cayuga words. International Journal of American Linguistics 48. 59–72. Frachtenberg, Leo 1922 Coos. Handbook of American Indian languages 2. 297–429. Frantz, Donald 1991 Blackfoot grammar. Toronto: University of Toronto Press. Frantz, Donald and Norma Jean Russell 1995 Blackfoot dictionary of stems, roots, and a‰xes. Second edition. Toronto: University of Toronto Press. Freeland, L.S. 1951 Language of the Sierra Miwok. Memoir 6 of the International Journal of American Linguistics. Freeland, L.S. and Sylvia Broadbent 1960 Central Sierra Miwok dictionary, with texts. University of California Publications 26. Berkeley: University of California Press. Gamble, Geo¤rey 1978 Wikchamni grammar. University of California Publications in Linguistics 89. Berkeley: University of California Press. Garvin, Paul 1948 Kutenai I: phonemics. International Journal of American Linguistics 14. 37–42. Goddard, Ives 1979 Delaware verbal morphology. New York: Garland Publishing. Goddard, Ives 1982 The historical phonology of Munsee. International Journal of American Linguistics 48. 16–48. Goddard, Ives 1996 Introduction. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 1–16. Gordon, Lynn 1986 Maricopa morphology and syntax. University of California Publications in Linguistics 108. Berkeley: University of California Press.

Accent in the native languages of North America

231

Gordon, Matthew K. 2002 A factorial typology of quantity-insensitive stress. Natural Language and Linguistic Theory. 1–62. Gordon, Matthew K. 2004 A phonological and phonetic study of word-level stress in Chickasaw. International Journal of American Linguistics 70. 1– 32. Gordon, Matthew K. 2005 Intonational phonology of Chickasaw. In Sun-Ah Jun (ed.), Prosodic typology: The phonology of intonation and phrasing. Oxford University Press. 301–330. Gordon, Matthew K. and Edmundo Luna 2004 An intergenerational investigation of Hupa stress. Proceedings of the 30th Annual Meeting of the Berkeley Linguistic Society. Berkeley: Berkeley Linguistic Society. Graczyk, Randolph 2007 A grammar of Crow. Apsa´alooke Alila´au. Lincoln: University of Nebraska Press. Granberry, Julian 1993 A grammar and dictionary of the Timucua language. Tuscaloosa: University of Alabama. Grubb, David 1977 A practical writing system and short dictionary of Kwakw’ala (Kwakiutl). Ottawa: National Museums of Canada. Haas, Mary 1946 A grammatical sketch of Tunica. In Harry Hoijer et. al. (editors) Linguistics Structures of Native North America. Viking Fund Publications in Anthropology 6. Facsimile reprint 1963, New York: Johnson. 337–366. Haas, Mary 1977 Tonal accent in Creek. In Larry Hyman (ed.) Studies in stress and accent. Southern California Occasional Papers in Linguistics 4, Department of Linguistics, University of Southern California. 195–208. Hagstrom, Paul 1995 When a Passamaquoddy unstressable schwa, that’s a mora. manuscript. MIT. Hale, Kenneth L. 1967 Toward a reconstruction of Kiowa-Tanoan phonology. International Journal of American Linguistics 33.2. 112–120. Hale, Kenneth 1985 A note on Winnebago metrical structure. International Journal of American Linguistics 51. 427–429. Hale, Kenneth L. and Josie White Eagle 1980 A preliminary metrical account of Winnebago accent. International Journal of American Linguistics 46. 117–132.

232

Keren Rice

Halle, Morris 1990

Respecting metrical structure. Natural Language and Linguistic Theory 8. 149–176. Halle, Morris and Jean-Roger Vergnaud 1987 An essay on stress. Cambridge: MIT Press. Halpern, Abraham 1946 Yuma. In Harry Hoijer et al. (editors). Linguistic Structures of Native North America. Viking Fund Publications in Anthropology 6. Facsimile reprint 1963, New York: Johnson. 249–288. Hargus, Sharon 2001 Initial consonant cluster moraicity in Yakima Sahaptin. Talk presented at the Annual Meeting of the Linguistic Society of America, Washington D.C. January 2001. Hargus, Sharon 2005 Prosody in two Athabaskan languages of British Columbia. In Sharon Hargus and Keren Rice (editors). Athabaskan Prosody. Amsterdam/Philadelphia: John Benjamins. Hargus, Sharon and Virginia Beavert 2001 Initial clusters and minimality in Yakima Sahaptin. University of Washington Working Papers in Linguistics 20: 1–24. Hargus, Sharon and Virginia Beavert 2002 Predictable versus underlying vocalism in Yakima Sahaptin. International Journal of American Linguistics 68. 316–340. Hargus, Sharon and Virginia Beavert 2006 Word-initial clusters and minimality in Yakima Sahaptin. Phonology 23-1. 21–58. Hargus, Sharon and Keren Rice (editors) 2005 Athabaskan prosody. Amsterdam/Philadelphia: John Benjamins. Hayes, Bruce 1995 Metrical stress theory. Principles and case studies. Chicago: University of Chicago Press. Hill, Jane H. 2005 A grammar of Cupen˜o. University of California Publications in Linguistics 136. Berkeley: University of California Press. Hill, Jane H. and Kenneth C. Hill 1968 Stress in the Cupan (Uto-Aztecan) languages. International Journal of American Linguistics 34. 233–241. Hill, Jane and Ofelia Zepeda 1992 Derived words in Tohono O’odham. International Journal of American Linguistics 58. 355–404. Hinton, Leanne 1984 Havasupai songs: a linguistic perspective. Tu¨bingen: Gunter Narr. Hockett, Charles 1948 Potowatomi I: phonemics, morphophonemics, and morphological survey. International Journal of American Linguistics 14. 1–10.

Accent in the native languages of North America

233

Hockett, Charles 1981 The phonological history of Menominee. Anthropological Linguistics 32. 51–87. Hoijer, Harry 1946 Tonkawa. In Harry Hoijer et al. (editors). Linguistic Structures of Native North America. Viking Fund Publications in Anthropology 6. Facsimile reprint 1963, New York: Johnson. 289– 311. Holton, Gary 2005 Pitch, tone, and intonation in Tanacross. In Sharon Hargus and Keren Rice (editors). Athabaskan Prosody. Amsterdam/ Philadelphia: John Benjamins. Hopi Dictionary 1998 Hopı`ikwa Lava`ytutuveni. A Hopi-English Dictionary of the Third Mesa Dialect. Hopi Dictionary Project. Bureau of Applied Research in Anthropology, University of Arizona. Tucson: University of Arizona Press. Howard, Philip 1990 A dictionary of the verbs of South Slavey. Yellowknife, Northwest Territories: Culture and Communication. Jacobson, Steven A. 1985 Siberian Yupik and Central Yupik prosody. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 25–46. Jany, Carmen 2009 Chimariko grammar: Areal and typological perspective. University of California Papers in Linguistics 142. Berkeley: University of California Press. Johnson, J. 1962 El idioma Yaqui. Mexico City: Departmento de Investigaciones Antropologicas, Instituto Nacional de Antropologica e Historia. Kari, James 1990 Ahtna dictionary. Fairbanks: Alaska Native Language Center, University of Alaska. Kaye, Jonathan 1973 Odawa stress and related phenomena. In Glyne L. Piggott and Jonathan Kaye (eds.) Odawa language project: Second report. Centre for Linguistic Studies, University of Toronto. 42–50. Kimball, Geo¤rey 1991 Koasati grammar. Lincoln: University of Nebraska Press. Kinkade, M. Dale 1997 How much does a schwa weigh? In Ewa Czaykowska-Higgins and M. Dale Kinkade (editors). Salish languages and linguis-

234

Keren Rice tics. Theoretical and descriptive perspectives. Berlin: Mouton de Gruyter. 197–216.

Krauss, Michael 1975 St. Lawrence Island Eskimo phonology and orthography. Linguistics 152. 39–72. Krauss, Michael 1985a Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. Krauss, Michael 1985b A history of the study of Yupik prosody. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 7–24. Langacker, Ronald, editor 1979 Studies in Uto-Aztecan grammar volume 2. Modern Aztec grammatical sketches. SIL and University of Texas at Arlington. Langdon, Margaret 1970 A grammar of Dieguen˜o. The Mesa Grande dialect. University of California Publications in Linguistics 66. Berkeley: University of California Press. Langdon, Margaret 1975 Boundaries and lenition in Yuman languages. International Journal of American Linguistics 41. 218–233. Lawrence, Erma (compiler) 1977 Haida dictionary. Fairbanks: Alaska Native Language Center, University of Alaska, Fairbanks. Leer, Je¤ 1985a Prosody in Alutiiq. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 77–134. Leer, Je¤ 1985b Evolution of prosody in the Yupik languages. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 135–158. Leer, Je¤ 1985c Toward a metrical interpretation of Yupik prosody. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 159–173. Leer, Je¤ry A. 1991 The schetic categories of the Tlingit verb. Ph.D. dissertation, University of Chicago.

Accent in the native languages of North America Leer, Je¤ry 2005

235

How stress shapes the stem-su‰x complex in Athabaskan. In Sharon Hargus and Keren Rice (editors). Athabaskan prosody. Amsterdam/Philadelphia: Benjamins. 277–318. LeSourd, Philip S. 1993 Accent and syllable structure in Passamaquoddy. New York: Garland. Lounsbury, Floyd 1942 Phonology of the Oneida language. MA thesis. University of Wisconsin. Mackenzie, Marguerite 1980 Toward a dialectology of Cree-Montagnais-Naskapi. Ph.D. dissertation, University of Toronto. Marlett, Stephen A. 1988 The syllable structure of Seri. International Journal of American Linguistics 54. 245–278. Marlett, Stephen A. 2008 Stress, extrametricality and the minimal word in Seri. Linguistic Discovery 6.1. Martin, Jack B. and Keith Johnson 2002 An acoustic study of ‘tonal accent’ in Creek. International Journal of American Linguistics 68. 28–50. Matthews, George 1965 Hidatsa syntax. The Hague: Mouton. McDonough, Joyce 1999 Tone in Navajo. Anthropological Linguistics 41. 503–540. McLendon, Sally 1973 Proto Pomo. University of California Publications in Linguistics 71. Berkeley: University of California Press. McLendon, Sally 1996 Sketch of Eastern Pomo, a Pomoan language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 507–550. Melinger, Alissa 2002 Foot structure and accent in Seneca. International Journal of American Linguistics 68. 287–315. Michelson, Karin 1988 A comparative study of Lake Iroquoian accent. Dordrecht: Kluwer Academic Publishers. Miller, Amy 2001 A grammar of Jamul Tipay. Berlin: Mouton de Gruyter. Miller, Wick 1996 Sketch of Shoshone, a Uto-Aztecan language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 693–720.

236

Keren Rice

Milligan, Marianne 2005 Menominee prosodic structure. Ph.D. dissertation, University of Wisconsin, Madison. Miner, Kenneth L. 1979 Dorsey’s law in Winnebago-Chiwere and Winnebago accent. International Journal of American Linguistics 45. 25–33. Miner, Kenneth L. 1986 Noun Stripping and loose incorporation in Zuni. International Journal of American Linguistics 52: 242–254. Miner, Kenneth 1989 Winnebago accent: the rest of the data. Anthropological Linguistics 31. 148–172. Mithun, Marianne 1999 The languages of Native North America. Cambridge: Cambridge University Press. Mixco, Mauricio 1997 Mandan. Mu¨nchen: Lincom Europa. Miyaoka, Osahito 1985 Accentuation in Central Alaskan Yupik. 1985. In Michael Krauss (editor). Yupik Eskimo prosodic systems: descriptive and comparative studies. Alaska Native Language Center Research Papers Number 7. Fairbanks; Alaska Native Language Center. 51–76. Munro, Pamela 1977 Toward a reconstruction of Uto-Aztecan stress. In Larry Hyman (editor). Studies in stress and accent. Southern California Occasional Papers in Linguistics 4. 303–326. Munro, Pamela 1990 Stress and vowel length in Cupan absolute nouns. International Journal of American Linguistics 56. 217–250. Munro, Pamela and Charles Ulrich 1984 Structure-preservation and Western Muskogean rhythmic lengthening. West Coast Conference on Formal Linguistics 3. 191–202. Newman, Stanley 1944 The Yokuts language of California. New York: Viking Fund Publications in Anthropology 2. Newman, Stanley 1965 Zuni grammar. University of New Mexico Papers in Anthropology. Albuquerque: University of New Mexico. Nicklas, Thurston Dale 1975 Choctaw morphophonemics. In James Crawford (ed.) Studies in southeastern Indian languages. Athens: University of Georgia. 237–249. Okrand, Marc 1977 Mutsun grammar. Ph.D. dissertation, University of California, Berkeley.

Accent in the native languages of North America

237

Oswalt, Robert 1961 A Kashaya grammar. Ph.D. dissertation, University of California, Berkeley. Parks, Douglas 1976 A grammar of Pawnee. New York: Garland. Parks, Douglas R. and Lula Nora Pratt 2008 A dictionary of Skiri Pawnee. Lincoln: University of Nebraska Press. Pesetsky, David 1979 Menomini quantity. In Ken Safir (ed.) MIT Working Papers in Linguistics 1: Papers on syllable structure, metrical structure, and harmony processes. Cambridge, Mass.: Department of Linguistics and Philosophy, MIT. 115–139. Piggott, Glyne 1978 Algonquian and other Ojibwa dialects: a preliminary report. In William Cowan (ed.) Papers of the Ninth Algonquian Conference. Ottawa: Carleton University. 160–186. Piggott, Glyne 1980 Aspects of Odawa morphophonemics. New York: Garland. Piggott. Glyne 1983 Extrametricality and Ojibwa stress. McGill Working Papers in Linguistics 1. Montreal: McGill University. 80–117. Pitkin, Harvey 1984 Wintu grammar. University of California Publications in Linguistics 94. Berkeley: University of California Press. Press, Margaret L. 1979 Chemehuevi. A grammar and lexicon. University of California Publications in Linguistics 92. Berkeley: University of California Press. Prince, Alan 1983 Relating to the grid. Linguistic Inquiry 14. 19–100. Proulx, Paul 1984 Proto-Algic 1: phonological sketch. International Journal of American Linguistics 50. 165–207. Radin, Paul 1929 A grammar of the Wappo language. University of California Publications in American Archaeology and Ethnology 27. 1–194. Redden, James 1966 Walapai I: phonology. International Journal of American Linguistics 32. 1–16. Rice, Curtis 1992 Binarity and ternarity in metrical theory: parametric extensions. Ph.D. dissertation, University of Texas. Rice, Keren 1989 A grammar of Slave. Berlin: Mouton de Gruyter.

238

Keren Rice

Rice, Keren and Sharon Hargus 2005 Introduction. In Sharon Hargus and Keren Rice (eds.), Athabaskan prosody. Amsterdam/Philadelphia: John Benjamins. Rigsby, Bruce and Noel Rude 1996 Sketch of Sahaptin, a Sahaptian language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 666–692. Rischel, Jørgen 1974 Topics in West Greenlandic phonology. Regularities underlying the phonetic appearance of word forms in a polysynthetic language. Copenhagen: Akademisk Forlag. Robinett, Florence 1995 Hidatsa I: morphophonemics. International Journal of American Linguistics 21. 1–7. Robins, Robert H. 1958 The Yurok language: grammar, texts, lexicon. University of California Publications in Linguistics 15. Berkeley: University of California Press. Rood, David S. 1996 Sketch of Wichita, a Caddoan language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 580–608. Rood, David S. and Allan R. Taylor 1996 Sketch of Lakhota, a Siouan language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 440–482. Sapir, Edward and Morris Swadesh 1960 Yana dictionary. Mary Haas, editor. University of California Publications in Linguistics 22. Berkeley: University of California Press. 1–267. Sapir, Edward 1912 The Takelma language of southwestern Oregon. Extract from Handbook of American Indian languages. Reprinted 1922 in Franz Boas, editor. 1–296. Reprinted 1990 in Victor Golla, editor. Sapir, Edward 1930 Southern Paiute. Proceedings of the American Academy of Arts and Sciences. 65. Saxton, Dean 1982 Papago. In Ronald Langacker (ed.). Studies in Uto-Aztecan grammar 3. 92–266. Seiler, Hans-Jakob 1965 Accent and morphophonemics in Cahuilla and in Uto-Aztecan. International Journal of American Linguistics 31. 50–59.

Accent in the native languages of North America

239

Seiler, Hans-Jakob 1977 Cahuilla grammar. Banning, California: Malki Museum Press. Shaul, David Leedom 1995 The Huelel (Esselen) language. International Journal of American Linguistics 61. 191–239. Shaw, Patricia A. 1985a Modularization and substantive constraints in Dakota lexical phonology. Phonology Yearbook 2. 173–202. Shaw, Patricia A. 1985b Coexistent and competing stress rules in Stoney (Dakota). International Journal of American Linguistics 51. 1–18. Shaw, Patricia A. 2009 Weight constraints as prosodic heads. Talk presented at the Canadian Linguistic Association, May 2009. Shaw, Patricia A., Susan J. Blake, Jill Campbell, and Cody Shepherd 1999 Stress in hnq’min’¼m’ (Musqueam) Salish. In Proceedings from the 4th Workshop on the Structure and Constituency of the Languages of the Americas. UBC Working Papers in Linguistics, vol. 2. Shipley, William W. 1964 Maidu grammar. University of California Publications in Linguistics. Berkeley: University of California Press. Silver, Shirley 1966 The Shasta language. Ph.D. dissertation, University of California, Berkeley. Sischo, William R. 1979 Michozca´n Nahuatl. In Ronald W. Langacker (editor). Studies in Uto-Aztecan grammar volume 2. Modern Aztec grammatical sketches. SIL and University of Texas at Arlington. 307–380. Snyder, Warren A. 1968 Southern Puget Sound Salish: phonology and morphology. Sacramento Anthropological Society Papers 8. Stonham, John 1999 Aspects of Tsishaath Nootka phonetics and phonology. LINCOM Studies in Native American Linguistics 32. Munich: LINCOM EUROPA. Stowell, Timothy 1979 Stress systems of the world, unite! MIT Working Papers in Linguistics 1. 51–76. Susman, Amelia 1943 The accentual system of Winnebago. Ph.D. dissertation, Columbia University. Swadesh, Morris 1946 Chitimacha. In Harry Hoijer (editor). Linguistic structures of Native America. New York: Viking Fund Publications in Anthropology. 312–336.

240

Keren Rice

Talmy, Leonard 1972 Semantic structures in English and Atsugewi. Ph.D. dissertation, University of California, Berkeley. Teeter, Karl 1964 The Wiyot language. University of California Publications in Linguistics 37. Berkeley: University of California Press. Teeter, Karl 1971 The main features of Malecite-Passamaquoddy grammar. In Jesse Sawyer (editor). Studies in American Indian languages. University of California Publications in Linguistics 65. Berkeley: University of California Press. 191–249. Teeter, Karl and Philip LeSourd 1983 Vowel length in Malecite. In William Cowan (ed.), Actes du quatorzie`me congre`s des algonquinistes. Ottawa: Carleton University. 245–248. Thomason, Sarah Grey and Terrence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thompson, Sandra A., Joseph Sung-Yul Park, and Charles N. Li 2006 A reference grammar of Wappo. University of California Publications in Linguistics 138. Berkeley: University of California Press. Todd, Evelyn 1970 A grammar of the Ojibwa language: the Severn dialect. Ph.D. dissertation, University of North Carolina, Chapel Hill. Tuggy, David H. 1979 Tetelcingo Nahuatl. In Ronald W. Langacker (ed.) Studies in Uto-Aztecan grammar. vol. 2. Modern Aztec grammatical sketches. SIL and University of Texas at Arlington. Tuttle, Siri G. 1998 Metrical and tonal structures in Tanana Athabaskan. Ph.D., University of Washington. Tuttle, Siri 2000 Locations and realizations of prominence in Athabaskan languages. Athabaskan Language Conference, June 2000. Moricetown, British Columbia, Canada. Tuttle, Siri 2005 Duration, intonation, and prominence in Apache. In Sharon Hargus and Keren Rice (eds.), Athabaskan Prosody. Amsterdam/ Philadelphia: John Benjamins. Ulrich, Charles 1986 Choctaw morphophonology. Ph.D. dissertation, University of California Los Angeles. Valentine, J. Randolph 1994 Ojibwe dialect relationships. Ph.D. dissertation, University of Texas, Austin.

Accent in the native languages of North America

241

Valentine, J. Randolph 2001 Nishnaabemwin reference grammar. Toronto: University of Toronto Press. van Eijk, Jan. 1997 [1985] The Lillooet language: Phonology, morphology, syntax. Native Languages Series, volume 1. Vancouver: UBC Press. [Ph.D. dissertation, University of Amsterdam]. Voegelin, Carl 1935 Tu¨batulabal grammar. University of California Publications in American Archaeology and Ethnology 34. 55–190. Voorhis, Paul 1974 Introduction to the Kickapoo language. Language Science Monographs 13. Bloomington: Indiana University Research Center for the Language Sciences. Watkins, Laurel 1984 A grammar of Kiowa. Lincoln: University of Nebraska Press. Willett, Elizabeth 1982 Reduplication and accent in Southeastern Tepehuan. International Journal of American Linguistics 48. 168–184. Williams, Marianne Mithun 1976 A grammar of Tuscarora. New York: Garland. Wolfart, H. C. 1996 Sketch of Cree, an Algonquian language. In Ives Goddard (ed.) Handbook of North American Indians, volume 17. Languages. Washington: Smithsonian Institution. 390–439. Wol¤, Hans 1948 Yuchi phonemes and morphemes, with special reference to person markers. International Journal of American Linguistics 14.4. 240–243. Woo, Nancy 1970 Tone in Northern Tepehuan. International Journal of American Linguistics 36. 18–30. Yu, Alan C.L. 2005 Quantity, stress, and reduplication in Washo. Phonology 23.3. 437–475. Yu, Alan C.L. 2008 The phonetics of quantity alternation in Washo. Journal of Phonetics 36. 508–520. Zigmond, Maurice L., Curtis G. Booth, and Pamela Munro 1990 Kawaiisu: a grammar and dictionary. University of California Publications in Linguistics 119. Berkeley: University of California Press.

242

Keren Rice

Consensus classification of the Native Languages of North America From: Ives Goddard. 1996. Introduction. In Ives Goddard (editor). Handbook of North American Indians. Volume 17. Washington: Smithsonian Institution. 4–8. See the source for notes on fonts. The dialect level is omitted below. 1) ESKIMO-ALEUT Eskimoan Sirenikski-Yupik: Sirenikski, Naukanski Yupik, Central Siberian Yupik, Central Alaskan Yupik, Pacific Yupik (Alutiiq) Inuit-Inupiaq: Seward Peninsula Inupiaq, North Alaska Inupiaq, Western Canadian Inuit, Eastern Canadian Inuit, Western Greenlandic, East Greenlandic Aleut: Aleut (Western, Eastern), Copper Island Aleut Creole 2) ALGIC Algonquian Blackfoot Cree-Montagnais Cree: Plains Cree, Mitchif, Woods Cree, Western Swampy Cree, Eastern Swampy Cree, Attikamek Montagnais-Naskapi: East Cree, Naspaki, Montagnais Arapahoan Arapaho-Gros Ventre: Gros Ventre (Atsina), Besawunena, Arapaho Nawathinehena Cheyenne Menominee Ojibwayan Northern Ojibwa: Severn Ojibwa, Northern Algonquin Southern Ojibwa: Saulteaux, Central Southern Ojibwa, Eastern Ojibwa, Old Algonquin, Ottawa Potawatomi Sauk-Fox-Kickapoo: Sauk-Fox, Kickapoo Shawnee Miami-Illinois Eastern Algonquian Micmac Abenakian: Maliseet-Passamaquoddy, Eastern Abenaki, Western Abenaki

Accent in the native languages of North America

243

Etchemin Southern New England: Massachusett-Narragansett, Loup, Mohegan-Pequot, Quiripi-Unquachog Delawaran: Mahican, Munsee Delaware, Unami Delaware Nanticoke-Conoy Virginia Algonquian Carolina Algonquian Ritwan Wiyot, Yurok 3) NADENE Tlingit Athapaskan-Eyak Eyak Athapaskan Southern Alaska: Ahtna, Tanaina Central Alaska-Yukon Koyukon-Ingalik: Ingalik, Holikachuk, Koyukon Tanana-Tutchone: Upper Kuskokwim (Kolchan) Tanana: Lower Tanana, Tanacross, Upper Tanana Tutchone: Northern Tutchone, Southern Tutchone Kutchin-Han: Kutchin, Han Northwestern Canada Cordillera Central Cordillera: Tagish, Tahltan, Kaska Southeastern Cordillera: Sekani, Beaver Mackenzie Slavey-Hare: Slavey, Mountain, Bearlake, Hare Dogrib Chipewyan Central British Columbia Babine-Carrier: Babine, Carrier Chilcotin Nicola Tsetsaut Sarcee Pacific Coast Athapaskan Kwalhioqua-Clatskanie Oregon Athapaskan: Upper Umpqua, Tututni, GaliceApplegate, Tolowa

244

Keren Rice

California Athapaskan: Hupa, Mattole, Eel River, Cahto Apachean Western Apachean: Navajo, Western Apache, MescaleroChiricahua Eastern Apachean: Jicarilla, Lipan Kiowa Apache 4) Haida 5) WAKASHAN Kwakiutlan (Northern Wakashan): Haisla, HeiltsukOowekyala, Kwakiutl (Kwak’wala) Nootkan (Southern Wakashan): Nootka, Nitinaht, Makah 6) CHIMAKUAN Chemakum, Quileute 7) SALISHAN Bella Coola Central Salish: Comox (Sliammon, Island Comox), Pentlach, Sechelt, Squamish, Halkomelem, Nooksack, Northern Straits, Clallam, Lushootseed, Twana Tsamosan: Quinault, Lower Chehalis, Upper Chehalis, Cowlitz Tillamook Interior Salish: Lillooet, Thompson, Shuswap, Okanagan, Kalispel, Coeur d’Alene, Columbian Proposed Penutian Superfamily (families 8–18) 8) TSIMSHIANIC Coast Tsimshian, Nass-Gitksan 9) CHINOOKAN Lower Chinook, Upper Chinook 10) ALSEAN Alsea, Yaquina 11) Siuslaw 12) COOSAN Hanis, Miluk 13) TAKELMAN Takelma, Kalapuyan, Yoncalla

Accent in the native languages of North America

245

14) WINTUAN Wintu-Nomlaki, Patwin 15) MAIDUAN Maidu, Konkow, Nisenan 16) UTIAN Miwokan Western Miwok: Coast Miwok, Lake Miwok Eastern Miwok: Bay Miwok, Plains Miwok, Northern Sierra Miwok, Central Sierra Miwok, Southern Sierra Miwok Costanoan Northern Costanoan: Karkin, Ramaytush, Chochenyo, Tamyen, Awaswas, Chalon Southern Costanoan: Mutsun, Rumsen 17) Yokutsan Palewyami, Buena Vista, Tule-Kaweah, Kings River, Gashowu, Valley Yokuts 18) PLATEAU PENUTIAN Klamath Sahaptian: Sahaptin, Nez Perce Molala Proposed Hokan Superfamily (families 19–28) 19) Karok 20) Chimariko 21) SHASTAN Shasta, New River Shasta, Okwanuchu, Konomihu 22) Palaihnihan Achumawi, Atsugewi 23) Pomoan Northeastern Pomo, Southeastern Pomo, Eastern Pomo, Northern Pomo, Central Pomo, Southern Pomo, Kashaya 24) Yana 25) SALINAN Antoniano, Miguelen˜o 26) COCHIMI-YUMAN

246

Keren Rice

Yuman Pai: Paipai, Upland Yuman River: Mohave, Quechan, Maricopa Delta-California: Diegueno, Cocopa Kiliwa Cochimi: Borjen˜o, Ignacien˜o, Cadegomen˜o, Laimo´n, Monqui-Didiu 27) Seri 28) Washoe 29) YUKIAN Yuki, Wappo 30) Esselen 31) CHUMASHAN Obispen˜o, Purisimen˜o, Inesen˜o, Barbaren˜o, Venturen˜o, Island Chumash 32) UTO-AZTECAN Numic Western Numic: Mono, Northern Paiute-Bannock Central Numic: Panamint, Shoshone, Comanche Southern Numic: Kawaiisu, Ute Tubatulabal Takic Serrano-Gabrielino: Serrano, Kitanemuk, Gabrielino Cupan: Cahuilla, Cupen˜o, Luisen˜o Tataviam (?) Hopi Tepiman: Upper Piman, Lower Piman, Northern Tepehuan, Southern Tepehuan Taracahitan Tarahumaran: Tarahumara, Guarijı´o Opatan: Opata, Eudeve Cahitan (Yaqui, Mayo) Tubar 33) KIOWA-TANOAN Kiowa, Jemez, Tiwa, Tewa, Piro 34) KERESAN Acoma-Laguna, Rio Grande Keresan

Accent in the native languages of North America

35) Zuni 36) Guaicura 37) OTOMANGUEAN 38) COAHUILTECO 39) COMECRUDAN Comecrudo. Mamulique, Garza 40) Cotoname 41) Aranama 42) Solano 43) Maritino 44) Quinigua 45) Naolan 46) Karankawa 47) Kootenai 48) Cayuse 49) SIOUAN-CATAWBA Siouan Missouri River: Hidatsa, Crow Mandan Mississippi Valley Dakotan: Sioux, Assiniboine, Stoney Dhegiha: Omaha-Ponca, Osage, Kansa, Quapaw Chiwere-Winnebago: Chiewere, Winnebago Ohio Valley (Southeastern): Ofo, Biloxi, Tutelo Catawban: Catawba, Woccon 50) Tonkawa 51) CADDOAN Northern Caddoan: Arikara, Pawnee, Kitsai, Wichita Southern Caddoan: Caddo, Yatasi (?) 52) Adai 53) ATAPAKAN Western, Atakapa, Eastern Atakapa

247

248

Keren Rice

54) Chitimacha 55) Tunica 56) MUSKOGEAN Western Muskogean: Choctaw, Chickasaw Central Muskogean: Hitchiti, Apalachee, Alabama, Koasati Creek (Creek and Seminole) 57) Natchez 58) Yuchi 59) TIMUCUAN Timucua, Tawasa 60) Calusa 61) IROQUOIAN Northern Iroquoian: Tuscarora, Nottoway, Huron, Laurentian, Seneca, Cayuga, Onondaga, Susquehannock, Mohawk, Oneida Cherokee 62) Beothuk

6.

The Languages of Middle America

Harry van der Hulst, Keren Rice, and Leo Wetzels 1. Introduction1 In this chapter, we present a survey of word prosodic systems in the languages of Middle America (Campbell 1997: Chapter 5). Middle America includes Central America and Mexico. The term ‘Middle America’, referring to a geographical unit, is not synonymous to ‘Mesoamerica,’ which refers to a culture area ‘‘defined on the basis of common characteristics that were present during the conquest times’’ (Sua´rez 1983: 11). Bearing this di¤erence in mind, we here o¤er a survey of the families and isolate languages that Sua´rez groups within the Mesoamerican group and that Campbell (1997) considers in his chapter on ‘Middle American’ languages; this does not include the ‘West Indies’ (Antilles, Bahamas, Turks and Caico islands) or Cuba.

2. The language families Some language families fall almost completely within the Middle American region, while others only have a few representatives. To the former group belong: (1)

– Oto-Manguean (section 5.1.) – Mixe-Zoquean (section 5.2.) – Totonacan (section 5.3.) – Tequistlatecan (section 5.4.) – Mayan (section 5.5.) – Misumalpan (section 5.6.)

1. During the writing of this chapter we have received useful input from various people, including Lyle Campbell, Esther Herrera, Diane Hintz, Jean-Le´o Le´onard, Inga McKendry, Irina Monich, Keith Snider, and Rebecca Yarrish.

250

Harry van der Hulst, Keren Rice, and Leo Wetzels

We adopt here a generally accepted classification; see Sua´rez (1983) and Campbell (1997) for detailed discussion as well as motivation and proposals for alternatives, involving some so-called macro-families. There also are a number of isolates in this region, among which the following are the better-known ones (see section 5.7.): (2) Extant isolates: Jicaque (Tol), Seri, Huave, Tarascan, Xincan Extinct isolates: Cuitlatec, Naolan, Maratiton, Guaicurian (a family), Alagı¨lac We will also include some languages spoken in this region that belong to language families that have most of their representatives outside Middle America, such as (to the north) the Aztecan languages of the Uto-Aztecan group (section 5.8.) and (to the south) some members of the Chibchan family (section 5.9.).

3. Our sources The present survey is based on a variety of sources. StressTyp contains information on only 15 Middle American languages. We complement this material with additional information from the literature on Middle American languages. Needless to say, we have been unable to consult all the sources. We hope that the present survey will encourage researchers to bring additional work to our attention so that, eventually, the information presented here and much additional information can be added to the StressTyp database.

4. Types of systems Like in many other parts of the world, the word prosodic systems of languages in this area fall in two broad categories. Many languages use pitch properties at the word level; cf. Weidert (1981), Sua´rez (1983), Yip (2002) and numerous studies by Kenneth Pike and his collaborators. These languages can roughly be grouped in two further categories. In a first category, pitch is used distinctively, with all or specific syllables having one of a set of distinctive tones (level tones such as H, L, M and, sometimes, contour tones). Such languages can be ‘fully tonal’, with tonal contrast on every syllable, or display characteristics of ‘restricted tone languages’ in that tone contrast occurs on only few syllable, sometimes only one in the word (Voorhoeve 1973). In the latter case, it is often sug-

The Languages of Middle America

251

gested that the syllable that attracts tonal contrast is accented, the accent being merely an anchor for tones or, in addition, showing other cues such as ‘stress’ (duration, intensity). Instead of this dependency of tone on accent, other tonal systems may display a dependency that goes in the other direction such that a notion of accent or stress is said to be dependent on the tonal structure of the word. Naturally, if a system is both tonal and has stress or accent, the two may be independent from each other (cf. Hyman 1977, 2006, van der Hulst, this volume; Wetzels and Meira, this volume). A second group contains languages that do not have distinctive tones, but rather choose pitch as the exclusive or dominant cue for ‘prominence’. Languages of this type are often called ‘pitch-accent’ languages (or nonstress accent languages; Beckman 1986), as opposed to stress-accent languages where the notion ‘stress’ is taken to refer to a set of phonetic cues within which pitch is not the only player. As pointed out in Wetzels and Meira (this volume) in stress-accent languages, pitch rather than duration may be a more important cue (possibly combined with a third common cue, ‘intensity’) when the language has contrastive vowel length, whereas duration may be a preferred cue in stress-accent languages that have distinctive tones. We thus end up with the following typology of word prosodic systems: (3) a. Tonal languages i. no accent ii. accent present but not related to tone iii. accent present and in a relationship with tone: a. tone location dependent on accent (‘‘tonal accent language’’) b. accent location dependent on tone b. Non-tonal languages i. no accent a. with stress b. without stress ii. accent a. with ‘stress’ as cue: ‘‘stress-accent language’’ b. with pitch as cue: ‘‘pitch-accent language’’2 c. with duration as cue: ‘‘duration-accent language’’ d. with . . . . 2. We always have to be aware of di¤erent uses of terminology, and of course di¤erent ways of analyzing any given system. Dol and van Zanten (this volume)

252

Harry van der Hulst, Keren Rice, and Leo Wetzels

We will use the term ‘accent’ for a lexical mark (predicable or unpredictable) of syllables that are somehow ‘special’, and ‘stress’ for a metrical structure and its associated phonetic cues. As shown in (3) non-tonal, nonaccentual languages may have stress or no stress (see van der Hulst, this volume; to appear) for full discussion. In this chapter, we mainly deal with languages in the categories (3a) and (3bii). In the remainder of this section, we first discuss examples of languages that combine accent, stress and tone in various ways. The material in this section is largely drawn from Sua´rez (1983) and Yip (2002). In section 5, we present languages representative of the various language families in this area, drawing on a wider range of resources. 4.1. Tonal aspects Sua´rez (1983) classifies the tone systems of languages of this area in terms of various criteria. Some systems have levels (such as high, mid and low), which occur as such (making up what others call ‘level tones’) or in combinations (making up what others call contour tones). He calls such systems ‘register systems,’ of which Tlapanec is mentioned as an example (see section 5.1.6.). In other languages, which may also have level tones, the contour tones can, apparently, not be defined as transitions between independently needed level tones, so that in these cases the contours are ‘primitives’; these kinds of systems he calls ‘contour systems’. Examples he mentions are Texmelucan Zapotec (with three contours and one low level tone) and Copola Trique (which has three level tones and five contours, some of which can be analyzed as sequences of level tones while others cannot). Clearly, the existence of ‘primitive’ contour tones is a crucial element in this typology. For a di¤erent point of view, see Yip (2002: 47–52), who proposes that perhaps all contours can be represented as sequences of level tones. Both Sua´rez (1983: 50–51) and Yip (2002: 217–219) remark that various languages in the area have a ‘terrace system,’ meaning that they have downstep or upstep. Other dimensions of di¤erence between languages of this area involve (a) the number of tones (paradigmatic contrast), (b) possible combinations

require that the syllable claimed to be accented is perceived as ‘prominent’ on independent grounds. Downing (this volume) on the other hand uses the term pitch accent system to include all systems in which accent and tone interact. See van der Hulst (to appear) for a general discussion of the notion pitchaccent.

The Languages of Middle America

253

in contour tones, (c) syntagmatic restrictions on tone sequences, (d) the presence of tone perturbation or tone sandhi, (e) the grammatical use of tone (tonal morphemes, tones as markers for morphological classes) and (f ) the relation between tone and stress. a. Tone numbers range from two (H, L) to perhaps five in register systems (mentioned are Usila Chinantec and Chicahuaxtla Trique, both of which also have various combinations, or contours). More common are four tone systems (Popolocan group, Lealao Chinantec, Yaitepec Chatino, Mixtec from San Antonio Atatlauhca), while three-tone systems are the most common (Zapotecan, Chinantecan, Miztecan, Popolocan, Tlapanec and Amuzgo). There are also many two-tone languages with contour tones (Otopamean languages: North and South Pame, Mezquital Otomi, Isthmus and Mitlazapotec). Two-tone systems without contours in this group are Matlatzinca and Chichimec. Outside the Otopamean languages, two-tone systems without contours are the only type and present in Huichol, Huave of San Mateco, Tzotzil of San Bartolo and Uspantec. Yip (2002: 214–5), based on Hollenbach (1977), takes a close look at two dialects of Trique, with five levels and various combinations, discussing the challenges that these systems pose for systems of tone and register features. She refers to San Juan Copola Trique, where the distribution of the tonal contrasts is such that many non-final syllables have no distinctive tone, whereas final syllables display an eight-way surface contrast (six-way if the vowel is short). San Andre´s Chicahuastla Trique has a fifteen-way contrast on final syllables (nine binary contours and two ternary contours), with a more limited contrast on the penult (four-way, all level) and antepenult (three-way, all level). Restrictions of this sort will be discussed further below. b. In some language combinations of tones into contours are free (as in Tlapanec, which also has a three-tone combination), while in others only certain combinations occur. Sometimes (as in Mixtec of San Miguel el Grande) combinations can only occur in long vowels. Yip (2002: 216–7) notes that surface contours can be the result of underlying contour tones, from the association of floating tones (with a morphological function) to syllables that bear tone, or from tone spreading rules. c. Sua´rez (p. 51) remarks that in two-tone systems it is rare to find forms that are di¤erentiated by tone only. This perhaps is an indication that such systems are on their way to using high tone/pitch as an accentual cue. This

254

Harry van der Hulst, Keren Rice, and Leo Wetzels

issue is related to the question of whether, in a two-tone system, both tones have equal status or not. In many cases it might seem as if the high tone must be lexically marked, with the low tone regarded as a default (literally ‘unmarked’) tone; see Yip (2002: 219–222) for discussion and the discussion of Huave below (where L rather than H is the lexically marked tone). Many syntagmatic restrictions are caused by the interaction between tone and accent or stress such that the tonal contrast is richer on or even restricted to accented syllables. Finally, syntagmatic restrictions may result from tonal assimilation. d. Some languages have tonal perturbation e¤ects, with tonal patterns a¤ected when a word is in a larger context. For instance, Huave, Northern Tepehuan, and Mixtec and Zapotecan languages show extensive tone perturbation e¤ects, while the Chinantec languages, Tlapanec, and Yucatec do not. Interestingly, while Soyaltepec Mazatec has tonal rules, the closely related language Huautla Mazatec does not. Tonal perturbation e¤ects may be progressive or regressive, and participation may depend on morphological or phonotactic and syntactic characteristics of the words that are a¤ected or that trigger the pertubations. Yip (2002: 224 ¤.) shows that the languages in this area display the ‘usual autosegmental e¤ects,’ which involve tone stability (tones staying behind after their underlying tone bearers have been deleted, which then may condition other tonal processes or dock on neighbouring tone bearers, sometimes creating additional contours), tonal spreading and lexical floating tones that condition processes of downstep and, finally, tonal a‰xes (factor (e) mentioned above). 4.2. Tone – accent interactions So far we have largely considered tone by itself, even though in some of the examples mentioned some form of accent or stress is present, either independent from the tonal system, or not. In terms of tone/accent dependency we distinguished two cases that we now discuss in more detail, taking examples from Sua´rez (1983) and Yip (2002), who both discuss the relationships between tone and accent/stress, and from the survey in section 5. 4.2.1. Tone dependent on accent Where tone is dependent on accent, Sua´rez refers to this as an accented syllable forming the domain of tone. We already mentioned the example

The Languages of Middle America

255

(discussed in Yip 2002, and Hollenbach 1977) of San Juan Copola Trique, which displays ‘shrinking’ tone inventories outside the final syllable. The final syllable is not described as being stressed just because it carries the tonal contrast; final syllables are richer in many phonotactic respects, with a more complex syllable structure and a greater number of segmental contrasts in consonants and vowels. We here mention some other examples from Sua´rez (1983) and Yip (2002). Isthmus Zapotec (Mock 1985, 1988) has two tones, which associate to the stressed syllable and from there spread rightward. Pre-stress syllables are low-toned. Stress is rootinitial (although according to Yip weight may play a role). Sua´rez mentions Northern Pame and Yaitepec Chatino as languages that have a tonal contrast only in the syllable that is said to be stressed (the last syllable in both cases, presumably of the stem). This can be compared to Huautla Mazatec where every syllable can have contrastive tone, with final stress. Between these extremes, we find cases where the contrast on certain nonaccented syllables is limited. In Palantla Chinantec, for example, there is no tonal contrast on a post-stress syllable. In cases of this type, the dependency of tone on accent is such that the tonal lexical contrast is limited in terms of reference to the accented syllables. A di¤erent type of dependency exists when there is no tonal contrast and accented syllables receive a specific tone (typically a high tone). Yip (2002: 232 ¤.) discusses various such examples. In Northern Tepehuan (Kim 1988), a H tone is assigned to the second syllable of trisyllabic words and the initial syllable of bisyllabic words; all other syllables have low tone. Huave, as discussed by Noyer (1991), presents an interesting case that combines (limited) lexical contrast and predictable tone. Stressed syllables may have either a H or a HL tone, which Noyer analyzes by assigning a lexical L to the latter and then a H to all stressed syllables; all other syllables are low toned. The location of stress is predictable. It occurs on the final syllable if closed, otherwise on the penultimate syllable. Sua´rez (1983: 52) refers to Huichol and Mazuhua where we find no tone contrast on the last two syllables or the last syllable, respectively. In these languages, inherent lexical tones are removed to free up space for intonational tones. 4.2.2. Accent dependent on tone According to Sua´rez (p. 52–3), stress in Huichol and Tlapanec is dependent on tone. In the latter language, bisyllabic word with one H tone are

256

Harry van der Hulst, Keren Rice, and Leo Wetzels

stressed on the syllable bearing that tone. If both syllables are H, or both are L and the second syllable has a long vowel, they are ‘equally stressed’. An interesting case is discussed in de Lacy (2002) and reported in Yip (2002). de Lacy discusses three Mixtec dialects. In Huajuapacan Mixtec (E. Pike and Cowan 1967), stress falls on the leftmost syllable when the tone is followed by a lower tone (HL, HM, ML) and, if this condition cannot be met, on the first syllable.

5. Languages by family3 In this section, we survey the accent systems of the languages of Middle America, organized by language family. 5.1. Oto-Manguean The Oto-Manguean (also Otomanguean) language stock is a large complex group of language families spoken in Mexico. In the past, languages of this group were spoken further south in Central American as well. The Ethnologue (Lewis 2009) list 177 languages for this family. The major groups within Oto-Manguean are Amuzgoan, Chiapanec-Mangue, Chinantecan, Mixtecan, Otopamean, Popolocan, Subtiaba-Tlapanecan, and Zapotecan. The bulk of attention to the phonology of these languages involves tone, laryngeal features, and syllable shapes. Both laryngeal features and tone are lexical. The languages exhibit two to five level tones and often contour tones as well. Tone is lexically marked and is also used grammatically, and complex tone sandhi exists in many of these languages. See Campbell (1997), Sua´rez (1983), and Rensch (1976), among others, for overviews. Hollenbach (1984) provides a clear set of minimal pairs from Copala Trique that illustrates the heavy load on tone and laryngeals in the languages of this group.

3. For a map of Meso-American languages see: http://www.mywire.com/a/Intl-Enc-Linguistics/MesoAmerican-Languages/ 9460795/?&pbl=105

The Languages of Middle America

257

(4) Copala Trique tones and laryngeals ‘palm mat’ yu3we35 3 32 ‘maguey’ yu we 3 32 ‘cli¤ ’ yu we h ‘marketplace’ yu3?we32 yu3?we32h ‘thread’ yue3?we3? ‘ice’ While much attention is given to tone, tone sandhi, laryngeal features, and the interaction between these, there is little work on stress, although stress is mentioned in almost every description of the languages of this stock.4 This section focuses on the predictable stress rather than the lexical and grammatical tone. Intonation is also discussed in some of the literature, but is largely set aside here as well. In general, stress in Oto-Manguean languages is morphologically conditioned, being attracted to a root or stem, with this morphological unit occurring near the end of the word. Rensch (1976: 11), in ground-breaking research on comparative Oto-Manguean phonology, reconstructs the stem as consisting of one syllable with a consonant onset, a vowel, an optional nasal, laryngeal elements, and a tone, together with possible preceding and following elements. In Proto-Oto-Manguean this syllable hosts the largest number of contrasts. Probably more than one syllable preceded this syllable, likely separated from it by a morpheme boundary; a syllable expressing person may have followed. Rensch notes that stress generally falls on the final syllable of the stem in contemporary Oto-Manguean languages, and posits that this was the case in Proto-Oto-Manguean as well (1976: 11); in a footnote, he remarks that in Cuicatec and Zapotec, stress is on the penultimate syllable. We will see that in many of the languages stems are generally single syllables, and other languages besides Cuicatec and Zapotec place stress on the initial syllable of the stem. Nevertheless, Rensch’s generalization about the placement of stress on the stem is overall confirmed across this complex. This stress-bearing unit, called the ‘stressed ultima’ or ‘couplet’ by Rensch (1976: 11) and sometimes referred to as the ‘binary couplet’, is 4. Since many sources do not distinguish between stress and accent (as we did in section 2), or use both terms interchangeably, we will use the terminology as we find it in the sources, where the use of the term ‘stress’ predominates.

258

Harry van der Hulst, Keren Rice, and Leo Wetzels

the topic of much study in the language stock. This stress unit, or couplet, is basically equated with the stem, with stress reconstructed as falling on its final syllable. The couplet determines the distribution of segments and tone in many of the languages. The following discussion is organized by language family within OtoManguean. There has been debate over the years about the internal organization of Oto-Manguean; see Rensch (1976) for a summary. The following classification is from the Ethnologue (Lewis 2009). It should be noted that the degree of detail presented di¤ers from language to language, and, for many of the languages, there are no comments about stress in the literature. Much of the work on these languages was done in the 1950’s and 1960’s, and the terminology used is often unfamiliar today; we have tried to make terminology clear, while at the same time giving the reader a sense of the language descriptions. No material is available on the Chipanec-Mangue subfamily, but each of the other groups receives some attention in the following discussion. 5.1.1. Amuzgoan The Ethnologue (Lewis 2009) lists three Amuzgoan languages, which are spoken in the Costa Chica region of the Mexican states of Guerrero and Oaxaca by about 35,000 speakers.5 According to Bauernschmidt’s (1965) study of syllable dynamics in Guerrero Amuzgo, the stem-final syllable is stressed; the stem can be followed by su‰xes. The stressed syllable obligatorily comprises an onset and a nucleus, and can carry a tone (see also Williams (2005) for a discussion of tones in this language). The phonetic realization of stress in Amuzgo is dependent on syllable type. Amuzgo exhibits a distinction, identified in many Oto-Manguean languages, between syllable types known in the literature as controlled and ballistic.6 There is considerable discussion of this contrast in the literature, which is more or less equivalent to a distinction between long and short syllables. Bauernschmidt (1965: 472) notes distinctions between con5. It should be kept in mind that here and elsewhere the numbers of speakers provided for a given language are only approximate, often based on a census that may date back as much as 20 years. Most of the time, the numbers provided are gathered from the Ethnologue. 6. For the distinction between ‘ballistic’ and ‘controlled’ in the Oto-Manguean languages, see the recent study by Herrera (2009), who shows that, at least in this family, the distinction is expressed by a contrast between breathy voice and creaky voice vowels.

The Languages of Middle America

259

trolled and ballistic syllables having to do with stress. In particular, when stressed, the ‘‘checked nucleus of the ballistic syllable shows minimum duration of voicing; that of the controlled syllable shows maximum duration of voicing [where voicing refers to duration of voicing in the nucleus HRW]. Although unchecked stressed syllables do not show as pronounced a contrast in this respect, here too voicing tends to be shorter in the ballistic syllable nucleus. Unstressed syllables show no consistent contrast in duration of voicing.’’ There are also some tonal di¤erences between controlled and ballistic stressed syllables: controlled syllables show high and mid tones with a downward drifting pitch contour, and low tone with an upward drift. Ballistic syllables realize high, mid, and low tones as rapid downglides. 5.1.2. Chinantecan The Ethnologue lists fourteen Chinantecan languages. The total number of Chinantec speakers is estimated at around 102,000. Most live in Oaxaca and Veracruz, Mexico, especially in the districts of Cuicatla´n, Ixtla´n de Jua´rez, Tuxtepec and Choapan. Usually in the Chinantecan languages there is a single stress, occurring on the final syllable, the stem; pre-tonic and post-tonic syllables are not stressed. The stressed syllable often has greater phonological and morphological complexity than the unstressed syllables. Contrasts are generally diminished on pre-stem and post-stem material. In several of the Chinantecan languages, su‰x tones are not contrastive, being determinable from the tone of the stem. Languages of this family exhibit the contrast between controlled and ballistic syllables introduced above; languages di¤er in constraints on ballistic syllables; see Silverman (2009) for a recent overview. This section is a brief review of the literature on stress in di¤erent languages of this subfamily. StressTyp contains an entry for Lealao Chinantec:7 Chinanteco, Lealao [LEX] Oto-Manguean, Chinantecan. North Oaxaca, San Juan Lealao, Latani, and Tres Rı´os (Mexico). f Stress is lexically distinctive. Words can be stressless. There is tone (not indicated). Niø ‘his/her voice’ fi ‘road’ ¨Niø ‘high’ ¨fi ‘handle’ 7. In Part II of this book all StressTyp entries are presented with the sources that they are based on.

260

Harry van der Hulst, Keren Rice, and Leo Wetzels

According to Rupp’s (1990: 63) description of Lealao Chinantec ‘‘the phonological word is typically just one stressed syllable, although a verb may be realized phonologically as a string of as many as six syllables. The stressed syllable is the last syllable of the word unless the word ends with one of five unstressed person markers’’; thus the stress falls on the last syllable of the stem. Rupp (1989: 3) further notes that most Chinantec words have simple roots or stems, and that the root is the domain for stress, by which he means that it is only the root that shows a contrast between controlled and ballistic syllables. Silverman (1997), in a study of tone sandhi in Comaltepec Chinantec, notes that words are generally monosyllabic, with rare polysyllabic roots. Inflection is often found through modification of the root vowel, resulting in monosyllabic stems that bear stress (Silverman 2009). Robbins (1968) and Gardner and Merrifield (1990: 93) observe for Quiotepec Chinantec that only one syllable per word is stressed, the stressed syllable representing a major class lexical item. Most such words are monosyllabic, but a few have a pretonic syllable marking categories such as tense-aspect and direction. In Quiotepec Chinantec, the pre-stem syllables occur with simple tones, not with contours, while the stem can have contours as well (Robbins 1968); post-tonic syllables have their tone determined by the tone of the tonic syllable. There are also post-tonic syllables that generally mark personal pronouns and possessors. These su‰xes can a¤ect the tone of the stem. Westley (1971: 160), in a study of Tepetotutla Chinantec, observes that there is one stressed syllable per word, the final syllable. Foris (2000) finds for Sochiapan Chinantec that stress occurs on the final syllable in polysyllabic words. 5.1.3. Mixtecan Mixtecan is a large group with subgroups of Mixtec-Cuicatec and Trique with a total of 57 languages, according to the Ethnologue (Lewis 2009). 5.1.3.1. Cuicatec The Cuicatecs inhabit the towns of Teutila and Tepeuxila in western Oaxaca. In 2000 they numbered around 23,000, of whom an estimated 65% still spoke the language. Two languages are listed under Cuicateco in the Ethnologue. Cuicatec generally has stress on the first syllable of the unit called the couplet (see the discussion below), although Needham and Davis (1946)

The Languages of Middle America

261

note that its location is not well understood. They also say that stress may be attracted to a high tone at the end of a morpheme, although the conditions are not totally clear. Bradley (1991), in a study of the syntax of Concepcio´n Pa´palo Cuicatec, makes some notes about phonology. Roots are monosyllabic or bisyllabic, with stress on the first syllable and with stressed vowels being long. 5.1.3.2. Mixtec At the end of the 20th century, the fifty-two Mixtecan languages listed in the Ethnologue taken together were estimated to be spoken by about 300,000 people, in the Mexican states of Oaxaca, Guerrero and Puebla. In the literature on Mixtec stress, the term ‘couplet’ is frequently encountered. K. Pike (1948) introduced the term ‘tonemic couplet’ in a study of San Miguel el Grande Mixtec, referring to this unit as the locus of contrastive tone (see also Gerfen (1996) for discussion). In the introduction of her detailed study of Mixtec phonological systems, Josserand (1983) notes that Proto-Mixtec and all modern and historically documented systems share a set of basic properties, one of which is the couplet, which she defines as ‘‘a microsegment composed of two syllables and carrying stress on the first of these two (1983: 180).’’ Josserand further notes that the first of these syllables is often marked by additional phonetic features such as vowel length, pre-aspiration of following consonants, or glottalization. It is also the syllable that is most likely to engage in tone perturbations, besides being the locus of paradigmatic consonant alternations and of stress. The couplet is claimed to be the basis for word building and is similar to a grammatical unit, but there is no categorical one-to-one relation between morpheme (root or stem) and couplet (181). In Josserand’s own terms ‘‘It is more satisfying to understand the couplet as a phonological or surface form for underlying morphological material to fit into, or accommodate itself to, for its spoken realizations. Nonetheless, the couplet is essential for the description of the morphological canons, i.e., the base forms and morpheme structure rules which precede the morphotactic processes so common in all Otomanguean languages’’ (183). This position is widely adopted in the literature. According to Sua´rez (1983: 53) stress in Mixtec languages usually falls on the first syllable of the root and remains invariable when a‰xes are added. North and Shields (1977) consider the couplet as the nucleus of both the phonological and grammatical word (citing K. Pike 1948, among others), as well as the relevant unit for tone contrasts. Thus, in Mixtecan in general, the couplet appears to be an

262

Harry van der Hulst, Keren Rice, and Leo Wetzels

important unit, attracting stress and allowing several contrasts that are not found elsewhere. In current terminology, the couplet might be referred to as a trochaic foot. In his discussion of the couplet in Mixtec, DiCanio (2008) comments on its status in several languages of this family. Faraclas (1983) states for Chalcatongo Mixtec that the basic stem is disyllabic, CVCV or CVV, sequences that were composed probably of two [or more] morphemes historically. In the same language, glottalization is restricted to couplet-initial syllables (Macaulay 1996). In Coatzospan Mixtec (Gerfen 1999), glottalization is licensed under stress, with the stressed syllable being initial in the word, and, in compounds, the initial syllable of the first morpheme is stressed. Pankratz and E. Pike (1967) discuss stress in Ayutla Mixtec, saying that the couplet coincides with the grammatical stem (288), with stress generally falling on the first syllable of the couplet. The couplet can be preceded and followed by a‰xes. Some examples are given below, where the acute accent represents stress and the numbers tone. (5) Ayutla Mixtec couplet only n˜u´2ma?3 precouplet þ couplet Si3-du´2?va3

‘wax’ (292) ‘spider web’ (292)

Pankratz and E. Pike remark that each phonological word has a word stress, occurring on the couplet or on a post-couplet syllable. Tone functions as a conditioner for the placement of word stress, which is attracted to what they call tone 1 (high tone) in a couplet or post-couplet syllable, whereas, in the absence of a tone 1, word stress goes on the first syllable of the couplet, as is illustrated in the examples below (cf. section 4.2.2.): (6) §1 n˜u1ra1 ku1-ta ta17 ra1 ku2nu3ra´1 sa3ta3-ra´1

‘his pineapple’ (293) ‘he is taking medicine’ (293) ‘his tobacco’ (293) ‘he bought’ (293) (Couplet plus postcouplet)

In Mortensen (2005), the distribution of stress in this language is defined by way of a more intricate set of conditions: (7) Tone-dependent stress in Ayutla Mixtec Within the stem þ su‰x domain, stress the leftmost H-toned syllable immediately followed by the L-toned syllable, else stress the leftmost M-toned syllable immediately followed by a L-toned syllable, else stress the leftmost H-toned syllable, else stress the leftmost syllable.

The Languages of Middle America

263

Interestingly, Mortensen suggests that Pankratz and E. Pike’s tonedependent stress is likely to represent phonetic pitch prominence. In Huajuapan Mixtec, E. Pike and Cowan (1967) also identify the couplet as an important unit, with stress on the first syllable of the couplet unless there is a following syllable within the word with a lower tone; in that case, word stress occurs on the syllable preceding the lower tone; stress never precedes the couplet. Tones are mostly predictable on nonstem morphemes. North and Shields (1977) discuss stress in their work on Silacayoapan Mixtec. This language also has stress on the first syllable of the couplet. There are conditions under which stress (and high tone) shift to the second member of the couplet. In their very clear description of the phonology of Mixtepec Mixtec, E. Pike and Ibach (1978) observe that in this language the first syllable of the couplet is marked by a lengthened vowel unless it is followed by a glottal stop. It is also stressed, indicated by loudness. They note that, if tone 1 (high) or tone 2 (mid) precedes tone 3 (low), the syllable preceding the tone 3 may be louder than the first syllable: vi1lu1-yu ‘my (polite) cat’ (272). They note that the loudness is particularly noticeable on the tone clusters 1-3 and 2-3. There are allotones as well as tone sandhi. Gerfen (1996, 1999) o¤ers a discussion of Coatzospan Mixtec. Gerfen (1996: 47) defines the couplet as the open class set of morphemes characterized by the shapes CVV and (C)VCV, with some rare exceptions. Stress is located on the penultimate syllable of the word, with enclitics ignored for the purposes of stress. The stressed syllable is longer, and is the site of contrastive tone. In a study of Alacatlazala Mixtec, Zylstra (1980: 16) remarks that stems consist of two or three syllables, where the three syllable stems are generally compounds. The word has a single stress, falling on the initial syllable of the couplet, but shifting to the second syllable when the first person postclitic is attached, and on the second syllable of a three-syllable stem. Tones are contrastive within the couplet. The language called Pinotepa Nacional Mixtec in the Ethnologue and Jicaltepec Mixtec by Bradley (1970) is analyzed as having two stresses, an initial ‘microstress’ which is part of the couplet, and thus on the initial syllable of the stem, and a ‘macrostress’, or sentence stress. The couplet is also relevant in the analysis of Molinos Mixtec (Hunter and E. Pike 1969). Hunter and E. Pike propose that the placement of stress depends on the tone sequence and the position of the couplet within the word, with stress occurring within the couplet. It can occur on either the first or the second syllable of the couplet depending on the tones. Hunter

264

Harry van der Hulst, Keren Rice, and Leo Wetzels

and E. Pike further propose that if the tones are all level 1 (high) or 2 (mid), then all syllables are equally stressed. Hunter and E. Pike note that the couplet is relevant for the placement of stress, for the distribution of some sounds and the distribution of allophones of others, and tone sandhi. E. Pike and Oram (1976) discuss stress in Diuxi Mixtec. As in many other Mixtec dialects/languages, stress falls on the first stem syllable and conditions lengthening of the first vowel or second consonant. According to E. Pike and Oram (1976), a second unpredictable stress occurs in Diuxi Mixtec on the second syllable of some stems, but Daly (1978) analyzes these alleged stem-final stresses in terms of tone. The Mixtec languages are very similar in terms of stress, as one concludes from the numerous studies devoted to this language group. Across Mixtec, stress is on the first syllable of the couplet, which can be equated with the stem, forming a trochee. This is the site where contrastive tones occur in many of the languages as well as other phonological features. 5.1.3.3. Trique The Trique languages di¤er from the other members of the Mixtecan group in the placement of stress. The Ethnologue identifies three Trique languages. DiCanio (2008) is a recent source on Trique, with interesting discussion of the historical development of this group. In the Trique languages, stress is word-final, falling on the root. This is also the position of greatest contrast of other phonological features. Tones contrast on this final syllable, being predictable on syllables that precede it (e.g., Hollenbach 1984: 10). In DiCanio’s (2008) analysis of Itunyoso Trique, stress is word-final, implemented by an obligatory bimoraic structure in final syllables (nonfinal syllables are light) and a larger set of licensed contrasts. Similar patterns are found in Copala Trique, where the final syllable carries tone and has a greater potential for contrasts in terms of consonants, vowels, and tone than other syllables (Hollenbach 1977, 1984). Similarly, in Chicahuaxtla Trique, the largest number of contrasts occurs in the final syllable: the opposition between fortis and lenis consonants is only functional in this syllable, while it is neutralized elsewhere. Also, the final syllable is longer than the prefinal syllables and hosts a greater number of tonal constrasts (Longacre 1952). Hollenbach (1977: 50) notes that there is phonetic stress on penultimate syllables with mid tone and on penultimate syllables before a final syllable with a medially-checked vowel (V?V, VhV). Sua´rez (1983: 52) remarks that Chicahuaxtla Trique has stress on the final

The Languages of Middle America

265

syllable of the word. Di Canio concludes that root-final stress is a shared characteristic of the Trique languages. 5.1.4. Otopamean Seventeen di¤erent languages compose the Otopamean family of Mexico, which are classified in four major subgroups: Chichimec (one language) spoken in the state of Guanajuato, Matlatzincan (two languages) in the state of Mexico, Pamean (two languages plus one extinct) in the state of San Luis Potosi, while Otomian is subdivided into Mazahuan (two languages), spoken in Michoacan, and Otomian (nine languages) spoken in the states of Puebla, Veracruz, Queretaro, Hidalgo, and Tlaxcala. The numbers of speakers vary greatly among the languages of this family, ranging from over 350,000 speakers for Central Mazahua to the nearly extinct Atzingo Matlatzinca. In reconstructions of the Proto-Otopamean vowel system, Bartholomew (1989) proposes that the root contained a vowel or vowel cluster and an optional glottal element; if the vocalic element was a single vowel, that vowel was rearticulated after the glottal element (V?V or VhV), if it was a vowel cluster, the glottal element was articulated between the vowels. 5.1.4.1. Chichimec Chichimeca-Jonaz (Lastra de Sua´rez 1984) is a tonal language, with a high tone on one or more syllables. Bisyllabic stems can be HL, LH or HH and, apparently rarely, LL. Trisyllabic stems can be HLL, HLH, LHL, LHH, and so on. No mention is made of stress. Lastra de Sua´rez does not describe the tonal system, but she marks tone in examples. 5.1.4.2. Matlatzincan Pe´rez (2007) is a study of Matlatzincan languages of the Oto-Pamean family, Matlatzinca and Ocuilteco, also known as Tlauica. Roots in Matlatzinca are generally monosyllabic, although consonant-final roots have an epenthetic final vowel, showing di¤erent patterning than the rare bisyllabic roots. Pe´rez does not discuss stress in Matlatzinca, and her discussion of tone is quite tentative. Escalante (1999) identifies two tones in Matlatzinca and discusses tone sandhi, without any mention of stress. Muntzel (1986), in a study of the structure of Ocuilteco, notes that

266

Harry van der Hulst, Keren Rice, and Leo Wetzels

primary accent falls on the first syllable of the verb or noun root, causing a slight lengthening of the vowel, which does not acquire the duration of the distinctively long vowels. Perez (2007: 128) notes that she did not hear stress in Teauica, suggesting the existence of dialect di¤erences. 5.1.4.3. Otomian 5.1.4.3.1. Otomi Nine Otomi languages are listed in the Ethnologue. For those Otomi languages with reasonable descriptions, the stem is reported to consist of two syllables, with stress on the first syllable. As in most other languages in the stock, that syllable is the locus of greater contrast in terms of both segmental properties and tone. Word stress in Mezquital Otomi is described by Wallis (1968) as falling on the first syllable of a bisyllabic sequence referred to as the ‘nucleus’, with the stressed vowel being long and with distinctive tone on the nucleus, or the stem. In Tenango Otomi, as described by Blight and E. Pike (1976), stress falls on the first syllable of the stem, which can be preceded by prefixes. Stress a¤ects the realization of consonants, as, for instance, in the case of the phonemic voiceless fortis stops that are preaspirated when they are initial in a stressed syllable (and not part of a cluster). The first syllable of the stem is stressed; it can be preceded by prefixes. The full range of tonal contrasts is found on stressed (stem) syllables, with a reduced set on prefixes. In Temoayan Otomi, Andrews (1949) reports that stress falls on the first syllable of the root. She remarks that stress and tone are independent, with some tones restricted to the stressed syllable, or the root, which is thus the locus of greater contrasts. 5.1.4.3.2. Mazahuan The Ethnologue lists two Mazahua languages. In these languages, stress occurs on the first syllable of the stem, which is composed of a root and a stem formative, as discussed in E. Pike (1951). Stress is accompanied by vowel lengthening (38) and, as expected, the greatest di¤erentiation of tone occurs on the stressed syllable. There is also an intonational pitch, which is final, and tone is not contrastive on the final syllable. 5.1.4.4. Pamean The Ethnologue identifies three Pamean languages. Both Northern Pame and Central Pame have many speakers, while Jilipian Pame is extinct. In

The Languages of Middle America

267

his detailed study of Northern Pame, Berthiaume (2003) remarks that stress falls on the lexical root (65), and contrastive tone is found only on the stressed syllable (41). With one exception, roots are monosyllabic. The pitch of the non-stressed syllables is determined by the pitch of the root. Berthiaume also reviews earlier work on Pamean. In a discussion of Manrique Castan˜eda’s work on Jilipian Pame, he reports that stress falls on the lexical root, which is also the domain of contrastive tone (Berthiaume 2003: 17). In Central Pame, based on work by Gibson (1956), stress is also on the lexical root, and tone is contrastive only in stressed syllables (Berthiaume 2003: 18). 5.1.5. Popolocan Popolocan is another major group of the the Oto-Manguean language stock, divided into Chocho-Popolocan, the Mazatecan group, for which the Ethnologue distinguishes eight di¤erent languages with about 185,000 speakers altogether living in the states of Oaxaca, Veracruz and Puebla, and Ixtecan, spoken by 128 individuals in Santa Marı´a Ixcatla´n in southern central Mexico. Chocho-Popolocan itself is subdivided into Chocho, spoken in the state of Oaxaca by around 770 speakers and Popolocan, comprising seven di¤erent languages with approximately 60,000 speakers in the state of Puebla. 5.1.5.1. Chocho-Popolocan In her study of Metzontla Popoloca, Veerman-Leichsenring (1984, 1991) notes that stress, which is realized primarily as duration, falls on the penultimate syllable of a word. Stops and a¤ricates are preglottalized, while liquids, approximants, and nasals are lengthened: /c&aku 7/ ‘face’ [c&a?ku 7]; /tı #ye´/ ‘night’ [tı #y‡e´] (page 21). With aspirated consonants, prenasalized consonants, and voiceless nasals, the consonants are essentially split, creating a phonetic sequence. When the syllable contains a long vowel or vowel sequence (including a vowel followed by a glottal stop), the vowel is long. In general, it is the consonant that shows the e¤ects of accent, and with monosyllabic words, a prothetic vowel may appear (/ku#/ ‘animal’ [ı`?ku#] (page 22). In Western Popoloca (Williams and E. Pike 1968: 379), the stressed syllable usually, but not always, coincides with the end of the grammatical stem. In Northern Popoloca (Stark 1976), tone is contrastive in stems; stem tones may override prefix tones. Kalstrom and E. Pike (1968) discuss stress in Eastern Popoloca. The penultimate vowel of the stem (the nucleus

268

Harry van der Hulst, Keren Rice, and Leo Wetzels

of the phonological word) receives stress, with the stressed component being long, either by consonant lengthening or vowel lengthening. (8) nte:4to2 ‘tall’ nka˜:3o˜2 ‘liver’ With certain su‰xes, the stress is realized as what the authors call consonant stress, lengthening the last consonant of the root. ‘metal object’ (9) tSi:3ka3 tSi4ka4n:a23 ‘my metal object’ Words with this stress type are noted to be infrequent. Consonant lengthening a¤ects the initial consonant of the last syllable. However, when certain su‰xes follow, the initial consonant of the su‰x lengthens. ‘woman’s shawl’ (10) to1t:e4 1 4 13 ‘my shawl’ to te n:a There are also vowels inserted in phrases, referred to as rhythmical syllables. (11) k?oe4n:a2tha˜3h:ma4 i2h:ma4 k?oe4n:a2tha˜3

‘she bought beans’ h:ma4 ‘beans she bought’ i2h:ma4

Stark and Machin (1977), in a study of Tlacoyalco Popoloca, also identify two types of stress, consonant stress and vowel stress, each marked by duration. Consonant stress occurs on the final syllable and moves when a su‰x is added. 5.1.5.2. Mazatecan In a study of comparative Mazatecan, Kirk (1966: 9) reconstructs for Proto-Mazatec heavy stress on the final syllable of the stem, with weak stress elsewhere. This system is also found in the contemporary Mazatec languages studied by this author (1966: 13). E. Pike (1956: 71) notes for Soyaltepec Mazatec that stress falls at the end of the grammatical word. She further points out that non-phonemic length of consonants is found in stressed syllables, which is di¤erent for Huautla Mazatec, where this length is not found. 5.1.6. Subtiaba-Tlapanecan The status of the Subtiaba-Tlapanecan group as part of Oto-Manguean has been of some debate. Its phonology is not well described in the avail-

The Languages of Middle America

269

able literature. Sua´rez (1983: 52) notes that in Me’phaa (formerly called Tlapanec) stress is determined by tones, with the second syllable of a bisyllabic word made prominent with a low-high tone sequence and the first syllable with a high-low tone sequence. 5.1.7. Zapotecan With approximately 482,000 speakers, the Zapotecan family of Mexico is one of the largest families in the Otomanguean stock, not only in terms of its number of speakers, but also as regards the number of di¤erent languages. The Chatino subbranch contains seven di¤erent languages, spoken in the state of Oaxaca, whereas for the Zapotecan subgroup, the Ethnologue distinguishes as many as fifty-seven di¤erent languages, all spoken in the states of Oaxaca and Veracruz. 5.1.7.1. Chatino StressTyp contains an entry for Yaitepec Chatino: Chatino, Yaitepec (dialect of Chatino) [U] Oto-Manguean, Zapotecan, Chatino. Oaxaca (Mexico). f Stress falls on ultimate syllable. ¨slya/ ‘cotton’ kwi/¨ya ‘eagle’

ti¨/a ‘water’ ts&i¨kwi/ ‘to talk’

Upson and Longacre (1965: 314) reconstruct final-syllable stress for ProtoChatino. Upson (1968: 3), referring to the contemporary Chatino languages, observes that stress falls on the last long vowel of the stem, and, in the absence of long vowels, on the stem-final syllable. Stressed vowels are longer than unstressed vowels. Pride (1984), in a study of Tataltepec Chatino, reports that the final syllable is the tonic or stressed syllable. The full set of vowel phonemes occur in this syllable, with phonetic lengthening. Pride notes that the tonic syllable is the minimal phonological word, which may be preceded by up to three syllables. Tone contrasts are also greatest on the tonic syllable, with seven contrastive tones. He proposes a four-way tone contrast on the penult, and a two-way contrast elsewhere. In Yaitepec Chatino as well, the final syllable is stressed (Pride 1963: 19). The final syllable is the sole carrier of contrastive tone and the contrastive presence or absence of nasalization and glottal stop.

270

Harry van der Hulst, Keren Rice, and Leo Wetzels

5.1.7.2. Zapotec Zapotec is a complex group of 57 languages. In Zapotecan, stress usually falls on the root, which is one or two syllables long, and, most of the time, the stressed syllable predictably lengthens. In Cajonos Zapotec, stress falls on the root (Nellis and Hollenbach 1980), which is typically one or two syllables long, with some exceptional three-syllable roots. Stress is reported to be penultimate in a vowel-final polysyllabic root and final otherwise. Root stress is retained under su‰xation. Fortis consonants are longer after a stressed vowel; stressed vowels lengthen before lenis consonants. Recent Spanish loans retain the stress position of Spanish. For Isthmus Zapotec, Marlett and Pickett (1987: 406) find that stress falls on the first rhyme of the root. In earlier work on the phonology of this language, Pickett (1951) argues that stress placement is predictable. Marlett and Pickett further report that branching rhymes are found only in stressed syllables. After a stressed vowel, obstruents lengthen phonetically; otherwise, the vowel lengthens. In Zoogocho Zapotec, stress falls on the penultimate syllable of the stem (Sonnenschein 2005); in a stem consisting of two roots, it falls on the second. The phonological shape of the word base is CV(C(V)). Sicoli (2007), in a study of three Zapotec languages of the Sierra Sur of Oaxaca, notes a bimoraic minimum. Stress-induced consonant lengthening occurs, similar to that found in Isthmus Zapotec (Mock 1985). In both cases, the vowel of the stressed syllable lengthens unless it is followed by a fortis consonant, in which case that consonant lengthens. In Guelavı´a Zapotec (Jones and Knudson 1977: 163) there is one stressed syllable per word, generally the penultimate syllable. The stressed syllable tends to have higher pitch, lengthening of a following fortis consonant, and lengthening or rearticulation of vowels with a following lenis consonant or when syllable-final. Tilquiapanz Zapotec (Merrill 2008) has stress on the root syllable. Merrill notes that it usually falls on the final root syllable, but sometimes on the penultimate syllable. Glottalized vowels are possible only in the root syllable, while word-final unstressed vowels usually devoice. Choapan Zapotec (Lyman and Lyman 1977) also has one stressed syllable per word. Stress must occur on a complex nucleus unless all nuclei are of equal length, in which case it is penultimate (Lyman and Lyman 1977: 149). A‰xes are more restricted in terms of syllable shape than stems are.

The Languages of Middle America

271

Beam de Azcona (2004) is a study of Coatla´n-Loxicha Zapotec. Most words are monosyllabic due to the historical deletion of non-tonic vowels. The only syllable of the stem bears stress. The few polysyllabic words, which are historical compounds, have final stress. Stressed syllables can take any tone; unstressed syllables only realize high tone. In San Francisco Ozolotepec Zapotec (Leander 2008), roots are composed of a single syllable, which also realizes the stress. Unstressed vowels are centralized and sometimes deleted (43), with the stressed syllable hosting the largest number of contrasts. San Juan Mixtepec Zapotec, as studied by Nelson (2004), has roots that are generally a single syllable, which bears the stress. Simple stems are monosyllabic in Guevea de Humboldt Zapotec (Marks 1980). In compounds, the final syllable is stressed. The stressed syllable undergoes a variety of morphophonemic changes. 5.1.8. Summary: Oto-Manguean In the above survey, we have discussed the word prosody of the OtoManugean languages for which we have found useful descriptions in the literature. In these languages, lexical and grammatical tone is usually found, as well as some interesting laryngeal contrasts. There is ample evidence that in languages of the Oto-Manguean family the root is the locus of stress and, expectedly, it is also the position that shows the greatest number of phonological contrasts. While Rensch (1976) reconstructs a two-syllable stem for Proto-Otomanguean with stress on the first vowel, there are languages in which the second stem-vowel is stressed, as well as languages that have single-syllable stressed roots. A common feature of all the languages of this family is that the root licenses more segmental and tonal contrasts than a‰xes do. Moreover, the root is obligatorily bimoraic in the majority of languages. In general, no mention is made of the phonetic properties of stress. Oto-Manguean languages have been cited as stress-attracting tone languages and as tone-attracting stress languages (see section 4.2). In the former, tone is lexical while stress is predictable. In our interpretation, the richness of tonal contrasts that occurs in the stressed syllable is a consequence of the fact that stress is a feature of the root; stress is morphologically controlled, and because it is, we expect to find the full range of contrasts in the stressed root syllable. Several of the languages have been argued to show tone-dependent stress systems, with stress attracted to certain tones, and, in some cases, more than one stress depending on

272

Harry van der Hulst, Keren Rice, and Leo Wetzels

tone. While this analysis may account for the facts, it might not be the only one possible. Higher-level intonation may play a role, as is evident in many of the descriptions, where stress is described at di¤erent levels of structure. This is clearly an area for further study. 5.2. Mixe-Zoque The Mixe-Zoque languages are spoken in and around the Isthmus of Tehuantepec, Mexico. Mixe has around 90,000 speakers, Zoque has some 60,000 speakers, and the Popoluca languages, of which some are Mixean and some Zoquean, have approximately 69,000 speakers. Wichman (1995) o¤ers a classification of the Mixe-Zoque group, for which the Ethnologue lists 17 languages, based on comparative evidence, including evidence from stress: (12) Mixean Oaxaca-Mixean

Zoque

þTapachulteco Oluta Popoluca8 Sayula Popoluca Gulf Zoque

Chimalapa Zoque (Oaxaca) Chiapas Zoque

Lowland Mixe: Coatla´n, Guichicovi Midland Mixe: Japtapec South Highland Mixe: Tlahuiltepec North Highland Mixe: Totontepec

Ayapa Zoque Texistepec Zoque Soteapan Zoque Santa Maria Chimalapa San Miguel Chimalapa

Here we review the information that Wichmann supplies about the Mixean languages and his reconstruction of proto-Mixe-Zoquean stress. Wichmann does not discuss the stress systems of the contemporary Zoquean languages, but refers to di¤erent bibliographical sources for information (page xiiv). In Tlahuiltepec (Wichmann 1995: 25 (based on Lyon 1980: 33)), primary accent is on the last heavy syllable of the word or, if there is no heavy syllable, on the final syllable. In Jaltapec (Wichmann 1995: 47), accent is on the final syllable if it is closed or ends in a long vowel, otherwise stress is on the penultimate syllable. This generalization does not apply to su‰xes or enclitics and loans from Spanish. In Coatla´n (Wichmann 1995: 8. The PopolUcan languages in this family should not be confused with the PopolOcan languages in the Oto-Manguean family.

The Languages of Middle America

273

47; Hoogshagen 1984: 5), ‘‘stress occurs on the last syllable of an uninflected word that ends in a consonant or a long vowel and on the next to last syllable of an uninflected word that ends in a short vowel. Stress occurs on the last syllable of the stem of an inflected word unless a stress carrying su‰x is present, in which case the su‰x will carry the stress. The stress-carrying su‰xes are -yi: (object person foxcus), -ç: (nominalizer) and -o/ (repetitive).’’ In Guichicovi (Wichmann 1995: 60), accent falls on the rightmost heavy syllable. Wichmann does not mention where stress falls in words that have no heavy syllables. According to Wichmann (1995: 81): ‘‘although both [Sayula and Oluta Popoluca] verge toward having phonemic stress, stress assignment rules in both languages can be shown to directly reflect [Proto-Mixean] stress rules which were automatic.’’ In Sayula Popoluca accent goes on the rightmost heavy syllable or else to the leftmost syllable in the rightmost root. Certain su‰xes are inherently accented (Wichmann 1995: 81). In Oluta Popoluca accent is final if the final syllable ends in a consonant or has a long vowel, otherwise it is penultimate. Wichmann remarks that in this language most syllables are closed because of a process of final glottalization. This may obscure the fact that Oluta Popoluca has a ‘right most heavy’ pattern (Wichmann 1995: 85). In Totontepec, accent is on the rightmost heavy syllable, otherwise on the final closed syllable, otherwise on the penultimate syllable (Wichmann 1995: 81, after Schoenhals and Schoenhals 1965: 302). Wichmann proposes that the stress pattern of Sayula Popoluca, where accent goes to the rightmost heavy syllable, or else to the leftmost syllable in the rightmost root, may be applicable in Totontepec as well, considering this from the perspective of reconstruction of proto-Mixean. Wichmann’s reconstruction for proto-Mixean emphasizes the recurrence of stress on the rightmost heavy syllable. He suspects that this feature goes back to proto-Mixe-Zoquean, but the Zoquean languages lost distinctive length. Zoquean then generalized in one of two directions: root stress (since roots would be the typical locus of heavy syllables) or, focussing on the default pattern when heavy syllables were absent, right-edge stress. For proto-Mixean languages Wichmann reconstructs the following rule: (13) a. b. c.

accent is on the rightmost heavy (V:, V:/, V/, Vh) else on the rightmost root: verbs: ultimate, nonverb: penultimate Some su‰xes are inherently accented and as such attract the primary accent

274

Harry van der Hulst, Keren Rice, and Leo Wetzels

His reconstruction for stress in proto-Zoquean is given in (14) (p. 89): (14) a.

Stress in on/in the root: verbs: ultimate, nonverb: penultimate/first syllable b. Some su‰xes are inherently accented and as such attract the primary accent

For proto-Mixe-Zoquean Wichmann reconstructs the system as essentially identical to that of Proto-Mixean. 5.2.1. Mixe(an) The Ethnologue (2009) lists three major Mixe languages, Eastern, Veracruz, and Western, each of which has several dialects. Here we describe the stress pattern of Ayutla Mixe (South Highland Mixe), based on RomeroMe´ndez (2008: 80 ¤.). In this language stress is word-final for all parts of speech except for verbs, which have stem-final stress. Phonetically, stress involves higher pitch and greater intensity. In compounds, the last syllable of the second root is stressed: (15) (a)

ka¨j-¨tse’e jaguar-pumpkin

‘chilacayote/cidra’

(b)

kapy-¨ka¨p bamboo-basket

‘bamboo basket’

Prefixes do not interact with stress patterns. When a prefix is added, the root still bears the stress (16): (16) (a) (b) (c)

ne¨-¨majtsk ordinal-two

‘second’

t-ak-¨jo’kx ‘s/he heats it’ 3a-caus-heat a-¨ke¨’e¨y-y ‘lid’ edge-cover.with.hands-dep

In the verb, it is possible to add as many as three syllables to the right of the root, while the main stress remains on the last syllable of the root: ¨pe¨jk-e¨-p ‘s/he got sick’ be.sick-inv-indep (b) ¨pe¨jk-e¨-te¨-p ‘they got sick’ be.sick-inv-pl-indep

(17) (a)

The Languages of Middle America

275

¨pe¨jk-e¨-ne¨-˙te¨-p ‘they have got sick’ be.sick-inv-perf-pl-indep (d) a¨me¨jk-e¨-ne¨-˙te¨-p ‘they have had nightmares’ have.nightmare-inv-perf-pl-indep

(c)

The plural marker, which can be used with only a few nouns, is the only inflectional su‰x that nouns can take. In this case, the su‰x receives stress: (18) (a) me¨¨ja¨’a¨y (b) me¨˙ja¨’a¨y-¨te¨jk

‘old person’ vs. ‘old people’

In contrast, the diminutive marker does not take primary stress: (19) ¨te¨jk-u’unk

‘little house’

Clitics cannot bear stress, even if they are in final position. In a trisyllabic word, the leftmost syllable receives secondary stress and the last syllable retains the primary stress, as in (20). If the third syllable in a trisyllabic word is a su‰x, then the stress goes on the second syllable and there is no secondary stress, as in (21). (20) a) [˙nçtç¨gçøk] ‘third’ b) [˙dagam¨bikJ] ‘s/he makes him angry’ c) (21) a)

[˙ajFø¨ˆ§p]

‘very poor’

a¨me¨k-e¨-p be.sick-inv-indep

b) am¨too-te¨-p hear-pl-indep

‘s/he has nightmares’ ‘they hear’

It is possible to have a secondary stress to the right of the stem, but only if there are more than two syllabic su‰xes, as in (17c–d) above. In cases like this, the last syllable of the root bears primary stress and the last syllable of the word has the secondary stress. Romero adds (p. 83): ‘‘As far as I have seen, some other Mixe languages seem to have a stress pattern very similar.’’ 5.2.2. Zoque(an) Three major Zoquean groups are indicated in the Ethnologue, Chiapas Zoque, Oaxaca Zoque, and Veracruz Zoque, each with some subdivisions. StressTyp contains one entry for this subfamily:

276

Harry van der Hulst, Keren Rice, and Leo Wetzels

Zoque, Copainala´ [P] Mixe-Zoque, Zoque, Chiapas Zoque. Copainala´, Northern Chiapas (Mexico). f Primary stress falls on the penultimate syllable of the stress group. f Secondary stress on the first syllable if there are three or more syllables. ˙min¨ke/tpa

‘he is coming again’

˙minsukke/tpa¨/itti

‘they are coming again’

According to Hurch and Rhodes (1996) clashes are allowed. All stressed vowels lengthen. (22) [/a`øna´øsa] ‘orange’ [he`/epı´kpa] ‘he is breathing’ In Sierra Popoluca (Gulf Zoquean), according to Foster and Foster (2008), the primary stress occurs on the penult of most parts of speech, but in verbs the stress is on the final syllable. 5.3. Totonacan The Totonacan family, consisting of twelve languages, is divided into two major groups in the Ethnologue (Lewis 2009), Tepehua and Totonac. A considerable amount of work has been done on this language family recently, which we discuss in this section. Recent work suggests that there might be a relationship between Totonacan and Mixe-Zoque (see Brown, Beck, Kondrak, Watters, and Wichmann forthcoming). McFarland (2009) cites an unpublished study by MacKay, who argues that primary stress in Totonaco-Tepehua falls on the final syllable when it is heavy and on the pre-final syllable when the final syllable is light. The definition of heavy syllable varies from language to language, but it is generally CVV or CV(V)C, where the final consonant is h, /, m, n, l, y, w. While this regular pattern is encountered in the various languages that are discussed below, in some languages this system is broken by construction-specific stress patterns and by forms which do not meet the generalization. For this reason, these languages are often described as having phonemic stress. 5.3.1. Tepehua Three languages are recognized under Tepehua in the Ethnologue, Huehuetla Tepehua (3000 speakers), Pisaflores Tepehua (4000 speakers), and Tlachichilco Tepehua (3000 speakers); all are spoken in western central Mexico.

The Languages of Middle America

277

5.3.1.1. Tlachichilco Tepehua According to Watters (1980), Tlachichilco Tepehua has one stressed syllable in a word. Prestressed syllables have mid pitch; stressed and post-stress syllables are high in pitch except before pause. There are adjustments in speech, with word boundaries often lost. Stress is maintained but the intonation pattern suggests a single word with high pitch falling on the last stressed syllable in the intonational phrase. Watters (1988) adds that Tlachichilco Tepehua has stress on a final syllable if it is heavy; otherwise stress is penultimate. There are two exceptions. 5.3.1.2. Huehuetla Tepehua Kung (2007), in a study of Huehuetla Tepehua, is concerned largely with stress placement. She distinguishes for this language three stress patterns, one in native words, a second in ideophones, and a third in loanwords (Kung 2007: 104). She also notes that stressed vowels are lengthened (2007: 31). In native, non-ideophonic words, there are two degrees of stress, with primary stress on the ultimate, penultimate, or antepenultimate syllable; its particular placement depends on syllable structure, word length, and word class. More generally, Kung notes that primary stress falls on the final syllable if it ends in a sonorant (m, n, l, r, ¸) or glide (h / w j) (23); otherwise the penult is stressed (24). Long vowels do not attract primary stress (25). Secondary stress is assigned to alternate syllables preceding the primary stress. The first column in the examples indicates the orthographic representation of the word. The examples below are taken from Kung (2007: 108–123) (23) Final sonorant-closed syllable 7akminaaw //a-k-min-a/-w/ [/˙ak.mi¨aøu] ‘we excl will come’ laqtz’in /laqts’in/ [laq.¨ts’in] ‘she sees it/him/her’ (24) Final syllable open or closed by non-sonorant k’i7ut’i /ki-/u-t’i/ [k’i.¨/u.ºi¥] ‘you eat me’ palata /palata/ [pa.¨la.ta] ‘more, better’ p’it’ilh /p’it’i-li/ [¨”i.ºI] ‘she scrubbed it’ (25) Final long vowel 7e7eyxtaa /aqejStaø/ [a.¨/eiS.taø]

‘tree species’

There are some prefixes that cannot take primary stress, with stress unexpectedly falling on a final syllable that ends in a non-sonorant. Kung

278

Harry van der Hulst, Keren Rice, and Leo Wetzels

also notes that there are noun su‰xes that cause violations of the stress patterns. Ideophones show their own stress pattern, with primary stress on the first syllable and secondary stress on all subsequent syllables. (26) chilili

[¨tSi.˙li.˙li]

‘sensation of fear’

In Spanish loanwords, primary stress falls on the syllable that corresponds to the stress-bearing syllable in Spanish, with secondary stress on alternate syllables. The Spanish word is shown in the third column, with its stressed syllable underlined. Only primary stress is shown. (27) 7abonalaa [/a.Bo.¨na.laø] kaapeen [kaø¨.peøn] 7aarreesgaalaa [˙/aø.reøs.¨gaø.laø]

‘fertilize’ ‘co¤ee’ arresgar ‘take a chance’

abonar cafe´

5.3.2. Totonac The Ethnologue (Lewis 2009) lists nine Totonac languages, altogether spoken by some 200,000 speakers in central western Mexico. As discussed above, stress generally falls on one of the last two syllables in this family, but there are numerous surface counterexamples to this generalization. Length and laryngealization are contrastive in these languages. 5.3.2.1. Papantla Totonac In Papantla Totonac (Levy 1990: 25), primary stress is contrastive: ¨paSa ‘take a bath 2p sg’ vs. pa¨Sa ‘take a bath 1p pl’, although it generally falls on the penultimate or final syllable. The first syllable of a word generally bears secondary stress (MacKay 1999: 409). Levy (1991) further proposes that inflectional a‰xes are outside the domain of stress assignment, and that stress falls on the final syllable of the imperfective and the penultimate syllable of the perfective; see MacKay (1999) for discussion. Levy (1987: 107) notes that primary and secondary stress are realized phonetically as intensity and perhaps as ‘higher’ pitch. 5.3.2.2. Upper Necaxa Totonac Upper Necaxa Totonac is analyzed by Beck (2004: 4) as having phonemic stress, with stress di¤erentiating words of di¤erent lexical classes and marking person-aspectual distinctions in some verb classes. Stress is generally final in words ending in a long vowel or a closed syllable and, except

The Languages of Middle America

279

in verbs, penultimate in words ending in a short vowel. In verbs, stress is on the final syllable, except for verbs in the perfective aspect or the second person, in which case it is penultimate. Stress is realized as increased duration and, to a lesser degree, as amplitude. Stressed short vowels are the same duration as long vowels, while stressed long vowels are not consistently longer than unstressed long vowels, but instead stress is marked with a rising tone. 5.3.2.3. Huehuetla Totonac Troiani (2004) treats accent in Huehuetla Totonac as lexical, given minimal pairs such as ¨kuku ‘uncle’ and ku¨ku ‘sand’ (p. 49). Nevertheless, there are some general tendencies: every lexical root has an accent. In the verb, su‰xation causes alternations in the stress pattern, such that stress is on the final syllable if it ends in m, n, y, l, w; otherwise it is on the prefinal syllable. Troiani notes that accent in nouns requires further study. Stress is realized as intensity and higher pitch, and sometimes as a slight lengthening of the vowel. 5.3.2.4. Misantla Totonac MacKay (1994, 1999) and MacKay and Treschsel (2005) are studies of Misantla Totonac. The sources treat nominal stress and verb stress separately. There are two degrees of stress in Misantla Totonac. Secondary stress is attracted to heavy syllables, which are syllables with a branching rhyme, either a long vowel or a vowel with a following consonant. Primary stress falls on either the final or penultimate syllable (MacKay 1999: 73). Phonetically, primary stress is stronger than other stresses, characterized by longer duration and greater loudness. Primary stress has the word as its domain, with clitics not a¤ected; clitics themselves are stressed and do not a¤ect the placement of primary stress. In nominals, stress falls on the penultimate syllable when the word-final syllable is light, and is attracted to a word-final heavy syllable. The examples (28–9) are taken from MacKay (1999: 74). (28) min-paa-luu [mı´mpa´alu´u] ukuk [uku´k]

‘your intestines’ ‘pierced’

However, word-final syllables closed by a coronal obstruent (t, s, S, ) do not receive primary stress, as shown in (29):

280

Harry van der Hulst, Keren Rice, and Leo Wetzels

(29) mukskut [mu´kskut] ‘fire’ kutSi [ku´tSi] ‘knife’ In general, final syllables with a long vowel are stressed even if they are closed by a coronal obstruent, although there are some exceptions. MacKay also proposes some instances of lexical stress, with some stressed CVC syllables ending in a voiceless coronal. In other Totonac varieties, these words are vowel-final, with the final vowel lost in Misantla Totonac. In verbs, primary stress falls on the final syllable regardless of its weight. Word-final inflectional su‰xes are never stressed in word-final position. There are various complications with su‰xed forms, for which MacKay introduces a type of lexical extrametricality. 5.3.2.5. Filomeno Mata Totonac McFarland (2009) argues that the regular stress pattern proposed by MacKay, although it is often found in Filomeno Mata Totonac (examples in (30)), does not always apply (examples in (31)). Note that final voiceless vowels as in the last two examples of (30) (called latent vowels by McFarland) do not count for the purposes of stress. On the other hand, the last example in (31) is considered to have a final sonorant-closed syllable. McFarland notes that forms with unpredictable stress occur in 25% of the monomorphemic items in her nominal database. Stress both fails to occur on final sonorant-closed syllables, and is found on final syllables that do not meet the conditions for stress. The following examples are taken from MacFarland (2009: 52–54): (30) tsı´isa su´wa tSı´kı´tS’i¥ tanta´l’a¥

‘morning’ ‘black sapote fruit’ ‘type of plant’ ‘naked’

(31) Stı´ilan ‘chicken’ tSaalı´ ‘tomorrow’ Since counterexamples are numerous, McFarland concludes that stress is lexical in Filomeno Mata Totonac. McFarland notes several cases of what she terms morphological stress, or stress assigned by a su‰x, as well as special stress patterns in compounds and ideophonic constructions. She argues that morphological

The Languages of Middle America

281

stress is associated with what she calls constructions, and not with the phonological surfacing of an a‰x. The stress rules are computed based on the underlying presence of su‰xes, some of which do not appear on the surface. For instance, the imperfective aspect construction has final stress whether the imperfective su‰x -aa is present or deleted (page 53). To give a few examples, the progressive aspect forms a construction that is associated with stress on the penultimate underlying syllable, as in (32) and (33). (32) /paati-nan-maa/ su¤er-hab-prog

[patina´ma]

(33) /S-ta-tata-maana-para/ [Statatamaanapa´] past-hear2-prog2-there2-2sub.sg

‘he is su¤ering’ ‘you were listening over there’

To complicate matters, the second person allomorph of the progressive -paa is always followed by the second person singular su‰x -ti¥. This su‰x assigns stress to the syllable preceding paa. (34) /taaskux-paa-ti¥/ [taasku´xpaatHi¥ ] ‘you are working’ work-prog2-2sub.sg In compounds, stress falls on the final syllable of the penult member of the compound. Sound symbolic adverbs are reduplicated forms, and take primary stress on each syllable of the reduplicant. Spanish borrowings tend to have penultimate stress, regardless of where stress is found in Spanish, although some have final stress (kape´ < cafe´). (35) bure´ko borrego ‘sheep’ puusiku´lan saecula ‘century’ (Latin saecula; puu ¼ loc) Overall, there is a single stress per word which falls within a threesyllable window at its right edge, the exact position being lexically determined in non-derived words and determined by the rightmost stresschoosing su‰x in derived words. 5.3.2.6. Summary Accent in Totonacan languages is of interest for several reasons. First, there are category di¤erences in stress patterns, with di¤erent patterns in nouns and verbs as well as in the onomatopoetic vocabulary in some of the languages. Second, there are construction-specific stress patterns.

282

Harry van der Hulst, Keren Rice, and Leo Wetzels

Third, in Huehuetla Tepehua, consonant-closed syllables attract primary stress, while lexical long vowels do not, a pattern that has been claimed to not exist. 5.4. Tequistlatecan Tequistlatecan, also known as Chontal of Oaxaca, has two surviving varieties, Highland Chontal, with 3600 speakers, and Lowland Chontal (also known as Huamelultec, Campbell 1997), with only 950 speakers (Lewis 2009, the Ethnologue). There is relatively little documentation available for these languages (see O’Connor 2007 and Maddieson, Avelino, and O’Connor 2009 for discussion), and there is little discussion of the stress system in the literature. 5.4.1. Highland Chontal Highland Chontal, as discussed by Turner (1966, 1967), has a single primary stress per word. This syllable is longer than unstressed syllables. Turner (1966) states that stress falls on the penultimate syllable of the word unless it is displaced to the final syllable because of its position in the sentence. In his 1967 work, he further indicates that word stress is complex, with grammatical-phonological interactions also playing a role. He notes that pluralization can be indicated by a stress shift: lena´la ‘the squash stem’, lenala´ ‘the squash stems.’ Turner also comments on clausal stress, manifested as loudness on the final word. 5.4.2. Lowland Chontal Waterhouse and Morrison (1950) make brief comments about stress in Lowland Chontal, remarking that it is phonemic. They cite pairs such as sa´nna ‘small crawfish’ and sanna´ ‘star’ to support this analysis. O’Connor (2007) provides a brief introduction to the phonology of Lowland Chantal. With respect to stress, she notes that vowel length is not phonemic but it occurs especially in penultimate stressed syllables (33). Maddieson, Avelino, and O’Connor (2009) is a phonetic study with brief comments about stress. They note that there are five vowels (i e a o u), all of which show some allophonic lengthening, frequently in pretonic syllables. 5.5. Mayan Mayan is a major language family, with 69 languages listed in the Ethnologue. There are five major divisions, Cholan-Tzeltalan, Huastecan,

The Languages of Middle America

283

Kanjobalan-Chujean, Quichean-Mamean, and Yucatecan, each with a number of subdivisions. Three languages of the Mayan linguistic family have contrastive tone: Yucatec, Uspantec, and one dialect of Tzotzil (San Bartolo Tzotzil), and there seems to be incipient tone developing in Mocho´9. In all these languages the use of tone is restricted. All other Mayan languages are stress-accent languages without contrastive tone. In the following a number of stress systems in Mayan languages are discussed. In our classification of the Mayan languages we follow Campbell, Kaufman and Smith-Stark (1986). One Mayan language is included in StressTyp. Pokomchı´; Poconchi [U] Mayan, Quichean-Mamean, Greater Quichean, Pocom. North Central Guatemala. f Stress falls on the final syllable. /e¨tal ‘to signal’

/au¨as

‘older brother’

ni¨naa ‘my head’

5.5.1. Huastecan Huastec is a Mayan language spoken in a small area of southeastern San Luis Potosı´ and northern Veracruz, Mexico. With the now extinct Chicomuceltec, it forms the Huestecan branch of the Mayan linguistic family. According to Edmonson (1988), stress in Huastec falls on the rightmost long vowel of the word or, if there is no long vowel, on the leftmost vowel (see also Larsen and E. Pike 1949). The language has no secondary stresses, except for a single case in which primary stress is reduced to secondary stress in words derived with the compound su‰x /-talaab/, which forms unpossessed abstractive nouns and seems to act as an independent phonological word: /wa´lab/ ‘sin’ P /wa`labtala´ab/ ‘sinfulness’. Some su‰xes may trigger changes in the stress pattern. Unstressed syllables show a tendency towards centralization. A similar analysis of Huastec is proposed in Herrera (2010), who provides the following examples (observe the lengthening of the stressed vowel in the adaptation of loans from Spanish (36h–i)):

9. Lyle Campbell, personal communication.

284

Harry van der Hulst, Keren Rice, and Leo Wetzels

(36) Huastec (Herrera 2010: 3) (a) ta´.mu (b) kı´.Tib (c) pa´k.laT (d) pa.t’a´al (e) la´ab.lij (f ) tek’Ta´al

‘I have met him’ ‘arena’ ‘bent’ ‘iron’ ‘to bless’ ‘to cook’ ‘very soft’

pa.lu.lu´ul (h) le´etsa (

‘pace’

e>¨lEktrn ‘electron’

¨Alfa>˙bEt

‘alphabet’

˙pa>ra>¨si>t

‘parasite’

English [(U-)P/A] Approximation: f Primary stress falls on the final syllable in nouns if the vowel is long, in verbs if the vowel is long or there are two closing consonants. f In other cases, stress falls on the penult if it contains a long vowel or coda. f Else stress is antepenultimate. f Secondary stress falls on alternate syllables to the left (many exceptions).

446 o¨bey

Harry van der Hulst

a¨genda

mo¨lest

¨discipline

as¨tonish mon¨soon

tre¨mendous ¨innocent ho¨rizon ¨croco˙dile ¨ali˙bi

¨cinnamon

Old English [I] f Primary stress falls on the first syllable (of the root/stem). f Secondary stress reported on alternate syllables to the right respecting weight. ¨mo#dgidanc ‘thought’

¨QTeling ‘prince’

Additional information Jessen (1999) concludes that, even though the ‘‘Dutch’’ accent rule is not fully matched in German, they are very similar nonetheless (cf. Fery 1998). In fact, the German accent system appears to be more closely related to the Dutch system than to the English system. The following characterization seems to apply to both German and Dutch: a. Accent is final if the final syllable is superheavy (VVC, VCC) b. If the final syllable is open accent is penultimate c. If the final syllable is closed (but not superheavy) and the penultimate syllable is open, accent is on the antepenultimate syllable (s´ CV CVC) Although there are considerable similarities between the Dutch/German and the English accent systems, the English system is di¤erent in a number of respects. Like Dutch/German, English has a quantity sensitive right edge system. The most basic di¤erences between English and Dutch/ German lie in the nature of the extrametricality rule and the way in which quantity is computed. Unlike English, German/Dutch treats all open syllables as light, and only closed syllables as heavy. Vowel length correlates with syllable closure in that open syllables always contain long (or tense) vowels, whereas short (or lax) vowels can only occur in closed syllables (Vennemann 1990; Wiese 2000). Closed syllables with long vowels (so called superheavy syllables) occur, but almost only in word final position. Whereas in English all final syllables are extrametrical, in German/Dutch only final VC (which will form a monosyllabic foot because it is heavy) is extrametrical (cf. van der Hulst 1984, Kager 1989, Zonneveld and Trommelen 1999).

Word accent systems in the languages of Europe

447

In derived words, German/Dutch, like English, distinguishes between Class I and Class II a‰xes. In compounds accent falls on the rightmost member if it is an adjective (unless the ‘rhythm rule’ applies) or a preposition and on the leftmost member if the rightmost member is a noun or a verb.14 An interesting phenomenon of English, Dutch and German, which occurs in other Germanic languages as well, is the adjectival accent-shift, due to the so called Rhythm Rule. This phenomenon is treated in depth in Visch (1999). Word tone in Germanic languages Word-tones (or tonal accents, phonetically manifested as a high or low pitch15) occur in addition to accent in most dialects of Norwegian and Swedish.16 In Danish another phonetic property, generally referred to as stød, related to intensity, subglottal pressure and constriction of the vocal folds, occurs on accented syllables. Both Swedish and Norwegian distinguish two di¤erent tonal accents, mostly called Word Tone I and Word Tone II. Historical reconstruction of the origin of these word tones indicates that they derive from an original di¤erence between monosyllabic and polysyllabic words. When, through a number of morphological and phonological processes, many originally monosyllabic words became polysyllabic the two tones came to stand in phonological opposition.17 An alternative theory is that accent 2 stems from the context of two stresses (disyllabicity following trivially from this). The chief argument is laid out

14. In English stress in nominal compounds goes to the right-hand member if this is itself a compound. 15. Central Swedish has a H lexical tone in accent 2, whereas South Swedish has a L lexical tone. Similarly, East Norwegian, North Norwegian and Go¨ta dialects have a H lexical tone, whereas West Norwegian (Bergen), South Norwegian and Dala have a L lexical tone. 16. I preserve the term ‘accent’ for the abstract culminative property that Swedish shares with the other Germanic languages, using the term ‘word tone’ for the word-level pitch opposition. In a di¤erent terminological tradition Bruce uses the terms ‘stress’ and ‘tonal accent’, respectively. 17. We also refer to Lahiri et al. (1999) for a discussion of the historical background of the tonal opposition.

448

Harry van der Hulst

in Riad (1998). Bruce (1999) presents some general regularities by which the choice of the tone class in Swedish and Norwegian can, at least partly, be synchronically predicted. He shows that in simplex (non-compound) words the main factors determining the choice of word tone are primary accent location and the morphological make-up of a word. In compounds the factors determining the choice of word tone varies in di¤erent dialects. Danish stød seems to be historically related to Word Tone I, the equivalent to Word Tone II simply being the absence of stød, but this correlation is not perfect. Stød may occur on the primary accented syllable of simplex words and in compounds on the accented syllable of the compound members. The presence or absence of stød conditions seems to be determined by the weight of the accented syllable. In fact, both tone (in Norwegian and Swedish) and stød (in Danish) occur only in heavy syllables. But for Danish stød to be realized, the sonority of the stressed syllable has to be high. (This in itself is a strong indication that we are dealing with a tonal configuration here, not a separate phonological entity ‘stød’). Like Swedish and Norwegian, most Limburgian and Rhenish dialects (situated in the south-east of the Netherlands and the southern part of Germany) have a contrast between two word-level tones. These are usually referred to as ‘falling tone’ (Dutch: stoottoon) and ‘dragging tone’ (Dutch: sleeptoon). Hermans (1999) discusses the Limburgian dialect which is spoken in and around Maasbracht, a small village near Roermond in the Netherlands. In this dialect, the falling tone starts on a relatively high pitch and makes a rather steep fall to a relatively low pitch. The dragging tone is realised at a level high pitch when followed by another syllable in the utterance, but as a fall-rise in utterance-final position. The dragging tone is accompanied by significant lengthening of the syllable. There are quite a number of monosyllabic words which di¤er only at the tonal level, but it is hard to find minimal pairs among polysyllabic words, although near minimal pairs are found. Hence, the distribution of tone is not (fully) predictable in monosyllabic or in polysyllabic words. Still, there are a number of environments in which the choice of tone is predictable. Hermans (1999) sums up the relevant environments and presents an analysis that accounts for the predictability of tone in these environments. Two of the main restrictions on the distribution of tone are related to

Word accent systems in the languages of Europe

449

primary accent: tone contrast is only possible on the syllable that bears primary accent, and only dragging tone occurs when the primary accented syllable is followed by another (secondary) accented syllable in the same word. Other restrictions concern the number of sonorant segments in the nucleus of the syllable, the voicing of the elements in the coda of the accented syllable, and the status of the vowel. In Hermans’ analysis, the falling tone is lexically marked by the presence of a L tone, whereas the dragging tone is lexically unmarked. I also refer to Gussenhoven (2004) for extensive discussion and analysis of Limburgian tone. Generalizations As in the case of Celtic, the Germanic languages must be subdivided into two groups with respect to accent: an initial and a final group. The first group consists of Icelandic and Faroese which have retained the old Germanic system of initial weight-insensitive accent. In all other Germanic languages, a synchronic analysis leads to postulating a right edge, quantitysensitive system with, for most morphologically complex words, dependence on morphology. Within this group, English occupies a special position, di¤ering from the other members in several ways. Firstly, although all Germanic languages allow antepenultimate accent (which is regular in case words end in VV-VC, and irregular otherwise), antepenultimate accent is more widely present in English, due to rules of ‘extrametricality’ that can be said to be a systematic part of the system, at least in the nominal system (see Zonneveld and Trommelen 1999). A second di¤erence involves the definition of syllable weight. In English closed syllables and syllables with long vowels count as heavy, whereas in at German and Dutch only syllable closure appears to perform this role. The latter di¤erence may be dependent on the phonological character of tense vowels in the various languages. It has been argued, for example for Dutch (van der Hulst 1984b, van Oostendorp 1995) and German (Vennemann 1990) that tense vowels are not long. If this route is taken, the di¤erence between the systems of the Germanic languages (except Icelandic and Faroese) are thus very slight (involving extrametricality, the status of tense vowels and the possibility of final geminates). Lahiri et al. (1999) o¤er an historical perspective on the accentual systems of the modern Germanic languages, taking the (root) initial primary accent of Proto-Germanic as a point of departure.

450

Harry van der Hulst

4.1.3. Baltic (< Indo-European) Genetic information BALTO-SLAVIC BALTIC: WEST: UOld Prussian (in 18th century) EAST: Latvian (also Lettish), Lithuanian SLAVIC (see section 4.1.4.) Baltic and Slavic are commonly grouped into one superfamily. In this section I will first discuss the Baltic subgroup which is generally considered to be the more conservative subgroup. The oldest information regards Old Prussian with records from 1300 AD to the time it went extinct (in the 18th century). Little is known about its word prosodic system, however. Latvian and Lithuanian are among all the IE languages, the most conservative, i.e. the closest to PIE. StressTyp extracts Latvian [I] f Stress falls on the first syllable. ¨baga#t"Æba

‘wealth’

¨kra#sa

‘colour, paint, dye’

¨kokvilna

‘cotton’

¨ma#kslinieks

‘artist’

Lithuanian [F/F] f Stress falls on the first syllable marked for accent, else on the first. f Secondary stress falls on the second syllable after the tone. ¨kiœSkis ‘rabbit’ va¨do)vas ‘leader’

¨giria ‘praises’ ¨mo@ky˙tojas ‘teacher’

gi¨ria

‘forest’

Additional information and Generalizations According to Dogil (1999b) the Latvian word prosodic system establishes a link between the West Slavic languages on the one hand, and the South and East Slavic languages (excluding Macedonian) and Lithuanian on the other hand. Latvian has strictly initial accent, like most of the West Slavic languages. Unlike, the West Slavic languages, however, it does not appear to have alternating secondary accent. The link with Lithuanian and most of the South and East Slavic languages can be found in the rural southern dialects of Latvian. These dialects show some of the accentual di¤erences that are characteristic of Lithuanian, i.e. they have lexical accents.

Word accent systems in the languages of Europe

451

Lithuanian has a system in which primary accent is, as in (according to most analyses) Russian, dependent on lexical accents. Roots, su‰xes and inflectional endings may be accented or unaccented. Primary accent occurs on the first accented syllable of a word, and if the word is completely unaccented, on the first syllable. It thus qualifies as a FIRST/ FIRST system, like Russian and indeed like PIE. The special feature of the prosodic system of Lithuanian is that primary accent occurs with three di¤erent phonetic exponents. Bimoraic syllables may have either a socalled ‘‘acute’’ accent or a ‘‘circumflex’’ accent. The choice between these two exponents is again a matter of lexical marking. Monomoraic syllables have a ‘‘grave’’ accent, which is phonetically very much like the circumflex accent, but shorter. According to Dogil these phonetic exponents do not, as often has been proposed, primarily involve pitch, since this is not the main distinctive mark of increased prominence in Lithuanian (cf. also Dogil and Williams 1999). Although the type of exponent does not play a role in the basic principle which governs accent-assignment, the position of accent is influenced by the exponent-type. This is the result of an accent advancement rule, called Saussure’s Law, which causes a rightward shift of the accent when circumflex or grave, but not in case of acute accent, in certain situations. Dogil finds that in every situation the circumflex and the grave accent indeed behave as a natural class. He therefore proposes an analysis of the Lithuanian system in which the acute accent has a di¤erent underlying representation than the grave and the circumflex accent. See Halle and Vergnaud (1987: 190–199) and Blevins (1993) for analyses of the Lithuanian system. 4.1.4. Slavic (Slavonic) (< Indo-European) Genetic information SLAVIC EAST: NORTH: Russian, Byelorussian SOUTH: Ukranian WEST: NORTH: Polish, Kashubian, UPolabian CENTRAL: Upper Sorbian, Lower Sorbian SOUTH: Czech, Slovak SOUTH: Old Church Slavonic WEST: Slovene, Serbian, Croatian EAST: Bulgarian, Macedononian

452

Harry van der Hulst

Commenting on Balto-Slavic accent, Beekes (1995: 150–153) notes that the reconstruction of Proto-Slavic accent is extremely unclear; cf. Kortland (1978, 1979). Anderson (1998: 428, 433) ascribes to Proto-Slavic a ‘free and mobile’ accent, inherited from PIE, including the distinction between accented (orthotonic) and unaccented (enclinomena) words. (This account is not incompatible with the BAP (Basic Accentuation Principle) proposed by Kiparsky and Halle (1977) for PIE, but stresses the unpredictable presence of accent in morphemes, rather than the choice of the first accent as primary, or, if no accent is present, the choice of the first syllable, by default.) Due to pre-Slavic developments involving (loss of ) laryngeal distinctions, Proto-Slavic also had an opposition between an acute (hightoned) and circumflex tone. This distinction was then lost in unaccented syllables. In Proto-Slavic we end up with a tonal distinction in accented syllables: acute (high followed by low) and neo-acute (high followed by high). The various Slavic languages either change or inherit a system of this kind. In Russian, pitch distinctions (and vowel length) are lost, but the accent location is essentially preserved. In Czech, the default initial accent location prevails. In Serbo-Croatian the di¤erence between the two accents is reinterpreted as a di¤erence in quantity. Some further remarks on a reconstruction of the proto-Balto-Slavic accentual system are made at the end of this section. Comrie and Crobett (1993) and Comrie (1981) contain overviews of Slavic languages. General sources on Slavic accentuation are Stankiewicz (1993) and Garde (1976); cf. Kortland (1978, 1979). EAST-SLAVIC StressTyp extracts Russian [F/F LEX] f Stress falls on the first syllable lexically marked for accent, else on the first. ras¨prava ‘reprisal [nom.sg]’

zje¨ny ‘woman [gen.sing]’

Additional information In most analyses, Russian accent is analyzed as being dependent on lexical accents that morphemes may have or cause to be placed or deleted in neighbouring morphemes. Given the lexical accents, word accent is located on the leftmost accented syllable or on the first syllable if there are no accents. Thus, in this analysis Russian is a FIRST/FIRST system. Ukranian appears to have the same kind of accent system.

Word accent systems in the languages of Europe

453

WEST-SLAVIC StressTyp extracts Polish [P] f Primary stress falls on the penultimate syllable. f Secondary stress on alternate syllables counted from the left (not on the antepenult). ¨rozpraw ˙saksofo¨nista

‘discussion [gen.pl]’ ‘saxophone player’

re¨porter ‘reporter’ ˙revo˙lucjo¨nista ‘revolutionary’

Slovincian (dialect of Kashubian) [L/F] f Stress falls on the last syllable with a long vowel or lexical accent mark. f If there are no such vowels, stress is initial. ¨kokoS ‘chicken’

je¨zoørka ‘lake dim.pl.’

svjiø¨niø

‘pig gen.sg.’

Polabian [U/P] f Stress falls on the final syllable if it contains a long vowel, else on the penult. Sorbian [I] f Primary stress falls on the initial syllable. f Secondary stress falls on the penultimate syllable of four syllable or longer words. ¨wojs@cojska ‘countryside’

¨domo˙wina ‘nation’

¨po@droz@nij˙kojstwo ‘going for a journey’ Czech [I] f Primary stress falls on the first syllable. f Secondary stress falls on alternate syllables thereafter. ¨name˙sti

‘[city] square’

¨nadra˙Zi

‘station’

¨kniha ‘book’

Slovak [I] f Primary stress falls on the first syllable. f Secondary stress falls on odd numbered syllables to the right. ¨nepo˙veziem ‘I will not carry’ ¨nepo˙vezie˙me or ¨nepo˙vezieme ‘we will not carry’ Additional information Dogil (1999a) describes the accent systems of Czech, Slovak, Sorbian, Polish, and the now extinct languages Polabian and Kashubian. All the

454

Harry van der Hulst

West-Slavic languages that are still spoken today have weight-insensitive systems, i.e. the position of accent is independent of the phonological make-up of the word. It is also insensitive to morphological structure. In Czech, Slovak and Sorbian primary accent falls on the initial syllable. The rules for secondary accent in Czech are di¤erent for formal and colloquial speech. In formal speech, secondary accents fall on odd-numbered syllables counting from the left edge of the word (i.e. away from the primary accent), whereas in more colloquial speech they fall on alternating syllables counting from the right edge of the word (i.e. on the penultimate syllable and on alternating syllables before it). In Slovak, as in formal Czech, secondary accents occur on alternating syllables counting from the left edge, but (depending on the accentual norm) the final syllable of odd-numbered words may or may not be accented. According to Dogil, Sorbian can be seen as a bridge between Czech with primary accent on the initial syllable and Polish with primary accent on the penultimate syllable. Sorbian has primary accent, realised as higher or rising pitch, on the initial syllable and secondary accent, realised as lengthening of the vowel, on the penultimate syllable of words with four or more syllables. This pattern is the same as that of words in Polish under narrow focus, since in words under narrow focus the ‘‘normal’’ prominence relation is reversed: primary accent occurs on the initial syllable and secondary accent on the penultimate syllable. In a subset of Polish words, which must be regarded as lexically marked, primary accent occurs on the antepenultimate or final syllable. The extinct Slavic language Polabian seems to have had a bounded weight-sensitive accent system. Accent is said to have occurred on the final syllable if it was heavy and on the penultimate syllable if the final syllable was light. Unfortunately not all the logical combinations of long and short syllables within the final two-syllable window occur in surviving Polabian materials. Slovincian, the northernmost dialect of Kashubian, also had a weightsensitive system. It may have formed a bridge between the South and East Slavic languages on the one hand and the weight-insensitive West Slavic languages on the other. Like Slovene, Serbian and Croatian it appears to have had some sort of tonal accent system, with lexical marking like the East-Slavic languages. The default in Slovincian, however, is initial accent. The Slovincian system can in fact be described as an unbounded LAST/FIRST system: accent occurs on the rightmost tonal accented syllable or on the first syllable. Since many of the historical weight and

Word accent systems in the languages of Europe

455

accent distinctions were eliminated through levelling processes the major part of Slovincian words ended up with initial accent, like the other West-Slavic languages. In the other Kashubian dialects initial accent has indeed become the rule. SOUTH-SLAVIC StressTyp extracts Slovenian; Slovene [F/L/L (tone)] f Stress falls on the first syllable with a strong low tone. f If there are no strong low tones, stress the last tone. f If there are no tones, stress the last syllable. me¨du@ø

‘honey gen.sg’

me¨do$øv ‘honey gen.pl’

od¨me@ødu

‘of honey gen.sg’

me¨do@øvih ‘honey.loc.pl.’

Serbo-Croatian [F/F (pitch)] f Stress is assigned to the first syllable with a tone, else to the initial syllable. bra¨to&vStina ‘brethren community’ Macedonian [A] f Stress falls on the antepenultimate syllable (with lexical exceptions). vo¨denitSar ‘miller’

¨polkovnik ‘colonel’

¨vetSer ‘evening’

Additional information There are Old Church Slavonic texts from the 9th century. Church Slavonic is still in use, but the accent system of Old Church Slavonic is not known. The South Slavic linguistic area can be divided into eastern South Slavic (Bulgarian and Macedonian) and western South Slavic (Slovene, Serbian and Croatian). The word prosodic systems of the eastern and the western South Slavic languages di¤er considerably. Whereas Macedonian has a fixed accent system, with regular antepenultimate accent and Bulgarian has a lexical accent system, in which accent has to be diacritically marked (much like Russian, see below), the western South Slavic languages (or at least their standard varieties) have tonal-accent systems and surface primary accent that is dependent on pitch. Gvozdanovic´ (1999) focuses on these latter systems. Of the three

456

Harry van der Hulst

western South Slavic languages standard Serbian and Croatian (which until recently were classified as one language: Serbo-Croatian; Browne 1993) have the same word-prosodic systems. In her description, they have lexically marked high tone which spreads one syllable to the left where possible. Primary accent falls on the first syllable bearing tone. In words with more than one lexical tone (or tonal accent), all but the rightmost tone is deleted before primary accent is assigned; cf. also Inkelas & Zec (1988). In toneless words accent falls on the leftmost syllable. Serbian and Croatian appear to have a LAST/FIRST accent system: primary accent falls on the rightmost high tone or on the first syllable. The Slovene system di¤ers considerably on the surface. It has lexically marked low tone as well as high tone, and the low tones may be diacritically marked as strong. Accent falls on the leftmost strong tone in a word or, in the absence of a strong tone, on the rightmost ‘‘normal’’ tone. In toneless words accent falls on the rightmost syllable. Accent in Slovene, then, can be described as a FIRST/LAST/LAST system. Gvozdanovic´ shows that the complicated word prosodic system of this language is undergoing a simplification process whereby alternations within the paradigm of a stem are for the most part eliminated. Generalizations The Balto-Slavic languages show a great diversity of accentual systems, ranging from very complex ones (in which word accent is based on tone) to very simple edge-based accent systems. A significant split seems to be that in languages where word accent is bounded and usually purely edgebased (i.e. weight-insensitive) and unbounded systems. Czech, Slovak, Sorbian, Polish, and Macedonian belong to the former category while Russian, Ukrainian, Byelorussian, Bulgarian, Serbian, Croatian, Slovene, and Lithuanian belong to the latter. Only the now extinct languages Polabian and Kashubian (Slovincian) are somewhat di‰cult to classify, but it would seem that the latter was probably unbounded (LAST/ FIRST), whereas the former could be the only example of a right-edge bounded weight-sensitive system (penult/final). Within the group of languages that have bounded (and as it turns out weight-insensitive) accent for at least the great majority of their lexicon a division can be made between systems in which accent is assigned to the left edge of the word and those were the position of accent must be determined with respect to the right edge of the word. The first group comprises Czech, Slovak, Sorbian and Latvian, all with fixed initial accent. Polish

Word accent systems in the languages of Europe

457

and Macedonian, with penultimate and antepenultimate accent respectively, make up the second group of bounded weight-insensitive systems. The weight-sensitive systems can also be subdivided, depending on the criterion for weight. Serbian, Croatian, Slovene, and Lithuanian (as well as Slovincian) are usually analysed as pitch-accent systems (but see Dogil 1999b for an analysis of Lithuanian as a lexical accent system). In such systems the position of word accent is dependent on tone, which is itself analyzed as either lexically present, as assigned to lexical accent marks (cf. van der Hulst 1999, section 5). Zec (1994) provides an analysis of Neo-Stockavian in which the position of the lexical tone/accent is predictable from phonological and morphological information. The East-Slavic languages and Bulgarian are languages in which weight is ‘‘diacritic’’ (cf. van der Hulst 1999, section 2.2.2). Thus the position of accent must be specified lexically for an important part of their morpheme inventory. Systems of this kind are usually called lexical accent systems. The diversity of the stress patterns of Slavic languages has always been a challenge to typological and historical linguistic research. The endeavour has been to find a common core for languages which: i.

ii.

iii.

show quantity sensitivity (apparent in by now extinct Polabian and marginal Slovincian (West Slavic), and profoundly present in Slovene and Serbo-Croatian (South Slavic)). show an apparently quantity insensitive stress pattern with a syllabic trochee as a base foot type like Polish, Czech, Slovak, Sorbian, Macedonian and Kashubian. show prosodic marks on specific – mostly lexically designated – morphemes (Russian, Ukrainian, Byelorussian (East Slavic); Bulgarian, Slovene, Serbo-Croatian (South Slavic)).

Traditional research into Slavic accentology has attempted to reconstruct the diversity of the observed system in a single proto-system (cf. Kuryłowicz 1952; Stang 1957; Garde 1976; Dybo 1981; Stankiewicz 1993). Most of the researchers (see Stankiewicz 1993, for a di¤erent position) see such a proto-system in the vicinity of the accentual system as represented by the Baltic languages, and particularly Lithuanian. The main accentual properties of historical Baltic (and present day Lithuanian) are rich enough to comprise the parameters found in the Slavic accentual systems: – Lithuanian has ‘‘accent’’ realizations (‘acute’ and ‘circumflex’) which are quantity sensitive (see 1 above). – dialects of Lithuanian, and particularly Latvian, show regular foot based fixed stress patterns (see 2 above).

458

Harry van der Hulst

– Lithuanian has a wide range of root morphemes, which have an ability to receive and preserve stress in a paradigm, as well as a‰xal morphemes, which may exert an influence on the accentual properties of stems (see 3 above).18 4.1.5. Italic (< Indo-European) Genetic information ITALIC UOSCO-UMBRIAN (also called SABELLIAN): Oscan, Umbrian, Sabellian LATINO-FALISCAN: UFaliscan, ULatin ROMANCE (i.e. the development of the Latin that was spread throughout the Roman empire) SOUTHERN ROMANCE/SARDINIAN: Sardinian, Corsican CONTINENTAL: EASTERN NORTH: Rumanian, Istro-Rumanian SOUTH: Megleno-Rumanian, Arumanian, Moldovian WESTERN: ITALO-ROMANCE (Italo-Dalmation) UDALMATIAN: Dalmatian ITALIAN: Italian, Sicilian, Neopolitan RHAETO-ROMANCE: Friulian, Ladin, Romansch GALLO-IBERO-ROMANCE: GALLO-ROMANCE: NORTH: French SOUTH: Provencal IBERO-ROMANCE: NORTH EASTERN: Catalan CENTRAL: Spanish WESTERN: Galician, Portuguese USOUTH: Mozarabic

18. The information on diachronic aspects was kindly furnished by Grzegorz Dogil.

Word accent systems in the languages of Europe

459

Latin is recorded from the 3rd century BCE, although there are inscriptions from the 6th century BCE. Latin suppressed other varieties of Italic (Faliscan, Oscan, Umbrian, Venetic) that died out soon after the beginning of the Common Era. Around 500 AD we find local dialects of Vulgar Latin (which was di¤erent from literary ‘Classical Latin’) which began to develop into di¤erent Romance languages. Of the modern languages Italian is closest to Latin. Other languages underwent considerable influence, French from Celtic and German, Rumanian from Slavic languages. Modern Italian has important dialectal di¤erentiation, for example: Neapolitan-Calabrese, Piedmontese, Sicilian and Venetian. Harris and Vincent (1997) contains sketches of the Romance languages. StressTyp extracts Latin [P/A] f Stress falls on the penultimate syllable if it contains a long vowel or is closed. f Else stress is antepenultimate. re¨feøcit

re¨fectus

¨reficit

Sardinian, Campidanese [U/P] f Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). f Final stress is obligatory when the final syllable is closed by a palatal. f In verbs, stress has become morphologised. ¨mendul a ‘almond’

mengi¨an u

‘morning’

¨atom u

a¨tomic u

‘atomic’

‘atom’

Rumanian [U/P] f Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). f In verbs, stress has become morphologised. ¨lingur « cu¨m«tru

‘spoon’ ‘godfather’

mui¨er e ‘woman’ maha¨la ‘suburb’

460

Harry van der Hulst

Italian [U/P] f Stress falls on the final syllable if lexically marked as accented, else stress is on the penultimate syllable of the stem (without desinence). do¨menic a ‘sunday’

pal¨to

‘coat’

¨atom o ‘atom’

Romansch, Berguner (dialect of Rheto-Romance) [U/P] f Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). f Final stress is obligatory when the final syllable is closed by a palatal. f In verbs, stress has become morphologised. mu¨leł l«¨venj«

‘mill’ ‘avalache’

bu¨lef ¨p«le

‘mushroom’ ‘swamp’

Provenc¸al; Occitan [U/P] f Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). f Final stress is obligatory when the final syllable is closed by a palatal. f In verbs, stress has become morphologised. ca¨dena

‘chain’

fi¨nestra ‘window’

espi¨tal

‘hospital’

tara¨bast

‘uproar’

French [U/P] f Primary stress falls on the final syllable, except when that is a /«/. f A likely alternative is that French has just a phrase accent. f Secondary stress claimed to exist, but there is not much evidence. aøbriø¨koø

‘apricot’

ystEn¨sil ‘utensil’

¨kEl«

‘which [fem.]’

Catalan-Valencian-Balearic [U/P] f Stress falls on the final syllable if closed or lexically marked, else on the penult. pa¨ awl«

‘word’

cam¨pana ‘bell’

bǬrEts

‘hats’

ele¨fan

‘elephant’

Spanish, Castilian [U/P] f Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). f Final stress is obligatory when the final syllable is closed by a palatal. f In verbs, stress has become morphologised.

Word accent systems in the languages of Europe

¨termino ‘terminus, end’

su¨til

¨util

bu¨fand a ‘scarf ’

‘useful’

461

‘subtle’

Portuguese [U/P] f Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). f Final stress is obligatory when the final syllable is closed by a palatal. f In verbs, stress has become morphologised. ¨perola

‘pearl’

pa¨nela

‘pan’

chami¨ne

‘chimney’

pai¨nel

‘panel’

Portuguese, Brazilian (dialect of Portuguese) [P] f Stress falls on the penultimate syllable. ¨ama ‘he loves’

ago¨nia ‘agony’

Additional information and generalizations Vineis (1998) says that Early Latin had a stress accent on the initial syllable and that the shift to the right occurred in Classical Latin, perhaps due to influence of Greek. Silvestri (1998: 238) and Wallace (2004a: 822) ascribe initial stress accent to the UOsco-Umbrian (Sabellian) group. Wallace (2004b) thinks that Venetic also had initial or near-initial accent. Lahiri et al. (1999) o¤er a study of the development from Early Latin to the modern Romance lanuages. The modern Romance language family comprises five o‰cial state languages, a number of other languages with an o‰cial status, and numerous dialects. Roca (1999) analyses the accentual systems of these five state languages (French, Spanish, Portuguese, Italian and Rumanian), as well as of four additional languages that are traditionally considered in Romance philological studies: Occitan, Catalan, Sardinian and RhaetoRomance (Romansh). Of these, the accent system of French is the odd one out and will be discussed separately in this summary. Because the accent systems of the other Romance languages have so much in common they will not be discussed individually but as a group. The position of accent in the Romance languages (minus French) is generally restricted to the three-syllable window at the right edge of the word. In this respect, then, the Romance languages (except French) are typologically very similar to the Germanic languages (except Icelandic &

462

Harry van der Hulst

Faroese), except for the e¤ect of stress-neutral su‰xes in the latter which can locate accent outside the three-syllable window. While StressTyp has the Romance languages (except for French) codes as [U/P] all of them also allow antepenultimate stresses. Portuguese, although coded as [P], does not have a very di¤erent system from the other Romance languages, except again for French. French has predominantly final accent. Only words ending in a syllable containing a schwa have penultimate accent, but word-final schwa in French is only pronounced under very specific circumstances nowadays. It has been argued, however, that accent in French is not a word-level phenomenon (e.g. Pulgram 1970, Dell 1984) but operates on the phrase level. In all Romance languages, most nouns consist of a stem and a desinence. In stems that occur without a desinence accent is usually restricted to the final two syllables. Desinences are systematically accentless. Roca (1999) achieves this by formulating a rule which assigns extrametricality to all desinences, the Romance Extrametricality Rule (RER). The RER allows for antepenultimate accent if one assumes that feet are left-headed binary. This system, then, seems very close to that of Classical Latin with its trochaic feet and final extrametricality. Classical Latin, however, had a quantity sensitive accent system in which closed syllables and syllables with long vowels were heavy. Accent fell on the penultimate syllable if heavy and otherwise on the antepenultimate syllable. Although contrastive vowel length has disappeared in the Modern Romance languages, accent still falls on the penultimate syllable in most words which have a closed syllable in that position and also in many words with an open penultimate syllable (i.e. those that used to have a long vowel (in most cases). Accent has thus become partly unpredictable and potentially contrastive. Roca proposes to analyse words with penultimate accent as undergoing the Romance Accent Rule (RAR), which places an accent on the stem-final syllable. Words which do not undergo the RAR must be lexically marked as [- RAR]. In these words a left-headed (trochaic) binary foot will be built on the final two syllables of the stem, resulting in antepenultimate accent when the stem is followed by a desinence. Due to the RAR or the final trochee, words will bear an accent on either the final or penultimate stem syllable. Primary accent results from elevating this accent to primary accent status.19 19. In fact, given the final accent assigned by the RAR, all nouns could be said to undergo trochaic footing. With an accent on the final syllable, the trochaic foot will necessarily be monosyllabic.

Word accent systems in the languages of Europe

463

In the verbal system accent is computed by means of a number of, partly language specific, rules. The most general of these is the Romance Verb Accent Rule (RVAR). This rule places an accent on the theme vowel, the vowel that in some way expresses the conjugation class of the verb. In the future tenses this rule is overruled by the Future Accent Rule (FAR), which places accent on the future su‰x. In the present tense the position of accent is predicted best by the RAR, which also determines the position of accent in nouns. Some languages have a fourth rule, the 1st plural/2nd plural Accent Rule (1/2 pl.), which moves the accent one syllable to the right only in the first and second person plural. This rule does not apply across the board in the languages in which it exists, but needs additional morphosemantic contextualisation. All non-finite forms generally undergo the RVAR. Thus Roca’s analysis suggests that the accent system of nouns and verbs is di¤erent in nature. In the former primary accent location is based on accent rules (the RAR and trochaic footing), whereas in the latter we rather deal with a lexical accent system in which the rightmost accent is elevated to primary accent. The complexity of the above-mentioned rules can lead to ‘minimal’ triplets, as for example in Italian: (10) ca´pito ‘I turn up’ capı´to ‘understood’ capito´ ‘he turned up’ Stress can occur outside the thee-syllable window when clitics are added; again the following examples are from Italian: (11) ca´pitano ‘they turn up’ fa´bbricamelo ‘make it for me’ Apart from some compound-like su‰xes (to be mentioned below) Romance a‰xes may be divided into two classes, like in Germanic. Roca describes this division as one between cyclic and non-cyclic a‰xes. Cyclic a‰xes are marked such that they trigger cyclic application of the stress rules. Before application of a cyclic rule all metrical information and structure of the preceding cycle is erased by the Stress Erasure Convention (SEC). Non-cyclic su‰xes, crucially, do not trigger the SEC does not apply. Cyclic a‰xes, then, are comparable to the Germanic Class I a‰xes, in that the same accent rules apply to a stem with Class I a‰xes as to a stem in isolation. Non-cyclic a‰xes can be compared to Class II a‰xes in that the original accent pattern of the stem is left intact. The di¤erence

464

Harry van der Hulst

between Romance and Germanic languages is that whereas the Germanic Class II a‰xes are completely accent-neutral, the Romance non-cyclic a‰xes are a¤ected by the Romance Accent Rule. Thus, although the metrical structure of the stem is left intact when it enters the non-cyclic phase, the non-cyclic su‰xes may a¤ect the surface position of accent all the same. There do not seem to be any completely accent-neutral a‰xes in Romance. In Romance compounds primary accent falls on the second member (the opposite of what we find in Germanic where the initial member usually carries primary accent). There has been a lot of debate on whether or not the accent of the first member of compounds is retained. Although this first member does not have a clearly prominent syllable in all languages, certain segmental processes (or the non-occurrence of such processes) indicate that the accent on that member is present at an earlier cycle. Thus, the syllable that would be accented when the first member occurs in isolation will fail to undergo reduction in e.g. Romansh, is lengthened like other accented syllables in Italian, diphthongises in Spanish, etc. The Romance languages are perhaps the most uniform group in the Indo-European family with respect to accent. Except for French, they all have the same basic system with only very slight language-specific deviations. Basically, accent occurs on one of the final two syllables of the stem. Synchronically the exact position is not always predictable because syllable quantity distinctions in Latin have not carried over into the Romance languages, although the Latin accent pattern has been retained. Thus, the position of accent in the modern languages must be governed by some sort of accent rule. According to some researchers, French seems to lack a word accent rule, accent being more of a phrase-final phenomenon. As will be clear, Roca (1999) study entails some obvious corrections of and additions to the StressTyp information. 4.1.6. Armenian (< Indo-European) Genetic information ARMENIAN UClassical Armenian, Armenian Armenian is closest to Greek in terms of ancestry and has been influenced by Iranian and perhaps UPhrygian. The heaviest influence on (Modern) Armenian is from Turkish.

Word accent systems in the languages of Europe

465

StressTyp extracts Armenian [L/ F] f Primary stress falls on the last full vowel, else on the first. f In some dialects secondary stress falls on the first syllable. Jer¨pHem«n

‘sometimes’

¨kutemn«

‘cress’

Additional information Proto-Armenian had fixed accent on the penultimate syllable, perhaps under the influence of UUrartian (Ajello 1998: 202). Clackson (2004: 927) attributes final stress to Classical Armenian. Modern Armenian has two standard dialects, an eastern and a western dialect, which can be subdivided into about 36 (sub)dialects. Most of these dialects have accent on the final full (¼non-epenthetic) vowel (e.g. yerphe´m«n ‘sometimes’) but some have accent on the penultimate full vowel (e.g. ku´temn« ‘cress’) or, in a sequence full-schwa-full on the final full vowel. It is also reported that in all dialects secondary accent regularly falls on the initial syllable. In words that do not have any full vowels primary accent falls on the initial syllable. Vaux (1994, 1998) presents an analysis of both LAST/FIRST and penultimate systems. The default first location is reminiscent of the PIE system, which, in the non-default case, would accent the first rather than the last heavy syllable. 4.1.7. Albanian (< Indo-European) Genetic information ALBANIAN Albanian: Gheg, Tosk Albanian (spoken in Albania, in Kosovo, Greece, Italy, Turkey) has two major mutually unintelligible dialects: Gheg (in the north), Tosk (in the south), with further internal divergence. The o‰cial language is based on Tosk. The extinct languages UIllyrian and UThracian have been advanced as ancestor languages but this is controversial. Accent information Accounts of accent location in Albanian range from unpredictable to statements to the e¤ect that accent generally falls on the final syllable of

466

Harry van der Hulst

stems, unless this syllable contains a schwa (orthographically , which, however, is not pronounced anyway in most varieties of standard Albanian) in which case accent falls on the penultimate syllable. Most derivational su‰xes are accent-sensitive. When occurring in final position they will bear the primary accent, while the originally accented syllable will bear secondary accent: pu´ne¨ ‘work’, pu`ne¨to´r ‘worker’. Certain derivational su‰xes (notably all non-verbal su‰xes ending in ) and nearly all inflectional su‰xes are accent-neutral; when added to a stem they do not change the position of the primary accent. Thus, word accent in Albanian remains invariant throughout the inflectional paradigm of a stem: ma´l ‘mountain’, ma´le ‘mountains’, ma´leve ‘of mountains’. The definite article, which in Albanian is added to the end of the noun, is also accent-neutral. In compounds primary accent falls on the second member: ze´me¨r ‘heart’ þ gje´re¨ ‘broad’ ¼ ze`me¨rgje´re¨ ‘generous’. Phraseaccent falls on the final word in a phrase (cf. Bevington 1974; Newmark, Hubard and Prifti 1982). Trommer (n.d.) contains the most detailed analysis of Albanian accent, presenting an analysis of 17077 stems, which in 11.006 cases have final stress and only about 50 cases having accent before the penultimate stem syllable. He shows that the placement of accent is sensitive to syllable structure. Trimoraic final syllables almost always have final stress (1084 out of 1088), whereas bimoraric syllables have final stress in 8663 out of 9238 cases. In the case of vowel final stem the location of stress is sensitive to vowel quality (there is no vowel length) with peripheral vowels (/a,i,u/) being more likely to be stressed than mid vowels (/e,o/). Final closed syllables with a schwa do not tend to be stressed unless they are trimoraic. Demiraj (1998: 486) notes that in ‘the oldest nouns’ stress most frequently falls on the first syllable. Does this indicate that Old Albanian had an initial pattern? Apparently not; Beekes (1995: 149) says that in Old Albanian accent was located on the last syllable of verbal stems and prepositions, and often on the penult in nouns. 4.1.8. Greek ( CV: > CV. In each syllable type the vowels /a, e, o/ are ‘heavier’ than the vowels /i, u, y/. In nouns there is a general restriction that lighter syllables may not precede heavier syllables. This means that the final syllable is always at least as heavy as the syllables preceding it and hence, that if accent is weight-sensitive it is expected to occur on the final syllable. For verbs the accentual patterns and phontactic restrictions are slightly di¤erent. We see that in the Daghestanian languages a whole array of word prosodic systems occurs. The southern languages are accent languages in which the position of the accent is determined mainly by weight and edge. The northern languages, which seem to be more archaic phonologically, are tone languages in which accent, if it occurs, is tone-sensitive. Some of the northern languages may be in a transitional stage in that an accentual pattern seems to be developing out of a purely tonal pattern. Generalizations The Kartvelian languages as a whole exhibit a great variety of word prosodic systems, ranging from purely tonal systems in the northern

484

Harry van der Hulst

Daghestan languages to bounded weight-sensitive accent systems. In those Caucasian languages that have accent, its position may be lexical (Abkhaz), determined by pitch-accents or tone (Avar, Godoberi), quantity sensitive (Lezgian), sensitive to vowel-quality (Archi), or fixed (Tsakhur). Both bounded and unbounded systems occur. There even appear to be Caucasian languages in which not only the position but the very presence of accent is lexically determined (Bagvalal). In some of the languages accent seems to fluctuate (Adyghe or West Circassian). Thus, examples of most types of word prosodic systems can be found in the North Caucasian languages. The great typological diversity of accentual systems raises the question about the prosodic system of the proto-language. So far there has been no attempt to reconstruct this system and to follow its evolution into the modern systems. With respect to the Daghestanian accentual systems, only a very tentative opinion on the topic was expressed in Nikolajev & Starostin (1994). They believe, that the prosodic type presented by such Northern languages as Andi and Akhvakh (no stress, syllabic tones, open syllables) is the oldest. Stress seems to be a rather late development. Arising in the North, it reflected an inherent prosodic structure of words (i.e. tonal and quasi-tonal properties). Arising in the South, it was oriented towards the edge and quantity sensitive. This di¤erence is probably due to the fact that the inherent prosodic structure of words was already di¤erent in these two groups at the moment that stress arose.21 4.5. Uralic (minus Yukaghir 22) Genetic information URALIC SAMOYED (Siberia) NORTH: Nenets (Yurak), Enets (Yenisei Samoyed), Nganasan (Tavgi or Aram) SOUTH: Selkup (Ostyak Samoyeed), Mator, UKamas FINNO-UGRIC UGRIC: Hungarian OB-UGRIC: Ostyak (Khanty), Vogul (Mansi) FINNO-PERMIC: PERMIC: Udmurt (Votyak), Komi-Zyrian, Komi-Permyak

21. This information was kindly made available to us by Sandro Kodzasov. 22. Yukaghir is discussed in chapter 10, section 4.2.2.

Word accent systems in the languages of Europe

485

FINNIC: CHEREMITIC: Mari (Cheremis) FINNO-MORDVINIC: MORDVINIC: Mordvin (Moksha), Ezra FINNO-LAPPIC SAAMIC (¼LAPPIC): (East, South and West) Saami BALTO-FINNIC: Finnish, Ingrian, Karelian, Olonets, Ludic, Livonian, Votic, Vespian, Estonian

The Ethnologue has a much flatter structure with all major branches of Uralic as equal daughters (Finnic, (Finno-)Ugric, Mari, Mordvin, Permian, Sami, Samoyed, Khanty and Mansi). Proto-Uralic is located in time between 5,000 and 8,000 BCE, geographically in the Northern Ural Mountains. Samoyed moved northeast into Asia, while Ugric went south. Ugric split from Finnic some 3000 years ago. The oldest text is from the 12th century; it regards Hungarian. A link between Uralic and Altaic has been proposed, as well as links to IE, Dravidian, and to Eskimo-Aleut. The Uralic a‰liation of Yukaghir is controversial. StressTyp extracts Samoyed Nenets, Tundra [I] f Primary stress is initial. f Secondary stress occurs on all syllables preceding a syllable with schwa. f Secondary stress also falls on non-final syllables preceded by an unstressed syllable. ¨ya ¨xampol

‘earth’ ‘litter’

¨wada ¨xOrOr«

‘word’ ‘knife poss. nom.sg’

Taz Selkup (dialect of Selkup) [I/I] f Primary stress is initial, except when the first vowel is short and the second is long. f Secondary stress on odd syllables after the main stress. ¨syr« ‘snow’ ¨ta¨˙t«ki˙ne ‘reindeer.lat.dat.pl.’

¨syøre ‘cow’ am¨qeøNa ‘is to take’

Additional information The speakers of the three Samoyedic languages Nenets (Yurak), Nganasan and Selkup live scattered across vast areas of Northern and Central

486

Harry van der Hulst

Siberia. Of these only Nenets extends into Europe. It is spoken across a vast area stretching from the White Sea in European Russia to the delta of the Yenisei River in Asia. De´csy (1966) describes the position of accent in Nenets as follows: ‘‘Stress usually falls on one of the first, second, third, fifth, or seventh syllables’’. Since he presents only two-syllable examples it is not quite clear how this description should be interpreted. It may be the case that primary accent in Nenets falls on either the first or the second syllable and that secondary accents fall on odd-numbered syllables following the primary accent. Another possible interpretation of De´csy’s statement is that the system is unbounded, accent location being dependent on syllable weight. Salminen (1998: 519–520), the source for StressTyp says that accent occurs initially, preceding a syllable with a schwa, or on a syllable preceded by an unstressed syllable; final syllables are never stressed. The acoustic correlates of Nenets primary accent are analyzed in Kavitskaya (2006). As stated in StressTyp, Nenets accent is initial with some kind of alternating trochaic rhythm, but Kavistkaya reports that earlier descriptions provide other accounts, including statements about lexically marked accent and final accent. Kavitskaya restates the description given in StressTyp (which comes from Salminen 1998) as follows: if the schwa is initial it gets the accent. If it is non-initial it remains unpronounced unless it is followed by another schwa in which case it receives secondary accent. Selkup has been referred to as a FIRST/FIRST system (e.g. Idsardi 1992). That the position of accent cannot be entirely predicted from phonological factors appears from minimal pairs like: te´va ‘tail, stern’ and teva´ ‘to reach’. According to De´csy, vowels in unaccented syllables are often pronounced ‘‘unclearly’’ (reduced, weakened). According to Helimski (1998b), Selkup accent is placed on the last long vowel, or in the absence of long vowels, on the first vowel. Exceptions are due to some su‰xes that have short vowels that behave as if they are long; these attract stress unless followed by a long vowel (in accordance with the general rule). Other su‰xes create a separate accentual domain so the general rules apply to this domain and the remainder of the word. There are also cases in which morphologically di¤erent words have di¤erent accentual patterns, but it is not clear what the relevant factors are. Kuztnetsova et al. (1993) also suggest an LAST/FIRST analyses of Selkup, adding that accent is variable, and that sometimes words contain multiple accents. Gordon (2000) speculates that perhaps all heavy syllables have a degree of accent. Nganasan, according to Helimski (1998a: 486), has primary accent on the penultimate vowel. If that vowel is high or /«/ we can have a retraction to the antepenultimate syllable. Longer words (which are common) are

Word accent systems in the languages of Europe

487

divided into ‘rhythmic bisyllabic groups’, the rightmost being possibly trisyllabic. This latter statement does not mean that the main stress in that case is antepenultimate. (13) k«ørI_[g«ølji[tı´ni k«ørI_[g«ølji[tinı´n« k«ørI_[g«ølji[rI_øa[tinı´n«

‘in marches’ ‘in my marches’ ‘only in my marches’

Note that primary stress is always PU. In the second and third example, the first vowel of the last rhythmic group is not lengthened. Hence, perhaps one must assume that in this case the last but final group is trisyllabic. However, Helimski says that the breaks between breath groups are clearly noticeable, creating the impression of a glottal stop. Finno-Ugric: Ugric Hungarian [I] f Primary stress falls on the first syllable. f Secondary stress usually on alternates after the primary stress. f If the secondary stress would fall on a short vowel, the pattern might become ternary. f In the latter case, secondary stress is also reported to avoid final syllables. ¨alma

‘apple’

¨kpaø˙vl

‘with who’

¨tEriø˙tOøvEl

‘with tablecloth’

¨feølE˙mElE˙tEn

‘on mezzanine’

Mansi; Vogul [I] f Primary stress falls on the first syllable. f Secondary stress usually on alternates after the primary stress. f Sources di¤er on whether the final syllable can be stressed or not. ¨tAørAtI

‘to leave, to let’

¨tAørA˙tANkwe

‘to leave, to let.inf.’

Additional information The only Ugric language that is spoken in Europe is Hungarian, the other two members of the Ugric group, Khanty and Mansi, being spoken in Western Siberia. Hungarian is not only geographically far removed from the other two, but also linguistically. Proto-Obugrian probably had initial accent (Honti 1998: 332). Primary accent in Hungarian is fixed on the initial syllable. According to Varga (1994: 234) a primary accented syllable has extra intensity and

488

Harry van der Hulst

bears an intonational pitch-accent. Secondary accented syllables also have extra intensity but are not linked to intonational accents. There is some controversy as to the position of secondary accent in Hungarian and some scholars deny it is there (cf. Sipta´r and To¨rkenczy 2000). Szinnyei (1912) and Lotz (1939) describe secondary accent as falling on the third and fifth syllable or (if the third syllable is light) on the fourth and sixth syllable, but never on the last syllable. This would be a pattern very close to that of Finnish. On the other hand, the system described by Balassa (1890), Hall (1938) and So¨vijarvi (1956) is completely weight-insensitive with secondary accent on odd-numbered syllables following the primary initial accent. According to Hammond (1987) the non-primary accents are not all of equal intensity. He states that secondary and tertiary accents alternate, secondary accents occurring in odd-numbered non-initial feet. On this ground he postulates an intermediate level between the foot and word level in Hungarian, the so-called ‘‘cola’’ (sing. colon). In his examples odd-numbered syllables in final position do bear accent: me´gveszte`getheˆtetle`nekneˆk ‘to unbribable (ones)’ (secondary accent is marked /^/, tertiary accent / `/). The di¤erence in the descriptions may, according to Hayes (1995), represent a dialect split. Matthews (1951: 22) says that in Ostyak (Khanty) accent is usually final. Finno-Ugric: Finno-Permic: Permic Yazva (dialect of Komi-Zyrian) [F/L] f Stress on the first heavy vowel. f Heavy syllables have low vowels. Weight of high vowels fluctuates. f Words without heavy syllables have final stress. Komi-Permyak [F/L] f Stress falls on the first syllable with a long vowel. f If there are no long vowels, stress is final. Additional information In proto-Permian accent was probably initial. Today, we find a di¤erence between Udmurt (Votyak) and Zyrian-Permian. In Udmurt (Votyak) words have final primary accent (except in certain grammatical forms), and sometimes a secondary accent on the first syllable: ko`rkajosm ‘our room’. This, for a Finno-Ugric language unusual,

Word accent systems in the languages of Europe

489

accentual pattern may be explained as caused by the influence of Turkic languages, especially Tatar. Both Tatar and Votyak also have initial stress in the imperative form of the verb. Csu´cs (1998: 280–281) adds that initial stress is also found in negative verb forms, as well as reduplicative adjectives and adverbs. With respect to Zyrian-Permian, according to Riese (1998) sometimes a political or administrative distinction is made between Komi-Zyrian and Komi-Permyak, but this di¤erence, he says, represents a minor dialectal di¤erence. Below we see that some di¤erences in accent location do appear to exist. The Yaz’va dialect of the Eastern Permic language Komi-Zyran is an example of an unbounded weight-sensitive FIRST/LAST system. Accent falls on the leftmost heavy syllable, and if there is no heavy syllable on the rightmost syllable. The notion weight is rather complex in KomiZyran, however. The non-high vowels a, e, o¨ and o are always accented, while the accent-behavior of the vowels i, u¨ and u depends the origin; some of these used to be non-high and they continue to behave so. Hausenberg (1998: 310–311) remarks that in Komi-Zyrian accent is free (variable) although there is a slight (pitch-based) prominence on the first syllable. In southern Komi-Zyrian accent, he says, is partially ‘morpheme-bound’. Gordon (2000) interprets Hausenberg’s (1998) claim that all heavy syllables are stressed as saying that heavy syllables that are not primary accented have secondary accent. He also says that accent can be called morphological: accent never falls on inflectional su‰xes, whereas certain derivational su‰xes are always accented. In western Permyac, or Komi-Permyac, there are certain dialects which resemble the Zyran system, but there are also dialects which have a completely di¤erent type of accentuation. In these dialects the position of the accent cannot be predicted by phonological rules only but is partly morphological. Accent always occurs on the stem, su‰xes are never accented. Thus, minimal pairs can occur, like jua´n ‘a drink’ vs. ju´an ‘you drink’ (Itkonen 1955). Finno-Ugric: Finno-Permic: Finnic Mari, High; Mountain (Western) Cheremis [L/L or P/P] Sources di¤er. Some quote stress is penult except when the antepenultimate vowel is full and the penult is not. Others quote stress occurs on the last non-final long vowel or the penultimate syllable.

f

490 f

Harry van der Hulst

Examples point in the direction of the latter.

aa¨Baaxaa ‘pod’ ¨kaan`EsEr ‘sorrow’

¨«Sta¨S Bˆlaa¨˜aanEsˇtEsˇE

‘to sweep’ ‘comedian’

Mari, Low; Meadow (Eastern) Cheremis [L/F;IRR] f Stress the last full vowel, and the first in words with only reduced vowels. f Free variation of final and non-final stress reported. Slaa¨paaZ«m or Slaapaa¨Zem ¨t«l«z«n or t«l«¨z«n ¨pu¨g«lmo¨ or pu¨g«l¨mo¨ ¨BoSt«l

‘his hat.acc’ ‘moon.gen’ ‘cone’ ‘laugh’ BoSt«¨lam ‘I laugh’

Moksha; Mordvin; Mordva [F/F] f Stress the first full vowel, and the first in words with only reduced vowels. f Free variation of final and non-final stress reported. tu¨cJa¨nJa¨

‘cloud’

¨puv«nd«ms ‘to press’

Additional information The accent systems of both Mordvin and Mari (Cheremis) are described by Kenstowicz (1994). The Mokshan dialect of Mordvin has two groups of vowels that behave di¤erently with respect to accent: the ‘narrow’ vowels /i u «/ and the ‘wide’ vowels /e o a¨ a/. According to Tsygankin and Dabaev (1975: 32–33) wide vowels attract word accent, i.e. they function as heavy syllables for the purpose of accent assignment. Accent falls on the first syllable containing a wide vowel. In words with only narrow or only wide vowels accent falls on the initial syllable: tuc ja¨n ja¨ ‘cloud’, pu´v«nd«ms ‘to press’. The Mokshan system is thus an example of an unbounded weight-sensitive FIRST/FIRST system. Just like in Komi, weight is determined by vowel quality rather than quantity. It is interesting to note that the Erzyan dialect of Mordvin does not distinguish between heavy and light syllables but simply has fixed initial accent. Zaicz (1998: 190–191) says that stress in the Erzya dialect of Modrva is ‘free and non-distinctive’ which means that in any given word any syllable can be given ‘accentual prominence’; unstressed vowels are not reduced. This being said, first syllable stress is said to be quite common. In Mari, a sister language of Mordvin, the position of accent is also

Word accent systems in the languages of Europe

491

conditioned by vowel quality. Mari distinguishes full and reduced vowels, full vowels acting as heavy for the purpose of accent-assignment. In Literary Mari accent falls on the last full vowel and in words with only reduced vowels on the initial syllable. Literary Mari is thus an example of an unbounded weight-sensitive LAST/FIRST system, the mirror image of Komi. One complicating factor in Literary Mari is that final open syllables are never accented. We must assume that these are extrametrical. Generally, two Mari dialects are distinguished: Western (Hill) Mari and Eastern (Meadow) Mari (Kangasmaa-Minn 1998: 224). The Literary dialect has the western accentual pattern, i.e. it has a LAST/FIRST system sensitive to the di¤erence between full vowels and reduced vowels (« is a reduced vowel): (14) olma´ ‘apple’

mu´n«

‘egg’

t«=l« z« -m ‘moon’

Eastern Mari has a di¤erent rule: accent is PU even if the PU vowel is reduced. If, however, the vowel preceding the PU reduced vowel is full, accent may occur on the APU. These characterizations of the western and eastern dialect. Northwest Mari has been described in Ivanov and Tuzharov (1970) who say that accent falls on the rightmost non-final heavy syllable (syllables with a full vowel). In all light words, accent falls on the first syllable if the non-full vowel is unrounded, but on the first or second syllable if the first vowel is unrounded (cf. Kenstowicz 1994). Ivanov and Tuzharov add that this L/F pattern is consistently present in words, uttered in isolation. Words that are embedded in larger utterances show a preference for penultimate accent. Gordon (2000) suggests that the default initial accent is an intonational e¤ect, assuming that the fundamental frequency at the beginning of an utterance will start high and then decline. Hayes (1995: 297) lists Western Mari (Cheremis) as an example of a LAST/LAST system with final extrametrical syllables; this analysis is adopted in StressTyp. Examples like tUvU´rgasˇ ‘to curdle’, in which accent falls on the last syllable if one disregards the extrametrical final syllable. However, examples like mUrU´ktUlasˇ ‘to thunder’ show that it is the second not the last syllable that is accented. Moreover, examples with the unrounded reduced vowel /i/ show that accent may fall on either the initial or the second syllable: tsItI´ra¨sˇ P tsI´tIra¨sˇ ‘to shake-REFL’. Hence, it would seem possible to classify Northwestern Mari as another example of a FIRST/FIRST system, albeit with some complications. With three competing analyses (F/L, L/L and F/F) it would seem that the proper analysis of the Western dialect is not available.

492

Harry van der Hulst

Above, I quoted Kangasmaa-Minn (1998: 224) who mentions final accent for the eastern dialect. This could be due to influence of Tatar (Itkonen 1955). Eastern Mari has also been described as a L/F pattern in Sebeok and Ingemann (1961) (who say that this pattern varies freely with a weight-insensitive final accent). But Ristinen (1960) finds that the default initial accent is not present in medial contexts. Gordon (2000) o¤ers a discussion of Northwest Mari and Eastern Mari, both described as L/F systems. Finno-Ugric: Finno-Permic: Finnic: Finno-Mordvinic: Finno-Lappic Ruija (dialect of Northern Saami) [I] Primary stress is initial. f Secondary stress on odd non-final syllables. f A weak final secondary stress may occur in trisyllables. f

¨ba#tta˙raddat ¨gapperist or ¨gappe˙rist

‘to fly continually’ ‘in the cup’

Finnish [I] f Primary stress falls on the first syllable. f Secondary stress usually on alternates after the primary stress, but never final. ¨lope˙teta ¨teuras˙tamo ¨opet˙tele˙manani

‘finish[neg.]’ ‘slaughterhouse’ ‘as something I have been learning’

Estonian [I] Primary stress falls on the first syllable, exceptions are rare. f Secondary stress usually on alternates after the primary stress. f If the secondary stress would fall on a short vowel, the pattern might become ternary. f

¨sada

‘hundred’

¨suletud ‘closed’

¨kahekesi ‘together, two at a time’ Karelian [I] f Primary stress falls on the first syllable. f Secondary stress usually on alternates after the primary stress, but never final. f In some cases the secondary stress pattern is reported to be ternary.

Word accent systems in the languages of Europe

493

Liv; Livonian [I] f Primary stress falls on the first syllable. f Secondary stress occurs on the third syllable, tertiary stress on the fifth (if present). Vod; Votic [I] f Primary stress is initial. f Secondary stress on odd syllables after the main stress, but not on case su‰xes. ¨suvaø˙miøn

‘loving’

¨tu_tøa_˙riøkøo

‘girl’

Additional information All languages in the Balto-Finnic group have fixed initial accent. When it comes to secondary accent, however, there are a number of di¤erences. In Estonian secondary accents in words that do not contain an ‘overlong syllable’ ‘‘are assigned iteratively from left to right at intervals of two or three syllables’’ (Hayes, 1995: 317). Whether these intervals consist of two or three syllables depends on the weight of the second syllable. If this syllable is heavy (i.e. is closed and/or contains a long vowel) it is accented: pa´rima`tteltt (*pa´rimatte`ltt), if it is light (i.e. CV) secondary accent may optionally go to the third syllable, causing a ternary rhythm: te´rava`maltt P te´ravama`ltt. However, final syllables can only bear secondary accent if they contain a long vowel or are closed by at least two consonants. Thus, CV and CVC syllables never bear accent in word-final position: *o´savama`. The fact that CVC syllables behave like CV syllables in final position, whereas they pattern with the heavy syllables word-medially points to final consonant extrametricality in Estonian. The Estonian accent system is considerably complicated by the phenomenon of overlength. Overlength is analysed by Harms (1962: 11–12) as a type of ‘‘postponed’’ accent, which occurs only on long syllables and is accompanied by lengthening of the already long syllable. Prince (1980), based on Hint (1973) focuses on the nature of overlength in Estonian, presenting a metrical analysis, as does Hayes (1995). A detailed description is beyond the scope of this chapter. The accent system of Finnish resembles that of Estonian, but there are some di¤erences. Firstly, Finnish has no overlength, which makes the Finnish system less complex than the Estonian system. Secondly, ternary rhythm in Finnish only occurs when the second syllable following an accented syllable is light and is followed by a heavy syllable, in which case it is not optional but obligatory: ra´kastune`ita (*ra´kastu`neita) ‘infa-

494

Harry van der Hulst

tuated lovers’. When the second syllable following an accented syllable is followed by a light syllable there is always binary rhythm, irrespective of whether or not the second syllable is heavy: o´piske`lija (*o´piskelı`ja) ‘student’. Like Estonian, Finnish does not allow secondary accent on final CV or CVC syllables, which indicates that Finnish, too, has final consonant extrametricality. Of the other Balto-Finnic languages, Karelian apparently has a system which is in between Finnish and Estonian. Like Finnish, it does not have overlength but a ternary rhythm appears to be optional, like in Estonian. The (nearly) extinct language Votian appears to have secondary accents on odd-numbered syllables. Also, final syllables can take secondary accent irrespective of whether they are heavy or light, except for case su‰xes, which are never accented. Livonian is reported to have primary accent on the first, secondary accent on the second, and tertiary accent on the fifth syllable. Final odd-numbered syllables carry accent, independent of their weight. Central Norwegian Lappish (Northern Saami) has initial primary accent. Secondary accent occurs on odd-numbered syllables. Although a secondary accent has been reported on the third and final syllable of trisyllabic words, it does not seem to occur on the final syllable of five-syllable words. Sammallahti (1998: 52–53) says that words are divided in stress group consisting of a stressed syllable followed by one or two (rarely three) unstressed syllables. Discussing the Finnic language group in general, Viitso (1998b: 104– 105) mentions the fact that the first secondary stress, while regularly being on the third syllable since rhythm alternates in a trochaic fashion away from the initial primary accent, sometimes ends up on the fourth syllable, for example when the third vowel is short and the fourth is long. We saw this pattern described above. It is also mentioned that some su‰xes attract the secondary stress, but the examples given seem to involve a long vowel in the fourth syllable (or is it lengthened due to stress?). Also the first secondary stress can end up on the second syllable due to deletion of a second syllable, although in that case the first syllable ends up as VVC (‘superheavy’) which may imply that a secondary stress on the second syllable is rhythmically wellformed. Epenthesis of a vowel may also cause secondary stress to be on the fourth syllable, but again, in the example given, the fourth vowel is long, the third being short. Also see Viitso (1998a: 119) on Estonian accent.

Word accent systems in the languages of Europe

495

Generalizations It seems clear at least that initial accent is the unmarked option in the Finno-Ugric languages. Due to lack of data on the Samoyedic accent systems it is not clear if initial accent plays an equally important role there. Primary accent in the Finno-Ugric languages is either weight-insensitive or sensitive only to vowel quality (except in some Permic dialects). Secondary accent, which occurs only in the languages with weight-insensitive primary accent, may be sensitive to vowel length and syllable closure. According to Sammallahti (1987) primary stress was on the first syllable in Proto-Uralic as in most present-day Uralic languages. Secondary stress fell on non-final odd-numbered syllables. Through vowel reduction and an accompanying rejection of stress by these reduced vowels, the unbounded systems of the Volga region (Mordvin, Cheremis) came into being. Bereczki (1987) suggests that the LAST/FIRST system of Cheremis originated through the influence of the Turkic language Chuvash. Abondolo (1998: 9) makes the following general statement: word stress is initial in most Uralic languages. Exceptions to this general rule are found in a belt of west-central languages (Moksha Mordva, Mari and all of Permian except Komi) and in Samoyedic. In West Mari and Nganasan the location is penutlimate, in Udmurt it is final. In Permiak is location is dependent on morphology, while in Moksha Mordva and Tundra Nenets it refers to vowel sonority or vowel fullness. In Selkup both phonological and morphological factors play a role. 4.6. Etruscan (Isolate) Genetic information: Isolate Accent information Etruscan is attested from 700 BC to 50 AD through some 9000 to 10,000 inscriptions. Accent was on the initial syllable in native words, except that demonstrative pronouns (because of their use as enclitics) had final accent. Loan words had accent on the last long vowel; since Etruscan did not have vowel length distinctions, the long vowels were interpreted as being accented (Rix 2004). Gabor Z. Bodroghy (website URL http://users.tpg. com.au/etr/etrusk/tex/develop.html) reports that the loss of vowels in non-initial syllables was due to an ‘intensive stress accent which around 5000 BCE a¤ected Etruscan as well as other languages of Italy’.

496

Harry van der Hulst

5. Generalizations and conclusions Van der Hulst, Hendriks and van de Weijer (1999) which also o¤ers an overview of word accentual systems in the languages of Europe concludes with tables which displays the variety of accentual types and with maps which show the geographical distribution of these types. In this chapter I will not repeat that information. The preceding more extensive overview confirms the general fact that the European area contains all conceivable types of word accent, ranging from bounded to unbounded systems, from weight-insensitive to weight-insensitive systems and from stress-accent to pitch-accent systems. Although languages can be grouped in such broad types, most languages display specific properties which make them, in a sense, unique. Needless to say that much more information and analysis is required for most if not all languages reviewed here, and many others which we have not discussed at all. Deeper analysis must take into account the intricate relationships that exist between accent placement and morphological structure. The characterizations given here are broad, and possibly wrong, but it is hoped that this survey will nonetheless serve the purpose of sparking further research in this area.

References Abondolo, Daniel 1998 Finnish. In: Daniel Abondolo 1998 (ed.), 149–183. Abondolo, Daniel 1998 Hungarian. In: Daniel Abondolo 1998 (ed.), 428–456. Abondolo, Daniel 1998 Khanty. In: Daniel Abondolo 1998 (ed.), 358–386. Abondolo, Daniel (ed.) 1998 The Uralic languages. London: Routledge. Aitken, A.J. and Tom McArthur (eds.) 1979 Languages of Scotland. Chambers: Edinburgh. Ajello, Roberto 1998 Armenian. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 197–227. Anagnostopoulos, George 1926 Tsakonische Grammatik. Berlin: Urania-buchhandlung. Anderson, Henning 1998 Slavic. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 415–453.

Word accent systems in the languages of Europe

497

Anderson, Gregory 1997 Lak Phonology. In: Alan Kaye 1997 (ed.), 973–997. ´ rnason, Kristja´n A 1999 [Word stress in] Icelandic and Faroese. In: Harry van der Hulst 1999 (ed.), 567–604. Aronson, H.I. 1982 Georgian, A reading grammar. Columbus, Ohio: Slavica. Aronson, Howard I. 1997 Georgian phonology. In: Alan S. Kaye 1997 (ed.), 929–940. Balassa, Jo´zsef 1890 ‘‘Hangsu´ly a magyar nyelvben’’, Nyelvtudoma´nyi Ko¨zleme´nyek 21: 401–431. Baldi, Philip 1983 An introduction to the Indo-European languages. Carbondale and Edwardsville: Southern Illinois University Press. Ball, Martin J. and James Fife (eds.) 1993 The Celtic languages. London: Routledge. Beekes, Robert S. P. 1995 Comparative Indo-European linguistics. An introduction. Amsterdam/Philadelphia: John Benjamins Publishing Company. Bereczki, Ga´bor 1987 Geschichte der Wolgafinnischen Sprachen. In: D. Sinor 1987 (ed.), 314–350. Bernard Comrie, Matthews, Stephen and Maria Polinsky (eds.) 2003 The atlas of languages: The origins and development of languages throughout the world. 2nd edition. New York: Facts and File Inc. Bevington, G.R. 1974 Albanian Phonology. Wiesbaden: Harrassowitz. Blevins, Julliette 1993 A tonal analysis of Lithuanian nominal accent. Language 69/2: 237–273. Bodroghy, Gabor Z. (http://users.tpg.com.au/etr/etrusk/tex/develop.html): The Etruscan Liber Linteus. Brixhe, Claude 2004 Phrygian. In: Roger D. Woodard 2004 (ed.), 777–787. Browne, Wayles 1993 Serbo-Croat. In: Bernard Comrie and Greville G. Corbett 1993 (eds.), 306–387. Bruce, Go¨sta 1999 [Word stress in] Swedish. In: Harry van der Hulst 1999 (ed.), 554–566. Bruce, Go¨sta 1999 [Word tone in] Scandinavian languages. In: Harry van der Hulst 1999 (ed.), 605–632.

498

Harry van der Hulst

Clackson, James P.T. 2004 Classical Armenian. In: Roger D. Woodard 2004 (ed.), 922–942. Comrie, Bernard 1981 Languages of the Soviet Union. Cambridge: Cambridge University Press. Comrie, Bernard and Greville G. Corbett (eds.) 1993 The Slavonic languages. London: Routledge. Comrie, Bernard et al. 2003 The atlas of languages. Second Edition. London: Quarto Publishing Inc. Csu´cs, Sa´ndor 1998 Udmurt. In: Daniel Abondolo 1998 (ed.), 276–304. De´csy, G. 1966 Yurak Chrestomathy, Uralic and Altaic series 50. Bloomington: Indiana University. De¤ner, Michael 1881 Zakonische Grammatik. Berlin, Weidmannsche Buchhandlung. Dell, Francois 1984 L’accentuation des phrases en Francais. In: F. Dell, D. Hirst & J.R. Vergnaud (eds.), Forme sonore du langage, 65–122. Paris: Hermann. Demiraj, Shaban 1998 Albanian. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 480–501. Dogil, Grzegorz and Briony Williams 1999 The phonetic manifestation of word stress. In H. van der Hulst 1991 (ed.), 273–334. Dogil, Grzegorz 1999a West Slavic. In: Harry van der Hulst 1999 (ed.), 813–838. Dogil, Grzegorz 1999b Baltic languages. In: Harry van der Hulst 1999 (ed.), 877–896. Doherty, C. 1991 Munster Irish Stress. In Phonology at Santa Cruz, vol. 2: 19–32. Drachman, Gabriel and Anegeliki Malikouti-Drachman 1999 Greek word accent. In: Harry van der Hulst 1999 (ed.), 897–946. Dybo, V.A. 1981 Slavjanskaja akcentelogija. Moscow: Nauka. Eichner, Heiner 1986a Die Akzentuation des Lydischen. Sprache 32: 7–21. Eichner, Heiner 1986b Neue Wege im Lydischen I: Vokalnasalita¨t vor Nasalkonsonanten. Zeitschrift fu¨r Vergleichende Sprachforschung 99: 203– 19. Eska, Joseph F. 2004 Continental Celtic. In: Roger D. Woodard 2004 (ed.), 857–880.

Word accent systems in the languages of Europe

499

Faarlund, Jan Terje 1994 Old and Middle Scandinivian. In: Ekkehard Ko¨nig and Johan van de Auwera 1994 (eds.), 38–71. Fe´ry, Caroline 1998 German word stress in OT. Journal of Comparative Germanic Linguistics 2: 101–142. Garde, Paul 1976 Histoire de l’ accentuation slave. Paris. Goedemans, Rob and Harry van der Hulst 2009 StressTyp: A database for word accentual patterns in the world’s languages. In: Martin Everaert, Simon Musgrave and Alexis Dimitriadis (eds.). The use of databases in cross-linguistic studies. New York, Berlin: Mouton de Gruyter, 235–282. Gordon, Matthew 2000 Re-examining default-to-opposite stress. Berlekey Linguistic Society 26: 101–112. Gordon, Raymond (ed.) 2005 Ethnologue: Languages of the world, 15th edition. Dallas, Texas: SIL international. Green, A.D. 1997 The prosodic structure of Irish, Scots Gaelic and Manx. PhD. Dissertation, Cornell University. Gussenhoven, Carlos 2004 The Phonology of Tone and Intonation. Cambridge: Cambridge University. Gussmann, Edmund 1994 Putting Your Best Foot Forward, Stress in Munster Irish. Ms., Catholic University of Lublin. Gvozdanovic, Jadranka 1999 South Slavic. In: Harry van der Hulst 1999 (ed.), 839–851. Hall, R.A. jr. 1938 An analytical grammar of the Hungarian language, Language, monograph 18. Baltimore: Linguistic Society of America. Halle, Morris and Jean-Rogier Vergnaud 1987 An Essay on Stress. MIT press. Hammond, Michael 1987 Hungarian Cola. Phonology Yearbook 4: 267–269. Harms, Robert 1962 Estonian Grammar. Bloomington: Indiana University. Harris, Martin and Nigel Vincent (eds.) 1997 The Romance languages. London: Routledge. Hausenberg, Anu-Reet 1998 Komi. In: Daniel Abondolo 1998 (ed.), 305–326. Hayes, Bruce 1995 Metrical Stress Theory. Chicago: The University of Chicago Press.

500

Harry van der Hulst

Helimski, Eugene 1998a Selkup. In: Daniel Abondolo 1998 (ed.), 548–579. Helimski, Eugene 1998b Nganasan. In: Daniel Abondolo 1998 (ed.), 480–515. Hermans, Ben 1999 A description of tonal accent in a Limburgian dialect. In: Harry van der Hulst 1999 (ed.), 633–658. Hewitt, B.G. 1979 Abkhaz, Lingua Descriptive Studies 2. Amsterdam: NorthHolland Publishing Company. Hint, M. 1973 Eesti Keele So˜nafonoloogia I. Tallinn, Estonia: Eesti NSV Teaduste Akademia. Honti, La´szlo´ 1998 Ob-Ugrian. In: Daniel Abondolo 1998 (ed.), 327–357. Horrocks, Geo¤rey 1997 Greek: A history of the language and its speakers. London & New York: Longmans. Hualde, Jose´ Ignacio 1993 On the historical origin of Basque accentuation. Diachronica 10: 13–50. Hualde, Jose´ Ignacio 1997 A gap filled: Postpostinitial accent in Azkoitia Basque. Linguistics 36: 99–117. Hualde, Jose´ Ignacio 1999 Basque accentuation. In: Harry van der Hulst 1999 (ed.), 947– 994. Hualde, Jose´ Ignacio 2003 From phrase-final to post-initial accent. In: Fikkert & Jackobs (eds.) 2003, Development in prosodic systems. Berlin: Mouton de Gruyter, 249–282. Hualde, Jose´ Ignacio 2007 Historical convergence and divergence in Basque accentuation. In: T. Riad & C. Gussenhoven 2007 (eds.) Tones and tunes, 291–322. Berlin: Mouton de Gruyter. Hualde, Jose´ Ignacio, Gorka Elordieta, In˜aki Gaminde & Rajka Smiljanic´ 2002 From pitch-accent to stress-accent in Basque. In: Gussenhoven, Carlos & Natasha Warner (eds.), Laboratory Phonology 7, 547– 584. Berlin and New York: Mouton de Gruyter. Hualde, Jose´ Ignacio, O. Lujanbio & F. Torreira To appear Lexical tone and stress in Goizueta Basque. Journal of the International Phonetic Association. Hulst, Harry van der 1984a Syllable structure and stress in Dutch. Dordrecht: Foris.

Word accent systems in the languages of Europe

501

Hulst, Harry van der 1984b Ambisyllabiciteit en de struktuur van Nederlandse lettergrepen. Spektator 14/5: 319–326. Hulst, Harry van der 1999 Word accent. In: Harry van der Hulst 1999 (ed.), 3–116. Hulst, Harry van der (ed.) 1999 Word prosodic systems in the languages of Europe. Berlin & New York: Mouton de Gruyter. Hulst, Harry van der 2002 Stress and accent. In: Nadel, L. (ed. Encyclopedia of Cognitive Science. Vol. 4, 4, 246–254. London: Nature Publishing Group. Hulst, Harry van der 2006 Word stress. In: Keith Brown (ed.) The encyclopedia of language and linguistics. 2nd edition. Volume 13, 655–665. Oxford: Elsevier. Hulst, Harry van der 2009 Brackets and grid marks or theories of primary accent and rhythm. In: E. Raimy and C. Cairns (eds.) Contemporary views on architecture and representations in phonological theory, 225– 245. MIT press. Hulst, Harry van der, Bernadet Hendriks and Jeroen van de Weijer 1999 A survey of word prosodic systems of European languages. In: Harry van der Hulst 1999 (ed.), 425–476. Idsardi, William J. 1992 The Computation of Prosody, MIT Ph.D. dissertation, distributed by MIT Working Papers in Linguistics. Inkelas, Sharon & Draga Zec 1988 Serbo-Croatian Pitch-Accent: The Interaction of Tone, Stress and Intonation. Language 64: 227–248. Itkonen, E. ¨ ber die Betonungsverha¨ltnisse in der finnisch-ugrischen Spra1955 U chen. Acta Linguistica Hungaricae 5: 21–34. Ivanov, I.G. and G. M. Tuzarov 1970 Severo-zapadnoje narecije jazyka. Josˇkar-ola. Jasano¤, Jay H. 2004 Gothic. In: Roger D. Woodard 2004 (ed.), 881–906. Jessen, Michael 1999 [Word stress in] German. In: Harry van der Hulst 1999 (ed.), 515–544. Kager, Rene´ 1989 A metrical theory of stress and destressing in English and Dutch. Dordrecht: Foris Publications. Kangasmaa-Minn, Eeva 1998 Mari. In: Daniel Abondolo 1998 (ed.), 219–248. Kavitskaya, Darya 2006 Tundra Nenets accent. LSA annual meeting. Handout.

502

Harry van der Hulst

Kaye, Alan S. (ed.) 1997 Phonologies of Asia and Africa. Volume 1 and 2. Winona Lake: Eisenbrauns. Kenstowicz, Michael 1994 Sonority driven stress, Ms., MIT. Kiparsky, Paul and Morris Halle 1977 Toward a reconstruction of the IE accent. Studies in Stress and Accent. Southern California Occasional Papers in Linguistics 4: 209–238. Kodzasov, Sandro 1999 Caucasian: Daghestanian languages. In: Harry van der Hulst 1999 (ed.), 995–1020. Kodzasov, Sandro 1999 Russian. In: Harry van der Hulst 1999 (ed.), 852–876. Ko¨nig, Ekkehard and Johan van der Auwera (eds.) 1994 The Germanic languages. London: Routledge. Kortland, Frederick 1978 A history of Slavic accentuation. Lingua 44: 67–91. Kortlandt, Frederik 1979 On the history of Slavic accentuation. Zeitschrift fu¨r vergleichende Sprachforschung 92: 269–281. Kristo¤ersen, Gjert 2000 The phonology of Norwegian. Oxford: Oxford University Press. Kuryl owicz, J. 1952 L’accentuation des languages indo-europe´ens. Krako´w: Polska Akademia Nauk. Kuznetsova A. I., Kazakevitch O. A., Io¤e L. Yu., Helimski E. A. (eds.) 1993 Studies in the Selkup language. Taz Dialect. Vol. 2. Texts. Dictionary. Moscow, 1993 ( . Lahiri, Aditi, Tomas Riad and Haike Jacobs 1999 Diachronic prosody. In: Harry van der Hulst 1999 (ed.), 335– 421. Lockwood, W.B. 1972 A panorama of Indo-European languages. London: Hutchinson University Library. Lockwood, W. B. 1975 Languages of the British Isles: Past and present. London: Deutsch. Lotz, J. 1939 Das ungarische Sprachsystem. Stockholm. MacAulay, Donald (ed.) 1992 The Celtic Languages. Cambridge: Cambridge University Press.

Word accent systems in the languages of Europe

503

Matthews, W. K. 1951 Languages of the USSR. Cambridge: Cambridge University Press. Melchert, H. Craig 1994 Anatolian Historical Phonology. Amsterdam/Atlanta: Rodopi. Melchert, H. Craig 1997 Hittite phonology. In: Alan S. Kaye 1997 (ed.), 555–568. Melchert, H. Graig 2004a Lycian. In: Roger D. Woodard 2004 (ed.), 591–600. Melchert, H. Graig 2004b Lydian. In: Roger D. Woodard 2004 (ed.), 601–608. Melchert, H. Graig 2004c Palaic. In: Roger D. Woodard 2004 (ed.), 585–590. Melchert, H. Graig 2004d Carian. In: Roger D. Woodard 2004 (ed.), 609–613. Melchert, H. Graig 2004e Luvian. In: Roger D. Woodard 2004 (ed.), 576–584. Michelena, Luis 1958 A propos de l’accent Basque. Bulletin de la Socie´te´r de Linguistique 53, 304–333.

Michelena, Luis Fone´tica histo´rica vasca. 2nd revised ed. San Sebastian: Disputacio´n de Guipu´zcoa. Nakhlev, Luay, Don Ringe and Tandy Warnow 2005 Perfect phylogenetic networks: A new methodology for reconstructing the evolutionary history of natural languages. Language 81/2: 382–420. Newmark, L., P. Hubbard and P. Prifti 1982 Standard Albanian: A Reference Grammar for Students. Stanford: Stanford University Press. Nichols, Johanna 1997 Chechen Phonology Lak Phonology. In Alan Kaye 1997 (ed.), 941–972. Nikolajev, S.L. & S.A. Starostin 1994 A North Caucasian etimological dictionary. Moscow. ` Dochartaigh, Cathair O 1992 The Irish language. In: Donald MacAulay 1992 (ed.), 11–99. Oostendorp, Marc van 1995 Vowel Quality and Phonological Projection. [Unpublished doctoral dissertation, University of Tilburg.] Philippaki-Warburton, I. 1976 On the boundaries of morphology and phonology: A case study from Modern Greek. Journal of Linguistics 12: 259–278. Price, Glanville 1998 Encyclopedia of the languages of Europe. Oxford: Blackwell. 1977

504

Harry van der Hulst

Price, Glanville 2000 Languages of Britain and Ireland. Oxford: Blackwell. Prince, Alan 1980 A Metrical Theory for Estonian Quantity. Linguistic Inquiry 11: 511–562. Pulgram, Ernst 1970 Syllable, Word, Nexus, Cursus, Series Minor 81. The Hague, Paris: Mouton. Ramat, Anna Giagalome and Paolo Ramat (eds.) 1998 The Indo-European languages. London: Routledge. Revithiadou, Anthi 1999 Headmost accent wins. Head dominance and ideal prosodic form in lexical accent systems. LOT Dissertation Series 15 (HIL/ Leiden Universiteit The Hague: Holland Academic Graphics. Riad, Tomas 1998 Toward a Scandinavian accent typology. In: Wolfgang Kehrein and Richard Wiese 1998 (eds. Phonology and morphology of the Germanic languages, 77–112. Tu¨bingen: Max Niemeyer Verlag. Rice, Curt 1999 [Word stress in] Norwegian. In: Harry van der Hulst 1999 (ed.), 545–553. Riese Timothy 1998 Permian. In: Daniel Abondolo 1998 (ed.), 249–277. Rischel, Jørgen 1972 Compound Stress in Danish without a Cycle. ARIPUC 6: 211– 228. Rischel, Jørgen 1982 On Unit Accentuation in Danish – and the Distinction between Deep and Surface Phonology. ARIPUC 16: 191–239. Ristinen, E. K. 1960 An East Cheremis Phonology. Bloomington; Indiana Univerisity Press. Rix, Helmut 2004 Etruscan. In: Roger D. Woodard 2004 (ed.), 943–966. Roca, Iggy 1999 Stress in Romance languages. In: Harry van der Hulst 1999 (ed.), 659–812. Rowicka, Grazyna 1994 2 þ 2 ¼ 3 Stress in Munster Irish. HIL Manuscripts 3/1: 12–31. Ruhlen, Merrit 1987 A guide to the world’s languages Vol. 1: Classification. Stanford: Stanford University Press. Ruhlen, Merritt 1991 A guide to the world’s languages. Volume 1: Classification. Stanford, CA: Stanford University Press.

Word accent systems in the languages of Europe

505

Salminen, Tapani 1998 Nenets. In: Daniel Abondolo 1998 (ed.), 516–547. Sammallahti, Pekka 1987 Historical Phonology of the Uralic Languages. In: D. Sinor 1987 (ed.), 478–554. Sammallahti, Pekka 1998 Saamic. In: Daniel Abondolo 1998 (ed.), 43–95. Schrijver, Peter 1995 Studies in British Celtic Historical Phonology [Leiden Studies in Indo European 5]. Amsterdam – Atlanta: Rodopi. Sebeok, Thomas and Francis Ingemann 1961 An Eastern Cheremis manual [Uralic and Altaic Series 5]. Bloomington: Indiana University Press. Silvestri, Domenico 1998 The Italic languages. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 322–344. Sipta´r, Pe´ter and Miklo´s To¨rkenczy 2000 The phonology of Hungarian. Oxford: Oxford University Press. Smeets, R. 1984 Studies in West Circassian Phonology and Morphology. PhD. Dissertation, Leiden. So¨vijarvi, A. ¨ ber die phonetischen Hauptzu¨ge der finnischen und der ungari1956 U schen Hochsprache. Wiesbaden: Otto Harrasowitz. Stang, C. S. 1957 Slavonio accentuation. Oslo. Stankiewicz, Edward 1993 The accentual patterns of the Slavic languages. Stanford/California: Stanford University Press. Szemere´nyi, Oswald J. L. 1996 Introduction to Indo-European Linguistics. Oxford: Clarendon Press. Szinnyei, J. 1912 Ungarische Sprachlehre. Berlin. Ternes, Elmar 1992 The Breton language. In: Donald MacAulay 1992 (ed.), 371–542. Thomas, Alan R. 1992a The Cornish language. In: Donald MacAulay 1992 (ed.), 346–370. Thomas, Alan R. 1992b The Welsh language. In: Donald MacAulay 1992 (ed.), 251–345. Thomson, Robert L. 1992c The Manx language. In: Donald MacAulay 1992 (ed.), 100–136. Trommer, Jochen Ms. Albanian word stress. University of Osnabru¨ck.

506

Harry van der Hulst

Tsygankin, D.B. and C.Z. Dabaev 1975 Ocˇerk sravnitel’noj grammatiki mordovskix (moksˇanskogo I erz’anskogo) literaturnix jazykov. Saransk. Varga, L. 1994 Rhythmic Stress Alternation in Hungarian. In: The even Yearbook 1994. Budapest: Department of English Linguistics, Eo¨tvo¨s Lora´nd University, 233–254. Vaux, Bert 1994 Armenian phonology. Doctoral dissertation, Harvard University. Vaux, Bert 1998 The Phonology of Armenian. Oxford: Oxford University Press. Vennemann, Theo 1990 Syllable structure and simplex accent in Modern Standard German. In: Michael Ziolkowski, Manuela Noske and Karen Deaton 1990 (eds.) Papers from the Regional Meeting of the Chicago Linguistic Society 26, vol. II: The Parasession, 399– 412. Chicago Linguistic Society. Viitso, Tiit-Rein 1998a Estonian. In: Daniel Abondolo 1998 (ed.), 115–148. Viitso, Tiit-Rein 1998b Finnic. In: Daniel Abondolo 1998 (ed.), 96–114. Vineis, Eoardo 1998 Latin. In: Anna Giacalone Ramat & Paolo Ramat 1998 (eds.) The Indo-European languages. Routledge., 261–321. Visch, Ellis 1999 The rhythmic organization of compounds and phrases. In: Harry van der Hulst 1999 (ed.), 161–232. Voegelin, C. F. and F. M. Voegelin 1977 Classification and index of the world’s languages. Amsterdam, New York and Oxford: Elsevier. Wallace, Rex E. 2004a Sabellian languages. In: Roger D. Woodard 2004 (ed.), 812–839. Wallace, Rex E. 2004b Venetic. In: Roger D. Woodard 2004 (ed.), 840–856. Watkins, Calvert 1998 Proto-Indo-European: Comparison and reconstruction. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 25–73. Watkins, Calvert 2004 Hittite. In: Roger D. Woodard 2004 (ed.), 551–575. Wiese, Richard 2000 The phonology of German. Oxford: Oxford University Press. Williams, Briony 1999 The phonetic manifestation of word stress in Welsh. In: Harry van der Hulst 1999 (ed.), 311–334.

Word accent systems in the languages of Europe

507

Winter, Werner 1998 Tocharian. In: Anna Giacalone Ramat and Paolo Ramat 1998 (eds.), 154–168. Woodard, Roger D. 2004a Attic Greek. In: Roger D. Woodard 2004 (ed.), 609–649. Woodard, Roger D. 2004b Greek dialects. In: Roger D. Woodard 2004 (ed.), 650–672. Woodard, Roger D. (ed.) 2004 The Cambridge Encyclopedia of the World’s Ancient Languages. Cambridge, UK: Cambridge University Press. Zaicz, Ga´bor 1998 Mordva. In: Daniel Abondolo 1998 (ed.), 184–218. Zec, D. 1994 Sonority constraints on prosodic structure. New York, NY: Garland, Taylor and Francis. Zonneveld, Wim and Mieke Trommelen 1999 Word stress in West-Germanic languages. In: Harry van der Hulst 1999 (ed.), 478–514.

10.

Word accent systems in the languages of Asia

Rene´ Schiering1 and Harry van der Hulst 2 1. Introduction This chapter surveys accentual systems in the languages of Asia. Our objective has been to provide information on as many languages as we could lay our hands on, given the inevitable limitations on the time for this project and on access to sources in this period. This survey, then, does not claim anything near completeness. In a way, it presents an agenda for further studies especially witnessed by the many sections where little or no information on accentual systems is presented. We are not defending these limitations by claiming that the basic descriptive work has not been done in all these cases, although for many languages this is probably true. We have no doubt, however, that a lot of useful information on word accent has been gathered and is present in the countless language descriptions that we have been unable to consult. There may even be typological surveys that focus on word accent that we have overlooked. With all these limitations and shortcomings, we hope that this chapter still o¤ers a useful inventory which will stimulate further typological and theoretical research. In section 2, we outline the contents of this chapter, motivating its organization and introducing the conventions which guide the presentation of accentual data. Section 3 explains on which materials this survey is based. Sections 4, 5, 6 and 7 present data on word accent systems in 1. While working on this chapter Rene´ Schiering received funding from the German Science Foundation (DFG) in the context of the project ‘Typology and Theories of the Word’ at the University of Leipzig and from the FritzThyssen-Stiftung. He gratefully acknowledges the help of Balthasar Bickel, Kristine A. Hildebrandt, Nikolaus Himmelmann and Werner Drossard. He is indebted to his colleagues at the Department of Linguistics at the University of Cologne who granted him access to the invaluable treasure of sources which is their library. 2. Harry van der Hulst wishes to acknowledge the help of Rob Goedemans, Elinor Keane, Joan Baart, Arthur Abrahamson, Martin Haspelmath, Eon-Suk Ko, Irina Monich, Michael Noonan, Pramod Pandey, Jonathan Bobaljik and Oksana Tarasenkova.

510

Rene´ Schiering and Harry van der Hulst

the languages of North-East Eurasia, Asia Minor and Central Eurasia, South and West Asia, and Mainland South East Asia, respectively. Finally, in section 8 we o¤er some general observations and conclusions.

2. Contents of this chapter The following table of contents summarizes the organization of the present chapter: 1.

Introduction

2.

Contents of this chapter

3.

A note on the information on which this chapter is based

4. 4.1. 4.2. 4.2.1. 4.2.2. 4.2.3. 4.2.4.

North-East Eurasia (Eastern Siberia) Chukotko-Kamchatkan Isolates Ket Yukaghir Nivkh Ainu

5. 5.1. 5.2. 5.3. 5.4. 5.5.

Asia Minor and Central Eurasia (Altaic) Turkic Mongolic Tungusic Korean Japanese

6. 6.1. 6.1.1. 6.1.2. 6.2. 6.3. 6.4. 6.5.

Southern Asia Indo-European: Indo-Iranian Indo-Aryan Iranian Dravidian Austroasiatic: Munda Andamanese Burushaski

7. 7.1. 7.1.1.

Mainland East and South East Asia Sino-Tibetan Chinese

Word accent systems in the languages of Asia

7.1.2. 7.2. 7.3. 7.4.

511

Tibeto-Burman Austroasiatic: Mon-Khmer Hmong-Mien Tai-Kadai

In order to facilitate access to the wealth of accentual data included in this chapter, we subdivided Asia into four broad areas following roughly a north to south direction, starting with North-East Eurasia (roughly Eastern Siberia), proceeding to Asia Minor and Central Eurasia (covering the languages subsumed under Macro-Altaic), followed by South and West Asia, and finally reaching Mainland East and South East Asia. The coverage of language families in Section 4 to 7 will be outlined in a brief introductory note at the beginning of each section. The organization of the subsections with accentual data adheres to the following scheme: a. Genetic structure of the (sub)family. Language classifications are taken from Ruhlen (1991), Comrie et al. (2003), R. Gordon (2005) and other more specialized sources that o¤er classifications for the language families to be discussed. The genetic information is sometimes followed by some archeological-historical and dating information, often based on the draft edition of Ruhlen (1991). We have tried to strike a compromise in cases of conflicting groupings. In each case, (sub)family names are presented in capitals, while the names of sample languages appear in italics. Bold print marks languages that are included in StressTyp (see section 3), whereas languages which are discussed in this chapter but are not included in StressTyp are underlined. b. Extracts from StressTyp entries (language name followed by StressTyp Code and examples). More complete extracts (including references) are o¤ered in Part II of this volume and, of course, in StressTyp itself. This stress information is presented unchanged, i.e. as it can be found in the database and Part II. c. Additional accent information. This might involve additional information on languages already in StressTyp, or information on languages that are not in StressTyp. (In some larger families that are treated in one section we present the StressTyp extracts and additional information per subfamily.) d. Generalizations. We have tried to make general statements about the accentual patterns in the relevant (sub)families, in some cases accompanied by remarks about diachronic developments or the data’s impact on current theorizing.

512

Rene´ Schiering and Harry van der Hulst

If (b) is absent, (c) is labeled ‘accent information’. Sometimes (c) and (d) are conflated into a single section.

3. A note on the information on which this chapter is based The first source for the accentual data presented here is StressTyp (see chapter 1, 2, 11 and Part II of this book). In addition, for European languages and languages bordering Europe, i.e. Altaic, we rely on the collection of articles in van der Hulst (1999). Thirdly, we consulted books that o¤er surveys of language families or languages in a specific geographical area (e.g. Johanson and Csato´ (eds.) 1998, Thurgood and LaPolla (eds.) 2003, Comrie 1981, Kaye (ed.) 1997a, b). Fourthly, we have consulted grammars of individual languages as well as research articles discussing accentual data for a theoretically-oriented readership. Fifthly, we have sent email queries to colleagues; where we rely on information that they directly have given to us (via email or in personal communication) we note this in the text. In line with the second author’s previous work on word accentual systems, we prefer to use the term ‘word accent’ where many others would use the term ‘word stress’. We refer to van der Hulst (1999, 2002, 2006) and chapter 1 of this volume for a justification and clarification of this terminological choice.

4. North-East Eurasia (Eastern Siberia) This part of Eurasia is roughly coextensive with Eastern Siberia, stretching approximately from the river Ob’ in the West to the Bering Sea in the East and from the Northern borders of Mongolia and China in the South to the Laptev Sea, the East Siberian Sea and the Chukchi Sea in the North (Comrie 2007: 244–245). In this area, languages of five language families and several language isolates are spoken: f f f f f f

Indo-European (Russian, see Chapter 8, section 4.1.4.) Altaic (Turkic, Mongolian and Tungusic, see Section 5 of this chapter) Uralic (Samoyedic and Finno-Ugric, see Chapter 8, section 4.5.) Chukotko-Kamchatkan Eskimo-Aleut (see Chapter 5, section 3.1.) Isolates: Ket, Yukaghir, Nivkh, Ainu

Word accent systems in the languages of Asia

513

With most of these language families discussed in other chapters of this volume, the present survey focuses on the Chukotko-Kamchatkan family, located in the Far East of the area delimited above and on the language isolates spread over Eastern Siberia. 4.1. Chukotko-Kamchatkan Genetic information The Chukotko-Kamchatkan family, located in Eastern Asia Russia, is further subdivided into a Northern and a Southern branch, with the former encompassing the two sub-branches Chukot and Koryak-Alyutor. CHUKOTKO-KAMCHATKAN NORTHERN: CHUKOTIAN: Chukchi (Chukot) KORYAK-ALYUTOR: Alutor, Koryak, Kerek SOUTHERN: Itelmen StressTyp extracts Chukchee; Chukot [I;S] f f

Accent usually occurs at the beginning of the word. To give emphasis to the word, the accent may be placed on the last syllable, the vowel of which then changes to /o/.

¨pari´n˜in

‘shoulder blade’

pi¨n˜epi

‘snowstorm’

u¨wema

‘while cooking’

¨pirirkin

‘he takes’

Additional information According to Dunn (1999: 54), primary stress in the Telqep variety of Chukchi occurs on the first syllable of the word with a consonant onset and a full vowel. Secondary stress is placed on every second syllable before and after that. (1) a. b. c. d. e.

/¨nu.tec.˙q«.c«.˙ku.kin/ /q«.¨jet.˜?i/ /˙k«r.˜«.¨Re.c?«.˙kin/ /a.¨tok.toR.˙ka/ /a.¨mo.´e.˙qaj/

‘smth. from the surface of the ground’ ‘come!’ ‘smth. made of dry stumps’ ‘without a doctor’ ‘bark (DIM)’

514

Rene´ Schiering and Harry van der Hulst

For this variety, the emphatic vocative prosody is exemplified in (2), where the final syllable of the vocative form of T«lel ?«n appears with a lengthened /o/. (2) «nk?am

n-in-iw-i˜«m

T«lel?-oøø-n

and HAB-TR-say-1sg personal.name-E.VOC-3sgABS ? ‘And I said to him: ‘T«lel «n!’ ’ In what follows, we add some more detailed descriptions that are based on summaries and partial translations of Russian sources provided to us by Oksana Tarasenkova. We have edited and abbreviated this information, but in view of the fact that little is known about these languages we have tried to preserve as much information as possible. In particular, the data presented below exemplify in detail how di¤erent morphological structures relate to the stress domain in Chukchi. Skorik (1961) discusses word accent in ‘Standard Chukchi’ in some detail. The following is based on his account. Accent never occurs outside the domain of the word stem. Given that accent is ‘stem bound’, if a one-syllable stem is combined with a su‰x, the accent is always on the stem vowel: (3) a. c. e. g.

pojg-a ‘spear-instr.’ w«kw-a ‘stone-instr.’ puwt-e ret-«k

‘can-instr.’ ‘bring’

b. kejN-e d. mirg-e

‘brown bear-instr.’ ‘grandfather-instr.’

f. iw-«k h. tejk-«k

‘say’ ‘make’

The same pattern is attested for words in which the su‰x is a partial reduplicant of the stem in (4): (4) a. c. e. g.

tanN-«-tan tirk-«-tir tumg-«-tum t«lg-«-t«l

‘stranger’ ‘the sun’ ‘comrade’ ‘thaw’

b. d. f. h.

korg-«-kor tilm-«-til tut?-«-tut t«mg-«-t«m

‘joy’ ‘sea eagle’ ‘haze’ ‘silence’

(The linking vowel -«- creates an extra open syllable preceding the consonant initial reduplicative su‰x.) A polysyllabic stem that is combined with a su‰x always has accent on its last syllable, as shown in (5). (5) a. c. e. g.

ekwet-«k wiriN-«k

‘leave’ ‘defend’

b. d.

inejpel?-«n k«tg«ntat-«k

‘docker’ ‘to run’

f. h.

pelat-«k ‘stay’ reqoka-lg«n ‘polar fox’ migcˇiret-«k ‘to work’ g«nr«ret-«k ‘to guard’

Word accent systems in the languages of Asia

515

The accent location does not change when the number of su‰xes is increased. The accent, remaining on the last syllable of the stem, thus appears further from the end of the word in (6b–c). (6) a. b. c.

winret-«k winret-«rk«n

‘to help’ ‘help-3SG’

winret-«rk«nit«k

‘help-2PL’

Apparently exceptional are the cases where the stem ends in a vowel. Unlike the above observation, most of such stems are accented on the preceding syllable which is the first stem syllable in the examples in (7). (7) a. c.

wane-wan ‘no’ cˇeri-cˇer ‘dirt’

b. d.

weni-wen ‘bell’ keli-kel ‘paper, book’

Such a change in the location of the accent can be explained in two ways. The last vowel of the stem can be considered a linking vowel, similar to the linking vowel « in the earlier examples in (4) (tumg-«-tum ‘comrade’, etc.). Another account appeals to a ranking of vowels according to their height: /i/ P /e/ P /a/. We could then say that accent falls on the lowest, most sonorous vowel. This alternative is supported by the data in (8). (8) a. c.

k«lka-k«l piNe-piN

‘shell’ ‘snowfall’

b. nute-nut d. jil?e-jil

‘land, country’ ‘arctic ground squirrel’

The following examples suggest that in case of equal height, accent falls on the second vowel. (9) a.

jara-N«

‘house’

b. welo-lg«n

‘ear’

The accentual pattern is di¤erent when the word has no a‰x, or has a non-syllabic su‰x. If the stem is combined with a su‰x that is composed of a consonant only (10b, d, f ), the accent is located on the penultimate or, as in (10d) even antepenultimate syllable of the stem. (10) a. titi-N« ‘needle-SG’ c. melota-lg«n ‘hare-SG’ e. qora-Ne ‘deer-SG’

b. d. f.

titi-t ‘needle-PL’ milute-t ‘hare-PL’ qora-t ‘deer-PL’

If the word has no overt su‰x (i.e. a zero su‰x), accent also appears on the penultimate syllable of the stem in (11b, d, f, h).

516

Rene´ Schiering and Harry van der Hulst

(11) a.

ri§it-ti

‘belt-PL’

b. ri§it

‘belt-SG’

c. e.

warat-te jatjol-te

‘people-PL’ ‘fox-PL’

d. but warat ‘people-SG’ f. jatjol ‘fox-SG’

g.

jejwel-ti

‘orphan-PL’

h. jejwel

‘orphan-SG’

These examples suggest that primary accent cannot be word final. This would, however, require stem-penultimate rather than antepenultimate accent in the example milute-t ‘hare-PL’; another generalization, consistent with the facts in 10b, d, f, would be that final accent is avoided by placing the accent on the stem-initial syllable. The accent shift to the penultimate (or first) syllable in the word forms without su‰xes causes vowel reduction in the open last syllable, as in (12b, d, f, h): (12) a. wala-jp« c. aNqa-jp« e. g.

r«rka-jp« omqa-jp«

‘from knife’ ‘from sea’

b. wal« d. aNq«

‘knife’ ‘sea’

‘from sea lion’ ‘from polar bear’

f. r«rk« h. umq«

‘sea lion’ ‘polar bear’

In words with root repetition but no su‰xes the penultimate (or first) syllable of the stem is accented, cf. (13). This is also consistent with reluctance to have final accent. (13) a. kawkaw ‘bread crust, biscuit’ c. melmel ‘good weather’ e. n«mn«m ‘settlement’

b. p?oNp?oN ‘mushroom’ d. tintin ‘ice’

Prefixation, as illustrated in (14) usually does not influence the location of the accent. (14) a. b. c. d. e. f. g. h. i. j.

kojN«n ga-kojN«ma jarar ga-jararma kulil?et«k ge-qulil?et-lin ge-requlil?etlin cˇaat«k na-§aanm«k nara-saanmfik

‘cup’ ‘with cup’ ‘drum’ ‘with drum’ ‘to yell’ ‘yell-3SG-past’ ‘trying to yell-3SG’ ‘catch with a lasso’ ‘catch with a lasso-3SG-past-SUBJ/1PL-OBJ.’ ‘catch with a lasso-3SG-future-SUBJ/1PL-OBJ.’

Word accent systems in the languages of Asia

517

So far, the facts reported are compatible with the generalization that word accent is stem-final, or, if the stem-final syllable is word final, the accent is stem-initial. However, all such cases of stem-initial accent are penultimate (except for 10d), which raises the question whether the avoidance of word final accent really causes initial rather than penultimate accent. Some additional rules are necessary for stems with the vowel /«/. A stem-final syllable with the schwa can be accented, but only if no other syllable with another vowel precedes it, for example: (15) a. g«tg-«n c. e.

t«lw«lq-«n r«m«t-«k

‘lake’

b.

m«§«kw-«n

‘shirt’

‘fire site’ ‘to wash’

d.

r«kg«t-«k

‘to stick’

However, if there is a syllable with a full vowel preceding the last syllable of the stem with a schwa, this preceding syllable is the one that gets accented, as shown in (16). (16) a. patg«rg-«n c. e.

tatl«N-«k ro§g«p-«k

‘hole’

b.

‘to answer’ d. ‘to lose strength’

pipiq«lg-«n

‘mouse’

rent«N-«k

‘to get away’

The example in (16b) suggests that the penultimate location is chosen if the final location is not acceptable. This, perhaps confirms that the penultimate rather than the initial syllable is the alternative to the stem-final location. In the verbs consisting of one syllable with a short /«/ the location of the accent moves to the prefix if it contains a full vowel, for instance in the prefixed forms in (17). (17) a. r«w«k

‘to put aside’ b.

ge-nw«lin

c.

t«w«k

‘to inform’

d.

ga-tw«len

e.

t«m«k

‘to kill’

f.

ga-nm«len

‘put aside-part.’ (stem /r«w/ ~ /nw«/) ‘informed-part.’ (stem /t«w/ ~ /tw«/) ‘killed-part.’ (stem /t«m/ ~ /nm«/).

In sum, Chukchi accent is bound to the domain of the word stem, but weight sensitive within it. The accent is located on the last syllable of the word stem, if the last syllable is closed. If the last syllable is open, the location of accent is dependent on vowel height. We get a shift to the penultimate syllable, if the stem-final syllable is also word-final. We

518

Rene´ Schiering and Harry van der Hulst

also get a shift leftward to the penultimate syllable if the vowel in the designated accent syllable is a schwa. According to Muravyova (1979), Alutor word forms are divided into two classes depending on their rhythmic organization: (i)

Word forms in which one of the initial syllables is more prominent.

(ii) Word forms which are unaccented. The accentual pattern of the accented words can be described with reference to a distinction between light and heavy syllables. A light syllable is of the type C« (which cannot be accented), all other syllable types constitute heavy syllables. The following three informal rules capture accent placement in Alutor. (i) (ii) (iii)

The accent can be located either on the first or on the second syllable, the latter being a more preferred option. Only a heavy syllable can be accented, light syllables are always unaccented. The last syllable of the word cannot be accented.

As a result of these rules, monosyllabic accented words are prohibited since the only syllable of such forms constitutes the word-final syllable. Monosyllabic words are function words, such as, e.g. the conjunctions to ‘and’ and mej ‘hi’. In disyllabic words, accent is realized on the first heavy syllable, as exemplified with the data in (18). (18) a. ?ak«k c. k«ttil

‘son’ ‘forehead’

b. tatul ‘fox’ d. wala ‘knife’

With monosyllabic roots and disyllabic roots with an initial light syllable, several strategies which add syllabic substance are employed in the nominative case: i) a syllabic vowel « is inserted, as in law«t ‘head’ (/lawt/), ii) a su‰x is added, as in milg«n ‘fire’ (/milg-n/) and N«ralN«n ‘knee’ (/Nra-lN«n/), iii) reduplication applies, for example in piN-piN ‘ash’ or n«m-n«m ‘settlement’. Such complexity can also be achieved by the gemination of the last consonant and a schwa-insertion after the geminate, for instance in Najj« ‘mountain’ (/Naj/) and q« ipp« ‘bread’ (/q« ip/). Gemination does not only occur in una‰xed words, but also in forms with mono-consonantal su‰xes, e.g. N«vujj« ‘he began’ (/Nvu-j/), but not in piNku-j ‘jump-3SG/masc-past’. With some su‰xes, however, /i/ is inserted instead of a schwa vowel, cf. N«ra-tti ‘two knees’ (/Nra-t/) vs. g«tka-t ‘two legs’ and g«va-kki ‘to stumble’ (/gva-k/) vs. piNku-k ‘to jump’. y

y

Word accent systems in the languages of Asia

519

In trisyllabic words, accent falls on the second syllable if it is heavy, for example in vitat«k ‘to work’, N«ralN«n ‘knee’ and vag«lN«n ‘nail’. Otherwise, the first syllable is accented, if it is heavy and the second syllable is light, as in e.g. tilp«qal ‘shoulder’ or t«rg«t«r ‘meat’. In more formal terms, accent assignment in Alutor can be captured by the following rules. (i)

In a word consisting of one heavy syllable the accent is realized on this syllable: #H#!#H# e.g. Najj« (/Naj/) ‘mountain’, g«mm« (/g«m/) ‘I’

(ii)

In a disyllabic word, accent is assigned to the first syllable, if it is heavy or to the second syllable, if the first syllable is light:3 a. # H H # ! # H H # e.g. tatul ‘fox’, k«ttil ‘forehead’, paNkan ‘hat’ b. # L H # ! # L H # e.g. p«Nunn« (/p«Nun/) ‘mushroom’, g«vakki (/g«va þ k/) ‘to stick’4

(iii)

In a polysyllabic word, the second syllable is accented, if it is heavy or the first syllable is accented if it is heavy and the second syllable is light: a. # H/L H . . . ! # H/L H . . . e.g. quraNa ‘deer’, ?«np«qlavul ‘old man’, ?atN«lka ‘it hurts’, n«malqin ‘good’, n«c«qqin ‘cold’ b. # H L . . . ! # H L . . . e.g. /«np«Nav ‘old woman’, nilg«qin ‘white’, łot«qin ‘thin’

The rule in (iiia) has one exception: the root /aw(«)ji/ ‘to eat’. This root has two allomorphs, i.e. /aw«ji/ in the causative taw«jat«k ‘to feed’ and /awji/ (phonetically reduced to [oji]) in the non-causative verb /ojik. All forms of the verb /ojik are accented as if they were derived from the allomorph /aw«ji/, for example tojitk«n (¼*taw«jitk«n) ‘I eat’, but not as *tojitk«n. After the rules of accent assignment have been applied, it might turn out that the last syllable of the word is accented (see the rules in (i) and (iib)). In these cases, the following rules of syllable expansion apply. 3. Alutor has no accented bi-syllabic word forms with a second light syllable. 4. Alutor has no word forms of the type C«CV (where V is any vowel).

520

Rene´ Schiering and Harry van der Hulst

(iv)

In a form with an accented final syllable, the last consonant is geminated and a schwa is inserted: ¼ C1VC2 ¼ ! ¼ C1VC2 ¼ C2« ¼ e.g. Najj« (/Naj/) ‘mountain’, p«Nunn« (/p«Nun/) ‘mushroom’

(v)

In a form with an accented final syllable where the last consonant is a separate morpheme, this consonant is geminated and either schwa or /i/ is inserted. /i/ is inserted after –k or –t, schwa is inserted elsewhere: a. ¼ C1V þ C2 ¼ ! ¼ C1V þ C2 ¼ C2i ¼ / C2 ¼ k or C2 ¼ t e.g. g«va þ kki (/g«va þ k/) ‘to stick’, l«la þ tti (/l«la þ t/) ‘two eyes’ b. ¼ C1V þ C2 ¼ ! ¼ C1V þ C2 ¼ C2« ¼ / C2 A k or C2 A t e.g. t«nu þ nn« (/t«nu þ n/) ‘I ate him’

The cases of syllable expansion can best be described by already introducing two morphs in the lexicon (naj/najj«, k/kki, t/tti, n/nn«) because there is no information about accent at the stage of derivation where the word morphs have to be chosen. Finally, (19) provides some examples of rhythm assignment in accented words derived from the same root but having di¤erent structures. (19) a.

/# t þ pr þ n #/ ! ¼ t«p ¼ r«n ¼ (iia) ! ¼ t«p ¼ r«n ¼ ‘I took him o¤ ’

b. /# pr þ n #/ ! ¼ p« ¼ r« þ n ¼ (iib) ! ¼ p« ¼ r« þ n ¼ (vb) ! ¼ p« ¼ r« þ n ¼ n« ¼ ‘he took him o¤ ’ c. /# t þ ta þ pr þ N þ n #/ ! ¼ t« þ tap ¼ r« ¼ N«n ¼ (iiia) ! ¼ t« þ tap ¼ r« ¼ N«n ¼ ‘I will take him o¤ ’ d. /# ta þ pr þ N þ ni þ n#/ ! ¼ tap ¼ r«N ¼ nin ¼ (iiia) ! ¼ tap ¼ r«N ¼ nin ¼ ‘he will take him o¤ ’ e. /# ga þ pr þ lin #/ ! ¼ gap ¼ r« ¼ lin ¼ (iiib) ! ¼ gap ¼ r« ¼ lin ¼ ‘he is taken o¤ ’ Zhukova (1972) notes that accent in Koryak falls, in the majority of cases, on the first syllable of a disyllabic word (20).

Word accent systems in the languages of Asia

va-jat

‘people’

b. c.

a-nok wa-jew

‘spring’ ‘the base piles of the yarangha (type of house)’

d. e.

pe-lak aj-Non

‘leave someone / something’ ‘a long time ago’

f. g.

met-˜aN ‘good’ giw-lin ‘he said’

(20) a.

521

Evidence from longer words suggests that the second stem syllable is accented, cf. (21) and (22). However, the data thus far could also be accounted for if final accent avoidance is assumed, with penultimate accent being the chosen pattern: (21) a. b. c.

wa-la

‘knife’

wa-la-ta ‘knife – instr.’ gaj-kç-wa-la-ta ‘with knife

(22) a. ve-tat b. ve-ta-tçk c. ve-tal-laj d. ko-ve-ta-tçn

‘work’ ‘to work’ ‘they worked’ ‘he works’

Only example (22d) points to the second syllable rather than the penultimate pattern. In words consisting of four or more syllables, accented and unaccented syllables alter relatively rhythmically, as shown in (23). In these cases, the di¤erence between primary and secondary accent is not obvious. (23) a. b. c.

mçcˇ-cˇaj-go-cˇaw-Nç-la ‘we studied’ cˇaw-cˇç-wacˇ-˜e-naN ‘in Koryak’ ga-ve-tał-Nvo-la-ta

‘work a little’

All forms are incompatible with the penultimate interpretation of primary accent. If the first accent is primary then (23a) and (23c) confirm the second syllable pattern. This pattern is clearly present in the Palansk dialect of Koryak discussed below. (23b) perhaps has initial rather than second syllable accent because the initial vowel is more prominent than the second vowel, which is also confirmed by the form in (24a). An analysis which accounts for all attested accent placements thus begs future research.

522

Rene´ Schiering and Harry van der Hulst

Although word accent generally has no segmental e¤ect, the so-called a¤ective intonation, which sometimes shifts word accent to the final syllable, is accompanied by a vowel quality change in the accented syllable (see (24) and also (2) above for a similar phenomenon in Chukchi). (24) a.

amkçka

‘many’

b. amkçko

‘especially many’

c.

aNajqçka ‘bad’

d. aNajqçko ‘especially bad’

e.

qçjçm

‘negation – prohibition’

f. qçjom

g.

kujqetçN

‘he is in a hurry’

h. kujqetoN ‘he is really in an hurry’

i.

jaqam

‘at once’

j.

jaqom

‘absolutely at once’

k.

javacˇ

‘late’

l.

javoc

‘especially late’

m. qçgit

‘look – imperative’

n. kçget

‘look-imperative’ (for immediate reaction)

o.

‘female name

p. mite

‘female name’ emotionally pronounced

miti

‘especially strict prohibition’

In the Palansk dialect of Koryak the following accent patterns have been noticed (Zhukova 1980: 30–31). In disyllabic words, accent is placed on the first syllable (25). (25)

a.

temkem

‘many’

b.

rokak

‘in the tobacco case’

c.

kukek

‘in the cooking bowl’

d.

ivek

‘speak’

e.

iw·aq

‘devil’

f.

nuraq

‘for a long time’

g.

tite

‘when’

h.

/opta

‘all’

i.

pleku

‘sale base’

j.

lewte

‘head-instr.’

k.

tum˜u

‘comrades’

The same pattern is also attested in the context of loan words, as those given in (26). (26) a.

molok

‘milk’

c.

s*a··a

‘shawl’ d.

b.

pike

‘ox’

q·eppa

‘bread’

In trisyllabic words accent falls on the second syllable, for example: (27) a. roroNa c. kusineN e. ?a·amta g. aso/as

‘sleeping curtain’ b. ririNe ‘spoon’ ‘a fly’ ‘salmon’

‘polar dolphin’

d. mi·utpi ‘rabbit’ f. pujepuj ‘baked meat’ h. quliqul ‘voice’

Word accent systems in the languages of Asia

i.

timitim

‘raft’

k. junet«k ‘live’ m. ?elmulq«n ‘heap of snow’

523

j. riqutk«n ‘achieves, overcomes’ l. lep/et«N ‘swan-dat.’

In tetrasyllabic words, accent falls on the second syllable, as exemplified in (28). (28) a. Niter˜ere c. e.

‘together (both)’

b. jitevineN d. cˇaNetav«k

ułułupi ‘baby’ miletat«k ‘swipe’

f.

torovat«k

‘towel’ ‘get scared’ ‘say hello’

In a trisyllabic word in which the second syllable contains a schwa (/«/), the first syllable gets the accent instead of the second one, cf. (29). (29) a. wut«kku

b. /ott«Ntak ‘go get the wood’

‘here’

c. tom˜«lN«n ‘comrade’

d. ˜utt«lin

e. niwt«kin

‘low’

f.

g. ka˜«rN«n

‘the mouth of the river’

‘forest-like’

nikm«kin ‘short’

If all syllables in a word contain a schwa, accent appears on its default locus, which is the second syllable, i.e. m«n˜«lN«n ‘hand’. In the Palansk variety, secondary accent usually falls on the last syllable of a word, as illustrated in (30). (30) a. ilq«tvik c. taq«t?aw

‘close eyes very tightly’ b. /us’qe?us’ ‘bridge’ ‘alyki’

e. mull«mul ‘blood’ g. /utt«?ut

d. Nelvel?«n f.

ulu?ul

‘group of animals’ ‘sivuch’

‘tree’

To summarize, accent in Palansk Koryak is placed on the second syllable of a word in the default case, i.e. unless this is the final syllable and unless this syllable contains a schwa while the first syllable contains a full vowel. As such, this pattern is similar to the one proposed above for the other dialect of Koryak. According to Bobaljik (p.c.) accent in Itelmen is located on the first syllable (whether the head is a full vowel or schwa), excluding agreement prefixes (of which only two are syllabic). There is no rhythmic/metrical alternation, but the prominence tapers o¤ from the accented syllable towards the right edge of the word (much like a list intonation in English). Syllables containing a glottal stop or a glottalized consonant have an increased prominence relative to their neighbors, regardless of their linear position in the word.

524

Rene´ Schiering and Harry van der Hulst

Generalizations In all cases discussed, except Standard Chukchi (as described in Skorik 1961), accent is on the left edge (initial or second syllable). However, in Standard Chukchi there is a predominant stem-final pattern with avoidance of word final accent leading to either the penultimate or indeed initial syllable. In his comparative Chukotko-Kamchatkan dictionary, Fortescue (2005: 13) hypothesizes that primary word accent in the proto-language was presumably on the penultimate syllable. The domain for this accent assignment references the stem (including the first part of reduplicated forms and partial reduplications of disyllabic stems) or the combination of stem and syllabic su‰x. This principle of accent placement is still traceable in the Chukchi data presented above. It should be evident from the previous survey that Chukotko-Kamchatkan still o¤ers rich research opportunities in the study of word accent. In particular, the mapping of accent domains to morphological structures with varying degrees of complexity and the interaction of accent and phonotactics are poorly understood. The relevance of onsets in Chukchi accent placement (see (1d, e) above) is only one of various phenomena instanced in these languages which could be of major interest to phonological theory (see also Everett (1988) and M. Gordon (2005) on onset-sensitive accent systems). 4.2. Isolates Apart from the major language families located in East Siberia, we also find four language isolates which are spread over the area. We start with Ket in Central Siberia, proceeding to Yukaghir in the East, followed by Nivkh and Ainu in the South East of North-East Eurasia. 4.2.1. Ket Genetic information: Isolate Ket is spoken in several regions and valleys east of the Khanti and Mansi in central Siberia. It is related to Yugh with which it forms the Yeniseian language group. YENISEIAN Ket Yugh

Word accent systems in the languages of Asia

525

Accent information As the only language of the area, Ket has been reported to have phonemic tone oppositions (Comrie 1981: 262). Following Werner (1996), this typological feature characterizes the entire Yeniseian group, including Yugh and the extinct Kot language, and can be reconstructed for the protolanguage. According to Werner (1997b: 20–25), Ket distinguishes four tones: (i) a high-even or slightly rising tone distributed over a half-long vowel, e.g. 1a>m ‘mother’, (ii) a short, rising-falling tone accompanied by laryngeal stricture or a full glottal stop, e.g. 2i / ‘day’, (iii) a long risingfalling tone without laryngealization or glottalization, e.g. 3aøN ‘hot’, and (iv) a short falling tone 4aN ‘rope’.5 Apart from monosyllables, these four tones can, in principle, also occur on any syllable of a polysyllabic word, e.g. 1di> 3tˆøl’ ‘I’m cold’. However, in a number of polysyllabic words, and sometimes dependent on speech rate, tone is only retained on one syllable in the surface form, cf. (31). (31) a. b. c. d. e.

1di>jaq

‘I leave’ as’2kˆ?t ‘fairytale’ 3aøNGat ‘to heat up’ 4dcn’-tet ‘I hit him’ 3diøtaN ‘I carry her’

Assuming that tone is only realized on the accented syllable of these forms, the words in (31a, c–e) would be accented on the first syllable and the one in (31b) would be accented on the second syllable. Additionally, polysyllabic words may also surface with a high-low (32a–b) or low-high (32c–d) accent pattern. (32) a. b. c. d.

u´ta`q ‘to hold’ dc´nu`l’aN ‘handles of a knife’ a`mma´s ‘stepmother’ ta`vu´l’aN ‘bare-footed’

In such cases, the syllables with higher pitch, i.e. the first syllables in (32a– b) and the second syllables in (32c–d), respectively, are perceived as being accented. These accent patterns also distinguish the grammatical number opposition in pairs such as du´mgı`t ‘birdie’ vs. du`mgı´t ‘birdies’. 5. With respect to the phonetic realization of the fourth tone, the several Ket dialects show considerable variation, see Vajda (2004: 9) for a brief overview and Werner (1996) for details.

526

Rene´ Schiering and Harry van der Hulst

Although the manifestation of tonal opposition in Yugh deviates from the one encountered in Ket, its accentual system parallels the one described above. The interested reader is referred to Werner (1996, 1997a, b) and Vajda (2004) for detailed comparisons and further analyses of accent in Yeniseian. 4.2.2. Yukaghir Genetic information: Isolate Yukaghir, usually subdivided into a Northern and a Southern variety, is spoken in Yakutia and on the Kamchatka peninsula. Together with the now extinct languages Chuvantsy and Omak it constitutes the Yukaghir language group. YUKAGHIR Yukaghir, †Chuvantsy, †Omok StressTyp extracts Yukaghir [L/L] f Stress falls on the last syllable that is closed by a consonant or has a long vowel. f Else stress falls on the last syllable of the word. f There are exceptional bisyllabic (C)VCe forms with initial stress. ¨aøs’e a¨roøje sˇana¨ja#q mo¨ro

‘domestic deer’ ‘kind of fish’ ‘fur coat’ ‘hat’

le¨gul ‘food’ ¨a#olhoro ‘hare’ cˇolo#¨lok ‘just so’6 co¨bi¨ne ‘spear’

Additional information The exceptional behavior of bisyllabic (C)VCe forms is exemplified with the words in (33), taken from the same source that the StressTyp entry and all following discussion is based on, i.e. Maslova (2003: 58–59). (33) a. ¨leme b. ¨kOd’e

‘what’ ‘larva’

¨mure

‘shoe’

c.

6. This line gives additional examples not originally included in the StressTyp entry.

Word accent systems in the languages of Asia

527

The accent placement rules summarized above also apply within derived words, such as augmentative derivatives (34a–b) and causative derivatives (34c–d). (34) a. ¨sˇowhe ‘plate’ c. sˇel’¨ges’ ‘break (intr.)’

b. sˇow¨hotke ‘plate’ d. sˇel’ge¨dejm ‘break (trans.)’

In the first derivation, the concatenation of the su‰x -tke and resyllabification conspire to create a heavy penultimate syllable /hot/ which attracts accent. In the second derivation, the su‰x -dE-(j)- itself provides a final heavy syllable onto which the word accent moves. In her reconstruction of Proto-Yukaghir phonology, Nikolaeva (2006: 75) discusses two scenarios for the historical development of disyllabic stems. Starting from the observation that certain stems show synchronic variation with respect to accent placement and vowel length, e.g. ¨(C)V:C« vs. (C)V¨CV, the first reconstruction assumes that the di¤erence in initial vs. final accent placement was already present in the proto-language. In a second step, vowel lengthening of the accented initial syllable and vowel reduction in the unaccented final syllable yielded the segmental template ¨(C)V:C«. The second reconstruction, adopted by Nikolaeva, takes the vowel length to be prior and assumes accent attraction to this heavy syllable. This second reconstruction thus projects the analysis for modern Yukaghir to Proto-Yukaghir. 4.2.3. Nivkh Genetic information The Language Isolate Nivkh (Gilyak) is spoken in the far east of Russia, along the Amur River and on Sakhalin Island. Three dialects can be distinguished: the Amur dialect spoken on mainland Russia, and the East and North Sakhalin dialects spoken on the respective regions of Sakhalin Island. StressTyp extracts Gilyak; Nivkh [LEX] f f f

Stress often falls on the first vowel of the word. Sometimes stress is lexically determined. Su‰xes of the vocative, the imperative, and the conditional converb, are always stressed.

528

Rene´ Schiering and Harry van der Hulst

¨xaunt

‘(someone) calls’

xa¨unt

¨toølkar halçl ¨çtçk-aø

‘(it became) very flat’ ‘father’

¨patiøkur ‘slowly’ vi-¨ja ‘go.sec.sg.imp’

ymyk-¨a

‘mother.voc’

‘(someone) dries’

Additional information In addition to the general rules of accent placement and the examples given above, Gruzdeva (1998: 12–13) also notes some cross-dialectal differences in accentuation. For instance, the word for ‘(someone) walks/ walked’ is accented on the initial syllable in the Amur dialect (¨amamd’) and on the second syllable in the East Sakhalin Island dialect (a¨mamnt). In a similar vein, Panfilov (1962: 22) notes forms which are accented on the second syllable, e.g. um¨gu ‘woman’ and ut¨ku ‘man’. An example for the auto-stressed conditional converb marker is presented in (35). (35) vi-¨g aj. . .

‘If [I] go. . .’

Shiraishi (2006: 30–31) additionally discusses phrasal accent, which does not di¤erentiate between compounds and phrases, as is shown with the examples in (36). (36) a. ¨ła ajs b. ¨kins Nçłk c.

¨pilkar Nçłk

(eye gold) (devil face)

‘glasses’ ‘devil’s face’

(big face)

‘a big face’

In both compounds (36a) and phrases (36b–c), the first constituent has primary accent on the initial syllable. With respect to phonetic correlates of accent, Shiraishi (2006: 30) notes high pitch and for some speakers palatalization of the consonant before the front vowels /i, e/. 4.2.4. Ainu Genetic information Ainu is presently spoken on the Japanese Kuril Islands (Tsishima), Hokkaido and in Russia. Formerly, it has also been spoken on the South Sakhalin Island. From the at least 19 original dialects, the following dialects can be distinguished at present: Sakhalin (Saghilin), Taraika, Hokkaido (Ezo, Yezo) and Kuril (Shikotan, Tsishima). The last speaker of Sakhalin dialect died in 1994. (The question of whether Ainu should be included into the Altaic language family remains a controversial issue; see for instance Ruhlen 1991).

Word accent systems in the languages of Asia

529

Accent information Ainu has a pitch-accent system in which the rise from low tone to high tone marks the accented syllable. According to the descriptions in Refsing (1986: 73–74), Dettmer (1989: 43–50) and Tamura (2000: 21–23), accent placement is governed by the following rules. The following data demonstrate the application of the most basic weight-sensitive rule by which accent is placed on the initial syllable if it is closed or contains a diphthong, i.e. if it is heavy (37a–c), and on the second syllable otherwise (37d–f ). ¨nonno ¨tapsut ¨aynu

‘flower’ ‘shoulder’ ‘Ainu’

d.

sa¨pa

‘head’

e. f.

sa¨paha ‘his/her/its head’ ku¨sapaha ‘my head’7

(37) a. b. c.

There are four sources of deviation from this pattern. First, there are words which are realized with accent on an initial, open syllable, cf. ¨nisap ‘sudden’ vs. ni¨sap ‘shin’. Note that in such pairs, the locus of accent is distinctive. Diachronically, this aberrant accent placement may be due to the loss of vowel length in the first syllable. As Tamura (2000: 22) points out, the more conservative Sakhalin dialect has an initial long vowel in such words, which attracts accent following the weight-sensitive rules summarized above, e.g. ¨miina ‘to laugh’. In the Hokkaido dialect, the initial vowel got shortened, but initial accent placement got preserved, yielding the form ¨mina. A tendency to prolong the vowel of an accented open syllable has also been noted by Refsing (1986: 73) for the Shizunai dialect. Secondly, contracted forms with preposed personal forms may also exhibit irregular accent placement. For instance, when the first vowel of the word ko’¨otopihi ‘my hair’ is deleted, accent surfaces on the initial open syllable of the contracted form ¨kotopihi. Thirdly, certain personal prefixes, such as eci- (second person plural), are not integrated into the accent domain and do not a¤ect accent placement, e.g. ’eci-sa¨pa-ha ‘your (plural) hair’. Fourthly, in derived words, accentuation is sensitive to morphological 7. The accentuation of these last two words di¤ered in southwestern Hokkaido dialects around 1955, i.e. sapa¨ha ‘his/her/its head’ kusa¨paha ‘my head’, suggesting accent on the third syllable.

530

Rene´ Schiering and Harry van der Hulst

structure such that the stem receives accent in combinations like ¨e-re (eat-CAUS) ‘cause to eat’, and that compounds are accented on the first member, e.g. ¨re-kor (name þ have) ‘to have a name’ and ka¨muy-mosir (god þ country/homeland) ‘the Gods’ world’. 5. Asia Minor and Central Eurasia (Altaic) This section is concerned with the Altaic languages. Whereas the genetic a‰liation of Turkic, Mongolian and Tungusic under the (‘Micro-’)Altaic node is by now accepted (see Georg et al. 1999 for an overview), the additional inclusion of Korean, Japanese and Ainu into this family (see e.g. Ruhlen 1991) has long been debated. More recently, Robbeets (2005, 2007a, b) compiled substantial lexical and morphological evidence in favor of this ‘Macro-Altaic’ hypothesis. Adopting the broader conception of Altaic, this section presents accentual data from Turkic, Mongolic, Tungusic, Korean and Japanese and thus covers an area which stretches from the Mediterranean and Black Sea in the West (Turkey) to the Pacific Ocean in the East (Korea and Japan), forming a middle layer from Aisa Minor to Central Eurasia. 5.1. Turkic Genetic information Turkic is generally believed to be a direct descendant of Altaic proper. The Ethnologue (R. Gordon 2005) distinguishes six major branches within the family (see Johanson 1998 and Ro´na-Tas 1998 for the history and reconstruction of Turkic). The accentual data surveyed in this section cover all major subbranches, however, to the exclusion of Urum. TURKIC BOLGAR: Chuvash EASTERN: Ainu, Chagatai, Ili Turki, Uyghur, Northern Uzbek, Southern Uzbek, West Yugur NORTHERN: Southern Altai, Northern Altai, Shor, Dolgan, Karagas, Khakas, Yakut, Tuvin SOUTHERN: Crimean Turkish, Salar AZERBAIJANI: South Azerbaijani, Northern Azerbaijani, Turkic Khalaj, Qashqa’i, Sakchuq TURKISH: Balkan Gagauz Turkish, Gagauz, Khorasani Turkish, Turkish

Word accent systems in the languages of Asia

531

TURKMENIAN: Turkmen WESTERN: Urum ARALO-CASPIAN: Karakalpak, Kazakh, Kirghiz, Nogai PONTO-CASPIAN: Judeo-Crimean Tatar, Karaim, KarachayBalkar, Kumyk URALIAN: Bashkir, Chulym, Tatar StressTyp extracts Chuvash [L/F] Altaic, Turkic, Bolgar. Chuvash republic (Russian Federation). f Stress falls on the last syllable with a full vowel, else on the first syllable. la¨Sa sarla¨ka

‘horse’ ‘widely’

¨ala(k ¨e(sle(pe(r

‘door’ ‘we shall work’

Uzbek, Northern [U] f Primary stress normally falls on the final syllable. f In words of three syllables a secondary stress may appear on the first syllable. f In longer words a tertiary stress can appear in between the primary and secondary stress. dZc¨nim ‘my soul’

quS¨ca ‘little bird’

Turkish; Osmanli [F/L;LEX] f Stress normally falls on the final syllable. f Stress may also appear anywhere in the word on lexically marked syllables. f In placenames and loans stress is antepenultimate if that syllable contains a long vowel or is closed and the penult is light, otherwise stress falls on the penultimate syllable. ta¨n" tan"d"kla¨r"m tan"¨d"k ta¨n"mad"klar"m"z tan"d"k¨lar

‘know’ ‘my acquaintances’ ‘acquaintance’ ‘those we do not know’ ‘acquaintances’

ak¨Samleyin is¨tanbul ¨ankara

‘at evening’ ‘istanbul’ ‘Ankara’

532

Rene´ Schiering and Harry van der Hulst

Bashkir; Basquort [U] f Stress falls on the final syllable of the word (including su‰xes). kHi¨tHApH

‘book’

kHitHAp¨lAr

‘books’

kHitHApHlArF¨bFD ‘our books’ Additional information We will start our elaboration of the accentual patterns in Turkic with the most divergent Turkic language Chuvash, representing the Bolgar branch and being spoken in the Chuvash Autonomous Republic in the Russian Federation. This language has an accentual system that is remarkably similar to that of Armenian (cf. 8.4.1.6.). Clark (1998: 435–436) locates stress on the final syllable, unless this syllable contains a reduced vowel, in which case accent occurs on the penult. The accent keeps moving to the left to find the rightmost full vowel; if no such vowels occur in the word, accent is placed initially. Thus, Chuvash can be analysed as a LAST/FIRST system, in which full vowels count as heavy (cf. Krueger 1961 and Hayes 1995). Chuvash is divergent from other Turkic languages in that the position of accent is dependent on vowel quality. It may have inherited this feature from the Uralic languages which are its neighbours geographically. This could have been, for example, a mutual influence between Chuvash and Cheremis, Cheremis having inherited the LAST value from Chuvash, while Chuvash has inherited the default FIRST value from Cheremis. Gordon (2000) o¤ers a discussion of Chuvash word accent, based on Dobrovolsky’s (1999) measurements of intensity, duration and fundamental frequency. According to these results, intensity and/or durational increase is a property of heavy syllables that are accented according to the rule that the rightmost heavy syllable is accented (hH, lH, Hl). If the default (when there is no heavy syllable) is ‘leftmost’, we predict that in ll words the initial syllable carries accent. However, initial syllables in such words do not have such phonetic correlates. Instead, all initial syllables carry a fundamental frequency peak, with fundamental frequencies going down toward the end of the word. This is a normal declination e¤ect. Dobrovolsky sees this initial high frequency as intonational in nature, such that the pitch range tends to decline over time. If indeed the alleged initial accent (in words with light syllables only) is merely caused by an initial high fundamental frequency, one would expect that words would only surface with this feature if they occur in initial position of an intonational domain. Whether this is indeed the case, remains to be tested. Meanwhile, one must wonder whether it is ‘reasonable’ to assume that an accent language has words that are accentless.

Word accent systems in the languages of Asia

533

Comrie (1997c: 923–925) shows that Uyghur, another representative of the Eastern branch, deviates from the final accent pattern prevailing across Turkic in being sensitive to syllable weight. If the final syllable is light and the penult heavy, we find penultimate accent. (38) a. ma#sˇ´¯ına ‘machine’ b masˇ"n"-da´ o¨¯za¨´m ‘myself ’ As the second example shows, when a su‰x is added and the long vowel is outside the two-syllable window, stress is final. Stress is also final if both the penultimate and the final syllable are heavy. Vowels that do not have primary stress loose their length, but the rightmost non-stress (underlying) long vowel will have a secondary stress, so the second form is actually /masˇ`ın"-da´/ (see also Hahn 1991). Word accent in Yakut, our major representative from the Northern branch, normally falls on the last syllable of a word (Krueger 1962: 70). However, there are exceptions to this default accent placement, most evidently in deictics, as listed in (39). (39) a. b. c. d. e.

¨subu

‘this’

¨siti ¨uonna ¨ittene ¨b"cˇ"kay

‘that’ ‘then, and’ ‘backwards’ ‘little, itty-bitty’

We now turn to the Southern Turkic languages. Of these Turkish, the main language of Turkey, is most intensively studied (see Kornfilt 1997, Lewis 2000, and Go¨ksel and Kerslake 2005 for reference grammars). In the default case, Turkish primary accent falls on the final syllable of a word. There are, however, numerous deviations from this pattern (for discussion of Turkish accentual data see Sezer 1981, Kaisse 1985, Barker 1989, Comrie 1997a, Csato´ and Johanson 1998). The regular pattern can be illustrated with the following examples. (40) a. b. c.

tan"-dı´k tan"-d"k-la´r tan"-d"k-lar-ı´m

‘acquaintance’ ‘acquaintances’ ‘my acquaintances’

Special rules of accent placement apply within a part of the vocabulary consisting of native and foreign place names, person names and recent borrowings. Although the latter mostly conform to the phonotactics of

534

Rene´ Schiering and Harry van der Hulst

Turkish, their accent pattern is deviant.8 The examples in (41), taken from Sezer (1981) and Barker (1989), are arranged according to the weight of their final syllables. The lowered dots represent syllable boundaries. O.dı´.pus Go¨.re´.me Ke.ne´.di Pi.to.le´.mi In.di.ya.na.po´.lis b. Sa.mu.e´l.son Va.s¸´ıng.ton lo.ka´n.ta Ha.li.ka´r.nas c. a´n.ka.ra s¸a.ma´n.d".ra pe´n.ce.re s¸e´v.ro.le d. Men.de´l.son Kam.c¸a´t.ka Ay.z"n.ho´:.ver

(41) a.

‘Oedipus’ ‘Go¨reme’ ‘Kennedy’ ‘Ptolemy’ ‘Indianapolis’ ‘Samuelson’ ‘Washington’ ‘restaurant’ ‘Halicarnassus’ ‘Ankara’ ‘buoy’ ‘window’ ‘Chevrolet’9 ‘Mendelssohn’ ‘Kamchatka’ ‘Eisenhower’

On the basis of such data, Sezer and Barker draw the following conclusions with respect to accent placement in this part of the vocabulary. (42) a.

If the antepenult is heavy and the penult is open with a short vowel, accent falls on the antepenult b. otherwise accent falls on the penult

As a second major group of lexical exceptions, adverbs are usually accented on the first syllable, e.g. so´nra ‘after’, a´ncak ‘only’ and bu´rada ‘here’. Furthermore, accentuation in morphologically complex words is sensitive to the accentual properties of su‰xes. Some su‰xes, such as yap-a´rak ‘by doing’ and gid-ı´nce ‘having gone’, are auto-stressed in carry-

8. This class of items is also discussed by Sezer (1981) and Kaisse (1985). 9. Note that /vr-/ is not a licit syllable onset, so that s¸evrole must be syllabified as indicated.

Word accent systems in the languages of Asia

535

ing primary accent on their initial syllable. There are also bound morphemes which trigger primary accent on the syllable immediately preceding them. In the following examples, taken from Barker (1989), these morphemes are underlined. (43) a. b. c.

tanı´-ma-d"k-lar-"m-"z ‘those we do not know’ ‘with coalition’ koalisyo´n-la tan"-d"k-lar-"m-ı´z-mi ‘our acquaintances?’

The negation marker -mA in (43a) is a genuine verbal su‰x, the bound morpheme in (43b) is the su‰xed variant of the postposition ile ‘with’, while the question particle mI in (43c) is generally considered a clitic. In terms of their contribution to accent placement, these bound morphemes can be considered pre-accenting. The formal expression of this generalization has triggered a debate in which, amongst others, Kaisse (1985) and Barker (1989) have participated. More recently, the Turkish data fed a controversy between proponents of lexical pre-specification (Kabak and Vogel 2001) and proponents of co-phonologies (Inkelas and Orgun 2003). Bashkir, a Western Turkic language spoken in the Bashkir autonomous Republic of the Russian Federation is also described as having final accent in its native vocabulary (Poppe 1964). Loanwords from Arabic, Persian and Russian, among other languages, keep their original accent pattern. When su‰xes are added to a stem the accent shifts, with a small number of exceptions. As in Turkish, when several su‰xes are added the last one takes the accent: kita´p ‘book’, kitapla´r ‘books’, kitapları¨bı¨ð ‘our books’, kitapları¨bı¨ðða´n ‘from our books’. The same holds true for Tatar, another Western Turkic language spoken in Tatarstan and adjacent areas within the Russian Federation, e.g. bala´ ‘child’ and balala´r ‘children’. Comrie (1997b: 909–912) provides more detail on exceptional stress in Tatar, including initial stress in second person imperative forms of the verb and interrogative pronouns and adverbs. He also lists several unstressed su‰xes and shows that enclitics do not get stressed. The patterns described above, i.e. default final accent placement with lexically conditioned deviations, are mostly identical across the family. For more accentual data on individual languages see Boeschoten and Vandamme (1998) on Chaghatay, Boeschoten (1998) on Uzbek, Scho¨nig (1998a) on Azerbaijanian, Brendemoen (1998) on Turkish dialects, Kerslake (1998) on Ottoman Turkish, Scho¨nig (1998b) on Turkmen, Berta (1998a) on Tatar and Bashkir, Berta (1998b) on the West Kipchak languages

536

Rene´ Schiering and Harry van der Hulst

Kumyk, Karachay-Balkar, Crimean Tatar, Karaim, Kirchner (1998a) on Kazakh and Karakalpak, Csato´ and Karakoc¸ (1998) on Noghay, Kirchner (1998b) on Kirghiz, Hahn (1998a, b) on Uyghur, and Stachowski and Menz (1998) on Yakut. Generalizations Johanson (1998a: 34–35) makes the following general remarks about word prosodic phenomena in Turkic languages: accent is the capacity to bear high pitch. Most words have accent on their final syllable which will thus carry pitch. Su‰xes may be accented or unaccented. Personal su‰xes of the pronominal type, copula markers, negation su‰xes, and enclitic particles (such as /dA/ ‘and, too’) are unaccented, which means that they cause the accent to be on the syllable preceding them (cf. 43). He also notes an interacting changeable dynamic stress-accent, characterized by more energy of articulation. This stress accent tends to fall on the first syllable and is thought to be the source of the left-to-right vowel harmony system and rhyme patterns in Old Turkic poetry (cf. also Johanson 1998b: 111). Csato´ and Johanson (1998) remark that the initial stress-accent in Turkish often falls on the first syllable, especially when this syllable is heavy, i.e. containg a long vowel or being closed. Though typically at opposite ends of the words, both accents may coincide on the same syllable. This occurs when the location of accent is exceptional, non-final, as happens in the exceptional cases discussed above. In these cases, it is said that the pitch- and stress accent fall on the same syllable. In accordance with these generalization it is possible to describe the location of accent as follows: accent falls on the rightmost lexically marked syllable (i.e. the exceptional cases) otherwise on the last syllable. Stated this way, Turkish is a last/last system (cf. van der Hulst 1999). 5.2. Mongolic Genetic information Whereas the Ethnologue (R. Gordon 2005) considers Mongolic a direct descendant of Altaic, other language classifications, for instance Ruhlen (1991), posit an intermediate node ‘Mongolian-Tungus’. This latter view suggests that Mongolic is more closely related to Tungusic than to Turkic. Within Mongolic, an Eastern and a Western branch are distinguished. The following discussion, as well as the StressTyp entries, focuses on Khalkha Mongolian as a representative of the Eastern Mongolian branch.

Word accent systems in the languages of Asia

537

MONGOLIAN EASTERN: DAGUR: Daur MONGOUR: Kangjia, Tu, Bonan, Dongxiang, East Yugur OIRAT-KHALKHA: KHALKHA-BURIAT: BURIAT: China Buriat, Mongolia Buriat, Russia Buriat MONGOLIAN PROPER: Khalkha Mongolian, Peripheral Mongolian OIRAT-KALMYK-DARKHAT: Darkhat, Kalmyk-Oirat WESTERN: Mogholi StressTyp extracts Mongolian; Khalkha [F/F] Stress falls on the first syllable that contains a long vowel. f Otherwise stress falls on the first syllable of the word. f

bos¨guul ‘fugitive’ ¨uNSis«N ‘having read’

¨axe mori¨ooroo

‘elder brother’ ‘by means of his own horse’

Additional information In the philological literature, there is considerable disagreement about the formulation of accent placement rules in Mongolian. Bosson (1964) and Poppe (1970) describe the pattern as follows: primary accent is assigned to the rightmost non-final syllable containing a long vowel or diphthong. If such a syllable occurs only finally, accent is placed on the ultima. If no such syllables are present, accent is realized initially. This would make Mongolian a last/first system, instead of a first/first system as is stated in StressTyp. Street (1963) and Walker (1995) furthermore state that heavy syllables that do not carry primary accent surface with a secondary accent. Also, all initial syllables – heavy or light – may carry an initial accent in words that contain heavy syllables.10 Matthews (1951: 60), on the other hand, diagnoses default initial accent for Mongolian, accompanied by a tone on the final syllable. It is this pitch phenomenon which may give the false impression of final ‘stress’. Note 10. Walker (1995) notes that Buriat accentuation is essentially the same as Khalkha.

538

Rene´ Schiering and Harry van der Hulst

that in his description, accent placement is completely independent of vowel length. Such an analysis closely resembles Johanson’s (1998a: 34– 35) characterization of Turkic word prosody in terms of initial stress and final pitch-accent (cf. 5.1.). Contrary to what is suggested by the examples in StressTyp and the previous statements, Svantesson et al. (2005) claim that long vowels in Mongolian only occur in initial syllables. The interested reader is referred to their summary and discussion of the six di¤erent analyses proposed for Mongolian accent: i) initial accent, combined with final high pitch (cf. Matthews as cited above), ii) accent on the first long, phonemic, nonreduced vowel, initial accent elsewhere (cf. StressTyp), iii) accent on rightmost long vowels, initial accent elsewhere (cf. Bosson and Poppe as cited above), iv) final accent, v) accent on the first long vowel, final accent elsewhere, and vi) initial accent if the initial syllable contains a long vowel, accent on the second syllable elsewhere. They further speculate that researchers di¤er widely in their views on the location of accent because accent is not phonologically relevant, i.e. non-contrastive. 5.3. Tungusic Genetic information Following up on the discussion of Mongolic’s position in the Altaic family tree (cf. 5.2.), Tungusic is either considered a direct descendant of Altaic (R. Gordon 2005) or the sister of Mongolic in the subgroup ‘MongolianTungus’ (Ruhlen 1991). Within Tungusic, a Northern and a Southern branch are indubitably distinguished. In order to substantiate the empirical coverage of Southern Tungus, the StressTyp extracts in this section are complemented with additional accentual data from Udihe. TUNGUS NORTHERN: EVEN: Even EVENKI: Evenki, Oroqen NEGIDAL: Negidal SOUTHERN: SOUTHEAST: NANAJ: Nanai, Orok, Ulch UDIHE: Oroch, Udihe SOUTHWEST: Jurchen, Manchu, Xibe

Word accent systems in the languages of Asia

539

StressTyp extracts Evenki; Tungus [U/U] Tentative f In words of two syllables, stress mostly falls on the second syllable. f If a disyllabic word has either a long vowel or cluster of consonants in the middle of the stem, stress falls on the first syllable. f In a disyllabic word with two long vowels, stress falls on the second syllable. f If a word has more than two syllables with short vowels, stress falls on the last syllable. f Some polysyllabic words with final stress have an antepenultimate secondary stress. bi¨ra

‘river’

¨iøkte ‘tooth’

o¨ron

‘reindeer’ ¨halka

moø¨kaøn

‘stick’

‘hammer’

˙ngina¨kin ‘dog’

Orok; Oroc [U] f Stress falls on the final syllable of the word. Additional information Accent placement in Evenki can be further elaborated on with the following examples taken from Nedjalkov (1997: 316–317). In bisyllabic words, accent falls either on the initial syllable (44a–d) or, more often, on the second syllable (44e–h). (44) a. b. c. d. e. f. g. h.

u´lle o´llo hu´lla ha´lka bira´ gira´n sa:cha´:s mo:ka´:n

‘meat’ ‘fish’ ‘blanket’ ‘hammer’ ‘river’ ‘step’ ‘you knew’ ‘stick’

Initial accent is typically found if the first syllable of a bisyllabic word contains a long vowel or is closed by a coda consonant, as for example in (44d). In a similar vein, final accent placement is often found in words with final closed syllables, such as e.g. (44f ). If a bisyllabic word has two long vowels, as in (44g–h), accent falls on the second syllable.

540

Rene´ Schiering and Harry van der Hulst

The accentual patterns of words consisting of more than two syllables are exemplified in (45). (45) a.

emere´n

‘he came’

b. giramna´ c. biradu´k d. emed’e´n e. singı´lgen f. in’e´kted’eren

‘from the river’ ‘he will come’ ‘snow’ ‘(s)he is laughing’

g. bo`kono´n h. silkı`d’ara´n

‘he caught up with someone’ ‘she washes up’

‘bone’

In the absence of long vowels, such polysyllabic words are usually accented on the final syllable (45a–d). The words in (45e) and (45f ) show that non-final, closed syllables attract stress, resulting in patterns of penultimate accent and accented second syllables, respectively. Some polysyllabic words are realized with two accents: a primary accent on the final syllable and a secondary accent on the antepenultima, cf. (45g–h). In morphologically complex words, accent usually shifts to the su‰xes, unless the stem contains a long vowel, in which case accent will remain on this vowel. Certain su‰xes, such as the causative su‰x -vkAn and the volition marker -mu, are always accented. In (46), these su‰xes are underlined. u`llivke´nen b. gikumu´d’aran

(46) a.

‘she made someone sew’ ‘he wants to go’

All in all, these data support the StressTyp U/U-analysis which places accent on the rightmost heavy syllable in a right-edge two-syllable window and on the final syllable if this window contains only light syllables. Nikolaeva and Tolskaya (2001: 90–95) classify Udihe as a language with an unbounded quantity-sensitive accent system in which accent is placed on the rightmost heavy syllable or, if there are no heavy syllables, on the last syllable. The data in (47) provide the empirical basis for this characterization. (47) a. b. c. d.

u.ta.w¨a a.na.n¨a ¨a:n.ta.zi.ga e.m¨e:.me.i

‘that (ACC)’ ‘long ago’ ‘women (PL)’ ‘have come’

Word accent systems in the languages of Asia

e.

b¨’a.ta.wa

‘boy (ACC)’

f. g.

k¨’o.lo.lo dog.b¨o

‘in the mitten (LOC)’ ‘night’

h. i.

zab.da.l¨a ‘grass snake (LOC)’ na.uN.za.k¨a ‘boy’

j.

zu.e.z¨e

541

‘table’

In the default case, cf. (47a–b), primary accent falls on the final syllable of a word. If a word contains a bimoraic – i.e. long or laryngealized – vowel, primary accent shifts to this vowel, as shown in (47c–f ). Note that closed syllables (47g–h) and vowel clusters (47i–j) do not constitute bimoraic domains for the sake of accent placement. If a word contains more than one bimoraic syllable, accent falls on the rightmost bimoraic syllable, as illustrated in (48). (48) a. b. c.

te:.g¨iek

‘he sat’

su:.s¨i:.ni ‘he extinguishes’ s’a.i.d¨a:.ni ‘he salted’

These data confirm that the system is of the L/L variety. In other words Udihe displays the unbounded version of the bounded U/U system in Evenki. In morphologically complex forms, this general accent pattern gets obscured by the special prosodic behavior of di¤erent morpheme classes. First, word-final syllables with high vowels (/i/ and /u/) are extrametrical with respect to accent placement. However, this extrametricality is only restricted to inflectional a‰xes. (49) a. b. c. d.

a.di.l¨i zo.m¨i.mi de.gu.m¨u

‘net’ ‘stealing’ ‘poles for drying fish’

sa.kin.de.z¨e.mu ‘we will clap’

The word-final /i/ in (49a) is prosodically integrated into the accent domain and receives regular word-final primary accent. In (49b), on the other hand, the word-final /i/ shuns final accent as it belongs to the infinitive su‰x -mi. Parallel evidence is given for word-final /u/, such that the final syllable regularly receives accent if the high vowel belongs to the root in (49c), and shuns word-final accent if constituting the inflectional

542

Rene´ Schiering and Harry van der Hulst

su‰x -u in (49d). As an exception to this rule, the dative su‰x -du typically bears accent, e.g. o-d ¨u ‘here’ , zugdi-d ¨u ‘in the house’ . The second major deviation from the general accent pattern concerns the prosodization of enclitics. Unlike su‰xes containing non-high vowels, clitics are outside the domain of accent assignment. Consider the data in (50). (50) a.

abuga-l¨a

b. tada-w¨a

‘at the father (LOC)’ ‘arrow (ACC)’

c. abug¨a-da ‘and the father’ d. jaz¨a-ta ‘and of course’ (50a–b) demonstrate that the locative and accusative su‰x are prosodically integrated into the accent domain: in the absence of bimoraic syllables, the word-final syllable receive primary accent, which results in the accentuation of the su‰x vowel in both cases. In (50c–d), on the other hand, the focus enclitics -dA and -tA remain outside the accent domain and word-final accent is realized on the syllables immediately preceding them. Generalizations The uncertainty concerning Mongolian accent makes it di‰cult to draw conclusions. However, in most analyses, some weight-sensitive form of initial accent is typical of this language. The Tungusic languages Evenki and Udihe, on the other hand display a final weight-sensitive pattern, either bounded (U/U) or unbounded (L/L). Ther latter type of system is also prevalent in the Turkic branch of Altaic. According to Poppe (1960) these di¤erent systems have developed from a common source which had accent on the first heavy syllable and a musical pitch-accent on the final syllable. In the Turkic languages, the final pitch-accent has apparently attracted the default accent, while Mongolian seems to have capitalized on the left-edge aspect of this system. 5.4. Korean Genetic information In the organization of this chapter, we have adopted the hypothesis that Korean ultimately belongs to the Altaic phylum (see Robbeets 2005,

Word accent systems in the languages of Asia

543

2007a, b). Ruhlen (1991) considers Korean a sister of Japanese in his ‘Korean-Japanese’ subgroup of Altaic. R. Gordon (2005) makes no commitment to potential a‰liations of Korean with other languages or groups and thus classifies Korean as a language isolate. Accent information There is considerable variation in the word prosodic systems of Korean dialects. The following overview follows the typology proposed by Fukui (2003), which makes a major distinction between distinctive and nondistinctive tonal patterns. I. Distinctive tonal patterns In systems of this sort, words of equal length and syllabic make-up can have di¤erent tonal patterns. I.a. Multipattern systems In multipattern systems, the limit on the number of patterns is dependent on how many syllables the words/accentual domains can have. I.a.i. n (þ1) systems: Hamgyeongdo In this context Fukui (2003) mentions the Yanbian Longjing dialect in the north eastern part of China, which he says is similar to the Hamgyeongdo system that has been described in Ramsey (1978). Each word has at most one accent and each syllable in a word can bear that accent which, thus, has to be lexically marked. This accent is associated with a high tone. Words can also be unaccented in which case the H tone associates to the final syllable in the accentual domain. This system is comparable to Tokyo Japanese (which, in addition, has leftward spreading of the high tone; cf. section 5.5.). This is a n þ 1 system (n being the number of syllables in the word, each of which can carry an accent, þ1 adding the accentless option). In a system of this sort, the number of patterns increases with word length because the accent/high tone can be on any syllable. Fukui reports that younger speakers of this dialect seem to be losing the unaccented class (which merges with the final accent class), which would make their dialect an ‘n system’. This system is said to be closer to that of Middle Korean than any other system.

544

Rene´ Schiering and Harry van der Hulst

I.a.ii. n þ 2 systems: Gyeongsangdo (e.g. North G.: Daegu; South G.: Busan, Changnyeong dialects)11 According to Fukui (2003), the Changnyeong dialect (at the northern border of South Gyeongsangdo) has a class of words that, as in Hamgyeongdo, can carry an accent on every syllable. There do not seem to be unaccented words, so this part of the system is an n system. In addition, this dialect has two classes of words that have a ‘double H pattern’: initial HH followed by Ls or initial LHH followed by Ls. This adds two patterns, hence n þ 2. Additionally, Cho (2003) notes vowel length in North Gyeongsangdo, but not in South Gyeongsangdo. I.b. N-pattern systems: West Gyeongsangdo (including some Chonnam dialects) These systems have a fixed number of patterns (N), i.e. the number of patterns is not dependent on word length. Fukui (2003) presents the Chonnam Gwangyang Jinsang dialect, which has 4 patterns, hence N ¼ 4, as a representative of this type. Two of the tonal patterns are double high tone patterns as in the Changnyeong dialect. The two others are single H patterns with accent either falling on the initial or penultimate syllable. II. Non-Distinctive tonal patterns II.a. One pattern systems In these systems, there is no tonal contrast. All words have the same pitch contour. Fukui (2003) illustrates this type with two examples. In Pyeong’ando there is a phrase final high pitch, or, if the phrase is utterancefinal, a penultimate high pitch. A second type is found in many Chonnam dialects, which have high pitch either on the second syllable or on the first syllable, if the syllable starts with /s/, /h/, an aspirated or reinforced consonant, or contains a long vowel. One might say that in such systems the location of the high pitch is fully predictable and therefore cannot be contrastive. II.b. Zero-pattern systems Finally, it would seem that the Seoul dialect is not a pitch accent system at all. This does not mean that words are always ‘flat’ because intonational tones are still present. We now turn to a more detailed discussion of the Seoul system. 11. Gyeongsang ¼ Kyongsang ¼ Kyungsang; Hamgyeongdo ¼ Hamgyong.

Word accent systems in the languages of Asia

545

The Seoul dialect Several studies deal with the Seoul dialect, which is often referred to as (standard) Modern Korean. Indeed, instead of postulating a pitch-accent system, many studies assume that Modern Korean has a stress-accent system, for which several accent placement rules can be proposed (see Kim 1998, Lim 2000). (51) accent the first syllable if heavy; otherwise accent the second syllable12 a. kyo´:doso ‘prison’ b. sı´:caN ‘market’ c. na´ks«n ‘rejection’ d. iya´ki ‘story’ e.

k«u´l

‘a mirror’

(52) accent the first heavy syllable; otherwise accent the last syllable a. sa´:ram ‘people’ b. nu´nbora ‘snow storm’ c. bara´m ‘wind’ d. pihæ´Nki ‘an airplane’ e. au´ ‘a younger sibling’ f. ima´ ‘forehead’ All of the second set of examples have accent on either the first or the second syllable, which implies that the formulation ‘accent the first syllable if heavy; otherwise accent the second syllable’ can be said to hold for both sets of words. The perception study reported in Lim (2000) indicates that the third syllable in trisyllabic words (for example LLL, or LLH) is never perceived as prominent. We also find a third approach to the analysis of Korean prosody, which heavily relies on lexical accent (Ko 2010). Consider the data in (53). (53) a. b. c. d.

kı´:l-ta kil-ı´

‘be long’ ‘length’ ka´:m-to ‘persimmon-also’ ka´:m-i ‘persimmon-NOM’

12. CVV and CVC count as heavy, CV is light. Note that vowel length has disappeared in Modern Seoul Korean.

546

Rene´ Schiering and Harry van der Hulst

The analysis is based on the assumption that the initial long vowels in (53) are not underlyingly long but underlyingly accented. Accordingly, accent falls on the first syllable if underlyingly specified for this position and on the second syllable otherwise. The surface length of the vowel in the first syllable is considered a phonetic correlate of accent. This accentual analysis is supported by the fact that certain su‰xes cause a shift of the accent to the second syllable. Those su‰xes do not seem to have an overt phonological property causing the shift. In fact, in some cases there is a homophonous su‰x which does not cause the shift. In some cases, these su‰xes may cause a shift away from a closed syllable (if accented), as shown in (54). (54) a. sa´:lm-ta b. sa:lm-a [salma´]

‘boil-inf.’ ‘boil-connective’

These shifts point to an accentual analysis because they can be accounted for by assuming that the shifting su‰xes have an accent which causes the deletion of the stem-accent. In a non-accentual analysis, the shifts cannot so easily be explained. In still other accounts, word accent is denied and the impression of ‘stress’ is attributed to pitch contours which form part of the (phrasal) intonational phonology (Jun 1993, Kim 2004). In this analysis, reference is made to the fact that the location of the alleged word accent di¤ers depending on the broader prosodic context which, if true, undermines (or weakens) the idea of there being a word-level accent. These researchers suggest that the pitch patterns that were assumed to be properties of the word domain (and dependent on word accent) are properties of a larger domain, called the Accentual Phrase, a unit within the Intonational Phrase. From this perspective, the impression of stress accent on the first or second syllable is interpreted as resulting from an edge or boundary tone at the beginning of the so-called accentual phrase. However, in such an analysis it must be assumed that the alignment of the tone is sensitive to the weight of the initial syllable (Lim 2000; De Jong 1994). This is, of course, a possible analysis although it might be argued that sensitivity to syllable weight is perhaps more characteristic of word level accent placement than of phrasal tone anchoring. With the accent in place, it would then be the case that the intonational tone anchors to the accented syllable. One way or the other, the sensitivity to syllable structure has to be stated.

Word accent systems in the languages of Asia

547

Generalizations Lee and Ramsey (2000) provide a brief synopsis of Korean dialectology. In regard of tonal properties, they remark that distinctive tone/pitch and distinctive length are almost in complementary distribution, although there are some dialects that have both and some that have neither. On Cheju Island and scattered over North Korea there are such dialects. Presumably, Seoul Korean, for speakers who have lost the length opposition, also falls in this category. It would seem that the discussion on the interpretation of the Korean prosodic system is ongoing and it is, as yet, not clear whether there is a word-level prosodic system. 5.5. Japanese Genetic information For a putative Altaic subgrouping ‘Korean-Japanese’ in which Japanese is a sister to Korean see the introductory remarks in section 5.4. As stated there, we follow the Macro-Altaic hypothesis in treating Japanese as belonging to the Altaic phylum. Within Japanese, Ryukyuan and Japanese have to distinguished, the former encompassing several subbranches and dialects. Being considered a separate language of disputed origin, it exhibits Altaic traits in grammar, but also Austronesian traits in the sound system and prefixation. Like Korean, Japanese has many dialects which di¤er especially in their word prosodic systems. Representative of the Western dialects is Kyoto, of the eastern dialects Tokyo. See Shibatani (1990) for a general overview of the languages of Japan. JAPANESE RYUKYUAN: AMAMI-OKINAWAN: NORTHERN AMAMI-OKINAWAN: Southern AmamiOshima, Kikai, Northern Amami-Oshima, Toku-No-Shima SOUTHERN AMAMI-OKINAWAN: Oki-No-Erabu, Central Okinawan, Kunigami, Yoron SAKISHIMA: Miyako, Yaeyama, Yonaguni Japanese Accent information A broad variety of Japanese dialects exists which, among other aspects, di¤er in their word prosodic systems. An overarching property of all

548

Rene´ Schiering and Harry van der Hulst

systems is the relevance of pitch movements at the level of the ‘word’, or, as some researchers prefer to put it: the ‘accentual domain’. Even with reference to the Tokyo variety of Japanese, which is probably the best studied dialect, interpretations of the nature of the pitch contour vary, relating either to the characterization of the locus of pitch movement (through accents or lexically specified tone) or to the domain and/or interaction between word-level contours and intonational properties that are introduced at higher levels, such as the accentual phrase or the intonation phrase. Here we are mostly interested in the variety that results from di¤erences that relate to the contrastiveness or non-contrastiveness of tonal contours and di¤erences that involve the locus of accent (or lexically specified tone). Meanwhile the study of Japanese tonology has come a long way in particular in the context of studying the melodic structure of prosodic constituents of varying size and the interaction between word tone and intonational tones, in particular various boundary tones. An important work is Pierrehumbert and Beckman (1988). A recent discussion of Tokyo Japanese, as well as further references, can be found in Gussenhoven (2004, chapter 10). An interesting overview in the context of autosegmental theory of dialectal di¤erences is o¤ered by Haraguchi (1988), who divides Japanese dialects into two broad categories: pitch-accent systems and unaccented systems. Cross-classifying with this dichotomy, he suggests a ‘universal’ inventory of melodies (H, L, HL, LH and LHL) from which a system may pick one or two at most. Additional ways in which dialects can be di¤erent lie in the presence or absence of tone spreading rules. Thus in Tokyo Japanese, the H tone spreads leftward (leaving an initial mora low, possibly due to a boundary L tone that comes with the accentual domain; cf. below. Here we are mostly intersted in the accentual aspect of the Japanese systems, i.e. the location of the accent. The system of Tokyo Japanese is such that the constituents of words (stems, a‰xes) can be accented or unaccented (or, in the case of a‰xes, pre-accented). When more than one accent is present in the accentual domain (which can be larger than the word and therefore needs a careful definition; Gussenhoven (2004) calls it the a-domain, the first (or leftmost) accent predominates, i.e. will attract the high pitch/tone. If no accent is present, the high pitch occurs on the last (rightmost) syllable. This FIRST/LAST pattern constitutes an unbounded system. In fact, Haraguchi (1988) notes that three of the possible unbounded patterns occur in Japanese dialects.

Word accent systems in the languages of Asia

549

(55) Systems with unaccented words Systems without unaccented words First/First First/Last

Kumi Tokyo, Osaka

First Fukuoka

Last/First Last/Last

– Hirosaki

Last

–

Note that systems without unaccented words have no default. Haraguchi (1977, 1988) also recognizes unaccented systems, i.e. systems in which no word is accented. He mentions Sendai (H), Miyakonojo (LH) and Kagoshima (LH, LHL). Perhaps one might analyse systems of this type as cases in which, as one might put it, only the default can apply, such that the default accent would determine the direction of association: (56) Systems with only unaccented words First Last Haraguchi (1977, 1988), following Goldsmith (1975) marks the lexical position of the pitch-accent with a diacritic (e¤ectively ‘an accent mark’). The pitch/tone associates to this accent mark or to an edge syllable if there is no accent mark. This makes the pitch that associates to an accent representationally di¤erent from a pitch that associates to an unaccented edge syllable. (57)

H | * ssss

H | | ssss

This is a welcome result since the pitch height of finally accented words is higher than that of unaccented words (Gussenhoven 2004: 190–191). In unbounded accentual systems, the analogue to this situation would be that, whereas in words with heavy syllables, the rightmost or leftmost heavy syllable ends up with a grid column consisting of two marks, words without heavy syllables would have just one mark, cf. (58). (58)

* * | | * | ssss ssss

550

Rene´ Schiering and Harry van der Hulst

One could, in fact, argue that this representation (making no reference to tones) can be used for pitch-accent systems as well, at least in those cases in which the pitch properties are seen as phonetic interpretations of * rather than as phonological entities. We refer to a discussion of these two ways of dealing with pitch-accent systems in van der Hulst (1999). A phonological tonal analysis seems more likely for those systems in which either tone spreading takes place (that cannot be analyzed as phonetic interpolation) or more than one tonal melody is present, one of which minimally would need to be specified. Generalizations Japanese accentology displays a rich and interesting typological variation, just like Korean and, perhaps another comparable case, Basque accentology (cf. Chapter 8, section 4.2.). Many of the theoretical issues, as illustrated in our discussion of Japanese, apply to these other systems as well.

6. Southern Asia In our definition of Southern Asia, we essentially follow Asher (2007), who delimits the area with reference to Iran in the west and Bangladesh in the east. In this region, eight language families are spoken: f f f f f f f f

Turkic (see Section 5.1.) Indo-European: Indo-Iranian Dravidian Austroasiatic (see also Section 7.2.) Tibeto-Burman (see Section 7.1.) Tai-Kadai (see Section 7.4.) Andamnese Burushaski

Turkic has already been discussed in Section 5. Tibeto-Burman and TaiKadai will be surveyed in Section 7. With respect to Austroasiatic, we will discuss the Munda languages in this section, while elaborating on Mon-Khmer in Section 7. The main rationale for this decision stems from the former’s geographical location in the region delimited above.

Word accent systems in the languages of Asia

551

6.1. Indo-European: Indo-Iranian Genetic information Indo-Iranian, as a major daughter of Indo-European, consists of two large and ancient groups, namely Indo-Aryan (‘Indic’) and Iranian. Nuristani is sometimes mentioned as a third major branch of the family (Ruhlen 1991). Recent classifications, such as R. Gordon (2005), however, treat this group as a subbranch of Indo-Aryan (see Watkins 1993: 26–31 for details on the principal branches of Indo-European and Sims-Williams 2002 on the classification of Indo-Iranian languages). INDO-IRANIAN INDO-ARYAN IRANIAN UNCLASSIFIED: Badeshi, Luwati Indo-Iranian, as the largest Indo-European subfamily, broke o¤ around 2000 BCE and remained fairly homogenous until 1000 BCE. The modern Indic languages all derive from Classical Sanskrit, the Iranian languages from Avestan (6th century BCE). This family spread over Iran, Afghanistan, India, and Pakistan. Hindi and Urdu, two o‰cial languages of India and Pakistan are very similar. 6.1.1. Indo-Aryan Genetic information Most recently, at least eight subbranches of Indo-Aryan are distinguished which sometimes exhibit considerable genetic complexity. The following presentation follows R. Gordon (2005), however, without taking a stance towards its adequateness. The interested reader is referred to Masica (1991: 8–60) for exhaustive discussion. The eleven languages included in the StressTyp sample cover six subbranches of the family. INDO-ARYAN CENTRAL ZONE: BHIL: Pauri Bareli, Rathwi Bareli, Bauria, Bhili, Bhilali, Palya Bareli, Chodri, Dhodia, Dubli, Dungra Bhil, Adiwasi Garasia, Gamit, Rajbut Garasia, Mawchi, Nahali, Noiri, Pardhi, Rathawi, Wagdi DOM: Domari GUJARATI: Aer, Kachi Koli, Gujarati, Jandavra, Parkari Koli, Wadiyara Koli, Saurashtra, Vasavi, Vaghri

552

Rene´ Schiering and Harry van der Hulst

KHANDESI: Ahirani, Dhanki, Khandesi PANJABI: Eastern Panjabi Powari RAJASTHANI: MARWARI: Dhatki, Dhundari, Goaria, Godwari, Loarki, Marwari (Pakistan), Marwari (India), Merwari, Mewari, Shekhawati UNCLASSIFIED: Bagri, Gujari, Gurgula, Harauti, Lambadi, Gade Lohar, Malvi, Nimadi ROMANI: BALKAN: Balkan Romani NORTHERN: Carpathian Romani, Kalo Finnish Romani, Baltic Romani, Sinte Romani, Welsh Romani VLAX: Vlax Romani UNCLASSIFIED: Parya, Sonha, Dangaura Tharu, Kathoriya Tharu, Mewati WESTERN HINDI: BUNDELI: Bundeli HINDUSTANI: Hindi SANSI: Kabutra, Sansi Urdu UNCLASSIFIED: Haryanvi, Bhaya, Kanauji, Braj Bhasha, Chamari, Ghera, Gowli EAST CENTRAL ZONE: Awadhi, Bagheli, Dhanwar, Fijian Hindustani, Chhattisgarhi EASTERN ZONE: BENGALI-ASSEMESE: Assamese, Bengali, Bishnupriya, Chakma, Chittagonian, Hajong, Halbi, Kurmukar, Kharia Thar, Kayort, Mal Paharia, Nahari, Rajbanshi, Sylheti, Tangchangya, Mirgan BIHARI: Angika, Bhojpuri, Caribbean Hindustani, Kudmali, Magahi, Maithili, Majhi, Sadri, Oraon Sadri, Surajpuri, Musasa, Panchpargania ORIYA: Bodo Parja, Bhatri, Bhunjia, Desiya Oriya, Kupia, Oriya, Adivasi Oriya, Reli UNCLASSIFIED: Bote-Majhi, Degaru, Chitwania Tharu, Kochila Tharu, Rana Tharu, Buksa NORTHERN ZONE: CENTRAL PAHARI: Kumauni EASTERN PAHARI: Nepali, Palpa

Word accent systems in the languages of Asia

553

GARHWALI: Garhwali WESTERN PAHARI: Mahasu Pahari, Bhadrawahi, Bhattiyali, Chambeali, Churahi, Dogri, Gaddi, Hinduri, Jaunsari, Bilaspuri, Kullu Pahari, Harijan Kinnauri, Mandeali, Pangwali, PahariPotwari, Sirmauri, Kangri NORTHWESTERN ZONE: DARDIC: CHITRAL: Khowar, Kalasha KASHMIRI: Kashmiri KOHISTANI: Bateri, Chilisso, Kalami, Gowro, Indus Kohistani, Tirahi, Torwali, Wotapuri-Katarqalai, Kalkoti KUNAR: Dameli, Gawar-Bati, Grangali, Shumashti PASHAYI: Northeast Pashayi, Northwest Pashayi, Southwest Pashayi, Southeast Pashayi SHINA: Brokskat, Domaaki, Phalura, Kohistani Shina, Shina, Savi, Ushojo LAHNDA: Southern Hindko, Northern Hindko, Jakati, Mirpur Panjabi, Western Panjabi, Seraiki, Khetrani SINDHI: Jadgali, Kachchi, Lasi, Sindhi Bhil, Sindhi NURISTANI: Ashkun, Kati, Prasuni, Tregami, Waigali, Kamviri Sanskrit SINHALESE-MALDIVIAN: Maldivian, Sinhala, Veddah SOUTHERN ZONE: KONKANI: Goanese Konkani, Kukna, Katkari, Konkani, Phudagi, Samvedi, Varli Marathi UNCLASSIFIED: Bhalay, Deccan, Gowlan, Varhadi-Nagpuri UNCLASSIFIED: Chinali, Dhanwar, Darai, Kanjari, Kumhali, Lahul Lohar, Memoni, Mina, Od, Pali, Tippera, Usui, Vaagri Booli

StressTyp extracts Gujarati [U/P;P/A] In bisyllables, stress is final if the vowel is /a/, if both vowels are non-/a/ or the first syllable contains a schwa and is closed, stress is penult; if the penult is an open schwa there is free variation. f In longer words, stress the penult if it is not schwa (but schwa þ coda is stressed) or /i/ and the antepenult is not /a/. If the antepenult is /a/ f

554

Rene´ Schiering and Harry van der Hulst

there is free variation. If the penult is schwa and open, stress is antepenultimate. ¨bidi

‘a type of cigar’

nak¨san

‘damage’

kh«¨but«r ‘pigeon’

¨akr«m«n

‘invasion’

j«¨j«nti

¨tajeter or ta¨jeter ‘recently’

‘birthday’

Romani, North Russian (dialect of Romani, Baltic) [U] f Stress falls on the final syllable (with lexical exceptions). duratu¨no ‘far’

o¨mestir ‘from a gypsy’

Hindi [U%A] f Stress falls on final syllables of the form VVC or VCC. f Otherwise stress is on penultimate long vowels or closed syllables. f If the final is not superheavy, and the penult is not heavy stress shifts to a heavy antepenult (or even to the pre-antepenult). ka¨maal ‘wonders’ ¨anumati ‘permission’

in¨saaniyat ‘humanness’ ¨kamal ‘lotus’

Awadhi-Bagheli [P/P] f Stress falls on the penultimate syllable, except when the final is heavy and the penult is not, then stress is final. pi¨saøn ‘flour’ kˆ¨œisi¥ ‘he said’

¨baøis ¨saøœaø

‘twenty two’ ‘share’

Bengali [I] f Stress falls on the first syllable. ¨bissleSon

‘analysis’

¨cbostha ‘condition’

Bhojpuri [A-U/P] Primary stress falls on the antepenultimate syllable of words with 4 syllables or more. f Primary stress falls on the penultimate syllable elsewhere, except when the final vowel is checked, in which case stress is final. f Secondary stress falls on long vowels, final closed syllables and to the left of the main stress on the only syllable preceding it, or two to the left. f

˙ga¨lab ‘to melt’ ˙kHa¨tamka˙rab ‘to finish’

˙ba¨hini ˙kHaju¨aøib

‘sister’ ‘to scratch’

Word accent systems in the languages of Asia

555

Maithili (dialect of Maithili; Bihari) [P%A/P] f Primary stress falls on the penultimate syllable if it contains a long vowel. f If the penult contains a short vowel and the final is long, stress is final. f If both the penultimate and the final vowel are short, stress is antepenultimate if that vowel is long, else penultimate. f Secondary stress falls on initial syllables. ¨maøti ˙pat«¨hiø

˙adh«¨laøne ¨gaøbhinu

‘earth dir. base’ ‘thin’

‘bad’ ‘pregnant’

Nepali; Gurkhali; Gorkali [I/I] f Stress falls on the first syllable, except when the second vowel is long and the first is not. Kalami; Kalam Kohistani [U (Tone)] f Stress falls on the final syllable, except in words with a HL tone, where stress varies. Sˆr¨da ‘quickly’

¨Soøliø ‘rice in the field’ [hl tone]

Sindhi [L/P] f Stress falls on the last heavy syllable and on the penult if there are no heavy syllables. Sanskrit [F/F (pitch)] f Stress falls on the first high pitched syllable, else on the first. f Claimed to be pitch-accent system rather than stress. ¨apaciti ‘retribution’ nama¨syati ‘respects’

dhaø¨rayati ‘holds’ aparaøh¨a ‘afternoon’

Additional information Mistry (1997: 660) reports that Gujarati accent occurs on the first syllable except when the second syllable has /a/ and the first syllable has a vowel other than /a/ (cf. StressTyp above): (59) a. b. c. d.

¨sabu ¨pr«clit ¨tarikh u¨taru

‘soap’ ‘prevalent’ ‘day, date’ ‘passenger’

556

Rene´ Schiering and Harry van der Hulst

Hindi does not show the same e¤ect of /a/’s attracting stress away from the right edge: (60) a.

Hindi

Gujarati

ga¨r"#b

¨garib

b. nuk¨sa#n nuk¨sa#n

‘poor’ ‘loss’

Vijaykrishnan (p.c.) provided us with the following more detailed information concerning accent placement in Gujarati. (61)

Bisyllables: a. Stress is initial (penult), but: b. /a/ in final syllable always attracts stress (unless it is a nominative ending) c.

If schwa in open initial syllable, initial stress varies freely with non-initial stress (Schwa in closed syllable will get initial stress) Trisyllables:

d. Stress is penult, but: e. If the penult contains a schwa in open syllable, stress is initial f. If the initial syllable contains /a/ than there is free variation between initial and penultimate stress We summarize this as follows: the leftmost non-final occurrence of /a/ is stressed, otherwise the penult is stressed (if not open schwa in which case stress is leftward in trisyllables or rightward in bisyllables). With respect to Punjabi, Vijaykrishnan (p.c.) shared the following insights with us. There are three degrees of syllable weight: L(ight), H(eavy) and S(uperheavy), where superheavy syllables occur only in final or penultimate position (unless there are su‰xes). Accent has an impact on the prosodic system, such that it provides the domain for tone and that phonological and morphological rules make reference to it. The following rules predict accent placement. (62) a.

Accent the final syllable if S (inflectional su‰xes leave that stress in place) b. Stress the penultimate syllable if H and the final syllable is H or L c. Any sequence of two H or L syllables has accent on the first (both in disyllables and before a S syllable)

Word accent systems in the languages of Asia

d.

Heavy prefixes attract accent

e.

Superheavy su‰xes get accent (with secondary accent two syllables to the left)

f.

Causative forms receive accent on the causative su‰x

557

The only Indic language that is spoken in Europe is Romany, the language of the Gypsies. As a language of a nomadic people it is spoken all over Europe and the Near East. As a result of a long period of separation from the other Indic languages, Romany has developed many properties of its own, and many regional variants, although all of these are still clearly Indic in their morphology and basic vocabulary. Romany consists of three major mutually unintelligible dialects: Syrian (or Asiatic), Armenian and European (Comrie 1981). The European dialect group in turn consists of a number of subdialects. The dialect described here is a North Russian dialect. The following description is based on Ventzel (1983). Romany primary accent always falls on the final syllable in underived native stems. In derived words, accent only falls on the penultimate syllable in oblique case forms of nouns and in certain inflected verbs: rome´s ‘man, Gypsy (nom.)’, rome´ske ‘man, Gypsy (dat.)’. It would seem then that inflection is mainly accent-neutral. Antepenultimate accent only occurs on adjectives with possessive or diminutive su‰xes and abstract nouns. These all have disyllabic su‰xes which must be considered accentneutral. Other derivational su‰xes seem to be accent-sensitive. Vijaykrishnan (p.c.) gives the following information concerning the accentual system of Hindi. (63) a. b. c. d. e. f.

S syllables are always accented Penult H syllables are always accented H is accented if preceding an accented syllable LLH has antepenult accent (i.e. final H is not stressed) Disyllables without S have initial accent There are alternating accents to the left of the primary accent (not sensitive to L/H distinction)

Kaye (1997: 650) remarks that accent in Hindi-Urdu is not phonemic and that there is a considerable degree of free variation. Most words take penultimate accent as inherited from Middle Indo-Aryan. Old IndoAryan presumably had a pitch-accent which developed into a stress-accent system.

558

Rene´ Schiering and Harry van der Hulst

Baart (1997) o¤ers the following rule for the accent systems of HinduUrdu and Punjabi: If we assume that the final segment is extrametrical, then we can say that accent falls on the rightmost heavy syllable. In the default case, the penultimate syllable is accented. The causative su‰x -aa, which is always accented, would be an exception to this pattern. Vijaykrishnan (p.c.) describes Adwadhi word accent as follows: (64) a. the leftmost long vowel is accented b. the antepenultima is accented if the penult contains a long vowel or schwa c. if the initial syllable in disyllabes is closed, final long vowels can be overlooked According to Mahanta (2002), primary accent in Assamese is assigned to the second syllable, if it is heavy. Otherwise primary accent is realized on the initial syllable. Secondary accent is then placed on (i) the first of the remaining alternating light syllables, or (ii) a heavy syllable if not preceded by a stressed heavy syllable. Overall, the accentual pattern thus avoids stress clashes. Morphologically speaking, accent is insensitive. Note furthermore that only closed syllables count as heavy in the absence of distinctive vowel length. The data in (65) exemplify these patterns. [A@rAdœ EnA] [o´bœbœAbEk] [A@lusEnA] [EAdœrEn] [upo´stœpEn] [A@b ErzEnA] g. [A@NbidœAnı`k] h. [EttAlı`kA] i. [mEnurEnEn] j. [EmpErkı`tE] k. [Ennibı`stE]

(65) a. b. c. d. e. f.

l.

[EnukEmpA]

‘worship’ ‘guardian’ ‘criticism’ ‘extra-ordinary’ ‘present’ ‘garbage’ ‘constitutional’ ‘palace’ ‘entertainment’ ‘related’ ‘included’ ‘compassion’

The opinions on accent placement in present-day Bengali di¤er. Generally, default initial accent placement is assumed (see e.g. Hayes and Lahiri 1991: 55), presumably at a level higher than the word (Chatterji 1991).

Word accent systems in the languages of Asia

559

However, we also find cases where the leftmost (non-final) heavy syllable is accented, e.g. [cid«@mb«r«m] (a name). Hayes and Lahiri (1991) provide a detailed analysis of Bengali intonational phonology which also encompases a thorough description of accent assignment at various levels of prosodic structure. According to their rules, the initial syllable gets accented. Within the phonological phrase, the leftmost non-clitic word is the strongest. At the level of the intonational phrase (I), focus has an impact, such that a phonological phrase (P) bearing narrow focus will receive the strongest accent in the intonational phrase. Under neutral focus, on the other hand, the rightmost phonological phrase in the intonational phrase is strongest. The application of these rules is illustrated in the following data taken from Hayes and Lahiri (1991: 56). (66) a. x x x x x x x x xx x x xx x x x x h [[sˇæmoli]P [ram-er bari]P [d¥ ukecˇhilo] P] I ] Shamoli Ram-’s house entered ‘Shamoli entered Ram’s house.’ (neutral focus) b. x x xx [[sˇæmoli]P Shamoli

x x xx [or bari]P his/her house

x x x x x x [d¥ hukecˇhilo] P ] I ] entered

‘Shamoli entered his/her house.’ (neutral focus) c. x x xx

x x x

x x xx

x x x x x h [d¥ ukecˇhilo] P ] I ]

[[sˇæmoli]P [ram-er bari]P Shamoli Ram-’s house entered ‘Shamoli entered Ram’s house.’ (narrow focus)

560

Rene´ Schiering and Harry van der Hulst

Note that the proclitic possessive marker in (65b) shuns phrase-initial prominence. Furthermore, focus clitics in Bengali show a genuine prosodic behavior in contributing an H* tone as shown by Lahiri and FitzpatrickCole (1999). According to Wali and Omar (1997: 307), Kashmiri is a syllable-timed language and ‘stress’ does not play any role in the prosodic system. No lexical pitch distinctions or word-level pitch contours are reported. With respect to Kalami, Baart (1997) argues that it does not have pitch or stress accent, but that it is a pure tone language. The perception of accent is triggered by changes in the pitch contour from high to low or low to high. The perceived location can be di¤erent depending on the location of a word in the sentence. Tonal patterns (five of them) associate to words from right-to-left. Gilgiti Shina (Radlo¤ 1999) is described as a lexical pitch accent system. Each word has exactly one accent which is realized as high pitch. In long vowels, the accent can be on the first or the second mora. The system described in the StressTyp entries for Sanskrit is of course that of Vedic Sanskrit. Classical Sanskrit has di¤erent stress patterns. Sanskrit has a lexical accent (uda-tta) that is marked by high pitch. Lazzaroni (1993: 107) reports that since a mark indicates accent in Vedic texts, we base our knowledge of Sanskrit accent on this variety. The location of the pitch accent essentially corresponds to the location of accent in PIE. There is a present-day pronunciation of Vedic which has stress accent located within a right-edge bisyllabic window. For Marathi, Pandharipande (1997: 555–559) proposes the following, weight-sensitive accent placement rules which are equally applicable to nouns and verbs. In a word with only one heavy or super-heavy syllable, this very syllable is accented regardless of its position within the word: (67) a. ku¨t 8he b. ¨dzha#d8a c. ¨dus8t 8a d. sa¨masta e.

‘where’13 ‘trees’ ‘evil/wicked person’ ‘entire’

mha#¨ta#rpan8a ‘old ages’

In bisyllabic words, the initial syllable is accented if both syllables are either heavy (68a–c) or light (68d–g). 13. Note that /e/ and /o/ pattern with the long vowels /ı /# , /u#/, and /a#/ in terms of syllable weight.

Word accent systems in the languages of Asia

‘star’

b. c.

¨ta#ra# ¨s´a#l 8a#

‘school’ ¨mu#rt" # ‘statue’

d. e.

¨phal 8a ¨mul 8a

‘fruit (pl.)’ ‘roots’

f. g.

¨bara ¨suta

‘all right!’ ‘threads’

(68) a.

561

With respect to trisyllabic words, no such forms exist which contain only light syllables. If the first two syllables are heavy, or if all syllables are heavy, accent falls on the first syllable, cf. (69a–b) and (69c–f ), respectively. If only the second and third syllables of such words are heavy, the second syllable receives accent (69g–i). For trisyllabic words with only one heavy syllable, the first rule applies, cf. (67d–e) and gari¨bı # ‘poverty’. (69) a. b.

¨dh"#ra#na ‘courageously’ ¨bha#wa#tsa ‘brother’s’

c.

¨a#d8ha#wa#

‘estimate’

d. e. f. g. h.

¨mha#ta#ra# ¨a#dza#r" # ¨pa#wsa#l 8a# tSa¨la#kh"# sa¨ma#dh"#

i.

pu¨dZa#r"#

‘old man’ ‘sick’ ‘rainy season’ ‘cleverness’ ‘the state of transcendence from the phenomenal world’ ‘priest/worshipper’

One way to generalize over these patterns would be to formulate a rule which places accent on the leftmost heavy syllable within the word, where V: is heavier than VC (Vijaykrishnan, p.c.). Das (1973: 34) notes that accent in Malto is not phonemic; see also Steever (1998d). A syllable containing a long vowel is always stressed, e.g. qe#n ([qeøn]) ‘guiltless’. With two consecutive syllables containing a long vowel, the last one is accented. In polysyllabic words, it appears to accent the first long vowel from the right, otherwise, the initial syllable. Generalizations In the languages discusses we see many examples of weight-sensitivity, sometimes with three degrees of weight. In the majority of cases the loca-

562

Rene´ Schiering and Harry van der Hulst

tion of the accents are predictable (non-phonemic) which may be one of the causes for why di¤erent descriptions provide di¤erent rules. As Hayes (1995: 162) observes with reference to Hindi, ‘‘the published descriptions almost all disagree with one another, and seldom mention the disagreement.’’ Several languages have tonal properties. Masica (1991: 118–122) summarizes the word prosodic systems of Indo-Aryan languages as follows. There are a few languages that have contrastive tone, the best known example being Punjabi. In this language the domain of tone is said to be two syllables. There are three contrastive tones: (70) a.

ko`r2a

‘horse’

b. ko#r2a ‘whip’ c. ko´r2a ‘leper’ The so-called ‘neutral tone’ is the most common one. Historically, contrastive tone derived from the loss of aspiration. In some other languages, murmur (or breathy voice) evolved in the course of this diachronic process, e.g. in Gujarati (where murmur is accompanied by low tone) and in Marathi. Lahanda and the Western Pahari dialect have also been mentioned as having contrastive tone. Those languages that have accent are said to be syllable or mora-timed rather than stress-timed and the location of accent is generally predictable. In Assamese (Goswami 1966), however, there are minmal pairs like those in (71). (71) a.

¨pise

b. pi¨se

‘he is drinking’ ‘then’

A good example of a fixed initial pattern is Bengali where the domain is referred to as the ‘breath group’ rather than the word, cf. Chatterji (1991: 23–24). Sinhalese and Nepali are described as having ‘dominant weak initial stress’. For the remaining langauges, rather complicated rules have to be formulated in order to predict accent placement. These typically make reference to the number of syllables, whether they are open or closed, and the nature of their vowels. Baart (2003) provides a survey of 30 languages. 18 of those have tonal systems (for 5 of these there is still some uncertainty).14 12 have no tonal system (for 6 of these there is some doubt on that). The first mentioned 18 14. In Baart (2003) this number has gone up to 25.

Word accent systems in the languages of Asia

563

languages are grouped into 3 types of tonal systems (in some cases the assignment to this class is tentative): A.

B.

C.

As in Shina (see above), thus with a lexically determined accent and a high pitch associated with the accent: Burushaski, Dameli, Gawar-Bati, Khowar, Indus-Kohistani, Palula, Shina, Bateri, Chilisso, Gowro, Ushojo.15 As in Punjabi, thus with a three-way distinction (high-falling, lowrising, level). Baart proposes to specify low tones lexically in the first two and assign a default high tone which makes the level tone a high tone. If the level tone would be analyzed as low, we could analyse this type of system as Shina, with the level tone being the result of the absense of accent. As in Kalami, thus with more tonal distinctions which necessitates the specification of tonal melodies that spread over the word.

Baart says that in all tonal systems the tone features associate with the ‘stressed’ syllable, because the pitch pattern occur on or near this syllable. However, in type A and B the ‘stress’ would be where the lexical accent is, whereas in Kalami (as stated in Baart 1997), the impression of stress is caused by changes in the pitch contour. 6.1.2. Iranian Genetic information Two subbranches are generally distinguished within Iranian, namely the Eastern and Western Iranian languages. R. Gordon (2005), from which the following family tree has been adapted, treats Tangshewi as unclassifiable in terms of the two major branches (for details on the history of the Iranian languages see Schmitt 2000). IRANIAN EASTERN: NORTHEASTERN: Avestan, Osetin, Yagnobi SOUTHEASTERN: PAMIR: Munji, Sanglechi-Ishkashimi, Wakhi, Yidgha SHUGNI-YAZGULAMI: Shughni, Sarikoli, Yazgulyam 15. Baart (p.c.) observes a close similarity between Shina and Lithuanian. In a branching nucleus accent can be on either mora and both languages also have a regressive accent shift (called de Saussure’s law for Lithuanian).

564

Rene´ Schiering and Harry van der Hulst

PASHTO: Southern Pashto, Northern Pashto, Central Pashto, Waneci WESTERN: NORTHWESTERN: Khalaj BALOCHI: Southern Balochi, Western Balochi, Eastern Balochi, Bashkardi, Koroshi CASPIAN: Gilaki, Mazanderani, Shahmirzadi CENTRAL IRAN: Ashtiani, Northwestern Fars, Zoroastrian Dari, Gazi, Khunsari, Natanzi, Nayini, Parsi-Dari, Parsi, Sivandi, Soi, Vafsis KURDISH: Central Kurdish, Northern Kurdish, Laki, Southern Kurdish ORMURI-PARACHI: Ormuri, Parachi SEMNANI: Lasgerdi, Sangisari, Semnani, Sorkhei TALYSH: Alviri-Vidari, Eshtehardi, Gozarkhani, Harzani, Karingani, Koresh-e Rostam, Razajerdi, Rudbari, Shahrudi, Takestani, Talysh, Upper Taromi, Maraghei, Kho’ini, Kajali, Kabatei UNCLASSIFIED: Dezfuli ZAZA-GORANI: Bajelani, Dimli, Gurani, Kirmanjki, Shabak, Sarli SOUTHWESTERN: FARS: Southwestern Fars, Lari LURI: Bakhtiari, Northern Luri, Southern Luri, Kumzari PERSIAN: Aimaq, Bukharic, Dehwari, Darwazi, Hazaragi, Dzhidi, Western Farsi, Pahlavani, Eastern Farsi, Tajiki TAT: Judeo-Tat, Muslim Tat UNCLASSIFIED: Tangshewi StressTyp extracts Osetin; Ossete; Ossetic [I/S] f Stress falls on the first syllable if it contains a long vowel, else on the second. ¨suudzaag

‘burning’

bæ¨laas ‘a tree’

Pashto [LEX] Stress location is lexically determined.

f

¨guta ‘knot’ ¨vulambed« ‘he took a bath’

gu¨ta ‘pochard’ stomaønti¨aø ‘fatigue’

Word accent systems in the languages of Asia

565

Farsi, Western; Persian [U] f Stress falls on the final syllable. za¨nan ‘women’

sanda¨li ‘chair’

xari¨dam ‘i bought’

Dari (dialect of Persian) [U] f Primary stress falls on the final syllable. f Secondary stress is located on the initial syllable ˙zendæ¨gi ‘life’ ˙asˇpæzxa¨næ ‘kitchen’ Tajik (dialect of Persian) [U] f Stress falls on the final syllable. Additional information Concerning Avestan accentuation, we find di¤ering characterizations. Whereas Sims-Williams (1993) assumes free accent, Hale (2004: 753) describes it as having penultimate stress; see also Testen (1997a). Ossetic, an Iranian language spoken in Russia and Georgia, has accent on the initial syllable of the phrase, if this syllable contains a long vowel; otherwise the second syllable of the phrase is accented. Accentless short vowels in initial position tend to reduce or delete. The Ossetic accent system thus appears to be weight-sensitive and lacking a word-level accent rule. According to Testen (1997b: 727–729), Ossetic has two major dialects: Digor (the western dialect) and Iron (the eastern dialect on which the literary language is based). Accent placement is sensitive to vowel quality, such that certain vowels pattern together as ‘strong’ or ‘weak’. The vowel groupings for Iron and Digor are given in (72). (72)

Strong Iron aeoiu Digor a e o

Weak æç æiu

In the Iron dialect, accent falls on the first syllable, if it contains a strong vowel and otherwise on the second. Personal names are always stressed on the second syllable ustç@æ ‘women’ and cˇçzZ""@tæ ‘girls’. An Arabic loan such as saqa´t ‘shortcoming’ is accented on the second of two strong vowels. Some words synchronically have initial stress on a weak vowel because the initial syllable has been lost diachronically, cf. Iron s"@stin ‘stand (inf.)’ vs. Digor isistun. In the same vein, Digor has an indefinite article /i/ which was lost in Iron. As a consequence, we see accent location as a marker of definiteness, if the first vowel in the accentual domain is weak, for instance in kwçro´j ‘a mill’ vs. kw"@roj ‘the mill’.

566

Rene´ Schiering and Harry van der Hulst

Digor accentuation is di¤erent and less well understood. It looks like a Last/Last stem system since accent falls (a) on the rightmost strong vowel and (b) if no strong vowels are present, on the last vowel. The data in (73) illustrate this point. (73) a. raxasto´n b. fælva´ræ c. ˜urumu´q d. næ tikı´s e. fidæ´ f.

‘I bought’ ‘the year before (last)’ ‘rough’ ‘our car’ ‘father’

ær-min-cæ´˜dæ

‘play for me’

The examples in (73e–f ) also demonstrate that the weak /æ/ can only be accented in bisyllables. Elfenbein (1997c: 737–738) characterizes accent in Pashto as phonemic, which is unusual among the Iranian languages. Accent can fall on any syllable, but is mainly located on the first, penult or last syllable. He gives minimal pairs like those in (74). gora´ b. go´ra

(74) a.

‘European’ ‘look!’

c. palita´ ‘wick’ d. palı´ta ‘whore’ Accent also serves a grammatical function in distinguishing aspect and mood in verbal forms. Compare the pairs in (75a–b) and (75c–d), respectively. kxˇena#st«@l«m b. kxˇe@na#st«l«m

‘I was sitting’

c. preg de@m d. pre@gdem

‘I leave’ ‘that I leave’

(75) a.

‘I sat down’

Elfenbein further distinguishes three degrees of accent, i.e. weak, medium and strong. The following data exemplify the distribution of the di¤erent accent degrees within single words. (76) a. prewat«@l b. pre@wat«$l c. prewa@tay

‘to fall, they (m.) were falling’ (strong) ‘they (m.) fell’ (strong, weak) ‘fallen (m.sg.)’ (medium)

With respect to the domain of accent, Elfenbein states that accent rules do not apply within the word but rather within a larger, phrasal unit. Such

Word accent systems in the languages of Asia

567

units typically consist of strings of a modifier and its head, or a noun and its postposition, or an object and the following verb. These domains are also said to display secondary accent(s). This is interesting in view of Gordon (2000) who provides evidence that other unbounded systems would appear to operate in larger domains than the word. In Balochi, accent is located on the first long vowel or diphthong, or, if no such nuclei are present, on the first syllable (Elfenbein 1997a: 774). Final /i:/ is only stressed in the word marrocˇ" @ # ‘today’. It would thus appear that Balochi has a First/First unbounded system. McCarus (1997) analyzes Kurdish accent as predictable only in terms of morphological structure. Stem-final syllables are regularly accented, but some a‰xes inherently carry accent, such as, for instance, the definite marker -aka´, the plural su‰x -a´n, the comparative/superlative ending -tı´r/-trı´n., and the negative prefix. Vocative forms are accented on the first syllable. The data in (77) exemplify this morphologically conditioned accent placement. (77) a. b. c. d. e.

karaka´ kura#@n gawratı´r na#@num @ o#sta! ma#m

‘the donkey’ ‘sons’ ‘bigger’ ‘I don’t sleep’ ‘teacher’

Since accent falls on the last syllable of the stem, this e¤ectively means that inflectional su‰xes in the default case are not accented. Derivational su‰xes which create new stems do get accented, e.g. d"rezˇ ‘long’ vs. d"rezˇ-ı´ ‘length’. McCarus lists subregularities for both types of complex words. In some cases, inflectional su‰xes do get accent, and in some cases derivational su‰xes do not. Also, in verbs, certain prefixes (such as the negative prefix and the subjunctive/imperiative prefix) take primary accent, but the negative prefix for nouns and adjectives does not get accented. Both the imperative form of verbs and the vocative of nouns get initial accent. In adjectives, degree su‰xes, though inflectional, take accent. Many particles are accented on the first syllable. Schmitt (2004: 275) reports that for Old Persian the accent location is not well-known. In the development from Old to Middle Persian, final syllables disappeared. The free pitch accent of IE has left traces in Avestan and some Modern Iranian languages (such as Pashto) and was perhaps still present in around 500 BCE. Avestan and Old Persian accent is unknown (Beekes 1995: 149).

568

Rene´ Schiering and Harry van der Hulst

Windfuhr (1997: 684–685) characterizes Persian accent as non-phonemic, dynamic and predictable, with accent falling on the final syllable of the base word, or the derivational su‰x. However, there are subregularities. ‘Discourse insertives’ have initial accent, and vocatives also have initial accent (inherited from IE, also said to be present in Semitic and Turkish). The negative prefix is accented. Generalizations According to Sims-Williams (1993: 135–136) we can infer the location of accent in Old and Middle Iranian from phonological processes. Such evidence suggests that the location of accent in Avestan is free, perhaps on the same syllable as in corresponding Vedic forms. The older phases of the Iranian languages may or may not have retained pitch accent, but the middle and modern languages all have stress accent. Most of these are governed by weight-sensitive rules, while some modern eastern Iranian languages may reflect the IE free accent location. 6.2. Dravidian Genetic information The Dravidian family consists of four major, geographically delimited branches (see, e.g., Andronov 1970, Steever 1998b). R. Gordon’s (2005) family tree, however, leaves eight languages unclassified. In Comrie et al. (2003), Kolami-Parji is not grouped together with Telugu-Kui. Likewise Steever (1998b) and Krishnamurti (2003) treat both as separate groups, the former being referred to as Central Dravidian and the latter as South Central. Also, Steever groups Brahui together with Ruhlen’s (1991) Northeast group as North Dravidian. Steever (1998b: 37) characterizes the putative genetic linkage of Elamite and Dravidian, as in e.g. Ruhlen (1991), as ‘dubious’. DRAVIDIAN CENTRAL: KOLAMI-NAIKI: Northwestern Kolami, Southeastern Kolami PARJI-GADABA: Mudhili Gadaba, Pottangi Ollar Gadaba, Duruwa NORTHERN: Brahui, Kumarbhag Paharia, Kurux, Nepali Kurux, Sauria Paharia SOUTH CENTRAL: GONDI-KUI:

Word accent systems in the languages of Asia

569

GONDI: Dandami Maria, Eastern Muria, Far Western Muria, Southern Gondi, Northern Gondi, Khirwar, Maria, Western Muria, Nagarchal, Pardhan KONDA-KUI: KONDA: Konda-Dora MANDA-KUI: KUI-KUVI: Koya, Kui, Kuvi MANDA-PENGO: Manda, Pengo TELUGU: Chenchu, Manna-Dora, Savara, Telugu, Waddar SOUTHERN: TAMIL-KANNADA KANNADA: Badaga, Holiya, Kannada, Urali TAMIL-KODAGU: KODAGU: Kodagu, Kurumba, Mullu Kurumba, Alu Kurumba, Jennu Kurumba TAMIL-MALAYALAM: Mannan MALAYALAM: Aranadan, Kadar, Malayalam, Malapandaram, Malaryan, Malavedan, Paliyan, Paniya, Ravula TAMIL: Irula, Kaikadi, Muthuvan, Sholaga, Tamil, Betta Kurumba, Yerukula TODA-KOTA: Kota, Toda TULU: Bellari, Kudiya, Tulu KORAGA: Korra Koraga, Mudu Koraga UNCLASSIFIED: Ullatan UNCLASSIFIED: Allar, Bazigar, Bharia, Kamar, Kanikkaran, Kurichiya, Malankuravan, Vishavan Indus Valley civilizations Harappa and Mohenjo Daro have been suggested to be Dravidian. Uralic and Altaic a‰liations have also been proposed. The earliest documented Dravidian texts regard: Tamil (3rd century BCE), Kannada (5th century), Malayalam (9th century). Brahui lies isolated in Pakistan. StressTyp extracts Koi; Koya (dialect of Gondi) [I] f Primary stress falls on the initial syllable. f Secondary stress on syllables with a closing consonant or long vowel. ¨aaki ‘leaf ’

¨ginne ‘cup’

¨«n˙dooru ‘everyone’

570

Rene´ Schiering and Harry van der Hulst

Malayalam [I/I] f Primary stress falls on the initial syllable, except when the first vowel is short and the second is long. f Secondary stress on syllables with a long vowel. ¨kuuam

‘crowd’

¨pukavaå i

¨muta˙laaæi

‘boss’

pa¨aaæak˙kaaran ‘soldier’

‘train’

Additional information Kolami, as described by Emeneau (1961: 8–9), is our only representative of the Central Dravidian branch. It is reported that the first syllable of a word has a stress-accent of medium intensity. Within phrases, it is the first accent which is strongest. With respect to intonation, it has been impressionistically noted that the degree of accent does not impose restrictions on the realization of pitch; see also Subrahmanyam (1998). Turning to the Northern Dravidian languages, Brahui exhibits nonphonemic accent which is sensitive to vowel length, such that the first long vowel within a word gets accented. Polysyllabic words lacking long vowels receive default initial accent (Elfenbein 1998: 394, see also Elfenbein 1997b). More information is available on Kurukh accentuation (Hahn 1908: 9). In disyllabic words, accent is always initial, e.g. ¨pudda# ‘short’. Trisyllabic verbs, such as ti¨signa# ‘to open’, are usually accented on the second syllable; except for those ending in ba’ana# which are accented on the third syllable. Initially accented verbs, for instance ¨nisigna ‘to dress a wound’, are rare. Most tri- or quadrisyllabic nouns take initial accent; cf. ¨dumbari ‘fig tree’ and ¨da#rhimissi ‘beard’. Adverbs always exhibit initial stress, irrespective of their length. In compounds, the first syllable of the second member receives accent. South Central Dravidian is represented by Gondi, Konda, Koya and Telugu in our sample. Steever (1998a: 274) characterizes accent in Gondi as non-phonemic and word-initial. According to Krishnamurti and Benham (1998: 244–245), the initial syllable of polysyllabic words receives primary accent, if it contains a long vowel. If the initial syllable contains a short vowel, the second syllable receives primary accent instead. Within the word, primary or secondary accent alternates rhythmically with unstressed syllables, e.g. /ar8˙bazi¨nad/ ‘she is crying’ (see also Krishnamurti 2003: 60). In Telugu (Krishnamurti 1998; 2003: 59–60), accent is initial in words with two short syllables, or when the first syllable is long and the second

Word accent systems in the languages of Asia

571

one is short. If the second syllable is long or if both syllables are long, accent is placed on the second syllable. The data in (78) illustrate these rules of accentuation in bisyllabic words. (78) a. b.

¨gadi ¨ta#ta

‘room’ ‘grandfather’

c. d.

da¨ga# ba#¨k" #

‘deceipt’ ‘debt’

In trisyllabic words, the penultimate syllable is accented in the default case. But if the initial syllable is long, it attracts accent, cf. (79). (79) a. b. c. d. e.

¨ku#turu ‘daughter’ pa¨laka ‘slate’ ta¨passu ‘penance’ tu¨pa#k" # sa#¨ta#ni

‘gun’ ‘a weaver caste’

These rules of accent placement have also been tested in an experimental study by Lisker and Krishnamurti (1991). Finally, in the South Dravidian languages, the following patterns are observed. For Kannada, very little information about the word prosodic system is available. Whereas Sridhar (1990: 301) only recognizes accent in the realms of emphasis, Steever (1998c: 131) notes non-contrastive accent on the initial syllable of every word. Neither is accent contrastive in Kodava (Kodagu), as described by Ebert (1996: 9). Word-initial syllables carry a strong accent and word-final syllable are realized with a weaker accent. If medial syllables contain a long vowel or a vowel followed by a geminate, such syllables also carry accent. Word-medial syllables with the shape CV or CVC are unstressed and undergo vowel centralization and reduction. Tamil is often described as a language with very weak accent. Utterances are generally perceived as consisting of successions of evenly accented syllables (Asher 1982: 230). More subtle phonetic studies enable us to infer rules of word accent placement from higher level intonation patterns. An utterance comprising the words [kAøDE] ‘ear-acc.’, [t5A«Vi] ‘felt’ and [pAøt5t5A)ø] ‘he saw’ will be pronounced [¨kAøDEt5¨t5A«VippAt5t5A)ø] ‘He felt the ear’, with initial accent on the first and second word (Balasubramanian 1980: 456). Such evidence backs up brief descriptions which state that Tamil has non-phonemic accent on the first syllable of the word (Annamalai & Steever 1998: 104).

572

Rene´ Schiering and Harry van der Hulst

Ashthamurthy (2003) reports a minimal di¤erence between Malayalam and Tamil. Contrary to the StressTyp description, Malayalam retains accent on the first syllable (when the second contains a long vowel) if this syllable is closed, i.e. initial closed syllables count as heavy. In Tamil, which is claimed to have the same general pattern as Malayalam (retraction to second syllable if it contains a long vowel), initial closed syllables are not weightful and thus do not keep the accent initial. This, however, is not a commonly held position. Keane (2001) o¤ers a detailed phonetic analysis of accent correlates in Tamil. Vowel reduction data seems to support fixed initial accent rather than a quantity-sensitive system like the one that has been suggested for Malayalam. Generalizations Krishnamurti (2003: 59) concludes that ‘‘very little has been written about stress and intonation’’ in Dravidian languages.16 As a consequence, most of the accentual descriptions are tentative and invite further research. Nevertheless, initial, non-contrastive accent placement seems to be typical for the entire family (Steever 1998b: 18). In some languages, vowel length and other quantity factors have an impact on accentuation such that they attract main prominence. In the light of the overall weak phonetic correlates of accent, approaching word prosody from higher level intonational phonology seems to be a promising route for future research. 6.3. Austroasiatic: Munda Genetic information The discovery of the Austroasiatic phylum goes back to Francis Mason (1854), who observed a number of correspondences between the MonKhmer language Talaing (Mon) and the Munda language Kole (Ho). In the early 20th century, Wilhelm Schmidt pioneered in establishing the Austroasiatic family on a scientific basis and formulated a first explicit classification (Schmidt 1906). After him, Heinz-Ju¨rgen Pinnow contributed substantially to the unraveling of the genetic a‰liations within the family (Pinnow 1959, 1960, 1963). Di¿oth’s (2005) most recent classification distinguishes three major branches: Munda, Khasi-Khmuic, and KhmeroVietic/Nico-Monic. Although the inner structure of the phylum is still an open question, the genetic a‰liation of Munda and Mon-Khmer as two immediate daughters of Proto-Austroasiatic, as adopted in R. Gordon’s 16. For comparative Dravidian segmental phonology see, for instance, Emeneau (1970).

Word accent systems in the languages of Asia

573

(2005) tree given below, is generally agreed on (see Sidwell 2006 for the history of Austroasiatic studies). For the purposes of this chapter and following its geographically oriented organization, we discuss the word accent systems of the Munda languages spoken in India in this section. The accentual data from MonKhmer languages will be detailed in the section on Mainland East and South East Asian languages in 7.2. Within Munda, a northern and a southern branch are distinguished (see Anderson 2001 on the classification of South Munda and Anderson 2008 for an overview of the language family). While StressTyp contains Mundari as the only representative for the whole family, we complement the survey with additional data from selected north and south Munda languages. AUSTROASIATIC MON-KHMER (see Section 7.2.) MUNDA: NORTH MUNDA: KHERWARI: Agariya, Bijori, Koraku MUNDARI: Asuri, Birhor, Koda, Ho, Korwa, Mundari SANTALI: Mahali, Santali, Turi KORKU: Korku SOUTH MUNDA: KHARIA-JUANG: Juang, Kharia KORAPUT MUNDA: GUTOB-REMO-GETA’: GETA’: Gata’ GUTOB-REMO: Bondo, Boda Gadaba SORA-JURAY-GORUM: GORUM: Parenga SORA-JURAY: Juray, Lodhi, Sora StressTyp extracts Bhumij; Mundari [U/P] If the final syllable is closed, it is accented. f Else the penultimate syllable is accented. f

¨lija ‘a cloth’ Additional information In the North Munda language Santali, word prosody is based on trochaic footing, such that the initial syllable of a bisyllabic word gets accented.

574

Rene´ Schiering and Harry van der Hulst

If the second syllable of such a word, however, is bimoraic, this syllable will attract the accent (Neukom 2001: 8). The data in (80) illustrate this pattern. (80) a. ¨dare b. ¨era c. d. e. f. g.

se¨dae u¨p«i bc¨tcr o¨rak’ ¨edre

h. ¨ult«u i. ¨dalpaN

‘tree’ ‘wife’ ‘old times’ ‘measure’ ‘fear’ ‘house’ ‘anger’ ‘reverse’ ‘half-naked’

In the South Munda language Kharia, the phonological word consists of at least one prosodic foot and potentially a varying number of ‘extrametrical’ enclitics. Prosodic feet, which are usually bisyllabic, less commonly monosyllabic and rarely polysyllabic, begin with a low-tone pitch. Throughout the remainder of the foot, the pitch level gradually rises. Accordingly, the word rocho /b ‘side’, like all lexical elements in the language, will be pronounced with a low-high pitch pattern. In monosyllabic feet, such as, for instance, laN ‘tongue’, this pitch pattern is condensed into a rising contour. At the sentential level, Kharia speech is characterized by a gradual decrease of intensity and pitch over the utterance, i.e. falling phrasal prosody (Peterson 2006: 18–33). In contrast to Peterson, Rehberg’s (2003: 23–28) analysis explicitly makes reference to word accent. This accent is always, i.e. irrespective of the morphological composition of the form, assigned to the first syllable of a word. It is realized by a lower pitch, whereas the following unaccented syllables within the word have higher pitch, cf. (81). (81) a. b. c. d. e. f. g. h.

¨kerkea (kerketta) ¨cerocagordi (on.all.four.sides) ¨soub-se (all-ABL) ¨umbo}ig-mae (NEG.NPT.COP-3P) ¨col-ki-Ø (go-A.PT-S) ¨tar-ol-e-pe (kill-bring-B.IRR-2P) ¨tama ¨ił (now 1S) ¨kag-kom-ki ¨dhog-ke (bow-arrow-P grab-SEQ)

Word accent systems in the languages of Asia

575

Her phonetic analyses also help to understand how previous descriptions of Kharia accentuation, such as Pinnow’s (1959), who diagnoses penultimate and final accent in some words, are motivated. In Kharia, low pitch is the main phonetic correlate of accent. Intensity may increase concomitant with the raising of pitch in unaccented syllables. Due to a bias towards European accent systems, some researchers misinterpreted the correlation of higher intensity and pitch in unaccented syllables as a realization of stress-accent. According to Donegan (1993: 5–6), default accent placement is wordinitial in Sora, another South Munda language. This pattern is most evident in the disyllabic words with balanced light or heavy syllables given in (82). Note that expanding the word by su‰xation does not alter accent assignment. ¨bagu ¨yagi

‘two’ ‘three’

c.

¨unji

‘four’

d. e. f.

¨mcnlcy ‘five’ ¨tudru ‘six’ ¨gulji ‘seven’

g. h. i. j. k.

¨tamji ¨tinji ¨g«lji ¨g«lmuj ¨mçjg«l

(82) a. b.

‘eight’ ‘nine’ ‘ten’ ‘eleven’ ‘twelve’

However, in words with light initial syllables and heavy second syllables, accent gets attracted to the second syllable of the word. This situation is evidenced in monomorphemic words (83a) as well as morphologically complex words (83b–d). (83) a. b. c. d.

«¨bcy «-¨y«N-«n «r-¨ed-«n j-«¨r-om-«n

‘one’ ‘his/her mother’ ‘scratching instrument’ (ed- ‘scratch’) ‘eating instrument’ ( jom- ‘eat’)

Generalizations Across the family, initial accent seems to be a recurring pattern in Munda word prosody. In some languages, namely Mundari, Santali and Sora,

576

Rene´ Schiering and Harry van der Hulst

accent placement is also weight-sensitive such that a heavy second syllable within a word attracts accent. Based on such observations, Donegan & Stampe (1983) propose a holistic typology which aims at explaining basic structural properties of the Munda languages with appeal to their rhythmic organization. In their reasoning, trochaic word accent and falling phrase accent can be held responsible for, e.g., agglutinative, su‰xing morphology and SOV word order. The Mon-Khmer languages, which are presented as showing opposing typological features, for instance, iambic word accent, rising phrase accent, prefixing morphology and SVO word order, are said to preserve the Proto-Austroasiatic profile. In this scenario, the diachronic development of Munda constitutes a major typological drift towards trochaic prosody. 6.4. Andamanese Genetic information The Andaman Islands constitute of cluster of approximately 250 islands in the Bay of Bengal, stretching from north to south and located southeast of the Indian sub-continent. Administrationally, they are part of the union territory of the Andaman and Nicorbar Islands and belong to India. The Andamanese languages consist of two groups:17 the Great Andamanese group, comprising ten languages, and the South Andamanese languages Jarawa, Onge and Sentinel. At present, the majority of these 13 languages are highly endangered (see Abbi 2006 for details). ANDAMANESE GREAT ANDAMANESE: CENTRAL: Aka-Bea, Aka-Bale, Aka-Kede, Aka-Kol, Aka-Pucikwar, Oko-Juwoi NORTHERN: Aka-Cari, Aka-Kora, Aka-Jeru, Aka-Bo ¨ nge, Sentinel SOUTH ANDAMANESE: Jarawa, O Accent information Manoharan (1989: 30) notes that in Andamese, the syllable preceding the final syllable is accented to di¤erentiate the meaning of a statement type

17. Abbi (2006: 7) distinguishes between an Eastern (¼ Great Andamanese) branch and a Western (¼ South Andamanese) branch. She assumes a flat structure for the former, but further di¤erentiates a Central Western and a Southern Western subgroup for the latter.

Word accent systems in the languages of Asia

577

from a question or doubtfullness. Accentuation is thus conceived of as operating at the sentence level only. (84) a. b.

axolaøliyo

‘he was there’

axolaø¨liyo ‘was he there?’

However, the available phonological descriptions are still too premature to generalize the word prosodic systems of the languages in question. 6.5. Burushaski Genetic information Burushaski is a language isolate spoken in the Hunza-Nagar and Yasin area of the Gilgit district in Northern Pakistan. Next to the Yasin (Werchikwar) dialect, we have the standard Hunza dialect and the Nagar (Nagir) dialect. StressTyp extracts Burushaski, Yasin (dialect of Burushaski) [LEX] f Stress is lexically determined, and plays a role in the grammar. di¨cilikini ‘he hung’ ¨dicilikini

‘he would hang’

ga¨li

‘it broke’

‘he went’

¨gali

Additional information Anderson’s (1997) description of word accent refers to the Hunza dialect. He notes that long vowels may receive accent on the first or second mora, giving rise to a falling or rising pitch: ´ıi ‘himself ’, iı´ ‘his son’, o´omaltaras ‘to envelop them’, oo´maltaras ‘to not envelop’. In addition, accent on the first mora may have a low pitch, a pattern that seems to occur with diminutives: t8ak ‘attached’, t8a`ak ‘somewhat attached’. This moraic stress di¤erence is said to be less pronounced in the Yasin dialect. Long vowels only occur in accented syllables. In some prefixed forms, accent is root initial (a-su´sun ‘my elbow’), but prefixes may attract the accent in other cases (a´-lpur ‘my eyebrow’). Perhaps this means that su´sun is lexically accented, while lpur is not, with accent falling on the first syllable by default. A bisyllabic sequence of a heavy syllable (CV:, CVC) followed by a light syllable is generally initially accented (hu´nzE ‘arrows’). Words with the reverse weight pattern, frequently have final accent (haGu´r ‘horse’), but exception occur (ha´mal ‘neighbour’). There are also minimal pairs: d8ud8u´r ‘apricot species’ – d8u´d8ur ‘small hole’.

578

Rene´ Schiering and Harry van der Hulst

Baart (1997: 40–41) refers to the language as tonal, possibly a pitch accent language. For further discussion of accentuation in the various Burushaski dialects the interested reader is refererred to Berger (1974, 1998).

7. Mainland East and South East Asia Our delimitation of this area is based on Bradley’s (2007) definition of East and South East Asia. The region roughly stretches from the Himalayas in the West to the Pacific Ocean in the East, bordering Mongolia in the North and Malaysia in the South. In this area the following languages are spoken: f f f f f f f f f

Sino-Tibetan Austroasiatic: Mon-Khmer Austro-Thai: Tai-Kadai, Austronesian and Miao-Yao Japanese (see Section 5.5.) Korean (see Section 5.4.) Ainu (see Section 4.2.4.) Turkic (see Section 5.1.) Manchu-Tungus (see Section 5.3.) Mongolian (see Section 5.2.)

With most of these language families and language isolates already discussed in previous sections, this section will concentrate on the Sino-Tibetan languages, encompassing Sinitic and Tibeto-Burman, the Austroasiatic family Mon-Khmer, Tai-Kadai, and Miao-Yao. The area discussed in this section is often thought of as mostly containing tonal languages, the intended implication being that we do not find accentual patterns in these languages. However, as has been argued in Chapter 1 the presence of lexical tone is not incompatible with stress or accent. The tonal aspects of the languages in this area are covered in many specialized books and articles. For an accessible recent discussion we refer to Yip (2002), and the references therein, who discusses several examples of tonal systems in Chinese languages (Cantonese, Mandarin, Wu and Min), in Tibeto-Burman languages (Lhasa Tibetan, Jingpho, Burmese, Bai), in Austro-Thai languages (Standard Thai, Wuming Zhuang) and in MonKhmer languages (Vietnamese).

Word accent systems in the languages of Asia

579

7.1. Sino-Tibetan Genetic information The Sino-Tibetan phylum consists of two major branches. First, the Sinitic (or Chinese) family includes all the dialects/languages of Chinese. The second family, Tibeto-Burman, is considerably larger than the first and shows a complex inner structure with various subbranches. Whereas there is agreement about the classification of Sino-Tibetan languages at this level of resolution, the details of the actual subgroupings are still being debated (see Thurgood 2003 for details). SINO-TIBETAN SINITIC (or CHINESE) TIBETO-BURMAN 7.1.1. Chinese Genetic information R. Gordon (2005), as shown below, assumes a flat structure for the 14 varieties of Chinese of the Chinese language family. Ruhlen (1991), on the other hand, distinguishes two subbranches within Sinitic: Bai and Chinese, with the latter consisting of a Min and a Mandarin-Yue group. Bai is considered a daughter of Tibeto-Burman in the Ethnologue classification adopted here. CHINESE: Min Dong Chinese, Jinyu Chinese, Mandarin Chinese, Pu-Xian Chinese, Huizhou Chinese, Min Zhong Chinese, Dungan, Gan Chinese, Hakka Chinese, Xiang Chinese, Min Bei Chinese, Min Nan Chinese, Wu Chinese, Yue Chinese StressTyp extracts Chinese, Mandarin [LEX] Syllables with tone may also carry stress, but do not have to. f Bisyllabic words with two tones have either initial primary stress or initial secondary stress and final primary stress. f Longer forms might not be true words, and have diverse stress patterns. f Status of ‘‘stress’’ in Chinese highly debatable. f

¨bo#li ‘glass’ ˙ha´n˙shu´¨biaˇo ‘thermomenter’

˙maˇ¨da´ ‘motor’ ¨wu#zili ‘in the room’

580

Rene´ Schiering and Harry van der Hulst

Additional information The Chinese languages are tonal languages. Standard Chinese (Peking dialect) has four tones (high level, high rising, low, high falling). The language also has accent and the full range of tonal contrasts only surfaces on accented syllables. Unaccented syllables show tonal reduction, lack of contrast and a pitch that is determined by surrounding tones and intonation (Ramsey 1987: 46–47). The aforementioned interaction of tone and accent can be further illustrated using data from Mandarin Chinese. Following Chao (1968: 25f.), the four tones of the language can be described on a scale of five pitch levels: the 1st tone is characterized by a high-level 55 pitch, the 2nd tone is realized by a high-rising 35 contour, the 3rd dipping/falling-rising tone comprises a 214 pitch contour, the 4th high-falling tone, finally, comes with a 51 pitch contour. The phonemic nature of these pitch di¤erences is illustrated in the minimal pairs in (85) taken from Li & Thompson (1981: 8). (85) a.

y"# (tone 1: high-level 55) ‘clothes’ b. yı´ (tone 2: high-rising 35) ‘to suspect’ c. y"( (tone 3: dipping/falling-rising 214) ‘chair’ d. yı` (tone 4: high-falling 51) ‘meaning’

As soon as syllables are combined, a number of tonal sandhi rules apply. For instance, when a tone 3 syllable is followed by a syllable with any tone other than tone 3, it changes to a low tone with the pitch contour 21. More relevant for the issue of accentuation is the so-called fourth tonal sandhi rule of neutral tone. When not contrastive or weakly accented, all syllables surface with normal accent. If a syllable has weak accent or is unaccented, however, the tonal contrasts are reduced and the syllable will receive a tone according to the following pattern: after a tone 1 syllable, unaccented syllables will surface as a half-low tone, after a tone 2 syllable as a middle tone, after a tone 3 syllable as a half-high tone, and after a tone 4 syllable they will be realized as a low tone. Accordingly, the genitive su‰x de, which is unaccented and appears in the neutral tone, will be realized with a half-low tone when it follows the tone 1 pronoun ta in the a‰xed word ta-de ‘he-Gen ¼ his’ (cf. Chao 1968: 26¤., 35f.; Li & Thompson 1981: 8f.; and Lin 2001: 48¤. for discussion). Duanmu (2000) discusses accent in Standard Chinese more extensively. He dismisses the views that SC either has no accent at all, or final accent. The latter idea is, he argues, due to a (cross-linguistically quite general) final lengthening e¤ect when words are uttered in isolation. Duanmu pro-

Word accent systems in the languages of Asia

581

poses that, at the word level, syllables are groups of trochaic syllabic feet, from left to right with the leftmost foot being the strongest. Compounds and phrases receive the strongest stress on the ‘non-head’. Foot formation is sensitive to the di¤erence between full and reduced syllables; the latter do not have distinctive tone and are monomoraic. Full syllables have distinctive tone and are bimoraic.18 Reduced syllables are always weak members of a foot, or unfooted. Full syllables are strong or weak depending on their position in the foot. In polysyllabic words, then, accent placement is initial. In trisyllabic words, a secondary accent occurs on the third syllable. In quadrisyllabic words, this secondary accent occurs on the third or fourth syllable, depending on the phrasal context. The di¤erence in accent between full and reduced syllables is easier to perceive than that between full syllables, because the latter carry tone whether they are accented or not. Positions that lack accent according to this analysis trigger processes of tone neutralization. Duanmu presents several further arguments to support his analysis. This analysis of accent suggests that the Chinese lexicon is not dominated by monosyllabic words. Duanmu does include compounds in his analysis (which, he argues, are highly frequent in Chinese running text or language usage, good for 80% of all words) as well as polysyllabic loanwords (foreign names, etc.). With these words included, most words in Chinese are disyllabic or longer. Monosyllabic words often have disyllabic variants, where the choice of one or the other depends on the broader prosodic context. See also Yip (1980, 2002) for an analysis of the tonal system of various Chinese languages. 7.1.2. Tibeto-Burman Genetic information For the Tibeto-Burman branch of Sino-Tibetan, many di¤erent classifications for the languages of this family are available. Comrie et al. (2003) group Karen tentatively under Tibeto-Burman. Within Tibeto-Burman, Benedict (1972) has eight groups: Tibeto-Kanauri, Bahing-Vayu, Abor-Miri-

18. This distinction can be compared to the di¤erence between full and reduced vowels in English, a distinction that could be taken as lexical, as proposed in Bolinger (1981), in which stress assignment would be sensitive to this distinction.

582

Rene´ Schiering and Harry van der Hulst

Dafla, Kachin, Burmese-Lolo, Bodo-Garo, Kuki-Naga, Karen. Thurgood (2003) distinguishes Lolo-Burmese, Bodic, the ‘Sal’ languages, Kuki-ChinNaga, Rung and Karenic, leaving the issue of several other small subgroups and unsubgrouped languages unresolved. The Ethnologue classification, which has been adopted in what follows, posits 14 subgroups (R. Gordon 2005). TIBETO-BURMAN BAI: Central Bai, Northern Bai, Southern Bai HIMALAYISH: MAHAKIRANTI: KHAM-MAGAR-CHEPANG-SUNWARI: CHEPANG: Bujhyal, Chepang, Kusunda, Wayu KHAM: Gamale Kham, Eastern Parbate, Sheshi Kham, Western Parbate MAGAR: Eastern Magar, Western Magar, Raji SUNWARI: Bahing, Sunwar KIRANTI: Tomyang EASTERN: Athpariya, Bantawa, Belhariya, Chhintange, Chhulung, Chukwa, Eastern Meohang, Kulung, Northern Lorung, Limbu, Lambichhong, Southern Lorung, Lumba-Yakkha, Moinba, Nachering, Pongyong, Phangduwali, Puma, Dungmali, Camling, Western Meohang, Saam, Sampang, Waling, Yakha, Yamphu, Yamphe WESTERN: Dumi, Jerung, Koi, Khaling, Lingkhim, Raute, Thulung, Tilung, Wambule NEWARI: Newar TIBETO-KANAURI: LEPCHA: Lepcha TIBETIC: BODISH: TSHANGLA: Tshangla DHIMAL: Dhimal, Toto TAMANGIC: Chantyal, Eastern Gurung, Southern Ghale, Northern Ghale, Kutang Ghale, Western Gurung, Manangba, Nar Phu, Seke, Eastern Tamang, Western Tamang, Eastern Gorkha Tamang, Thakali, Northwestern Tamang, Southwestern Tamang

Word accent systems in the languages of Asia

583

TIBETAN: Gongduk, Lhokpu CENTRAL: Atuence, Central Tibetan, Dolpo, Humla, Jad, Kyerung, Nubri, Lhomi, Lowa, Mugom, Walungge, Panang, Stod Bhoti, Helambu Sherpa, Spiti Bhoti, Kagate, Tichurong, Tseku, Tsum EASTERN: Dakpakha, Bumthangkha, Nyenkha, Nupbikha, Olekha, Chalikha, Khengkha, Kurtokha NORTHERN: Amdo Tibetan, Choni, Khams Tibetan SOUTHERN: Adap, Brokkat, Chocangacakha, Dzongkha, Groma, Jirel, Lakha, Lunanakha, Layakha, Brokpake, Sikkimese, Sherpa UNCLASSIFIED: Naaba, Sherdukpen, Thudam WESTERN: Balti, Purik, Zangskari LADAKHI: Changthang, Ladakhi, Takpa UNCLASSIFIED: Dzalakha WESTERN HIMALAYISH: Rongpo ALMORA: Byangsi, Chaudangsi, Darmiya, Rangkas EASTERN: Baraamu, Thangmi JANGGALI: Rawat KANAURI: Gahri, Chitkuli Kinnauri, Jangshung, Kinnauri, Kaike, Pattani, Tinani, Bhoti Kinnauri, Shumcho, Sunam, Tukpa, Kanashi UNCLASSIFIED: Baima JINGPHO-KONYAK-BODO: JINGPHO-LUISH: JINGPHO: Jingpho, Singpho, Taman LUISH: Kado KONYAK-BODO-GARO: BODO-GARO: BODO: Bodo, Deori, Dimasa, Tiwa, Riang, Kok Borok, Kachari GARO: Garo?, Megam KOCH: A’tong, Koch, Rabha, Ruga KONYAK: Chang Naga, Konyak Naga, Nocte Naga, Khiamniungan Naga, Wancho Naga, Phom Naga, Tase Naga, Tutsa Naga KAREN: PA’O: Pa’o Karen PWO: Eastern Pwo Karen, Phrae Pwo Karen, Western Pwo Karen, Northern Pwo Karen SGAW-BGHAI: BGHAI:

584

Rene´ Schiering and Harry van der Hulst

EASTERN: Lahta Karen, Kayan UNCLASSIFIED: Bwe Karen, Geko Karen WESTERN: Geba Karen BREK: Brek Karen KAYAH: Eastern Kayah, Yinbaw Karen, Yintale Karen, Manumanaw Karen, Western Kayah SGAW: Paku Karen, S’gaw Karen, Wewaw UNCLASSIFIED: Zayein Karen KUKI-CHIN-NAGA: KUKI-CHIN: CENTRAL: Bawm Chin, Haka Chin, Ngawn Chin, Zotung Chin, Darlong, Hmar, Mizo, Pankhu, Senthang Chin, Tawr Chin NORTHERN: Aimol, Anal, Biete, Chiru, Siyin Chin, Tedim Chin, Falam Chin, Gangte, Hrangkhol, Kharam Naga, Kom, Lamkang, Chothe Naga, Monsang Naga, Moyon Naga, Paite Chin, Purum, Purum Naga, Ralte, Sakechep, Simte, Thado Chin, Tarao Naga, Vaiphei, Yos, Zome SOUTHERN: Mro Chin, Daai Chin, Nga La, Mara Chin, Mu¨n Chin, Welaung, Zyphe KHUMI: Khumi Awa Chin, Khumi Chin SHO: Bualkhaw Chin, Chinbon Chin, Asho Chin, Shendu NAGA: ANGAMI-POCHURI: Mao Naga, Angami Naga, Khezha Naga, Northern Rengma Naga, Pochuri Naga, Southern Rengma Naga, Chokri Naga, Sumi Naga, Poumei Naga AO: Lotha Naga, Ao Naga, Sangtam Naga, Yimchungru Naga TANGKHUL: Khoibu Naga, Tangkhul Naga, Maring Naga UNCLASSIFIED: Puimei Naga ZEME: Rongmei Naga, Liangmai Naga, Koireng, Inpui Naga, Thangal Naga, Maram Naga, Mzieme Naga, Zeme Naga LOLO-BURMESE: BURMISH: NORTHERN: Achang, Zaiwa, Pela, Hpon, Lashi, Maru SOUTHERN: Chaungtha, Intha, Arakanese, Burmese, Taungyo, Tavoyan, Yangbye UNCLASSIFIED: Xiandao LOLOISH: Ache Yi, Poluo Yi, Limi Yi, Mili Yi, Muji Yi, Pula Yi, Puwa Yi

Word accent systems in the languages of Asia

585

NORTHERN: Samei LISU: Lisu, Lipo YI: Sichuan Yi, Laghuu, Southern Yi, Mantsi, Guizhou Yi, Southeastern Lolo Yi CENTRAL YI: Central Yi, Dayao Yi, Miqie Yi, Southern Lolopho Yi EASTERN YI: Naluo Yi, Wumeng Yi, Wuding-Luquan Yi, Wusa Yi SOUTHEASTERN YI: Awu Yi, Axi Yi, Azhe Yi, Sani Yi SOUTHERN YI: Eshan-Xinping Yi, Yuanjiang-Mojiang Yi WESTERN YI: Xishan Lalu Yi, Eastern Lalu Yi, Western Lalu Yi, Western Yi SOUTHERN: Youle Jinou, Buyuan Jinou, Ugong AKHA: Mahei, Phana’ HANI: Sansu, Sila BI-KA: Biyo, Kaduo HAO-BAI: Honi HA-YA: Akha, Hani LAHU: Lahu Shi, Lahu PHUNOI: Bisu, Coˆoˆng, Mpi, Phunoi, Pyen UNCLASSIFIED: Laopang, Lopi, Nusu, Zauzou NAXI: Naxi UNCLASSIFIED: Phula MEITEI: Meitei MIKIR: Amri, Karbi MRU: Mru NORTH ASSAM: DENG: Darang Deng, Geman Deng TANI: Adi, Galo Adi, Apatani, Bugun, Idu-Mishmi, Nisi, DigaroMishmi, Miri, Miju-Mishmi, Na, Sulung NUNGISH: Drung, Lama, Norra, Nung, Rawang TANGUT-QIANG: QIANGIC: Northern Qiang, Ersu, Guiqiong, Muya, Namuyi, Northern Pumi, Southern Pumi, Queyu, Southern Qiang, Shixing, Zhaba RGYARONG: Horpa, Shangzhai, Guanyinqiao, Jiarong TUJIA: Northern Tujia, Southern Tujia UNCLASSIFIED: Anu, Ayi, Hruso, Khamba, Lui, Palu, Pao, Sajalong, Zakhring WEST BODISH: Dura

586

Rene´ Schiering and Harry van der Hulst

StressTyp extracts Chepang [I] f Stress is initial. ¨chaN

‘shelf ’

¨jiksa

¨sipru ‘snake’

‘to be sick’

Tibetan; Lhasa [F/F] Stress the leftmost syllable with a long vowel, else the leftmost. f Alternative pattern of two equal stresses in bisyllables with two long vowels reported. f

qhap¨tee` ¨Naamo

¨qo¨la ¨k«p

‘rival’ ‘sweet’

‘charcoal’ ‘behind’

Bawm [U] f Southern Bawm has tone. Northern Bawm has final stress. nu¨pi

‘wife’

nu¨fen ‘skirt’

fa¨nu

‘daughter’

Additional information The Mahakiranti language Kham (Watters 2002) is a fully tonal language with no reference to stress or accent. It has a four tone system, divided over two registers that originate from phonation di¤erences in consonants. The tonal distinction in each register predates the register distinction and may itself originate from an accent–no accent opposition. Apart from a few lexical exceptions, word accent is initial in the Eastern Kiranti language Belhare (Bickel 2003: 547). Secondary accent is distributed following a trochaic rhythm of bimoraic feet, with final open syllables always una¤ected. Unaccented open syllables in non-final position are subject to phonetic reduction. (86) a. ¨phagi˙det˙lem ‘butterfly’ b. u-¨hop-chi (3sgPOSS-calebash-nsg) c. ¨u-hop (3sgPOSS-calebash) In (86b), the phonological word, which provides the domain for accent placement, starts at the left stem edge, i.e. the prefix is excluded from the prosodic word. However, if stem-initial accent would result in word-final stress, as in (86c), the prefix is accented to ensure the trochaic rhythm of the language (see also Bickel 1998 for an OT analysis). Chintang words, as recently analyzed by Bickel et al. (2007), contain

Word accent systems in the languages of Asia

587

one primary accent which regularly falls on the last syllable of the lexical stem. The data in (87), taken from Hildebrandt (2007) and Schiering et al. (2007), summarize the word accent system of Limbu (van Driem 1987). (87) a. b.

/ku-la:p/ [¨kula:p] (3Poss-wing) ‘its wing’ /pe:g-i/ [¨pe:gi] (go-1pS) ‘We go.’

c.

/a-oN-e:/ [?a¨?oN˙Ne:] (1Poss-brother.in.law-Voc) ‘My brother in law!’ d. /ku-taN¼mE/ [ku¨taNmE] (3Poss-horn¼Ctr) ‘its horn, on the contrary’ e. /mE-thaN-e¼aN/ [mE¨t h aN˙jaN] (3ns-come.up-Pst¼and) ‘they come up and . . .’ In bisyllabic prefix-stem combinations, such as (87a–b), the prefix attracts accent yielding of trochaic accent foot. The longer forms in (87c–e) show that, parallel to the Belhare pattern described above, the default locus of the word accent is the left edge of the stem, i.e. prefixes are usually excluded from the accent domain. Furthermore, secondary accent is assigned within the word following a trochaic, weight-sensitive rhythm in (87c) and (87e). Turning to the Newari branch of Mahakiranti, the Dolakha# dialect of Newa#r has accent but no tone. Word accent is assigned to the first syllable of the root, with a secondary accent on the third syllable in longer words (Genetti 2003: 357). Within Tamangic, Chantyal is the only language which is not tonal. With respect to word accent, native vocabulary is accented on the first syllable. Napali borrowings retain accentuation governed by the Nepali accent rules (Noonan 2003a: 317). In Nar-Phu, words are accented on the first syllable of the root. Compounds carry accent on their first member (Noonan 2003b: 339). Although tone is the major suprasegmental feature of Lahu (Matiso¤ 1973: 34–35), a number of word forms exhibit accent asymmetries, such that the initial syllable of a word appears to be unaccented. First, in a few recent loanwords from Burmese, Shan or Thai which have a prefix with unaccented schwa, the initial syllable is also unaccented in Lahu, e.g. a¨khwa`n ‘permission’, ka¨na´n ‘number’ and ma¨pa#w ‘coconut’. Secondly, the native noun prefix /$-/ typically has reduced accent, cf. $-¨thıˆ/ ‘a packet’. Otherwise, all syllables receive the same degree of accent.

588

Rene´ Schiering and Harry van der Hulst

The Nungish language Dulong (Drung) exhibits a typical sesquisyllabic word structure, such that the first syllable of the word is unaccented, toneless and of the shape CV, whereas the second syllable is accented, carries tone and allows more phonotactic structures, e.g. d«&ı # ‘a kind of pheasant’ (LaPolla 2003b: 674). With respect to the Qiangic languages, Northern Qiang (LaPolla 2003a: 574) has generally trochaic word accent. In bisyllabic words, the phonetic reduction of the final, unaccented syllable leads to the loss of the final vowels, e.g. /s«-t h«/ [s«t h] ‘drink!’. Niuwozi Prinmi (Southern Pumi) exhibits a prosodic system in which various aspects of suprasegmental structure interact. First, three tones are distinguished in monosyllables: bjE)H ‘busy’ (high), bjE)F ‘urine’ (falling), bjE)R ‘to fly’ (rising). The contour tones can be conceived of as being composed of a sequence of high-low and low-high, respectively. In longer words, the surface realization of tone depends on the location of the high tone within the domain and on whether the high tone spreads or not. For example, in a quadrisyllabic word, the high tone may be located on the second syllable. If this tone spreads, we will find the surface tone sequence L-H-H-L; if not, we will find L-H-L-L. Ding (2003: 590–591) thus characterizes Prinmi as a pitch-accent language. According to Sun (2003: 491), Caodeng rGyalrong can be characterized as a pitch-accent language in which a pitch drop (H-L) within the phonological word is distinctive at the lexical and morphosyntactic level. In the default, the pitch-accent is located at stem-final position. The minimal pair in (88) illustrates how di¤erences in the accentuation of monosyllabic words become transparent in the course of morphological processes. vs. vs.

¨

H-XserL

a a

(88) a. (¨)XserH ‘gold’ ‘silver’ b. rNulH

L-rNulH

‘my gold’ ‘my silver’

The monosyllabic word Xser H ‘gold’ in (88a) is inherently specified for pitch-accent (marked by (¨) in the example given). The H-L pitch drop surfaces only when another morphological element is prefixed to this form. Note that the unaccented form in (88b) never surfaces with the pitch drop, irrespective of whether it is prefixed or not. Generalizations In Sino-Tibetan, we find a range of prosodic systems ranging from nontonal accent languages to fully tonal, accentless languages, with mixed accent/tone languages somewhere in-between. Again, the presence of lexical tone does not preclude an accentual structure. Unfortunately, since the

Word accent systems in the languages of Asia

589

phonological study of these languages traditionally focuses on segmental inventories, syllable phonology and tone, the complex interactions of accent and tone are still to be investigated for most languages. 7.2. Austroasiatic: Mon-Khmer Genetic information With respect to the language families to be discussed in the next three sections, a number of classifications and a‰liations have been proposed. Ruhlen (1991), for instance, includes Austroasiatic, Miao-Yao and AustroThai in her ‘Austric’ phylum. Both Benedict (1990) and Yip (2002) place Miao-Yao within the Austro-Tai group. Benedict also proposes that Japanese-Ryukyan is a sister to Austronesian. In this line of thought, he suggests that an accentual system might be attributed to Proto-AustroTai from which the tonal systems of Miao-Yao and Kadai, as well as the accentual system in Japanese-Ryukyuan might be derived. Following the policy of this chapter, we adopt R. Gordon’s (2005) more modest view on the groups involved and treat them as three independent phyla. With respect to the Mon-Khmer branch of Austroasiatic, Di¿oth’s (2005) most recent classification distinguishes between the Khasi-Khmuic and the Khmero-Vietic/Nico-Monic branch. R. Gordon’s (2005) tree given below is compatible with the hypothesis that the family consists of up to ten subgroups of comparable time depth, suggesting a rapid spread over Southeast Asia (Sidwell 2006). Comrie et al. (2003) also have Viet-Muong as a separate daughter of Mon-Khmer, and suggest the same as a possibility for the Nicobar branch. AUSTROASIATIC MON-KHMER: ASLIAN: JAH HUT: Jah Hut NORTH ASLIAN: CHEWONG: Chewong EASTERN: Batek, Jehai, Minriq, Mintil TONGA: Tonga WESTERN: Kintaq, Kensiu SENOIC: Lanoh, Sabu¨m, Semai, Semnam, Temiar SOUTH ASLIAN: Besisi, Semelai, Semaq Beri, Temoq EASTERN MON-KHMER: BAHNARIC:

590

Rene´ Schiering and Harry van der Hulst

CENTRAL BAHNARIC: Alak, Bahnar, Lamam, Romam, Tampuan, Kaco’ NORTH BAHNARIC: Katua EAST: CUA-KAYONG: Cua, Kayong TAKUA: Takua WEST: Trieng, Talieng DUAN: Halang Doan JEH-HALANG: Halang, Jeh RENGAO: Rengao SEDANG-TODRAH: SEDANG: Hre, Sedang TODRAH-MONOM: Monom, Todrah SOUTH BAHNARIC: Budeh Stieng SRE-MNONG: MNONG: EASTERN: Eastern Mnong SOURHERN-CENTRAL: Central Mnong, Southern Mnong, Kraol SRE: Maa, Koho STIENG-CHRAU: Chrau, Bulo Stieng WEST BAHNARIC: BRAO-KRAVET: Lave, Kru’ng, Kravet, Sou LAVEN: Laven NYAHEUN: Nyaheun OI-THE: Jeng, Oy, Sok, Sapuan, The KATUIC: CENTRAL KATUIC: TA’OIH: Ir, Kataang, Ong, Upper Ta’oih, Lower Ta’oih EAST KATUIC: KASENG: Kasseng KATU-PACOH: Eastern Katu, Western Katu, Pacoh, Phuong, Tareng NGEQ-NKRIANG: Khlor, Ngeq WEST KATUIC: BROU-SO: Eastern Bru, Western Bru, Soˆ, Khua KUAY-NHEU: Kuy KUAY-YOE: Nyeu KHMER: Central Khmer, Northern Khmer PEARIC:

Word accent systems in the languages of Asia

591

EASTERN: Pear WESTERN: CHONG: Chong, Sa’och SAMRE: Somray, Samre SUOY: Suoy MONIC: Nyahkur, Mon NICOBAR: CAR: Car Nicobarese CHOWRA-TERESSA: Chaura, Teressa GREAT NICOBAR: Southern Nicobarese NANCOWRY: Central Nicobarese SHOM PENG: Shom Peng NORTHERN MON-KHMER: KHASIAN: War, Khasi, Pnar KHMUIC: KHAO: Bit, Khao MAL-KHMU’: KHMU’: Khuen, Khmu, O’du MAL-PHRAI: Mal, Lua’, Phai, Pray MLABRI: Mlabri XINH MUL: Kha´ng, Phong-Kniang, Puoc MANG: Mang PALAUNGIC: EASTERN PALAUNGIC: DANAU: Danau PALAUNG: Pale Palaung, Rumai Palaung, Shwe Palaung RIANG: Riang, Yinchia WESTERN PALAUNGIC: ANGKUIC: Kon Keu, Hu, Man Met, Mok, Samtao, Tai Loi, U, Kiorr LAMETIC: Con, Lamet WAIC: BULANG: Blang LAWA: Western Lawa, Eastern Lawa WA: Parauk, Vo PALYU: Bogan, Bolyu UNCLASSIFIED: Bugan, Buxinhua, Kemiehua, Kuanhua VIET-MUONG: CHUT: Arem, Maleng, Chut CUOI: Hung, Tho

592

Rene´ Schiering and Harry van der Hulst

MUONG: Bo, Muong, Nguoˆn THAVUNG: Aheu VIETNAMESE: Vietnamese MUNDA (see Section 6.3.) StressTyp extracts Halang; Koyong [U] Stress is located on the final syllable.

f

Sedang [U] f Phonological words are maximally bisyllabic. f Stress is located on the final, or the only, syllable. f Status of ‘‘stress’’ in Sedang unclear. ¨a ‘i’

ko¨blin ‘to be very full’

Khmer, Central; Cambodian [U] f Disyllabic words consist of an unstressed pre-syllable followed by a stressed full syllable. f Trisyllabic words exist; they contain two pre-syllables. f Words with more full syllables are polymorphemic and carry more stresses. f Status of ‘‘stress’’ in Cambodian might be debatable. k n¨laeN ‘place’

krakh¨wak

‘dirty’

ura¨moh ‘house’

A

Khasi [U] f In isolation words have one stress which coincides with the falling pitch on the final syllable. In context these stresses and pitch contours are deleted on all words but the final. f Status of ‘‘stress’’ in Khasi highly debatable. ¨khlaa ‘tiger’ tara¨jur ‘scales’

paa¨troy ‘pull by the hair’

Khmu’ [U] f Like in Cambodian one or two pre-syllables and a final full syllable make up a word. f Stress is located on the full syllable. Words with two full syllables are polymorphemic and carry two stresses. f Status of ‘‘stress’’ in Khmu’ might be debatable. k«¨t$N

‘jar’

s«m¨lı`øk

‘fish scale’

c«r¨la`ø¨ta´øp ‘butterfly’

Word accent systems in the languages of Asia

593

Additional information With respect to the Aslian languages, Jahai (Burenhult 2005: 38) has noncontrastive accent that invariably falls on the last syllable of the word. No further levels of accentuation, such as secondary accent, are evidenced. Word-final accent placement applies as much as to recent loans from Malay as to native words. According to Kruspe (2004: 40), accent is a syllable-related phenomenon in Semelai. Within the phonological word, accent is on the final syllable and there is no secondary accent. In the course of su‰xation, the accent shifts from the root to the su‰x, cf. (89). ‘to follow’ p¨d r pd ¨ri? ‘to follow (s.th.)’ A A

(89) a. b.

In the Bahnaric language Chrau (Thomas 1971: 31), the phonological word can be defined with reference to accent, such that it constitutes a phonological stretch containing only one main, accented syllable. The main syllable may be complemented by one unaccented presyllable, e.g. ca¨mla˘h ‘deny’. For Mon,19 Bauer (1982: 99¤.) distinguishes four degrees of accent: zero (unaccented), primary accent, secondary accent and tertiary accent. Accent placement is sensitive to four di¤erent domains: polysyllables in isolation, compounds, phrases, and the sentence, where compounds and phrases behave alike. The following examples illustrate the various attested accent patterns with words of varying degrees of morphological complexity and phonological length. (90) a. b. c. d. e. f.

/¨tEm/ /p«¨tEm/

‘to know’ ‘to inform’

(stem, monosyllabic) (prefix þ stem, disyllabic)

/˙«¨kh˛$/ /˙ łı`/¨łE$?/ /˙cNh«¨ku`i/

‘during’ ‘a little (bit)’

(prefix þ stem, disyllabic) (stem þ stem, disyllabic)

‘to cause to burn’ (stem þ stem, trisyllabic) /h«˙tom¨cih/ ‘to fall down’ (stem þ stem, trisyllabic)

On the basis of such evidence, the following accent placement rules can be formulated.

19. See also Schiering (2006) for a summary of word-related phonological patterns in Mon.

594

Rene´ Schiering and Harry van der Hulst

(91) a.

Primary stress is placed on the final syllable of polysyllabic words. b. Secondary stress appears on the initial syllable of disyllabic and trisyllabic words. c. Tertiary stress is realized on the medial syllable of trisyllabic words. d. The initial syllable of native disyllabic words and trisyllables remains unstressed; in some cases, trisyllables have secondary stress on the initial syllable and unstressed medial syllables. (Note that tri- and tetrasyllables are loans).

In one case, accent placement seems to make a lexical contrast, cf. (92). /˙ łE$h¨ko`h/ ‘who?’ b. /¨łE$h˙ko`h/ ‘they, them, any’

(92) a.

However, the special prosodic status of the definiteness marker /ko`h/ sheds doubt on this minimal pair. As shown in (93), if this element closes a noun phrase, the preceding noun receives primary accent counter to the more general phrase-final accent pattern. /˙kwan ¨mo`a/ b. /¨kwan ˙ko`h/ c. /˙kwan ¨nc? ˙ko`h/

(93) a.

‘a village’ ‘the village’ ‘this village (def.)’

Rabel’s (1961: 30) observation that Khasi words, at least in isolation, have word-final primary accent which coincides with a falling pitch is supported by Khyriem’s (2001) study. The data in (94) show that accent falls on the final syllable in simple, complex and compound words. (94) a. b. c. d.

/kfintu´/ ‘to persuade’ /bfita´r/ ‘to be angry’ /fi@N þ pfi@N þ ´ım/ ! /fiNpfiNnı´m/ ‘salvation’ ‘coal’ /khfindEu þ @@N/ ! /dEuj@ N/

According to Thompson (1965: 106–107), each syllable in Vietnamese 20 carries one accent. Monomorphemic, disyllabic words are realized with final accent if uttered in isolation, e.g. va-¨li ‘suitcase’. At higher levels of prosodic structure, three levels of accent are distinguished: weak, medium 20. See also Schiering (2007) for a summary of word-related phonological patterns in Vietnamese.

Word accent systems in the languages of Asia

595

and heavy. The general pattern is that of iambic phrasing, but ultimately the degree of accent on the syllables within a pause group is determined by the information load of the di¤erent elements. Accentuation of compounds is illustrated in (95). (95) a. b. c. d.

ngu’o`’i ¨ta moˆ.t ¨mı`nh hoa ¨ho`ˆ ng Lieˆn-¨hieˆ.p quo´ˆ c

‘somebody’ ‘alone’ ‘rose’ ‘United nations’

In the default case, disyllabic compound words are realized with final accent. In (95d), the pseudo-compound has its origin in structural borrowing from Chinese. In such Sino-Vietnamese compounds, accent is assigned to the left branch of the construction, in which stress is realized on the final syllable. Disyllabic reduplications as in (96a) also exhibit final accent. Longer polysyllabic reduplicative strings are parsed as two accent domains with final accent, cf. (96b). (96) a. b.

no´i ¨no´i mo’ ¨mo’ ma`ng ¨ma`ng

‘keep talking and talking’ ‘deep in the state of dreaming’

At the phrase level, default final accent placement prevails. Accordingly, the phrases in (97) are realized with final accent. (97) a. b.

hoa ¨ho`ˆ ng Toˆi khoˆng ¨bie´ˆ t.

‘pink flower’ ‘I don’t know’

Generalizations The most obvious recurring accent pattern in the Mon-Khmer languages relates to the canonical sesquisyllabic word structure which implies the iambic rhythm of an unaccented presyllable and an accented main syllable. In Donegan & Stampe’s (1983) holistic typology, Mon-Khmer thus constitutes the opposite extreme pole to Munda and is said to retain Proto-Austroasiatic prosody. 7.3. Hmong-Mien Genetic information There seems to be agreement on the division of Hmong Mien into three daughters. Hmongic (Miao), Honte (with the language She) and Mienic

596

Rene´ Schiering and Harry van der Hulst

(Yao) (see Comrie et al. 2003). The following language classification is taken from R. Gordon (2005). HMONG-MIEN HMONGIC: BUNU: Younuo Bunu, Wunai Bunu, Bu-Nao Bunu, Jiongnai Bunu CHUANQIANDIAN: Hmong Njua, Southern Mashan Hmong, Central Huishui Hmong, Northeastern Dian Hmong, Eastern Huishui Hmong, Hmong Don, Southwestern Guiyang Hmong, Southwestern Huishui Hmong, Northern Huishui Hmong, Chonganjiang Hmong, Luopohe Hmong, Central Mashan Hmong, Northern Mashan Hmong, Hmong Doˆ, Western Mashan Hmong, Southern Guiyang Hmong, Hmong Shua, Northern Guiyang Hmong, Hmong Daw PA-HNG: Pa-Hng QIANDONG: Northern Qiandong Hmong, Eastern Qiandong Hmong, Southern Qiandong Hmong XIANGXI: Western Xiangxi Hmong, Eastern Xiangxi Hmong HO NTE: She MIENIC: BIAO-JIAO: Biao-Jiao Mien MIAN-JIN: Biao Mon, Iu Mien, Kim Mun ZAOMIN: Dzao Min Accent information Hmong-Mien is usually described as being exclusively tonal, but that does not imply that the languages in this family couldn’t be accentual as well, cf. Chinese. Unfortunately, we did not have access to sources that would allow us to elaborate on this point. 7.4. Tai-Kadai Genetic information For this family, most sources distinguish three main subgroups: the Tailanguages, the Kam-Sui group and the Kadai group. The first two are often taken together as the Kam-Tai group, for instance in Comrie et al. (2003). Crystal (1997) has a Tai family, which he does not link to Kadai

Word accent systems in the languages of Asia

597

and Kam-Sui. Ruhlen (1991) has a more intricate family which he calls Daic. The main di¤erence is that the Kadai languages have been split up in three inclusive groups. Finally, Tai-Kadai has Lati as a sister. We follow R. Gordon’s (2005) proposal for the purposes of this survey.

TAI-KADAI HLAI: Jiamao, Hlai KADAI: BU-RONG: Yerong GE-CHI: Gelao, Green Gelao, Red Gelao, White Gelao, Lachi, White Lachi YANG-BIAO: Buyang, Cun, En, Qabiao, Laha KAM-TAI: BE-TAI: BE: Lingao TAI-SEK: SEK: Saek TAI: Rien, Tay Khang, Tai Pao, Tai Do CENTRAL: Southern Zhuang, E, Cao Lan, Nung, Ts’u¨n-Lao, Ta`y EAST CENTRAL: NORTHWEST: Turung NORTHERN: Northern Zhuang, Bouyei, Tai Me`ne, Yoy SOUTHWESTERN: Tai Ya, Pa Di, Pu Ko, Tai Long, Tai Thanh, Ta`y Sa Pa EAST CENTRAL: CHIANG SAENG: Tai Dam, Northern Thai, Phuan, Thai Song, Thai, Tai Hang Tong, Tai Do´n, Thu Lao, Tai Daeng, Ta`y Tac LAO-PHUTAI: Lao, Nyaw, Phu Thai, Northeastern Thai NORTHWEST: Ahom, Aiton, Lu¨, Khamti, Khu¨n, Khamyang, Phake, Shan, Tai Nu¨a SOUTHERN: Southern Thai UNCLASSIFIED: Tai Hongjin, Yong UNCLASSIFIED: Kuan KAM-SUI: Ai-Cham, Biao, Cao Miao, Northern Dong, Southern Dong, Kang, Mak, Mulam, Maonan, Sui, T’en LAKKJA: Lakkia

598

Rene´ Schiering and Harry van der Hulst

Accent information All languages in this family are tonal and morphemes are monosyllabic. Abrahamson (p.c.) reports that in polysyllabic words, e.g. compounds, Thai has final stress, by and large, with shortening of vowels in the first syllable. Smyth (2002: 10) also describes disyllabic words as having accent on the final syllable. The unaccented vowel /a/ in a word-initial syllable is usually reduced to schwa and its tone is mid. When /aa/ occurs both in the first and the second syllable, it is normally shortened in the first syllable. (98) a. b. c. d.

pra¨tuu [pr«¨tuu] sa¨du`ak [s«¨du`ak] aa¨ha˘an [a¨ha˘an] phaa¨sa˘a [pha¨sa˘a]

‘door’ ‘convenient’ ‘food’ ‘language’

The phonotactic template for words in Lao is C0V0.C1V1V2C2, where only C1 and V1 are obligatory. The initial minor syllable (C0V0) is unaccented and does not show distinctive tone independent of the accented major syllable. De-accented a‰xes or clitics, e.g. class terms, modifier classifiers and some aspectual-modal markers, behave regularly like minor syllables in this respect (Enfield 2007: 33).

8. Generalizations and conclusions Obviously, with such a large amount of languages spread over Asia, no sweeping generalizations can be made with respect to the word prosodic systems found in this area. Everything under the accentual ‘sun’ seems to be represented in the survey. Furthermore, even within the more narrowly delimited regions discussed in the various sections, no major areal patterns emerge. Only the prominence of sesquisyllabic word structure and iambic prosody in the languages of Southeast Asia could be taken as a candidate for an areal pattern (cf. the noted prosodic di¤usibility noted by Matiso¤ 2001). However, the accentual data surveyed for di¤erent families within a phylum often o¤er significant insights into the diachrony of word prosody. For instance, the relative conformity of accentuation in the major branches of Altaic is noteworthy, even more so if they can be attributed to a single source in the proto-language. In Austroasiatic, on the other

Word accent systems in the languages of Asia

599

hand, we find a radically di¤erent picture: the word prosodic system of the proto-language can arguably only be found in the Mon-Khmer branch, whereas the Munda branch has undergone a major typological drift. Finally, the languages of Asia o¤er a high potential for the study of the incompatibility or interaction between accent and tone. In this context, it would be highly desirable if phonological descriptions would not focus on one of the two suprasegmental features to the exclusion of the other, as is often done in the context of ‘pure tone languages’.

References Abbi, Anvita 2006

Endangered Languages of the Andaman Islands. (Lincom Studies in Asian Linguistics 64.) Mu¨nchen: Lincom Europa. Anderson, Gregory D. S. 1997 Burushaski phonology. In: Alan S. Kaye 1997b (ed.), 1021–1041. Anderson, Gregory D. S. 1997 Lak phonology. In: Alan S. Kaye 1997b (ed.), 973–1000. Anderson, Gregory D. S. 2001 A new classification of South Munda: Evidence from comparative verb morphology. Indian Linguistics 62, 21–36. Anderson, Gregory D. S. (ed.) 2008 The Munda Languages. (Routledge Curzon Language Family Series.) New York: Routledge Curzon. Andronov, M. S. 1970 Dravidian Languages. Moscow: Izdatel’stvo ‘‘Nauka’’. Annamalai, E. and Sanford B. Steever 1998 Modern Tamil. In: Sanford B. Steever 1998 (ed.), 100–128. Asher, R. E. 1982 Tamil. (Lingua Descriptive Studies 7.) Amsterdam: NorthHolland. Asher, R. E. 2007 Southern Asia: from Iran to Bangladesh. In: R. E. Asher and Christopher Moseley 2007 (eds.), 211–228. Asher, R. E. and Christopher Moseley (eds.) 2007 Atlas of the World’s Languages. Second Edition. London and New York: Routledge. Ashthamurthy, K. 2003 The phonetics and phonology of primary prominence in Malayalam and Tamil: A typological investigation. PhD Dissertation. CIEFL, Hyderabad.

600

Rene´ Schiering and Harry van der Hulst

Baart, J. G. 1997

The sounds and tones of Kalam Kohistani. (Studies in the Languages of Northern Pakistan.) Dallas: Summer Institute of Linguistics.

Baart, J. G. 2003 Tone languages in Northern Pakistan. Ms. Balasubramanian, T. 1980 Timing in Tamil. Journal of Phonetics 8: 449–467. Barker, Christopher 1989 Extrametricality, the cycle, and Turkish word stress. In: Junko Itoˆ & Je¤ Runner (eds.) Phonology at Santa Cruz 1. Santa Cruz: Linguistics Research Center, 1–33. Bauer, Christian 1982 Morphology and Syntax of Spoken Mon. London: SOAS. Beekes, Robert 1995 Comparative Indo-European Linguistics. Amsterdam/Philadelphia: John Benjamins. Benedict, Paul 1972 Sino-Tibetan, a Conspectus. (Contributing editor, James A. Matiso¤.) Cambridge: Cambridge University Press. Benedict, Paul K. 1990 Japanese/Austro-Tai. Ann Arbor: Karoma. Berger, Hermann 1974 Das Yasin-Burushaski (Werchikwar Grammatik, Texte, Wo¨rterbuch. (Neuindische Studien 3.) Wiesbaden: Harrassowitz. Berger, Hermann 1998 Die Burushaski-Sprache von Hunza und Nager. Teil 1: Grammatik. (Neuindische Studien 13.) Wiesbaden: Harrassowitz. ´ rpa´d Berta, A ´ . Csato´ 1998 1998a Tatar and Bashkir. In: Lars Johanson and E´va A (eds.), 283–300. ´ rpa´d Berta, A ´ . Csato´ 1998b West Kipchak languages. In: Lars Johanson and E´va A 1998 (eds.), 301–317. Bickel, Balthasar 1998 Rhythm and feet in Belhare morphology. Rutgers Optimality Archive, http://www.ruccs.rutgers.edu/roa.html Bickel, Balthasar 2003 Belhare. In: Graham Thurgood and Randy J. LaPolla 2003 (eds.), 546–570. Bickel, Balthasar, Goma Banjade, Martin Gaenszle, Elena Lieven, Netra Paudyal, Ichchha Rai, Manoj Rai, Novel Kishor Rai and Sabine Stoll 2007 Free prefix ordering in Chintang. Language 83: 1–31. Boeschoten, Hendrik ´ . Csato´ 1998 (eds.), 357– 1998 Uzbek. In: Lars Johanson and E´va A 378.

Word accent systems in the languages of Asia

601

Boeschoten, Hendrik and Marc Vandamme ´ . Csato´ 1998 (eds.), 1998 Chaghatay. In: Lars Johanson and E´va A 166–178. Bolinger, D. L. 1981 Two Kinds of Vowels, Two Kinds of Rhythm (distributed by the Indiana University Linguistics Club, Bloomington, Indiana.) Bosson, J. 1964 Modern Mongolian: a primer and reader. Uralic and Altaic Series 38. Bloomington: Indiana University. Bradley, David 2007 East and South-East Asia. In: R. E. Asher and Christopher Moseley 2007 (eds.), 159–208. Brendemoen, Bernt ´ . Csato´ 1998 1998 Turkish dialects. In: Lars Johanson and E´va A (eds.), 236–241. Burenhult, Niclas 2005 A Grammar of Jahai. (Pacific Linguistics 566.) Canberra: Pacific Linguistics. Chao, Yuen Ren 1968 A Grammar of Spoken Chinese. Berkeley, Los Angeles and London: University of California Press. Chatterji, Suniti Kumar 1991 Bengali Self-Taught. Calcutta: Rupa. Cho, Young-mee Yu 2003 Syllable weight and tone in Korean. In: William McClure (ed.) Japanese/Korean Linguistics. Volume 12. CSLI Publications, 103–116. Clark, Larry ´ . Csato´ 1998 (eds.), 434– 1998 Chuvash. In: Lars Johanson and E´va A 452. Comrie, Bernard 1981 Languages of the Soviet Union. Cambridge: Cambridge University Press. Comrie, Bernard 1997a Turkish phonology. In: Alan S. Kaye 1997b (ed.), 883–898. Comrie, Bernard 1997b Tatar (Volga Tatar, Kazan Tatar) phonology. In: Alan S. Kaye 1997b (ed.), 899–912. Comrie, Bernard 1997c Uyghur phonology. In: Alan S. Kaye 1997b (ed.), 913–928. Comrie, Bernard 2007 Northern Asia and Eastern Europe. In: R. E. Asher and Christopher Moseley 2007 (eds.), 231–256. Comrie, Bernard (ed.) 1990 The World’s Major Languages. Oxford: Oxford University Press.

602

Rene´ Schiering and Harry van der Hulst

Comrie, Bernard, Maria Polinsky, and Stephen Matthews 1996 The Atlas of Languages. London: Quarto. Comrie, Bernard, Maria Polinsky, and Stephen Matthews 2003 The Atlas of Languages. Second Edition. London: Quarto Publishing Inc. Comrie, Bernard, Matthew Dryer, Martin Haspelmath and David Gil (eds.) 2005 World Atlas of Language Structures. Oxford: Oxford University Press. Crystal, David 1997 The Cambridge encyclopedia of language. 2nd edition. ´ . and Birsel Karakoc¸ Csato´, E´va A ´ . Csato´ 1998 (eds.), 344– 1998 Noghay. In: Lars Johanson and E´va A 356. ´ . and Lars Johanson Csato´, E´va A ´ . Csato´ 1998 (eds.), 203– 1998 Turkish. In: Lars Johanson and E´va A 235. Das, Sisirkumar 1973 Structure of Malto. Annamalai University. De Jong, Ken 1994 Initial tones and prominence in Seoul Korean. Ohio State University Working Papers in Linguistics 43: 1–14. Dettmer, Hans A. 1989 Ainu-Grammatik. Teil 1: Texte und Hinweise. (Vero¨¤entlichungen des Ostasien-Instituts der Ruhr-Universita¨t Bochum 38.) Wiesbaden: Harrassowitz. Di¿oth, Ge´rard 2005 The contribution of linguistic palaeontology to the homeland of Austroasiatic. In: Laurent Sagart, Roger Blench and Alicia Sanchez-Mazas (eds). The Peopling of East Asia: Putting Together Archaeology, Linguistics and Genetics, 79–82. New York: Routledge Curzon. Ding, Picus Sizhi 2003 Prinmi: A sketch of Niuwozi. In Graham Thurgood and Randy J. LaPolla 2003 (eds.), 588–601. Dobrovolsky, Michael 1999 The phonetics of Chuvash stress: implications for phonology. Proceedings of the 14th International Conference of Phonetic Sciences: 539–42. Donegan, Patricia 1993 Rhythm and vocalic drift in Munda and Mon-Khmer. Linguistics of the Tibeto-Burman area 16: 1–43. Donegan, Patricia J. and David Stampe 1983 Rhythm and the holistic organization of language structure. In: John F. Richardson, Mitchell Marks and Amy Chukerman (eds.) Papers from the Parasession on the Interplay of Phonology,

Word accent systems in the languages of Asia

603

Morphology and Syntax, 337–353. Chicago: Chicago Linguistic Society. Duanmu, San 2000

The phonology of Standard Chinese. Oxford: Oxford University Press. Dunn, Michael John 1999 A grammar of Chukchi. Doctoral dissertation, Australian National University. Ebert, Karen 1996 Kodava. (Languages of the World/Materials 104.) Mu¨nchen and Newcastle: Lincom Europa. Elfenbein, Josef 1997a Balochi phonology. In: Alan S. Kaye 1997b (ed.), 761–776. Elfenbein, Josef 1997b Brahui phonology. In: Alan S. Kaye 1997b (ed.), 797–814. Elfenbein, Josef 1997c Pashto phonology. In: Alan S. Kaye 1997b (ed.), 733–760. Elfenbein, Josef 1998 Brahui. In: Sanford B. Steever 1998 (ed.), 388–414. Emeneau, M. B. 1961 Kolami: A Dravidian Language. (Annamalai University Publications in Linguistics 2.) Annamalainagar: Annamalai University. Emeneau, M. B. 1970 Dravidian Comparative Phonology. Annamalainagar: Annamalai University. Enfield, N. J. 2007 A Grammar of Lao. (Mouton Grammar Library 38.) Berlin and New York: Moutn de Gruyter. Everett, Daniel L. 1988 On metrical constituent structure in Piraha˜ phonology. Natural Language & Linguistic Theory 6: 207–246. Fortescue, Michael 2005 Comparative Chukotko-Kamchatkan dictionary. (Trends in Linguistics/Documentation, 23.) Berlin and New York: Mouton de Gruyter. Fukui, Rei 2003 Pitch accent systems in Korean. In: Shigeki Kaji (ed.) Proceedings of the symposium ‘Cross-linguistic studies of tonal phenomena’ 275–286. Research Institute for Languages and Cultures of Asia and Africa (ILCAA Tokyo: University of Foreign Studies.) Genetti, Carol 2003 Dolakha# Newa#r. In: Graham Thurgood and Randy J. LaPolla 2003 (eds.), 355–370. Georg, Stefan, Peter A. Michalove, Alexis M. Ramer & Paul J. Sidwell 1999 Telling general linguists about Altaic. Journal of Linguistics 35: 65–98.

604

Rene´ Schiering and Harry van der Hulst

Go¨ksel, Aslı and Celia Kerslake 2005 Turkish: A Comprehensive Grammar. (Comprehensive Grammars.) London and New York: Routledge. Goldsmith 1975 Autosegmental Phonology. PhD dissertation. MIT. Gordon, John Matthew 2000 Re-examining default-to-opposite stress. Berkeley Linguistic Society 26: 101–112. Gordon, John Matthew 2005 A perceptually-driven account of onset-sensitive stress. Natural Language & Linguistic Theory 23: 595–653. Gordon, Raymond G., Jr. (ed.) 2005 Ethnologue: Languages of the world. 15th edition. Dallas, Tex.: SIL International. Online version: http://www.ethnologue.com/. Goswami, Golok Chandra 1966 An Introduction to Assamese Phonology. Poona: Deccan College. Gruzdeva, Ekaterina 1998 Nivkh. (Languages of the World/Materials 111.) Mu¨nchen and Newcastle: Lincom Europa. Gussenhoven, Carlos 2004 The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Hahn, Ferdinand 1908 Grammar of the Kurukh language. Bengal Secretariat Press. Hahn, Reinhard F. ´ Csato´ 1998 (eds.), 379– 1998a Uyghur. In: Lars Johanson and E´va A 396. Hahn, Reinhard F. ´ . Csato´ 1998b Yellow Uyghur and Salar. In: Lars Johanson and E´va A 1998 (eds.), 397–402. Hahn, Reinhart F. 1991 Spoken Uyghur. Seattle: University of Washington Press. Hale, Mark 2004 Avestan. In: Roger D. Woodard 2004 (ed.), 742–763. Hale, Mark 2004 Pahlavi. In: Roger D. Woodard 2004 (ed.), 764–776. Haraguchi, S. 1977 The Tone Pattern of Japanese: An Autosegmental Theory of Tonology. Tokyo: Kaitakusha. Haraguchi, S. 1988 Pitch accent and intonation in Japanese. In: Harry van der Hulst and Norval Smith (eds.) Autosegmental Studies on Pitch Accent, 123–150. Dordrecht: Foris Publications. Hayes, Bruce 1995 Metrical Stress Theory. Chicago: The University of Chicago Press.

Word accent systems in the languages of Asia

605

Hayes, Bruce and Aditi Lahiri 1991 Bengali intonational phonology. Natural Language and Linguistic Theory 9: 47–96. Hildebrandt, Kristine A. 2007 Prosodic and grammatical domains in Limbu. Himalayan Linguistics Journal 8: 1–34. Hulst, Harry van der 1999 Word accent. In: Harry van der Hulst 1999 (ed.), 3–116. Hulst, Harry van der 2002 Stress and accent. In: Nadel, L. (ed.) Encyclopedia of Cognitive Science. Vol. 4, 4, 246–254. London: Nature Publishing Group. Hulst, Harry van der 2006 Word stress. In: Keith Brown (ed.) The encyclopedia of language and linguistics. 2nd edition. Volume 13, 655–665. Oxford: Elsevier. Hulst, Harry van der (ed.) 1999 Word prosodic systems in the languages of Europe. (Empirical approaches to language typology/Eurotyp 20.4.) Berlin: Mouton de Gruyter. Hulst, Harry van der, Bernadet Hendriks and Jeroen van de Weijer 1999 A survey of word prosodic systems of European languages. In: Harry van der Hulst 1999 (ed.), 425–476. Inkelas, S. 1999 Exceptional stress-attracting su‰xes in Turkish: representations versus the grammar’. In: Kager, R., van der Hulst, H. & Zonneveld, W. (eds.) The Phonology-Morphology Interface, 134–187. Cambridge: Cambridge University Press. Inkelas, Sharon and Cemil Orhan Orgun 2003 Turkish Stress: a review. Phonology 20: 139–161. Johanson, Lars ´ . Csato´ 1998 1998a The history of Turkic. In: Lars Johanson and E´va A (eds.), 81–125. Johanson, Lars ´ . Csato´ 1998b The structure of Turkic. In: Lars Johanson and E´va A 1998 (eds.), 30–66. ´ . Csato´ (eds.) Johanson, Lars and E´va A 1998 The Turkic languages. London: Routledge. Jun, Sun-Ah 1993 The Phonetics and Phonology of Korean Prosody. Colmubus: Ohio State University Dissertation. [1996 Garland] Kabak, Barıs¸ and Irene Vogel 2001 The phonological word and stress assignment in Turkish. Phonology 18, 315–360. Kaisse, Ellen 1985 Some Theoretical Consequences of Stress Rules in Turkish.

606

Rene´ Schiering and Harry van der Hulst Papers from the Regional Meetings of the Chicago Linguistic Society 21, 199–209. Chicago: Chicago Linguistic Society.

Kaye, Alan S. 1997 Hindu-Urdu phonology. In: Alan S. Kaye 1997b (ed.), 637–652. Kaye, Alan S. (ed.) 1997a Phonologies of Asia and Africa. Volume 1. Winona Lake: Eisenbrauns. Kaye, Alan S. (ed.) 1997b Phonologies of Asia and Africa. Volume 2. Winona Lake: Eisenbrauns. Kaye, Alan S. 1997c Hindu-Urdu phonology. In: Alan S. Kaye 1997b (ed.), 637–652. Keane, Elinor 2001 Echo words in Tamil. PhD Thesis, Merton College, Oxford. Kerslake, Celia ´ . Csato´ 1998 1998 Ottoman Turkish. In: Lars Johanson and E´va A (eds.), 179–202. Khyriem, Barika 2001 Some Aspects of Syllable Structure in Khasi Unpublished Master’s Thesis, Deaprtment of Linguistics, University of Delhi, New Delhi. Kim, J. K. 1998 Anti-trapping e¤ects in an iambic system: Vowel shortening in Korean. The 8th Japanese/Korean Linguistics: 289–302. Kim, Sahyang 2004 The Role of Prosodic Phrasing in Korean Word Segmentation. Doctoral dissertation: UCLA. Kirchner, Mark ´ . Csato´ 1998a Kazakh and Karakalpak. In: Lars Johanson and E´va A 1998 (eds.), 318–332. Kirchner, Mark ´ . Csato´ 1998 (eds.), 344– 1998b Kirghiz. In: Lars Johanson and E´va A 356. Ko, Eon-Suk 2010 Stress and long vowels in Korean: Chicken or egg, first? Japanese Korean Linguistics 17, CSLI, pp. 377–390. Kornfilt, Jaklin 1997 Turkish. (Descriptive grammars.) London and New York: Routledge. Krishnamurti, Bhadriraju 1998 Telugu. In: Sanford B. Steever 1998 (ed.), 202–240. Krishnamurti, Bhadriraju 2003 The Dravidian Languages. Cambridge: Cambridge University Press. Krishnamurti, Bhadriraju and Brett A. Benham 1998 Konda. In: Sanford B. Steever 1998 (ed.), 241–269.

Word accent systems in the languages of Asia

607

Krueger, John R. 1961 Chuvash Manual. Bloomington: Indiana University. Krueger, John R. 1962 Yakut Manual. Area Handbook, Grammar, Graded Reader and Glossary. (Uralic and Altaic Series 21.) Bloomington: Indiana University. Kruspe, Nicole 2004 A Grammar of Semelai. (Cambridge Grammatical Descriptions.) Cambridge: Cambridge University Press. Lahiri, Aditi and Jennifer Fitzpatrick-Cole 1999 Emphatic clitics and focus intonation in Bengali. In: Rene´ Kager and Wim Zonneveld (eds.) Phrasal Phonology, 119–144. Nijmegen: University of Nijmegen Press. LaPolla, Randy J. 2003a Qiang. In: Graham Thurgood and Randy J. LaPolla (eds.) 2003, 573–587. LaPolla, Randy J. 2003b Dulong. In: Graham Thurgood and Randy J. LaPolla (eds.) 2003, 674–682. Lazzaroni, Romano 1993 Sanskrit. In: Anna Giacalone Ramat and Paolo Ramat 1993 (eds.), 98–124. Lee, Iksop and S. Robert Ramsey 2000 The Korean Language. Albany: State University of New York Press. Lewis, Geo¤rey 2000 Turkish Grammar. Second Edition. Oxford: Oxford University Press. Li, Charles N. & Sandra A. Thompson 1981 Mandarin Chinese. A Functional Reference Grammar. Berkeley, Los Angeles and London: University of California Press. Lim, Byung-Jin 2000 The role of syllable-weight and position on prominence in Korean. In: Miheharu Nakayama and Charles J. Quinn (eds.) Japanese/Korean linguistics Volume 9, 139–150. CSLI Publications. Lin, Hua 2001 A Grammar of Mandarin Chinese. (Languages of the World/ Materials 344.) Mu¨nchen: Lincom Europa. Lisker, L. and Bh. Krishnamurti 1991 Lexical stress in a ‘stressless’ language: judgments by Teluguand English-speaking linguists. Proceedings of the XII International Congress of Phonetic Sciences (Universite´ de Provence) 2, 90–93.

608

Rene´ Schiering and Harry van der Hulst

Mahanta, Shakuntala 2002 Some Aspects of Prominence in Assamese and Assamese English. Unpublished M. Phil. Dissertation. Hyderabad: CIEFL. Manoharan, S. 1989 A Descriptive and Comparative Study of Andamanese Language. Calcutta: Government of India, Ministry of Human Resource Development. Masica, Colin P. 1991 The Indo-Aryan Languages. (Cambridge Language Surveys.) Cambridge: Cambridge University Press. Maslova, Elena 2003 A grammar of Kolyma Yukaghir. (Mouton Grammar Library 27.) Berlin and New York: Mouton de Gruyter. Mason, Francis 1854 The Talaing Language. Journal of American Oriental Society 4, 277–288. Matiso¤, James A. 1973 The Grammar of Lahu. (University of California Publications, Linguistics 75.) Berkeley, Los Angeles and London: University of California Press. Matiso¤, James A. 2001 Genetic vs. areal linguistics in Southeast Asia: prosodic di¤usibility in Southeast Asian languages. In Aleksandra Y. Aikhenvald and R. M. W. Dixon (eds.), Areal Di¤usion and Genetic Inheritance, 291–327. Cambridge: Cambridge University Press. Matthews, Stephen, Bernard Comrie and Maria Polinsky (eds.) 1996 The Atlas of Languages: The Origins and Development of Languages throughout the World. New York: Facts and File. Matthews, W. K. 1951 Languages of the USSR. Cambridge: Cambridge University Press. McCarus, Ernest N. 1997 Kurdish phonology. In: Alan S. Kaye 1997b (ed.), 691–706. Mistry P. J. 1997 Gujarati phonology. In: Alan S. Kaye 1997b (ed.), 653–674. Muravyova, I. A. 1979 The comparative study of Chuckchi, Koryak and Alutor morphology. Candidate degree dissertation. Moscow (Муравьева

И.А. 1979 Сопоставительное исследование морфологии чукотского, корякского и алюторского языков. Диссертация на соискание ученой степени кандидата филологических наук. Москва.)

Nedjalkov, Igor 1997 Evenki. London and New York: Routledge.

Word accent systems in the languages of Asia

609

Neukom, Lukas 2001 Santali. (Languages of the World/Materials 323.) Mu¨nchen: Lincom Europa. Nikolaeva, Irina 2006 A historical dictionary of Yukaghir. (Trends in Linguistics/ Documentation 25.) Berlin and New York: Mouton de Gruyter. Nikolaeva, Irina and Maria Tolskaya 2001 A Grammar of Udihe. (Mouton Grammar Library 22.) Berlin and New York: Mouton de Gruyter. Noonan, Michael 2003a Chantyal. In: Graham Thurgood and Randy J. LaPolla 2003 (eds.), 315–335. Noonan, Michael 2003b Nar-Phu. In: Graham Thurgood and Randy J. LaPolla 2003 (eds.), 336–352.

Pandharipande, Rajeshwari V. 1997 Panfilov, V. Z. 1962 Peterson, John 2006

Marathi. (Descriptive Grammars.) London and New York: Routledge. Grammatika nivkhskogo iazyka 1. Moscow and Leningrad: Nauka.

Kharia. A South Munda language. Volume 1: Grammatical Analysis. University of Osnabru¨ck. Pierrehumbert, J. B. and M. Beckman 1988 Japanese tone structure. LI monograph series #15. Cambridge, MA: MIT press. Pinnow, Heinz-Ju¨rgen 1959 Versuch einer historischen Lautlehre der Kharia-Sprache. Wiesbaden: Harrassowitz. Pinnow, Heinz-Ju¨rgen ¨ ber den Ursprung der voneinander abweichenden Strukturen 1960 U der Munda- und Khmer-Nikobar-Sprachen. Indo-Iranian Journal 4: 81–103. Pinnow, Heinz-Ju¨rgen 1963 The position of the Munda languages within the Austroasiatic language family. In: H. L. Shorto (ed.) Comparison in Southeast Asia and the Pacific, 140–152. London: SOAS. Poppe, Nicholas 1960 Vergleichende Grammatik der Altaischen Sprachen. Harrassowitz: Wiesbaden. Poppe, Nicholas 1964 Bashkir Manual. Bloomington: Indiana University. Poppe, Nicholas 1970 Mongolian Language Handbook. Washington, D.C.: Center for Applied Linguistics.

610

Rene´ Schiering and Harry van der Hulst

Rabel, Lili 1961

Khasi: A Language of Assam. (Louisiana State University Studies, Humanities Series 10.) Baton Rouge: Louisiana State University Press.

Radlo¤, Carla F. 1999 Aspects of the sound system of Gilgiti Shina. (Studies in the Languages of Northern Pakistan, 4.) Dallas: Summer Institute of Linguistics. Ramat, Anna Giagalome and Paolo Ramat (eds.) 1998 The Indo-European Languages. London: Routledge. Ramsey, S. R. 1978 Accent and Morphology in Korean Dialects. Seoul: Tower Press. Ramsey, S. Robert 1987 The Languages of China. Princeton: Princeton University Press. Refsing, Kirsten 1986 The Ainu language. The morphology and syntax of the Shizunai dialect. Aarhus: Aarhus University Press. Rehberg, Kerstin 2003 Phonologie des Kharia. Prosodische Strukturen und segmentales Inventar. Master’s Thesis, University of Osnabru¨ck. Robbeets, Martine 2005 Is Japanese related to Korean, Tungusic, Mongolic and Turkic? (Turcologica 64.) Wiesbaden: Harrassowitz. Robbeets, Martine 2007a How the actional su‰x chain connects Japanese to Altaic. Turkic Languages 11: 3–58. Robbeets, Martine 2007b The causative-passive in the Trans-Eurasian languages. Turkic Languages 11: 235–278. Ro´na-Tas, Andra´s 1998 The reconstruction of Proto-Turkic and the genetic question. In: ´ . Csato´ 1998 (eds.), 67–80. Lars Johanson and E´va A Ruhlen, Merritt 1991 A guide to the world’s languages. Volume 1: Classification. Stanford, CA: Stanford University Press. Schiering, Rene´ 2006 Phonological domains in Mon. Autotyp Language Reports. Schiering, Rene´ 2007 Phonological domains in Vietnamese. Autotyp Language Reports. Schiering, Rene´, Kristine A. Hildebrandt and Balthasar Bickel 2007 The prosodic word is not universal. Manuscript, University of Leipzig and University of Manchester. Schmidt, Wilhelm 1906 Die Mon-Khmer-Vo¨lker, ein Bindeglied zwischen Vo¨lkern Zentralasiens und Austronesiens. Archa¨ologische Anthropologie 5, 59–109.

Word accent systems in the languages of Asia

611

Schmitt, Ru¨diger 2000 Die iranischen Sprachen in Geschichte und Gegenwart. Wiesbaden: Reichert. Schmitt, Ru¨diger 2004 Old Persian. In: Roger D. Woodard 2004 (ed.), 717–741. Scho¨nig, Claus ´ . Csato´ 1998 (eds.), 1998a Azerbaijnian. In: Lars Johanson and E´va A 248–260. Scho¨nig, Claus ´ . Csato´ 1998 (eds.), 261– 1998b Turkmen. In: Lars Johanson and E´va A 272. Sezer, Engin 1981 On non-final stress in Turkish. Journal of Turkish Studies 5: 61– 69. Shibatani, Masayoshi 1990 The Languages of Japan. New York: Cambridge University Press. Shiraishi, Hidetoshi 2006 Topics in Nivkh phonology. (Groningen Dissertations in Linguistics 61.) Groningen: Rijksuniversiteit Groningen. Sidwell, Paul 2006 Lecture series: Mon-Khmer languages. www.mon-khmer.com, 10.06.2007. Sims-Williams, Nicholas 1993 The Iranian languages. In: Anna Giacalone Ramat and Paolo Ramat 1993 (eds.), 125–153. Sims-Williams, Nicholas (ed.) 2002 Indo-Iranian Languages and Peoples. Oxford: Oxford University Press. Skorik, P. Ja. 1961 Chuckchi Grammar. Part 1: Phonetics and Morphology of Nominal Parts of Speech. USSR Academy of Science ( Грамматика чукотского языка. Ч.1. Фонетика и морфология именных частей речи. Изд-во Академии наук СССР. 1961.) Smyth, David 2002 Thai: An Essential Grammar. (Essential Grammars.) London and New York: Routledge. Sridhar, S. N. 1990 Kannada. (Descriptive Grammars.) London and New York: Routledge. Stachowski, Marek and Astrid Menz ´ . Csato´ 1998 (eds.), 417– 1998 Yakut. In: Lars Johanson and E´va A 433. Steever, Sanford B. 1998a Gondi. In: Sanford B. Steever 1998 (ed.), 270–300.

612

Rene´ Schiering and Harry van der Hulst

Steever, Sanford B. 1998b Introduction to the Dravidian languages. In: Sanford B. Steever 1998 (ed.), 1–39. Steever, Sanford B. 1998c Kannada. In: Sanford B. Steever 1998 (ed.), 129–157. Steever, Sanford B. 1998d Malto. In: Sanford B. Steever 1998 (ed.), 359–387. Steever, Sanford B. (ed.) 1998 The Dravidian Languages. (Routledge Language Family Descriptions.) London and New York: Routledge. Street, John. 1963 Khalkha structure. Bloomington. Indiana University Publications. Subrahmanyam, P. S. 1998 Kolami. In: Sanford B. Steever 1998 (ed.), 301–327. Sun, Jackson T.-S. 2003 Caodeng rGyalrong. In Graham Thurgood and Randy J. LaPolla 2003 (eds.), 490–502. Svantesson, Jan-Olof, Anna Tsendina, Anastasia Karlsson and Vivan Franze´n 2005 The phonology of Mongolian. Oxford: Oxford University Press. Tamura, Suzuko 2000 The Ainu language. (ICHEL Linguistic Studies 2 Tokyo: Sanseido. Testen, David 1997a Old Persian and Avestan phonology. In: Alan S. Kaye 1997b (ed.), 569–600. Testen, David 1997b Ossetic phonology. In: Alan S. Kaye 1997b (ed.), 707–732. Thomas, David D. 1971 Chrau Grammar. (Oceanic Linguistics Special Publication 7.) Honolulu: University of Hawaii Press. Thompson, Laurence C. 1965 A Vietnamese Grammar. Seattle: University of Washington Press. Thurgood, Graham 2003 A subgrouping of the Sino-Tibetan languages: the interaction between language contact, change, and inheritance. In Graham Thurgood Graham and Randy J. LaPolla 2003 (eds.), 3–21. Thurgood, Graham and Randy J. LaPolla (eds.) 2003 The Sino-Tibetan Languages. (Routledge Language Family Series.) London: Routledge. Vajda, Edward J. 2004 Ket. (Languages of the World/Materials 204.) Mu¨nchen: Lincom Europa. van Driem, George 1987 A Grammar of Limbu. (Mouton Grammar Library 4.) Berlin: Mouton de Gruyter.

Word accent systems in the languages of Asia

613

Ventzel, T. V. 1983 The Gypsy Language. Moscow: Nauka. Wali, Kashi and Omar N. Koul 1997 Kashmiri. A Cognitive-Descriptive Grammar. (Descriptive Grammars.) London and New York: Routledge. Walker, Rachel 1995 Mongolian stress: Typological implications for nonfinality in unbounded systems. Phonology at Santa Cruz 4, 85–102. Watkins, Calvert 1993 Proto-Indo-European: Comparison and reconstruction. In: Anna Giacalone Ramat and Paolo Ramat 1993 (eds.), 25–73. Watters, David E. 2002 A Grammar of Kham. Cambridge: Cambridge University Press. Werner, Heinrich 1996 Vergleichende Akzentologie der Jenissej-Sprachen. (Vero¨¤entlichungen der Societas Uralo-Altaica 46.) Wiesbaden: Harrassowitz. Werner, Heinrich 1997a Das Jugische (Sym-Ketische). (Vero¨¤entlichungen der Societas Uralo-Altaica 50.) Wiesbaden: Harrassowitz. Werner, Heinrich 1997b Die ketische Sprache. (Tunguso Sibirica 3.) Wiesbaden: Harrassowitz. Windfuhr, Gonnet L. 1997 Persian phonology. In: Alan S. Kaye 1997b, 675–690. Woodard, Roger D. (ed.) 2004 The Cambridge Encyclopedia of the World’s Ancient Languages. Cambridge: Cambridge University Press. Yip, Moira 1980 The Tonal Phonology of Chinese. Bloomington: Indiana University Linguistics Club. Yip, Moira 2002 Tone. Cambridge: Cambridge University Press. Zhukova, A. N. 1972 The Grammar of Koryak. Phonetics and Morphology. Leningrad, ‘‘Nauka’’. ( Жукова А.Н. 1972 Грамматика корякского

языка. Фонетика. Морфология. Издательство «Наука», Ленинградское отделение, Ленинград)

Zhukova, A. N. 1980 The Language of Palansk Koryak. Leningrad, Nauka. (A. H. Жукова. 1980. Язык поланских коряков. Ленинград, «Наука».)

11.

Word accent systems in the Middle East

Harry van der Hulst and Sam Hellmuth 1. Introduction1 This chapter deals with accentual systems in the languages spoken in the Middle East and includes a discussion of a number of isolate ancient languages and some Afro-Asiatic languages, notably Egyptian and Semitic languages. The notion ‘Middle East’ is not well-defined linguistically, and some languages that could have been included here are dealt with in other chapters. All Altaic languages, Indo-Iranian languages (such as Kurdish and Avestan) are discussed in Schiering and van der Hulst (this volume). All Caucasian languages as well as Indo-European languages such as Hittite (and related Anatolian languages) and Armenian are treated in van der Hulst (this volume). Afro-asiatic languages spoken in North Africa can be found in Downing (this volume). This survey, incomplete as it is, was included because the relevant area was not included in the areas covered in the other chapters of this volume. The organization of each section (or subsection) with accentual data is as follows: a. Genetic structure of the (sub)family; these are based on sources such as Ruhlen (1991), Comrie et al (2003), the Ethnologue (15th edition) and several others of the many sources that o¤er classifications. We have tried to strike a compromise in cases of conflicting groupings and no claim is made here that the resulting groupings are superior to those o¤ered in other sources. In each case (sub)family names are presented in capitals. We have not included information of the numbers of languages per (sub)family and in most cases list only a (sometimes rather arbitrary) subset of the languages in each (sub)family. Languages that are included in StressTyp (see section 3) are indicated in bold. Languages that are not in StressTyp about which this chapter provides information are underlined.

1. We wish to acknowledge the help of various people who gave feedback on parts of this chapter: Bob Hoberman, Janet Watson and Alan Kaye.

616

Harry van der Hulst and Sam Hellmuth

b. Extracts from StressTyp entries (language name followed by StressTyp Code, and examples). More complete extracts (including references) are o¤ered in Part II of this volume and, of course, in StressTyp itself. We included the Stress information unchanged, i.e. as it is in the database and Part II, despite the fact that in some cases this information has been questioned by specialists. (Future work on StressTyp must, of course, aim at removing or changing such information.) c. Additional accent information: This might involve additional information on languages already in StressTyp, or information on languages that are not in StressTyp. (In some larger families that are treated in one section we have presented the StressTyp extracts and additional information per subfamily.) d. Generalizations: We have tried to make general statements about the accentual patterns in the relevant (sub)family, in some cases accompanied by remarks about diachronic developments. If (b) is absent, (c) will be labeled ‘accent information’. Sometimes (c) and (d) are conflated into a single section. 2. A note on the information on which this chapter is based The first source for the accentual data presented here is StressTyp (see chapter 1, 2, 11 and Part II of this book for information on this database). Secondly, we consulted books that o¤er surveys of language families or languages in a specific geographical area. Thirdly, we have consulted grammars of individual languages and, fourthly, we have sent email queries to colleagues; where we rely on information that they directly have given to us (via email or in personal communication) we note this in the text. In line with previous work on word accentual systems (van der Hulst 1999, 2002, 2005), we use the term ‘word accent’ where many others would use the term ‘word stress’ (as in done in StressTyp). We refer to chapter 1 of this volume for a justification and clarification of this terminological choice. 3. Isolate ancient languages There are many now extinct languages from this area. Here we provide some information about four languages. Others that are not dealt with include: Median, Ancient Macedonian, Had(r)ani, Minaean, Old Nubian,

Word accent systems in the Middle East

617

Sabaean, many of which are known too fragmentedly to provide researchers with information on word accent. USumerian. This language isolate (once suspected to be an artificial, secret language), spoken in Southern Mesopotamia, is known first from clay tablets found in Uruk from around 3200 BCE. It was superseded by the unrelated Semitic language Akkadian (cf. below) in the beginning of the 2nd millenium, but remained in limited use for 3000 years. Michalowski (2004) dismisses the suggestion that Sumerian, an agglutinative language, might have been a tonal language which has been suggested because the language otherwise has an unusually high number of homophones. He does not provide further information about word prosodic properties. Hayes (1997: 1013) agrees that very little is known about word accent in Sumerian. Vowel deletions suggest a strong stress accent: amar-utu-(k) ‘bull of the Sun’ > Akkadian: marduk suggests a second syllable accent: ama´rutuk. Not enough such examples are known to suggest whether stress was lexically marked or predictable by rule. A recent grammar is Edzard (2003). UElamite (also known as Scythian, Median, Susian, Anzanite), the o‰cial language of the Persian empire from the 6th to the 4th century BCE. A genetic link to the Dravidan family has been suggested. Stolper (2004: 73) thinks that accent could have been non-final, perhaps initial. UHurrian and UUrartian (Vannian, Chaldean). These languages are related to each other, with no known genetic connection to any other language family, although links to Northeast Caucasian (notably Georgian) have been proposed. Wilhelm (2004a: 100) says that Hurrian had a penultimate accent on words (including su‰xed words), but excluding enclitics. Wilhelm (2004b: 123) suggest the same pattern for Urartian.

4. Afro-Asiatic: Egyptian and Semitic Genetic information The Afro-Asiatic family contains: EGYPTIAN SEMITIC BERBER CHADIC CUSHITIC OMOTIC

618

Harry van der Hulst and Sam Hellmuth

Omotic is sometimes regarded as a sister to the rest forming a group, while Semitic and Berber are usually seen as a subgroup. This family extends over North Africa, the Middle East and Asia Minor. In this section we will deal with the Afro-Asiatic languages that are spoken outside Africa, roughly in the Middle East. This means that we will deal with the Egyptian branch and the Semitic branch (excluding Ethiopian Semitic and Western Arabic languages). Egyptian, Semitic and Berber do not have tonal languages, whereas the other three subfamilies do. We refer to Downing (this volume) for some brief remarks about Afro-Asiatic languages spoken outside Africa and more elaborate discussion of Afro-Asiatic languages in Africa. StressTyp contains information about one Cushitic language outside Africa, Beja (spoken in parts of Egypt, Sudan and Eritrea):2 Beja; Bedawi; Bischari [A/P] f Stress falls on heavy penultimate syllables. f If the penult is light, stress falls on the antepenultimate syllable. f In bisyllabic words stress is penultimate unless the penult is light and the final is not. ga¨naøj

‘gazelle’

¨enton

‘here’

¨an/alan ‘I cursed’ sa¨noøku ‘your brother’ We refer to Downing (this volume) for a discussion of this and other Cushitic languages. 4.1. Egyptian (< Afro-Asiatic) Genetic information ANCIENT EGYPTIAN UAncient Egyptian, UCoptic There is a written record from around 3200 to 400 BCE. A distinction is made (with further divisions) between early Egyptian (3200–1300 BCE) and late(r) Egyptian (1300 BCE–5th century AD), the last phase being called Demotic Egyptian. Coptic, as known since the 4th century AD, the latest phase of this language, went extinct in the late 17th. (The language name ‘‘Modern Egyptian’’ refers to a form of Arabic; cf. below.). See Loprieno (1995) for a general overview.

2. This language is sometimes seen as a separate branch of Afro-Asiatic.

Word accent systems in the Middle East

619

Accent information Loprieno (1997, 2004) provides the following information about ‘earlier’ or ‘historical’ Egyptian. The accent was on the ultimate or penultimate syllable: (1) wabaa´X ‘to become white’ saa´tpaw ‘is chosen’

jafdaa´w ‘four’ saa´øtap ‘to choose’

CVC and CV syllables occur in all positions in the word, but accented penultimate vowels in open syllables are always long. Word-finally, an extra C could be present, thus allowing CVCC. Perhaps there was an earlier APU accent location which became integrated in the above pattern by loss of the PU vowel: (2) Xu´piraw > Xu´praw ‘transformation’ Indeed, in later periods unaccented vowels would reduce and delete which leads to complex consonant clusters, as can be noted in Coptic. 4.2. Semitic (< Afro-Asiatic) Genetic information SEMITIC EAST: UAkkadian, UEblaite WEST: UAmorite, UUgaritic ARAMAIC: Western: UNabataean Western Middle Aramaic languages Samaritan Aramaic, UJewish Palestian Ar. Western Neo-Aramaic: Ma’aloula Eastern: UBiblical A., Syriac, Turoyo, UMlahso, Mandaic, Judaeo-Aramaic CANAANITE: UEdomite, UMoabite, UAmmonite Hebrew: Biblical (Tiberian) Hebrew, Mod. Hebrew UPhoenician/Punic ARABIC: See below SOUTH Western ANCIENT SOUTH ARABIAN: Sabean, Qatabanian, Hadhramautic, Minean ETHIOPIAN: North: UGe’ez, Tigrinya, Tigre´, Dahlik South: Amharic, Harari, Gurage Eastern: Mehri, Harsusi, Bathari, Hobyot, Jibbali, Soqotri

620

Harry van der Hulst and Sam Hellmuth

Here Arabic is seen as a separate branch of Central Semitic. Woodard (2004) has a distinction between Ancient South Arabian and Ancient North Arabian. However, the latter group concerns older forms of Arabic, which is here placed in Central Semitic. See also Bergstra¨sser (1983), Hetzron (1997) and Faber (1997). The living languages that are spoken outside Africa are Neo-Aramaic languages, Hebrew, Eastern South Semitic and varieties of Arabic. Accent information East Semitic UAkkadian (which splits into Babylonian and Assyrian), known from the later half of the 3rd millennium, replaced Sumerian (see above) in the beginning of the 2nd millennium BC and was then itself replaced by Aramaic during the first half of the first millennium (although surviving until 100 AD) (G. Gordon 1997). Buccellati (1997) subscribes to the view that accent falls on the rightmost long vowel (excluding the final morphemic vowel). According to an alternative view accent is initial, but it might be the case that this perhaps obtains in the absence of long vowels. This would indicate that Akkadian had an unbounded last/first system. This is confirmed by Huehnergard and Woods (2004: 234) who say that accent falls on the ultimate syllable if closed and has a long vowel (superheavy), or when containing a circumflex vowel (a vowel resulting from contraction, perhaps counting as two syllables). Otherwise the accent falls on the rightmost nonfinal syllable which is either closed or contains a long vowel. If there are no heavy syllables at all, accent is initial: (3) idu #k ‘s/he killed’ ibnuˆ= ‘they built’ ipa´rras ‘s/he will cut’ ma #rum ‘son’ ´ı lu na´din ‘is given’ ‘gods’ Knudsen (1980) also presents an analysis of Akkadian accent. He notes that there is no evidence for secondary accents. Primary accent falls on the final or penultimate syllable. If the final syllable was short and the penultimate had a long vowel, accent was penultimate. A final closed syllable with a long vowel would have primary accent. This would also be the case if the final syllable contained a circumflex vowel. He also notes that accent would be penultimate if the last two syllables were light, even if the antepenultmate vowel were long. In case words only have light

Word accent systems in the Middle East

621

syllable, he notes, that there is no evidence for accent. These observations are not incompatible with those made by Huehnergard and Woods (cf. above), except that their unbounded account would suggest that in the last mentioned case (heavy–light–light) accent would be antepenultimate. Knudsen’s observations suggest a bounded rather than an unbounded system. No information on accent could be found for UEblaite and UAmorite (only known through proper names in Akkadian texts); cf. Gordon (1997a,b,c). West Semitic: Central StressTyp extracts Aramaic [P] Afro-Asiatic, Semitic, Central, Aramaic. East Kurdistan, Israel. Palmyra, Sinai. f Stress falls on the penultimate syllable. f Epenthetic vowels do not count for stress assignment. ¨arba ‘four’

cˇay¨xaøna

‘teashop’

da¨qiqa

‘fine’

We did not find information on UAmmonite, UUgaritic (Gordon 1997a,b; Dennis 1997, 2004). Turning to Aramaic, Creason (2004: 400) says that in Proto-Aramaic a final closed syllable was accented, otherwise accent was on the penult (even if the final vowel was long). Then final short vowels would be deleted, or lengthened. In case of deletion, the penultimate open syllable now would become a final closed syllable, with accent. This basic pattern (final if closed) remains constant throughout the history of Aramaic, but in some late Aramaic dialects, final accent shifted to the penult in some or all words. In line with this Kaufman (1997: 121) states that Classical Aramaic is said to have final accent (where he apparently ignored the penultimate location when the final syllable is open), while the modern languages have penultimate stress. Western Aramaic: In his discussion of Neo-Aramaic languages (E.g. Ma’aloula) Jastrow (1997: 336) says that in Western Neo-Aramaic word accent is usually on the penultimate. The last syllable is accented if it has a long vowel or ends in two or more consonants.

622

Harry van der Hulst and Sam Hellmuth

Eastern Aramaic: Daniels (1997: 137) discussing Syriac accent refers to the view that accent is penult among the ‘Nestorians’, but on final closed syllables and otherwise on the penult for ‘Maronites’. The latter pattern is the norm today among ‘Chaldeans’ (these terms referring to di¤erent Christian traditions or groups). Malone (1997: 156) proposes that accent in Classical Mandaic falls on the last vowel of the word. In Modern Mandaic, according to Malone (1997: 149), accent falls on the rightmost, non-final full vowel. This suggests an unbounded system, but no default clause is mentioned in case no full vowel is present. As for the Eastern Neo-Aramaic (ENA) languages Jastrow (1997: 353) reports that in Mlahso and in the North-Eastern (NENA) languages nouns are accented on the last syllable, but in the majority of ENA languages accent has shifted to the penult. Accent moves to the (new) penult when su‰xes are added, although in verbal forms accent will stay on the original penult when further su‰xes are added. This penult location is, for example, found in Turoyo: (4) Turoyo: h` oze´no ‘I (m.) see’

h` oze´nole

‘I (m.) see him’

He also says that perhaps in these varieties accent has become phonemic in the sense that across verb classes di¤erent patterns may obtain: (5) Turoyo: ma´lı˘m

‘he collects’

malı˘m ‘collect!’

Whereas Jastrow says that in NENA languages accent is final in nouns, Hoberman (1997: 330 ¤.) reports that accent in the NENA dialects spoken in northwestern Iraq, except in verbs, uniformly falls on the penult. In verbs, where penult is still the default, accent placement is governed by the morphology. Consider the following ‘minimal’ pair (representative of Modern Aramaic NENA dialects): (6) ma´ølçple ‘teach him (imp. sing.)’

maølı´ple ‘that he teach him’

He adds the following generalizations: (7) a.

In the imperative form accent is initial

b.

Accent is APU when a surface penult /i:/ or /u:/ is derived from an underlying non-syllabic semi-vowel.

c.

Certain verbal su‰xes cannot be accented, such as the past tense su‰x /wa:/. When such su‰xes occur accent occurs earlier in the word.

Word accent systems in the Middle East

623

Two word combinations in which the second word cliticizes to the first show a reduction of the accent on the second word. Canaanite StressTyp extracts Hebrew, Modern [U;P] Afro-Asiatic, Semitic, Central, South, Canaanite. Israel. f

f

Primary stress is mostly final, but sometimes phonemically on the penult. Secondary stress on alternate syllables to the left of the main stress.

g a¨dol

‘big’

ta¨am

me˙vuga¨rim

‘adults’

¨taam ‘taste’

‘tasted’

Hebrew, Tiberian (dialect of Hebrew; Ivrit) [U/P] Afro-Asiatic, Semitic, Central, South, Canaanite. Lake Tiberia (Israel). f f f

Main stress is assigned to the final syllable if it is closed. Otherwise stress is penultimate (but may shift again to the final syllable). Secondary stress two syllables to the left of the main stress and long vowels before that.

¨qaømuø

‘arise third.pl.’

ktab¨tem ‘write scnd pl masc’

Additional information UPunic is a late dialect of UPhoenician through we which we get most information about earlier phases of Phoenician (Hackett 2004; Segert 1997a,b). This language disappears in the eastern Mediterranean area during the first century BCE and in North Africa in the fifth century AD. Segert (1997: 63) suggests that the position of word accent can be determined on the last syllable in most cases. We have no information on UMoabite, UEdomite and UAmmonite. Turning to Hebrew, Steiner (1997: 149) says that in Biblical (Tiberian) Hebrew primary stress is on the ultimate or penultimate syllable, with an increased tendency toward the ultimate; see also Khan (1997) and Rendsburg (1997). In Modern Hebrew (Bat-El 1993, Bolozky 1997, Berman 1997, Co‰n and Bolozky 2005) word accent follows the ‘Sephard’ tradition. It is final, with (sometimes systematic) exceptions:3 3. We thank Shmuel Bolozky for supplying us with additional information.

624

Harry van der Hulst and Sam Hellmuth

(9) a.

In general, penultimate accent may result when su‰xes remain ‘neutral’ such that the accent remains on the final syllable of the base. This happens in the verbal system where (except in the present participle) accent lies on the stem-final vowel: kata´vti ‘I wrote’ diba´rta ‘you spoke’ nimce´nu ‘we were found’ ipı´la ‘she dropped (trans.)’

b.

Also in nouns that are acronyms or frequently used, accent can be ‘stable’ under su‰xation: manka´l ‘general manager’ manka´lim (plural) rasa´r ‘first sergeant major’ rasa´rim (plural) tu´t ‘water melon’ tu´tim (plural) cxo´k ‘laughter’ cxo´kim ‘joyous moment’

c.

Stable accent is also found in names for residents: telavı´v ‘Tel Aviv’ telavı´vi ‘resident of . . .’ xolo´n ‘Holon’ xolo´ni ‘resident of . . .’

d.

Certain derived (‘segolate’) nouns like me´ser ‘messenger’; here, historically a vowel broke up a final ‘impermissible’ cluster.

e.

Insertion of /a/ causes penultimate stress in another group of words: gavo´a ‘tall (m.sg.)’ pote´ax ‘open (m.sg.)’

f. Penultimate accent occurs in names of children’s games: Shı´ra (name) and kla´tim (children’s game); compare: klafı´m ‘cards’ kla´fim ‘gard game’ monopo´l ‘monopoly’ mono´pol ‘Monopoly (game)’ g.

Due to the influence of Yiddish accent, we find a penultimate location in colloquial style of proper names: Yae´l P Ya´el xaı´m ‘life’ Xa´im ‘proper name’ In general it would seem that in informal speech, penultimate accent may take over from the ‘more formal’ final accent.

h.

In nonverbal forms there are sporadic exceptions to the final accent pattern: la´ma ‘where’ (h)e´na ‘here’

i.

Loan words (like akade´miya, instala´tor) can be exceptional,

Word accent systems in the Middle East

625

In words that have exceptional accent (when no su‰xes are involved) accent appears to be stable under su‰xation (10) tı´ras tı´ras – im ‘corn – plural’ This especially applies to borrowings which keep their original accent location even if non-final and this position is also maintained under su‰xation. Bat-El (1993) analyses the accent system of Modern Hebrew nouns, which involves lexical marking of stems and su‰xes. It would seem that primary accent location follows a LAST/LAST pattern: it falls on the last lexically accented syllable, or, if no lexical accents are present, on the last syllable. Modern Hebrew has regular alternating secondary accents going leftward away from the primary accent. South Semitic: For the Eastern South Semitic (the modern Subarabic or south Arabian) languages, Lonnet and Simeone-Senelle (1997: 354) describe the location of accent as follows: accent falls on the final CVC(C) or CV: (C) syllable and if there is no such syllable on the first CVC syllable. This suggests an unbounded Last/First pattern, although no mention is made of the location in words that also lack CVC; some examples suggest that accent is initial in that case. Simeone-Senelle (1997: 386) states that in Mehri, Hobyo#t, Batƒh` ari and H`arsu#si accent is on the last strong syllable, or on the first syllable if all vowels are short. This confirms the Last/First pattern, though Janet Watson (p.c.) points out that in at least some dialects of Mehri the Last/First pattern only holds of stems (not inflected words) and that in any case strings of short syllables are rare and thus initial stress is itself rare. In contrast, Simeone-Senelle (1997: 386) notes that in Jibba#li a word can have several accents, whereas in Soqot` there is a general trend to have initial accent. As for the Old/Ancient South Arabian languages belonging to the Western South Semitic, Graag (1997) and Nebes and Stein (2004) provide no information on accent. 4.3. Arabic languages (< Semitic < Afro-Asiatic) Genetic information A standardized version of Arabic, the language of the Quran and early Islamic literature arose in the 8th century. This version came to be known as Classical Arabic (Fischer 1997; Owens 2006) which develops in the 19th

626

Harry van der Hulst and Sam Hellmuth

and 20th century into Modern Standard Arabic which is used in most sectors of public life. Meanwhile, many more conservative Bedouin (rural) and sedentary (urban) varieties of Arabic develop which are usually grouped into a Western and the Middle Eastern group. The Western group (the Maghrebi or North African dialects) covers Libya, Tunisia, Algeria, Morocco, Mauritania, and the outlying Maltese variety (Borg and Azzopardi-Alexander (1997) (as well as now extinct Andalusian Arabic and Siculo Arabic). The Middle Eastern group covers the Arabian Peninsula, Iraq, Iran, Sudan, Syria-Lebanon-Palestine, Jordan, Egypt and areas in Afghanistan and Uzbekistan, as well as an outlying variety in Cyprus (Alexander (1997). Most entries in StressTyp (nearly all of which are analyzed in Hayes 1995) come from the Middle Eastern group which can be geographically subdivided into a southern group (Arabian Peninsula), a northern group (Levant and Mesopotamia) and a central group (including Egypt and Sudan). Linguistic studies on Arabic have a rich tradition (cf. Owens 2006; Versteegh et al. 2009) and in more recent times the variety of accentual systems has attracted wide attention (cf. Watson, to appear for an overview of di¤erent approaches and some case studies). Kaye (1997) and Kaye and Rosenhouse (1997) provide general discussions of the phonology. Studies of specific accentual systems can be found in Birkeland (1954), Janssens (1972), Angoujard (1990), Hayes (1995) and Kager (2009). StressTyp extracts Arabic, Classical [L/F] f Stress the rightmost non-final syllable that has a long vowel or closing consonant. f If there are no such syllables, stress the first. ki¨taabun ‘book’ ¨mamlakatun ‘kingdom’ ¨balaatun ‘date’

ta¨mamtumaa ‘you both completed’

Middle East, Southern Arabic, Bedouin Hijazi (dialect of Arabic) [U-P/A] f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. f In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult.

Word accent systems in the Middle East

mak¨tuøb

‘written’

D‡a¨rabt

¨maølana gaø¨bilna

‘our property’ ‘meet us m.s.’

mak¨tuøfah ‘tied f.s.’ ¨kitab ‘he wrote’

¨bgarah

‘cow’

¨˜aza

627

‘I hit’

‘he raided’

Arabic, Gulf [U/P] f If the vowel of the final syllable is long, it bears stress. f In all other cases stress is on the penult. a¨jaattin ‘their life’ jSuu¨fuun ‘they see’ ¨beettum ‘their house’ mus¨taSfa ‘hospital’ ¨qaæbitS ‘your heart’ mithaaw¨Siin ‘having quarrelled with each other’ Middle East, Northern: Arabic, Beirut/Lebanese (dialect of Arabic, North Levantine) [U-P/A]4 f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. f In other cases, stress falls on heavy penults, otherwise on the antepenult. ¨d‡arab sˇa¨Zaratun al¨lam na

‘hit’ ‘tree’ (Classical) ‘we teach’

d‡a¨rabna ‘he hit’ ¨d‡arabu ‘they hit’ ¨ allamu ‘they teach’

Arabic, Damascene (dialect of Arabic, North Levantine) [U-P/A] f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. f In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult. ma¨daøres ‘schools’ dar¨rast ‘I/you m.s. taught’ mut¨taide ‘united’ al¨lamna ‘teach pl.’ Arabic, South Levantine; Palestinian Arabic [L (CNT)] f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. 4. StressTyp has two di¤erent entries for Beirut and Lebanese Arabic which we have here collapsed because they represent the same dialect.

628 f f f

f

Harry van der Hulst and Sam Hellmuth

Otherwise, stress penultimate syllables if heavy, and in bisyllabic words. Heavy antepenults followed by two light syllables carry stress. Four syllable (Classical) words with only light syllables have initial stress. All other words ending in three light syllables have antepenultimate stress.

da¨rast ‘I studied’ ¨katab ‘he wrote’ ¨sˇajaratun ‘a tree’ (Classical)

¨maktab ¨baørako sˇaja¨ratuhu

‘o‰ce’ ‘he blessed him’ ‘his tree’ (Classical)

Arabic, Negev Bedouin [L (CNT)] f Stress the last syllable with a long vowel or vowel cluster (VVC or VCC). f Otherwise stress the second syllable. f There is some variation in four syllable words. ˜ana¨maøt ‘sheep (several)’ ˜a¨nam ‘sheep’ fa¨rasah ‘his horse’ an¨kitalaw ‘he was killed’ Arabic, Bani Hassan [U-P/A/P]5 f Stress a final superheavy syllable. f Stress the penultimate light syllable if not preceded by a heavy syllable. f Stress penultimate or antepenultimate heavy. Si¨baøb ‘youth’ al¨lamt ‘I taught’ mak¨tabti ‘my library’ mi¨kaøtib ‘o‰ces’ ¨walad ‘boy’ ¨smikah ‘a fish’ alla¨matuh ‘she taught him’

baøra¨katuh

‘she blessed him’

5. This system is coded in StressTyp as [L/F] but no examples are given and we find the following verbal description: f f

Stress the rightmost syllable with a long vowel or coda, otherwise the first. Long vowels preceding the main stress have secondary stress.

The generalisation above is based on the data in Irshied & Kenstowicz (1984). Hayes (1995) sees Bani Hassan Arabic as similar to Palestinian Arabic with the di¤erence that in a final HLL] sequence the accent is on the H rather, as in Palestinian Arabic, on the penultimate light. He also says that accent is penultimate in HLLL]. We discuss an analysis below.

Word accent systems in the Middle East

629

Middle East, Central Arabic, Egyptian Radio (dialect of Arabic, Egyptian Spoken) [L (CNT)] f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. f Otherwise, stress penultimate syllables if heavy, and always in bisyllabic words. f Three syllable words with only light syllables have antepenultimate stress. f Four syllable words with only light syllables have antepenultimate or pre-antepenultimate stress in free variation. f If the antepenult is heavy and the penult and final light, either it or the penult is stressed. f If the pre-antepenult is heavy, stress is always antepenultimate. f Rhythm is assigned to non-final heavy syllables and odd light ones from left to right. sa¨laøm ‘peace’ qad¨damna ‘we presented’ ¨malik ‘king’

˙kata¨bahu or ¨kata˙bahu ‘he wrote it’ ¨/abadan ‘never’ ˙mux¨talifa ‘di¤erent’

Arabic, Egyptian; Cairene Arabic (classical6) [L (CNT)] f Superheavy (long vowel þ coda, or double coda) final syllables carry stress. f Otherwise, stress penultimate syllables if heavy. f In all other cases, stress the penult or antepenult, whichever is separated from the last heavy syllable, or the left word edge, by an even number of syllables (or zero). ¨buxala ‘misers’ a¨malti ‘you f.s. did’ mux¨talifa ‘di¤erent f.s.’ mar¨taba ‘mattress’ saka¨kiin ‘knives’ Sajarata¨humaa ‘their (dual) tree nom.’ (Classical) Western group (Maghrebi) Arabic, Libyan; Cyrenaican Bedouin [L (CNT)] f Bisyllables of which the first syllable is open and the vowel is short have final stress. 6. Note that this refers to Classical Arabic as pronounced by speakers of Cairene Arabic; of the examples given only the last word is exclusively Classical.

630 f

f

f

Harry van der Hulst and Sam Hellmuth

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. Otherwise, stress heavy penults or heavy antepenults followed by two light syllables. Otherwise, stress the penult or the antepenult, whichever is separated from the closest preceding heavy syllable or the left word edge by an odd number of syllables.

tç¨rAøfçgAn ‘they accompanied’ ki¨tab ‘he wrote’ ma rç¨kçtta ‘her quarrel’

¨maktab ‘o‰ce’ ki¨tabtan ‘you wrote’ fina¨Ziøl ‘cups’

Maltese [U/P] f Stress falls on the final syllable if it has a long vowel or is closed by two consonants. f In all other cases stress is penultimate. ber¨quøq ‘apricot’ ˜a¨zaqt ‘I have dug’ ¨tifla ‘girl’ Additional information Middle East, Southern San‘aani Arabic is spoken in the Old City of the capital city of Yemen (Watson 2002, to appear), and has a complex stress system, summarized in the following algorithm from Watson (2002: 82): a) stress the rightmost non-final CVV or CVG syllable ([¨haøkaDahaø] ‘like this’, [mit¨axxiraøt] ‘late f.pl.’), otherwise, b) stress a final CVVC/CVCC syllable ([ba¨naøt] ‘girls’), otherwise c) stress the rightmost non-final CVC syllable up to the antepenult ([¨madrasih] ‘school), otherwise d) stress the leftmost CV syllable ([¨ragabatih] ‘his neck’). However, as Watson (2002: 81) points out, aside from the unusual fact that a CVV/CVG can attract stress away from a final superheavy, the San‘aani stress system is in other respects similar to a Last/First pattern: a) stress a final superheavy CVCC or CVVC syllable, otherwise b) stress the rightmost non-final heavy syllable (up to the antepenult), otherwise c) stress the leftmost light syllable. Watson (2002: 98–121) analyses the variable behaviour of CVG/CVV vs. CVC syllables in San‘aani by appeal to a two-layer metrical grid (Hayes 1995). She notes that there is considerable fluctuation in stress position in connected speech, particularly in pre-pausal and post-pausal position, and that secondary stress is observed in San‘aani in words containing two or more feet.

Word accent systems in the Middle East

631

The dialect that McCarthy (1979: 461) refers to as Yemen Plateau Arabic is a cover term for a grouping of dialects spoken in the high plateau (‘Hochebene’) regions of northern Yemen, taking in the following geographical areas of Yemen (Diem 1973: 127): San‘aa’ and environs, the plateau areas to the north and south of the capital, the North East and al-Jawf, and the South East (Al-Bayda and Hari:b). Diem (1973: 11) gives the following basic algorithm for all Yemeni dialects: a) stress a final superheavy CVCC or CVVC syllable, otherwise b) stress the rightmost non-final heavy syllable, else c) stress the first syllable of the word. Diem (1973: 11) notes particularly that in the high plateau areas (only) stress is not confined to the last three syllables of the word but can occur further forward in the word, as in the example [¨amalateh] ‘she carried it’. This is the evidence which causes McCarthy (1979: 461) to cite Yemen Plateau Arabic as one of the few contemporary spoken dialects which still show the Classical Last/First stress assignment pattern (beyond the last three syllables of the word), but contemporary speakers in fact vary in their accentuation of words of this type between the initial and penult (Janet Watson p.c.). Prochazka (1988) generalizes over all of the dialects of Saudi Arabia (including North Arabian, Najdi and Hijazi) and claims that all display a ‘rightmost heavy else antepenult’ pattern, matching the pattern described above for Bedouin Hijazi Arabic. Middle East, Northern The stress patterns of Baghdadi Arabic are described in general terms in Erwin (1963: 40–42) and can be paraphrased as: a) stress a final superheavy, else b) stress the rightmost heavy syllable, else b) stress the antepenult (penult in disyllables). This would place Baghdadi in the U-P/A category. McCarthy & Ra¤ouli (1964: 10–11) state that in words containing only light syllables stress will fall on the initial syllable, but give no examples of words which are longer than 3 syllables (e.g. they give [¨keleme] ‘word’). It is possible that the three syllable window does indeed apply, but that McCarthy & Ra¤oulli’s generalisation holds because, as in Palestinian Arabic (Kenstowicz 1983), any 4 syllable word containing only light syllables will undergo syncope and be reduced to 3 syllables. De Jong’s (2000) survey of North Sinai Bedouin Arabic dialects includes a number of distinctive stress patterns as potential dialectal markers. A small number of dialects in the survey are reported to have initial stress in words containing 4 light syllables (Ca´CaCaCv, without syncope/

632

Harry van der Hulst and Sam Hellmuth

resyllabification to e.g. CCv´CCv), which suggests that they retain the Classical Last/First pattern. Pickett’s (2006) survey of North Arabian Arabic dialects spoken by present-day nomadic groups in Syria and Lebanon,7 are all reported to show the Classical pattern: a) stress a final superheavy (CCVVC, CVVC, CVCC) e.g. [tSi¨Tiør] ‘many’, else b) stress the rightmost heavy syllable (CCVV, CVC, CVV, CCVC) e.g. [¨madrasa] ‘school’, else c) stress the initial syllable e.g. [¨wasat‡] ‘medium (sized)’. The Arabic dialect described in the literature as Bani Hassan Arabic is spoken by a Bedouin-origin community now settled in the north of Jordan (Kenstowicz 1983, Irshied & Kenstowicz 1984). The dialect displays trisyllabic vowel deletion, a common marker of Bedouin-origin varieties (e.g. /samak-ah/ ‘a fish’ is realised as [¨smikah], Irshied & Kenstowicz 1984: 137). The accent system no longer retains the Classical pattern however, since stress is confined to the last three syllables of the word, with a noninitial accent appearing in words containing 4 syllables: e.g. [ alla¨matuh] ‘she taught him’. Note that this pattern is not due to the special status of the 3f.s. su‰x, as it would be in Egyptian Arabic and Lebanese Arabic in which the 3f.s. su‰x always attracts stress, regardless of the syllabic structure of the word; in BHA the su‰x does not automatically attract stress: LA: [Saø¨fitu] ‘she saw him’/BHA [¨Saøfatuh] ‘she saw him’ (Irshied & Kenstowicz 1984: 129). Middle East, Central Egyptian Sa‘iidi Arabic is spoken along the Nile Valley in Upper Egypt (south of Cairo as far as Aswan), and McCarthy (1979: 461) mentions it as a spoken dialect which retains the Classical Last/First stress assignment pattern. Khalafallah (1969) describes the stress patterns as follows: a) if there is a long vowel in the word it will bear stress (no words contain more than one long vowel) e.g. [xala¨gaøtu] ‘his clothes’/[xuz¨naø] ‘we kept it’, else b) stress the rightmost closed syllable (non-final CVC or final CVCC) e.g. [ al¨lamt] ‘I taught’, [ma¨saktu] ‘I caught him’, else c) stress the first syllable [¨katab] ‘he wrote’. Crucially, c) holds in words containing 4 light syllables such as [¨katabatu] ‘she wrote it’, confirming that this is in essence a Last/First pattern (though a final CVV attracts stress).

7. Dialects of the Syro-Mesopotamian, ‘Anazi, and Shammari groups in Johnstone’s (1967) classification.

Word accent systems in the Middle East

633

Nishio (1994) reports the same Classical style pattern in the dialect spoken in the Nile Valley village of Qift (near Qena). Hamid (1984: 37) also describes the stress patterns of Sudanese Arabic in terms of a Classical style Last/First pattern, which reduces to: a) stress the rightmost heavy syllable, else b) stress the initial syllable (cf. Kenstowicz 1984: 129). However, no sample words are given that contain more than 3 light syllables, and there are a number of counterexamples to the ‘initial stress’ rule, which Hamid explains by appeal to interaction with segmental phenomena, which might be amenable to a L(CNT) analysis. Western group (Maghrebi) Boudlal (2001: 107¤.) describes the stress patterns of the dialect of Moroccan Arabic spoken in Casablanca, for words in isolation, as follows: a) stress the final syllable if it is heavy e.g. [li¨mun] ‘oranges’, otherwise, b) stress the penult (regardless of weight) e.g. [¨r«mla] ‘sand’. Only CVC syllables containing full vowels are heavy; C«C is treated as light. For isolation forms then, Moroccan Arabic displays a U/P system similar to that observed in Maltese; in connected speech Boudlal found stress was invariably wordfinal. Guella (m.s.) describes the stress patterns of the dialect of Algerian Arabic spoken near Tlemcen as follows, barring some morphologically constrained exceptions: a) stress the leftmost long vowel (e.g. [m«S¨duødiøn] ‘tied up m.pl.’), otherwise b) stress the penultimate syllable (e.g. [t«q¨lebkum] ‘she will overturn you pl.’). Talmoudi (1980) describes the stress patterns of the dialect of Tunisian Arabic spoken in the Old City of Sousa as follows: a) stress the leftmost syllable containing a long vowel, otherwise b) stress the initial syllable. Closed syllables, both CVC and C«C, are treated as light. It is not clear whether the patterns described hold for isolation forms or in connected speech. Abumdas (1985) gives an account of the dialects of Libyan Arabic spoken in Tripoli, Ben Ghazi and Zliten (a Bedouin-origin variety). He notes that there are stress minimal pairs (as also reported by Mitchell (1960) for Cyrenaican Bedouin Arabic), e.g. [¨xalaq] ‘creating’ P [xa¨laq] ‘he created’, but states that stress is nonetheless predictable in many cases, giving rules parallel to those put forward for Cyrenaican Bedouin Arabic (¼Eastern Libyan Arabic) by Mitchell (1960) and Owens (1984).

634

Harry van der Hulst and Sam Hellmuth

Generalizations The following table summarizes the accentual types that we have mentioned in this chapter: (11) An overview of accentual types in Arabic Classical Arabic [L/F] Middle

Eastern

Central

Western group

Southern group

Northern Group

Central Group

Gulf: [U/P]

Beirut/Lebanese: [U-P/A]

Radio Egyptian: [L(CNT)] P [U-P/A]

Cyrenaican Bedouin (East. Libyan): [L(CNT)]

San‘aani: [L/F] and [U-P/A]

Damascene: [U-P/A]

Cairene: Trochaic [L(CNT)]

Libyan (Tripoli): [L(CNT)]

Yemen Plateau: [L/F]

South Levantine/Palestinian: Trochaic [L(CNT)] or [L/F] and [U-P/A]

Egyptian Sa’iidi: [L/F]

Moroccan: [U/P]

Bedouin Hijazi: [U-P/A]

Negev Bedouin: Iambic [L(CNT)]

Sudanese: [L/F]

Nigerian: [U-P/A]

Saudi Arabic dialects: [U-P/A]

Bani Hassan: [U-P/A/P]

Chadic: [U-P/A]

Urban Hijazi Arabic: Trochaic [L(CNT)]

Baghdavi: [U-P/A]

Tunisian: [F/F]

North Sinai Bedouin: [L/F]

Algerian [F/P]

North Arabian dialects: [L/F]

Maltese: [U/P]

Because the Arabic languages show such an interesting variety of accentual systems, their proper analysis has been the subject of both descriptive and theoretical work. Here, following Hayes (1995), Kager (2009) and Watson (to appear), who o¤er detailed analyses, we will briefly discuss the variety of systems and their possible relations, using some ingredients of the accentual theory proposed in van der Hulst (in prep.).

Word accent systems in the Middle East

635

The table in (12) summarizes the various accentual types: (12) A summary of accentual types in Arabic [L/F]

Classical; Yemen Plateau; North Sinai Bedouin; North Arabian Dialects; Egyptian Sa’iidi; Sudanese

[F/F]

Tunesian

[F/P]

Algerian

[U-P/A]

Bedouin Hijazi; Saudi Arabic dialects; Beirut/Lebanese; Damascene; Baghdavi; Nigerian; Chadic

[U-P/A/P]

Bani Hassan

[L/F] and [U-P/A]

San‘aani

[U/P]

Moroccan; Gulf; Maltese

[L(CNT) tr]

þ EM(syll): Palestinian; þ /EM (syll): Radio Egyptian EM(syll): Cairene (~[U/P]) EM (syll): Urban Hijazi (no post-heavy strong light)

[L(CNT) ia]

Negev Bedouin; Cyrenaican Bedouin

Extrametricality It seems that all Arabic languages treat final CVXC as heavy and final CVC as light. The status of final CVV, which is rare or absent, is not clear, but apparently not uniform across the dialects. Final C is always ‘invisible’, but on top of that we need syllable EM. In final position, if CVV is either absent, or present and invisible (like CVC and CV), we can simply say that syllable-EM applies. If final CVV is stressed, we have to say that there is (a) final C-extrasyllabicity (to cover final CVC) and, additionally, final LIGHT syllable EM (to cover final CV). Final CVC, then, is (harmlessly) ‘doubly invisible’. Extrasyllabicity of C makes it light and as such extrametrical together with final CV. However, Cairene does not have final syllable EM, while it does ignore final C, because final CVC acts as light. In U/P languages we also do not have final syllable EM, but still final C is invisible, except in Moroccan where final CVC is heavy. In this language C«C is light together with final CV. To see a separate role for consonant-EM we can look at bisyllabic words. In most languages, in bisyllabic words, one would suppose that

636

Harry van der Hulst and Sam Hellmuth

syllable-EM is suppressed to guarantee word-minimality. Most of these CVCVC (LH) words have initial accent, suggesting that the final C does not contribute to weight, but in Negev Bedouin such words have final accent, which suggest that there is no final C-EM. This may be connected to the rise of so-called iambic patterns. Watson discusses the case of San‘aani Arabic in which the rightmost non-final CVV or CVG (syllables closed by a geminate consonant) attract accent away from final superheavy syllables and CVV syllables (and of course CVC syllables). This language then treats CVXC and CVV as ‘monosyllabic’ and as such they fall under syllable-EM. Bounded systems All Arabic languages agree in certain patterns: (13) a. Accent is U if the U syllable is VXC b. Accent is P if the P syllable is heavy and the U syllable is light (A final CVC counts as light) Systems start to di¤er once we look at final sequences of light syllables. Here we see a rich variety. The standard Latin like pattern adds the following third clause to the two clauses in (13) (14) c.

Accent is A if the two final syllables are light

This creates the [U-A/P] pattern of Damascene Arabic and many other varieties. However Cairene Arabic has P accent if a word ends in two light syllables preceded by a heavy syllable or two (or rather an even number of) other light syllables (counting from the word beginning or the rightmost heavy syllable): (15) HLL ]

HLLL ]

XLLLL ]

(X ¼ # or H)

This is a count system. In van der Hulst (in prep.) a count system is treated as having two bounded accentual domains, one on the left (which in this case is right-headed if the right hand syllable is heavy, otherwise left-headed) and one on the right that is not headed: (16) x >> (ss)sssss(ss)*s An alternating trochaic pattern of rhythmic beats echoes rightward, away from the initial accent (shown in 17 as Perfect Gridding-trochaic (Left to Right)) and ‘invades’ the right-hand domain as follows:

Word accent systems in the Middle East

(17) Cairene Arabic x x xx x x (HL)L ] H(LL)L ]

x xx x XL(LL)L ]

637

ER(R) PG-tr (LR)

Note that a rhythmic beat is assigned to a post-heavy light syllable (since rhythm in this language is moraic, as shown in Hayes 1996). In the first case the domain has two beats (one by weight and one by rhythm) and the rightmost wins, which is shown in (17) as End Rule (Right)). Palestinian Arabic is minimally di¤erent from Cairene: (18) Palestinian Arabic x x xx x x (HL)L ] H(LL)L ]

x xx x XL(LL)L ]

ER(R) PG-tr (LR)

The accent is further to the left, which Hayes (1996) in his metrical account handles by imposing foot extrametricality). We suggest an alternative. To account for the HLL] pattern we need to assume that the rhythmic beat on the penultimate light syllable is deleted; in other words: within the accentual domain a heavy syllable prevails over a rhythmically strong light syllable.8 However, we also need to account for the pre-antepenultimate pattern in case a word ends in a string of four light syllables which is claimed to exist because speakers pronounce quadrisyllabic light-syllabled words with initial accent. To us, this looks like a ‘Classical’ application of the Classical Arabic unbounded stress rule (specifically its default clause which applies if all syllables are light). We therefore submit that Palestinian Arabic embodies a hybrid aspect: in case of a long final string of light syllables it applies the unbounded initial default that is characteristic of Classical Arabic; this is indicated in table 11 by specifying that the system is both [L/F] and [U-P/A]).9 Urban Hijazi Arabic is like Palestinian Arabic, although it misses its hybrid character; it is simply [U-P/A]: (19) Urban Hijazi x x xx x x (HL)L ] H(LL)L ]

x xx x XL(LL)L ]

ER(L) PG-tr (LR)

8. We cannot say that the ER is Left, because in case of (HH)L] stress is penultimate, in all Arabic dialects. 9. Radio Egypt Arabic vacillates between the Cairene and the Palestinian pattern.

638

Harry van der Hulst and Sam Hellmuth

There is further variation among the count systems. Negev Bedouin has iambic rhythm rather trochaic rhythm: (20) Negev Bedouin x x xx x x (HL)L ] H(LL)L ]

x x x x XL(LL)L ]

ER(L) PG-ia (LR)

As in Cairene and Palestinian, the location of accent is not constant within a window that has two light syllables. This means that Negev is also a count system. However, the rhythmic pattern is iambic rather than trochaic. Finally we look at two non-count patterns: (21) Bani-Hassan10 x x xx x x (HL)L ] H(LL)L ]

x x x XL(LL)L ]

ER(R) Def (R)

This system di¤ers from Damascene Arabic in that a domain with two light syllables is right headed (indicated by Default (Right)), whereas the default is Left in Damascene: (22) Damascene x x x x x (HL)L ] H(LL)L ]

x x x XL(LL)L ]

ER(R) Def (L)

The di¤erences between these various systems are small, but real, although the diagnostic sequences may not be so easy to obtain, because of the fact that (final) sequences of light syllables are rare. It is interesting to note that count systems come very close to being weight-sensitive right-edge systems which we realize when we compare the languages discussed above: (23)

SH ] Damascene U Bani-Hassan Arabic U Negev U Cairene Arabic U Palestinian Arabic U Urban Hijazi U

HL ] HLL ] HLLL ] XLLLL ] P A A A P A P P P A P A P P A P P A A I / PA P A A P

10. This pattern possibly also occurs in Riyadh Arabic as analyzed in Halle and Kenstowicz (1989).

Word accent systems in the Middle East

639

As for the first three sequences, we only see a di¤erence in the HLL case which seems to reflect a di¤erence in extrametricality (‘no’ in Cairene). But if we add the fourth sequence (HLLL]), syllable-EM (no) can no longer be correct for Cairene, and this is where we have to resort to the count analysis. For Palestinian we do not have to do that yet. We can treat Bani-Hassan and Negev alike in that both treat a domain with two light sylables as right-headed, whereas the others have a left-headed choice. Only by adding the fourth sequence of syllables (four final light syllables) can we di¤erentiate all the systems, fully enforcing a count analysis of Negev, Urban Hijazi and Palestinian. The point of this exercise is to demonstrate that the crucial evidence for deciding the precise nature of the system (especially whether the system is a count system or not) lies in sequences of light syllables that are not frequent in the data.

Unbounded systems There is a third type of system that occurs with some frequency in the Arabic languages and is exemplified by Classical Arabic which has an unbounded LAST/FIRST system with some form of extrametricality. In unbounded systems the domain of accent assignment comprises the whole word. If the word contains heavy syllables, one of these (in Classical Arabic the rightmost non-final one) attracts accent; if there is no (non-final) heavy syllable in the word, the default option is a light syllable at one of the edges (in Classical Arabic it is the first). This LAST/FIRST pattern is reconstructed for Classical Arabic and as such it is controversial. An alternative interpretation of the comparative evidence is that accent never falls further leftward than the antepenultimate syllable (cf. Angoujard 1990 and Kager 2009 for discussion). However, as we have seen, Classical Arabic is not unique in having an unbounded system in the Arabic family. McCarthy (1979: 461) remarks that the Classical pattern is only preserved in a few modern dialects (Egyptian Sa‘iidi, Yemen Plateau). Our survey here above suggests that it is also found in Bedouin-origin North Arabian dialects in Syria, Lebanon and Sinai, and possibly also in Sudanese Arabic. Among the unbounded systems we also see cases that di¤er from Classical Arabic in the choice of the leftmost heavy syllable to bear accent. Algerian Arabic has a FIRST/FIRST system (with the default clause as in Classical Arabic, i.e. first), while Tunisian Arabic chooses the penultimate syllable as a default and hence has FIRST/PENULT [F/P]

640

Harry van der Hulst and Sam Hellmuth

Historical change If the unbounded L/F pattern can indeed be attributed to Classical Arabic (or perhaps proto-Arabic), two kinds of systems have developed from this unbounded system. We find ‘count systems’ in which the primary accent is on the right side of the word, but computation starting on (i.e. ‘counting’ from) the left side, and then we find right-edge systems of the ‘Latin type’ with penultimate or antepenultimate accent. McCarthy (1979) sees the count system type (of, for example, Cairene Arabic) as a later development than the Latin type (found in, for example, Damascene Arabic). In van der Hulst (1997), it has been suggested that the Cairene (count) system might be a transitional phase, forming the link between the initial (default) accent of the unbounded Classical Arabic and the bounded Latin-like accent rule of Damascene Arabic. Interpreted within the accentual theory in van der Hulst (in prep.), the transitions that have taken place can be seen as follows: (24) a.

Unbounded domain (þs-extrametricality) [(.............)] L/F: ER(R), Def (L): Classical Arabic F/F: ER(L), Def (L): Tunesian F/P: ER(L), Def (R): Algerian

b. Count system (þs-extrametricality, except in Cairene which means that final C-EM ‘emerges’) [(ss)ssss(ss)*] L(CNT)-trochaic: Cairene, Palestinian L(CNT)-iambic: Negev Bedouin c.

Right-edge bounded system (þs-extrametricality) [.............(ss)] U-A/P: ER(R), Def(L): Damascene Place accent on last heavy or first light in the rightmost bisyllabic domain

d. Right-edge bounded system (s-extrametricality, but þ C-EM except in Moroccan.) [.............(ss)] U/P: ER(R), Def(L): Gulf Place accent on last heavy or first light in the rightmost bisyllabic domain Classical Arabic is unbounded. The count systems display a ‘fracturing’ of the unbounded domain into two polar bounded domains, with the rightmost domain being the strongest. It would seem that count systems really

Word accent systems in the Middle East

641

are intermediate systems in that on the one hand ‘the whole word domain’ (at least both edges) is relevant, while, on the other hand, accent is confined to a two-syllable window on the right-edge. This provides for the re-interpretation in terms of a bounded right-edge system, one variety of which (namely the one that displays antepenultimate accent) maintains syllable-extrametricality, while the other does not (Gulf, Moroccan and Maltese Arabic). Final accent in [LH] words in, for example, Negev Bedouin Arabic, points to a loss of consonant extrametricality in specific cases, which may be connected to the rise of so-called iambic patterns. Throughout, all systems maintain weight-sensitivity.

4. Conclusions and Generalization Very little, if anything can be said about the isolate extinct languages mentioned in section 3. The pattern for Egyptian suggests a weightsensitive right-edge system which fits with the general pattern found in the Semitic languages where accent is either unbounded or confined to the right edge: SEMITIC EAST: Last/First (UAkkadian) WEST: CENTRAL: ARAMAIC: Latin-type CANAANITE: Final (Hebrew), Latin type (Tiberian Hebrew) ARABIC: Last/First / L (count) / Latin type SOUTH Western (not discussed here) Eastern: Last/First Knudsen (1980) in his analysis of Akkadian accent concludes (p. 15): ‘‘viewed in a historical perspective, most features in the Akkadian system of stress as outlined above are common to Akkadian, Hebrew, Aramaic, and Medieval Arabic.’’ It is tempting to see the Last/First system as fundamental with, as suggested earlier, count systems and Latin-type systems having derived from that. The loss of weight-sensitivity and extrametricality as exemplified by Modern Hebrew seems a final step in the direction of fixed right-edge accent.

642

Harry van der Hulst and Sam Hellmuth

References Abumdas, A.H.A. 1985 Libyan Arabic Phonology. University of Michigan. Abu-Salim, I.M. 1987 Vowel Harmony in Palestinian Arabic: A Metrical Perspective. Journal of Linguistics, 23, 1: 1–24. Angoujard, Jean-Pierre 1990 Metrical structure of Arabic. Dordrecht: Foris. Bergstra¨ßer, G. 1983 Introduction to Semitic Languages. Winona Lake, Ind.: Eisenbrauns. (Translation by P.T. Daniels of the German original, einfu¨hrung in die semitischen Sprachen, Munich: Max Hueber, 1928; 2nd. ed. 1963. Berman, Ruth 1997 Modern Hebrew. In: Robert Hetzron 1997 (ed.), 312–333. Birkeland, Harris 1954 Stress patterns in Arabic. Oslo: Jacob Dybwad. Bolozky, Shmuel 1997 Isreali Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 287– 312. Borg, Albert and Marie Azzopardi-Alexander 1997 Maltese. London and New York: Routledge. Borg, Alexander 1997 Cypriot Arabic Phonology. In: Alan S. Kaye 1997 (ed.), 219– 244. Borg, Alexander 1997 Maltese Phonology. In: Alan S. Kaye 1997 (ed.), 245–286. Boudlal, A. 2001 Constraint interaction in the phonology and morphology of Casablanca Moroccan Arabic. Universite´ Mohammed V, Rabat, Morocco. Buccellati, Giorgio 1997 Akkadian and Amorite Phonology. In Alan S. Kaye 1997 (ed.), 3–38. Co‰n, Edna A. & Shmuel Bolozky 2005 A Reference Grammar of Modern Hebrew. Cambridge: Cambridge University Press. Comrie, Bernard, Maria Polinsky, and Stephen Matthews 2003 The Atlas of Languages. Second Edition. London: Quarto Publishing Inc. Creason, Stuart 2004 Aramaic. In: Roger D. Woodard 2004 (ed.), 391–426. Daniels, Peter T. 1997 Classical Syriac Phonology. In: Alan S. Kaye 1997 (ed.), 127– 140.

Word accent systems in the Middle East de Jong, R. 2000

Diem, W. 1973

643

A Grammar of the Bedouin Dialects of the Northern Sinai Littoral Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden, Brill.

Skizzen Jemenitischer Dialekte. Wiesbaden, Kommission bei Franz Steiner Verlag. Edzgard, Dietz Otto 2003 Sumerian Grammar. Leiden: Brill. Erwin, W. 1963 A short reference grammar of Iraqi Arabic. Washington, Georgetown University Press. Faber, Alice 1997 Genetic subgrouping of the Semitic languages. In: Robert Hetzron 1997 (ed.), 3–15. Fischer, Wolfdietrich 1997 Classical Arabic. In: Robert Hetzron 1997 (ed.), 187–219. Gordon, Cyrus H. 1997a Ugaritic Phonology. In: Alan S. Kaye 1997 (ed.), 49–54. Gordon, Cyrus H. 1997b Amorite and Eblaite. In: Robert Hetzron 1997 (ed.), 100–115. Gordon, Cyrus H. 1997c Eblaite Phonology. In: Alan S. Kaye 1997 (ed.), 39–48. Gordon, Giorgio 1997 Akkadian. In: Robert Hetzron 1997 (ed.), 69–99. Graag, Gene 1997 Old South Arabian Phonology Phonology. In: Alan S. Kaye 1997 (ed.), 161–168. Guella, N. ms. On syllabication, stress and intonation in an Urban Arabic dialect. King Saud University, Riyadh. Hackett, Jo Ann 2004 Phoenician and Punic. In: Roger D. Woodard 2004 (ed.), 365– 385. Halle, M. and M. Kenstowicz 1989 On Cyclic and Non-Cyclic Stress. Ms. Dept. of Linguistics, M.I.T., Cambridge, MA. Hamid, A.H.M. 1984 A descriptive analysis of Sudanese colloquial Arabic phonology. University of Illinois. Hayes, B. 1995 Metrical stress theory: principles and case studies. Chicago, University of Chicago Press. Hayes, John 1997 Sumerian phonology. In: Alan S. Kaye 1997 (ed.), 1001–1020.

644

Harry van der Hulst and Sam Hellmuth

Hetzron, Robert (ed.) 1997 The Semitic languages. London: Routledge. Hoberman, Robert D. 1997 Modern Aramaic phonology. In: Alan S. Kaye 1997 (ed.), 313– 336. Huennergard, John and Christopher Woods 2004 Akkadian and Eblaite. In: Roger D. Woodard 2004 (ed.), 218– 287. Hulst, H.G. van der 1977 Primary accent is non-metrical. Rivista di Linguistica 9/1: 99– 127. Hulst, H.G. van der 2000 Issues in foot typology. In: M. Davenport & S.J. Hannahs (eds Issues in phonological structure, 95–127. Amsterdam: John Benjamins Publishing Company. Irshied, O. & M. Kenstowicz 1984 Some phonological rules of Bani-Hassan Arabic: a Bedouin dialect. Studies in the Linguistic Sciences 14: 109–148. Janssens, Gerard 1972 Stress in Arabic and word structure in the modern Arabic dialects. Leuven: Peeters. Jastrow, Otto 1997 The Neo-Aramaic languages. In: Robert Hetzron 1997 (ed.), 334–377. Kager, Rene´ 2009 Stress. In: Kees Versteegh et al (eds.), The Encyclopedia of Arabic Language and Linguistics, vol. 4, 344–353. Leiden: Brill. Kaufman, Stephen A. 1997 Aramaic. In: Robert Hetzron 1997 (ed.), 114–130. Kaye, Alan S. 1997 Arabic Phonology. In: Alan S. Kaye 1997 (ed.), 187–204. Kaye, Alan S. (ed.) 1997 Phonologies of Asia and Africa. Volume 1 and 2. Winona Lake: Eisenbrauns. Kaye, Alan S. and Judith Rosenhouse 1997 Arabic dialects and Maltese. In: Robert Hetzron 1997 (ed.), 263–311. Kenstowicz, Michael 1983 Parametric variation and accent in the Arabic dialects. In Chukerman, A., M. Marks, & J.F. Richardson (eds.) Papers from the nineteenth regional meeting of the Chicago Linguistic Society. CLS 19, 205–213. Chicago, Chicago Linguistic Society. Khalafallah, A.A. 1969 A Descriptive Grammar of Sa9i:di Egyptian Colloquial Arabic. The Hague, Mouton.

Word accent systems in the Middle East

645

Khan, Geo¤rey 1997 Tiberian Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 85– 102. Knudsen, Ebbe Egede 1980 Stress in Akkadian. Journal of Cuneiform Studies 32/1: 3–16. Lonnet, Antoine and Marie-Claude Simeone-Senelle 1997 La phonologie des langues subarabiques modernes. In: Alan S. Kaye 1997 (ed.), 337–372. Loprieno, Antonio 1995 Ancient Egyptian: A linguistic introduction. Cambridge: Cambridge University Press. Loprieno, Antonio 1997 Egyptian and Coptic phonology. In: Alan S. Kaye 1997 (ed.), 431–460. Lorpieno, Antonio 2004 Ancient Egyptian and Coptic. In: Roger D. Woodard 2004 (ed.), 160–217. Malone, Joseph L. 1997 Modern and Classical Mandaic Phonology. In: Alan S. Kaye 1997 (ed.), 141–160. McCarthy, John 1979 On stress and syllabification. Linguistic Inquiry 10; 443–466. McCarthy, R.J. & F. Ra¤ouli 1964 Spoken Arabic of Baghdad. Beirut, Librarie Orientale. Michalowski, Piotr 2004 Sumerian. In: Roger D. Woodard 2004 (ed.), 19–59. Mitchell, T.F. 1960 Prominence and syllabification in Arabic. Bulletin of the School of Oriental and African Studies 23: xxx–269. Nebes, Norbert and Peter Stein 2004 Ancient South Arabian. In: Roger D. Woodard 2004 (ed.), 454– 487. Nishio, T. 1994 The Arabic dialect of Qift (Upper Egypt Tokyo, Institute for the Study of the Languages and Cultures of Asia and Africa. Owens, J. 1984 A short reference grammar of Eastern Libyan Arabic. Wiesbaden, Harrassowitz. Owens, Jonathan 2006 A linguistic history of Arabic. Oxford: Oxford University Press. Pardee, Dennis 1997 Ugaritic. In: Robert Hetzron 1997 (ed.), 131–144. Pardee, Dennis 2004 Ugaritic. In: Roger D. Woodard 2004 (ed.), 288–318.

646

Harry van der Hulst and Sam Hellmuth

Pickett, I. 2006

Some aspects of dialect variation among nomads in Syria and Lebanon. SOAS.

Prochazka, T. 1988 Saudi Arabian dialects. London, Keegan Paul International. Rendsburg, Gary A. 1997 Ancient Hebrew Phonology. In: Alan S. Kaye 1997 (ed.), 65–84. Ruhlen, Merritt 1991 A guide to the world’s languages. Volume 1: Classification. Stanford, CA: Stanford University Press. Segert, Stanislav 1997 Old Aramaic Phonology. In: Alan S. Kaye 1997 (ed.), 115–126. Segert, Stanislav 1997a Phoenician and Punic Phonology. In: Alan S. Kaye 1997 (ed.), 55–64. Segert, Stanislav 1997b Phoenician and the Eastern Cannanite languages. In: Robert Hetzron 1997 (ed.), 174–186. Simeone-Senelle, Marie-Claude 1997 The Modern South Arabian languages. In: Robert Hetzron 1997 (ed.), 378–423. Steiner, Richard C. 1997 Ancient Hebrew. In: Robert Hetzron 1997 (ed.), 145–173. Stolper. Matthews W. 2004 Elamite. In: Roger D. Woodard 2004 (ed.), 60–94. Versteegh, Kees et al (eds.) 2006–9 The Encyclopedia of Arabic Language and Linguistics. Leiden: Brill. Watson, J.C.E. 2002 The phonology and morphology of Arabic. Oxford, OUP. Watson to appear Arabic word stress. In: Colin Ewen, Marc van Oostendorp and Keren Rice (eds.) Phonological Compendium. Blackwell. Wilhem, Gernot 2004a Hurrian. Roger D. Woodard 2004 (ed.), 95–118. Wilhem, Gernot 2004a Urartian. Roger D. Woodard 2004 (ed.), 119–137. Woodard, Roger D. (ed.) 2004 The Cambridge Encyclopedia of the World’s Ancient Languages. Cambridge: Cambridge University Press.

12.

A typology of stress patterns1

Rob Goedemans 1. Introduction One of the prime advantages of having a database is that it allows for the verification of typological claims in a quantitative manner. As such, StressTyp can be put to good use in checking the validity of assumptions that have been made in the past about the abundance, rarity, absence or implicational properties of certain metrical patterns. In this chapter, I will make an attempt to use StressTyp in this fashion, and cover some topics that will be of interest to anyone working on the general characteristics of stress patterns. Perhaps surprisingly, stress is not one of the linguistic properties that feature prominently in traditional typological studies. Many of these focus on syntax and morphology, and where phonology is mentioned, it is usually limited to remarks on the segmental inventories of languages.2 Hence, many of the figures on primary, or non-implicational, factors that will be presented below have not appeared in the literature on earlier occasions. In the light of the above, it is to be expected that almost no implicational universals of the type: ‘If a language has A, then does it also have B.’ concerning stress are addressed in purely typological studies either. In fact, such rules rather show up in theoretically oriented work on the modeling of metrical phonology, such as Prince (1983), Halle and Vergnaud 1. I thank Harry van der Hulst, Ellen van Zanten, and Dik Bakker for valuable comments on this chapter. The figures mentioned here are based on a snapshot of StressTyp made several years ago and, hence, may deviate slightly from what one may find if one was to query StressTyp now. Three other typological studies on stress have come to our attention since this chapter was written. Matt Gordon’s database on QI systems and Todd Bailey’s stress system database have been discussed in Goedemans & van der Hulst (2009). Jef Heinz’ stress pattern database combines the two and expands on them. This database can be consulted at http://phonology.cogsci.udel.edu/dbs/ stress/. A survey similar to the one in this chapter, but limited to Austronesian an Papuan languages, can be found in van Zanten & Goedemans (2007). 2. The seminal work in this (segmental) area is Maddieson (1984).

648

Rob Goedemans

(1987), Kiparsky (1991), Hayes (1995), Kager (1995) and van der Hulst (1996a,b). In these cases, (suspected) typological observations are often used as arguments in favour of a certain modification or extension of the abstract model that is used to generate all possible metrical systems. However, su‰cient means to verify these typological observations have not always been available to everyone. All too often, I fear, rather farreaching conclusions have been drawn from a survey of too small a dataset. Because of the enormous diversity we find in the metrical patterns of natural languages, the set of primary stress data one uses to test typological hypotheses must be quite sizeable. StressTyp, I submit, is such a source. As far as I know, a larger body of stress data does not exist. Currently, it covers over 500 languages, for which all the available information on basic stress patterning is stored. Additional information, for which fields are reserved, is often lacking, but this information is never needed for the type of questions that will be discussed in this chapter. Obviously, therefore, StressTyp can be instrumental in checking the claims concerning the typology of stress that have been made in the literature. In the following sections, I will first present some data concerning primary properties of stress systems. Section 2 will contain the results of relatively simple searches of the database that involve only one field, such as: ‘How many languages have fixed initial stress.’ In section 3, some implicational universals will be discussed which usually involve combinatory searches of two or more database fields. Though the primary data can be quite significant for many theoretical issues, it is very often the case that these theoretical issues are too intricate, and must be addressed with ‘multiple-field searches’. Let me emphasise that the results presented here far from exhaustively cover the information one may extract from StressTyp to validate or enrich metrical theories. The queries in the next two sections merely serve as examples of how we can use this database. We encourage its use in the testing of other hypotheses. Wherever appropriate, some discussion will precede and/or follow the data. However, since the focus of this chapter should really be on the figures themselves, I will keep these discussions as brief as possible. 2. Primary Data As was noted in the introduction, StressTyp currently contains information on the stress patterns of 510 languages. It is important to note that these 510 do not form a representative sample of the world’s languages.

A typology of stress patterns

649

Due to its specialist nature, StressTyp only contains languages that do indeed have word stress, and these are not evenly distributed geographically or genealogically. Both African and South-East Asian languages, for instance, are seriously underrepresented for obvious reasons. The major drawback of the decision to exclude languages that have no stress is that we cannot use StressTyp to calculate the percentage of the world’s languages that employ stress. Fortunately, however, data from StressTyp appeared in the World Atlas of Language Structures (Comrie, Haspelmath, Dryer and Gil 2005). Preparations regarding this project involved, among other things, expanding StressTyp such that it includes a 200 language sample that was compiled for WALS. As is to be expected, many languages in this sample do not have stress, and since it is quite representative, we can get a fair idea of the percentage of stress languages in the world by looking at this sample. To begin with, I found prosodic information for 176 of the 200 sample languages. Of these, 141 (80%) use stress, 28 (16%) have only (lexical) tone or pitch-accent,3 and for 7 languages (4%) it is explicitly stated that they have no (fixed) stress while no information on tones was found either. There is yet another way in which the languages in StressTyp might not form a representative sample. In Goedemans, van der Hulst and Visch (1996) we have clearly stated that some language families, most notably the Australian languages, were overrepresented. Later work, especially the inclusion of the WALS sample, has diminished the uneven distribution somewhat. However, since work on the database has been done under the umbrella of a project on the prosody of Indonesian languages for a period of four years, after the 1996 publication, a natural result is that the Austronesian language family must now also be counted among those that are overrepresented. The advantage of such extensive coverage of certain families is that we can present fairly exhaustive surveys (see chapters 2 3. Pitch-accent languages are languages in which accent placement is rule based. The only di¤erence with stress accent languages is the realisation of the accent. In the former, prominence is only achieved by pitch movements, while in the latter duration, intensity, and vowel quality also play a role. Strictly speaking, then, pitch accent languages can easily be included in StressTyp, and, as far as we know the rules for the location of the pitch-accent, such languages are indeed represented in StressTyp, though not abundantly (only 10, a situation which we must remedy in the future). In many cases, however, languages are classified by authors as pitch-accent languages, while they refrain from giving further details about the location of that accent. Those languages are not included in StressTyp and form part of the 16% mentioned above.

650

Rob Goedemans

and 3). It stands to reason that such an unevenly distributed sample should be used for our typological purposes with great caution. Therefore, some e¤ort has been put into the selection of a representative subset from the 500 StressTyp languages.4 This subset contains 238 languages. For the queries that will be presented below, both the subset and the entire dataset have been used as input. Whenever the results for the subset deviate substantially from those found for all the languages, the 238-language sample percentages will also be presented. In the next section, I will begin the presentation of the actual data, starting with the primary features. 2.1. Fixed or variable? One of the most straightforward questions we can ask concerning stress is: ‘where is it located’. This question is already quite di‰cult to answer directly. Before we can answer it we must distinguish between the quantity-sensitive (QS) systems, in which stress is not located on one specific syllable in every word, and the quantity-insensitive (QI) systems in which it is.5 Hence, the first question concerning stress should not be, ‘where is it located’, but ‘how many languages have only one possible stress location for every word’. Figure 1 below divides the stress languages into those that place word stress in one fixed position, and those in which the location of stress varies, depending on factors such as syllable weight, prominence or lexical marking. Needless to say, figure 1 also gives a good impression of the geographical distribution of the languages that use stress. It may be clear from this figure that the variable-stress languages do not form a minority group, even though their stress rules are more complicated. Let me give some details on the internal composition of the QS group. Firstly, 54% of the languages in this group are strictly QS; main stress 4. We are greatly indebted to Dik Bakker for his help in selecting the right languages for a representative sample. The number (500) di¤ers from that mentioned in the introduction since this chapter is based on an earlier snapshot, see note 1. 5. Note that we lump all the systems that have variable stress locations together under the QS label here. Languages in which stress depends on prominence factors or lexical marking are not strictly QS. Prominence-sensitive languages are usually regarded as QS in the literature though, and, in StressTyp, we have chosen to look upon lexical marking as just another factor to which the stress rule can be sensitive. Hence, languages with lexical stress are QS to the same degree as languages in which the stress rule is sensitive to, for instance, high tone.

A typology of stress patterns

651

Figure 1. Areal patterning of languages with QI stress (black dot: 278 languages ¼ 55.5%) and QS stress (white square: 222 languages ¼ 44.5%), see section 2.5 for adjustment of percentages.

locations solely depend on the length of the vowel or the presence of a coda in the syllables in the stress domain. 28% of the QS languages are sensitive to other syllabic factors, usually referred to as prominence (see Goedemans 1996 for a list of these), often in combination with normal quantity-sensitivity (10% of total). Finally, 18% have lexical stress, again sometimes in combination with other QS factors (3% of total).6 6. Note that languages can have lexical stress in a subset of their words (for which stress may even be contrastive). In this survey we place the languages in the group that describes the behaviour of the largest subset of the words. Hence, English and Dutch, for instance, are not counted among the lexical stress languages, while Russian is. Combinations of these various types of quantity sensitivity come in two flavours. Firstly, languages may make di¤erent sensitivity choices for primary and secondary stress, eg. Yaqui: primary stress initial or second, lexically determined, secondary stress on long vowels and closed syllables. Secondly, a single stress rule may assign stress on the basis of more criteria, eg. Iraqw: stress penultimate long vowels, else final vowels with a high tone, else initial vowels. Finally, for 2% of the languages we know they are QS, but have no information on the exact nature of heavy syllables.

652

Rob Goedemans

Figure 2. QI stress by position.

2.2. Stress location in QI languages In quantity-insensitive languages stress clearly has a demarcative function. It occurs on one syllable at the left or the right word edge, signaling the beginning or the end of the word, respectively. One would expect then, that the initial and the final syllable are highly favoured for this signal function. In figure 2 we see that this prediction is not entirely borne out by the facts (for areal patterning of these types, see WALS chapter 14). Stress at the right word edge more often falls on the penultimate syllable than on the final (though in the representative sample the di¤erence is much smaller: 33% vs. 27%). This fact can easily be explained with the help of a second observation that we can find in the literature, and which is confirmed by figures from StressTyp in the next section. Languages have a clear preference for trochaic (left-headed) feet. When such feet are used to derive stress locations, we naturally end up with stress on the first or the penultimate syllable, depending on the edge at which we place the bisyllabic foot. We also observe that many more languages mark the end, rather than the beginning of a word with a stress. 2.3. Rhythm type As was noted above, we find an unequal division of languages over the two foot types that we use to describe the rhythmic properties of stress

A typology of stress patterns

653

Figure 3. Rhythm type. No rhythm: source mentions specifically there is no secondary stress. Undefined: rhythm type of secondary stresses not clear or uses both foot types.

systems. In general, languages prefer the trochaic foot over the iambic foot. Note that we divide the languages into trochaic or iambic systems on the basis of their secondary stress properties. Main or primary stress is assigned in a bisyllabic domain in a separate procedure (at least, in StressTyp) that may deviate drastically from the (regular) assignment of stresses in the rest of the word (see section 3.3). Because of this capricious behaviour, we assume that main stress is not the perfect indicator for the trochaicity or iambicity of the language. Instead, we look at the rhythmic footing that derives the secondary stress locations. Unfortunately, that means many languages in StressTyp are left out of the equation, simply because we do not have any data on secondary stress for these languages. The statistics for the 326 languages for which we do know something about secondary stress can be found in figure 3. The hugely skewed distribution between iambic and trochaic systems in figure 3 is hard to miss even in a cursory look at the typology of stress systems. Though it has never been quantified in this way before, the scarcity of iambic systems has already led some researchers on the path to total denial of the iamb’s existence as an abstract metrical entity. Hayes (1987, 1995) dispenses with all quantity-insensitive iambs, while among the most

654

Rob Goedemans

drastic of these studies, van de Vijver (1998) claims that all iambic systems can be reanalysed as trochaic ones. In StressTyp we choose no sides in this, obviously highly theoretical, debate. We will continue to o¤er the iambic foot as an option, however, because there are enough languages, in our view, of which the stress patterns are clearly more straightforwardly described with the help of these iambic feet. As far as rhythm is concerned, one other division can be made. In the discussion above, I stepped lightly over the issue that stress assignment in ‘‘the rest of the word’’ does not necessarily mean that the entire word is indeed supplied with stresses. Rhythm assignment can be non-iterative, which means only one secondary-stress foot is placed on the rhythm plane (see Chapter 2 for further discussion of such systems). StressTyp reveals that the iterative option is clearly favoured. 70% of the 227 languages for which we have information on the placement of secondary stress choose this option, while only 18% reveal non-iterative rhythm (no iterativity information for 12%). For the sake of completeness I mention here that 6 (4%) of the iterative secondary stress languages show ternary rhythm (each rhythm beat is followed, or preceded, by two unstressed sylables).7 2.4. Stress location in QS languages For the quantity-sensitive languages it is not feasible to present all the possible stress locations in this chapter. There are simply too many options. For both edges there are two domains (with and without Extrametricality (henceforth EM)) in which stress can be assigned in various ways. Moreover, stress may move out of these domains under certain special conditions. What I will do below, therefore, is present the domain locations, limiting myself to the most common cases in which the domain is either located at the left or right edge, one syllable away from either edge (with syllable EM), or covers the entire word (unbounded systems). Hence, figure 4 does not provide us with exact stress locations, but it gives us the syllables on which stress may fall, depending on their weight. As for the QI systems, we divide the languages into groups according to their preference for either the right or the left edge. Languages in which stress may fall on any syllable in the word (unbounded or lexical systems) are represented as well (for areal patterning of these types, see WALS chapter 15). 7. In the representative sample, secondary stress information is lacking for no less than 37% of the languages. Hence the percentages for iterativity are quite di¤erent: 49% iterative, 11% non-iterative and 3% ternary.

A typology of stress patterns

655

A: stress located on one of the leftmost 2 syllables B: stress located on one of the rightmost two syllables C: stress located on left word edge, but not restricted to leftmost two syllables (2 languages only) D: stress located on right word edge, but not restricted to rightmost two syllables, or stress located on either penult or antepenult (never final). E: stress may be located on any syllable in the word (unbounded). F: combination of 3 or 4 and 5. G: stress location is not predictable/cannot be determined, it is either lexical, completely irregular, or there is no primary stress (all stresses are equally prominent). Figure 4. Variable Primary Stress locations of the 222 QS languages.

With respect to the unbounded systems (E), a further remark is in order. We could assign these to either the left or the right-oriented group on the basis of their choice for the edge at which they assign stress in case no heavy syllable is present in the word, assuming that this default edge, and not the choice for first or last heavy syllable in case there are heavy syllables in the word, represents their edge preference. Seen in this light, 29% of the unbounded systems are right-edge oriented, while 55% prefer the left edge (16% undetermined). The grand totals for all QS systems thus add up to 31% left and 54% right (15% undetermined). Perhaps it is also worthwhile to present the grand totals for all stress systems. Averaged over the QS and QI systems, 35% of the world’s lan-

656

Rob Goedemans

guages have a clear preference for stress at the left word edge, while in 58% of the languages stress is preferably located at the right edge of the word. The low number of languages that assign stress to either the second or the third syllable is predicted. In Goedemans (1996, note 18) it was already mentioned that left-edge extrametricality is extremely rare. Indeed we find only 13 such cases in StressTyp, all but one of which are suspect, either because they use EM-left in an unconventional way (the Aranda type languages, 6 cases, see Goedemans 1998), or because the stress rule is unclear to begin with (6 cases). The only undisputable case is Winnebago. Opposed to this we find 67 languages in which EM-right seems to operate in a straightforward manner. A further division according to the prosodic unit that is extrametrical shows that syllable EM appears in the majority (66%) of cases, while segment EM (23%), foot EM (6%) and mora EM (2%) are used much less frequently (3% unique cases with, for example, morphologically determined EM). A final remark concerning stress location in QS systems features the number of languages in which stress may shift out of the domain for main stress assignment. We observe that only 9 (4%) of the QS languages show this behaviour, while in the vast majority of these (8) deviant languages main stress is located at the right word edge. 2.5. Matters of weight In section 2.1, we have noted that quantity-sensitive stress systems assign stress to particular syllables in the main stress domain on the basis of either (1) syllable weight, (2) syllable prominence, or (3) lexical marking. The first category comes in three flavours. There is always a division between light and heavy syllables, but the division is not identical across languages: (a) Some choose to count only syllables with long vowels as heavy, (b) in others just syllables that are closed by a consonant are heavy, (c) while yet other languages count both these types among the heavy syllables. In all cases, open syllables with short vowels (CV) are light. When syllables are divided according to their prominence, the flavours are too many to mention here. To name just a few: high tone, sonority of either the vowel or the coda consonant, quality of the vowel, and rhythm beats (according to the view adopted in StressTyp, see Visch 1996) can all make syllables ‘heavy’. Finally, in the lexicon, syllables that are indistinguishable from others where prominence and weight are concerned can still be marked as ‘heavy’ for the metrical rules.

A typology of stress patterns

657

A: only long vowels make syllables heavy (and/or only diphthongs) B: only closing consonants (codas) make syllables heavy C: both long vowels and closing consonants make syllables heavy D: prominence factors (tone, sonority etc.) make syllables heavy8 E: syllables are ‘‘heavy’’ by lexical marking Figure 5. Weight criteria (NB: in the representative sample only 16% of the languages fall into category C while 20% are lexical stress (cat. E) languages).

If we were to look at all the possible weight categories, the figures and numbers would soon become dazzling and uninformative. So, in figure 5 I only distinguish between 5 categories: VV is heavy, VC is heavy, VV þ VC are heavy, a ‘prominence’ category in which I lump all the prominence systems together, and finally, the lexical stress languages. The totals will not add up to the percentages that were mentioned in section 2.1, because there we were only looking at primary stress, ignoring cases that were QI for primary stress but QS for secondary stress. In figure 5 all the systems that show some trace of quantity sensitivity are included, which will of course influence the results. For one, the total percentage of QS systems rises to 49%, leaving 51% of true QI systems (compare percentages below figure 1). 7 QS languages were not included in figure 5 8. In this category we also include the so called ‘‘count’’ systems that assign primary stress on the last syllable that receives a rhythm beat in secondary stress assignment.

658

Rob Goedemans

because their weight criteria were unclear or unknown. (For areal patterning of these types see WALS chapter 16). A further search within the set of prominence systems reveals that, apart from the 21% of languages that are sensitive to rhythm beats (the non-traditional prominence category we have added for the sake of count-systems), 6% are sonority related, 22% are tone related, 30% use vowel quality as a criterion, 16% use vowel height (which could be added to sonority) and 5% are unique unrelated cases. It has often been claimed that, if a language counts closed syllables as heavy for stress, then syllables with long vowels, should the language have these, will also be heavy for stress. A search in StressTyp reveals that only 3 of the 29 languages in which only closed syllables are heavy for stress do have long vowels (hence, do not behave in the predicted way). Not all these cases are clear-cut, but the majority shows that the prediction seems to hold true in general.9 Strictly speaking, the last remark belongs in the next section. It is a clear example of the type of ‘if then’ statement we can check with the help of a database. Let us proceed, then, to section 3, in which we will look at some more statements of this type and cast them in a StressTyp query to determine their validity.

3. Implicational cases 10 3.1. Quantity-sensitive stress and long vowels The first case continues on the same note as the final remark in the previous section. Wetzels (2001) cites Trubetzkoy’s claim that all languages that have QS stress must also have long vowels. He continues to argue

9. A language in which the long vowels clearly show behaviour that deviates from the prediction is Amele. Languages that seem to fall in this category, but for which further research is needed to determine it with certainty are Berguner Romansch and Bagvalal. Dutch was long thought to be a clear counterexample to this prediction. It has been shown convincingly, however, that the distinction between the Dutch ‘long’ and ‘short’ vowels is rather attributable to a di¤erence between tense and lax vowels (see, among others, van Oostendorp 1995). 10. The hypotheses in this section were not tested on the representative sample, because that sample often contained too few relevant cases, which means the percentages would be meaningless.

A typology of stress patterns

659

Table I. Number of languages in which codas (and in some cases also lexical marking or prominence) make syllables heavy for stress while long vowels do not, broken down into languages that do (probably) have long vowels and languages that (probably) do not. vø absent

vø probably absent

vø present

vø probably present

20

6

1

2

69%

21%

3%

7%

that this claim is falsified by the facts he finds in Portuguese, concluding that Trubetzkoy’s notion of quantity must have di¤ered greatly from the way in which we use it today. Table 1 shows the languages in StressTyp that support Wetzels’ arguments. Even if we discard the questionable cases in the second column, there are still enough languages left to question the universality of the claim that all QS languages need to have long vowels. 3.2. Sub-minimal words and degenerate feet Many languages have a restriction on the minimal size of (content) words, which are mostly required to contain at least two syllables. Languages that do not adhere to this restriction are said to contain sub-minimal words. In StressTyp such sub-minimal words are defined as monosyllabic in QI languages, while that single syllable must necessarily be light in QS systems for the word to be sub-minimal. In Kiparsky (1991) we find the claim that left-to-right trochaic systems which have sub-minimal words will stress the rightmost syllable of a word when that syllable is ‘left over’ after the word is parsed into feet and vice versa.11 To be able to stress this final syllable, these languages are said to use degenerate feet (monosyllabic or monomoraic feet) or ‘catalexis’ (Kiparsky 1991). Languages without sub-minimal words leave such syllables unparsed (i.e. do not allow degenerate feet or catalexis). 11. The discussion on degenerate feet and sub-minimal words in this section is solely based on the theory found in Kiparsky (1991) and Kager (1995). Di¤erent views on this issue can be found in Hayes (1995). In Goedemans, van der Hulst and Visch (1996), we also discuss the sub-minimal word and degenerate feet issue. For a discussion of Hayes’ view, and the relevance of the StressTyp data for this view, I refer to that work.

660

Rob Goedemans

Table II. Frequencies of presence or absence of degenerate feet in languages with and without sub-minimal words. Kager (44) sub-minimal words deg. feet y n

StressTyp I (46) sub-minimal words

StressTyp II (45) sub-minimal words

y

n

y

n

y

n

13

10

15

7

11

11

1

20

6

18

3

20

Kager (1995) checks this claim against 44 languages that fit the left-toright trochaic pattern and concludes that 75% of his languages behave according to Kiparsky’s predictions. Kager’s results can be found alongside the results of the StressTyp query in the first two sub-tables in table II. The StressTyp data clearly corroborate Kager’s findings (72% correct), though the di¤erence between the ‘sub-minimal words (y) – degenerate feet (n)’ cases in the Kager and StressTyp I sub-tables is striking. However, we must note that Kager’s definition of sub-minimal words deviates slightly from that used in StressTyp. In his view, bimoraic words are not sub-minimal in systems that have a long-short vowel contrast, even if the stress pattern is QI, while in StressTyp such words are considered sub-minimal if the stress pattern does not use weight. Hence, Kager’s subset of left-to-right trochaic cases contains comparatively few systems with subminimal words, which explains the di¤erence between the first two sub-tables in II. When we use Kager’s definition of sub-minimal words (and throw out the one language for which we lack the necessary information), we end up with the StressTyp II sub-table. Seen this way, 69% of the relevant languages in StressTyp behave according to Kiparsky’s prediction. Not the overwhelming support, we feel, that one would expect to find for an across-the-board generalization of this type. 3.3. Separating primary and secondary stress Some multiple queries that are very closely related to the ideas that underly StressTyp concern arguments for the separation of main stress and rhythm. Such a separation has been proposed and defended on theoretical grounds in van der Hulst (1984, 1996a, 1996b). Moreover, during the storage of stress systems in the database it appeared that the separate parameters for primary and secondary stress were practical in the description of certain ‘exotic’ stress patterns (see, for instance, the discussion on

A typology of stress patterns

661

Table III. Frequencies of languages with left or right edged main stress cross spliced with direction of footing (sample ¼ 154). Starting edge of footing Main stress domain

Left

Right

Left

63

12

Right

27

53

Table IV. Mismatches in quantity-sensitivity between primary and secondary stress (sample ¼ 97, QS languages for which we have no info on rhythm were not taken into account). Both QS

Primary QS Secondary QI

Primary QI Secondary QS

Both QS, but di¤erently

40

32

17

8

41%

33%

18%

8%

Gaanay on page 60–61). Van der Hulst could not have foreseen the existence of these patterns. Yet, the fact that they do exist lends further support to the Primary Accent First (PAF) claim. Slightly less exotic, but all the more relevant, are those languages that have stress patterns going against the tendency that the edge at which main stress is located is also the edge at which rhythmic footing starts. For such languages, any metrical theory would need to separately specify both the edge for main stress assignment and the edge at which the assignment of secondary stress feet starts. Van der Hulst uses the existence of these languages as an argument in favour of his PAF theory. In Table III we see exactly on how many languages this argument is based. In total, 39 languages (8% of all the languages in StressTyp) assign main and secondary stress from opposite directions. It seems the subset of languages for which we need separate direction parameters for main stress and rhythm is large enough to provide serious support for the PAF theory. A second argument for the separation of primary and secondary stress is formed by languages in which the rules for main stress and rhythm do not agree in quantity-sensitivity, or more intricately, in the way in which they use it. Table IV shows the number of languages in which

662

Rob Goedemans

main stress is QS and secondary stress is QI, or vice versa. The last column in Table IV shows the languages in which main and secondary stress disagree in their definition of what constitutes a heavy syllable. In such languages, closed syllables may, for instance, be heavy for stress but not for rhythm.12 In total, we need separate quantity statements for primary and secondary stress in 57 languages (11% of all the languages in StressTyp). All in all 17% of the languages in StressTyp support the separation hypothesis on the basis of opposite footing or QS properties for primary and secondary stress (Tables III and IV combined, 10 overlapping languages). 3.4. Rhythm type and the direction of footing An interesting correlation that has been noted many times in the past concerns the foot type used in secondary stress assignment and the edge at which this assignment starts. The claim is that languages with an iambic rhythmical pattern assign rhythm from left-to-right (cf. Kager 1993; Alber 2001). The reverse claim, that languages with trochaic rhythm start footing at the right edge, is not commonly deduced from the observation with respect to iambic languages. People seem hesitant to claim anything in this case. I present the data for both possibilities in Table V. Table V. Di¤erences in preference for direction of footing for both foot types. Starting edge of footing Foot Type

Left

Right

Trochaic

83

58

Iambic

22

8

We observe that the correlation between iambic rhythm and left-toright footing is indeed quite high. It turns out to be even higher when 12. We have subtracted from these numbers the count systems in which primary stress is sensitive to rhythmically strong syllables while secondary stress is sensitive to normal weight properties (or is even QI), since this di¤erence in what constitutes a heavy syllable is induced by our analysis of these count systems, and is not an intrinsic phenomenon of the language. We cannot, of course, count idiosyncratic properties of our framework as evidence for that same framework. In most other metrical theories, sensitivity of the main stress rule to rhythm beats is the normal default option, as the End Rule promotes the leftmost or rightmost secondary stress to main stress.

A typology of stress patterns

663

we remove the ‘suspicious’ cases from the list of right-to-left iambic languages. It appears that in many of these languages the choice between iambic and trochaic feet is not easily made, and where the descriptive source mentions the language is iambic, this choice is often based on superficial evidence. For only 3 of these cases can we be quite certain that rhythm is iambic. The trochaic opposite is far more problematic indeed. In fact, most trochaic languages start foot assignment at the left instead of the right edge. Yet, the di¤erence is not large enough to speak of a clear correlation in this case. We may conclude, therefore, that the choice to state the claim only for the iambic systems is justified in hindsight by the StressTyp data. 4. Conclusion I hope to have shown here that StressTyp can be a very useful tool for any linguist working on stress. It is perhaps wise, however, to conclude this chapter with some words of caution. It is imperative to remember that the design of a database imposes its limitations on the data that is stored. The specific definitions of the fields in the database may often require recoding of some language material found in a grammar before it can be incorporated. In other words, the language facts must be translated to database code. For the most straightforward typological properties of stress patterns, such translation is fairly unproblematic, since the fields with their possible values were carefully designed to accommodate these properties. When more unusual properties, or more exotic values of standard properties come into play, however, translation may prove quite di‰cult. In such cases, and admittedly, even in less di‰cult ones, the translation fully depends on the interpretation of the linguist. This step adds an extra uncertainty factor to the risk we are already taking in trusting the descriptive sources (Croft 1990: 25), in which primary language data have already been translated by the author to linguistically coded material.13 Moreover, there will be errors of the ‘normal’ sort in StressTyp. No matter how often one checks such large bodies of data, errors always 13. One could add to this yet another degree when using secondary sources as input for a database, as was common in the early days of StressTyp. The data in such secondary sources is often coloured by the hypothesis of the analyst (Croft 1990: 25). For all but about 10 of the languages in StressTyp, the primary sources have now been checked to guard against this source of errors.

664

Rob Goedemans

remain. Furthermore, it is by no means uncommon to find only extremely brief descriptions of the stress pattern in grammars. Hence, for many languages in StressTyp, information that goes beyond the basic stress pattern is missing. Finally, I must add that we should never overlook StressTyp’s weak side. As was noted above, and in all the previous chapters time and again, it is sometimes impossible to correctly represent a language family in the StressTyp sample, simply because there are no (good) grammars of the languages in question. Moreover, even if there are grammars, they might easily be mistaken in their stress description (see especially the chapters on African and Austronesian languages in this volume). These factors added together force us to view the results of StressTyp queries with somewhat more caution than, say, queries on a database filled with information on the behaviour of particles after atom collisions. Nevertheless, we can safely assume that, through the sheer size of StressTyp, the correct data will outweigh the errors, so that we may treat them as mere ‘noise’ in the larger queries. However, as numbers get smaller in more specialised queries, one must always be cautious when interpreting the results. In such situations, it is wise to check all the languages generated by the query (perhaps even reconsulting the primary source) to see whether they are all really of the type one intended to find with that particular query. Needless to say, it is of vital importance to get thoroughly acquainted with the ins and outs of the StressTyp fields before one can even begin to formulate these queries. With all this in the back of my mind, I can still safely say that StressTyp will be of much use to the linguistic community. The results presented in this chapter, but also any future results obtained from a judicious use of StressTyp, may serve as solid foundations for any linguist who wishes to base a theory, or modification thereof, on facts rather than speculation.

References Alber, Birgit 2001

Right-Alignment as Avoidance of Stress Lapse and Stress Clash. Paper presented at WCCFL 20, USC Los Angeles, 23–25 Feb. 2001. Comrie, B., M.S. Dryer, M. Haspelmath & D. Gil 2005 World Atlas of Language Structures. Oxford: Oxford University Press.

A typology of stress patterns Croft, William 1990

665

Typology and Universals. Cambridge: Cambridge University Press. Goedemans, Rob W.N. 1996 Syllable Weight and Prominence. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 115–163. Goedemans, Rob W.N. 1998 Weightless Segments: A Phonetic and Phonological Study Concerning the Metrical Irrelevance of Syllable Onsets. LOT Dissertations 9. The Hague: Holland Academic Graphics. Goedemans, Rob W.N. & Harry G. van der Hulst 2009 StressTyp: A database for word accentual patterns in the world’s languages. In: M. Everaert, S. Musgrave & A. Dimitriadis (Eds.), The Use of Databases in Cross-Linguistic Studies (Empirical approaches to language typology), 41, pp. 235–282. Berlin/New York: Mouton de Gruyter. Goedemans, Rob W.N., Harry G. van der Hulst and Ellis A.M. Visch 1996 Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics. Halle, Morris and Jean-Roger Vergnaud 1987 An Essay on Stress. Cambridge: MIT. Hayes, Bruce 1987 A Revised Parametric Metrical Theory. Northeastern Linguistic Society 17: 472–489. GLSA. Amherst, Ma. Hayes, Bruce 1995 Metrical Stress Theory. Chicago: University of Chicago Press. Hulst, Harry G. van der 1984 Syllable Structure and Stress in Dutch. Dordrecht: Foris. Hulst, Harry G. van der 1996a Separating Primary and Secondary Accent. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 1–25. Hulst, Harry G. van der 1996b Primary Accent is Non-metrical. Revista di Linguistica 9/1: 1– 119. Kager, Rene´ 1993 Alternatives to the Iambic-Trochaic law. Natural Language and Linguistic Theory 11: 381–432. Kager, Rene´ 1995 Consequences of catalexis. Harry G. van der Hulst and Jeroen van de Weijer (eds.) Leiden in Last: HIL Phonology papers I. HIL Publications 1. The Hague: Holland Academic Graphics, 269–298.

666

Rob Goedemans

Kiparsky, Paul 1991

Catalexis. Ms, Stanford University and Wissenschaftkolleg zu Berlin.

Maddieson, Ian 1984 Patterns of Sounds. Cambridge: Cambridge University Press. Oostendorp, Marc van 1995 Vowel Quality and Syllable Projections. PhD Dissertation, Tilburg University. Prince, Alan 1983 Relating to the Grid. Linguistic Inquiry 14: 19–100. Smith, Norval 2000 Lenition: the search for a relevant hierarchical framework. Paper presented at the 8th Manchester Phonology meeting. Vijver, Ruben van de 1998 The Iambic Issue: Iambs as a Result of Constraint Interaction. HIL dissertations 37. The Hague: Holland Academic Graphics. Visch, Ellis A.M. 1996 The Linguistic Structure of StressTyp. Rob W.N. Goedemans, Harry G. van der Hulst and Ellis A.M. Visch (eds.) Stress Patterns of the World: Part 1 Background. HIL Publications 2. The Hague: Holland Academic Graphics, 69–114. Wetzels, Leo 2001 On the Relation Between Quantity-Sensitive Stress and Distinctive Vowel Length: The History of a Principle and its Relevance for Romance. In Caroline Wiltshire and Joaquim Camps (eds.) Selected Papers from the XXXth Linguistic Symposium on Romance Languages. Amsterdam: John Benjamins. Zanten, Ellen A., van & Rob W.N. Goedemans 2007 A functional typology of Austronesian and Papuan stress systems. In: V.J. van Heuven & E.A. van Zanten (Eds.) Prosody in Indonesian Languages (LOT Occasional Series) 9, pp. 63–88. Utrecht: Landelijke Onderzoekschool Taalkunde.

Part II.

StressTyp Data

Australian languages

Alawa [P] Australian, Maran, Alawic. Mid-north of Australia. f

Primary stress on the penultimate syllable.

¨mama ‘food’ ¨nula ‘he’

al¨awal ‘properly’ bal¨palpi ‘slowly’

Sharpe, M.C. (1972). Alawa Phonology and Grammar. Australian Institute of Aboriginal Studies, Canberra. Alyawarr; Iliaura; Alyawarra [S] Australian, Pama-Nyungan, Arandic, Urtwa. East Central Northern Territory. f f

Primary stress on the first syllable if it has an onset, else on the second. Secondary stress on alternates after the primary stress.

iyl¨pa ‘ear’ wa¨liymparra ‘pelican’

¨kwiya ‘girl’ apm¨pirni˙tjika ‘will cook’

Yallop, C. (1977). Alyawarra: an aboriginal language of Central Australia. Australian Institute of Aboriginal Studies, Canberra. Arrarnta, Western; Aranda, Western [S] Australian, Pama-Nyungan, Arandic, Urtwa. Central Australia. f f f

Stress in bisyllables on the first. In other words: Primary stress on the first syllable if it has an onset, else on the second. Secondary stresses on alternates after the primary stress.

¨kama ‘to cut’ i¨batja ‘milk’

¨ilba ¨lelan˙tinama

‘ear’ ‘to walk along’

Strehlow, T. (1942). Arrandic Phonetics. Oceania XII, pp. 255–302. Baadi; Bardi; Badimaya [I] Australian, Nyulnyulan. Western Australia. f f

Primary stress on the first syllable. Secondary stress on odd non-final syllables.

670 ¨Nud

Australian languages

‘horse’

¨wanara ‘long, thin’

¨mada

‘stone’

¨wanal˙dJiliNa ‘scorpion’

Dunn, L. (1988). Badimaya, a Western Australian Language. Papers in Australian Linguistics 17, Pacific Linguistics, Series A-71, 19–149. Metcalfe, C.D. (1971). A tentative phonetic statement of the Bardi Language. Papers on the languages of the Australian Aborigines. AAS 38, 82–92. Australian Institute of Aboriginal Studies, Canberra. Bagundji (Dialect of Darling) [I] Australian, Pama-Nyungan, Baagandji. New South Wales. f f

Primary stress on the initial syllable. Secondary stress on alternate syllables after the first.

¨nidja ‘one’ ¨bira˙duda ‘hawk’

¨guni˙ga ‘fire’ ¨baødju˙ga ‘moon’

Hercus, L.A. (1982). The Bagandji Language. Pacific Linguistics, Series B, No. 67. Australian National University, Canberra. Banggarla; Parnkalla [A] Australian, Pama-Nyungan, South-West, Yura. Lake Eyre, South Australia. f

Primary stress is on the antepenultimate syllable (initial in disyllables).

¨pulba ‘dust’ pal¨larriti ‘to shine’

¨marniti ‘good’ wa¨yandata ‘to throw, pour, fill’

Schu¨rmann, C.W. (1844). A Vocabulary of the Parnkala Language. Adelaide: George Dehane, Australia. Burarra; Barera [I] Australian, Burarran. Maningrida, Arnhem Land, Northern Territory. f

f

Primary stress always occurs on the first syllable of the first root in the word. If there is prefixation, a secondary stress occurs on the prefix.

¨ama

‘mother’

¨kalamaN ‘axe’

a¨ma

‘will get him’

˙gu ¨r imaNa ‘he is holding it’

Australian languages

671

Glasgow, D. & K. Glasgow (1967). The phonemes of Burera. Papers in Australian Linguistics 1, pp. 1–14. Pacific Linguistics, Series A. Occasional papers, No. 10. Glasgow, K. (1981) Burarra. In: Bruce Waters (ed.) AAB SIL Working papers. Series A-5, 63–90. Dhurga; Thaua [I/I] Australian, Pama-Nyungan, Yuin-Kuric, Yuin. New South Wales, Bermagui to Jervis Bay. f f

Most words have initial primary stress. Most words with long vowels in the second syllable carry the main stress there.

¨amaga ‘good’

wa¨raøNa ‘boomerang’

Eades, D.K. (1976). The Dharawal and Dhurga Langages of the New South Wales South Coast. Australian Aboriginal Studies, Research and Regional Studies, No. 8. Australian Institute of Aboriginal Studies, Canberra. Dieri; Diyari [I] Australian, Pama-Nyungan, Karnic, Karna. South Australia, Leigh Creek. f f

Primary stress on the first syllable. Secondary stress on odd non-final syllables of roots and su‰xes.

¨mankara ‘girl’ ¨nanda-yi ‘hit-pres’

¨Nanda˙walka ‘to close’ ¨kana-ni-˙mata ‘man-loc-ident’

Austin, P. (1978). A Grammar of the Diyari Language of North-East South Australia. Ph.D. Dissertation, Australian National University, Canberra. Austin, P. (1981). A Grammar of Diyari, South Australia. Cambridge: Cambridge University Press. Djingili; Tjingili [P] Australian, West Barkly, Jingalic. Northern Territory, Ash Burton Range area. f f

Primary stress on the penultimate syllable. Secondary stress on the pre-antepenult.

672 ¨Naødi

Australian languages

‘I am’

˙Nuru¨ala ‘we all’

bi¨aNga

‘later’

abur˙mai¨gala

‘one at a time’

Chadwick, Neil (1975). A Descriptive study of the Djingili Language. Australian Institute of Aboriginal Studies, Canberra. Dyirbal [I] Australian, Pama-Nyungan, Dyirbalic. Inland and Coast of East Central Queensland. f f

Primary stress on the first syllable. Secondary stress on the odd syllables of all morphemes.

¨gumburu ‘mountain mist’ ¨łinay-˙mariłu ‘sit-com-instr-refl’ Dixon, R.M.W. (1972). The Dyirbal Language of North Queensland. Cambridge Studies in Linguistics 9, Cambridge: Cambridge University Press. Gaalpu (Dialect of Dhangu) [I] Australian, Pama-Nyungan, Yuulngu. Northeast tip of Arnhem Land. f f

Primary stress on the first syllable. Secondary stress on the penult or antepenult of words longer than three syllables.

¨rakal

‘greedy’

¨rawaktji ‘to shrivel up’

¨pat5

‘after’

¨tjinpulkku˙wanam ‘to sharpen’

Wood, P. (1978). Some Yuulngu Phonological Patterns. Pacific Linguistics, Series A, No. 51, 53–118. Australian National University, Canberra. Garawa [I] Australian, Garawan. Borroloola, Northern Territory and Doomadgee, Queensland. f f f

Primary stress on the first syllable. Secondary stress on even syllables counted from the right. No secondary stress adjacent to the main stress.

¨yami ¨watjim˙paNu

‘eye’ ‘armpit’

¨punjala ‘white’ ¨tjal ¨nariNin˙mukun˙jina˙mira

‘flower’ ‘at your own many’

Furby, C. (1974). Garawa Phonology. Pacific linguistics, Series A, No. 37. Australian National University, Canberra.

Australian languages

673

Gayardilt; Kayardild [I] Australian, Pama-Nyungan, Tangic. Queensland, Bentinck Island, Gulf of Carpentaria. f f

f

Primary stress is initial. Secondary stress on all long vowels and syllables that have a trill in the coda. Secondary stress on the penult (except in trisyllables with liquid or semivowel onsets).

¨kunja

‘little’

¨kuka˙lutha ‘rebound’

¨ka˙laaja

‘cut oneself ’

¨da˙barra ‘tree’

Keen, Sandra (1969). Ms, Australian Institute of Aboriginal Studies, Canberra. Evans, Nicholas D. (1995). A Grammar of Kayardild. Berlin: Mouton de Gruyter. Gidabal (dialect of Bandjalang) [I;S (NMS)] Australian, Pama-Nyungan, Bandjalangic. Northeast corner of New South Wales. f f

f

Stress is located on the first syllable, regardless of its weight. It also occurs on the odd numbered vowels of sequences of syllables with long vowels. Systematic exceptions with stress on the second instead of the first syllable exist.

¨gawaN

‘mother’s brother’

¨gabalNinj ‘scrubdweller’

¨manda¨r aøm

‘wild raspberry’

¨gawar i¨waøla

‘is definitely running’

Geytenbeek, B. & H. Geytenbeek (1971). Gidabal Grammar and Dictionary. Australian aboriginal studies 43, Canberra. Gooniyandi; Gunian; Konejandi [I;P (NMS)] Australian, Bunaban. Western Australia, Fitzroy River basin southeast of Derby. f

f

f

Mostly primary stress is initial (in trisyllables sometimes on a heavy final or penult). Secondary stress on the penult of four or five syllable words, or the final of a trisyllable. Primary and secondary stress may reverse, or the distinction may vanish.

674

Australian languages

¨ngaboo

bil¨gaali

‘father’

¨ngadda¨nyoo ‘mother’

‘midnight’

¨lawagi¨mana ‘white’

McGregor, W. (1990). A functional grammar of Gooniyandi. Amsterdam: John Benjamins. Guguyimidjir; Koko Imudji [I] Australian, Pama-Nyungan, Yalandyic. Queensland. f f

Primary stress on the first syllable. Secondary stress on the penult, or on odd numbered syllables (undecided, not indicated).

¨Nalowari ‘sweat’

¨dabaNadinu ‘ask’

Zwaan, J.D. de (1969). A Preliminary Analysis of Gogo-Yimidji: A Study of the Structure of the Primary Dialect of the Aboriginal Language Spoken at the Hopevale Mission in North Queensland. Australian Aboriginal Studies, No. 16, Linguistic Series, No. 5. Australian Institute of Aboriginal Studies, Canberra. Gunwinggu; Mayali [I;P] Australian, Gunwingguan, Gunwinggic. Western Arnhem Land. f

Primary stress falls on the first or the penultimate syllable.

Na ¨djiridburen

‘i wash’

Na ¨bugan ‘i will show you’

Oates, L.F. (1964). A Tentative Description of the Gunwiggu Language (of Western Arnhem Land). Oceania Linguistic Monographs, No. 10. University of Sydney, Australia. Jaru; Djaru [F/F] Australian, Pama-Nyungan, South-West, Ngumbin. Southeastern Kimberley region. f f

Primary stress is located on the first long vowel or the first syllable. Stress may also fall on the first syllable of a trisyllabic second morpheme or a disyllabic one containing a heavy syllable (in which case the initial stress may be deleted).

¨maNari ‘vegetable food’

gida¨łaøNił ‘very good’

Tsunoda, T. (1981). The Djaru Language Family of kimberley, Western Australia. Pacific Linguistics, Series B, No. 78. Australian National University, Canberra.

Australian languages

675

Juat [I] Australian – Pama-Nyungan – Southwest Pama-Nyungan. Western Australia. f f

Primary stress is located on the first syllable. A secondary stress appears on the penult.

¨pini˙juku ‘generation level’ Douglas, W.H. (1968). The Aboriginal Languages of South-West Australia. Australian Aboriginal Studies, No. 14, Linguistic Series, No. 4. Australian Institute of Aboriginal Studies, Canberra. Juwalarai; Yuwaalaraay (dialect of Yualyai) [all/F] Australian – Pama-Nyungan – Wiradjuric. Central New South Wales. f

f

Primary stress on all long vowels, or in the absence thereof on the first syllable. Secondary stress on alternate syllables preceding and following the main stress.

¨gidul˙gara

‘snail’

¨maø¨laøbi˙di ‘big tree’

Williams, C.J. (1980). A Grammar of Yuwaalaraay. Pacific Linguistics, Series B, No. 74. Australian National University, Canberra. Kalkutung [I] Australian, Pama-Nyungan, Galgadungic. Western Queensland. f f

f

Primary stress is word initial. Secondary stress on alternate syllables after the main stress, but not the final. Secondary stress on all non-final odd syllables of polysyllabic morphemes.

¨jalka˙pari

‘boomerang’

¨kunku˙ju-˙ati ‘two daughters’

Blake, B.J. (1969). The Kalkatungu Language. Australian Institute of Aboriginal Studies, Canberra. Kokata; Gugada [I] Australian, Pama-Nyungan, South-West, Wati. SA, Pimba, Mt. Eba, Coober Peby.

676 f f

Australian languages

Primary stress is word initial. Secondary stress on alternate syllables after the main stress, but not the final.

¨rabidna ‘rabbit-obj-clitic’

¨wala˙binjdjala

‘being angry’

Platt, J.T. (1972). An Outline Grammar of the Gugada dialect: South Australia. Australian Aboriginal Studies, No. 48, Linguistic Series, No. 20. Australian Institute of Aboriginal Studies, Canberra. Kuku-Yalanji [I] Australian, Pama-Nyungan, Yalandyic. Queensland. f f

Primary stress on the first syllable. Secondary stress on the penult of words with more than three syllables.

¨yaba ‘elder brother’ ¨marbaji ‘stringy bark’

¨juluNgarr ¨baja˙burray

‘flood’ ‘tired’

Oates, H.H.J. & L.F. Oates (1964). Gugu-Yalanji Linguistic and Anthropological Data. In Gugu-Yalanji and Wik Munkan Language Studies 1–17, Australian Institute of Aboriginal Studies, Canberra. Kunjen [U;P] Australian, Pama-Nyungan, Paman, Central Pama. Cape York Peninsula. f f f

Monosyllables are stressed. Two syllable words have final stress. Words longer than two syllables have unpredictable final or penultimate stress (not given).

¨eg

‘head’

a¨may ‘big’

Sommer, B.A. (1969). Kunjen Phonology: Synchronic and Diachronic. Pacific Linguistics, Series B, No. 11. Australian National University, Canberra. Kuuku Ya’u [L/F] Australian, Pama-Nyungan, Paman, Northeastern Pama. Cape York Peninsula. f

f

Primary stress falls on the last long vowel, on the first vowel in words with only short ones. Secondary stress on the second syllable if the first is a short main stressed one.

Australian languages f f

677

Secondary stress on the initial if the main stress falls elsewhere. Secondary stress on all non-final long vowels.

˙wa¨taaya

‘permit’

¨pim˙/intaka ‘hawk’

¨wiimunu ‘large number of ants’ ¨pa˙ma

‘aboriginal person’

Thompson, D.A. (1976). A Phonology of Kuuku-Ya/u. In P. Sutton (ed.) Languages of Cape York, 213–235. Australian Institute of Aboriginal Studies, Canberra. Lamu-lamu; Lamalama [S] Australian, Pama-Nyungan, Paman, Lamalamic. Cape York Peninsula. f

Primary stress on the first syllable if it has an onset, else on the second.

¨tutulu ‘black’

ar¨kulan ‘moon’

Laycock, D. (1971). Three Lamalamic Languages of North Queensland. Papers in Australian Linguistics 4, Pacific Linguistics, Series A, No. 17, pp. 71–97. Australian National University, Canberra. Laragia; Larakia [S%T/I] Australian, Laragiyan. Near Darwin, Western Arnhem Land. f f

Primary stress is initial unless the second or third syllable is long. Secondary stress on alternate syllables following the main stress.

¨madawa ‘animal, bird, meat’ ¨gwaøra˙bila ‘bandicoot’

gwaø¨gwaøgba ‘crow’ mi¨luølula ‘species of kangaroo’

Capell, A. (1984). The Laragia Language. Pacific Linguistics, Series A, No. 68, pp. 55–106. Australian National University, Canberra. Mangarayi [P;P%A;I] Australian, Gunwingguan, Mangarayic. Mataranka and Elsey stations, Northern Territory. f f

f

f

Stress in disyllables is initial. In trisyllables it is penultimate with a few exceptions that have intial stress. In longer words it is initial and penultimate (often unclear which is primary). In some longer words stress is antepenultimate.

678

Australian languages

¨gaŒugu

‘woman’

gu¨rajn)in

¨wuru˙mumu yi¨riniwun

‘hornet’ ‘hard tree wax’

˙warin¨jalan ‘Exocarpus Latifolius’ ¨baŒ jar ‘water goanna’

‘blood’

Merlan, F. (1989). Mangarayi. New York: Routledge. Mantjiltjara; Mandjildjara (dialect of Martu Wanka) [I] Australian, Pama-Nyungan, South-West, Wati. Gibson Desert, Western Australia. f f f

Primary stress on the first syllable. Secondary stress on alternates in words with more than three syllables. Secondary stress may vary in counting from the penult or the main stress.

¨yapu

‘stone’

¨waæ pu˙kur u ‘tall’

¨kakar a

‘east’

¨pakalku˙}aku

‘let him come out’

Marsh, J. (1969). Mantjiltjara Phonology. Oceanic Linguistics 8, 131–152. Maranunggu [I] Australian, Daly, Bringen-Wagaydy, Wagaydy. Hermit Hill/Daly River. f f

Primary stress falls on the initial syllable. Secondary stress is located on all odd syllables after the main stress.

¨mere˙pet

‘beard’

¨yangar˙mata

¨tQni˙kulkul

‘[kind of ] rabbit’

¨langka˙rate˙ti ‘prawn’

‘the pleiades’

Tryon, D.T. (1970). An Introduction to Maranungku. Pacific Linguistics Monographs, Series B, No. 14. Australian National University, Canberra. Margany [I (I/I)] Australian, Pama-Nyungan, Maric, Mari. South Queensland. f f f

Primary stress on the first syllable. Secondary stress on the third syllable. Alternatively, primary stress on a long /a/ in the second syllable.

¨d5 au˙bira ‘waddy’

¨bad5 in5 iya ‘i am sick’

¨wabata˙ban5 i ‘is going along’

Breen, J.G. (1981). Margany. In: Bob Dixon & Barry Blake. The Handbook of Australian Languages 2, 275–393. Amsterdam: John Benjamins.

Australian languages

679

Martuyhunira; Martuthunira [I (NMS)] Australian, Pama-Nyungan, South-West, Coastal Ngayarda. Northwest coast near Dampier. f

f

All morphemes larger than one syllable have a stress on the first syllable. Alternate non-final syllables after the first also carry an equally prominent stress.

¨pangu

‘good’

¨kanyara

‘man’

¨warra¨ma-l

‘to make’

¨mangkuru-¨thuni

‘peter creek’

Dench, A.C. (1987). Martuthunira: A language of the Pilbara Region of Western Australia. Ph.D. Dissertation. Australian National University, Canberra. Dench, A.C. (1995). Martuthunira: A Language of the Pilbara Region of Western Australia. Pacific Linguistics, Series C, No. 125. Australian National University, Canberra. Mathimathi [S/I] Australian, Pama-Nyungan, Karnic. New South Wales. f

f

f f

Primary stress on the second syllable if it is closed or contains a diphthong. Primary stress on the second if the first is light and the second has a coronal onset. Otherwise stress is initial (always in bisyllables). Secondary stress on the first if the second has main stress, and on all alternates.

¨bugi

‘bad’

˙wi¨ridab

˙di¨bargi˙maTa

‘to adhere’

¨geNgi˙nin ‘[your] uncle’

‘whirlwind’

Hercus, L. (1986). The languages of Victoria: A late survey. Pacific Linguistics, Series C, No. 77. Australian National University, Canberra. Maung [P;A/P] Australian, Yiwaidjan, Yiwaidjic. Goulburn Island, Arnhem Land, Northern Territory. f f f f

Words of up to three syllables are equally stressed on the first two. Longer words have penultimate stress. Secondary stresses on alternates before the penult. When a closed syllable precedes the penult, it may take stress.

680

Australian languages

¨gum¨bil

‘chest’

a˙wuni¨laNuN ‘he was eating them’

¨ba¨ladji

‘bag’

˙jini¨wudbunjan

‘he started it’

Capell, A. & M.E. Hinch (1970). Maung Grammar: Text and Vocabulary. The Hague: Mouton. Mayapi; Mayi; Mayi-Yapi [I] Australian, Pama-Nyungan, Paman, Mayabic. Queensland. f Primary stress is word initial. f Secondary stress is penult in trisyllables. f In longer words, secondary stress appears on non-final alternates after the first. ¨wana˙tuNkari

‘moon’

¨ikala-˙mondu ‘grass-abl’

Breen, J.G. (1981). The Mayi Languages of the Queensland Gulf Country. Australian Institute of Aboriginal Studies, Canberra. Mbabaram [S] Australian – Pama-Nyungan – Pama-Maric. Cape York Peninsula. f The first syllable with an onset is stressed. a¨Nal ‘boomerang’

¨beirubu ‘dog’

Dixon, R.M.W. (1970). Languages of the Cairns Rain Forest Region. In: S.A. Wurm & D.C. Laycock (eds.) Pacific Linguistic Studies in Honour of Arthur Capell, pp. 651–687. (PL Series C, 13). Canberra. Mpakwathi; Anguthimri [I] Australian, Pama-Nyungan, Paman, Northern Pama. Cape York Peninsula. f f

Primary stress is word initial. Secondary stress on alternate syllables after the main stress, but not the final.

¨Du/u ‘yamstick’ ¨paøna ‘level’

¨kalipwa ¨/unu˙wana

‘gully’ ‘blister’

Crowley, T. (1981). The Mpakwathi dialect of Anguthimri. In R.M.W. Dixon & B. Blake. The Handbook of Australian languages. Vol. 2, 146–149. Amsterdam: John Benjamins.

Australian languages

681

Mullukmulluk; MalakMalak [F (CNT)] Australian, Daly, Malagmalag, Malagmalag Proper. Daly river, Northern Territory. f In words with an odd number of syllables stress is on the second syllable and all even syllables thereafter. f In words with an even number of syllables stress is on the first syllable and all odd syllables thereafter. ¨yo_nto_n ‘he’ ¨munan˙kara ‘beautiful’

an¨kini˙yaNka ‘you and i will stand’ ¨no_Nko_˙r o_no_˙yuNka ‘you [pl] will lie down’

Birk, D. (1975). The Phonology of Malakmalak. Pacific Linguistics, Series A, No. 39. Australian National University, Canberra. Muruwari; Murawari [F/F] Australian, Pama-Nyungan, Muruwaric. Border of New South Wales/ Queensland. f f f

Primary stress falls on the first long vowel. Or on the first syllable if there are no long vowels. Secondary stress on the penultimate syllable of long enough words.

pa¨laø ‘plain’

¨kuthara ‘child’

¨karraka˙rranti ‘mirage’

Oates, L.F. (1988). The Muruwari Language. Pacific Linguistics, Series C, No. 108. Australian National University, Canberra. Narrinyeri; Jabulajabula [I] Australian, Pama-Nyungan, Ngarinyeric-Yithayithic. New South Wales. f f

Primary stress is word initial. Secondary stress on the third syllable of words longer than two syllables.

¨poøli ‘child’ ¨mura˙lapi ‘small’

¨maøra˙wi ‘sister’ ¨tiøtu˙maørunelaør ‘they play’

Yallop, C. (1975). Narinjari: an outline grammar of the language studied by George Taplin with Taplin’s notes and comparative tables. Oceanic Linguistic Monographs, No. 17. University of Sydney, Australia. Ngalakan [P] Australian, Gunwingguan, Ngalakanic. North of Roper River, Arnhem Land.

682 f

Australian languages

Primary stress is largely penultimate, but there are many exceptions.

¨alki/

‘still’

didi¨bawaba ‘lotus bird’

¨baragal 8/

‘spear’

barabara¨dakul5 ‘crested pigeon’

Merlan, F. (1983). Ngalakan Grammar Texts and Vocabulary. Pacific Linguistics, Series B, No. 89. Australian National University, Canberra. Ngalkbun; Dalabon; Boun [P;I] Australian, Gunwingguan, Gunwinggic. Bulman Gorge and Wilton R., Northern Territory. Capell: f Main stress falls on the initial syllable, f Secondary stress on alternate syllables thereafter, but not on the final. Sandefur & Jenhan: f Primary stress on the penult. f In trisyllables, primary stress may be on the initial or the penult. f Secondary stress on the first syllable of words longer than three syllables. ¨motjan/ ‘wild honey’ na¨komtutj or ¨nakomtutj ‘little boy’ ˙wulkun¨tjanNan ‘my younger sister’ ˙Na/yenjyenjtju¨Niyan ‘i will talk’ Capell, A. (1962). Some Linguistic Types in Australia. Oceania Linguistic Monographs, No. 7. University of Sydney. Sandefur, J. & D.N. Jenhan (1977). A tentative description of the phonemes of the Ngalkbun language. Work papers of the SIL-AAB. Series A1, 57–96. Darwin. Ngarinyin; Ungarinjin [I/I] Australian, Wororan, Ungarinjinic. Derby to King River, Kimberley, Western Australia. f

Stress on the first syllable of single morpheme words unless the second has a long vowel.

¨Nudmanda ‘sweat’ ¨daøwuru ‘beard’ wa-Nan¨daørenjiriwalu ‘they two are not coming to me’ Capell, A. (1970). Stress and intonation. In Oates, H.H.J. & L. Oates (eds.) A Grammar of Ngarinjin: Western Australia. Australian Aboriginal Studies 25. Linguistics Series 10, Australian Institute of Aboriginal Studies, Canberra.

Australian languages

683

Nunggubuyu [P] Australian, Gunwingguan, Nunggubuan. Numbulwar, Arnhem Land, Northern Territory. f f

Primary stress on the penultimate syllable. Secondary stress on the first and all alternates thereafter, but never the antepenult.

wu¨rugu

‘billabong’

˙ambala¨lari ‘poor’

˙ngalaa¨ligi

‘turtle’

˙mara˙garri¨jinyung

‘shark species’

Hore, M. (1981). Syllable length and stress in Nunggubuyu. In: AAB SIL Working papers. Series A5, B. Waters (ed.). 1–62. Nyawaygi [F (CNT)] Australian, Pama-Nyungan, Nyawaygic. East Coast of Queensland near Townsville. f f

f f

Initial long vowels always receive primary stress. If there are no long vowels, stress the first syllable in two and four syllable words. And stress the second in three and five syllable words. Secondary stress on alternates after the primary, but never the last.

¨gina

‘man’

¨giøba}i ‘south’

¨biya˙gala ‘water snake’ bul¨biri

‘quail’

Dixon, R.M.W. (1981). Nyawaygi. In R.M.W. Dixon & B.J. Blake (eds.) Handbook of Australian Languages 2, 430–531. Amsterdam: John Benjamins. Pintupi-Luritja [I] Australian, Pama-Nyungan, South-West, Wati. Eastmost part of Western Australia. f f

Primary stress is word initial. Secondary stress on alternate syllables after the main stress, but not the final.

¨paNa

‘earth’

¨puæiN˙kalatju ‘we [sat] on the hill’

¨tjuaya

‘many’

¨tjamu˙limpa˙tjuNku ‘our relation’

Hansen, K.C. & L.E. Hansen (1969). Pintupi Phonology. Oceanic Linguistics 8, 153–170.

684

Australian languages

Pitjantjatjara [I] Australian, Pama-Nyungan, South-West, Wati. Western Australia. f f

Primary stress is word initial. Secondary stress reported on alternate syllables after the main stress, but not the final.

¨pitilyir ‘temple’ Douglas, W.H. (1959). An Introduction to the Western Desert Language of Australia. Oceania Linguistic Monographs, Sydney. Glass, A.D. & D. Hackett (1970). Pitjantjara Grammar: A Tagmemic View of the Ngaanyatjara (Warburton Ranges) Dialect. Australian Aboriginal Studies, No. 34, Linguistic Series, No. 13. Australian Institute of Aboriginal Studies, Canberra. Pitta pitta; Bidhbidha [I] Australian, Pama-Nyungan, Karnic, Palku. Southwest Queensland. f f

Primary stress is word initial. Secondary stress on alternate syllables after the main stress, but not the final.

¨kupi ‘fish’ ¨tatji˙liNa ‘to eat’

¨miNkara ¨kaå a˙kaiaNu

‘nails, claws’ ‘come-past hither’

Blake, B.J. & I. Green (1971). The Pitta Pitta Dialects. Linguistic Communications 4. Monash University, Melbourne. Blake, B.J. (1979). Pitta-Pitta. In R.M.W. Dixon & B.J. Blake (eds.) Handbook of Australian Languages 1, 182–242. Amsterdam: John Benjamins. Ritarungo [I] Australian, Pama-Nyungan, Yuulngu. Northern Arnhem Land. f f

Primary stress is word initial. No secondary stress.

¨baøpa ‘father’

¨murawat i

‘a fruit bearing tree’

Heath, J. (1980). Basic Materials in Ritharugu: Grammar, texts and dictionary. Pacific Linguistics, Series B, No. 62. Australian National University, Canberra.

Australian languages

685

Thaypan; Kuku-Thaypan [S] Australian, Pama-Nyungan, Paman, Rarmul Pama. Cape York Peninsula. f

Stress falls on the first syllable with an onset.

al¨ka ‘spear’

¨knbalaN ‘male grey kangaroo’

Rigsby, B. (1976). Kuku-thaypan Descriptive and Historical Phonology. In P. Sutton (ed.) The Languages of Cape York. Australian Aboriginal Studies, RRS-6, AIAS, Canberra. Thurawal [I/I] Australian, Pama-Nyungan, Yuin-Kuric, Yuin. New South Wales. f

f

Primary stress on the first syllable, unless the vowel of the second is the only long one. No secondary stress.

¨gaøndi ‘smoke’

wa¨raøNa ‘boomerang’

Eades, D.K. (1976). The Dharawal and Dhurga Languages of the New South Wales South Coast. Australian Aboriginal Studies Research and Regional Studies, No. 8. Australian Institute of Aboriginal Studies, Canberra. Tiwi [P] Australian, Tiwian. Bathurst and Melville Islands, Nguiu, Northern Territory. f f f

Primary stress is located on the penultimate syllable. In words longer than three syllables there is an initial secondary stress. Su‰xes longer than one syllable get their own stress.

¨poja ‘seed’

tu¨wara

‘tail’

˙kalam¨para

‘sweat’

Osborne, C.R. (1974). The Tiwi Language. AIATSIS. Australian Aboriginal Studies, No. 55. Linguistics Series 21. Canberra. Lee, J. (1987). Tiwi Today: A Study of Language Change in a Contact Situation. Pacific Linguistics, Series C, No. 96. Australian National University, Canberra. Uradhi [S] Australian, Pama-Nyungan, Paman, Northern Pama. Cape York Peninsula.

686

Australian languages

Hale: f Stress the first postconsonantal vowel (is always in the first or second syllable). Crowley: f Stress on heavy syllables, else on the antepenult. ¨jukuk

‘tree’

u¨tagak ‘dog’

i¨pik

‘water’

¨minhik5 jik5

‘bird’

Hale, K. (1976). Phonological Developments in a Northern Paman Language: Uradhi. In: P. Sutton (ed.) The Languages of Cape York. Australian Aboriginal Studies RRS-6. AIAS, Canberra. Crowley, T. (1983). Uradhi. In: B. Dixon & B. Blake (eds.) The Handbook of Australian Languages 3, 306–428. Amsterdam: John Benjamins. Waalubal (dialect of Bandjalang) [I/I] Australian, Pama-Nyungan, Bandjalangic. New South Wales. f

f

Stress is initial except in cases where the first vowel is short and the second vowel is long. Secondary stress on long vowels.

¨bandaN

‘other’

¨baøłbini

¨wurgu˙luøm

‘magpie’

¨Na˙maølu ‘tree goanna-erg’

‘took out’

Crowley, T. (1978). The Middle Clarence Dialects of Bandjalang. Australian Institute of Aboriginal Studies, Canberra. Walmajarri; Walmatjari [I] Australian, Pama-Nyungan, South-West, Ngumbin. South of Fitzroy Crossing. f

f f

In disyllables and words longer than three syllables primary stress is initial. In trisyllables primary stress varies between initial and second. Secondary stress on alternates after the primary or on the penult.

¨yapa ¨walaku˙waå i

‘child’ ‘sons’

¨kaNani or ka¨Nani ‘carried’ ¨njantu˙ku}aæu ‘his-subj’

Hudson, J. & E. Richards (1969). The Phonology of Walmatjari. Oceanic Linguistics 8, pp. 171–189.

Australian languages

687

Wambaya [I/I] Australian, West Barkly, Wambayan. Northern Territory, Barkly Tablelands. f

f f

Stress is initial except in cases where the first vowel is short and the second vowel is long. Secondary stress on alternates after the main stress, but not the final. Sequences of two or more su‰xes form a stress domain.

¨bulyungu ‘little.iv’ ga¨rdaala ‘gidgee tree’

¨ngurra˙ramba ¨dagumaj-˙barli

‘night time’ ‘hit-agnt-i’

Nordlinger, R. (1998). A Grammar of Wambaya, Northern Territory (Australia). Pacific Linguistics, Series C, No. 140. Australian National University, Canberra. Wangaybuwan-Ngiyambaa [F/F] Australian, Pama-Nyungan, Wiradhuric. New South Wales, Darling, Barwon, Bogan. f

f

f

f

Primary stress on the first long vowel, or the first syllable when long vowels are absent. Secondary stress on polysyllabic su‰xes by the same rule as for primary stress. Secondary stress on the first syllable if it, or the second, is not already stressed. Secondary stress on alternates after the main stress.

¨girala ‘star’ ˙gaba¨daø-ga ‘moon-loc’ ¨yana-˙buna-˙wa-ha ‘go along back’ Donaldson, T. (1977). A Description of Ngiyamba: the language of the wanga:ybuwan people of Central New South Wales. Ph.D. Dissertation. ANU, Canberra. Donaldson, T. (1980). Ngiyambaa: The Language of the Wangaaybuwan. Cambridge: Cambridge University Press. Wardaman [P;A (NMS)] Australian, Gunwingguan, Yangmanic, Yibwan. Northern Territory, upper Daly River.

688 f f

Australian languages

Stress on the penultimate syllable and alternate syllables to the left. In trisyllables and longer words there is an alternative antepenultimate pattern.

¨yi-bam

‘head’

¨jarnarran ‘pelican’

ja¨warrga

‘liver’

¨gura¨mana

‘gums’

Merlan, F. (1994). A Grammar of Wardaman. Berlin: Mouton de Gruyter. Warlpiri; Walbiri; Wailbry [I] Australian, Pama-Nyungan, South-West, Ngarga. West central part of Northern Territory. f f f

Primary stress is initial. Secondary stress on alternates after the main stress, but not the last. Start secondary stress assignment anew with every morpheme.

¨watiya ‘tree’ ¨wati˙ya-rla ‘tree-loc’

¨manaN˙karra ¨yaparta-˙Nurlu

‘spinifex plain’ ‘father’s mother-elat’

Nash, D. (1986). Topics in Warlpiri Grammar. Garland, NY. Reece, L. (1970). Grammar of the Wailbri Language of Central Australia. Oceania Linguistic Monographs 13, Sydney. Warrgamay [F (CNT)] Australian, Pama-Nyungan, Dyirbalic. Queensland, between Cairns and Townsville. f f

f f

Initial long vowels always receive primary stress. If there are no long vowels, stress the first syllable in two and four syllable words. And stress the second in three and five syllable words. Secondary stress on alternates after the primary, but never the last.

¨giøba}a ‘fig tree’ ¨gia˙wulu ‘freshwater jewfish’

ga¨gara u¨}agay˙miri

‘dilly bag’ ‘from niagara vale’

Dixon, R.M.W. (1981). Wargamay. In R.M.W. Dixon & B.J. Blake (eds.) Handbook of Australian Languages 2, 1–144. Amsterdam: John Benjamins. Wongkumara; Wankumara (dialect of Ngura) [I] Australian, Pama-Nyungan, Karnic, Ngura. Queensland.

Australian languages f f f

689

Primary stress is initial. Secondary stress on alternates rightward, but not on the final. Disyllabic su‰xes receive a secondary stress on their first syllable.

¨mirki˙riłi ‘moon’

¨maku˙ra-ni

‘stick-nom’

McDonald, M. & S.A. Wurm (1979). Basic Materials in Wangkumara: Grammar, Sentences and Vocabulary. Pacific Linguistics, Series B, No. 65. Australian National University, Canberra. Yanyuwa; Anyula; Janjula [P (NMS)] Australian, Pama-Nyungan, Yanyuwan. Gulf of Carpentaria region Limmen Bight river. f

f

Stress is on the penultimate syllable, and sometimes on the first of trisyllables. In words longer than three syllables the first syllable is also stressed.

¨Nawu

‘cloud’

¨mar u¨war a ‘cousin’

ni¨wanji

‘animal flesh’

ganj¨mada or ¨ganjmada ‘two’

Kirton, J.F. (1977). Anyula Phonology. Pacific Linguistics, Series A, No. 10. Australian National University, Canberra. Yidiny [F/F] Australian, Pama-Nyungan, Yidinic. Inland and coast of East Central Queensland. f f

Primary stress on the first syllable with a long vowel, else the first. Secondary stress on alternates preceding and following the main stress.

gu¨luødu ¨yalu˙Nunda ˙wuNa¨baødji˙ łunda wa¨raøbu˙ga

‘dove’ ‘today, nowadays’ ‘hunt.dji.dat.subord’ ‘white apple tree’

Dixon, R.M.W. (1977). Some Phonological Rules in Yidiny. Linguistic Inquiry 8, 1–34. Dixon, R.M.W. (1977). A Grammar of Yidin y. Cambridge: Cambridge University Press. Nash, D. (1979). Yidiny Stress: a metrical account. In: E. Battistella (ed.) Proceedings of the ninth annual meeting of the North East Linguistic Society, CUNYForum, No’s. 7–8.

690

Australian languages

Yindjibarndi; Jindjibandi [I/ I] Australian, Pama-Nyungan, South-West, Coastal Ngayarda. Coast of Western Australia. f

f f

Stress is initial except in cases where the first vowel is short and the second vowel is long. Secondary stress on alternates after the main stress. If the second long vowel is high, it may break and carry a secondary stress on the second part, main stress remains initial in this case.

¨wilar˙ra

¨piri˙i

‘moon’

nyi¨laarti ‘native mead’

‘match’

¨purta˙tyirri ‘impassibly rough ground’

Wordick, F.J.F. (1982). The Yindjbarndi Language. Pacific Linguistics, Series C, No. 71. Australian National University, Canberra. Yir Yoront [I] Australian, Pama-Nyungan, Paman, Western Pama. Cape York Peninsula. f

Primary stress on the first syllable.

¨lalpuym ‘bundle’

¨melungkurr ‘one-wire spear’

Alpher, B. (1991). Yir Yoront Lexicon. Sketch and Dictionary of an Australian language. Trends in Linguistics, Documentation 6. Berlin: Mouton de Gruyter, New York.

If the a‰liation is not in italics this means it is not from the Ethnologue but from Voegelin & Voegelin.

Austronesian languages

Aceh Atjeh; Atjehnese; Achehnese, Achinese [U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Mlayic, Achinese-Chamic, Achinese. Northern Aceh Province, Sumatra (Indonesia). f f

Primary stress on the final syllable. Secondary stress on the first syllable, unless this syllable contains the vowel //.

˙ma¨ta ‘eye’

˙kandu¨ri ‘feast’

[k¨dE]

‘shop, market’

Cowan, H.J.K. (1981). An outline of Achehnese phonology and morphology. Bulletin of the Society of Oriental and African Studies 44: 522–549. Durie, M. (1985). A grammar of Achehnese on the basis of a dialect of north Aceh. Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 112. Dordrecht: Foris. Durie, M. (1995). Aceh. In D.T. Tryon (ed.) Comparative Austronesian Dictionary, Part 1: 407–420. Berlin/New York: Mouton de Gruyter. Langen, K.F.H. van (1889). Handleiding voor de Beoefening van de Atjesche Taal. The Hague: M. Nijho¤ (KITLV). Aklanon; Aklan [P/U;P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Meso Philippine, Central Philippine, Bisayan, West, Aklan. Panay Isl. (Philippines). f

f f

Primary stress on the penultimate syllable when it is closed, else on the final. Secondary stress on alternates before the main stress, respecting weight. Secondary stress on the final syllable when primary stress is penultimate.

pi¨tu ‘seven’ /a¨sir˙tar ‘lucky’

¨gas˙ta ‘spend’ ki˙napu¨tus ‘wrap instrument-foc-past-post’

Chai, N. (1971). A grammar of Aklan. Unpublished doctoral dissertation, University of Pennsylvania, Philadelphia.

692

Austronesian languages

Alune; Sapalewa; Patasiwa, Alfoeren [P;A (IRR)] Austronesian, Malayo-Polynesian, Central-Eastern, Malayo-Polynesian, Central Maluku, East, Seram, Nunusaku, Three Rivers, Amalumute, Northwest Seram, Ulut Inai. Seram, Ulut Inai. f f f

Stress falls on the penultimate syllable, even when su‰xes are added. Stress does not move rightward in verbs formed with -ke. Demonstrative enclitics carry primary stress.

mu¨sute ‘heat’ bu¨nikele ‘to hide it’

¨suku ‘to bathe’ tamata¨re ‘that person’

Taguchi, Y. & T. Taguchi (1990). A phonology of Alune. Workpapers in Indonesian Languages and Culture 8: 95–127. Ambon: Pattimura University þ SIL. Ambrym, Southeast [U+P/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, North and Central Vanuatu, Northeast Vanuatu-Banks Islands, East Vanuatu. Ambrym Island (Vanuatu). f f

Stress falls on the penultimate syllable if the final is short. If the final syllable contains a long vowel or is closed stress is variable or is constant in both the penult and the final.

¨nim ‘house’ ¨ku¨lii ‘dog’

re¨vul ‘hole’ a¨tuli ‘girl’

Parker, G.J. (1968). Southeast Ambrym Phonology. Oceanic Linguistics 7, pp. 81–91. Apoze (dialect of Kela) [A] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, North New Guinea, Huon Gulf, North. Coastal. Morobe District (Papua New Guinea). f

Stress falls on the antepenultimate syllable (initial in short words).

Collier, K. & M. Collier (1975). A tentative Phonemic Statement of the Apoze dialect, Kela Language, in R. Loving, (ed.) Phonologies of five Austronesian Languages. Summer Institute of Linguistics, Ukarumpa, Papua New Guinea, pp. 129–161.

Austronesian languages

693

Aralle-Tabulahan (dialect of Pitu Ulunna Salu) [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, South Sulawesi, Pitu Ulunna Salu. South Sulawesi, between Mandar and Kalumpang (Indonesia). f

Stress falls on the penultimate syllable.

¨tama

‘to enter’

pa¨tama

¨dasaN

‘house’

da¨saNku ‘my house’

‘to cause to enter’

McKenzie, R. (1991). Phonology of Aralle-Tabulahan. Workpapers in Indonesian Languages and Culture 12: 99–149. Ujung Pandang: SIL. Arosi [I] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Southeast Solomonic, Malaita-San Cristobal, San Cristobal. West San Christobal (Solomon Isl). f

Stress falls on the initial syllable.

¨dangi

‘day’

¨mwQrahu ‘a chief ’

Capell, A. (1971). Arosi Grammar. Pacific linguistics, Series B, No. 20. Australian National University, Canberra. Balantak; Kosian [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, Central Sulawesi, Eastern. East Central Sulawesi, Banggi District (Indonesia). f f

Primary stress falls on the penultimate syllable. Secondary stress on the first syllable of four syllable words.

¨liNkoN ‘fern’

ba¨koko/ ‘machete’

˙bolu¨sukon ‘durian’

Busenitz, R.L. & M.J. Busenitz (1991). Balantak Phonology and Morphophonemics. NUSA Monographs 33: 29–47. Jakarta: NUSA. Bali; Balinese [U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Bali-Sasak. Island of Bali, northern Nusapenida, Lombok Islands, east Java, south Sulawesi. f

Stress falls on the final syllable.

694

Austronesian languages

ba¨p« ‘father’ Barber, C.C. (1977). A grammar of the Balinese language, University of Aberdeen Occasional Publications 3 (Aberdeen University Library). Clynes, A. (in prep.) A grammar of a Balinese dialect, Ph.D. Dissertation, ANU, Canberra. Kersten, J. (1948). Balische grammatica. The Hague: Van Hoeve. Clynes, A. (1995). Balinese. In D.T. Tryon (ed.) Comparative Austronesian Dictionary, Part 1: 495–510. Berlin/New York: Mouton de Guyter. Ward, J.H. (1973). Phonology, morphophonemics and the dimensions of variation in spoken Balinese, Ph.D. Dissertation, Cornell University. Bambam; Pitu Ulunna Salu; Pitu Ulunna Binanga [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Northern, Pitu Ulunna Salu. South Sulawesi, Polmas District (Indonesia). f

Stress falls on the penultimate syllable.

da¨haNku ‘my horse’ Campbell, P.J. (1991). Phonology of Pitu Ulunna Salu. Workpapers in Indonesian Languages and Culture 12: 1–52, Ujung Pandang, Sulawesi: SIL. Banggai; Aki [U/P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, Central Sulawesi, Banggai. Central Sulawesi (Indonesia). f

f

f

Primary stress falls on the penultimate syllable, unless the final syllable is long. Secondary stress falls on the first syllable of words containing four syllables or more. If longer words contain a diphthong before the main stress, it carries secondary stress.

maN¨kabi ‘to ask’ maNka¨bai ‘to make’

a¨saaN ‘breath’ ˙saliNu¨NudZon ‘to lament’

Bergh, J.D. van den (1953). Spraakkunst van het Banggais. The Hague: M. Nijho¤.

Austronesian languages

695

Bantik [P;A;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Sangiric. Northeast section of the north Sulawesi; three villages near Manado. f f

Stress falls on the penultimate syllable. There are exceptions with antepenultimate or final stress.

¨sahaN

‘tree’

¨likudu/ ‘back’

sa¨haN

‘ant’

ma¨muku ‘to kneel’

Sneddon, J.N. (1984). Proto Sangiric and the Sangiric Languages. Pacific Linguistics, Series B, No. 91. Australian National University, Canberra. Betawi; Jakarta Malay; Batavi, Batawi, Melayu Jakarte [P/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Malayic. Jakarta, Java (Indonesia). f f

Stress mostly falls on the penultimate syllable. If the penult contains a schwa and the final does not, stress is final.

tu buku ¨mErah ‘that red book’

tu buku g«¨de ‘that big book’

Ikranegara, K. (1980). Melayu Betawi Grammar. Linguistic Studies in Indonesian and Languages in Indonesia 9. Jakarta: NUSA. Wallace, S. (1976). Linguistic and Social Dimensions of Phonological Variation in Jakarta Malay. Ph.D. Dissertation, Cornell University. Biak; Numfor; Nufoor; Mafoorsch [P;U] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, South Halmahera-West New Guinea, West New Guinea, Cenderawasih Bay, Biakic. Islands of Biak and Numfor. f f

Stress falls on the penultimate syllable. There are exceptions with final stress.

¨Bumi ‘grandmother’ ra¨muma ‘to doubt’

fara¨rur ‘to work’ maNg«¨nem ‘too much’

Hasselt, F.J.F van (1905). Spraakkunst der Nufoorsche Taal. The Hague: M. Nijho¤. Bicolano, central [U/LEX] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Meso

696

Austronesian languages

Philippine, Central Philippine, Bikol, Inland, Iriga. Provinces of Albay, Camarines Sur, Catanduanes and Sorsogon (Philippines). f f

Stress always falls on the final syllable if the two last syllables are closed. Otherwise stress is unpredictable.

la¨bot

‘aperture, cavity, hole’ ¨labot

tug¨tog ‘music’

‘business, concern, care’

apo¨don ‘to call’

Mintz, M. (1971). Bikol grammar notes. University of Hawaii Press, Honolulu. Mintz, M. & J. Del Rosario Britanico (1985). Bikol-English Dictionary. Quezon City: New Day Publishers. Bima; Bimanese [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Bima-Sumba. Sunda Islands, eastern Sumbawa Island (Indonesia). f

Stress falls on the penultimate syllable and stays there when su‰xes are added.

¨ina ‘mother’

¨inamu ‘your mother’

Verheijen, J.A.J. (1982). Komodo, het Eiland, het Volk en de Taal. Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 96. The Hague: M. Nijho¤. Bugis; Buginese; ugi; Boegineesch, Ugi, De’, Rappang Buginese [P;A] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Bugis. South Sulawesi and other parts of Indonesia. f f

Stress falls on the penultimate syllable. Loans ending in /l, r, s/ get an extrametrical syllable V/, resulting in antepenultimate stress.

¨bola ‘house’

¨kambara/ (< /kambar/)

‘twins’

Abas, H. & C.E. Grimes (1995). Bugis. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1, 549–562. Berlin/New York: Mouton de Guyter. Matthes, B.F. (1875). Boegineesche Spraakkunst. The Hague: M. Nijho¤. Samsuri (1965). Introduction to a Rappang Buginese grammar, Ph.D. Dissertation, Indiana University.

Austronesian languages

697

¨ . (1983). The Buginese language (translated by E.H. Tsipan; edited Sirk, U by Lina Skharban), Moscow: Nauka. Buli [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, South Halmahera-West New Guinea, South Halmahera, Southeast. East coast of central Halmahera, north Maluku (Indonesia). f f

Stress falls on the final syllable if it contains a long vowel. Otherwise stress is located on the penult.

sada¨kala ‘lizzard’

kaka¨laø ‘red’

Maan, G. (1951). Proeve van een Bulische Spraakkunst. Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 10. The Hague: M. Nijho¤. Buru; Boeroe; Buruese [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Central Maluku, Buru. Buru Island, Central Maluku, Ambon, Jakarta (Indonesia). f

Stress falls on the penultimate syllable.

Hendriks, H. (1897). Het Burusch van Ma`sa`rete. The Hague: M. Nijho¤. Grimes, C.E. (1995). Buru. In D.T. Tryon (ed.) Comparative Austronesian Dictionary, Part I: 623–636. Berlin/New York: Mouton de Gruyter. Cebuano [U/U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Meso Philippine, Central Philippine, Bisayan, Cebuan. Central Philippines and Mindanao. f

f

f

Primary stress falls on the final syllable if the two last syllables are both heavy or light. If the penult contains a long vowel or coda and the final does not contain a long vowel, stress is located on the penult. Secondary stress falls on alternate syllables to the left of the Primary stress.

¨tinda ‘sell’ /i˙kadu¨ha ‘second’

ba¨to ‘rock’ ˙makig¨daa ‘bring.active.pot.fut’

698

Austronesian languages

Shryrock, A. (1993). A Metrical Analysis of Stress in Cebuano. Lingua 91, 103–148. Chamorro [P/A] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Chamorro. Guam. f f

Stress falls on the penultimate syllable if it has a coda. Otherwise stress falls on the antepenult.

¨pikaru

‘sly’

¨ma_mati ‘reef ’

¨kanay

‘hand’

pu¨lonnum ‘trigger fish’

Chung, S. (1983). Transderivational Relationships in Chamorro Phonology. Language 59: 1, 35–66. Seiden, W. (1960). Chamorro Phonemes. Anthropological Linguistics 2: 4, 6–35. Da’a; Bunggu (dialect of Kaili) [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West Central, Kaili-Pamona, Kaili. Central and South Sulawesi (Indonesia). f f

Stress falls on the penultimate syllable, even when su‰xes are added. Consecutive vowels count as two syllables.

¨tara ‘look at’

ta¨ra.a

‘pineapple’

ku¨leNe

‘tired’

Barr, D.F. & S.G. Barr (1988). Phonology of Da’a, central Sulawesi. In H. Steinhauer (ed.) Papers in Western Austronesian Linguistics 4, Pacific Linguistics Series A, No. 79. Australian National University, Canberra. Barr, D.F. (1995). Da’a. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1, 529–538. Berlin/New York: Mouton de Gruyter. Dehu; Lifu [I] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Loyalty Islands. Lifu Islands, Loyalty Islands. f f

Primary stress falls on the initial syllable. One secondary stress falls on the third syllable.

¨kem ‘father’

¨teDin ‘brother’

¨ala˙mek

‘eye’

Austronesian languages

699

Tryon, D.T. (1967). Dehu Grammar. Pacific Linguistics, Series B, No. 7. Australian National University, Canberra. Dobel; Kobro’or; Kobroor; Doibel [P;U] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Southeast Maluku, Kei-Aru, Aru, South. Aru Islands, Southeast Maluku (Indonesia). f f

Stress falls on the penultimate syllable. There are lexical exceptions with final stress.

da¨tabaj

‘they carry [on shoulders]’

data¨baj ‘they hit’

Hughes, J. & K. Hughes (1990). Dobel Phonology. Workpapers in Indonesian Languages and Culture 8: 43–76. Ambon: Pattimura University þ SIL. Hughes, J. (1995). Dobel. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1, 637–650. Berlin/New York: Mouton de Guyter. Emae; Mae [A] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, Samoic-Outlier, Futunic. Emae Island. f

Stress falls on the antepenultimate syllable.

¨nanafi ‘yesterday’

¨taNata ‘man’

Capell, A. (1962). The Polynesian language of Mae (Emwae), New Hebrides. Linguistic Society of New Zealand, Auckland. Enggano [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sumatra, Enggano. Enggano Island and nearby islands (Indonesia). f

Stress falls on the penultimate syllable, sometimes also when su‰xes are added.

Ka¨hler, H. (1939). Grammatischer Abriss des Enggan. Zeitschrift fu¨r Eingeborenen-Sprachen 30: 81–117. Fijian [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, East Fijian. Fiji Islands.

700 f f f

Austronesian languages

Primary stress falls on the final syllable if it contains a long vowel. Otherwise primary stress falls on the penultimate syllable. Secondary stress falls on remaining long vowels and alternate syllables to the left.

¨mata

ki¨laø

‘eye’

˙nreø¨nreø ‘di‰cult’

‘know’

nda¨liNa ‘ear’

Schutz, A.J. (1985). The Fijian Language. Honolulu: University of Hawaii Press. Dixon, R.M.W. (1988). A Grammar of Boumaa Fijian. Chicago: University of Chicago Press. Fordata; Larat (dialect of Kei-Fordata) [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, South-East Maluku, Kei-Tanimbar, Kei-Fordata. Southeast Maluku (Indonesia). f

Stress falls on the penultimate syllable (long vowels and diphthongs are disyllabic).

¨la.uk ‘protrude’

a¨ka.a ‘what’

¨sumar ‘mouth’

Drabbe, P. (1926). Spraakleer der Fordaatsche Taal. Verhandelingen van het Bataviaasch Genootschap 67/1. The Hague: M. Nijho¤/Batavia: Albrecht & Co. Gayo; Gajo [P/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Gayo. North Sumatra (Indonesia). f

f

Primary stress falls on the penultimate syllable, unless it contains a schwa and the final vowel is not schwa. Secondary stress on the antepenult if main stress is final.

¨tulu

‘three’

k«¨ro ‘fried rice’

˙mur«¨las

‘to collapse’

Hazeu, G.A.J. (1907). Gayosch-Nederlandsch Woordenboek, met Nederlands Gayosch register. Batavia: Landsdrukkerij. Hanuno´o [P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Meso Philippine, South Mangyan, Hanunoo. Mindoro (Philippines).

Austronesian languages f f f

701

Primary stress falls on the penultimate syllable. There are lexical exceptions with final stress. Prefixes may have inherent secondary stress.

paRa/uli¨/an ‘home’

˙nagpaka¨yamu ‘too much, too many’

/um¨puk

¨/umpuk

‘rebound’

‘dispersion, disintegration’

Conklin, H.C. (1953). Hanunoo-English Vocabulary. University of California Publications in Linguistics 9. Hawaiian [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East, Central, Marquesic. Ni’ihau (Hawaii). f f

Primary stress falls on the penultimate syllable. Secondary stress falls on long vowels and alternate syllables to the left (no examples).

a¨kaøka ‘clear’

¨hale

‘house’

hana¨uma [place name]

Elbert, S.H. & S.A. Keala (1961). Conversational Hawaiian. Honolulu: University of Hawaii Press. Elbert, S.H. & M.K. Pukui (1979). Hawaiian Grammar. Honululu, University of Hawaii Press. Iaai; Iai; Yai [I] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Loyalty Islands. Uvea, Loyalty Islands. f f

Primary stress falls on the initial syllable. Secondary stress final in three syllable words and penultimate in longer words.

¨ko ‘whale’ ¨ema˙kan ‘eye’

¨momo ‘woman’ ¨bara˙konen ‘ear’

Tryon, D.T. (1968). Iai Grammar. Pacific Linguistics, Series B, No. 8. Australian National University, Canberra. Indonesian; Bahasa Indonesia [P;no stress] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Malayic, Malayan, Local Malay. Indonesia.

702 f f

f

Austronesian languages

Primary stress falls on the penultimate syllable. Secondary stress on the initial and alternate syllables before the main stress. Serious debate on location and even status of stress (accent only language?).

pErEm¨puan

‘woman’

˙xatulis¨tiwa ‘equator’

¨cari

‘sit’

˙difErensi¨asi ‘di¤erentiation’

Halim, A. (1974). Intonasi. Jakarta: Djambatan. Samsuri (1971). Ciri-ciri prosodi dalam kalimat Bahasa Indonesia. Flores: Nusa Indah. Irarutu; Irahutu [U;LEX] Austronesian, Malayo-Polynesian, Central-Eastern, Unclassified. East Bomberai Peninsula (Irian Jaya, Indonesia). f

Stress falls on the final syllable (with lexical exceptions).

‚a¨nE ‘pig’ wa¨Rada ‘knife’

d«¨bE ‘to hit’ waRa¨Ru ‘news’

Matsumura, T. (1991). Irarutu Phonology. Workpapers in Indonesian Languages and Culture 10: 37–74. Jayapura, Irian Jaya: Cendera-wasih University & SIL. Voorhoeve, C.L. (1995). Irarutu. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 651–658. Berlin/New York: Mouton de Gruyter. Javanese; Jawa [P/U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Javanese. Central Java, Irian Jaya, Sulawesi, Sumatra etc. (Indonesia). f f f

Primary stress falls on the penultimate syllable if it contains a full vowel. If the penultimate vowel is schwa, stress is final. When more than one su‰x is added, stress shifts to the new penultimate syllable, leaving a secondary stress on the fourth syllable from the right.

pA¨NEstu ‘blessing’ k«¨bAN«t«n ‘too bad’

b«¨n«r ‘correct’ pa˙luNgu¨Ane ‘his seat’

Dudas, K.M. (1976). The phonology and morphology of modern Javanese. Ph.D. Dissertation, University of Illinois.

Austronesian languages

703

Herrfurth, H. (1964). Lehrbuch des modernen Djawanish. Leipzig: VEB Verlag Enzyklopa¨die. Horne, E.C. (1961). Beginning Javanese. Yale Linguistics Series 3. New Haven and London: Yale University Press. Noordenbos, C. (1941). Klank en teken van het Javaans. Groningen: Wolters. Prijohoetomo, M. (1937). Javaansche Spraakkunst. Leiden: E.J. Brill. Ras, J.J. (1985). Inleiding tot het Modern Javaans. Dordrecht: Foris. (1st edition: 1977) Kambera (dialect of Sumba) [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Bima-Sumba. Sumba Island south of Flores, Lesser Sundas (Indonesia). f

Stress falls on the penultimate syllable (which is in most cases also initial).

¨nomu

‘six’

¨daøNu ‘be many’

Klamer, M. (1994). Kambera, a language of Eastern Indonesia, HIL Dissertations 11. The Hague: Holland Academic Graphics. Kara [L/F] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Meso Melanesian, New Ireland, Lavongai-Nalik. New Ireland Province, kavieng subdistrict (PNG). f f f

f f

Primary stress falls on the last syllable with an /a/ nucleus. Otherwise on the last closed syllable, or the first syllable. Secondary stress on the first syllable if main stress is not initial or second. Secondary stress on the final syllable if main stress is antepenult. Secondary stress on the penult if main stress is located even further to the left.

¨qHaqsa˙˜« ‘one leg’ ˙ninaN¨pHap ‘stepmother’

‚«¨˜aR«˙d«q ‘straighten’ ¨m«tH«‚«˙tH«m« ‘a family’

Schlie, P. & G. Schlie (1993). A Kara Phonology. In J. Clifton (ed.). Phonologies of Austronesian Languages 2, 99–130. SIL Ukarumpa.

704

Austronesian languages

Karo Batak [P/P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sundic, Sumatra, Batak, Northern. West and nortwest of Lake Toba, Sumatra (Indonesia). f

Stress falls on the penult, unless it contains an open schwa and the final does not.

¨kuda ‘horse’

m«¨dem ‘to sleep, to lie’

Joustra, M. (1901). Over Schrift en Uitspraak van het Karo-Bataksch. Tijdschrift voor Indische Taal- Land- en Volkenkunde 44: 12–23. The Hague: M. Nijho¤/Batavia: Albrecht & Co. Neumann, J.H. (1922). Schets der Karo-Bataksche Spraakkunst, Verhandelingen van het Bataviaasch Genootschap 63. The Hague: M. Nijho¤/ Batavia: Albrecht & Co. Woollams, G. (1996). A Grammar of Karo Batak, Sumatra. Pacific Linguistics, Series C, No. 130. Australian National University, Canberra. Kei; Kai; Evav [U] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Southeast-Maluku, Kei-Tanimbar, Kei-Fordata. Kei Kecil, Kei Basar isl, South Maluku (Indonesia). f

Stress falls on the final syllable.

den¨kut ‘little’

to¨mat ‘human being’

Geurtjens, H. (1922) Spraakleer der Keieesche taal, Verhandelingen van het Bataviaasch Genootschap 63. The Hague: M. Nijho¤/Batavia: Albrecht & Co. Kilivila [U/P;A] Austronesian, Malayo-Polynesian, Central Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Papuan Tip, Peripheral, KilivilaLouisiades, Kilivila. Milne Bay Province, Trobriand Islands. f f

Stress falls on the final syllable if it is closed or contains a diphthong. Otherwise stress is penultimate (or antepenultimate in some special cases).

¨waga ‘canoe’ bobwe¨lila ‘gift’

ba¨kam ‘i will eat’ kukatu¨poi ‘you ask’

Austronesian languages

705

Senft, G. (1986). Kilivila, The Language of the Trobriand Islanders. Berlin: Mouton de Gruyter. Kisar; Meher; Loikera [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalyoPolynesian, Timor. Southwest Maluku, Kisar-Roma. Kisar Island, South Maluku (Indonesia). f

Stress falls on the penultimate syllable.

¨horo

‘cough’

ma¨weke ‘girl’

Christensen, J. & S. Christensen (1992). Kisar Phonology. In D.A. Barquest & W.D. Laidig (eds.) Phonological Studies in four languages of Maluku. Publications in Linguistics 108: 33–65. Dallas: SIL and University of Texas at Arlington. Kola; Warilau; Kulaha; Marlasi [P;U] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Aru. North Aru Islands, South Maluku (Indonesia). f f

Primary stress falls on the penultimate syllable. There are lexical exceptions with final stress.

ta¨lah

‘we sail’

¨‚anen ‘a bird of paradise’

¨talah

‘sit’

‚a¨nis

‘bat’

Takata, M. & Y. Takata (1992). Kola Phonology. NUSA Monographs 34: 31–46, Jakarta. Konjo; Kondjo; Konyo; Tiro; Pegunungan [P;A] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Makassar. South Sulawesi (Indonesia). f f

f

Primary stress falls on the penultimate syllable. Loans ending in /l, r, s/ get an extrametrical syllable V/, resulting in antepenultimate stress. Secondary stress falls on the initial syllable of long enough words.

¨galuN

‘wet rice field’

ha¨dZikaN ‘to be better [than]’

Friberg, T. & B. Friberg (1989). Notes on Konjo Phonology, Jakarta: NUSA. Friberg, T. (1995). Konjo. In D.T. Tryon (ed.) Comparative Austronesian Dictionary, Part 1, 563–572. Berlin/New York: Mouton de Guyter.

706

Austronesian languages

Kuanua; Tolai [S] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Meso Melanesian, New Ireland, South New Ireland-Solominic Patpatar-Tolai. East New Britain Province, Rabaul District, Gazelle Peninsula. f

Stress falls on the second syllable (and on the only one of monosyllables).

¨pap

‘dog’

ki¨ta ‘to hit’

di¨tal

‘they’

va¨lubana

‘to forget’

Franklin, K.J., H.B. Kerr & C.H. Beaumont (1974). Tolai Language Course. Asian Pacific, Series 7. SIL. Huntington Beach, California. Kulamanen (dialect of Manobo, Matigsalug) [P/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Southern Philippine, Manobo, Central, South, Ata-Tigwa. Mindanao (Philippines). f

Stress falls on the penult, unless its vowel is /e/ (most likely schwa) and the final vowel is full.

be¨nal

‘true’

ete¨bay ‘sister’

¨deleg

‘continue to do something’

mene¨nagat ‘fisherman’

Dubois, C. (1976). Sarangani Manobo: an introductory guide. Philippine Journal of Linguistics. Special monograph issue; No. 6. Meiklejohn, P. & K. Meiklejohn (1958). Accentuation in Sarangani Manobo. In: Studies in Philippino Linguistics. Oceanic Linguistic Monographs, No. 3, pp. 1–3. Kwaio [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Southeast Solomonic, Malaita-San Cristobal, Malaita, Northern. Solomon Islands. f

Stress falls on the penultimate syllable (diphtongs are disyllabic).

¨ma.e ‘be dead’

ma¨ena ‘his death’

Keesing, R.M. (1985). Kwaio Grammar. Pacific Linguistics, Series B, No. 88. Australian National University, Canberra.

Austronesian languages

707

Lamaholot; Solor; Solorese [P/P] Austronesian, Malayo-Polynesian, Central-Central-Eastern, Central Malayo-Polynesian, Timor, Flores-Lembata. Lesser Sunda Islands, Solor Island, other islands (Indonesia). f

f

Stress falls on the penult, unless its vowel is schwa and the final vowel is full. When su‰xes are added, stress remains in its original position.

d«¨ra ‘still’ ¨olaja ‘he works’

¨ola ‘to work’ b«lebo¨lebo ‘the flood’

Arndt, P. (1937). Grammatik der Solor-Sprache. Ende, Flores: Arnoldus. Lampung [U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Lampungic, Pesisir. Lampung Province, southern Sumatra (Indonesia). f

Stress falls on the final syllable.

an¨dZau ‘to visit’

di/andZau¨i ‘to be visited’

Walker, D.F. (1975). A Grammar of the Lampung Language: The Pesisir Dialect of Way Lima. NUSA Monographs 1. Jakarta: Nusa. Larike (dialect of Larike-Wakasihu) [P] Austronesian, Malayo-Polynesian, Central-Central-Eastern, Central Malayo-Polynesian, Central Maluku, East, Seram, Nunusako, Piru Bay, West Piru Bay – Hoamoal, East. Ambon Island, Maluku (Indonesia). f f

Primary stress falls on the penultimate syllable of the root. Secondary stress falls on alternate syllables left and right of the main stress.

pa¨lEdi ‘to sell’

¨sa/a

‘to ride’

˙aka¨sa/a˙ta ‘passenger’

Laidig, W.D. & C.J. Laidig (1991). Tarus Sou Rikedu ¼ Tata Bahasa Larike ¼ Larike Grammar, Ambon, Maluku: SIL and Pattimura University. Laidig, C.J. (1992). Segments, Syllables, and Stress in Larike. In D.A. Barquest & W.D. Laidig (eds.) Phonological Studies in four languages of Maluku. Publications in Linguistics 108: 67–126. Dallas: SIL and University of Texas at Arlington.

708

Austronesian languages

Lauje; Laudje; Tinombo; Ampibabo [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West-Central, Tomini. Central Sulawesi (Indonesia). f f

Stress falls on the penultimate syllable (long vowels are disyllabic). When su‰xes (except paragogic /e/) are added stress remains penultimate.

ba¨/inye ‘his/her head’

meNg¨u.uN ‘noise, sound’

Himmelmann, N.P. (1991). Tomini-Tolitoli Sound Structures. NUSA Monographs 33: 49–69, Jakarta: NUSA. Ledo; Palu; Paloesch (dialect of Kaili, Ledo) [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West-Central, Kaili Pomona, Kaili. Central and South Sulawesi (Indonesia). f

Stress falls on the penultimate syllable (long vowels are disyllabic).

banu¨ana ‘his/her house’

na¨da.a ‘bad’

Esser, S.J. (1934). Handleiding voor de beoefening van de Ledo-taal, Verhandelingen van het Bataviaasch Genootschap 72/1. Bandoeng: Nix & Co. Lenakel (dialect of Tanna) [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, South Vanuatu, Tanna. Tanna Island. f f

f

f

Primary stress falls on the penultimate syllable. In nouns, secondary stress falls on alternate syllables before the main stress. In verbs, secondary stress falls on the first and alternate syllables thereafter. No two stresses may be adjacent.

¨koley ‘sweet potato’ rI¨maugin ‘he ate’

kay˙Elaw¨Elaw ‘kind of dance’ ˙tIna˙gamyasI¨N«vIn ‘you will be copying it’

Lynch, J. (1974). Lenakel Phonology, Doctoral Dissertation. In: University of Hawaii Working Papers in Linguistics 7.

Austronesian languages

709

Lynch, J. (1978). A Grammar of Lenakel. Pacific Linguistics, Series B, No. 55. Australian National University, Canberra. Leti [U/P] Austronesian, Malayo-Polynesian, Central-Central-Eastern, Central Malayo-Polynesian, Timor, Southwest Maluku, Luang. Leti Island in Maluku (Indonesia). f

f f

Primary stress falls on the final syllable if it contains a long vowel (no examples). Otherwise primary stress is located on the penult. Secondary stress falls on remaining long vowels and alternate syllables to the left.

da¨Bna ‘potter wasp’ ˙pwrsal¨jra ‘seaside gate’ ˙rø¨nenu

‘they eat turtle’

Engelenhoven, P.G. van (1995). A description of the Leti Language. Ph.D. dissertation, Leiden University. Ma’anyan; Maanyak Dayak; Ma’anjan; Siang [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Borneo, Barito, East, Central-South, South. South around Tamianglayang (Kalimantan, Indonesia). f f

Stress falls on the penultimate syllable. When su‰xes are added stress moves to the penult of the new form, but not when the short form of pronominal enclitics is added.

¨anak ‘child’

¨anaknu ‘your child’

Sundermann, H. (1913). Der Dialek der Olon Maanjan (Dajak) in Su¨d-Ost Borneo. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 67: 203–267. The Hague: M. Nijho¤. Makassar [P;LEX] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Makassar. South of South Sulawesi (Indonesia). f f

Stress falls on the penultimate syllable. Vowel deletion rules in su‰xation process of non-neutral su‰xes may cause stress to be superficially final.

710 ¨bawa

Austronesian languages

‘mouth’

sas¨saN [sassa.aN] ‘laundry’

ran¨tasa/

‘messy’

ał¨jama/ [aN.jama.a/] ‘i work’

Manyambeang, A.K., A.K. Mulya & Nasruddin (1996). Tata Bahasa Makassar. Jakarta: Pusat Pembinaan dan Pengembangan Bahasa. Basri, H. (1999). Phonological and syntactic reflections of the morphological structure of Selayarese. Ph.D. Dissertation, SUNY, stony brook. Malagasy [P;U;A] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian Borneo, Barito, East, Malagasy. Madagascar. f f

Stress falls on the penultimate syllable. There are lexical exceptions with final or antepenultimate stress.

o¨me ‘to give’ mi¨hafy ‘to endure’

¨kely ‘small’ ¨tanana ‘hand’

¨soratra ‘writing’ ta¨nana ‘village’

Rajaonarimanana, N. (1994). Grammaire moderne de la langue malgache. Methode de Malgache, Vol. 1. Paris: L’asiathe`que. Malay [P/U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Malayic, Malayan, Local Malay. Malaysia. f f

Stress falls on the penultimate syllable if it contains a full vowel. If the penultimate vowel is schwa, stress is final.

s«n¨dar ‘to snore’ s«¨madi ‘concentration’

¨datang ‘to come, to arrive’ lak¨sana ‘quality’

Winstedt, R.O. (1927). Malay Grammar. Oxford: Oxford University Press. Lewis, M.B. (1947). Teach yourself Malay. London: English University Press. Mamasa [P] Austronesian, Malayo-Polynesian, Western Western Malayo-Polynesian, Sulawesi, South Sulawesi, Northern, Toraja-Sa’dan. South Sulawesi, Polmas District (Indonesia). f

Stress falls on the penultimate syllable, even when su‰xes are added.

Austronesian languages

711

kali¨bambaN ‘butterfly’ ¨dua i¨rusan

‘two’ ‘thing used for drinking (< iru/-an)’

Matti, D.F. (1991). Phonology of Mamasa. Workpapers in Indonesian Languages and Culture 12: 53–98. Ujung Pandang, Sulawesi: SIL. Manam [U%A/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, North New Guinea, Schouten, Kairiru-Manam, Manam. Manam Isl. Madang (New Guinea). tentative f Primary stress falls on the final syllable if it is closed. f Else it falls on the antepenult if closed, else on the penult. f Secondary stress falls on alternate syllables to the left (no examples). ta¨ma da

‘our incl. father’

ta¨nepwa

‘chief ’

¨embe/i

‘sacred flute’

¨moare

‘flower’

Lichtenberk, F. (1983). A Grammar of Manam. Honolulu: University of Hawaii Press. Chaski, C. (1985). Linear and Metrical Analysis of Manam Stress. Oceanic Linguistics 25, pp. 167–209. Manggarai [P/U;P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Bima-Sumba. Western third of Flores Island, Lesser Sundas (Indonesia). f

f

Stress falls on the penultimate syllable, unless it contains a schwa, then stress is final. Before voiceless (pre-nasalised) consonants stress may fall on penultimate schwas.

¨pitak

‘mud’

¨«mpo ‘grandchild’

«¨ma ‘father’

Burger, A. (1946). Voorlopige Manggaraise spraakkunst. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 103: 15–256. The Hague: M. Nijho¤. Verheijen, J.A.J. & C.E. Grimes (1995). Manggarai. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 585–592. Berlin/New York: Mouton de Guyter.

712

Austronesian languages

Maori [F/F] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East, Central, Tahitic. New Zealand. f

Stress falls on the first long vowel (no examples), else on the first syllable.

¨tangata ‘man’

¨papa

‘a slab’

Hohepa, P.W. (1967). A Profile Generative Grammar of Maori. In: International Journal of American Linguistics, Memoir 20. Ngata, A (n.d.) Maori Grammar and Conversation. Christchurch: Whitcombe & Tombs. Mentawai [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sumatra, Mentawai. Mentawai Islands, o¤ west coast of Sumatra (Indonesia). f

Stress falls on the penultimate syllable, even when su‰xes are added.

¨kulup ‘to sneak’

ai¨lampa ‘jambulanum domesticum’

Morris, M. (1900). Die Mentawai Sprache. Berlin. Mentu; Mentuh Tapuh (dialect of Dayak, Land) [U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Borneo, Land Dayak. Southern Sarawak (Malaysia). f

Stress falls on the final syllable.

da¨jukN ‘woman, female’

[bU¨r˛tn] ‘moon, month’

Court, C. (1972). The Phonological Structure of the Word in Mentu Land Dayak. Oceania Linguistic Monographs 15: 86–93. Minangkabau; Minang, Padang [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Malayic, Malay-Moklen, Malayan, Para-Malay. West Central Sumatra and throughout Indonesia. f

Stress falls on the penultimate syllable of the root.

¨pisaN ‘banana’

Austronesian languages

713

Adelaar, K.A. (1995). Minangkabau. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 433–442. Berlin/New York: Mouton de Gruyter. Toorn, J.L. van der (1899). Minangkabausche Spraakkunst. The Hague: M. Nijho¤. Mongondow; Bolaang Mongondow; Mongondou; Minahassa [P;P/U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Gorontalo, Mongondowic. Northeast Sulawesi (Indonesia). f f

f

Stress falls on the penultimate syllable when the final is open and light. When the final vowel is a diphthong speakers vary between final and penultimate stress. If the final diphthong is in a closed syllable, it always carries stress.

Dunnebier, W. (1929). Spraakkunst van het Bolaang Mongondowsch, part I. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 85: 297–468. The Hague: M. Nijho¤. Muna; Wuna; Mounan [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Muna-Buton. Muna Island o¤ coast Sulawesi, part of Buton Island, Ambon (Indonesia). f f

Primary stress falls on the penultimate syllable of the word. Secondary stress falls on alternate syllables to the left of the main stress.

¨ani ‘bee’

ne˙tula¨tula

‘he tells a story’

Berg, R. van de (1989). A grammar of the Muna language. Ph.D. Dissertation, Rijksuniversiteit Leiden (KITLV). Dordrecht: Foris. Nambas, Big [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, North and Central Vanuatu, Malekula Interior, Malekula Central. Malekula Island. f

f

Primary stress falls on the penultimate syllable (with some lexical exceptions). Secondary stress falls on alternate syllables to the left but not on the first.

i¨pali ‘he will burn it’ ita¨mamu ‘he was first’

¨lepu i˙putak¨mani

‘rat’ ‘he will spoil it’

714

Austronesian languages

Fox, G.J. (1979). Big Nambas Grammar. Pacific Linguistics, Series B, No. 60. Australian National University, Canberra. Napu; Pekurehua [P] Austronesian, Malayo-Plynesian, Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West-Central, Kaili-Pamona, Pamona. Central Sulawesi (Indonesia). f

Stress falls on the penultimate syllable of the word.

me¨sule ‘to return home’

¨manu ‘chicken’

ma¨nuna ‘his chicken’

Hanna, R. & L. Hanna (1991). Phonology of Napu. Workpapers in Indonesian Languages and Culture 12: 150–178. Ujung Pandang: SIL. Nengone [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Loyalty Islands. Mare´, Loyalty Islands. f f

Primary stress falls on the penultimate syllable. Secondary stress is located two syllables to the left of the main stress.

¨ke ‘but’

¨moma ‘old man’

a˙niti¨tini

‘to growl’

Tryon, D.T. & M.-J. Dubois (1969–71). Nengone Dictionary. Australian National University, Canberra. Tryon, D.T. (1967). Nengone Grammar. Linguistic Circle of Canberra Publications, Series B, No. 6. Canberra. Ngada; Nga’da; Ngadha; Nad’a; Bajava; Badjava [P/U] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Bima-Sumba. South central Flores (Indonesia). f

Stress falls on the penultimate syllable, unless it contains a schwa, then stress is final.

¨beli ‘witness’

v«¨li ‘buy’

Arndt, P. (1933). Grammatik der Ngad’a-Sprache (VBG 72/3). Bandoeng: Nix & Co. Djawanai, S. (1977). ‘‘A description of the basic phonology of Ngad’a and the treatment of borrowings’’. In NUSA Monographs 5: 10–18. Jakarta: NUSA.

Austronesian languages

715

Djawanai, S. & C.E. Grimes (1995). In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 593–600. Berlin/New York: Mouton de Guyter. Moore, P. (1980). Nga’da Phonology. In Naylor (ed.) Austronesian Studies: Papers from the second Eastern Conference on Austronesian Languages: 203–212. Michigan: Ann Arbor. Nias; Batu [P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sumatra, Northern. Nias and Batu Islands o¤ the west coast of Sumatra (Indonesia). f f

Stress falls on the penultimate syllable. There are morphological exceptions with final stress.

¨dela

‘bridge’

¨gowi

i¨dano) ¨bongi

‘water, stream’ ‘night, it is night’

ha¨lo)w)o) ‘work’ bo¨ngi ‘tonight’

‘potato’

Sundermann, H. (1913). Niassische Sprachlehre. The Hague: M. Nijho¤. Nuaulu; Patakai; Fatakai [P] Austronesian, Malayo-Polynesian, Central-Central-Eastern, Central Malayo-Polynesian, Central Maluku, East, Seram, Nunusaku, PatakaManusela. Seram Island, Central Maluku (Indonesia). f

f

Primary stress on the penult of the root (with some antepenultimate exceptions). Secondary stress falls on the first syllable of four syllable roots.

ka¨ranate ‘greater’

˙kara¨tupa ‘chili’

Bolton, R.A. (1989). Nuaulu Phonology. Workpapers in Indonesian Languages and Culture 7: 89–119. Ambon: Pattimura University þ SIL. Bolton, R.A. (1990). A Preliminary description of Nuaulu phonology and grammar. Ph.D. Thesis, University of Texas at Arlington. Oloh Mangtangai (dialect of Ngaju; Ngaju Dayak) [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Borneo, Barito, West, South. Trade language for most of Kalimantan, Borneo (Indonesia). f

Stress falls on the penultimate syllable of the word.

716

Austronesian languages

¨karas ‘strong’

manga¨rasan ‘strengthen’

Muller, F. (1858). Versuch einer Grammatik der Dajakschen Sprache. Amsterdam: C.A. Spin & son. Paama; Paamese [A (IRR)] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, North and Central Vanuatu, Notheast Vanuatu-Banks Islands, East Vanuatu. Paama Isl. (Vanuatu). f

Stress falls on the antpenultimate syllable (with some lexical exceptions).

vi¨sokono ‘morning’ ¨natahosi ‘he is good’

¨mesay ‘he is sick’ to¨hulue ‘bush vegetable type’

Crowley, T. (1982). The Paamese Language of Vanuatu. Pacific Linguistics, Series B, No. 87. Australian National University, Canberra. Padoe; South Mori; Padoe´; Alalao [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West Central, Bungku-Mori, Mori. South Sulawesi, Eastern Luwu District (Indonesia). f

Stress falls on the penultimate syllable of the word.

¨raha ‘house’

ra¨haNgu ‘my house’

raha¨mi.u

‘your house’

Karhunen, M. (1991). Phonology of Padoe. Workpapers in Indonesian Languages and Culture 12: 179–196. Ujung Pandang: SIL. Paiwan [P] Austronesian, Formosan, Paiwanic. Taiwan. Southern, southeastern mountains. f

Stress falls on the penultimate syllable of the phrase.

¨tsiqaw ‘fish’ tsa.u¨tsa.u ‘person’

si¨ka.u ‘net bag’ va.ika¨ngaken ‘I am really leaving’

Ferrell, R. (1982). Paiwan dictionary. Pacific linguistics, Series C, No. 73. Australian National University, Canberra. Pamona; Bare’e; Baree; Poso; Taa; Wana [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi,

Austronesian languages

717

Central Sulawesi, West Central, Kaili-Pamona. Central and South Sulawesi (Indonesia). f

Stress falls on the penultimate syllable of the root.

¨pela ‘skin’

¨kobati from ¨kobat ‘cradle’

Adriani, N. (1931). Spraakkunst der Bare’e taal, Verhandelingen van het Bataviaasch Genootschap 70. Bandoeng: Nix & Co. Paulohi [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Central Maluku, East, Seram, Nunusakum, Piru Bay, East, Seram Straits, Solehua. Central Maluku, West Seram (Indonesia). f

Stress falls on the penultimate syllable of the word (su‰xes count).

¨nife

‘tooth’

nife¨niru ‘their teeth’

Stresemann, E. (1918). Die Paulohi Sprache. The Hague: M. Nijho¤ (KITLV). Pulopetak (dialect of Ngaju) [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Borneo, Barito, West, South. Kapuas and other rivers, south. Borneo (Indonesia þ Malaysia). f f

Primary stress falls on the penultimate syllable of the word. Except when followed by 3sg. possessive or place/time marker, then stress remains on the penult of the root.

¨anak ‘child’

¨anak«h ‘his child’

Mihing, T.W.J. & W.A.L. Stokhof (1977). On the Ngaju Dayak Sound System (Pulau Petak dialect). NUSA Monographs 4: 49–59. Jakarta: NUSA. Puluwatese; Puluwat [L/L] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Micronesian, Micronesian Proper, Ponapeic-Trukic, Trukic. Trukic area of Micronesia. tentative f Stress falls on the last syllable beginning with /h/, else on the last.

718 ki¨yo@

Austronesian languages

pa¨halo

‘outrigger boom’

yi¨we ‘then’

‘to drift away’

yapwa¨hano@ ‘to dry out’

Elbert, S.H. (1974). Puluwat Grammar. Pacific Linguistics, Series B, No. 29. Australian National University, Canberra. Rantepao; Kesu’ (dialect of Toraja Sa’dan) [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Northern, Toraja-Sa’dan. South Sulawesi, Tana Toraja District (Indonesia). f f

Stress falls on the penultimate syllable of the root. Some su‰xes cause stress to shift to the right.

¨kande kan¨deku

‘to eat, food’ ‘my food’

ku¨kandemo ‘I have eaten’ Sande, J.S. & W.A.L. Stokhof (1977). On the Phonology of the Toraja Kesu/dialect. NUSA Monographs 5: 19–34. Jakarta: NUSA. Rapanui; Easter Island; Pascuense [P;U] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East, Rapanui. Easter Island, Chile, Tahiti, USA. f

Stress falls on the penultimate syllable (lexical exceptions with final stress).

¨kiko

‘meat’

¨Nao

¨po

‘night’

ma¨ta ‘axe head’

‘neck’

Du Peu, V. (1996). Rapanui. London: Routledge. Ratahan; Bentenan; Pasan [P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Sangiric. North Sulawesi (Indonesia). f

Stress falls on the penultimate syllable (lexical exceptions with final stress).

mun¨dupa ‘to put’ mundu¨pa ‘to measure a span’ ¨baley ‘house’

Austronesian languages

719

Sneddon, J.N. (1984). Proto Sangiric and the Sangiric Languages. Pacific Linguistics, Series B, No. 91. Australian National University, Canberra. Roro; Waima [I/S] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Papuan Tip, Peripheral, Central Papuan, West Central Papuan, Nuclear. Central Province Bereina District, near Kairuku, Hall Sound (PNG). f f

f

Stress falls on the first syllable of disyllabic words. In longer words stress falls on the first syllable if it has a long vowel, else on the second. Stress on third syllable of longer words reported.

Strong, W.M. (1913–14). The Rora and Mekeo Languages of British New Guinea. Zeitschrift fu¨r kolonialsprachen 4, pp. 286. Rote; Rotti; Rotinese; Roti [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Timor-Flores, Timor, unclassified. Roti Island and adjacent islands (Indonesia). f

f

Primary stress on the penultimate syllable (long vowels and diphthongs are disyllabic). Secondary stress two syllables to the left of the main stress.

¨tutu

‘pound’

tu¨tuuk ‘seat’

˙tasi¨oe ‘sea water’

Jonker, J.C.G. (1915). Rottineesche spraakkunst. Leiden: E.J. Brill. Fox, J.J. & Grimes C.E. (1995). Roti. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 611–622. Berlin/New York: Mouton de Guyter. Rotuman [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, West Fijian-Rotuman, Rotuman. Rotuma (Fiji). f

Stress falls on the final vowel if it is long, else on the penult.

¨taka ‘lie down’

kara¨raa ‘snore’

hunu¨nuka ‘gasp for breath’

Churchward, C.M. (1940). Rotuman Grammar and Dictionary. Australian Medical Publishing Co.

720

Austronesian languages

Sakao, Northern (dialect of Sakao) [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, North and Central Vanuatu, East Santo, North. Santo Island and small adjacent islands (New Hebrides). f

f

f

Primary stress described to occur phrase final (on the pretonic or final syllable). Secondary stress (which could be word stress) on the final syllable if it has a vowel nucleus. Else on the penult.

a¨mas ‘club’

yapi¨lyD

‘they quarrel’

ne¨ssar ‘broken’

Guy, J.B.M. (1974). A Grammar of the Northern Dialect of Sakao. Pacific Linguistics, Series B, No. 33. Australian National University, Canberra. Sama Baangingi, Bajan [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, SamaBajaw, Sulu-Borneo, Inner Sulu Sama. Sulu Archipelago (Philippines). f

Stress falls on the penultimate syllable of the word.

kol¨lo/in ‘remove it’

/a¨sannaN ‘peaceful’

Gault, J.A. (1979). Phonemics and Morphophonemics of Sama Baangingi. Pacific Linguistics, Series A, No. 50, 49–67. Australian National University, Canberra. Sangir; Sangihe´; Sangirese; Sangi [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Sangiric. North Sulawesi, Great Sangir Island, and north Maluku (Indonesia). f f

Primary stress falls on the penultimate syllable (with lexical exceptions). The paragogic syllable /«/ is extrametrical, causing apparent antepenultimate stress.

¨bale ‘house’ ¨kapu ‘to rub’

ba¨leku ‘my house’ ka¨pu ‘desirous’

Adriani, N. (1893). Sangireesche Spraakkunst. Leiden: A.H. Adriani. Sneddon, J.N. (1983). Proto-Sangiric and the Sangiric Languages, Pacific Linguistics, Series B, No. 91. Australian National University, Canberra.

Austronesian languages

721

Sawai; Weda; Were [P/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, South Halmahera-West New Guinea, South HalmaheraGeelvink Bay, South Halmahera, Central East. South Halmahera, North Maluku (Indonesia). f

Stress falls on the penultimate syllable, unless its vowel is /E/ and the final vowel is not /E/.

bo¨bane ‘bay’

¨jEgEt ‘oil’

musE¨la ‘woven mat’

Whisler, R. (1992). Phonology of Sawai. In D.A. Barquest & W.D. Laidig (eds.) Phonological Studies in four languages of Maluku. Publications in Linguistics 108: 7–32. Dallas: SIL and University of Texas at Arlington. Whisler, R. & J. Whisler (1995). Sawai. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 659–665. Berlin/New York: Mouton de Guyter. Selaru; Salaru [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Southeast Maluku. Tanimbar, Selaru Island, Yamdena Island (Moluccas, Indonesia). f

f f

Primary stress falls on the final syllable if it is closed by a consonant and semivowel. Otherwise stress is penultimate. Secondary stress on the first syllable (sometimes this becomes the primary stress).

/a¨mana ‘octopus’

¨/owa˙nare ‘clouds’

hEt¨tilj ‘corn’

Coward, N.E. (1989). A phonological sketch of the Selaru language. Workpapers in Indonesian Languages and Culture 7: 1–38. Ambon: Pattimura University þ SIL. Drabbe, P. (1932). Beknopte Spraakkunst en Korte Woordenlijst der Slaroe‘esche Taal, Verhandelingen van het Bativiaasch Genootschap 71/4. Bandoeng: Nix & Co. Selayar; Salayar; Selayarese [P;LEX] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, South Sulawesi, Makassar. South Sulawesi, Selayar Island (Indonesia).

722 f f

Austronesian languages

Stress falls on the penultimate vowel. There are lexical exceptions with antepenultimate and pre-antepenultimate stress.

¨pao

‘mango’

¨allo ‘day’

pa¨oku

‘my mango’

¨katala

‘itch’

pappal¨luaN ‘place for cooking’

Mithun, M. & H. Basri (1986). The Phonology of Selayarese. Oceanic Linguistics 25, 1–2: 210–254. Sikule; Sichule [U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sumatra, Northern. Central Simeulue Island, Sumatra (Indonesia). f

Stress falls on the final syllable, even final syllabic possessive su‰x -m.

ba¨xa ‘the will’

baxa¨m

‘your will’

Ka¨hler, H. (1955). Die Sichule-Sprache auf der Insel Simalur an der West¨ bersee (Beiheft 27). Berlin: Reimer. ku¨ste von Sumatra, Afrika under U Sio [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, North New Guinea, Ngero-Vitiaz, Vitiaz, Sio. Morobe Province, Wasu District, mainland near Sio Island (PNG). f

Stress falls on final diphthongs, else on the penultimate syllable.

¨dewa

‘yam species’

pataRa¨wNa ‘sacrifice’

ka¨pula ‘egg’ Na¨lai

‘large’

Soetenga Clark, D. (1993). The phonology of the Sio language. In J. Clifton (ed). Phonologies of Austronesian Languages 2. 25–70. Ukarumpa: SIL. Sunda; Sundanese; Priangan [P/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sundanese. Western third of Java Island (Indonesia). f

Stress falls on the penultimate syllable, unless it contains a schwa and the final does not.

¨isuk ‘morning’ g«¨de ‘big’

i¨sukan ‘tomorrow’ Nab«¨dil ‘shoot’

Austronesian languages

723

Clynes, A. (1995). Sundanese. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 459–468. Berlin/New York: Mouton de Guyter. Taba; East Makian [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, South Halmahera-West New Guinea, South Halmahera, East Makian-Gane. Islands of Makian, Mori, Kayoa, Halmahera, Bacan and Obi, north Maluku. f f

Primary stress falls on the penultimate syllable. Secondary stress falls on the pre-antepenultimate syllable of long enough words.

¨pojo ‘head’ ˙manu¨si.a ‘people’

ka¨mudu ka˙kumpap¨pido

‘be thick’ ‘large woven rice basket’

Bowden, J. (1997). Taba (Makian Dalam): Description of an Austronesian Language of Eastern Indonesia. University of Melbourne Ph.D. Thesis. Tagalog [P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Meso Philippine, Central Philippine, Tagalog. Greater Manila (Philippines). f

Stress falls on the penultimate syllable (lexical exceptions have final stress).

da¨mit ‘dress’ ¨mura ‘cheap’

magan¨da ‘beautiful’ ma¨husay ‘good’

Bloomfield, L. (1917). Tagalog texts with grammatical analysis. University of Illinois Studies in Language and Literature III, 3, Urbana, Illinois. Llamzon, T.A. (1966). Tagalog Phonology. Anthropological Linguistics. 8: 1, pp. 30–39. Schachter, P. & F.T. Otanes. (1972). Tagalog Reference Grammar. Berkeley: University of California Press. Tahitian [F/L] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East, Central, Tahitic. Tahiti, Society Islands.

724 f

Austronesian languages

Primary stress falls on the first long vowel, else on the penultimate syllable.

(seen as first heavy or last with final syllable extrametricality). /o¨hipa

‘work’

ta¨maahine ‘daughter’

¨fare

‘house’

ta¨maaroa ‘boy’

Tryon, D. (1976). Conversational Tahitian: An introduction to the Tahitian Language of French Polynesia. Canberra: ANU press. Vergnaud, J.R. & M. Halle (1978). Metrical Structures in Phonology. Unpublished ms. MIT, Cambridge Massachusetts. Talaud; Talaut; Talodda [P;A] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Sangiric. North Sulawesi (Indonesia). f f

Primary stress falls on the penultimate syllable. When the last vowel is paragogic /a/, stress is antepenultimate.

Sneddon, J.N. (1983). Proto Sangiric and the Sangiric Languages. Pacific Linguistics, Series B, No. 91. Australian National University, Canberra. Talens, J.P. (1911). Ee´n en ander over het Talaoetsch, medegedeeld door Dr. N. Adriani. Verhandelingen van het Bataviaasch Genootschap 59/1. The Hague: M. Nijho¤/Batavia: Albrecht & Co. Tanna, Southwest (dialect of Tanna) [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, South Vanuatu, Tanna. Tanna Island in Vanuatu (New Hebrides). f f f

f

Primary stress falls on the final syllable if it contains a long vowel. Else stress the penult. In nouns, secondary stress falls on alternate syllables before the main stress. In verbs, secondary stress falls on the first and alternate syllables thereafter.

¨ilu ‘to be deaf ’ nç¨kleen ‘his chest’

nç¨pwanak ˙lçmne˙liuok¨pihiu

‘my forehead’ ‘he walked north’

Lynch, J.D. (1982). Southwest Tanna Grammar and Vocabulary. In J. Lynch (ed.) Papers in the Linguistics of Melanesia 4, 1–91. Australian National University, Canberra.

Austronesian languages

725

Tarangan, West; Trangan, West [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Southeast Maluku, Kei-Aru, Aru, South. Southern Aru Islands, South Maluku (Indonesia). f

f

Stress falls on the final syllable of the root if it is closed and has a mid vowel. Else stress the penult.

¨gw« ‘blossom’

gu¨rep

‘nail’

˙kab«¨dar«w ‘bird species’

Nivens, R. (1992). ‘‘A Lexical Phonology of West Tarangan’’, In: D.A. Barquest & W.D. Laidig (eds.) Phonological Studies in four languages of Maluku. Publications in Linguistics 108: 127–227. Dallas: SIL and University of Texas at Arlington. Tawala [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Papuan Tip, Nuclear, North Papuan Mainland-D’Entrecasteaux, Are-Taupota, Taupota. Milne Bay Province, East Cape (Papua New Guinea). f f

Primary stress falls on the penultimate syllable. Secondary stress falls on alternate syllables to the left of the main stress.

¨bada ‘man’

hi¨puli ‘earth’

˙kedu¨luma ‘woman’

Ezard, B. (1991). Tawala Grammar: A functional approach. Ph.D. Dissertation. Australian National University, Canberra. Tigak [I] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Western Oceanic, Meso Melanesian, New Ireland, Lavongai-Nalik. New Ireland province (Papua New Guinea). f f

Primary stress falls on the first syllable, but not on a prefix. Words with prefixes have secondary stress.

¨etok ‘talk’

¨bogo ‘pig’

pa¨loNani ‘hear it’

Beaumont, C.H. (1979). The Tigak Language of New Ireland. Pacific Linguistics, Series B, No. 58. Australian National University, Canberra. Timor; Atoni; Timorese; Timol; Dawan; Ambenu; Vaikenu [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central Malayo-

726

Austronesian languages

Polynesian, Timor-Flores, Timor, unclassified. Western Timor Island (Indonesia). f

Stress falls on the penultimate syllable of the word.

¨pEna ‘corn’

¨lEko ‘good’

¨anah ‘child’

Middelkoop, P. (1950). Proeve van een Timorese Grammatica. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 106: 375–514. The Hague: M. Nijho¤. Tiruray [A;P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, South Mindanao, Tiruray. Upi, Cotabato, Mindanao. Philippines. f f

Stress falls on the penultimate or antepenultimate syllable. When both the penultimate and antepenultimate vowel are shortened, stress is final.

Schlegel, S.A. (1971). Tiruray-English Lexicon. Berkeley: University of California Press. Toba-Batak [P;U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sundic, Sumatra, Batak, Southern. Sumatra (Indonesia). f

Stress falls on the penultimate syllable (lexical exceptions have final stress).

mar la¨tak ‘to snap’

la¨bu.an ‘harbour’

Adelaar, K.A. (1995). Toba Batak. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1: 421–432. Berlin/New York: Mouton de Gruyter. Nababan, P.W.J. (1981). A grammar of Toba Batak, Pacific Linguistics, Series D, No. 37. Australian National University, Canberra. Percival, W.K. (1981). A grammar of the urbanized Toba Batak of Medan, Pacific Linguistics, Series B, No. 76. Australian National University, Canberra. Tuuk, H.N., van der (1864, 1867). Tobasche spraakkunst (2 parts). Amsterdam: Muller. Tondano [U/P%A/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Sangiric. North Sulawesi (Indonesia).

Austronesian languages

727

tentative f Stress falls on the final syllable if it contains a long vowel. f In other cases stress is penultimate, unless its vowel is schwa. f In that case stress may shift to the final or the antepenult under some conditions. ¨wale ‘house’

kar i¨maNka ‘spider’

m«Naa¨Naan ‘is continually eating’

s«¨r a/ ‘fish’

¨t«mpok

¨kaan«na

‘tip’

‘will be eaten by him’

Sneddon, J.N. (1975). Tondano Phonology and Grammar. Pacific Linguistics, Series B, No. 38. Australian National University, Canberra. Tongan [U/P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polyn., Polynesian, Tongic. Tonga Islands. f

f

Primary stress falls on the final syllable if it has a long vowel, else on the penult. Secondary stress falls on remaining long vowels and alternate syllables to the left.

¨tuki

‘strike’

˙fee¨fee

˙kaa¨taki

‘sorry’

˙maa˙loo¨loo ‘rest’

‘how’

Chrurchward, C.M. (1953). Tongan Grammar. London: Oxford University Press. Feldman, H. (1978). Some notes on Tongan Phonology. Oceanic Linguistics 17, 133–139. Tontemboan; Tompakewa; Tountemboan; Pakewa [U-P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Sangir-Minahasan, Minahasan. North Sulawesi (Indonesia). f

Stress falls on the penultimate syllable, except when the penultimate vowel is schwa, and the final syllable is closed and contains /a, o, e/, then stress is final.

¨sisil ‘story’

sisi¨sil«n ‘tale’

wǬka/

‘stick, sta¤ ’

Adriani, N. & M.L. Adriani (1908). Hoofdstukken uit de Spraakkunst van het Tontemboaansch. The Hague: M. Nijho¤ (KITLV).

728

Austronesian languages

Tuamotuan; Pa’umotuan [P] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East, Central, Tahitic. Tuamotuan Archipelago, Tahiti. f f

Primary stress falls on the penultimate syllable. Longer words have two primary stresses; two prosodic words.

¨gutu

‘lips’

ta¨gata ‘man’

¨fara¨oa ‘flour’

Kuki, H. (1970). Tuamotuan Phonology. Pacific Linguistics, Series B, No. 17. Australian National University, Canberra. Tukang Besi [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Muna-Buton, Tukangbesi-Bonerate. Islands o¤ the south(-east) coast of Sulawesi (Indonesia). f f

Primary stress falls on the penultimate syllable. Secondary stress falls on alternate syllables to the left of the main stress.

¨BandE ku˙pamo˙ro/u¨kemo

‘rain’ ku˙Bali¨ako ‘i’ll go home’ ‘i made him\her drink’

Donohue, M. (1999). A Grammar of Tukang Besi. Mouton Grammar Library 20. Berlin, New York: Mouton de Gruyter. Ulithian [IRR] Austronesian, Malayo-Polynesian, Central-Eastern, Eastern MalayoPolynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Micronesian, Micronesian Proper, Ponapeic-Trukic, Trukic. Ulithi, Sonsoral and Woleai Islands. f

Stress is non-phonemic and accompanies vowels preceding a long consonant.

Sohn, H.M. & B.W. Bender (1973). A Ulithian Grammar. Pacific Linguistics, Series C, No. 27. Australian National University, Canberra. Uma; Pipikoro; Uma Aria; Oema [P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Central Sulawesi, West Central, Kaili-Pamona, Kaili. Central Sulawesi, Kulawi Subdistrict (Indonesia).

Austronesian languages f

729

Stress falls on the penultimate syllable (diphthongs are disyllabic unless the second vowel is higher than the first).

¨uda ‘rain’

ke¨ni.a ‘baggage’

¨paiBa ‘feeding trough’

Esser, S.J. (1964). De Uma-taal (West Midden-Celebes): spraakkunstige schets en teksten, uitgegeven en van een woordenlijst voorzien door Dr. J. Noorduyn. Verhandelingen van het Koninklijk Instituut voor Taal, Land, en Volkenkunde 43. The Hague: M. Nijho¤. Martens, M.P. (1988). Phonology of Uma. In H. Steinhauer (ed.) Papers in Western Austronesian Linguistics 4, Pacific Linguistics, Series A, No. 79. Canberra: Australian National University. Martens, M.P. (1995). ‘‘Uma’’. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1. Berlin & New York: Mouton de Gruyter. Wetan; Wetang (dialect of Luang) [P;IRR] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Timor, Southwest Maluku, Luang. Northwest Babar Islands east of Timor, South Maluku (Indonesia). f

Stress falls on the penultimate syllable (but it is weak and there are many exceptions).

¨kerna

‘dry land, dry’

Jong, J.B.P. de (1987). Wetan Fieldnotes. Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 130. Dordrecht: Foris. Wolio; Baubau; Buton; Butung; Butonese; Boetonees [U/P] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Sulawesi, Muna-Buton, Buton. Southwestern part of Buton Island, southeast Sulawesi (Indonesia). f

Stress falls on the penultimate vowel.

¨taNi ‘cry’ ma¨waø ‘red hot’

taNi¨aka

‘cry about something’

Anceaux, J.C. (1952). The Wolio language: outline of grammatical description and texts, Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 11. The Hague: M. Nijho¤. Anceaux, J.C. & C.E. Grimes (1995). Wolio. In D.T. Tryon (ed.) Comparative Austronesian Dictionary Part 1, 573–584. Berlin/New York: Mouton de Guyter.

730

Austronesian languages

Yamdena; Jamdena; Jamden [P] Austronesian, Malayo-Polynesian, Central-Eastern, Central MalayoPolynesian, Southeast Maluku, Kei-Aru. Southeast Maluku, Selaru Island, Yamdena Island (Indonesia). f

Stress falls on the penultimate syllable of the root.

¨lete ‘garden’

¨letar

‘gardens’ ¨lendan ‘hill’

Drabbe, P. (1926). Spraakleer der Jamdeensche Taal, Verhandelingen van het Bataviaasch Genootschap 67/2. The Hague: M. Nijho¤/Batavia: Albrecht & Co. Mettler, T. & H. Mettler (1990). Yamdena Phonology. Workpapers in Indonesian Languages and Culture 8: 29–79. Ambon: Pattimura University þ SIL. Yapese [U/U] Austronesian, Malayo-Polynesian, Western Malayo-Polynesian, Yapese. Yap Island. f

f

f f

Primary stress on the penultimate syllable if it is long and the final is short. Primary stress on the penultimate syllable if the word ends in an open long vowel. Else stress is final. Long vowels and closed syllables get secondary stress.

¨saalap ‘expert’ ˙mag¨paa/ ‘wedding’

˙moe¨roes ˙marun˙gaq¨gean

‘savage’ ‘about it’

Jensen, J.T. (1977). Yapese Reference Grammar. Honolulu: University Press of Hawaii.

Papua New Guinea and Irian Jaya languages

East Bird’s Head Meah; Mansibaber [LEX] East Bird’s Head, Meax. East Bird’s Head, north coast (Irian Jaya). f

Stress location is lexically determined.

¨EREsa ‘go visit’ ¨mohwEkEni ‘bride cloth’

ERE¨sa ‘child’ ohoto¨ru ‘gather’

Gravelle, G. & G. Gravelle (1991). Meah Phonology. Workpapers in Indonesian Languages and Culture 10: 1–36. Jayapura, Irian Jaya: Cendera-wasih University & SIL. East Papuan Lavukaleve [I;LEX] East Papuan, Yele-Solomons-New Britain, Yele-Solomons, central Solomons. Russell Islands, northwest of Guadalcanal, central Solomons. f

f

Most words have primary stress on their initial syllable (there are lexical exceptions). Secondary stress on every second syllable in both directions from the primary stress.

¨tata

‘spider’ ¨urio

˙mita¨keu ‘dog’

¨mano˙giri˙giri

‘crab’ ‘seagull sp’

Terrill, A. (1999). A grammar of Lavukaleve: a Papuan language of the Solomon Islands. Ph.D. Thesis, Australian National University. Yele; Yeletnye [I] East Papuan, Yele-Solomons-New Britain, Yele-Solomons, Yele. Rossel Island, Papua New Guinea. f f

Primary stress falls on the first syllable. Secondary stress falls on the third syllable (not in trisyllables if the second is more open than the first or the word is vowel initial).

¨yi ‘leg’ ¨mbala ‘breadfruit

¨kEbe ‘paddle’ ¨tobu˙gada ‘crocodile’

732

Papua New Guinea and Irian Jaya languages

Henderson, J.E. (1975). Yeletnye, the language of Rossel Island. In T.E. Dutton, (ed.), Studies in the Languages of Central and South East Papua, Pacific Linguistics, Series C, No. 19, pp. 817–833. Australian National University, Canberra. Henderson, J.E. (1995). Phonology and grammar of Yele, Papua New Guinea. Pacific linguistics, Series B, No. 112. Australian National University, Canberra. Geelvink Bay Doutai; Tolitai; Taori-So [I (Tone)] Geelvink Bay, Lakes Plain, Tariku, East. Lakes Plain Area (Irian Jaya). f

Primary stress falls on the first syllable.

¨a@d"@ ‘rope’ ¨sa$d" $ ‘flower’

¨do$ba@ ‘spider’ ¨sa@d" $ ‘good’

McAllister, L. & K. McAllister (1991). The Process of Phonological Change in Doutai. Workpapers in Indonesian Languages and Culture 9: 121–142. Jayapura, Irian Jaya: Cendera-wasih University & SIL. Obokuitai; Obogwitai; Baburiwa [L/L (Tone)] Geelvink Bay, Northern, Lakes Plain, Tariku, East. Lakes plain north of mid Rou¤aer River (Irian Jaya). f

Stress falls on the last high tone, or on the last syllable if there are no high tones.

¨hE^t" ‘same’

k"@¨b"@ ‘leech’

a¨R"@

‘father’s sister’

Scott, D.S. & P.B. Jenison (1991). Obokuitai Phonology. Workpapers in Indonesian Languages and Cultures 9: 69–90. Jayapura, Irian Jaya: Cendera-wasih University & SIL. Sikaritai; Aikwakai [L/F (Tone)] Geelvink Bay, Lakes Plain, Tariku, East. Lakes Plain Area (Irian Jaya). f

Stress falls on the last high tone, or on the first syllable if there are no high tones.

sE@bE@¨k"@ ‘narrow’ hu@¨Ra@RE$ ‘male’ ¨a&‚a&rE$ ‘handle’ Martin, D.L. (1991). Sikaritai Phonology. Workpapers in Indonesian Languages and Culture 9: 91–120. Jayapura, Irian Jaya: Cendera-wasih University & SIL.

Papua New Guinea and Irian Jaya languages

733

West Papuan Galela [P;A] West Papuan, North Halmahera, North, Galela-Loloda. North Maluku, Galela Bay and north of Tobelo (Indonesia). f f

Primary stress falls on the penultimate syllable. Development towards antepenultimate stress in words with paragogic vowels (and loans).

¨tuma bobara¨mahi

‘egret’ ‘kind of fish’

go¨tola ‘cockatoo’ ¨pugili ‘thigh’

Baarda, M.J. van (1891). Beknopte spraakkunst van de Galillareesche taal. Utrecht: Kemink. Baarda, M.J. van (1908). Leiddraad bij het bestudeeren van ’t Galela’sch dialect op het eiland Halmaheira. The Hague: M. Nijho¤ (KITLV). Shelden, H. (1990). Galela phonemes and stress. Workpapers in Indonesian Languages and Culture 8: 77–88. Ambon: Pattimura University þ SIL. Mai Brat; Maybrat [I] West Papuan, Bird’s Head, North-Central Bird’s Head, Central Bird’s Head. Central Bird’s Head around Ayamaru Lakes, Papua New Guinea. f

f

Primary stress most often falls on the initial syllable (many lexical exceptions). Secondary stress falls on the last syllable of trisyllabic words.

¨amAx ‘house’

¨rapu˙wx ‘forest’

Dol, P.H. (1999). A Grammar of Maybrat. A language of the Bird’s Head, Irian Jaya, Indonesia. Ph.D. Dissertation, Leiden University. Pa’disua; Palisua (dialect of Sahu) [A;IRR] West Papuan, North Halmahera, North, sahu. North Maluku, soutwestern north Halmahera Island (Indonesia). f

Commonly stress falls on the antepenultimate syllable, but there are many exceptions.

¨igono

‘coconut’

be¨le/asa ‘shoulder’

Visser, L.E. & C.L. Voorhoeve (1987). Sahu-Indonesian-English Dictionary and Sahu Grammar Sketch, Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 126. Dordrecht: Foris.

734

Papua New Guinea and Irian Jaya languages

Pagu; Pago; Pagoe [P] West Papuan, North Halmahera, North, Kao River. North Maluku, interior Halmahera. f

Stress falls on the penultimate syllable underlyingly, which is often antepenult overtly because of the addition of echo vowels in words with final consonants.

¨osam ‘to enter’ ¨oyomo ‘to eat’

¨lepe ‘many’ ¨sasauku ‘hot’

Wimbish, S.G. (1992). Pagu Phonology. NUSA Monographs 34: 69–89. Jakarta: NUSA. Tabaru; Tobaru [P] West Papuan, North Halmahera, North, Tobaru. North Maluku, Ibu, Jailolo and Oba districts (Indonesia). f

Stress falls on the penultimate syllable underlyingly, which is often antepenult overtly because of the addition of echo vowels in words with final consonants.

¨gakaka ‘palate’

¨toteke ‘to knock’

Fortgens, J. (1928). Grammatikale aantekeningen van het Tabaroesch, Tabaroesche volksverhalen en raadsels. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 84: 300–544. The Hague: M. Nijho¤. Tobelo [U/P] West Papuan, North Halmahera, North, Tobelo. North Maluku, Raja Ampat Islands of Irian Jaya (Indonesia). f

f

Stress falls on the final syllable if it contains a diphthong, else on the penult. Words that have an echo vowel at the end have apparent antepenultimate stress.

¨biki ‘tail’ ba¨kau [no translation provided] da¨maø [no translation provided] ¨wtiti ‘navel’ Hueting, A. (1908). Tobe`loreesch-Hollandsch Woordenboek met Hollandsch-Tobe`loreesche Inhoudsopgave. The Hague: M. Nijho¤ (KITLV).

Papua New Guinea and Irian Jaya languages

735

Hueting, A. (1936). Iets over de spraakkunst van de Tobeloreesche taal. Bijdragen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 94: 295–325. The Hague: M. Nijho¤. Sepik Ramu Alamblak [P/A%L/F] Sepik-Ramu, Sepik, Sepik Hill, Alamblak. Middle Karawari and Wagupmeri rivers. Papua New Guinea. f

f

Stress falls on the penultimate syllable if it is not /ç/, else on the antepenult if it is not /ç/. If both are /ç/, stress a heavy su‰x if present, else stress the last non-/ç/ nucleus in the word. If the word has only /ç/ nuclei, stress the first.

bi¨dan çm ‘shell type .pl’ xe ¨m«tçbçn ‘she gave to you’ kçpç¨mEt ‘it’s a sago carrying basket’ ¨kçpçmçt ‘sago carrying basket’ Bruce, L. (1984). The Alamblak Language of Papua New Guinea (East Sepik). Pacific Linguistics, Series C, No. 81. Australian National University, Canberra. Awtuw; Autu [P;I] Sepik-Ramu, Sepik, Ram. Northwestern Papua New Guinea. f f

Primary stress falls on the penultimate or the first syllable. In longer words, primary stress is usually penult and secondary stress initial.

¨kinik ‘sit’ wiy¨tape ‘river’

¨waruke ˙orkewey¨naywo

‘big’ ‘four’

Feldman, H. (1986). A Grammar of Awtuw. Pacific Linguistics, Series B, No. 94. Australian National University, Canberra. Murik; Nor [F/F] Sepik-Ramu, Nor-Pondo (Lower Sepik), Nor. East Sepik Province, along the coast (Papua New Guinea). f

Stress falls on the first long vowel, and if there are none, on the first syllable.

¨damag ‘garden’ dIsi¨uøtH ‘sea waves’

ya¨goøt an«npHa¨r&EøtH

‘sick’ ‘lightning’

736

Papua New Guinea and Irian Jaya languages

Abott, S. (1985). A Tentative Multilevel Multiunit Phonological Analysis of the Murik Language. Papers in New Guinea Linguistics 22, Pacific Linguistics, Series A, No. 63, 339–373. Australian National University, Canberra. Yimas [I] Sepik-Ramu, Nor-Pondo, Pondo. East Sepik Province, Papua New Guinea. f f

Primary stress falls on the initial syllable. Secondary stress falls on the third syllable of words longer than three syllables.

¨awak ‘star’ ¨awtmayNi ‘sugarcane’

¨warpayn ‘close’ ¨wura˙takay ‘turtle’

Foley, W.A. (1991). The Yimas Language of New Guinea. Stanford: Stanford University Press. Torricelli Au [F/F] Torricelli, Wapei-Palei, Wapei. West Sepik Province (Papua New Guinea) f

f

Stress the first vowel that is not high central unrounded or mid central unrounded. Else stress the first.

¨mçtˆ ‘woman’ kç¨sawçn ‘he hides’

¨pˆrpˆrˆ ‘flying fox’ kˆ¨wat ‘he gives’

Scorza, D. (1985). A sketch of Au Morphology and Syntax. Papers in New Guinea Linguistics 22, Pacific Linguistics, Series A, No. 63, 215–273. Australian National University, Canberra. Bukiyip; Arapesh, Mountain [P] Torricelli, Kombio-Arapesh, Arapesh. East Sepik,Yangoru District (Papua New Guinea). f

Stress falls on the penultimate syllable.

¨yu h ‘completely’ u¨tabal ‘money’

¨buwul ‘pig’ na¨mudak ‘like that’

Papua New Guinea and Irian Jaya languages

737

Conrad, R.J., and Kepas Wogiga (1991). An Outline of Bukiyip Grammar. Pacific Linguistics, Series C, No. 113. Australian National University, Canberra. Fortune, R.F. (1942). Arapesh. Publications of the American Ethnological Society 19. J.J. Augustin: New York. Gerstner, A. (1963). Grammatik der Alubansprache. Micro-Bibliotheca Anthropos 37. St. Augustin bei Bonn: Anthropos Institut. Ningil [I (NMS)] Torricelli, Wapei-Palei, Wapei. West Sepic District (Papua New Guinea). f f

Stress is located on the first syllable and alternates thereafter. There is no prominence di¤erence between these stresses.

¨sçg ‘leaf ’ ¨tapa¨pi ‘small, few’

¨nˆmçk ‘eye’ ¨lçgçl¨tEki ‘jaw, chin’

Manning, M. & N. Jaggers (1977). A Tentative Phonemic Analysis of Ningil. In R. Loving (ed.) Phonologies of Five Papua New Guinea Languages, Workpapers in Papua New Guinea Languages 19, 49–72. Summer Institute of Linguistics, Ukarnumpa, Papua New Guinea. Yil [I/I] Torricelli, Wapei-Palei, Wapei. West Sepic District (Papua New Guinea). f

Stress is located on the first syllable, except when its vowel is schwa and the vowel in the second syllable is not schwa.

¨lawag ‘strong’ n«¨pin ‘red’

¨t«p«n ‘cassowary’ n«m¨rik ‘green bird’

Martens, M. & S. Tuominen (1977). A Tentative Phonemic Statement in Yil in West Sepik District. In R. Loving (ed.) Phonologies of five Papua New Guinea Languages. Workpapers in Papua New Guinea Languages 19, 29–48. SIL, Ukarumpa, PNG. Trans New Guinea Amele [F/F] Trans-New Guinea, Madang-Albert Range, Madang, Mabuso, Gum. Madang Province, around Madang (Papua New Guinea).

738 f

Papua New Guinea and Irian Jaya languages

Stress falls on the first closed syllable, else the first.

du¨an

ja¨walti ‘wind from north’

‘cold’

¨nifula ‘species of beetle’

iti¨tom

‘righteous’

Roberts, J.R. (1987). Amele. Guilford: Biddles. Asmat, Central [U;P (IRR)] Trans-New Guinea, Main Section, Central and Western, Central and South New Guinea-Kutubuan, Central and South New Guinea, Asmat-Kamoro. Along the coast of South-West Irian Jaya, Flamingo Bay. f f

Primary stress falls on the final syllable (there are exceptions). Secondary stress falls on the first syllable in three syllable words.

u¨si ‘camp’

e¨co

‘revenge’

˙usa¨wic ‘banana’

Voorhoeve, C.L. (1965). The Flamingo Bay dialect of the Asmat Language, Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 46. The Hague: M. Nijho¤. Ekari; Ekagi; Kapauku [S/- (pitch)] Trans-New Guinea, Main Section, Central and Western, Wissel LakesKemandoga, Ekari-Wolani-Moni. West Central Highlands of Irian Jaya. f

f

Syllable structure and pitch-accent seem to be connected in a very intricate way. Many words end up with stress on the second syllable (indicated with acute accent mark).

aw"@ø ‘to put in’ na@økagi ‘smoke’

ena@ø ‘ripe’ awe@øta ‘tomorrow’

Doble, M. (1962). Essays on Kapauku grammar. Nieuw Guinea Studien 6: 152–155. Doble, M. (1987). A Description of some Features of Ekari Language Structure. Oceania Linguistics 26: 55–113. Drabbe, P. (1952). Spraakkunst van het Ekagi, Wisselmeren, Nederlands Nieuw Guinea. The Hague: Nijho¤. Fore [F/F] Trans-New Guinea, Main Section, Central and Western, East New Guinea Highlands, East-Central, Fore. Eastern Highlands Province, Okapa District.

Papua New Guinea and Irian Jaya languages f f

739

Stress the first syllable with a high tone. In words without high tones stress the first syllable.

yo¨g"

‘knife’

wa¨n" ne

‘water’

nake.¨we

‘my ear’

keqpa.ta¨ra we

‘sand two.ind’

Nicholson, R. & R. Nicholson (1962). Fore Phonemes and their Interpretation. Oceanic Linguistic Monographs, No. 6. University of Sydney. Kaure [L/LEX (tone)] Trans-New Guinea, Kaure, Kaure Proper. Southwest of Lake Sentani (Irian Jaya). f

f f f

f f

Primary stress falls on a syllable that has a high level tone or a high-mid falling tone. If one high-mid falling tone is present it gets the stress. If two high-mid falling tones occur in sequence, the first is stressed. If more high-mid falling tones occur but not in sequence, the last is stressed. Other high-mid falling tones get secondary stress. Stress on syllables with high tones is unpredictable.

ka#¨t"sfi# )Ø l ‘will fish’ ¨haÆ my)# ‘wasp’

 l ˙t"s"#n")¨sfi^ ha#¨my)

‘wants to fish’ ‘honey’

Dommel, P.R. & G.E. Dommel (1991). Kaure Phonology. Workpapers in Indonesian Languages and Culture 9: 1–68. Jayapura, Irian Jaya: Cenderawasih University & SIL. Ketengan [LEX] Trans-New Guinea, Mek, Eastern. Eastern Highlands (Irian Jaya). f

Stress is phonemic.

¨bumE ‘horsefly’ ¨wENdoNona ‘to enter’

bu¨mE ‘short’ goRom¨dana ‘to bump into’

Sims, A. & A. Sims (1982). Ketengban Phonology. Workpapers in Indonesian Linguistics 1: 50–74. Ambon: SIL. Kewa [LEX] Trans-New Guinea, Main Section, Central and Western, East New Guinea Highlands, West-Central, Angal-Kewa. Southern Highlands Province, Kagua district.

740 f

Papua New Guinea and Irian Jaya languages

Stress is phonemic, but never occurs on the final syllable.

na¨maya ‘my father’ ke¨p" r" sa ‘it was covered’

¨kabe ‘wild cane’ ¨monane ‘at the side there’

Franklin, K.J. (1967). Kewa Sentence Structure. Papers in New Guinea Linguistics, No. 7, Pacific Linguistics, Series A, No. 13. Australian National University, Canberra. Franklin, K.J. (1971). A Grammar of Kewa, New Guinea. Pacific Linguistics, Series C, No. 16. Australian National University, Canberra. Franklin, K.J., and Joice Franklin. (1978). A Kewa Dictionary, with Supplementary Grammatical and Anthropological Materials. Pacific Linguistics, Series C, No. 53. Australian National University, Canberra. Kobon [P] Trans-New Guinea, Main Section, Central and Western, East New Guinea Highlands, Kalam, Kalam-Kobon. Madang and Western Highlands (Papua New Guinea). f f

Stress usually falls on the penultimate syllable. In the 3rd Sg Pres stress is final.

¨Maj«

‘big’

a¨lafe ‘tree’

a¨rab

‘he is going’

dan¨abel ‘they are bringing’

Davies, H.J. (1980). Kobon Phonology. Pacific Linguistics, Series B, No. 68. Australian National University, Canberra. Lower Grand Valley Dani [U] Trans-New Guinea, Main Section, Central and Western, Dani-Kwerba, Southern, Dani. Central Highlands, Baliem Grand Valley (Irian Jaya). f

f

Stress usually falls on the final syllable of the word (stressless clitics may follow). The first syllable of words with three syllables or more is usually pronounced louder and with a secondary upward pitch movement.

pI¨r˛ ‘two’ Bromley, H.M. (1961). The Phonology of Lower Grand Valley Dani, Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 34. The Hague: M. Nijho¤.

Papua New Guinea and Irian Jaya languages

741

Mairasi [U/P (pitch)] Trans-New Guinea, Main Section, Central and Western, MairasiTanamerah, Mairasi. Bomberai Peninsula (Irian Jaya). f f f

f

Primary stress is located on the final syllable if it contains a long vowel. Else stress is penultimate. Secondary stress falls on the first syllable of words with three syllables or more. Stress is realised as high or neutral pitch.

¨EømE

‘we’

˙aba¨ba/a ‘toy’

/a¨Boø

‘close’

˙ururusa¨neve ‘will swing’

Peckham, L. (1991). Mairasi Phonology. Workpapers in Indonesian Languages and Culture 10: 111–145. Jayapura, Irian Jaya: Cendera-wasih University & SIL. Marind [U;P] Trans-New Guinea, Main Section, Central and Western, Marind, Marind Proper. South coast of Irian Jaya. f

f

Primary stress is usually located on the final syllable (with penultimate exceptions). Secondary stress on alternating syllables before the main stress.

ka˙daha¨but ‘island’

˙dira¨ri

‘watercooker’

so¨padem ‘name’

Drabbe, P. (1955). Spraakkunst van het Marind. Studia Instituti Anthropos 11. Geurtjens, H. (1926). Spraakleer der Marindeesche taal. Verhandelingen van het Bataviaasch Genootschap 68/2. Bandoeng: Nix & Co. Ono [I] Trans-New Guinea, Finisterre-Huon, Western Huon. Northeast coast of Huon Peninsula, Morobe Province (Papua New Guinea). f f

Primary stress is initial. Secondary stress on alternates thereafter, but not necessarily on all alternates.

¨sekao ‘space under a house’ ¨mesi˙kene ‘you will sit’

¨lolot˙ne ¨ari˙mage˙ake

‘many’ ‘he always goes’

742

Papua New Guinea and Irian Jaya languages

Phinnemore, T.R. (1985). Ono Phonology and Morphophonemics. Papers in New Guinea Linguistics 22, Pacific Linguistics, series A-63, 173–214. Australian National University, Canberra. Selepet [I] Trans-New Guinea, Main Section, Central and Western, Huon-Finisterre, Huon, Western. Pumune, Kiari rivers (Papua New Guinea). f Primary stress is initial. f Secondary stress on alternates thereafter, optional for three syllable words. ¨lek ‘man’

¨pato ‘large’

¨boko˙sok ‘mud’

McElhanon, K.A. (1970). Selepet Phonology. Pacific Linguistics, Series B, No. 14. Australian National University, Canberra. Sentani [U/P] Trans-New Guinea, Main Section, Central and Western, Sentani, Sentani proper. Pumune, northeastern West Irian. f Primary stress is final if the last syllable is closed, otherwise stress is penultimate. f There are exceptions, some of them depending on openness of the penult. f Secondary stress follows a ternary pattern, respecting closed syllables. ˙«d«ka¨wale

‘I saw thee’

a˙dil«mi¨hibe

haba¨kay

‘tobacco’

ha˙b«wnoko¨kale ‘I struck him [aor]’

ha˙b«wdo¨koke ‘he hit me [aor]’ adi˙l«d«mi¨him

‘you two will collect them’ ‘let me collect them’

Cowan, H.J. (1965). Grammar of the Sentani language. Verhandelingen van het Koninklijk Instituut voot Taal-, Land- en Volkenkunde 47. The Hague: Martinus Nijho¤. Foley, W.A. (1986). The Papuan Languages of New Guinea. Cambridge: Cambridge University Press. Siroi [S] Trans-New Guinea, Madang-Adelbert Range, Madang, Rai Coast, Kabenau. Pumune, Madang Province (Papua New Guinea). f Stress occurs on the second, or only, syllable of the stem. ¨soN ‘parcel’ ku¨mah ‘dead’ am¨bonga ‘at first’

ku¨bele

‘yesterday’

Wells, M.A. (1979). Siroi Grammar. Pacific Linguistics, Series B, No. 51. Australian National University, Canberra.

Papua New Guinea and Irian Jaya languages

743

Tabla; Tanh Merah [P/A] Trans-New Guinea, Main Section, Central and Western, Sentani, Sentani Proper. North coast around Tanahmerah Bay (Irian Jaya). f

Stress occurs on the penultimate syllable, unless its vowel is schwa and there is an antepenultimate syllable to which stress can shift.

¨suaN‘

‘kind of wood’

¨k«ndo ‘coconut cream’

¨dimi

‘cry’

¨suNk«nuN ‘mosquito’

Collier, K. & M. Collier (1986). Tabla Phonology. Workpapers in Indonesian Languages and Culture 3: 1–16. Jayapura, Irian Jaya: Cenderawasih University & SIL. Usan [LEX] Trans-New Guinea, Madang-Adelbert Range, Adelbert Range, PihomIsumrud-Mugil, Pihom, Numugenan. Madang Province (Papua New Guinea). f

Stress is unpredictable and plays a role in the morphology.

¨gugum ‘all’ ¨gobi ‘taro seedling’

gu¨gum ‘cold’ go¨bi ‘a tree’

Reesink, G.P. (1987). Structures and their functions in Usan: a Papuan language of Papua New Guinea. Amsterdam: John Benjamins. Wahgi [U;P;LEX (tone)] Trans-New Guinea, Main Section, Central and Western, East New Guinea Highlands, Central, Wahgi. Highlands (Papua New Guinea). f

f f

Primary stress is final in most words. Proper names and some disyllabic words ending in vowels may have penultimate stress (the latter may also have two equal stresses). Secondary stress reported. In many words stress is indistinguishable from high tone.

¨Ne$n"@ka$ ‘a clan name’

¨ka@ n5˙d5 z5" $p

‘star’

˙ko$n5d5z5"@¨ka@

‘clan name’

Luzbetak, L.J. (1956). Middle Wahgi Phonology. Oceania Linguistic Monographs 2. University of Sydney, Sydney. Phillips, P.J. (1976). Wahgi Phonology and Morphology. Pacific Linguistics, Series B, No. 36. Australian National University, Canberra.

744

Papua New Guinea and Irian Jaya languages

Wambon [P;U (IRR)] Trans-New Guinea, Main Section, Central and Western, Central and South New Guinea-Kutubuan, Central and South New Guinea, Awyu-Dumut, Dumut. South coast (Irian Jaya). f

Stress usually falls on the penult. In some words stress can be either final or penultimate.

tam¨bulop or tambu¨lop ‘kind of fruit’ Vries, L. de & R. de Vries-Wiersma (1992). The Morphology of Wambon of the Irian Jaya Upper-Digul Area, with an Introduction to its Phonology. Verhandelingen van het Koninklijk Instituut voor Taal, Land en Volkenkunde 151. Leiden: KITLV Press. Weri [U] Trans-New Guinea, Main Section, Eastern, Central and Southeastern, Goilalan, Kunimaipa. Morobe District (Papua New Guinea). f f f

Primary stress usually falls on the final syllable. Secondary stress on alternate syllables before the main stress. In words of four or five syllables the antepenultimate and the final stress may have equal strength (reportedly the antepenult may even be stronger in some cases).

¨kan ‘road’ ˙kuli¨pu ‘hair of arm’

Nin¨tip ‘bee’ ˙aku˙nete¨pal ‘times’

Boxwell, H. & M. Boxwell (1966). Weri Phonemes. In S.A. Wurm (ed.) Papers in New Guinea Linguistics, No. 5. Australian National University, Canberra, pp. 77–93. Woisika [IRR] Trans-New Guinea, South Bird’s Head-Timor-Alor-Pantar, Timor-AlorPantar, Makasai-Alor-Pantar, Alor. Alor Island (East Indonesia). f f

Stress is distinctive. It is completely free in bisyllables. Words of three syllables or more with final stress are rare.

¨sibe ‘chicken’

ta¨pui ‘crab’

la¨mata

‘greedy’

Stokhof, W.A.L. (1979). Woisika II. Phonemics. Pacific Linguistics, Series B, No. 59. Australian National University, Canberra.

North American languages

Algic Cree, plains [A] Algic, Algonquian, Central, Cree. Eastern Canada. f

f

Primary stress falls on the antepenultimate syllable, but is final in bisyllabic words. Secondary stress falls on alternate syllables before and after the primary stress.

is¨kweøw ‘woman’ ¨neøhi˙yaw ‘cree indian’

mih¨ti ˙pasa˙kwaøpi¨simo˙win

‘piece of firewood’ ‘shut eye dance’

Wolfart, H.C. (1996). Sketch of Cree, an Algonquian Language. In: I. Goddard (ed.) Handbook of North American Indians, Volume 17. Washington: Smithsonian Institution, 390–439. Okimasis, J. and S. Ratt. (1999). Cree: Language of the Plains/ nehiyawewin: paskwawi-pikiskwewin. Canadian Plains Research Center, University of Regina. Menomini [L/ ] Algic, Algonquian, Central. Wisconsin (US). f f

f

Primary stress falls on penultimate long vowels and diphthongs. Primary stress also falls on long vowels followed by a syllable with a short vowel. Secondary stress comes on even numbered syllables in a succession of short ones, but never on the last syllable.

e¨nEøniw ‘man’ pia¨tEøtuaq ‘when they arrive’ ne¨piøhti˙kEmenaw ‘we enter’ ¨meøwe˙nEse˙hEwak ‘they drive him away’ Bloomfield, L. (1962). The Menomini Language. Yale University Press, New Haven, Conn. Pesetsky, D. (1979). Menomini Quantity. In K. Safir (ed.) MIT Working papers in Linguistics, Vol. 1, pp. 115–139.

746

North American languages

Munsee; Delaware [L (CNT)] Algic, Algonquian, Eastern. Delaware River, lower Hudson (Ontario, US). f f

Secondary stress on all even syllables counted from the left. Primary stress falls on the last non-final stressed syllable.

w«˙lama¨l«s«w

‘he is well’

n«˙kak«¨tak«˙ka

‘i do a fast dance’

Goddard, I. (1979). Delaware Verbal Morphology. New York: Garland Publishing. Goddard, I. (1982). The Historical Phonology of Munsee. International Journal of American Linguistics 48: 1, pp. 16–48. Passamaquoddy; Malecite [L (CNT)] Macro-Algonquian – Algonquian. Maine (US), New Brunswick, Quebec (Canada). f

f f

Secondary stress falls on even syllables counted from the left, respecting the weight of full vowels (long as opposed to schwa). These cannot be in the weak position of the iambic feet, schwa can be. Primary stress falls on the stressed syllable of the prefinal foot. There is reportedly an alternating prominence di¤erence between the stresses themselves.

˙pe¨tek«po ˙poh¨tayak

‘it comes to be located here’ ‘bottle.loc’

˙kin«¨w«s«w«k ‘they are certain ones’ ˙et«l«¨lohket ‘he who is working’ ˙mek«˙nut«¨s«p«nik ‘those [animate] who must have been chosen’ LeSourd, P.S. (1993). Accent and Syllable Structure in Passamaquoddy. Outstanding Dissertations in Linguistics. New York: Garland. Teeter, K.V. (1971). ‘‘The main features of Malecite-Passamaquoddy Grammar’’. In Jesse Sawyer (ed.) Studies in American Indian Languages. University of California Publications in Linguistics 65, Univ. of California Press, Berkeley, pp. 191–249. Teeter, K.V. & P. LeSourd (1983). Vowel Length in Malecite. In William Cowen (ed.) Actes du quatorzieme congres des algonquinistes. Carleton University, Ottawa, pp. 245–248.

North American languages

747

Unami; Delaware [L (CNT)] Algic, Algonquian, Eastern. Oklahoma, New Jersey, Delaware valley (US). f

f f

Secondary stress on initial long vowels or closed syllables and alternates thereafter. Otherwise secondary stress on all even syllables counted from the left. Primary stress falls on the last non-final stressed syllable.

˙eønta¨maxka˙wiøt

‘when he found me’

Goddard, I. (1979). Delaware Verbal Morphology. New York: Garland Publishing. Goddard, I. (1982). The Historical Phonology of Munsee. International Journal of American Linguistics 48: 1, pp. 16–48. Wiyot [F/F] Algic, Wiyot. Northwestern California (US). f

Stress on formerly long vowels or initial if there are none.

bǬlid

‘eye’ from ‘meliøni’

¨pumip«/l ‘stone knife’

Proulx, P. (1984). Proto-Algic I: Phonological Sketch. International Journal of American Linguistics 50: 2, pp. 165–207. Caddoan Kitsai [IRR] Caddoan, Northern, Pawnee-Kitsai, Kitsai. Texas, Oklahoma (now extinct). f

Stress is completely irregular.

Bucca, S. & A. Lesser (1969). Kitsai Phonology and Morphophonemics. International Journal of American Linguistics, No. 35, pp. 7–19. Wichita [all/L] Caddoan, Northern, Wichita. Texas, Oklahoma (now extinct). f

f f f

Many counterexamples, but generally all high pitched syllables have primary stress. If there is no high pitch all long vowels have primary stress. If there are no long vowels the last voiced vowel has primary stress. Secondary stress on alternating vowels left and right of primary stress, avoiding clash.

748

North American languages

˙naheø¨ha@rih ‘creek’

˙tika˙siøskW

‘he has a beard’

˙tika¨/acs

¨arasi¨citeø˙riw

‘you finished it’

‘he is eating it’

Rood, D.S. (1996). Sketch of Wichita, a Caddoan Language. In Ives Goddard, ed., Handbook of American Indians, Vol. 17: Languages, pp. 580–608. Washington: Smithsonian Institute. Eskimo-Aleut Greenlandic, west (dialect of Greenlandic; Inuktitut) [NMS] Eskimo-Aleut, Eskimo, Inuit. Greenland. f f

Stressing unclear. Prominence of heavy syllables reported. All stresses are equally prominent.

Schultz-Lorenzen (1945). A Grammar of the West Greenlandic Language. Copenhagen: C.A. Reitzel. Fortescue, M. (1984). West Greenlandic. Croom Helm Descriptive Grammars. London: Croom Helm. Rischel, J. (1974). Topics in West Greenlandic Phonology. Copenhagen: Akademisk Forlag. Norton Sound (dialect of Central Alaskan Yupik) [I/(NMS)] Eskimo-Aleut, Eskimo, Yupik, Alaskan. Norton Sound, Southwestern Alaska (US). f f f

f f

Stress all long vowels and closed first syllables. Also stress the even numbered syllables in a light syllable string. Remaining unstressed syllables that are followed by a heavy syllable are stressed. Final syllables are unstressed. All stresses are equally prominent.

¨aNyaX¨paøka

‘my big boat’

qa¨yaXpaN¨yuxtuq ‘he wants to get a big kayak’ ma¨qi¨kaøtxun ‘with their [other’s] future steambath material’ qa¨ya¨pix¨kaøni ‘in his [another’s] future authentic kayak’ Jacobson, S.A. (1985). Siberian Yupik and Central Yupik Prosody. In M. Krauss (ed.) Yupik Eskimo Prosodic Systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 25–46.

North American languages

749

Yupik, Chevak (dialect of Central Alaskan Yupik) [I/(NMS)] Eskimo-Aleut, Eskimo, Yupik, Alaskan. Chevak, north central coast of Alaska (US). f

Stress pattern as in General Central Yupik, but with destressing in case there is a clash only to the left (see second example).

¨aNyax¨kami i¨kamilut«N

‘how about my two boats’ ‘they having sleds’

Jacobson, S.A. (1985). Siberian Yupik and Central Yupik Prosody. In M. Krauss (ed.) Yupik Eskimo Prosodic Systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 25–46. Woodbury, A. (1981). Study of the Chevak Dialect of Central Yup’ik Eskimo. Doctoral Dissertation, University of California, Berkeley. Yupik, General Central [I/(NMS)] Eskimo-Aleut, Eskimo, Yupik, Alaskan. Alaska (US). f f f f f f f

Long vowels and initial closed syllables are always stressed. A closed syllable preceding an open, light, syllable is stressed. Syllables following unstressed syllables are stressed. Unstressed remaining syllables preceding heavy syllables are stressed. Before final CV, posttonic CVC is not stressed. Final syllables are unstressed. All stresses are equally prominent.

¨akNix¨tatNa

‘they hurt me interrogative’

¨aN¨yaXpaka ‘my big boat’

pi¨sutu¨lini¨luni ‘he apparently always hunted’ Jacobson, S.A. (1985). Siberian Yupik and Central Yupik Prosody. In M. Krauss (ed.) Yupik Eskimo Prosodic Systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 25–46. Leer, J. (1985). Toward a metrical interpretation of Yupik prosody. In M. Krauss (ed.) Yupik Eskimo Prosodic systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 159–172. Miyaoka, O. (1985). Accentuation in Central Alaskan Yupik. In M. Krauss (ed), Yupik Eskimo Prosodic Systems: Descriptive and comparative studies, Alaska Native Language Center, Fairbanks. pp. 51–76. Yupik, Pacific Gulf; Alutiiq; Chugach [I/S (NMS)] Eskimo-Aleut, Eskimo, Yupik, Alaskan. Prince William Sound, Kodiak Island (US).

750 f f

f f

North American languages

Stress all long vowels and closed first syllables. In words with only light syllables the second and every third thereafter are stressed. If only two syllables remain after a stress, the last is stressed. All stresses are equally prominent.

¨ag¨Nuaqu¨tar¨tuaNa ‘i am going to dance’

¨taøta¨qa

‘my father’

ta¨qumalu¨ni ‘apparently getting done’ ¨naømi¨ci¨qua ‘I will su‰ce’ ¨atsaxsu¨quøtaqu¨ni ‘if he [refl] is going to get berries’ Kraus, M. (1985). Yupik Eskimo Prosodic Systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. Leer, J. (1985). Prosody in Alutiiq. In M. Kraus (ed.), Yupik Eskimo Prosodic systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 77–133. Leer, J. (1985). Toward a Metrical Interpretation of Yupik Prosody. In M. Kraus (ed.), Yupik Eskimo Prosodic systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 159–172. Yupik, St. Lawrence Island (dialect of Central Siberian Yupik) [NMS] Eskimo-Aleut, Eskimo, Yupik, Siberian. St. Lawrence Island. (Alaska) and South of Chuckchee peninsula (Siberia). f f f

Stress all long vowels and even syllables in a light syllable string. Word final syllables are not stressed. All stresses are equally prominent.

¨NaøNa ‘his mother’ aN¨yaX´aX¨´aNyuxtuq ¨aøNqaX¨´aX´aN¨yuxtuq

a¨taøta ‘uncle’ ¨tuø¨naø˜aø ‘he’ll give it’ ‘he wants to make a big boat’ ‘he wants to make a big ball’

Jacobson, S.A. (1984). The stress conspiracy and stress repelling bases in Central Yupik and Siberian Yupik Eskimo languages. International Journal of American Linguistics 50: 3, 312–324. Jacobson, S.A. (1985). Siberian Yupik and Central Yupik Prosody. In M. Krauss (ed.) Yupik Eskimo Prosodic Systems: Descriptive and Comparative Studies. Alaska Native Language Center, Fairbanks. pp. 25–46. Kraus, M. (1975). St Lawrence Island Eskimo Phonology and orthography. Linguistics 152, 39–72.

North American languages

751

Gulf Tunica [I] Gulf. Central Louisiana (US). f

f

The first syllable of every morpheme seems to be stressed (inferred from dictionary). No di¤erence between primary and secondary stresses reported.

¨rus&ta ‘rabbit’ ¨kohina ‘clay vessel, cup’

¨rus&ta¨hipu ‘rabbit dance’ ¨kohina¨makini ‘bowl i.e. deep clay vessel’

Haas, M.M. (1953). Tunica Dictionary. University of California Publications in Linguistics 6. Berkeley and LA: University of California Press. Hokan Karok [F/L] Hokan, Northern, Karok-Shasta. Northwestern California, along the banks of the Klamath River (US). f f

If a long voweled syllable is present, the stress falls on it. Otherwise the stress falls on the final syllable.

/im¨ma>n

‘tomorrow’

kuk¨/u>mi ‘go there’

/iS¨puk ‘money’

Bright, W. (1957). The Karok Language. Berkeley: University of California Press. Kashaya; Southwestern Pomo [S-T/F] Hokan, Northern, Pomo, Russian River and Eastern, Russian River, Southern. Northern California (US). f f f f

f

The first syllable does not count for stress assignment. If the second syllable is short and open, accent the following syllable. If the second is short and closed it gets stress. If the second contains a long vowel the third syllable gets stress if it is closed or long. If in this case the third syllable is open and short, the fourth syllable gets stress.

/ime¨/ahqol ‘long fuzz’ /i¨mehmi ‘real fuzz’ /icHaø¨suøle ‘spider web’

/icHaøtHi¨neømu /ima¨tahmi

‘that is not a spider’ ‘a real woman’

752

North American languages

Oswalt, R.L. (1961). A Kashaya Grammar (Southwestern Pomo). Ph.D. Dissertation, University of California, Berkeley. Oswalt, R.L. (1988). The Floating Accent of Kashaya. In W. Shipley (ed.) In Honor of Mary Haas, 611–622. Berlin: Mouton de Gruyter. Maricopa [U] Hokan, Esselen-Yuman, Yuman, River Yuman. Gila River and Salt River reservations near Phoenix, Arizona (US). f

Stress falls on the final syllable.

iø¨Do ‘face, eye’ ¨Dik ‘here’

/a¨ve ‘snake’ /a¨veø ‘mouse’

Gordon, L. (1986). Maricopa Morphology and Syntax. University of California Publications in Linguistics 108. Berkeley: University of California Press. Meso Grande Dieguen˜o (dialect of Dieguen˜o) [U/P] Hokan – Yuman – Delta-Californian. Southern California (US). f

f f

Stress falls on the final syllable if it contains a long vowel or closing consonant, unless the vowel before that consonant is schwa. If not, stress the penult. Secondary stress on long vowels and the first of two unstressed syllables reported.

˙ace¨pac ‘he took it out’ ˙aø¨kwal ‘he licks’

/«¨naøv«k ‘if we go’ /«¨mati ‘on the ground’

Couro, T. & C. Hutcheson (1973). Dictionary of Mesa Grande Dieguen˜o. Banning: Malki Museum Press, California. Langdon, M. (1970). A Grammar of Dieguen˜o (The Mesa Grande Dialect). University of California Publications in Linguistics 66. Berkeley and LA: University of California Press. Pomo, Eastern [S] Hokan, Northern, Pomo, Russian River and Eastern, Eastern. Around Clear Lake, Sacramento River, California (US). f f

Primary stress on the second syllable. Secondary stress on alternate syllables after the main stress.

North American languages

diø¨leø ‘middle’

diø¨le

753

‘forehead’

McLendon, S. (1975). A Grammar of Eastern Pomo. University of California Publications in Linguistics, 74. Quechan; Yuma; Kwitchyan [U] Hokan, Esselen-Yuman, Yuman, River Yuman. Ft. Yuma Reservation, southeastern corner of California (US). f f

Stress falls on the final syllable of the stem. Some su‰xes have their own (primary) stress.

a¨/im ‘he says’

mas&a¨Dek ‘he is afraid’

/a¨Du¨wu>m ‘would i do’

Halpern, A.M. (1946). Yuma I: phonemics. IJAL 12, pp. 60–63; Yuma II: morphophonemics. IJAL 12, pp. 147–151. Yana [F/F] Hokan, Northern, Yana. California (Extinct). f

Stress falls on the first closed syllable or long vowel, else on the first.

i¨taal/pa ¨put/uk/u

‘head scratcher’

¨galu

‘arm’

‘skull’

hap/a¨laamaubiiwi

‘mud’

Sapir, E. & M. Swadesh (1960). Yana Dictionary. University of California Publications in Linguistics, No. 22. Berkeley and LA: University of California Press. Iroquoian Cayuga (dialect of Seneca) [L (CNT)] Iroquoian, Northern Iroquoian, Five Nations, Seneca-Onondaga, SenecaCayuga. Ontario (Canada), New York, Oklahoma (US). f f f

Main stress falls on the last even numbered syllable that is not final. If the penult contains a long vowel it always has main stress. Secondary stress on even syllables counted from the left.

ka¨nesta/ e)˙kata˙to)kw/e¨to)ønye)/

‘board’ ‘I will make some people for myself ’

Chafe, W.L. (1977). Accent and related phenomena in the Five Nations Iroquois languages. In L. Hyman (ed.) Studies in Stress and Accent, pp. 169–181. Southern California Occasional Papers in Linguistics, No. 4. Los Angeles: University of Southern California Press.

754

North American languages

Foster, M. (1982). Alternating Weak and Strong Syllables in Cayuga Words. International Journal of American Linguistics 48, pp. 59–72. Mohawk, Akwesasne (dialect of Mohawk) [P] Iroquoian, Northern Iroquoian, Five Nations, Mohawk-Oneida. Ontario, Quebec (Canada), New York (US). f

The stressed vowel is usually the penult.

o¨nuøta/ ‘hill’

¨ohskyv ‘bone’

ok¨waho

‘wolf ’

Bonvillain, N. (1973). A Grammar of Akwesasne Mohawk. Ethnology Division Paper 8. Ottawa: National Museum of Man. Michelson, K. (1988). A comparative study of Lake-Iroquian Accent. Dordrecht: Kluwer Academic Publishers. Oneida [P] Iroquoian, Northern Iroquoian, Five Nations, Mohawk-Oneida. South Ontario (Canada). Central New York, eastern Wisconsin (US). f f

The stressed vowel is usually the penult. Shift of pitch to the final syllable reported.

teyutoh¨talhos

‘she tidies up’

teyohˆ¨letha/

‘she is hollering’

Michelson, K. (1988). A comparative study of Lake-Iroquoian accent. Dordrecht: Kluwer Academic Publishers. Seneca [L/- (CNT:tone)] Iroquoian, Northern Iroquoian, Five Nations, Seneca-Onondaga, SenecaCayuga. Ontario (Canada), New York (US). f

f

Stress on the last even-numbered non-final closed syllable, or on the last even-numbered non-final open syllable immediately followed by a closed syllable. If criteria for accenting are not met, the word will bear no accent at all.

he)¨no)hto)nyo)h ‘he is thinking’

ho¨tino)hso)øniøh

‘Iroquois league’

Chafe, W.L. (1960). Seneca Morphology I: Introduction. IJAL 26: 11–22. Chafe, W.L. (1977). Accent and related phenomena in the Five Nations Iroquois languages. In L. Hyman (ed.) Studies in stress and Accent, pp. 169–181. Southern California Occasional Papers in Linguistics, No. 4. Los Angeles: University of Southern California Press.

North American languages

755

Chafe, W.L. & M.K. Foster (1981). Prehistoric divergences and recontacts between Cayuga, Seneca and other Northern Iroquoian languages. IJAL 47: 121–142. Muskogean Chocktaw [pitch] Muskogean, Western. Oklahoma (US). f

f

The selection of which syllable is most prominent in the word is determined by the pitch accent system, which is entirely independent of metrical structure. Rhythmic stress on long vowels, closed syllables, and every second light syllable.

li¨tiøha¨tok

pi¨saøli

‘it was dirty’

¨okc&a¨liølih ‘I woke him up’

‘see I [subj]’

sa¨liøti¨haøtok ‘I was dirty’

Munro, P. & C. Ulrich (1984). Structure-preservation and Western Muskogean Rhythmic Lengthening. Proceedings of the Third West Coast Conference on Formal Linguistics. Stanford University. Nicklas, T.D. (1972). The elements of Chocktaw. Doctoral dissertation. University of Michigan Ann Arbour. Nicklas, T.D. (1975). Chocktaw morphophonemics. In J.M. Crawford (ed.) Studies in Southeastern Indian Languages. University of Georgia Press. Athens, 237–250. Ulrich, C. (1986). Chocktaw morphoponology. Doctoral dissertation. UCLA. Koasati [P/U (pitch)] Muskogean, Eastern. Koasati reservation, Louisiana, and AlabamaKoasati reservation, Texas (US). f

f

If the penultimate syllable has a long vowel, that syllable bears the pitch accent. If not, the ultimate syllable bears the pitch accent.

salakla@ ‘wild goose’ pala@øna ‘plate’

taktalasbas"@ ‘handkerchief ’ ibisa@øni ‘nose’

Kimball, G. (1991). Koasati Grammar. Lincoln, Nebraska: University of Nebraska Press.

756

North American languages

Muskogee; Creek [L (CNT, pitch)] Muskogean, Eastern. Oklahoma (US). f

f

f

f

f

Primary stress will be final if the final syllable is closed or contains a long vowel. Otherwise it will bepenultimate if the that syllable is closed or contains a long vowel. In all other cases primary stress is located on the last even syllable counted from the last long vowel, closed syllable or in case there are none, the left word edge. There is some debate as to whether the secondary stresses are actually realised. The accent is tonal.

i¨fa

ko¨fotSka

‘dog’

apata¨ka ‘pancake’ am apa¨taka ‘our pancake’ taøsho¨kita ‘to jump’ [dual subj.]

‘mint’

tiøniøt¨kiø ‘thunder’ iNkosapi¨ta ‘one to implore’ taøski¨ta ‘to jump’ [sg. subj.]

Haas, M. (1977). Tonal Accent in Creek. In L. Hyman (ed.) Studies in Stress and Accent. Southern California Occasional Papers in Linguistics, No. 4. University of Southern California, LA, pp. 195–208. Na-Dene Masset Haida; Northern Haida (dialect of Haida) [L/L (tone)] Na-Dene, Haida. Queen Charlotte Island, British Columbia (Canada). f f

f

Primary stress falls on the last high tone, else on the last syllable. Secondary stress falls on the last syllable if primary stress is located elsewhere. Secondary stresses also fall on alternates before the main stress and before the final secondary stress (if present).

˙guuding¨ee ˙gudi˙sahlda¨hid dla˙dahlda˙gada¨hahl kil˙aldaanh¨la@ng˙id/wa˙gan

‘giant purple urchin’ ‘worry.inceptive’ ‘fall down.into water.caus.dir.that’ ‘caus.pot.be.slave.inceptive.pl.past’

Enrico, J. (1991). The Lexical Phonology of Masset Haida. Alaska Native Language Center Research Papers 8. Fairbanks: Alaska Native Language Center, University of Alaska.

North American languages

757

Slave [LEX (tone)] Na-Dene, Nuclear Na-Dene, Athapaskan-Eyak, Athapaskan, Canadian, Hare-Chipewyan, Hare-Slave. Alberta, BC, Nw Territories (Canada). f

f

Stress generally falls on the vowel immediately preceding the stem vowel of a noun. Many intricate interactions between tone assignment, morphology and stress.

/o@ne@de@/

‘s/he gave away’

/o@ne@¨de/

‘I gave away’

Rice, K. (1989). A Grammar of Slave. Berlin: Mouton de Gruyter. Penutian Hanis Coos [LEX (P)] Penutian, Oregon Penutian, Coos. Southern Oregon (US). f f

Stress is reported to be lexically distinctive. Many words, however, seem to have penultimate stress.

¨helaq ‘to get, to arrive’ ¨alqas ‘fear’

he¨laq wi¨xilis

‘to climb up’ ‘food’

Frachtenberg L.J. (1922). Coos. In F. Boaz (ed.) Handbook of American Indian Languages, part 2. Smithsonian Institution Bureau of American Ethnology Bulletin 40, pp. 297–429. Washington: Government Printing O‰ce. Kalapuya [I] Penutian, Oregon Penutian, Kalapuyan. Northwest Oregon (US). f

Stress is fixed on the first syllable of the stem.

Berman, H. (1990). An Outline of Kalapuya Historical Phonology. IJAL 56: 1, pp. 27–59. Klamath-Modoc [L/P/A] Penutian, Plateau Penutian, Klamath-Modoc. Klamath, Oregon; Modoc, California (US). f f f

Primary stress falls on the last long vowel. If there is no long vowel stress closed penults, else the antepenult. Secondary stress on closed penults if the primary stress is located elsewhere.

758

North American languages

nis¨qaøk ‘little girl’

gat¨bambli

‘returns home’

¨cawiga

ga¨wiønapgabli

‘is going among them’

‘is crazy’

Barker, M.A.R. (1964). Klamath Grammar. University of California Press Publications in Linguistics 31, Berkeley. Maidu, Mountain [I/S;I] Penutian, California Penutian, Maidu. Northern California (US). f f f f

Primary stress is initial if the first syllable is closed. Otherwise stress falls on the second syllable. There are many words with lexicalized initial stress. Secondary stress falls two syllables to the right of the main stress (optional when final).

ky¨le

‘woman’

wi¨setpem ‘frightened’ ca¨tata˙k’a ‘rattlesnake’

¨jukbom

‘bear dance’

¨synda˙ka ‘forehead’ ¨/omka˙najdi ‘under the rock’

Robbins, S. (1991). Lexicalized Metrical Foot Structure in Maidu. In A. Mester & S. Robbins (eds.) Phonology at Santa Cruz 2, 95–116. Syntax Research Center, University of California, Santa Cruz. Shipley, W.F. (1964). Maidu Grammar. University of California Publications in Linguistics 41. Berkeley and Los Angeles: University of California Press. Miwok, Central Sierra [I/S] Penutian, California Penutian, Miwok-Costanoan, Miwok, Eastern. California (formerly upper valleys of the Stanislause and Tuolumne, US). f f

Stress is initial if the first syllable is closed or contains a long vowel. Otherwise stress falls on the second syllable.

ta¨naøja ‘evening star’ ¨koøkotSu ‘clover’

ku¨ma/sa ‘aunt’ ¨nopta ‘to drop’

Freeland, L.S. (1951). Language of the Sierra Miwok. Memoir 6 of the International Journal of American Linguistics. Freeland, L.S. & S.M. Broadbent (1960). Central Sierra Miwok Dictionary, with texts. UCPL 23. Berkeley: University of California Press.

North American languages

759

Miwok, Northern Sierra [I/S] Penutian, California Penutian, Miwok-Costanoan, Miwok, Eastern. California, upper valleys of Mokelumne and Calaveras rivers (US). f

f f

Primary stress is initial if the first syllable is closed or contains a long vowel. Otherwise stress falls on the second syllable. Secondary stress occurs on alternating syllables after the primary stress.

¨toøno ‘bull’ pa¨latta˙ta ‘woodpecker’

¨poku ‘to burst, explode’ la¨piøsa˙jy ‘fish’

Callaghan, C.A. (1987). Northern Sierra Miwok Dictionary. Berkeley: University of California Press. Miwok, Southern Sierra [I/-] Penutian, California Penutian, Miwok-Costanoan, Miwok, Eastern. California (US). f f

f f

Primary stress falls on the first long vowel of a word. This is always the first or the second syllable, words without long vowels do not exist. Secondary stress occurs on other long vowels. In a sequence of long vowels the odd numbered ones are more heavily stressed.

hi¨saøk ‘to hiss’ ha¨kaø§a/ ‘golden cup oak’

¨huø§u/ ‘buzzard’ ¨toøko§u/ ‘ear’

Broadbent, S.M. (1964). The Southern Sierra Miwok Language. University of California Publications in Linguistics, No. 38. Freeland, L.S. (1951). Language of the Sierra Miwok. Memoir 6 of International Journal of Ameriacan Linguistics. Indiana University, Bloomington, Indiana. Nez Perce [LEX] Penutian, Plateau Penutian, Sahaptin. Northern Idaho (US). f f

Stress can occur throughout the word. Sometimes stress has a lexical function.

¨tehes ‘ice’ ceøptukt¨x8e/pt ‘to crawl under’

qu¨quke/ykt wisteq¨neømit

‘to gallop’ ‘gift’

760

North American languages

Aoki, H. (1970). Nez Perce Grammar. Berkeley: University of California Press. Tsimshian, Coast; Sm’algyax (dialect of Tsimshian) [U;P] Penutian, Tsimshian. Northern coast of British Columbia (Canada). f f

Stress generally occurs on the final syllable. There are some words with lexical penultimate stress (not caused by su‰xation).

ksa¨woox

‘dream’

a´di¨qaws ‘humming bird’

nip¨/a/la

‘button’

tkad¨zemsk ‘boiled whole fish’

Dunn, J.A. ed. (1995). Sm’algyax: A reference dictionary and grammar for the Coast Tsimshian Language. Seattle: University of Washinton Press, Juneau: Sealaska Heritage Foundation. Yawelmani (dialect of Valley Yokuts) [P;A] Penutian, California Penutian, Yokuts. California (US). f f

Stress generally occurs on the penultimate syllable. There are many words with antepenultimate stress.

/oh¨yoohin ‘searched for’

yoloo¨winhin ‘assembled.intrans’

¨hooyexot

¨/ohyoxo/

‘is being sent’

‘is searching’

Newman, S. (1944). The Yokuts Language of California. The Viking Fund Publications in Anthropology, No. 2. NY. Salishan Salish, Southern Puget Sound [LEX] Salishan, Central Salish, Twana. Puget Sound, Washington (US). f

The placing of stress is phonemic.

¨l«q«bId ‘understand it.imp’ u¨b«tSalIkW ‘someone laid this down’

l«q«¨bId ‘after’ ub«¨tSalIkW ‘someone bet’

Snyder, W.A. (1968). Southern Puget Sound Salish: Phonology and Morphology. Sacramento: The Sacramento Anthropological Society, Sacramento State College.

North American languages

761

Spokane [LEX] Salishan, Interior Salish, Southern. Washington state (US). f f

Stress is morphologically determined. There are weak and strong roots and su‰xes that determine where the stresses fall.

¨kWI.n.¨cut

‘he made himself ’

s&l.n.¨texW ‘you chopped it’

Carlson, B.F. (1989). Reduplication and Stress in Spokane. IJAL 55: 2, 204–213. Squamish [P] Salishan, Central Salish, Squamish. British Columbia (Canada). f f

Morphologically simple forms have penultimate stress. In more complex constructions sequences of stresses, some of which are adjacent, occur.

¨t/aqwal/ ‘dry’

qwu¨layus

‘high tide’

Demers, R.A. & G.M. Horn (1978). Stress Assignment in Squamish. IJAL 44: 3, 180–191. Siouan Dakota; Sioux [S] Siouan, Siouan Proper, Central, Mississippi Valley, Dakota. Minnesota, North and South Dakota (US). f

Stress falls on the second syllable, and on the only syllable of monosyllabic words.

¨kte

‘he kills’

wa ¨kte

‘i kill’

ma ¨ya kte ‘you kill me’ o ¨wicha ya kte ‘you kill them there’ wi¨c ha ya kte ‘you kill them’ Shaw, P.A. (1985). Coexisting and competing stress rules in Stoney Dakota. IJAL 51: 1, 1–18. Shaw, P.A. (1985). Modularisation and substantive constraints in Dakota lexical phonology. Phonology 2, 173–202. Stoney Dakota (dialect of Dakota; Sioux) [S;U/P] Siouan, Siouan Proper, Central, Mississippi Valley, Dakota. Montana (US), Alberta, Saskatchewan (Canada).

762 f

North American languages

The placement of stress in Stoney Dakota indicates an ongoing development from the basic Dakota pattern of second syllable stress to an emerging pattern of final or penultimate stress. Stress the final syllable when it is closed by two consonants (no examples), otherwise the penult.

ma¨yakte ‘you kill me’

wa¨kte

‘i kill’

a¨kida

akida¨biktac&

‘they will look at it’

‘look at it’

Shaw, P.A. (1985). Coexisting and competing stress rules in Stoney Dakota. IJAL 51: 1, 1–18. Hoca´k; Winnebago [T] Siouan, Siouan Proper, Central, Mississippi Valley, Winnebago. Nebraska, Wisconsin (US). f

f

f

Primary stress is final in disyllables and on the third vowel in longer words. So, if one of the two first syllables contains a long vowel, stress falls on the second syllable.1 Secondary stress is located on every second vowel after the primary stress.

wa¨je ‘dress’ kiri¨ina ‘returned’

waghi¨ghi ‘ball’ haa¨kitu˙jiksha˙na ‘he pulls it out.decl’

Hale, K. & J. White Eagle (1980). A Preliminary Metrical Account of Winnebago Accent. IJAL 46: 117–132. Miner, K.L. (1979). Dorsey’s Law in Winnebago – Chiwere and Winnebago Accent. IJAL 45: 25–33. Miner, K.L. (1981). Metrics, or Winnebago made Harder. IJAL 47: 340– 342. Oto; Chiwere (dialect of Iowa-Oto) [I/S] Siouan, Siouan Proper, Central, Mississippi Valley, Chiwere. Iowa, Minnesota, Missouri, Oklahoma, Kansas, Nebraska (US). f

f

Stress the first syllable if it has a long vowel, otherwise the second (historical). Long vowel distinction has been lost, so now the pattern is lexicalised. 1. Hale and White Eagle claim that long vowels are disyllabic in Winnebago. Hence, in their view, stress is always located on the third syllable. This view forms the basis of our encoding. QS alternatives are possible.

North American languages

wa¨sose ‘brave’

wa¨naxe ‘spirit’

¨kiDa

¨loha7

‘fight’

763

‘much many’

Whitman, W. (1947). Descriptive Grammar of Ioway-Oto. IJAL 13: 233– 248. Miner, K.L. (1979). Dorsey’s Law in Winnebago-Chiwere and Winnebago Accent. IJAL 45: 25–33. Uto-Aztecan Cahuilla, Desert and Mountain (dialects of Cahuilla) [I] Uto-Aztecan, Northern Uto-Aztecan, Takic, Cupan, Cahuilla-Cupeno. Around Palm Springs, California (US). f f

f f f

Primary stress regularly falls on the first syllable of the root. Secondary stress on alternating syllables preceding and following the main stress. Secondary stress is sensitive to weight of long vowels. A syllable preceding an inserted /// always bears stress. If a secondary stress would fall on a voiceless vowel, according to the rules above, it shifts to the preceding vowel, even if that means it ends up adjacent to another stress.

¨n)a/a˙c&eh ˙papen¨tule˙qale˙veh ¨qaan˙kitSem

‘sit down’ ‘where i was grinding it’ ‘palo verde pl.’

¨pukaw˙temih

‘gopher snakes.obj.pl’

Seiler, H. (1965). Accent and morphophonemics in Cahuilla and in UzoAztecan. IJAL 31/1. Seiler, H. (1967). Structure and Reconstruction in some Uzo-Aztecan Languages. IJAL 33/2. Seiler, H. (1972). Cahuilla Grammar. Banning: Malki Museum Press. Comanche [I] Uto-Aztecan, Northern Uto-Aztecan, Numic, Central. Western Oklahoma (US). f

Primary stress falls on the first syllable of the word (with some exceptions).

764 f

f

North American languages

Secondary stress on alternating syllables following the main stress (not final). Secondary stress is reportedly sensitive to weight of long vowels and diphthongs.

¨yçtsçn ‘he got up’ ¨puhiwI ‘money’

¨waha˙tima˙to/i˙katç

‘twelve’

Charney, J.O. (1993). A grammar of Comanche. Lincoln: University of Nebraska Press. Osborn, H. and W.A. Smalley. (1949). Formulae for Comanche stem and word formation. IJAL 15: 93–99. Cupen˜o [I;LEX] Uto-Aztecan, Northern Uto-Aztecan, Takic, Cupan, Cahuilla-Cupeno. Southern California (US). f f

Primary stress regularly falls on the first syllable of the word. In some words a vowel other than the first used to be long and bear main stress. This vowel length has been reduced in Cupen˜o but the stress remains on the original syllable. In such cases stress is lexical.

¨naq« l

‘ear’

k«¨r« t ‘bird’

Munro, P. (1990). Stress and Vowel Length in Cupan Absolute Nouns. IJAL 56: 2, 217–250. Hopi [I/S] Uto-Aztecan, Northern Uto-Aztecan, Hopi. Northeast Arizona (US). f

f f f f

Primary stress falls on the first syllable if it is closed or contains a long vowel. Otherwise stress falls on the second syllable. Stress in bisyllables is always initial. Stress is realised as high tone. Secondary stress reported but its patterning is poorly understood.

¨naatIhota ‘to hurt oneself ’ ¨sipmasmi ‘silver bracelet’

ka¨wayom ‘horses’ qo_¨to_sompi ‘headband’

¨koho ‘wood’ ¨hop" $it ‘hopis’

The Hopi Dictionary Project (1998). Hopi Dictionary. Tucson: The University of Arizona Press. Jeanne, L. (1978). Aspects of Hopi Grammar. Unpublished Doctoral Dissertation, MIT, Cambridge, Massachusetts.

North American languages

765

Kalectaca, M. (1978). Lessons in Hopi. Tucson: University of Arizona Press. Seaman, P.D. (1985). Hopi dictionary. Dept of Anthropology, Northern Arizona University, Flagsta¤. Whorf, B.L. (1946). The Hopi language, Toreva dialect. In Cornelius Osgood (ed.) Linguistic Structures of Native America. Viking Fund Publications in Anthropology 6, pp. 158–183. Kawaiisu [U/P] Uto-Aztecan, Northern Uto-Aztecan, Numic, Southern. California (extinct?). f f

Primary stress falls on the final syllable if it contains a long vowel. Otherwise stress falls on the penultimate syllable.

/a¨taavi ‘tree worm’

/ikç¨vadç

paa¨yaa

ka/a¨pç.ni ‘my food’

‘surface, front, wall’

‘finger’

Zigmund, M.L., C.G. Booth & P. Munro (1990). Kawaiisu: A Grammar and Dictionary. University of California Publications in Linguistics 119. Berkely & Los Angeles: University of California Press. Luisen˜o [?/I;?/S] Uto-Aztecan, Northern Uto-Aztecan, Takic, Cupan, Luiseno. Pala, California (US). f

f

If there is a long vowel in either the first or the second syllable, it is stressed. Otherwise stress falls on the first syllable in verbs and the second in nouns.

¨huønal

‘badger’

pu¨huø ‘his arrow’

qa¨ruøt

‘bird’

Kroeber, A.L. & G.W. Grace (1960). The Sparkman Grammar of Luiseno. University of California Publications in Linguistics 16. Munro, P. (1990). Stress and Vowel Length in Cupan Absolute Nouns. IJAL 56: 2, 217–250. Nahuatl, Tetelcingo [P] Uto-Aztecan, Southern Uto-Aztecan, Aztecan, Gen. Aztec, Aztec. Tetelcingo (Mexico).

766 f

North American languages

Stress occurs on the penultimate syllable.

¨kaktli ‘sandal’

ka¨laki

‘enter’

a¨yohtli ‘squash’

Pittman, R.S. (1954). A Grammar of Tetelcingo (Morelos) Nahuatl. Language. Volume 30 supplement. Dissertation 50. Paiute, Southern [S] Uto-Aztecan, Northern Uto-Aztecan, Numic, Western. Western Nevada (US). f

f

f

Primary stress occurs on the second vowel, but is initial in bisyllabic words. Secondary stresses fall on alternate vowels thereafter, but not on the last syllable. Long vowels are disyllabic.

¨kani ‘house’ kWa¨axWi ‘grass seed’ a¨c&akki ‘to fall’ o¨ta.api ‘pole’ c&a¨ka.i˙cini ‘my younger brother’ tç. xw¨i .i ˙na. tç ˙Bicu xw˙a i ˙/i Nwa ‘go and ask him to tell a story’ Harms, R.T. (1966). Stress, voice and length in Southern Paiute. IJAL 32, 228–235. Sapir, E. (1930). Southern Paiute, A Shoshonean Language. In Proceedings of the American Society of Arts and Sciences, Vol. 65, No. 1–3. Shoshone, Panamint; Koso (dialect of Shoshoni) [I/I] Uto-Aztecan, Northern Uto-Aztecan, Numic, Central. Eastern California (US). f

f

f

Primary stress occurs on the second syllable if it contains a long vowel and the first does not. In all other cases stress is initial. Secondary stresses fall on alternate syllables thereafter, respecting vowel length. Secondary stress on final syllables is optional, the vowel is devoiced if unstressed.

kuk¨kwiippu_h ‘smoke’ tsi¨too˙hin ‘push’ pih¨naa˙witu_n ‘bee’ ¨nattu˙su/ung˙kantu_n ‘doctor’ ¨tukum˙mahan˙ningku_n˙na ‘cook for’

North American languages

767

Dayley, J.P. (1989). Tu¨mpisa (Panamint) Shoshone Grammar. University of California Publications in Linguistics 115. Berkeley: University of California. Dayley, J.P. (1989). Tu¨mpisa (Panamint) Shoshone Dictionary. University of California Publications in Linguistics 115. Berkeley: University of California Press. Tepehuan, Southeastern (dialect of Tepehuan) [I/I] Uto-Aztecan, Southern Uto-Aztecan, Sonoran, Tepiman, Southern Tepehuan. Chihuahua, Durango, Nayarit, Sonoro, Jalisco (Mexico). f

f

f

Primary stress occurs on the second syllable of the stem if it contains a long vowel and the first does not. In all other cases stress is initial. Secondary stresses on alternate syllables, respecting vowel length, reported. Prefixes do not count for stress assignment.

¨vaavaS

‘pheasant’

hir ko¨maarak ‘it is wide’

ha¨hoon)i/

‘wives’

hir ¨koo/kmarak

‘they are wide’

Kim, J.S. (1988). Lexical Phonology and Morphology of Southeastern Tepehuan. Master’s Report, University of Texas at Austin. Willet, E. (1982). Reduplication and Accent in Southeastern Tepehuan. IJAL 48: 2, 168–184. Tu¨batulabal [U (NMS)] Uto-Aztecan, Northern Uto-Aztecan, Tubatulabal. Kern River, California (US). f f

f

Stress falls on final vowels and long vowels. Stress also falls on short vowels situated two syllables to the left of a stress. All stresses are equally prominent.

¨taa¨hawi¨laap pi¨tipi¨tiidi¨nat ¨tciNi¨yal

‘in the summer’ ‘he is turning it over repeatedly’ ‘the red thistle’

¨yuøu¨duø¨yuøu¨dat

‘the fruit is mashing’

Voegelin, C.F. (1935). Tu¨batulabal Grammar. Berkeley: University of California Press.

768

North American languages

Yaqui [I;S] Uto-Aztecan, Southern Uto-Aztecan, Sonoran, Cahita. Mexico. f f f

A stress usually falls on the first or second syllable. Stress also falls on long vowels and closed syllables. Primary/secondary relation between these is unclear.

ba¨ka/apo

‘kind of tree’

¨taahi¨bepne ‘hockey’

¨baka¨noora

‘alcoholic maguey drink’

¨si/ita¨rokka

‘green beard of maiz’

Johnson, J.B. (1962). El idioma Yaqui. Departamento de Investigaciones Atropologicas, Publicaciones 10. Mexico: Instituto Nacional de Atropologica e Historia. Wakashan Kwakiutl; Kwakw’ala [F/L] Wakashan, Northern. Coast of British Columbia and end of Vancouver Island (Canada). f

Stress leftmost heavy syllable, otherwise the rightmost syllable. Heavy syllables are those with long vowels or closed by a sonorant consonant. Bach, E. (1975). Long vowels and stress in Kwakiutl. Texas Linguistic Forum 2, 9–19. Boas, F. (1947). Kwakiutl Grammar with a Glossary of the Su‰xes. Transactions of the American Philosophical Society, N.S. 37/3, 201–377. f

Makah; Kwe-Nee-Chee-Aht; Kweedishchaaht [I/S] Wakashan, Southern. Vancouver Island (Canada), Washington (US). f

If the first vowel is long it receives stress, else stress is on the second syllable.

¨t´uøyas ‘dry place on ground’ du¨duøk ‘to sing’

t´u¨yas ‘dry place on beach’ ¨baødapat´ ‘to practice’

Jacobsen, W.H. (1979). First lessons in Makah. Makah Language Program. Makah Cultural and Research Center, Washington. Isolates Kutenai [P;U] Language Isolate. Idaho, Washington (US), Montana, B. Columbia, Alberta (Canada).

North American languages f f

f

769

Primary stress is penultimate. Many exceptions with final stress. Some words have distinctive secondary stress (˙˙) on any syllable preceding main stress. Automatic secondary stress occurs on every second syllable preceding the distinctive secondary stress, or in case there is none, the primary stress. Syllables in between the distinctive secondary stress and the primary stress are unstressed.

¨sa/npa´ ˙nIs a˙kIk´u¨namIs ˙kqaqa¨na´kqac

‘but indeed’ ‘the town obv.’ ‘automobile’

huc˙˙/IsnI´qan¨qamek ‘I will sit there myself ’ Garvin, P.L. (1948). Kutenai I: Phonemics. IJAL 14: 1, 37–43. Mocama [I] Language Isolate. Florida (US). f f

Primary stress is initial. Words of more than three syllables get a penultimate or antepenultimate secondary stress.

¨yuwu ‘stone’ ¨ati˙muku ‘frost’

¨nipita ‘mouth’ ¨tira˙kukula ‘best’

Granberry, J. (1956). Timucua I: Prosodics and phonemics of the Mocama dialect. IJAL 22: 2, 97–105. Yuchi [LEX] Language Isolate. Among the Cree people in east central Oklahoma (US). f f

Primary stress is lexically determined, but usually final in bisyllables. Secondary stress exists and is also unpredictable.

y)c¨pi

‘walnut’

¨caya

‘squirrel’

ca¨ya ‘weeds’ ta¨hah)˙n) ‘the older one’ h)wEta˙nE¨cE)dji ‘the road that he used to go’ Wagner, G. (1933–1938). Yuchi. In Franz Boas (ed.) Handbook of American Indian Languages, Part 3. Bureau of American Ethnology, Bulletin 40, Part 3. New York: Columbia University Press. Also published 1934 as separate monograph.

770

North American languages

Zuni; zun˜i [I] Language Isolate. New Mexico, south of Gallup (US). f

Stress always occurs on the first syllable.

¨s&iwani

‘priest’

¨telikina ‘prayer stick’

¨SontSi

‘claw, fingernail’

¨melika ‘white American’

Bunzel, R. (1934). Zuni. In F. Boas (ed.) Handbook of American Indian Languages 3, 383–515. New York: Columbia University Press. Miner, K.L. (1986). Noun Stripping and Losse Incorporation in Zuni. IJAL 52: 242–254.

Middle American languages [For Nahuatl, Tepehuan and Yaqui, belonging to the Uto-Aztecan family, see the North America section] Mayan Aguacateco [L/L] Mayan, Quichean-Mamean, Greater Mamean, Ixilan. Guatemala. f

Stress on the last long vowel, or on the last syllable in case there are no long vowels.

¨/aøk’ah

‘new’

/u¨mul

‘rabbit’

McArthur, H. & L. McArthur (1956). Aguacatec (Mayan) Phonemes in the Stress Group. IJAL 22, 72–76. Huasteco [L/F] Mayan, Huastecan. Vera Cruz, San Luis Potosi, Oaxaca (Mexico). f

Stress on the last long vowel, or on the first syllable in case there are no long vowels.

Larsen, R.S. & E.V. Pike (1949). Huasteco Intonations and Phonemes. Language 25: 268–277. Jacalteco [I] Mayan, Kanjobalan-Chujean, Kanjobalan, Kanjobal-Jacaltec. Huehuetenango (Guatemala). f

Stress falls on the first syllable of the stem.

¨tSeh ‘horse’

¨tohol ‘straight’

¨nahata ‘distant’

Day, C. (1973). The Jacaltec Language. Indiana University Publications, Language Science Monographs 12. Bloomington: Indiana University. Mam [L/P] Mayan, Quichean-Mamean, Greater Mamean, Mamean. Western Guatemala, Southern Chiapas (Mexico).

772 f

f

Middle American languages

Stress falls on the last long vowel, else on the last syllable closed by a glottal stop. If these rules do not apply, stress the vowel preceding the last consonant of the stem.

/a¨ˇuøntl ‘work’

waˇ ¨naøya ‘i worked’

§&pi¨tSaˇ

pu/¨la/

‘racoon’

‘dipper’

England, N. (1983). A Grammar of Mam, a Mayan Language. Austin: University of Texas Press. Pokomchı´; Poconchi [U] Mayan, Quichean-Mamean, Greater Quichean, Pocom. North Central Guatemala. f

Stress falls on the final syllable.

/e¨tal ‘to signal’

/au¨as

‘older brother’

ni¨naa ‘my head’

Mayers, M. (1960). The Phonemes of Pocomchi. Anthropological Linguistics 2/9, 1–39. Quiche´ [U] Mayan, Quichean-Mamean, Greater Quichean, Quichean, Quiche-Achi. Guatemala, Mexico. f

Stress falls on the final syllable.

wixo¨kil ‘my wife’

cho¨mak ‘fat ones’

Mayers, M. ed. (1966). The Languages of Guatemala. Janua Linguarum, Series Practica 23, The Hague: Mouton. Tzotzil, Zinacanteco (dialect of Tzotzil) [I] Mayan, Cholan-Tzeltalan, Tzeltalan. Zinacantan village, Chiapas Highlands (Mexico). f

Stress falls on the first syllable of the root.

¨xvabaxm

‘musician’

¨ats/is ‘you sewed it’

¨kok ‘my leg’

J. Haviland, S. Robinson & E. Gutierrez. An On-line Tzotzil Grammar. http://www.zapata.org/Tzotzil

Middle American languages

773

Misumalpan Mı´skito [I] Misumalpan. Nicaragua, Honduras. f

Stress falls on the first syllable of the word.

¨wal ‘two’

¨utla ‘house’

¨lawana ‘song’

No author (1985). Miskitu Bila Aisanka. Managua: Centro de investigaciones y Documentacio´n de la costa Atla´ntica. Mixe-Zoque Zoque, Copainala´ [P] Mixe-Zoque, Zoque, Chiapas Zoque. Copainala´, Northern Chiapas (Mexico). f f

Primary stress falls on the penultimate syllable of the stress group. Secondary stress on the first syllable if there are three or more syllables.

˙min¨ke/tpa ˙minsukke/tpa¨/itti

‘he is coming again’ ‘they were going to come again’

Knudson, L.M. (1975). A Natural Phonology and Morphophonemics of Chimalapa Zoque. Papers in Linguistics 8, 283–346. Wonderly, W.L. (1951). Zoque II: phonemes and morphophonemics. IJAL 17: 105–123. Oto-Manguean Chatino, Yaitepec (dialect of Chatino) [U] Oto-Manguean, Zapotecan, Chatino. Oaxaca (Mexico). f

Stress falls on ultimate syllable.

¨slya/

‘cotton’

ti¨/a

‘water’

kwi/¨ya

‘eagle’

ts&i¨kwi/

‘to talk’

Upson, B.W. & R.E. Longacre (1965). Proto-Chatino Phonology. IJAL 31: 1, 312–322. Chinanteco, Lealao [LEX] Oto-Manguean, Chinantecan. North Oaxaca, San Juan Lealao, Latani, and Tres Rı´os (Mexico).

774 f

Middle American languages

Stress is lexically distinctive. Words can be stressless. There is tone (not indicated).

Niø

‘her/his voice’

¨Niø ‘high’

fi

‘road’

¨fi ‘handle’

Rupp, J.E. (1989). Lealao Chinantec Syntax. Studies in Chinantec Languages 2. Summer Institute of Linguistics and the University of Texas at Arlington. Isolates Pure´pecha; Tarasco; Tarascan [I;S] Language Isolate. Michoaca´n (Mexico). f

Stress is initial or on the second syllable.

¨patani

‘to extinguish the fire’

ka¨rani ‘to write’

xa¨potSHaakani ‘i will wash my head’ Foster, M.L. (1969). The Tarascan Language. University of California Publications in Linguistics 56. Tol; Jicaque [U/P] Language Isolate. Honduras. f

Stress is usually final if the last syllable is closed, otherwise stress is penultimate.

¨bele ‘he talks’ naNku¨pwepe ‘left’

na¨c/om ‘my wife’ phes¨mas ‘skunk’

Fleming, I. & R.K. Dennis (1977). Tol (Jicaque): Phonology. IJAL 43: 2, 121–127. Seri [F/F] Hokan, Salinan-Seri. Mexico, Gulf of California. f f f

Stress falls on the first complex nucleus of the root, else on the first. Other complex nuclei bear secondary stress. See page 290 for [U/P] analysis based on the more recent source.

¨xpeyo ‘sailfish’

xa¨poø ‘sea lion’

la¨Xiøktim ‘placename’

Marlett, S.A. (1988). The Syllable Structure of Seri. IJAL 54: 3, 245–278. Marlett, S.A. (2008). Stress, extrametricality and the minimal word in Seri. Linguistic Discovery 6: 1, 1–14.

South American languages

Araucanian Mapuche; Araucanian; Aucan [S] Araucanian. Chile and Argentina. f f

Primary stress falls on the second syllable. Secondary stress falls on alternate syllables thereafter.

wu¨le ‘tomorrow’ e¨lumu˙yu ‘give us’

t.i¨panto ‘year’ ki¨mufa˙luwu˙lay ‘he pretended not to know’

Echeverria, M.S. & H. Contreras (1965). Araucanian Phonemics. 3I: IJAL 31, pp. 132–135. Arawakan Apurina˜ [U/P] Arawakan, Maipuran, Southern Maipuran, Purus. Purus River from Rio Branco to Manaus. Amazon region, Brazil. f

f

Primary stress falls on the final syllable if it contains a nasal vowel, else on the penult. Secondary stress falls on alternate syllables to the left of the main stress.

¨taka ‘put’ ç¨a) ‘tooth of ’ a˙na)na¨pe) ‘pineapple juice’

¨pa)ma pa¨taro a˙nçrç˙ma)nE¨kata

‘pama fruit’ ‘chicken’ ‘with our relatives’

Facundes, S. (2000). A grammar of Apurina. Ph.D. Dissertation. SUNY at Bu¤alo. Banawa´ [L (CNT)] Arawakan, Arauan. Southern Amazone, Brazil. f

f

Secondary stress on alternate syllables starting at the left edge and skipping initial vowels. The last secondary stress is promoted to primary stress.

˙su.i¨ri ‘penis’ ˙wana¨kori ‘spider’

o˙wari.¨a ‘one’ ˙yeye˙roka¨na ‘crank’

776

South American languages

Buller, B. & E. Buller (1986). Banawa´ phonemic statement. Ms., Summer Institute of Linguistics, Porto Velho, Brazil. Campa, Pajonal Ashe´ninca; Atiri [LEX;NMS] Arawakan, Maipuran, Southern Maipuran, Campa. Peru, Central Gran Pajonal area and western Ucayali River. f

Primary stress falls on multiple syllables in an unpredictable fashion.

¨sam¨pakItI ‘a type of bird’ ¨ma¨here¨tatSIrI ‘the one who is sitting quietly’ ¨so¨ro¨me/ ‘a type of caterpillar’ Pike, E.V. & W. Kindberg (1956). A Problem in Multiple Stresses. Word 12, pp. 415–428. Paumarı´ [A] Arawakan, Arauan. Brazil, Amazon. f

Stress falls on the antepenultimate syllable.

¨xooni ‘younger brother’ ¨topava ‘shell’

¨joma ‘night’ oni¨manari ‘seagull’

Chapman, S. & D. Derbyshire (1991). Paumari. In Desmond C. Derbyshire & Geo¤rey K. Pullum. Handbook of Amazonian languages 3, 161–352. Berlin: Mouton. Piro [P] Arawakan, Maipuran, Southern Maipuran, Purus. Peru. f f f

Primary stress falls on the penultimate syllable. Secondary stress falls on the first syllable. Tertiary stress on other odd syllables in long words, but not adjacent to the main stress.

¨walo ˙kaxruu˙kakhi˙manatat¨kana

‘rabbit’ ‘they were joking together, it is said’

Matteson, E. (1965). The Piro (Arawakan) Language. Berkeley and LA: University of California Press. Suriname Arawak (dialect of Arawak, Lokono) [I;IRR] Arawakan, Maipuran, Northern Maipuran, Caribbean. Surinam.

South American languages f

f

777

Word stress is initial, but may fluctuate though interference of phrasal stress and rhythm. Phrasal stress falls on the penultimate syllable, rhythm on every other syllable leftward.

ba˙hç lo¨ko7ro& or ˙bahç lo¨ko7ro&

‘into the house’

da ˙Sima¨kabo

‘I am calling’

Pet, W.J.A. (1979). Another look at Suriname Arawak phonology. IJAL 45: 4, 323–331. Aymaran Aymara [P] Aymaran. Bolivia, Peru, Chile. f

Stress falls on the penultimate syllable.

ku¨laka ¨qala

‘sister’ ‘stone’

kula¨kaxa ‘my sister’ a¨maya ‘spirit’

Hardman-de-Bautista, M.J., J. Vasquez, & J. de Dios Yapita. (1988). Aymara, compendio de estructura fonologica y gramatical. Gainesville: Aymara Foundation. Jaqaru [P] Aymaran. Lima Dept, Yauyos Province, Tupe village and Cachuy village. f f

Primary stress falls on the penultimate syllable. Secondary stress falls on the syllable immediately to the left of the main stress.

¨mac&a ‘to irrigate’

˙waø¨raha ‘star’

˙mus¨c&atHa ‘slowly’

Hardman, M.J. (1966). Jaqaru: Outline of phonological and morphological structure. The Hague, Mouton. Carib Carib; Kalihna [. . ./L] Carib, Northern, Galibi. Surinam, Guyana, Venezuela, Brazil. f

Primary stress falls on a long vowel, diphthong or VCC sequence that is not the first of such syllables in the word. If there are no such syllables, or there is only one, stress is final.

778

South American languages

tonoø¨ro

‘large bird’

karaxsa¨wa kaøra¨waøsi

‘fish species’ ‘rattle sp. of tree’

iweixtoøko¨nimbo ‘their former [way of ] being’ Ho¤, B.J. (1968). The Carib Language: Phonology, Morphology, Texts and Word Index. The Hague: M. Nyho¤. Derbyshire, D. (1999). Carib. In B. Dixon & A. Aikhenwald (eds.) The Amazonian Languages, 23–64. Cambridge: CUP. Hixkarya´na [NMS] Carib, Southern, Southern Guiana. Amazonas, upper Nhamunda´ River to Mapuera and Jatapu´ Rivers. f f f

Stress falls on long vowels and closed syllables. Stress also falls on the even members of a string of light syllables. There is no di¤erence between these stresses, but there seems to be a final or penultimate pitch accent that is independent of the metrical structure.

mç¨haøna¨nçhno

‘you taught him’

¨owto¨hoøna ‘to the village’

Derbyshire, D.C. (1985). Hixkaryana and Linguistic Typology. SIL Publications in Linguistics 76. Summer Institute of Linguistics, Dallas. Derbyshire, D. (1979). Hixkaryana. Lingua Descriptive Studies 1. Amsterdam: North-Holland. Macushi [U (pitch)] Carib, Northern, East-West Guiana, Macushi-Kapon, Macushi. Guyana, Northern Brazil. f f

Primary stress falls on the final syllable. (tentative) Secondary stress on long vowels and even syllables counted from the left.

piri¨pi ‘spindle’

s&i/miri¨k" _ ‘little’

umanari¨r" _ ‘cassava grater’

Hawkins, W. (1950). Patterns of vowel loss in Macushi (Carib). International Journal of American Linguistics 16, pp. 87–90. Abbott, M. (1991). Macushi. In D.C. Derbyshire & G.K. Pullum (eds.) Handbook of Amazonian Languages 3, 23–160. Berlin: Mouton de Gruyter.

South American languages

779

Chapacura-Wanham Pakaa´snovos; Wari [U;P] Chapacura-Wanham, Madeira. Rondonia, Western Brazil. f f

Primary stress falls on the final syllable. Except when that syllable is a stress avoiding member of a verbal compound, in which case stress will go on the syllable to its immediate left.

to¨mi/ ‘speak’ pi¨ye/ ‘child’

¨kom ‘water’ tSi¨tot ‘garden’

Everett, D.L. & Barbara Kern (1997). Wari. London: Routledge. Chibchan Rama [NMS] Chibchan, Rama. Nicaragua. f

f

Long vowels and closed syllables are stressed, all stresses are equally strong. Sequences of two light syllables are stressed on the first (no examples).

¨kaat

‘tree’

ku¨maa ‘woman’

¨sula

‘deer’

a¨bung ‘fire’

¨tiiskama ‘child’

Grinevald, C. A grammar of Rama. Unpublished draft ms. Choco Embera Saija; Epera Pede´e [U/P;U] Choco, Embera, Southern. Southern Pacific coast, Colombia. f f f

In nouns and adjectives stress falls on final long vowels or diphthongs. Else stress falls on the penultimate syllable. Morphological factors may cause stress in verbs to be final.

pa¨nia ‘water’

a)i¨surra

wa)¨hi

kHo¨pata ‘eat.hab.pl’

‘go.pst’

‘monkey’

war¨raa ‘flavorful’ ¨taama

‘snake’

Harms, P.L. (1994). Epena Pedee Syntax. Studies in the Languages of Colombia 4. SIL and U. Texas at Arlington. Embera-Chami [U/P] Choco, Embera, Southern. Risaralda, Caldas, Antioqu’a, Valle, Colombia.

780 f

f

South American languages

Stress falls on the final syllable if that syllable has an autonomous stress mark. Else stress falls on the penultimate syllable.

¨we)ra

e¨ter

‘woman’

‘chicken’

wa)wa¨de ‘kindergarden’

Aguirre Licht, D. (1998). Embera. Languages of the World Materials 208. Munchen: Lincom Europa. Chon Tehuelche [I] Chon. Patagonia, Argentina. f

Stress falls on the initial syllable.

¨catenue ‘fork’

¨cone ‘grandmother’

¨yomeno

‘now, at once’

Schmid, T. (1910). Two linguistic treatises on the Patagonian or Tehuelche language. Buenos Aires: Coni Bros. Macro-Chibchan Shiriana; Ninam? (dialect of Waica) [U] Macro-Chibchan – Waican (Yanomam?). Braziland (Venezuela). f f

Primary stress falls on the final syllable. Secondary stress falls on alternate syllables to the left skipping single V syllables.

ka˙rihi¨ham

‘to the other side’

˙aoya¨/im

‘pumpkin’

Migliazza, E. & J.E. Grimes (1961). Shiriana Phonology. Anthropological Linguistics 3.6, pp. 31–41. Macro-Ge Canela-Krahoˆ [U] Macro-Ge, Ge-Kaingang, Ge, Northwest, Timbira. Maranha˜o, southeastern Para´. f

Stress falls on the final syllable.

¨po ‘deer’ ihmu¨tri ‘over there’

cu¨ra ‘kill’ ipica¨hur ‘run’

South American languages

781

Popjes, J. and J. Popjes (1986). Canela-Kraho. In Desmond C. Derbyshire & Geo¤rey K. Pullum, (eds.), Handbook of Amazonian Languages 1. Berlin: Mouton de Gruyter. Mataco Mataco; Wichı´ Lhamte´s [I] Mataco-Guaicuru, Mataco. Argentina. f f

Primary stress falls on the first syllable in the absence of su‰xes. When su‰xes are added primary stress is final and secondary stress is initial.

¨hinno

‘man’

˙hin¨no-l ‘men’

Vin˜as Urquiza, M.T. (1974). Lengua Mataca. Tomos 1 & 2. Buenos Aires: Centro de Estudios Linguisticos. Mura Mu´ra-Piraha˜; Piraha˜ [A-U/U] Mura. Brazil. f

f

Stress falls on one of the last three syllables according to the following hierarchy: CVV > GVV > VV > CV > GV (C ¼ voiceless, G ¼ voiced). In the event of ties, stress falls on the rightmost syllable.

¨kaøgai

‘word’

soioa@ga¨hai ‘thread’

/apa¨baøsi ‘square’ ¨/ibog"@ pao@hoa¨hai ‘anaconda, rainbow’

‘milk’

poø¨ga@ihia"@ ‘banana’ /aba¨pa

[proper name]

Everett, D.L. & K. Everett (1984). Syllable Onsets and Stress Placement in Piraha. Proceedings of the West Coast Conference on Formal Linguistics 3. Stanford Linguistics Association, Stanford, CA, 105–116. Everett, D.L. (1988). On metrical constituent structure in Piraha. Natural Language and Linguistic Theory 6, 207–246. Nambiquaran Nambiquara, Southwest; Nambikua´ra, Southern [U] Nambiquaran. Northwestern Mato Grosso, Rondonia (Brazil). f

Stress falls on the final syllable (variants exist).

782

South American languages

in ¨tsu ‘man’

huk ¨ksu ‘bow’

Price, P.D. (1976). Southern Nambiquara Phonology. IJAL 42: 4, 338–348. Panoan Capanahua [S/I] Panoan, North-Central. Peru, Tapiche-Buncuya rivers area. f

f f

Primary stress falls on the second syllable if it is closed by a non-/// coda. Else primary stress is initial. Secondary stress falls on alternate syllables after the main stress (no examples).

¨mapo

‘head’

¨c&ic&ika ‘knife’

pis&¨kap

‘small’

wi¨rankin

‘he pushed it’

Safir, K. (1979). Metrical structure in Capanahua. In K. Safir (ed.) MIT working papers in Linguistics, Vol. 1, pp. 95–114. Loos, E. (1969). The Phonology of Capanahua and its Grammatical Basis. Ph.D. Dissertation. SIL, University of Oklahoma publication 20. Shipibo-Conibo [S/I] Panoan, North-Central. Northeastern middle Ucayali River area, Peru. f

f

Stress falls on the second syllable if it is closed by a consonant or nasal vowel. Else stress is initial.

¨mapo ‘head’ ¨berochon ‘species of fish’

yan¨con ‘blue, green’ ma¨nans_hahue ‘hammer’

Faust, N. (1973). Lecciones para el aprendizaje del idioma Shipibo-Conibo. Instituto Linguistico de Verano. Quechuan Inga [U/P] Quechuan, Quechua II, B. Colombia. f

Primary stress falls on the final syllable if it has a sonorant coda, else on the penult.

South American languages f

783

Secondary stress falls on alternate syllables to the left of the main stress.

ya¨war

‘blood’

¨kanc&is ‘seven’

apa¨muy

‘to bring’

kam¨kuna

‘you pl.’

Levinsohn, S.H. (1976). The Inga Language. The Hague: Mouton. Quechua [P] Quechuan, Quechua I or II. Peru (Bolivia, Chile). f

Stress on the penultimate syllable, but it is final if a first person length marker is added.

al¨quø ‘my dog’

¨aks&u ‘potato’

Adelaar, W.F.H. (1984). Grammatical Vowel Length and the Classification of Quechua Verbs. IJAL 50: 1, 25–47. Quichua, Highland, Imbabura; Otavalo Quichua [P] Quechuan, Quechua II, B. Northern highlands, Imbabura Province (Ecuador). f

Stress falls on the penultimate syllable.

¨jana ‘think-inf’ yaju¨nata ‘think-prog-inf-acc’

ja¨nata ‘think-inf-acc’

Cole, P. (1982). Imbabura Quechua. Lingua Descriptive Studies 5. Amsterdam: North-Holland. Tacanan Cavinen˜a [P] Tacanan, Araona-Tacana, Cavinena-Tacana, Cavinena. Northern Bolivia, southeast of Riberalta, along the Beni River. f f

Primary stress falls on the penultimate syllable. Secondary stress falls on alternate syllables leftward.

ki¨rika ‘paper, book’

¨hida ‘good’

ma˙toha¨iwa ‘a cactuslike plant’

Key, M.R. (1968). Comparative Tacanan Phonology. The Hague: Mouton. Ese Ejja; Chama [P;U] Tacanan, Tiatinagua. Northwestern region Bolivia, Tambopata and Heath rivers in Peru.

784 f

f

South American languages

Primary stress falls on the penultimate syllable (lexical exceptions with final stress). Secondary stress falls on alternate syllables to the left (no examples).

¨ena ‘water’

e¨na

‘blood’

M.R. Key (1968). Comparative Tacanan Phonology. The Hague: Mouton. Tacana [P;A] Tacanan, Araona-Tacana, Cavinena-Tacana, Tacana Proper. Bolivia. f

f

Primary stress is generally on the penult of words without a‰xes spoken in isolation. Exceptions with antepenultimate stress reported (no polysyllabic examples provided in source).

¨bi ‘palm’

¨di ‘mosquito’

M. Key (1968). Comparative Tacanan Phonology. The Hague: Mouton. Tupi Guarani [U] Tupi, Tupi-Guarani, Guarani (I). Paraguay. f

Stress falls on the final syllable (only the conjunctive copula ‘ha’ is stressless).

gua¨ta ‘walk’

ky¨ra ‘fat’

pura¨hei ‘to sing’

Ayala, J.V. (1993). Gramatica Guarani. Paraguay. Kaiwa´ [U;IRR] Tupi, Tupi-Guarani, Guarani (I). Paraguay. f

Stress is final, but there are many exceptions depending on sentence rhythm.

ajepo¨ta ‘i light a fire’

eju¨vI

‘you strangle something’

Bridgeman, L. (1971). Kaiwa (Guarani) Phonology. IJAL 27: 4, 329–334. Tupi [P;A] Tupi, Tupi-Guarani, Tupi (III). Eastern Brazil. f

Stress is penultimate, and antepenultimate in some five syllable or longer words.

South American languages

¨it" )g

‘it is white’

tai¨titu ‘wild pig’

pati¨uapE

785

‘bark pan’

Abramson, A. (1968). Contrastive distribution of Phoneme Classes in Icua Tupi. Anthropological Linguistics 10.6, pp. 11–22. Urubu´-Kaapor [U] Tupi, Tupi-Guarani, Oyampi (VIII). Maranha˜o, Brazil. f f

Primary stress falls on the final syllable. Secondary stress falls on alternate syllables to the left of the main stress.

ta¨ta ‘fire’

˙war&u¨wa

‘glass, mirror’

mi˙r&ar&a¨/ir ‘small tree’

Kakumasu, J. (1986). Urubu-Kaapor. In Desmond C. Derbyshire & Geo¤rey K. Pullum, (eds.), Handbook of Amazonian Languages 1, pp. 326–403. Berlin: Mouton de Gruyter. Witotoan Huitoto, Nipode; Witoto; Huitoto, Miunane [I] Witotoan, Witoto, Witoto Proper, Nipode. Peru. f

Stress falls on the first syllable (with lexical exceptions).

¨jabe ‘now’

¨n)egana ‘work’

¨enenorie ‘separately’

Minor, E. & D. Hendrich de Minor (1971). Vocabulario Huitoto Muinane. Serie Linguistica Peruana 5. Instituto Linguistico de Verano. Minor, E. (1956). Witoto vowel clusters. IJAL 22: 131–137. Yanomam Sanuma´ [P] Yanomam. Auaris River, Roraima. f

f

Primary stress is initial in two and three syllable words, penultimate in longer words. Secondary stress on the initial syllable of words with penultimate stress.

¨nako_ ‘tooth’

¨ipoko

‘wind’

˙kapu¨lumo_

‘yam’

Borgman, D.M. (1990). Sanuma. In Desmond C. Derbyshire & Geo¤rey K. Pullum, (eds.), Handbook of Amazonian Languages 2, pp. 15–248. Berlin: Mouton de Gruyter.

786

South American languages

Isolates and Creoles Aukaans; Ndyuka [?/P] Creole, English based, Atlantic, Surinam, Djuka. Surinam, French Guyana. f

f f f f

If the first has a low tone and the second a high tone, both are prominent. If there is a long vowel or closed syllable in the word it carries the stress. Words with two heavy syllables have two equal stresses. In two syllable words of all light syllables, the first is stressed. In words with more than two light syllables, the penultimate is stressed.

¨ba@ka si¨do@N ko¨mo@to musoko¨so@ko

‘back’ ¨ba@aka ‘black’ ‘to sit’ ¨bansa ‘side’ ‘to come out’ ¨feemu@su ‘bat’ ‘bird of paradise flower’

ba¨ka@a ‘white person’ ¨pii¨si@i ‘pleasure’ sapa¨ka@a ‘lizard’

Huttar, G.L. & M.L. Huttar (1994). Ndyuka. London: Routledge. Cayubaba; Cayuvava [A (NMS)] Language Isolate. Beni Department, west of Mamore River, north of Santa Ana (Bolivia). f f

f

Mono- and bisyllables have initial stress. Else stress occurs on the antepenultimate syllable and every third syllable preceding it. Dispute over whether the antepenult and the rest of the stresses have di¤erent strength.

¨daru ‘hand’ ari¨kajahi ‘he has already fallen’

¨sakahe ‘stomach’ iki¨tapara¨repeha ‘the water is clean’

Key, H. (1961). The Phonotactics of Cayuvava. IJAL 27: 2, pp. 143–150. Key, H. (1963). Morphology of Cayuvava. Ph.D. Dissertation, University of Texas Austin. Key, H. (1967). Morphology of Cayuvava. The Hague: Mouton. Key, H. and M. Key (1967). Bolivian Indian Tribes, Mexico, Summer Inst. of Linguistics. Trumai [U] Language Isolate. Xingu´ Park, source of Xingu´ River, Mato Grosso (Brazil).

South American languages f

787

Stress always falls on the last syllable of the word.

t5s" $¨pu@p

‘firewood’

¨kº

p" $¨kE@

‘house’

wa$ru$/¨t5a@ ‘pumpkin’

‘wax’

Guirardello, R. (1999). A reference grammar of Trumai. Ph.D. Thesis, Rice University, Houston, Texas. Waorani; Auca; Sabela [NMS (I)] Language Isolate. Northern Equador. f f

f

On the stem, stress every odd syllable from the left. On the sequence of su‰xes, stress every even syllable from the right until the stem or the fourth syllable is reached. No di¤erence in prominence between the stresses.

¨go¨tamo¨napa ‘we two went’ ¨go¨kQdomo¨naimba ‘we two would have gone’ ¨kQga¨kamba ¨e¨Nakan¨dapa

‘his tooth hurts’ ‘he was born’

Pike, K. (1964). Stress trains in Auca. In D. Abercrombie (ed.) In honour of Daniel Jones. Longman: London. Warao; Guarao [P] Language Isolate. Venezuela. f f

Primary stress falls on the penultimate syllable. Secondary stress on every alternate syllable before the main stress.

¨tira

‘woman’

ko¨ranu

‘drink it’

n" )˙ha)ra¨paka

‘alligator’

˙yapu˙ruki˙tane¨hase

‘verity to climb’

Osborn, H. (1966). Warao I: Phonology and Morphophonemics. IJAL 32, pp. 108–123. Romero-Figeroa, A. (1997). A reference grammar of Warao. Studies in Native American Linguistics 6. Mu¨nchen: Lincom Europa.

Eurasian languages*

Basque Genetic a‰liations given here are from Hualde (1999). Ethnologue does not distinguish between all the dialects we present here. Basque, Gernica [LEX (pitch)] Basque, Western Basque, Northern Biscayan Basque. Northern Biscay (Spain). f

f f

f f f

There is a fundamental distinction between unaccented and accented words. Unaccented words receive final prominence only at the phrase level. Accented words present a rise in pitch culminating on the accented syllable, which is followed by a sharp drop on the following syllable. Accent is lexically determined, but never falls on the final syllable. The leftmost accent determines the location of the prominent syllable. Su‰xes can be inherently accented, or pre-accenting.

ka@ npotarak ‘the outsiders’ pelotaRi or pelotaR"@ ‘ball player’

xixo@naRi

‘to the men’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basaburua and Imoz [LEX-I;S] Basque, Navarrese, Southern High Navarrese. Navarre (Spain). f

f

In the singular the primary stress falls on the first or second syllable and a secondary stress on the last syllable. In the plural, on the other hand, primary stress falls on the syllable that contains the plural su‰x, which may be the last or the penultimate syllable of the word.

¨xiDo˙nak ‘the man erg.’ e¨maku˙mek ‘the woman erg.’

xiDo¨nak ‘the men abs.erg.’ emaku¨mek ‘the women abs.erg.’

* This section includes all the languages spoken in Europe as well as the languages spoken in India that belong to the Proto-Indo-European language family.

Eurasian languages

789

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Ibarra, O. (1994). Ultzamako euskararen azterketa. Ph.D. dissertation, University of the Basque Country, Vitoria-Gasteiz. Basque, Bidasoa Valley [S] Basque, Central Basque. Border area between Gipuzkoa and Navarre (Spain). f f

Stress normally falls on the second syllable of the stem. Monosyllables are stressed.

inpor¨tantzi ‘importance’

¨lur etik ‘from the land’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Northern High Navarrese [P;U (IRR)] Basque, Navarrese. Navarre (Spain). f f

Stress normally falls on the penultimate syllable. Uninflected consonant-final words may optionally receive final stress.

¨etxja ‘the house’ Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Ibarra, O. (1994). Ultzamako euskararen azterketa. Ph.D. dissertation, University of the Basque Country, Vitoria-Gasteiz. Michelina, L. (1967). Notas fonolo´gicas sobre el salacenco. Anuario del Seminario de Filologı´a Vasca Julio de Urquijo 1: 163–177. (Repr. in Michelena, L. (1987). Palabros y textos, 221–234. Leioa: Universidad de Paı´s Vasco) Michelina, L. (1972). A Note on Old Labourdin Accentuation. Annuario del Seminario de Filologı´a Vasca Julio de Urquijo 6: 110–120. (Repr. in Michelena, L. (1987). Palabros y textos, 110–120. Leioa: Universidad de Paı´s Vasco)

790

Eurasian languages

Michelina, L. (1976). Accentuacio´n alto-Navarra. Fontes Linguisticae Vasconum 23: 147–162. (Repr. in Michelena, L. (1987). Palabros y textos, 245–260. Leioa: Universidad de Paı´s Vasco) Salaburu Etxeberria, P. (1984). Arau fonologikoak. Bilbao: Universidad del Paı´s Vasco. Basque, Hondarribia [U/P] Basque. Border area between Gipuzkoa and Navarre (Spain). f f f

f

If the stem is monosyllabic stress falls on its only syllable. If the stem is bisyllabic stress falls on the second syllable. If the stem has three or more syllables and ends with a vowel stress falls on the penultimate syllable of the stem. Else on the last syllable. Inflectional su‰xes do not have any relevance for the assignment of stress.

¨ar a

‘the worm abs.’

ka¨to

‘cat’

txistu¨lari ja

‘the ugly one abs.’

irabaz¨tun

‘the winner’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Lekeitio [P (pitch)] Basque, Western Basque, Northern Biscayan Basque. Northern Biscay (Spain). f

f

f

f

There is a fundamental distinction between unaccented and accented words. Most words are unaccented and receive final prominence only at the phrase level. Accented words present a rise in pitch culminating on the accented syllable, which is followed by a sharp drop on the following syllable. Accent always falls on the penultimate syllable.

sagara ‘apple abs.’ sagare@tik ‘apple abl.’

saga@ rak ‘apples abs.’ sagareta@ tik ‘apples abl.’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter.

Eurasian languages

791

Basque, On˜ati [S;I] Basque, Central Basque. Gipuzkoa (Spain). f f

f

f

Monosyllables have stress. In bisyllabic stems stress falls on the initial syllable of the uninflected form and the second of longer, inflected forms. Some bisyllables always have initial stress. Longer words are stressed on the second syllable of the stem, su‰xes have no influence. Accent always falls on the penultimate syllable.

a¨bare ‘priest’

¨xiTon ‘man’

xi¨Tona

‘man abs.’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Roncalese [P;U;A] Basque, Eastern Basque. Navarre (Spain). f f f

f

Uninflected forms may be stressed on any of the last three syllables. Penultimate stress is the most common option. Inflectional su‰xes do not influence stress placement, but may receive a secondary stress. Plural su‰xes other than the absolutive attract stress to themselves.

mar¨DuDa ‘strawberry’

ama¨łi ‘grandmother’

¨Damari ‘horse’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Sakana [I;S] Basque, Navarrese, Southern High Navarrese. Navarre (Spain). f f f

f

Stress generally falls (noncontrastively) on the initial syllable. A secondary stress falls on the last syllable of the phrase. In words of three or more syllables the primary stress may fall on either the first or the second syllable. Unclear whether stress variation is rhythmical or lexical (perhaps phrasal accent).

¨kari˙ka ‘street’

¨arbizua˙ra or ar¨bizua˙ra ‘person from Arbizu’

792

Eurasian languages

Hualde, J.I. (1991). Basque Phonology. London & New York: Routledge. Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Souletin [U/P] Basque, Eastern Basque. Pays de Soule (France). f f

Stress generally falls on the penultimate syllable. Some su‰xes attract stress to themselves when they occur in final position.

¨gizun

‘man’

gi¨zuna ‘man abs.’

gizu¨nek

‘the men erg.’

Hualde, J.I. (1993). Topics in Souletin Phonology. In J.I. Hualde & J. Ortiz de Alibina (eds.) Generative Studies in Basque Linguistics, 289–327. Amsterdam & Philadelphia: Benjamins. Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter. Basque, Zeberio [LEX] Basque, Western Basque, Southwestern Biscayan. Southwestern Biscay (Spain). f f

f

f

f

Accent location is lexically determined. Unlike in the northern Biscay dialects, Zeberio accent is not just marked with pitch. Uninflected words can be accented or unaccented, but all inflected words are accented. In words with several accented morphemes only the first is given surface prominence. Su‰xes may influence accenting, depending on morphology and stem size.

mama¨roa ‘bug abs.’ mama¨roaRi ‘bug dat.’

ma¨maroak ‘bugs abs.’ ma¨maroaRi ‘bugs dat.’

Hualde, J.I. (1999). Basque Accentuation. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 947–993. Berlin & New York: Mouton de Gruyter.

Eurasian languages

793

Caucasian Abkhaz [LEX] North Caucasian, Northwest, Abkhaz-Abazin. Georgia. f

Stress location is unpredictable, there are minimal pairs.

¨arat Wa ‘plurality’

a¨rat Wa

‘coal’

ara¨t Wa ‘bark from a walnut tree’ Hewitt, B.G. (1979) Abkhaz. Lingua Descriptive Studies 2. Amsterdam: North Holland Publishing Company. Archi; Archin [S/S] North Caucasian, Northeast, Lezgian. Southern part of Dagestan Republic (Russian Federation). f f

Only one of the first two syllables may be stressed. Comparison of vowel quality determines stressing: The second syllable is stressed if its vowel is not inferior in weight to the first vowel, else stress is initial.

bu¨Xut ‘veins’

¨belum

‘spades’

Kibrik, A.E. et al. (1977). Opyt strukturnogo opisanija arcinskogo jazyka, 4 vols. Moscow: Izdatel’stvo MGU. Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter. Avar; Dagestani [I/I (tone)] North Caucasian, Northeast, Avaro-Andi-Dido, Avar. Southern part of Dagestan Republic, southern part of Azerbaijan (Russian Federation). f

f

Stress is initial, except when the first syllable has a low tone and the second a high tone. If the tone value of the two first syllables is equal stress shifts to the second syllable if that syllable is closed.

¨ro@su$ ‘village’

mu$¨t’u@ ‘mirror’

mo@ ¨q’o@q’ ‘partridge’

Bokarev, A.A. (1949). Sintaksis avarskogo jazyka. Moscow-Leningrad. Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter.

794

Eurasian languages

Bagvalal; Bagulal; Bagval; Bagvalin [LEX-U/P] North Caucasian, Northeast, Avaro-Andi-Dido, Andi. Southern part of Dagestan Republic (Russian Federation). f f f

Words can have strong stress, weak stress or no stress. Strong stress can occur anywhere in the word. Weak stress occurs on the final syllable if it is closed, else on the penult.

¨buk’a ‘it was’

tSi¨ba ‘bitch’

tS’i˙lu´ ‘cheek gen.’

˙tS’ila ‘cheek nom.’

awal¨la ‘in Agwali village’

Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter. Georgian [A;I (NMS)] South Caucasian, Georgian. Georgia. f f

f

Stress in Georgian is extremely weak and has no e¤ect on vowel quality. In words of four syllables or less, stress is either initial or antepenultimate. In longer words stress is both initial and antepenultimate.

¨sakartvelo or sa¨kartvelo ‘Georgia’ ¨mdgoma¨reoba

‘situation’

Aronson, H.J. (1982). Georgian – A reading Grammar. Columbus, Ohio: Slavica. Rudenko, B.T. (1972). Grammatika gruzinskogo jazyka. (Janua Linguarum, series anastatica 7) The Hague: Mouton. Tschenke´li, K. (1958). Einfu¨hrung in die georgische Sprache. Bd. 1–2. Zu¨rich: Amirani. Vogt, H. (1971). Grammaire de la langue ge´orgienne. Oslo: Universitetsforlaget. Ghodoberi [LEX-(F/)L (Tone)] North Caucasian, Northeast, Avaro-Andi-Dido, Andi. Southern part of the Dagestanian Republic (Russian Federation). f

f

In nouns stress depends on high tone, articulatory accent, lexical marking and breathy voice in an intricate way. Generally, the heaviest syllable or the final is stressed. In verbs only the aorist has initial stress, otherwise stress is final.

Eurasian languages

¨"@tSa@´" $ ‘mare dat.’

bu@a@¨´" $

‘bull.dat’

¨u@m" $ ‘sleep aor.’

u@mi"@t´"@¨bu@

‘sleeping fut.part.’

795

Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter. Saidova, P.A. (1973). Godoberinskij jazyk. Maxackala. Hunzib [U/P] North Caucasian, Northeast, Avaro-Andi-Dido, Dido. Southern part of the Dagestanian Republic (Russian Federation). f

Stress the final syllable if it contains a long vowel, else the penult.

¨iyu

‘mother’

hi¨naa ‘how’

¨k’ot’t’u ‘good’ k’i¨s&aa

‘play’

Berg, H. van den (1995). A Grammar of Hunzib. Lincom Studies in Caucasian Linguistics 1. Munich: Lincom Europa. Ingush [I] North Caucasian, North Central, Chechen-Ingush. Chechen Ingushetia, northern Caucasus, west of the Chechen (Russian Federation). f

Stress always falls on the initial syllable.

¨mott ‘tongue, language’ ¨beaccara ‘green’

¨tassa ‘sprinkle’ ¨kinas&ka ‘book’

Nichols, J. (1994). Ingush. In Rienk Smeets (ed.) The indigenous languages of the Caucasus, volume 4, 79–146. New York: Caravan Books. Lezgi; Lezgian; Kiurintsy [I/I (IRR)] North Caucasian, Northeast, Lezgian. Southern part of the Daghestan Republic (Russian Federation). f

f

Stress always falls on the second syllable if it is closed and the first is open. If the first is closed stress is usually initial, with many exceptions, especially in loans.

c&u¨k’ul ‘knife’

q’ic¨ti ‘jug.erg’

¨s&ekdi ‘lamb.erg’

796

Eurasian languages

Haspelmath, M. (1993). A Lezgian Grammar. Mouton Grammar Library 9. Berlin: Mouton de Gruyter. Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter. Tsakhur; Tsaxur; Caxur [U;I] Caucasian, Nakh-Dagestan, Dagestanian, Lezgian. Dagestan, Azerbaijan. (Russian Federation). f f f

f

Words come in two classes, stressed and weakly stressed (or unstressed). The stress in stressed nouns is final, in verbs it is initial. In weakly stressed forms long vowels (only in nouns) and closed syllables are prominent. Status of weak stress unclear (perhaps phrase accent red.)

t’y¨bytS’ ‘spindle’ dop¨pi ‘clay jug’

tS’i¨ka ‘knife’ ¨eøk’alor ‘jump.pres’

Ibramigov, G.X. (1990). Caxurskij jazyk. Moscow: Nauka. Kodzasov, S. (1999). Caucasian: Daghestanian Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 995–1020. Berlin & New York: Mouton de Gruyter. Indo-European Armenian [L/F] Indo-European, Armenian. Armenia, Eastern Turkey, Middle East. f f

Primary stress falls on the last full vowel, else on the first. In some dialects secondary stress falls on the first syllable.

Jer¨pHem«n ‘sometimes’

¨kutemn« ‘cress’

Abeghian, A. (1936). Neuarmenische Grammatik. Berlin-Leipzig. Fairbanks, G.H. (1958). Spoken East Armenian. New York: American Council of Learned Societies. Gulian, K.H. (1957). Elementary modern Armenian Grammar. New York: Frederick Ungar. Kogian, S.L. (1949). Armenian grammar (west dialect). Vienna: Mekhitarist Press.

Eurasian languages

797

Vaux, B. (1995). The Phonology of Armenian. Ph.D. Dissertation, Harvard University. Awadhi-Bagheli [P/P] Indo-European, Indo-Iranian, Indo-Aryan, East Central zone. India. f

Stress falls on the penultimate syllable, except when the final is heavy and the penult is not, then stress is final.

pi¨saøn ‘flour’ kˆ¨œisi¥ ‘he said’

¨baøis ¨saøœaø

‘twenty two’ ‘share’

Saksena, B. (1971). Evolution of Awadhi. Delhi: Motilal Banarsidass. Bengali [I] Indo-European, Indo-Iranian, Indo-Aryan, Eastern zone, Bengali-Assamese. India, Bangladesh. f

Stress falls on the first syllable.

¨bissleSon ‘analysis’

¨bostha

‘condition’

Bykova, E.M. (1981). The Bengali Language. Translation of Bengalskii jazyk. Languages of Asia and Africa. Bhojpuri [A-U/P] Indo-European, Indo-Iranian, Indo-Aryan, Central zone,Western Hindi, Hindustani. India. f

f

f

Primary stress falls on the antepenultimate syllable of words with 4 syllables or more. Primary stress falls on the penultimate syllable elsewhere, except when the final vowel is checked, in which case stress is final. Secondary stress falls on long vowels, final closed syllables and to the left of the main stress on the only syllable preceding it, or two to the left.

˙ga¨lab ‘to melt’ ˙kHa¨tamka˙rab ‘to finish’

˙ba¨hini ˙kHaju¨aøib

‘sister’ ‘to scratch’

Trammel, R.L. (1971). The phonology of the northern standard dialect of Bhojpur. Anthropological Linguistics 13, pp. 126–141. Breton [P] Indo-European, Celtic, Insular, Brythonic. Western Brittany, but also dispersed in Eastern Brittany (France).

798 f

Eurasian languages

Word stress falls on the penultimate syllable.

¨pesk«t ‘fish pl.’

ki¨der«s

‘butcherswife’

peske¨ter«s ‘fishing’

Bothorel, A. (1982). Etude Phonetique et Phonologique du Breton Parle a Argol (Finistere-Sud), Atelier National Reproduction Theses, Universite Lille, Lille. Stephens, J. (1993). Breton. In M.J. Ball (ed.) The Celtic Languages. London & New York: Routledge. Ternes, E. (1992). The Breton Language. In D. MacAulay (ed.) The Celtic Languages. Cambridge: Cambridge University Press. Catalan-Valencian-Balear [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Ibero-Romance, North, Eastern. Spain, France, Alghero (Sardinia). f

Stress falls on the final syllable if closed or lexically marked, else on the penult.

pa¨ awl« ‘word’ b«¨rEts ‘hats’

cam¨pana ‘bell’ ele¨fan ‘elephant’

Fabra, P. (1897). E´tude de phonologie catalane. Revue Hispanique 4: 5–30. Fabra, P. (1956 [1912]). Grama´tica catalana. Teide, Barcelona. Gulsoy, J. (1982). Catalan. Trends in Romance Linguistics and Philology 3, 189–296. The Hague: Mouton. Hualde, J.I. (1992). Catalan. London & New York: Routledge. Roca, I. (1999). Stress in the Romance Languages. Chapter 10 in H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin: Mouton de Gruyter. Mascaro, J. (1978). Catalan Phonology and the Phonological Cycle. Unpublished doctoral dissertation, MIT Cambridge, Massachusetts, Distributed by the IULC. Wheeler, M. (1979). Phonology of Catalan, Oxford, Basil Blackwell. Wheeler, M. (1988). Catalan. In M. Harris & N. Vincent (eds.) The Romance Languages. London: Routledge. Cornish [P] Indo-European, Celtic, Insular, Brythonic. Cornwall (England, but now extinct).

Eurasian languages f f

799

Stress falls on the penultimate syllable. Secondary stress on the fourth and sixth from the end (no examples).

¨estren ‘foreigner’

es¨trenyon ‘foreigners’

¨arluD

‘lord’

George, K. (1993). Cornish. In M.J. Ball (ed.) The Celtic Languages. London & New York: Routledge. Pool, P.A.S. (1958). Cornish for beginners. Cornish Language Board. Thomas, A.R. (1992). The Cornish Language. In D. McAuley (ed.) The Celtic Languages. Cambridge: Cambridge University Press. Czech [I] Indo-European, Slavic, West, Czech-Slovak. Western Czechoslowakia, Bohemia, Moravia, Silezia. f f

Primary stress falls on the first syllable. Secondary stress falls on alternate syllables thereafter.

¨name˙sti

‘[city] square’

¨nadra˙Zi

‘station’

¨kniha

‘book’

Harkins, W.E. (1953). A Modern Czech Grammar. New York: Kings Crown Press. Jakobson, R. (1926). Contributions to the study of Czech Accent. Reprinted in Roman Jakobson, Selected Writings 1 (1971), pp. 641–25. Danish [L/P] Indo-European, Germanic, North, East Scandinavian. Denmark, Greenland, Northern Germany. f f

Primary stress falls on the last long vowel. If there are none, stress falls on final closed syllables, else on the penultimate.

vio¨l"ø$n ‘violin’

pe¨troø$leum

‘para‰n’

¨foto

kom¨plot

‘conspiracy’

‘fotograph’

[ $ is stOd]

Rischel, J. (1970). Morpheme stress in Danish. Aripuc 4: 111–144. Rischel, J. (1972). Compound Stress in Danish without a Cycle. Aripuc 6: 211–228. Rischel, J. (1982). On Unit Accentuation in Danish – amd the Distinction between Deep and Surface Phonology. Aripuc 16: 191–239.

800

Eurasian languages

Dari (dialect of Persian) [U] Indo-European, Indo-Iranian, Iranian, Western, Southwestern, Persian. Afghanistan. f f

Primary stress falls on the final syllable. Secondary stress is located on the initial syllable.

˙zendQ¨gi

‘life’

˙as&pQzxa¨nQ ‘kitchen’

Mueller Bing, J. (1980). Linguistic Rhythm and Grammatical Structure in Afghan Persian. Linguistic Inquiry 11–3, 437–463. Dutch [U-P/P;A] Indo-European, Germanic, West, Continental, Low, Dutch. The Netherlands, Belgium. f

f f

f

Primary stress falls on the final syllable if it has three segments in the rhyme. Otherwise stress falls on the penult if the final syllable is open. When it is closed, stress is penultimate if also closed, otherwise antepenultimate. Secondary stress falls on alternate syllables to the left (with many exceptions).

¨tEmpo>

‘pace’

¨Alfa>˙bEt

‘alphabet’

e>¨lEktrn

‘electron’

˙pa>ra>¨si>t

‘parasite’

Hulst, H.G. van der (1984). Syllable Structure and Stress in Dutch. Dordrecht: Foris Publications. Kager, R. (1989). A Metrical Theory of Stress and Destressing in English and Dutch. Dordrecht: Foris Publications. Trommelen M. & W. Zonneveld (1999). Dutch. H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe. Mouton de Gruyter, Berlin. Chapter 8.1.2, 492–514. English [(U-)P/A] Indo-European, Germanic, West, North Sea, English. World. Approximation: f Primary stress falls on the final syllable in nouns if the vowel is long, in verbs if the vowel is long or there are two closing consonants. f In other cases, stress falls on the penult if it contains a long vowel or coda.

Eurasian languages f f

801

Else stress is antepenultimate. Secondary stress falls on alternate syllables to the left (many exceptions).

o¨bey

a¨genda

mo¨lest

¨discipline

as¨tonish mon¨soon

tre¨mendous ¨innocent ho¨rizon ¨croco˙dile ¨ali˙bi

¨cinnamon

Trommelen, M. & W. Zonneveld (1999). English. H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe. Mouton de Gruyter, Berlin. Chapter 8.1.1, 478–491. Faroese [I] Indo-European, Germanic, North, West Scandinavian. Faroer Islands. f f

Primary stress falls on the first syllable. Secondary stress on rhythmic basis.

¨tomur ‘empty’

¨hestarnir ‘the horses’

¨seyDa˙fylgi ‘flock of sheep’

´ rnason, K. (1999). Icelandic and Faroese. H. van der Hulst (ed.) Word A Prosodic Systems in the Languages of Europe. Mouton de Gruyter, Berlin. Chapter 8.2.3, 567–604. Lockwood, W.B. (1955). An Introduction to Modern Faroese, Copenhagen: Ejnar Munksgaard. Farsi, Western; Persian. [U] Indo-European, Indo-Iranian, Iranian, Western, Southwestern Persian. Iraq, Iran, Tajikistan, Turkmenistan, Uzbekistan; Bahrain, Oman Qatar. f

Stress falls on the final syllable.

za¨nan ‘women’

sanda¨li ‘chair’

xari¨dam ‘i bought’

Lambton, Ann K.S. (1953). Persian Grammar. Cambridge: Cambridge University Press. French [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Gallo-Romance, North. France, Quebec (Can), Switserland, Belgium, Corsica, Africa, E. Asia. f f f

Primary stress falls on the final syllable, except when that is a /«/. A likely alternative is that French has just a phrase accent. Secondary stress claimed to exist, but there is not much evidence.

aøbriø¨koø ‘apricot’

ystEn¨sil ‘utensil’

¨kEl« ‘which [fem.]’

802

Eurasian languages

Dell, F. (1984). L’accentuation des phrases en Franc¸ais. In F. Dell, Hurst, D. & J.R. Vergnaud (eds.) Forme sonore du langage, 65–122. Paris: Hermann. Roca, I. (1999). Stress in the Romance Languages. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Gaelic, Irish [I] Indo-European, Celtic, Insular, Goidelic. Ireland. f f

Primary stress falls on the first syllable. Secondary stress often on non-initial long vowels.

¨bach˙log ‘bud’

¨uachta˙raøn ‘president’

Bammesberger, A. (1983). A Handbook of Irish, Vol. 2. An outline of Modern Irish Grammar. Heidelberg: Carl Winter Universita¨tsverlag. O’Siadhail, M. (1989). Modern Irish, Grammatical Structure and Dialectal Variation. Cambridge: Cambridge University Press. Gaelic, Munster (dialect of Gaelic, Irish) [F/F] Indo-European, Celtic, Insular, Goidelic. Southwest of Irish Republic. f f

Stress falls on the first long vowel, else on the first. Stress may shift to -(e)acht, when preceded by a syllable with a short vowel.

bean¨nacht ‘blessing’

cai¨liøniø ‘girls’

Dillon, M. (1961). Irish. Teach Yourself books. London: Hodder & Stoughton. O’Siadhail, M. (1989). Modern Irish: Grammatical Structure and Dialectal Variation. Cambridge, Cambridge University Press. Gaelic, Scottish [I] Indo-European, Celtic, Insular, Goidelic. Northwest Scotland. f f

Primary stress falls on the first syllable. Location of secondary stress unclear.

¨djeSal«x«˜ ‘preparing’ ¨ah«r«xi«n ‘changes’ ¨kl«nt« ‘complete, fulfilled’

Eurasian languages

803

Dorian, N.C. (1978). East Sutherland Gaelic. Dublin Institute for Advanced Studies. Gillies, W. (1993). Scottish Gaelic. In M.J. Ball (ed.) The Celtic Languages, 145–227. London & New York: Routledge. German [U-P/A] Indo-European, Germanic, West, Continental, High. Germany, Austria, Switserland, Hungary, Chechoslowakia, Liechtenstein. Approximation: f Primary stress is final if the vowel is long or there are two closing consonants. f In other cases, stress falls on the penult if it is closed by a consonant. f Else stress is antepenultimate. f Secondary stress falls on alternate syllables to the left (many exceptions). hi¨bIskus ‘hibiscus’ ¨anorAk ‘anorak’

a¨lArm ‘alarm’ ¨alibi ‘alibi’

Jessen, M. (1999). German. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, chapter 8, 515–545. Berlin & New York: Mouton de Gruyter. Greek, Modern [A;P;U (LEX)] Indo-European, Greek, Attic. Greece, Cyprus, Egypt, Turkey, Italy. f

Stress is located on one of the last three syllables (lexically determined).

¨skrinjo ‘desk’ ¨maTima ‘lesson’

la¨os ‘nation’ po¨litis ‘citizen’

Drachman, G. & A. Malikouti-Drachman (1999). Greek Word Accent. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 897–945. Berlin & New York: Mouton de Gruyter. Malikouti-Drachmann, A. & G. Drachmann (1981). Slogan Chanting and Speech rhythm in Greek. In W. Dressler (ed). Phonologica 1980, Innsbruck, IBS. Gujarati [U/P;P/A] Indo-European, Indo-Iranian, Indo-Aryan, Central zone, Gujarati. Gujarat (India).

804 f

f

Eurasian languages

In bisyllables, stress is final if the vowel is /a/, if both vowels are non-/a/ or the first syllable contains a schwa and is closed, stress is penult, if the penult is an open schwa there is free variation. In longer words, stress the penult if it is not schwa (but schwa þ coda is stressed) or /i/ and the antepenult is not /a/. If the antepenult is /a/ there is free variation. If the penult is schwa and open, stress is antepenultimate.

¨bidi ‘a type of cigar’ kh«¨but«r ‘pigeon’ j«¨j«nti ‘birthday’

nak¨san ‘damage’ ¨akr«m«n ‘invasion’ ¨tajeter or ta¨jeter ‘recently’

Cardona, G. (1965). A Gujarati Reference Grammar. University of Pensylvania, South Asia Regional Studies, Philadelphia: University of Pennsylvania Press. Hindi [U%A] Indo-European, Indo-Iranian, Indo-Aryan, Central zone, Western Hindi, Hindustani. India. f f f

Stress falls on final syllables of the form VVC or VCC. Otherwise stress is on penultimate long vowels or closed syllables. If the final is not superheavy, and the penult is not heavy stress shifts to a heavy antepenult (or even to the pre-antepenult).

ka¨maal ‘wonders’ ¨anumati ‘permission’

in¨saaniyat ‘humanness’ ¨kamal ‘lotus’

Fairbanks, C. (1981). The development of Hindi oral narrative meter. Ph.D. Dissertation. Univ. of Wisconsin, Madison. McGregor, R.S. (1977). Outline of Hindi Grammar. Second Edition. Delhi: Oxford University Press. Icelandic [I] Indo-European, Germanic, North, West Scandinavian. Iceland. f f

Primary stress falls on the first syllable. Secondary stress falls on alternate syllables to the right.

¨taka ‘to take’ ¨alma˙nak ‘calendar’ ¨bio˙grafi˙a

‘biography’

´ rnason, K. (1980). Quantity in Historical Phonology, Icelandic and A Related Cases. Cambridge: Cambridge University Press.

Eurasian languages

805

Italian [U/P] Indo-European, Italic, Romance, Italo-Western, Italo-Romance. Italy, Switzerland, Corsica, San Marino. f

Stress falls on the final syllable if lexically marked as accented, else stress is on the penultimate syllable of the stem (without desinence).

do¨menic a ‘sunday’

pal¨to ‘coat’

¨atom o

‘atom’

Bertinetto, P.M. (1981). Strutture Prosodiche dell’Italiano. Accento, Quantita`, Sillaba, Giuntura, Fondamenti Metrici. Academia della Crusca, Firenze. Lepschy, G.C. (1978). The Italian Language Today. London: Hutchinson. Malagoli, G. (1946). L’Accentazione Italiana. Firenze. Muljacic, Z. (1972). Fonologia della Lingua Italiana. Bologna: Il Mulino. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin: Mouton de Gruyter. Saltarelli, M. (1970). A Phonology of Italian in a Generative Grammar. The Hague: Mouton. Vincent, N. (1988). Italian. In M. Harris & N. Vincent (eds.) The Romance Languages, 279–313. London: Routledge. Vogel, I. & S. Scalise (1982). Secondary stress in Italian. Lingua 58: 213– 242. Kalami; Kalam Kohistani [U (Tone)] Indo-European, Indo-Iranian, Indo-Aryan, Northwestern zone, Dardic, Kohistani. Northern Pakistan. f

Stress falls on the final syllable, except in words with a HL tone, where stress varies.

Sˆr¨da ‘quickly’

¨Soøliø ‘rice in the field’ [hl tone]

Baart, Joan L.G. (1997). The Sounds and Tones of Kalam Kohistani. Studies in Languages of Northern Pakistan 1. Islamabad: Horsleys Green. Latin [P/A] Indo-European, Italic, Latino-Faliscan. Italy.

806 f

f

Eurasian languages

Stress falls on the penultimate syllable if it contains a long vowel or is closed. Else stress is antepenultimate.

re¨feøcit re¨fectus ¨reficit Latvian [I] Indo-European, Baltic. Latvia. f

Stress falls on the first syllable.

¨baga#t"Æ ba ‘wealth’ ¨kokvilna ‘cotton’

¨kra#sa ‘colour, paint, dye’ ¨ma#kslinieks ‘artist’

Ekblom, R. (1933). Die lettischen Akzentarten. Uppsala: Almquist & Wiksells Boktryckerei. Fennel, T.G. & H. Gelsen (1980). A grammar of modern Latvian. Berlin: Mouton. Markus, D. (1987). Types of syllable tonemes in the Ziemeri variant of High Latvian Dialect. Proceedings of the 11th ICPhS 5, 107–110. Tallinn, Estonia. Markus, D. (1991). Syllable tonemes in Latvian. Proceedings of the 12th ICPhS 2, 242–245. Aix en Provence. Lithuanian [F/F] Indo-European, Baltic. Lithuania. f f

Stress falls on the first syllable marked for accent, else on the first. Secondary stress falls on the second syllable after the tone.

¨kiœSkis ‘rabbit’ va¨do)vas ‘leader’

¨giria ‘praises’ ¨mo@ ky˙tojas ‘teacher’

gi¨ria

‘forest’

Ambrazas, V., et al. (1997). Lithuanian Grammar. Lithuania: Baltos Lankos. Girdenis, A. (1985). Grammatika Litoskogo jazyka. Vilnius: Mokslas. Leskien, A. (1919). Litauisches Lesebuch. Heidelberg: Winter Verlag. Mathiassen, T. (1996). A Short Grammar of Lithuanian. Columbus: Slavica Publisers. Otrebski, J. (1958–1965). Gramatyka jezyka litewskiego, 3 volumes. Warszawa: Panstwowe Wydawnictwo Naukowe.

Eurasian languages

807

Young, S.R. (1991). The Prosodic Structure of Lithuanian. New York: University Press of America. Macedonian [A] Indo-European, Slavic, South, Eastern. Macedonia. f

Stress falls on the antepenultimate syllable.

vo¨denitSar ‘miller’

¨polkovnik ‘colonel’

¨vetSer ‘evening’

Lunt, H. (1952). A Grammar of the Macedonian Literary Language. Skopje. Maithili (dialect of Maithili; Bihari) [P%A/P] Indo-European, Indo-Iranian, Indo-Aryan, Eastern zone, Bihari. India, Nepal. f

f f

f

Primary stress falls on the penultimate syllable if it contains a long vowel. If the penult contains a short vowel and the final is long, stress is final. If both the penultimate and the final vowel are short, stress is antepenultimate if that vowel is long, else penultimate. Secondary stress falls on initial syllables.

¨maøti

‘earth dir. base’

˙adh«¨laøne ‘bad’

˙pat«¨hiø

‘thin’

¨gaøbhinu

‘pregnant’

Davis (1984). Basic Colloquial Maithili. Dehli: Motilal Banarsidass. Jha (1958). The Formation of the Maithili Language. London: Luzac. Yadav (1984). Maithili phonetics and Phonology. Mainz: Seldan & Tamm. Manx [I] Indo-European, Celtic, Insular, Goidelic. Isle of Man (UK; now extinct). f

Stress falls on the first syllable.

¨begAn ‘a little’

¨puøs« ‘wedding’

Broderick, G. (1993). Manx. In M.J. Ball (ed.) The Celtic Languages, 228–285. Routledge: London & New York. Thomson, R.L. (1992). The Manx Language. In D. McAuley (ed.) The Celtic Languages. Cambridge: Cambridge University Press.

808

Eurasian languages

Nepali; Gurkhali; Gorkali [I/I] Indo-European, Indo-Iranian, Indo-Aryan, Northern zone, Eastern Pahari. Nepal. f

Stress falls on the first syllable, except when the second vowel is long and the first is not.

Meerendonk, M. (1949). Basic Gurkhali Grammar. (4th ed. 1971), Singapore: Sen Wah Press and Co. Norwegian [U-P/A] Indo-European, Germanic, North Germanic. Norway. Approximation: Primary stress is final if the vowel is long or there are two closing consonants. f In other cases, stress the penult if it is closed by a consonant or the vowel is long. f Else stress is antepenultimate. f Secondary stress falls on alternate syllables to the left (many exceptions). f

maka¨roøni ¨leksikon

‘macaroni’ ‘lexicon’

¨pasta ‘pasta’ a¨gurk ‘pickle’

Rice, C. (1999). Norwegian. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 545–567. Berlin & New York: Mouton de Gruyter. Old English [I] Indo-European, Germanic, West, North Sea, English. England. f f

Primary stress falls on the first syllable. Secondary stress reported on alternate syllables to the right respecting weight.

¨mo dgidanc ‘thought’

¨QTeling ‘prince’

Hamer, R.F.S. (1967). Old English Sound Changes for Beginners. Oxford: Basil Blackwell. Mitchell, B. & F.C. Robinson (1986). A guide to Old English. 4th ed. Oxford: Basil Blackwell.

Eurasian languages

809

Osetin; Ossete; Ossetic [I/S] Indo-European, Indo-Iranian, Iranian, Eastern, Northeastern. NorthOssetia, Georgia, South-Ossetia. f

Stress falls on the first syllable if it contains a long vowel, else on the second.

¨suudzaag ‘burning’

bQ¨laas

‘a tree’

Abaev, V.I. (1964). A Grammatical Sketch of Ossetic. Indiana University Center in Antropology, Folklore and Linguistics, Bloomington, Indiana. Pashto [LEX] Indo-European, Indo-Iranian, Iranian, Eastern, Southeastern, Pashto. Pakistan, Afghanistan. f

Stress location is lexically determined.

¨guta ‘knot’ ¨vulambed« ‘he took a bath’

gu¨ta ‘pochard’ stomaønti¨aø ‘fatigue’

Becka, J. (1969). A study in Pashto stress. Dissertationes Orientales 12, Academia, Prague. Shafeev, D.A. (1964). A Short Grammatical Outline of Pashto. The Hague: Mouton. Polabian [U/P] Indo-European, Slavic, West, Lechitic. Lower Sachsony and Brandenburg (now Extinct). f

Stress falls on the final syllable if it contains a long vowel, else on the penult.

Olesch, R. (1974). Der dravaenopolabische Wortakzent. Mainz: Verlag der Akademie der Wissenschaften. Polish [P] Indo-European, Slavic, West, Lechitic. Poland. f f

Primary stress falls on the penultimate syllable. Secondary stress on alternate syllables counted from the left (not on the antepenult).

¨rozpraw ˙saksofo¨nista

‘discussion [gen. pl]’ re¨porter ‘saxophone player’ ˙revo˙lucjo¨nista

‘reporter’ ‘revolutionary’

810

Eurasian languages

Franks, S. (1985). Extrametricality and Stress in Polish. Linguistic Inquiry 16: 144–151. Portuguese [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Ibero-Romance, North, Western. Portugal, Brazil, Spain, Guinea, Angola, Mozambique. f

f f

Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). Final stress is oligatory when the final syllable is closed by a palatal. In verbs, stress has become morphologised.

¨perola

‘pearl’

chami¨ne ‘chimney’

pa¨nela ‘pan’ pai¨nel

‘panel’

Abaurre, M.B.M. & W.L. Wetzels (1992). Fonologia do Portogueˆs. Caderneos de Estudos Linguisticos 23. Universidade de Campinas, Campinas. Caˆmara, Jr. & J. Mattoso (1979). The Portuguese language. Chicago: University of Chicago Press. Martins, M.R. Delgado (1982). Aspects de l’accent en Portugais. Hamburg: Buske Verlag. Mateus, M.H. Mira et al. (1989). Grama´tica da Lı´ngua Portuguesa. Lisboa: Caminho. Parkinson, S. (1988). Portuguese. In M. Harris & N. Vincent (eds.) The Romance Languages, 131–169. London: Croom Helm. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Portuguese, Brazilian (dialect of Portuguese) [P] Indo-European, Italic, Romance, Italo-Western, Western, Ibero-Romance, North, Western. Brazil. f

Stress falls on the penultimate syllable.

¨ama ‘he loves’

ago¨nia ‘agony’

Lopez, B. (1979). The Sound Pattern of Brazilian Portugese (Cariocan Dialect). Doctoral Dissertation, UCLA.

Eurasian languages

811

Provenc¸al; Occitan [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Ibero-Romance, North, Eastern. Southern France. f

f f

Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). Final stress is oligatory when the final syllable is closed by a palatal. In verbs, stress has become morphologised.

ca¨dena

‘chain’

fi¨nestra ‘window’

espi¨tal

‘hospital’

tara¨bast

‘uproar’

Alibe`rt, L. (1976). Grammatica Occitana, segons los parlars lengadocians. Centre d’Estudis Occitans, Montpelhie`r. Bec, P. (1963). La Langue Occitane, ‘‘Qui Sais-je?’’, No. 1059. Paris: Presses Universitaires de France. Cardaillac, K.R. (1971). A Descriptive Analysis of Gascon (Donzac). The Hague: Mouton. Maas, U. (1969). Untersuchungen zur Phonologie und Phonetik der Mundart von Couzou (Lot). Ph.D. dissertation, Freiburg im Breisgau. Maurand, G. (1974). Phone´tique et phonologie du parler occitane d’Ambialet (Tarn). [The`se d’Etat, Universite´ de Toulouse-Le Mirail] Cercle Occitane, Villeneuve-sur-Lot. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Sauzet, P. (1986). Les clitiques Occitan: analyse me´trique de leur variation dialectale. Morphosyntx des langues Romanes, Actes du XVIIeme Congres International de Linguistique et Philologie Romanes 4, 155–180. Universite´ de Provence-Jeanne Lafitte, Marseille. Sauzet, P. (1994). Attenance, gouvernement et mouvement en phonologie. Les constituents dans la phonologie et la morphologie de l’occitan. Ph.D. Dissertation, Universite´ de Paris VIII; distributed by the Centre d’Estudis Occitans, Montpellier. Romani, North Russian (dialect of Romani, Baltic) [U] Indo-European, Indo-Iranian, Indo-Aryan, Central zone, Romani, Northern. Across Europe and the Near East.

812 f

Eurasian languages

Stress falls on the final syllable (with some lexical exceptions).

duratu¨no ‘far’

ro¨mestir ‘from a Gypsy’

Ventzel, T.V. (1983). The Gypsy Language. Moscow: Nauka. Romansch, Berguner (dialect of Rheto-Romance) [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Gallo-Romance, Rheto-Romance. Switzerland. f

f f

Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). Final stress is oligatory when the final syllable is closed by a palatal. In verbs, stress has become morphologised.

mu¨leł ‘mill’ l«¨venj« ‘avalache’

bu¨lef ¨p«le

‘mushroom’ ‘swamp’

Kamprath, C. (1987). Supra-segmental Structures in a Rhaeto-Romance Dialect: A Case Study in Metrical and Lexical Phonology. Ph.D. Dissertation, University of Texas, Austin. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Rumanian [U/P] Indo-European, Italic, Romance, Eastern, North. Rumania, Greece, Soviet Union, Albania, Yogoslavia, Bulgaria. f

f

Stress falls on the final syllable if that is marked for accent, else on the penultimate syllable of the stem (without desinence). In verbs, stress has become morphologised.

¨lingur «

‘spoon’

cu¨m«tru ‘godfather’

mui¨er e

‘woman’

maha¨la

‘suburb’

Augerot, J.E. (1974). Romanian Phonology. A Generative Sketch of the Core Vocabulary of Standard Romanian. Idaho Research Foundation, University of Idaho, Idaho. Avram, A. (1978). Romanian and General Phonetics and Phonemics. Current trends in Romanian Linguistics, Revue Romaine de Linguistice 23; Cahiers de Linguistique The´orique et Applique´e 15, 71–109. Bucuresti: Editura Academiei.

Eurasian languages

813

Graur, A. & A. Rosetti (1938). Esquisse d’une phonologie du Roumain. Bulletin Linguistique 6, 5–29. Lombard, A. (1935). La Pronunciation du Roumain. Uppsala. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Steriade, D. (1984). Glides and vowels in Romanian. Berkeley Linguistics Society 10, 47–64. Ulivi, A. (1977). Quelques remarques sur la relation entre la syncope et l’accent dans les parlers dacoroumains. Revue Romaine de Linguistique 22, 63–71. Ulivi, A. (1985). Quelques remarques sur l’accent secondaire dans les parlers dacoroumains. Revue Romaine de Linguistique 31, 505–512. Russian [F/F LEX] Indo-European, Slavic, East. CIS. f

Stress falls on the first syllable lexically marked for accent, else on the first.

ras¨prava ‘reprisal [nom.sg]’

zje¨ny ‘women’

Ungebaum, B.O. (1957). Russian Grammar. Oxford. Hamilton, W.S. (1980). Introduction to Russian Phonology and Word Structure. Columbus: Slavica Publishers, Ohio. Kodzasov, S. (1999). Russian. In H. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 852–876. Berlin & New York: Mouton de Gruyter. Sanskrit [F/F (pitch)] Indo-European, Indo-Iranian, Indo-Aryan. India (Now extinct). f f

Stress falls on the first high pitched syllable, else on the first. Claimed to be pitch-accent system rather than stress.

¨apaciti ‘retribution’ nama¨syati ‘respects’

dhaø¨rayati ‘holds’ aparaøh¨åa ‘afternoon’

Burrow, T. (1955). The Sanskrit Language. London: Faber and Faber.

814

Eurasian languages

Sardinian, Campidanese (dialect of Italian) [U/P] Indo-European, Italic, Romance, Southern, Sardinian. Sardinia (Italy). f

f f

Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). Final stress is oligatory when the final syllable is closed by a palatal. In verbs, stress has become morphologised.

¨mendul a ‘almond’

mengi¨an u

‘morning’

¨atom u

a¨tomic u

‘atomic’

‘atom’

Blasco, F.E. (1986). La Lingua Sarda Contemporanea. Edizione della Torre, Cagliari. Jones, M.A. (1988). Sardinian. In M. Harris & N. Vincent (eds.) The Romance Languages, 314–350. Croom Helm, London & Sydney. Porru, V.R. (1975). Saggio di grammatica sul dialetto sardo meridionale. Liberia Editrice Dessı`, Sassari. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Virdis, M. (1978). Fonetica del dialetto sardo campidanese. Cagliari: Edizione della Torre. Wagner, M.L. (1984). Fonetica storica del sardo. Cagliari: Trois. Serbo-Croatian [F/F (pitch)] Indo-European, Slavic, South, Western. Serbia, Bosnia-Hercegovina, Montenegro. f

Stress is assigned to the first syllable with a tone, else to the initial syllable.

bra¨to&vStina ‘brethren community’ Hodge, Ch. (1946). Serbo-Croatian Phonemes. Language 22: 112–120. Gvozdanovic, J. (1999). South Slavic. In H.G. van der Hulst (ed.) Word Prosodic Systems in European Languages, 839–852. Berlin & New York: Mouton de Gruyter. Sindhi [L/P] Indo-European, Indo-Iranian, Indo-Aryan, Northwestern zone, Sindhi. India and Pakistan.

Eurasian languages f

815

Stress falls on the last heavy syllable and on the penult if there are no heavy syllables.

Khubchandani, L.M. (1969). Stress in Sindhi. Indian Linguistics 30, 115–118. Slovak [I] Indo-European, Slavic, West, Czech-Slovak. Slovakia. f f

Primary stress falls on the first syllable. Secondary stress falls on odd numbered syllables to the right.

¨nepo˙veziem

‘I will not carry’

¨nepo˙vezie˙me or ¨nepo˙vezieme ‘we will not carry’ Dogil, G. (1999). West Slavic Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 813–839. Berlin & New York: Mouton de Gruyter. Kra´lj, A. (1988). Pravidla´ slovenskej vyslovnosti. Bratislava´: Slovenske´ Pedagogicke´ Nakladatela`stvo. Letz, B. (1950). Gramatika slovenske´ho jazyka. Bratislava´: Sˇta´tne Nakladatela`stvo. Rubach, J. (1991). The Lexical Phonology of Slovak. Oxford: Oxford University Press. Slovenian; Slovene [F/L/L (tone)] Indo-European, Slavic, South, Western. Slovenia. f f f

Stress falls on the first syllable with a strong low tone. If there are no strong low tones, stress the last tone. If there are no tones, stress the last syllable.

me¨du@ø

‘honey gen.sg’

od¨me@ødu

me¨do$øv

‘honey gen.pl’

me¨do@øvih ‘honey.loc.pl.’

‘of honey gen.sg’

Gvozdanovic, J. (1999). South Slavic. In H.G. van der Hulst (ed.) Word Prosodic Systems in European Languages, 839–852. Berlin & New York: Mouton de Gruyter. Slovincian (dialect of Kashubian) [L/F] Indo-European, Slavic, West, Lechitic. West of Gdansk, Southwest of Gdynia (Poland, now extinct).

816 f f

Eurasian languages

Stress falls on the last syllable with a long vowel or lexical accent mark. If there are no such vowels, stress is initial.

¨kokoS ‘chicken’

je¨zoørka ‘lake dim.pl.’

svjiø¨niø

‘pig gen.sg.’

Kurylowicz, J. (1960). Uwaki o akcencie kaszubskim. Slavia Orientalis 20, 71–77. Lehr-Splawinski, T. (1917). Ze studio´w nad akcentem slowianskim. Cracow: Polska Akademia Umiejetnosci. Lorentz, F. (1903). Slovinzische Grammatik. St. Petersburg. Sorbian [I] Indo-European, Slavic, West, Sorbian. Eastern Germany, centering on the upper reaches of the Spree river. f f

Primary stress falls on the initial syllable. Secondary stress falls on the penultimate syllable of four syllable or longer words.

¨wojs@cojska ‘countryside’ ¨domo˙wina ‘nation’ ¨po@droz@nij˙kojstwo ‘going for a journey’ Dogil, G. (1999). West Slavic Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 813–839. Berlin & New York: Mouton de Gruyter. Janasˇ, P. (1976). Niedersorbische Grammatik. Bautzen: VEB DomowinaVerlag. Mucke, K.E. (1965). Laut- und Formlehre der Niedersrbischen Sprache. Leipzig: Zentral-Antiquariat der DDR. (Reprint of original from 1891) Sjwela, B. (1952). Grammatik der Niedersorbischen Sprache. Bautzen: VEB Domowina-Verlag. Wowcerk, P. (1951). Obersorbischen Grammatik. Volk und Wissen Verlag: Leipzig. Spanish, Castilian [U/P] Indo-European, Italic, Romance, Italo-Western, Western, Ibero-Romance, North, Central. Spain, Central þ South America, Caribbean. f

f f

Stress falls on the final syllable if that is closed or marked for accent, else on the penultimate syllable of the stem (without desinence). Final stress is oligatory when the final syllable is closed by a palatal. In verbs, stress has become morphologised.

Eurasian languages

¨termin o

‘term’

su¨til

¨util

‘useful’

bu¨fand a ‘scarf ’

817

‘subtle’

Navarro, I.T. (1965). Manual de Pronunciacio´n Espanola. Cosejo superior de Investigaciones Cientificas Instituto ‘‘Miguel de Cervantes’’ Publicaciones de la Revista de Filologia Espanola III, Madrid. Harris, J. (1983). Syllable Structure and Stress in Spanish. Cambridge Massachusetts: MIT Press. Alarcos Llorach, E. (1965). Fonologı´a Espan˜ola. Madrid: Gredos. Green, J. (1988). Spanish. In M. Harris & N. Vincent (eds.) The Romance languages, 79–130. London: Croom Helm. Harris, J.W. (1969). Spanish Phonology. Cambridge Massachusetts: MIT Press. Roca, I. (1991). Stress and Syllables in Spanish. In H. Campos & F. Martı´nez-Gil (eds.) Current Studies in Spanish Linguistics, 599–635. Washington: Georgetown University Press. Roca, I. (1999). Stress in the Romance Languages. In H.G. van der Hulst (ed.) Word Prosodic Systems in the Languages of Europe, 659–811. Berlin & New York: Mouton de Gruyter. Swedish [U%A] Indo-European, Germanic, North, East Scandinavian. Sweden, Finland. f f f

Stress falls on the final syllable if it is closed or contains a long vowel. Stress falls on the penultimate syllable if it is heavy and the final is not. If both the final and the penult are light, stress is antepenultimate.

tEk¨nik ‘technique’ in¨fErno ‘inferno’

bu¨feø ‘bu¤et’ ¨vide.o ‘video’

Bruce, G. (1999). Swedish. In H. van der Hulst (ed.) Word Prosodic Systems in the languages of Europe, 605–633. Berlin & New York: Mouton de Gruyter. Tajik (dialect of Persian) [U] Indo-European, Indo-Iranian, Iranian, Western, Southwestern, Persian. Iran, Afghanistan, Tadzhik SSR, Iraq, Persian Gulf Islands. f

Stress falls on the final syllable.

818

Eurasian languages

Rastorgueva, V.S. (1963). A Short Sketch of Tajik Grammar. Published by the Indiana University Linguistics Club, Bloomington. Welsh; Cymraeg; Cymric [P] Indo-European, Celtic, Insular, Brythonic. Wales (UK), Canada. f

Stress falls on the penultimate syllable.

k«¨muçnas ‘pleasure’

k«v¨leçs ‘convenient’

Ball, M.J. (ed.) (1993). The Celtic Languages. London/New York: Routledge. Jones, J.M. (1955). A Welsh Grammar. Oxford: Clarendon Press. Jones, J.M. & A.R. Thomas (1977). The Welsh Language: studies in its syntax and semantics. Cardi¤: University of Wales Press. MacAuley, D. (ed.) (1992). The Celtic Languages. Cambrdige: Cambridge University Press. Watkins, T.A. (1992). Kurze Beschreibung des Kymrischen. Innsbruck. Williams, S.J. (1980). A Welsh Grammar. Cardi¤: University of Wales Press. Uralic Estonian [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Balto-Finnic. Estonia. f f f

Primary stress falls on the first syllable, exceptions are rare. Secondary stress usually on alternates after the primary stress. If the secondary stress would fall on a short vowel, the pattern might become ternary.

¨sada ‘hundred’ ¨kahekesi ‘together, two at a time’

¨suletud

‘closed’

Eek, A. (1975). Observations on the durations of some word structures: II. Estonian papers in Phonetics 4. Academy of sciences of the Estonian SSR, Institute of Language and Literature, 7–54. Hint, M. (1973). Eesti Keele So˜nafonoloogia I. Eesti NSV Teaduste Akadeemia, Talinn, Estonia.

Eurasian languages

819

Finnish [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Balto-Finnic. Finland, Russian Federation. f f

Primary stress falls on the first syllable. Secondary stress usually on alternates after the primary stress, but never final.

¨lope˙teta ‘finish[neg.]’ ¨teuras˙tamo ‘slaughterhouse’ ¨opet˙tele˙manani ‘as something i have been learning’ Fromm, H. (1982). Finnische Grammatik. Heidelberg: C. Winter, Universitatsverlag. Hungarian [I] Uralic, Finno-Ugric, Ugric, Hungarian. Hungary. f f f

f

Primary stress falls on the first syllable. Secondary stress usually on alternates after the primary stress. If the secondary stress would fall on a short vowel, the pattern might become ternary. In the latter case, secondary stress is also reported to avoid final syllables.

¨alma ¨tEriø˙tOøvEl

‘apple’ ‘with tablecloth’

¨kpaø˙vl ‘with who’ ¨feølE˙mElE˙tEn ‘on mezzanine’

Kerek, A. (1971). Hungarian Metrics: Some Linguistic Aspects of Iambic Verse. Indiana University Publications, Uralic and Altaic Series 117. The Hague: Mouton. Szinnyei, J. (1912). Ungarische Sprachlehre. Berlin: Go¨schen. Karelian [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Balto-Finnic. Karelian Republic (Russian Federation), Finland. f f

f

Primary stress falls on the first syllable. Secondary stress usually on alternates after the primary stress, but never final. In some cases the secondary stress pattern is reported to be ternary.

820

Eurasian languages

¨ ber die Phonemsystem der Karelischen Sprache. Leskinen, H. (1984). U In P. Hajdu´ & L. Honti (eds.) Studien zur phonologischen Beschreibung uralischer Sprachen, 247–257. Budapest: Akade´miai Kiado´. Raun, A. (1964). Karelian Survey. Reasearch and Studies in Uralic and Altaic Studies 9. Cleveland: Bell & Howell, Ohio. Komi-Permyak [F/L] Uralic, Finno-Ugric, Finno-Permic, Permic. Komi Republic (Russian Federation). f f

Stress falls on the first syllable with a long vowel. If there are no long vowels, stress is final.

Austerlitz, R. (1964). Permian (Votyak-Zyrien) manual. Research and Studies in Uralic and Altaic languages 64. Cleveland: Bell & Howell, Ohio. ¨ ber die Betonungsverha¨ltnisse in den FinnischItkonen, E. (1955). U Ugrischen Sprachen. Acta Linguistica Academiae Scientarium Hungaricae 5: 21–34. Lytkin, V.I. (1961). Komi-iazvinskii dialekt. Moscow: Izdatel’stvo Akademii Nauk. Liv; Livonian [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Balto-Finnic. Latvia. f f

Primary stress falls on the first syllable. Secondary stress occurs on the third syllable, tertiary stress on the fifth (if present).

Sjogren, J.A. (1861). Livische Grammatik Nebst Sprachproben. Commisionare der Kaiserlichen Akademie der Wissenschaften, St. Petersburg. Mansi; Vogul [I] Uralic, Finno-Ugric, Ugric, Ob-Ugric. Along the Ob-river in Western Siberia (Russian Federation). f f f

Primary stress falls on the first syllable. Secondary stress usually on alternates after the primary stress. Sources di¤er on whether the final syllable can be stressed or not.

¨tAørAtI

‘to leave, to let’

¨tAørA˙tANkwe ‘to leave, to let.inf.’

Eurasian languages

821

Ka´lma´n, B. (1965). Vogul Chrestomathy. Indiana University, Bloomington, Indiana. Lako´, G. (1957). Nordmansische Sprachstudien. Acta Linguistics Academiae Scientarum Hungaricae 6: 347–423. Rombandeeva, E.I. (1973). Mansijskij (Vogul’skij) Jazyk. Moscow: Isdatel’stvo ‘‘Nauka’’. Mari, High; Mountain (Western) Cheremis [L/L or P/P] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Cheremisic. GornoMariy, and some in Bashkortostan. (Russian Federation). f

f

Sources di¤er. Some quote stress is penult except when the antepenultimate vowel is full and the penult is not. Others quote stress occurs on the last non-final long vowel or the penultimate syllable. Examples point in the direction of the latter.

aa¨Baaxaa ‘pod’ ¨kaanEsEr ‘sorrow’

¨«Sta_S Bˆlaa¨˜aanEs&tEs&E

‘to sweep’ ‘comedian’

¨ ber die Betonungsverha¨ltnisse in der FinnischItkonen, E. (1955). U Ugrischen Sprachen. Acta Linguistica Academiae Scientarium Hungaricae 5: 21–34. Kangasmaa-Minz, D. (1998). Mari. In D. Abondolo (ed.). The Uralic Languages. Routledge language family descriptions. London: Routledge. Mari, Low; Meadow (Eastern) Cheremis [L/F;IRR] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Cheremisic. GornoMariy, and some in Bashkortostan (Russian Federation). f

f

Stress the last full vowel, and the first in words with only reduced vowels. Free variation of final and non-final stress reported.

Slaa¨paaZ«m or Slaapaa¨Zem ‘his hat.acc’ ¨t«l«z«n or t«l«¨z«n ‘moon.gen’ ¨pu_g«lmo_ or pu_g«l¨mo_ ‘cone’ ¨BoSt«l ‘laugh’ BoSt«¨lam ‘I laugh’ Gruzov, L.P. (1960). Sovremennyi marijskij jazyk: fonetika. Joskar-Ola. Ingemann, F. & T. Sebeok (1961). An Eastern Cheremis Manual. Indiana University, Bloomington, Indiana.

822

Eurasian languages

¨ ber die Betonungsverha¨ltnisse in der FinnischItkonen, E. (1955). U Ugrischen Sprachen. Acta Linguistica Academiae Scientarium Hungaricae 5: 21–34. Kangasmaa-Minz, D. (1998). Mari. In D. Abondolo (ed.). The Uralic Languages. Routledge language family descriptions. London: Routledge. Timofeeva, V.T. (1961). Sovremennyi marijskij jazyk: sintaksis sloznogo predlozenija. Joskar-Ola. Moksha; Mordvin; Mordva [F/F] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Mordvinic. Mordvin Autonomous Republic (Russian Federation). f

f

Stress the first full vowel, and the first in words with only reduced vowels. Free variation of final and non-final stress reported.

tu¨cJa_nJa_ ‘cloud’

¨puv«nd«ms ‘to press’

¨ ber die Betonungsverha¨ltnisse in der FinnischItkonen, E. (1955). U Ugrischen Sprachen. Acta Linguistica Academiae Scientarium Hungaricae 5: 21–34. Koljadenkov, M.N. (1959). Struktura prostogo predlozenija v mordovskix jazykax. Saransk. Paasonen, H. (1909). Mordwinische Chrestomathie mit Glossar und grammatikalischem Abriss. Helsinki: Finnisch-Ugrische Gesellschaft. Raun, A. (1964). Mordvin Manual. (Research and Studies in Uralic and Altaic Languages, project 39). Cleveland, Ohio: Bell & Howell. Tsygankin, D.B. & C.Z. Dabaev (1975). Ocerk sravitel’noj grammatiki mordovskix (moksanskogo i erz’anskogo) literaturnix jazykov. Saransk. Zavodova, R.A. & Koljadenkov, M.N. (eds.) (1964). Grammatika mordovskix (moksanskogo i erzjanskogo) jazykov. Nenets, Tundra [I] Uralic, Samoyedic, Northern Samoyedic. Northwest Siberia (Russian Federation). f f f

Primary stress is initial. Secondary stress occurs on all syllables preceding a syllable with schwa. Secondary stress also falls on non-final syllables preceded by an unstressed syllable.

Eurasian languages

¨ya

‘earth’

¨xampol ‘litter’

¨wada

823

‘word’

¨xOrOr« ‘knife poss. nom.sg’

Salminen, T. (1999). http://www.helsinki.fi/~tasalmin/tn.html. Ruija (dialect of Northern Saami) [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Lappic, Northern. Norway, Sweden, Finland. f f f

Primary stress is initial. Secondary stress on odd non-final syllables. A weak final secondary stress may occur in trisyllables.

¨ba#tta˙raddat ‘to fly continually’

¨gapperist or ¨gappe˙rist ‘in the cup’

Itkonen, E. (1958). Uber die Betonungsverhaltnisse in den FinnishUgrischen Sprachen. Acta Linguistica Academiae Scientarium Hungaricae 5: 21–34. Taz Selkup (dialect of Selkup) [I/I] Uralic, Samoyedic, Southern Samoyedic. Yamalo-Nenets District, Krasnoyarski krai and Tomskaya oblast (Russian Federation). f

f

Primary stress is initial, except when the first vowel is short and the second is long. Secondary stress on odd syllables after the main stress.

¨syr«

¨syøre

‘snow’

¨ta_˙t«ki˙ne ‘reindeer.lat.dat.pl.’

‘cow’

am¨qeøNa ‘is to take’

McNaughton, A.R. (1976). Taz Selkup: The phonology and morphology of the verb. Ph.D. Dissertation, Colombia University. Vod; Votic [I] Uralic, Finno-Ugric, Finno-Permic, Finno-Cheremisic, Finno-Mordvinic, Finno-Lappic, Balto-Finnic. Ingria, near Estonia and Leningrad (Russian Federation). f f

Primary stress is initial. Secondary stress on odd syllables after the main stress, but not on case su‰xes.

¨suvaø˙miøn ‘loving’

¨tu_tøa_˙riøkøo ‘girl’

824

Eurasian languages

Ariste, P. (1968). A Grammar of the Votic Language. Indiana University, Bloomington. Yazva (dialect of Komi-Zyrian) [F/L] Uralic, Finno-Ugric, Finno-Permic, Permic. Komi Republic (Russian Federation). f f f

Stress on the first heavy vowel. Heavy syllables have low vowels. Weight of high vowels fluctuates. Words without heavy syllables have final stress.

Austerlitz, R. (1964). Permian (Votyak-Zyrien) manual. Research and Studies in Uralic and Altaic languages 64. Cleveland: Bell & Howell, Ohio. ¨ ber die Betonungsverha¨ltnisse in den FinnischItkonen, E. (1955). U Ugrischen Sprachen. Acta Linguistica Academiae Scientarum Hungaricae 5: 21–34. Lytkin, V.I. (1961). Komi-iazvinskii dialekt. Moscow: Izdatel’stvo Akademii Nauk.

Asian languages [For Indo-Aryan languages see European Languages section] Altaic Bashkir; Basquort [U] Altaic, Turkic, Western, Uralian. Bashkir Autonomous Republic (Russian Federation). f

Stress falls on the final syllable of the word (including su‰xes).

kHi¨tHApH ‘book’

kHitHApH¨Ar ‘books’ kHitHApHlArF¨bFD ‘our books’

Dmitriev, N.K. (1948). Grammatika bashkirskogo jazyka. Moscow. Poppe, N. (1964). Bashkir Manual. Indiana University, Bloomington. Chuvash [L/F] Altaic, Turkic, Bolgar. Chuvash republic (Russian Federation). f

Stress falls on the last syllable with a full vowel, else on the first syllable.

la¨Sa ‘horse’ sarla¨ka ‘widely’

¨ala(k ‘door’ ¨e(sle(pe(r ‘we shall work’

Kruger, J.R. (1961). Chuvash Manual. Indiana University, Bloomington, Indiana. Evenki; Tungus [U/U] Altaic, Tungus, Northern, Evenki. Russia, China, Mongolia. Tentative f In words of two syllables, stress mostly falls on the second syllable. f If a disyllabic word has either a long vowel or cluster of consonants in the middle of the stem, stress falls on the first syllable. f In a disyllabic word with two long vowels, stress falls on the second syllable. f If a word has more than two syllables with short vowels, stress falls on the last syllable. f Some polysyllabic words with final stress have an antepenultimate secondary stress.

826

Asian languages

bi¨ra

‘river’

o¨ron

¨iøkte

‘tooth’

moø¨kaøn ‘stick’

‘reindeer’

¨halka

‘hammer’

˙ngina¨kin ‘dog’

Nedjalkov, I. (1996). Evenki. London: Routledge. Mongolian; Khalkha [F/F] Altaic, Mongolian, Eastern, Oirat-Khalkha, Khalkha-Buriat, Mongolian Proper. Mongolia. f f

Stress falls on the first syllable that contains a long vowel. Otherwise stress falls on the first syllable of the word.

bos¨guul ‘fugitive’

¨axe

¨uNSis«N ‘having read’

mori¨ooroo ‘by means of his own horse’

‘elder brother’

Poppe, N. (1970). Mongolian Language Handbook. Washington: Center for Applied Linguistics. Strees, J.C. (1963). Khalkha Structure. Indiana University, Bloomington, Indiana. Orok; Oroc [U] Altaic, Tungus, Southern, Southeast, Nanaj. Eastern Siberia (Russian Federation). f

Stress falls on the final syllable of the word.

Avrorin, V.A. & E.P. Lebedeva (1968). Orocsky Jazyk. In V.V. Vinogradov (ed.) Jazyki Narodov SSSR V: Mongolskie Tungusko-Manczurskie i Paleoaziatskie Jazyki. Moscow: Nauka, 191–209. Turkish; Osmanli [F/L;LEX] Altaic, Turkic, Southern, Turkish. Turkey, Bulgaria, Cyprus, Russian Federation. f

Stress normally falls on the final syllable. Stress may also appear anywhere in the word on lexically marked syllables. f In placenames and loans stress is antepenultimate if that syllable contains a long vowel or is closed and the penult is light, otherwise stress falls on the penultimate syllable. ta¨nfi ‘know’ tançdçkla¨rçm ‘my acquaintances’ tanfi¨dfik ‘acquaintance’ ta¨nçmadçklarçmçz ‘those we do not know’ tançdçk¨lar ‘acquaintances’ ak¨Samleyin ‘at evening’ is¨tanbul ¨ankara f

Asian languages

827

Kornfilt, J. (1997). Turkish. London: Routledge. Lees, R. (1961). The Phonology of Modern Standard Turkish. Uralic and Altaic Series 6. Indiana University Publications, Indiana. Sezer, E. (1983). On Non-final Stress in Turkish. In: Journal of Turkish Studies 5, 61–69. Underhill, R. (1976). Turkish Grammar. Cambridge: MIT Press. Uzbek, Northern [U] Altaic, Turkic, Eastern. Uzbekistan, Turkmenistan, Kirgizistan, Tadzhikistan, Kazakhstan (Russian Federation). f f

f

Primary stress normally falls on the final syllable. In words of three syllables a secondary stress may appear on the first syllable. In longer words a tertiary stress can appear in between the primary and secondary stress.

dZ¨nim ‘my soul’

quS¨ca ‘little bird’

Raun, A. (1969). Basic Course in Uzbek. Indiana University, Bloomington. Austro-Asiatic Bhumij; Mundari [U/P] Austro-Asiatic, Munda, North Munda, Kherwari, Mundari. Mayurbhanj District, Orissa and Singhbhum District, Bihar. India. f f

If the final syllable is closed, it is accented. Else the penultimate syllable is accented.

¨lija

‘a cloth’

Cook, W.A. (1966). A descriptive analysis of Mundari. Ph.D. Thesis, Georgetown University. Halang; Koyong [U] Austro-Asiatic, Mon-Khmer, Eastern Mon-Khmer, Bahnaric, North Bahnaric, West, Jeh-Halang. Vietnam, Laos. f

Stress is located on the final syllable.

Cooper, J. & N. Cooper (1966). Halang Phonemes. Mon-Khmer Studies II, Linguistic Circle of Saigon, Pub. No. 3.

828

Asian languages

Khasi [U] Austro-Asiatic, Mon-Khmer, Northern Mon-Khmer, Khasian. Assam. India Bangladesh. f

f

In isolation words have one stress which coincides with the falling pitch on the final syllable. In context these stresses and pitch contours are deleted on all words but the final. Status of ‘‘stress’’ in Khasi highly debatable.

¨khlaa ‘tiger’

paa¨troy ‘pull by the hair’

tara¨jur

‘scales’

Rabel, L. (1961). Khasi, a Language of Assam. Baton Rouge: Louisiana State University Press. Khmer, Central; Cambodian [U] Austro-Asiatic, Mon-Khmer, Eastern Mon-Khmer, Khmer. Cambodia, Thailand, Vietnam, Laos. f

f f

f

Disyllabic words consist of an unstressed pre-syllable followed by a stressed full syllable. Trisyllabic words exist; they contain two pre-syllables. Words with more full syllables are polymorphemic and carry more stresses. Status of ‘‘stress’’ in Cambodian might be debatable.

kffin¨laeN ‘place’

krakh¨wak ‘dirty’

ura¨moh ‘house’

Gorgoniyev, Y.A. (1966). The Khmer Language. Moscow: Nauka. Jacob, Judith M. (1968). Introduction to Cambodian. London: Oxford University Press. Gri‰th, T. (1991). Cambodian as an iambic language. Ms. Department of Linguistics, University of California, Irvine. Hu¤man, F.E. (1970). Modern Spoken Cambodian. New Haven: Yale University Press. Nacaskul, K. (1978). The Syllabic and Morphological Structure of Cambodian Words. Mon-Khmer Studies 7, 183–200. Khmu’ [U] Austro-Asiatic, Mon-Khmer, Northern Mon-Khmer, Khmuic, Mal-Khmu’, Khmu’. Thailand, Laos, Vietnam. f

Like in Cambodian one or two pre-syllables and a final full syllable make up a word.

Asian languages f

f

829

Stress is located on the full syllable. Words with two full syllables are polymorphemic and carry two stresses. Status of ‘‘stress’’ in Khmu’ might be debatable.

kǬt$N

‘jar’

s«m¨l" $øk ‘fish scale’

c«r¨la$ø¨ta@øp ‘butterfly’

Premsrirat, S. (1987). A Khmu Grammar. Papers in South-East Asian Linguistics, No. 10, Pacific Linguistics, Series A, No. 75. Australian National University, Canberra. Sedang [U] Austro-Asiatic, Mon-Khmer, Eastern Mon-Khmer, Bahnaric, North Bahnaric, West, Sedang-Todrah, Sedang. Vietnam. f f f

¨a

Phonological words are maximally bisyllabic. Stress is located on the final, or the only, syllable. Status of ‘‘stress’’ in Sedang unclear. ‘i’

ko¨blin ‘to be very full’

Smith, K.D. (1979). Sedang Grammar, Phonology and Syntactic Structure. Pacific Linguistics, Series B, No. 50. Australian National University, Canberra. Chukotko Chukot; Chuckchee [I;S] Chukotko-Kamchatkan, Northern, Chukot. Chuckchi peninsula, northeastern Siberia. f f

Accent usually occurs at the beginning of the word. To give emphasis to the word, the accent may be thrown upon the last syllable, the vowel of which then changes to /o/.

¨pari´n)in ‘shoulder blade’ u¨wema ‘while cooking’

pi¨n)epi ‘snowstorm’ ¨pirirkin ‘he takes’

Bogoras, W. (1922). Chuckchee. In F. Boaz (ed.) Handbook of American Indian languages, Part 2, 631–903. Smithsonian Institution Bureau of American Ethnology Bulletin 40. Washington: Government Printing O‰ce. Dravidian Koi; Koya (dialect of Gondi) [I] Dravidian, South-Central, Gondi-Kui, Konda-Kui, Manda-Kui, Kui-Kuvi. Madhya Pradesh (India).

830 f f

Asian languages

Primary stress falls on the initial syllable. Secondary stress on syllables with a closing consonant or long vowel.

¨aaki ‘leaf ’

¨ginne ‘cup’

¨«n˙dooru ‘everyone’

Tyler, S.A. (1969). Koya: An outline Grammar. University of California Publications in Linguistics, No. 54. Berkerly and LA: University of California Press. Malayalam [I/I] Dravidian, Southern, Tamil-Kannada, Tamil-Kodagu, Tamil-Malayalam, Malayalam. Kerala (India). f

f

Primary stress falls on the initial syllable, except when the first vowel is short and the second is long. Secondary stress on syllables with a long vowel.

¨kuuam ‘crowd’ ¨muta˙laaæi ‘boss’

¨pukavaå i ‘train’ pa¨aaæak˙kaaran ‘soldier’

Asher, R.E. & T.C. Kumari (1997). Malayalam. London: Routledge, New York. Sino-Tibetan Bawm [U] Sino-Tibetan, Tibeto-Burman, Baric, Kuki-Naga, Kuki-Chin, Central, Unclassified. India, Myanmar, Bangladesh. f

Southern Bawm has tone. Northern Bawm has final stress.

nu¨pi

‘wife’

nu¨fen ‘skirt’

fa¨nu

‘daughter’

Reichle, V. (1981). Bawm Language and Lore. Bern: Peter Lang. Chepang [I] Sino-Tibetan, Tibeto-Burman, Bodic, Eastern Himalayan, Kiranti, Western, Marginal Western. Nepal, central hills. f

Stress is initial.

¨chaN ‘shelf ’

¨jiksa ‘to be sick’

¨sipru

‘snake’

Caughley, R.C. (1982). The Syntax and Morphology of the Verb in Chepang. Pacific Linguistics, Series B, No. 84. Australian National University, Canberra.

Asian languages

831

Chinese, Mandarin [LEX] Sino-Tibetan, Chinese. China. f

Syllables with tone may also carry stress, but do not have to. Bisyllabic words with two tones have either initial primary stress or initial secondary stress and final primary stress. f Longer forms might not be true words, and have diverse stress patterns. f Status of ‘‘stress’’ in Chinese highly debatable. ¨bo li ‘glass’ ˙ma¨&da@ ‘motor’ ˙ha@n˙shu@¨bia&o ‘thermomenter’ ¨wu zili ‘in the room’ f

Henne, H., O.B. Rongen & L.J. Hansen (1977). A handbook on Chinese language structure. Oslo: Universitetsforlaget. Tibetan; Lhasa [F/F] Sino-Tibetan, Tibeto-Burman, Bodic, Bodish, Tibetan, Central. Tibet. f f

Stress the leftmost syllable with a long vowel, else the leftmost. Alternative pattern of two equal stresses in bisyllables with two long vowels reported.

qhap¨tee ‘rival’ ¨Naamo ‘sweet’

¨qo_la ‘charcoal’ ¨k«p ‘behind’

Goldstein, M.C. & N. Nornang (1978). Modern Tibetan: Lhasa Dialect. Kathmandu: Ratna Pustak Bhandar. Yukaghir Yukaghir [L/L] Yukaghir. Yakutia and the Kamchatka Peninsula (Russian Federation). f

f f

Stress falls on the last syllable that is closed by a consonant or has a long vowel. Else stress falls on the last syllable of the word. There are exceptional bisyllabic (C)VCe forms with initial stress.

¨aøs’e ‘domestic deer’ a¨roøje ‘kind of fish’ mo¨ro ‘hat’

le¨gul ‘food’ ¨a#olhoro ‘hare’ co_bi¨ne ‘spear’

Maslova, E. (2003). A Grammar of Kolyma Yukaghir. Berlin: Mouton de Gruyter.

832

Asian languages

Isolates Burushaski, Yasin (dialect of Burushaski) [LEX] Isolated; Unknown. Hunza, Nagir states and Jasin district of Pakistan. f

Stress is lexically determined, and plays a role in the grammar.

di¨cilikini ‘he hung’

¨dicilikini ‘he would hang’

ga¨li

¨gali

‘he went’

‘it broke’

Berger, H. (1974). Das Yasin-Burushaski. Wiesbaden: Harrassowitz. Morgenstierne, G. (1942). Notes on Burushaski Phonology. In a Tripet of Burushaski Studies. Norsk Tideskrift for Sprogvidenskap. Bind XIII, Soertrykk. Gilyak; Nivkh [I;LEX] Language Isolate. Sakhalin Island, Russia. f f f

Stress often falls on the first vowel of the word. Sometimes stress is lexically determined. Su‰xes of the vocative, the imperative, and the conditional converb, are always stressed.

¨xaunt ‘(someone) calls’ ymyk¨a ‘mother.voc’

xa¨unt vi¨ja

‘(someone) dries’ ‘go.sec.sg.imp’

Gruzdeva, E. (1998). Nivkh. Languages of the world materials 111. Mu¨nchen: Lincom Europa.

Middle Eastern and African languages

Afro-Asiatic Arabic, Bani Hassan [L/F] Afro-Asiatic, Semitic, Central, South, Arabic. Southwest Sahara. f

f

Stress the rightmost syllable with a long vowel or coda, otherwise the first. Long vowels preceding the main stress have secondary stress.

Irshied, O. & M. Kenstowicz (1984). Some phonological rules of BaniHassan Arabic: a Bedouin dialect. Studies in the linguistic sciences 14.1, pp. 109–148. Arabic, Bedouin Hijazi [U-P/A] Afro-Asiatic, Semitic, Central, South, Arabic. North red sea coast, Sinai. f

f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult.

mak¨tuøb ‘written’ mak¨tuøfah ‘tied.f.s.’

Da¨rabt ‘i hit’ ¨maølana ‘our property’ gaø¨bilna ‘meet us.m.s.’

Al Mozainy, H. (1981). Vowel alternations in a Bedouin Hijazi Arabic Dialect: Abstractness and Stress. Ph.D. Dissertation, University of Texas, Austin. Arabic, Beirut (dialect of Arabic, North Levantine) [U-P/A] Afro-Asiatic, Semitic, Central, South, Arabic. Beirut. f

f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. In other cases, stress falls on heavy penults, otherwise on the antepenult.

¨darab ¨darab u

‘hit’ ‘they hit’

da¨rab na al¨lam na

‘he hit’ ‘we teach’

s&a¨jaratun ‘tree’ ¨ allam u ‘they teach’

Abdul-Karim, K. (1979). Aspects of the Phonology of Lebanese Arabic. Ph.D. Dissertation, University of Illinois, Urbana.

834

Middle Eastern and African languages

Arabic, Classical [L/F] Afro-Asiatic, Semitic, Central, South, Arabic. Northern Africa, Middle East. f

f

Stress the rightmost non-final syllable that has a long vowel or closing consonant. If there are no such syllables, stress the first.

ki¨taabun

‘book’

¨balatahun ‘date’

¨mamlakatun

‘kingdom’

ta¨mamtumaa ‘you both were completely’

McCarthy, J. (1979). Formal Problems in Semitic Phonology and Morphology. Unpublished doctoral dissertation, MIT Cambridge, Massachusetts. McCarthy, J. (1979). On Stress and Syllabification. Linguistic Inquiry 10, pp. 443–466. Wright, W. (1967). A Grammar of the Arabic Language. Cambridge: Cambridge University Press. Arabic, Damascene (dialect of Arabic, North Levantine) [U-P/A] Afro-Asiatic, Semitic, Central, South, Arabic. Damascus (Syria). f

f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. In other cases, stress falls on the penult if it is heavy, otherwise on the antepenult.

ma¨daøres ‘schools’ mut¨tahide ‘united’

dar¨rast ‘i/you masc.sg. taught’ al¨lamna ‘teach 1pl.’

McCarthy, J. (1980). A Note on the Accentuation of Damascene Arabic. Studies in the Linguistic Sciences 10.2, pp. 77–99. Arabic, Egyptian Radio (dialect of Arabic, Egyptian Spoken) [L (CNT)] Afro-Asiatic, Semitic, Central, South, Arabic. Egypt. f

f

f

f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. Otherwise, stress penultimate syllables if heavy, and always in bisyllabic words. Three syllable words with only light syllables have antepenultimate stress. Four syllable words with only light syllables have antepenultimate or pre-antepenultimate stress in free variation.

Middle Eastern and African languages f

f f

835

If the antepenult is heavy and the penult and final light, either it or the penult is stressed. If the pre-antepenult is heavy, stress is always antepenultimate. Rhythm is assigned to non-final heavy syllables and odd light ones from left to right.

sa¨laøm

‘peace’

˙qad¨damna ‘we presented’ ¨malik ‘king’

˙kata¨badu or ¨kata˙badu

‘he wrote it’

¨/abadan ˙mux¨talifa

‘never’ ‘di¤erent’

Harrel R.S. (1960). The phonology of Colloquial Egyptian Arabic. In C.A. Fergusson (ed.). Contributions to Arabic Linguistics, 3–77. Cambridge: Harvard University Press. Arabic, Egyptian Spoken; Cairene Arabic (classical) [L (CNT)] Afro-Asiatic, Semitic, Central, South, Arabic. Egypt, Tunesia. f f f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. Otherwise, stress penultimate syllables if heavy. In all other cases, stress the penult or antepenult, whichever is separated from the last heavy syllable, or the left word edge, by an even number of syllables (or zero).

¨buxala ‘misers’ mux¨talifa ‘di¤erent f.sg.’ Sajarata¨humaa ‘their dual tree nom.’

ca¨malti mar¨taba saka¨kiin

‘you f.sg. did’ ‘mattress’ ‘knives’

McCarthy, J. (1979). Formal problems in Semitic phonology and morphology. Unpublished doctoral dissertation, MIT Cambridge, Massachusetts. McCarthy, J. (1979). On Stress and Syllabification. Linguistic Inquiry 10, pp. 443–466. Arabic, Gulf [U/P] Afro-Asiatic, Semitic, Central, South, Arabic. Saudi Arabia, Quatar, Oman etc. f f

If the vowel of the final syllable is long, it bears stress. In all other cases stress is on the penult.

Ka¨yaattin ‘their life’ ySuu¨fuun ‘they see’ ¨qaæbitS ‘your heart’

¨beettum ‘their house’ mus¨taSfa ‘hospital’ mithaaw¨Siin ‘having quarreled with each other’

836

Middle Eastern and African languages

Qafisheh, H.A. (1977). A short reference grammar of Gulf Arabic. Tucson: University of Arizona Press. Arabic, Lebanese (dialect of Arabic, North Levantine) [P/A] Afro-Asiatic, Semitic, Central, South, Arabic. Lebanon. f

f

Stress is penultimate in bisyllables, and in longer words if it has a long vowel or closing consonant. In all other cases, stress is on the antepenultimate syllable.

Kenstowicz, M.J. & K. Abdul-Karim (1980). Cyclic Stress in Levantine Arabic. Studies in the Linguistic Sciences 10.2. Department of Linguistics, University of Illinois, Urbana. Arabic, Libyan spoken; Cyrenaican Bedouin [L (CNT)] Afro-Asiatic, Semitic, Central, South, Arabic. Eastern Lybia. f

f

f

f

Bisyllables of which the first syllable is open and the vowel is short have final stress. Superheavy (long vowel þ coda, or double coda) final syllables carry stress. Otherwise, stress heavy penults or heavy antepenults followed by two light syllables. Otherwise, stress the penult or the antepenult, whichever is separated from the closest preceding heavy syllable or the left word edge by an odd number of syllables.

ta¨raøfigan ‘they accompanied’ ki¨tab ‘he wrote’ ma ri¨kitha ‘her quarrel’

¨maktab ‘o‰ce’ ki¨tabtan ‘you wrote’ fana¨Ziøl ‘cups’

Owens, J. (1984). A short reference grammar of eastern Lybian Arabic. Wiesbaden: Otto Harrassowitz. Arabic, Negev Bedouin [L (CNT)] Afro-Asiatic, Semitic, Central, South, Arabic. Negev Desert (East Israel). f f f

Stress the last syllable with a long vowel or vowel cluster. Otherwise stress the second syllable. There is some variation in four syllable words.

ga¨da ‘lunch’ fa¨rasah ‘his horse’

a¨naam ‘sheep’ ¨ankital ‘he was killed’

Middle Eastern and African languages

837

Blanc, H. (1971). The Arabic Dialect of the Negev Bedouins. In: Proceedings of the Israel Academy of Sciences and Humanities, Iv. 7, pp. 112–150. Arabic, South Levantine Spoken; Palestinian Arabic [L (CNT)] Afro-Asiatic, Semitic, Central, South, Arabic. Middle East, esp. West Bank and Gaza. f

f f f f

Superheavy (long vowel þ coda, or double coda) final syllables carry stress. Otherwise, stress penultimate syllables if heavy, and in bisyllabic words. Heavy antepenults followed by two light syllables carry stress. Four syllable words with only light syllables have initial stress. All other words ending in three light syllables have antepenultimate stress.

da¨rast

‘I studied’

¨katab ‘he wrote’ ¨s&ajaratun ‘a tree’

¨maktab

‘o‰ce’

¨baørako ‘he blessed him’ s&aja¨ratuhu ‘his tree’

Abu-Salim, I.M. (1980). Epenthesis and geminate consonants in Palestinian Arabic. Studies in the Linguistic Sciences 10.2, 1–11. Dept. of Linguistics, University of Illinois, Urbana. Abu-Salim, I. (1982). A Reanalysis of Some Aspects of Arabic Phonology: A Metrical Approach. Ph.D. Dissertation, University of Illinois. Brame, M. (1973). On stress assignment in two Arabic dialects. In S.R. Anderson & P. Kiparsky (eds.). A festschrift for Morris Halle, 14–25. Holt, Reinhart & Winston, New York Brame, M. (1974). The Cycle in Phonology: Stress in Palestinian, Maltese and Spanish. Linguistic Inquiry 5, pp. 39–61. Kenstowicz, M. & K. Abdul-Karim (1980). Cyclic Stress in Levantine Arabic. Studies in the Linguistic Sciences 10.1, pp. 55–76. Aramaic [P] Afro-Asiatic, Semitic, Central, Aramaic. East Kurdistan, Israel, Palestine, Sinai. f f

Stress falls the penultimate syllable. Epenthetic vowels do not count for stress assignment.

¨/arba ‘four’

c&ay¨xaøna ‘teashop’

da¨qiqa ‘fine’

Khan, G. (1999). A grammar of neo-Aramaic: the dialect of the Jews of Arbel. Leiden: Brill.

838

Middle Eastern and African languages

Arsanis, G.V. (1968). Sovremenny Assirijskij Jazyk. In: Vinogradov (ea) Jazyki Narodov SSSR V: Mongol’skie, Tungusko-man’czurskie i paleoaziatskie Jazyki. Nauka, Moscow, pp. 489–507. Assyrian Neo Aramaic [P] Afro-Asiatic, Semitic, Central, Aramaic, Eastern, Central, Northeastern. Iraq, Iran, Syria, Turkey, Georgia, Armenia, Azerbaijan. f f

Stress falls the penultimate syllable. Unstressed syllables often have reduced vowel quality or are deleted entirely.

a_r¨milta_ ‘widow’

qir¨qipta

‘skull’

Tsereteli, K.G. (1978). The Modern Assyrian Language. Moscow, Nauka. Beja; Bedawi; Bischari [A/P] Afro-Asiatic, Cushitic, North. Sudan, Eritrea, Egypt. f f f

Stress falls on heavy penultimate syllables. If the penult is light, sterss falls on the antepenultimate syllable. In bisyllabic words stress is penultimate unless the penult is light and the final is not.

ga¨naøj ‘gazelle’ ¨an/alan ‘i cursed’

¨enton ‘here’ sa¨noøku ‘your brother’

Almkvist, H. (1881). Die Bischari-Sprache, Vols. 1 and 2. Upsala: Druck der Akademischen Buchdruckerei. Gorowa [F/L] Afro-Asiatic, Cushitic, South. Kondoa Central Province, Mbulu District (Tanzania). f f

Stress the leftmost long vowel or diphthong. Else, stress a final closed syllable, otherwise the penult.

¨duugonoo ‘thumb’

giram¨booda ‘snu¤ ’

am¨rami

axe¨mis

‘ivory arm ring’

‘hear’

Seidel, A. (1900). Die Sprache von Ufromi in Deutsch-Ostafrika. In: Zeitschrift fur Afrikanische und Oceanische Sprachen 5, pp. 165–175.

Middle Eastern and African languages

839

Hebrew, Modern [U;P] Afro-Asiatic, Semitic, Central, South, Canaanite. Israel. f

f

Primary stress is mostly final, but sometimes phonemically on the penult. Secondary stress on alternate syllables to the left of the main stress.

g&a¨dol

‘big’

ta¨am ‘tasted’

me˙vuga¨rim

‘adults’

¨taam ‘taste’

Bolozky, S. (1982). Remarks on Rhythmic stress in Modern Hebrew. Journal of Linguistics 18, pp. 275–289. Glinert, L. (1989). The Grammar of Modern Hebrew. Cambridge: Cambridge University Press. Hebrew, Tiberian (dialect of Hebrew; Ivrit) [U/P] Afro-Asiatic, Semitic, Central, South, Canaanite. Lake Tiberia (Israel). f f f

Main stress is assigned to the final syllable if it is closed. Otherwise stress is penultimate (but may shift again to the final syllable). Secondary stress two syllables to the left of the main stress and long vowels before that.

¨qaømuø

‘arise third.pl.’

ktab¨tem ‘write scnd pl masc’

Coetzee, A.W. (1999). Tiberian Hebrew Phonology: Focussing on Consonant Clusters. Studia Semitica Neerlandica. Assen: Van Gorkum. Iraqw [P/U%F] Afro-Asiatic, Cushitic, South. Tanzania. f f f

Stress is on the penultimate syllable if it has a long vowel. Otherwise stress is on the final if it has a high tone. In all other words, stress is initial.

ba¨/eeso ‘bushbucks’

mu¨gul

¨hlooro

¨wawitmo ‘king’

‘grasshopper’

‘collarbone’

Mous, M. (1993). A Grammar of Iraqw. (Kuschitische Sprachstudien 9) Hamburg: Buske. Kera [F/L] Afro-Asiatic, Chadic, East, A, A.3. Chad, Cameroon.

840 f f

f

Middle Eastern and African languages

Stress falls on a syllable with a long vowel. If there are none, stress the first syllable with a high tone, unless it is a prefix, ending in Cn/ng or contains an inserted vowel. If there are no long vowels or high tones, stress the final syllable as long as it is not the nominal ending -i or schwa.

¨tiiru@

‘his skin’

ki¨s"@rk"@ ‘black’

g«$¨g«$«r«$m

¨m"@rg"@ ‘greetings’

‘your knee’

kormo¨yom ‘your son’

¨kasi

‘hand’

Ebert, K. (1979). Sprache und tradition der Kera (Tschad). Teil III Grammatik. Berlin: Verlag von Dietrich Reimer. Maltese [U/P] Afro-Asiatic, Semitic, Central, South, Arabic. Malta. f

f

Stress falls on the syllable if it has a long vowel or is closed by two consonants. In all other cases stress is penultimate.

ber¨quøq ‘apricot’

˜a¨zaqt ‘I have dug’

¨tifla ‘girl’

Sutcli¤e, E.F. (1936). Grammar of the Maltese language with chrestomathy and vocabulary. London: Oxford University Press. Aquilina, J. (1959). The Structure of Maltese. Malta: Royal University. French, E. (1978). Contemporary journalistic Maltese: an analytical and comparative study. (Studies in Semitic Languages and Linguistics 8). Leiden: E.J. Brill Saho [LEX/- (pitch?)] Afro-Asiatic, Cushitic, East, Saho-Afar. Ethiopia: Eritrea. f f

Stress is contrastive, and accompanied by high pitch. Stressless words exist, and are pronounced in isolation with high pitch on the final syllable. This must not be confused with stress.

¨baŒa ‘a son’ yi¨baŒa ‘my son’

baŒa ‘a daughter’ ¨yibaŒa ‘my daughter’

Welmers, W.E. (1952). Notes on the structure of Saho. Word 8, pp. 145–162. Tachelhit; Shilha [U] Afro-Asiatic, Berber, Northern, Atlas. Morocco.

Middle Eastern and African languages f f

f

841

Stress falls on the final syllable of the stem. If su‰xes are added, weak stress occurs on the final vowel of the stem and a‰x. If prefixes are added, stress on their final vowel is weaker than the main stress.

ta¨mart ‘beard’

taga¨na ‘sleep’

har¨ma

adu¨ku

‘clown’

‘shoe’

Applegate, J.R. (1958). An Outline of the Structure of Shilha. ACLS, New York. Tamazight; Central Atlas [U] Afro-Asiatic, Berber, Northern, Atlas. Morocco western Algeria, Atlas Mountains. f

Stress falls on the final syllable of the word, even if it only has inserted schwas.

adi¨ni˜ ‘i will say’ før [«¨fø«r] ‘to hide’

nda¨wa tfsørnt [t«f«sø«¨r«nt]

‘we cured’ ‘you f.p. explained’

Abdel-Massih, E.T. (1971). A Reference Grammar of Tamazight. Ann Arbor: Center for Near Eastern and North African Studies, The University of Michigan. Niger-Congo Diola-Fogny [I] Niger-Congo, Atlantic-Congo, Atlantic, Northern, Bak, Diola Proper, Diola-Gusilay, Diola-Fogny. Senegal, Gambia. f f

f

f

Stress is placed on the first syllable of neutral themes. Su‰xed person markers when in final position or when appearing before the simple subordinate and contingent markers are also stressed. Definite articles and reduplicated neutral themes receive slight secondary stress. In all other situations stress is even throughout the word.

E ¨ja mEn ni ¨ka n

‘goat’ ‘i dress myself ’

E ¨ja mEn ¨nm ni ¨ka ¨n

‘my goat’ ‘i dress him’

Sapir, J.D. (1965). A Grammar of Diola Fogny. A language spoken in the basse-casamance region of senegal. Cambridge: Cambridge University Press.

842

Middle Eastern and African languages

Grebo [LEX] Niger-Congo, Atlantic-Congo, Volta-Congo, Kru, Western, Grebo, Liberian. Liberia. f

Words can be di¤erentiated only by the place of the stress.

¨sono ‘sting’

so¨no

¨mina

‘smell’

‘spoon’

pu)¨u) ‘ash’

Innes, G. (1966). An introduction to Grebo. London: SOAS, University of London. Kongo [I] (status doubtful, this might be a non-stress language) Niger-Congo, Atlantic-Congo, Volta-Congo, Benue-Congo, Bantoid, Southern, Narrow Bantu, Central, H, Kongo (H.10). Zaire, Angola, Congo. f f f

Primary stress in Kongo falls upon the first syllable of the stem. Secondary stress on the penult of four syllable and longer words. Secondary stress on the first of two monosyllabic prefixes.

¨meka ‘to bleat’ ¨dodo˙kela ‘appeal, request’

¨tantika ‘to stretch’ ¨bindala˙kana ‘to interlace’

Holman Bentley, Rev. W. (1887). Dictionary and Grammar of the Kongo Language. London: Baptist Missionary Society. Koromfe [I] Niger-Congo, Atlantic-Congo, Volta-Congo, North, Gur, Central, Northern, Kurumfe. Burkina faso. f

Stress is located on the first syllable of the word.

¨hama ‘believe’

¨jilenga

‘frog’

¨sumbotru ‘open’

Rennison, J.R. (1997). Koromfe. London: Routledge. Luvale [P] Niger-Congo, Atlantic-Congo, Volta-Congo, Benue-Congo, Bantoid, Southern, Narrow Bantu, Central, K, Chokwe-Luchazi K(20). Angola, Zambia. f f

Primary stress is located on the penultimate syllable of the stem. Secondary stress occurs in forms where more than 2 syllables follow the main stress.

Middle Eastern and African languages

¨kuya

‘to go’

a¨mbile ‘he said’

843

ngutam¨buke ‘that I may walk’ cila¨njilo

‘market’

Horton, A.E. (1949). A Grammar of Luvale. Johannesburg: Witwatersrand University Press. Nyambo; Kinyambo [I] Niger-Congo, Atlantic-Congo, Volta-Congo, Benue-Congo, Bantoid, Southern, Narrow Bantu, Central, J, Haya-Jita (J.20). Uganda, Kenya, Tanzania. f f

Underlyingly the first syllable of the root is stressed. Stress is realised as high tone in a later stage.

o mu ¨ka@ ma ‘chief ’

o mu ¨ru@ muna ‘sibling’

o mu ¨ka@ ate ‘bread’

Bickmore, L. (1989). Kinyambo Prosody. Ph.D. Dissertation, UCLA, Los Angeles. Bickmore, L. (1992). Multiple stress levels in Kinyambo. Phonology 9: 2, 155–198. Senoufo, Supyire [I] Niger-Congo, Atlantic-Congo, Volta-Congo, North, Gur, Senufo, SuppireMamara. Sikasso, Mali. f f

In the majority of lexical roots the initial syllable is stressed. A‰xes clitics and other morphemes may combine with each other to form phonological words complete with stress.

¨ba ¨nuRuNo…

‘river’ ‘return’

¨yiRi ‘get up’ ¨kapana ‘reason for coming’

Carlson, R. (1994). A Grammar of Supyire. Berlin: Mouton de Gruyter. Swahili [P] Niger-Congo, Atlantic-Congo, Volta-Congo, Benue-Congo, Bantoid, Southern, Narrow Bantu, Central, G, Swahili (G.40). Tanzania, Kenya. f f

Stress is penultimate. Exceptions with antepenulitmate stress (e.g. Arabic loans).

¨nge ‘scorpion’ dho¨ruba ‘hurricane’

¨ona thela¨thini

‘see’ ‘thirty’

844

Middle Eastern and African languages

Wald, B. (1987). Swahili and the Bantu Languages. In B. Comrie (ed.) The World’s Major Languages. Oxford: Oxford University Press. Ashton, E.O. (1947). Swahili Grammar. Second Edition. London: Longmans, Green & Co. Nilo-Saharan Bagirmi [U] Nilo-Saharan, Central Sudanic, West, Bongo-Bagirmi, Sara-Bagirmi, Bagirmi. Chad, Niger. f f

Stress falls on the final syllable. Later source does not mention stress, only tone.

Gaden, H. (1909). Essai de grammaire de la langue Baguirmienne. Paris: Lecroux. Stevenson, R.C. (1969). Bagirmi Grammar. Linguistic Monograph, Series No. 3. Khartoum: Sudan Research Unit, University of Khartoum. Kenuzi-Dongola [L/F] Nilo-Saharan, Eastern Sudanic, Eastern, Nubian, Central, Dongolawi. Northern Sudan. f f f

Stress falls on the last syllable with a long vowel or closed syllable. Final closed syllables are light unless they contain a long vowel. In words without heavy syllables, stress is initial.

¨aru ‘rain’ in¨tillE ‘needle’

gE¨riøf ‘small net’ ¨durukanE ‘old age’

¨nosogid ‘length’ du¨roøkanE ‘thickness’

Armbruster, C. (1960). Dongolese Nubian: A Grammar. Cambridge: Cambridge University Press. Kunama [LEX (A/)P/U] Nilo-Saharan, Kunama. Ethiopia. f f

Stress is mostly lexical. Elsewhere it is often penultimate or final.

¨toma ‘fasting’ ser¨ga ‘left hand’

to¨ma ‘fire’ fa $¨naka ‘time’

Middle Eastern and African languages

845

Tucker, A.N. & M.A. Bryan (1966). Linguistic Analysis: The Non-Bantu Languages of North-Eastern Africa. London: Oxford University Press. Bender, Lionel M. (1996). Kunama. Munich: Lincom Europa. Lango; Lwo [I] Nilo-Saharan, Eastern Sudanic, Nilotic, Western, Luo, Southern, Luo-Acholi, Alur-Acholi, Lango-Acholi. Central, north of Lake Kyoga, Lango Province (Uganda). f

Stress falls on the first syllable of the root.

¨r" $N.o@ ‘meat’ ¨ma$ka c ‘scissors’

o$¨.t" $t ¨@ d.

‘firefly’ ‘she scoops’

Noonan, M. (1992). A Grammar of Lango. Berlin: Mouton de Gruyter. Murle [F/F] Nilo-Saharan, Eastern Sudanic, Eastern, Surmic, South, Southwest, Didinga-Murle, Murle. Southern Sudan, Upper Nile Province. f f f

Stress falls on the first syllable with a long vowel, else on the first. In longer words (three syllables and up) this rule is less often adhered to. Stress and high tone co-occur.

¨ka@ dak ‘to eat’ ¨k"@ziwan ‘bu¤alo’

ka¨da@ ak ‘to die’ mo¨to@ oNtoc ‘tamarind tree’

Arensen, J. (1982). Murle grammar. Juba: SIL Sudan.

Appendix: The StressTyp Codes Part II contains extracts from the StressTyp database. Each extract starts with the name of the language followed by a StressTyp code between square brackets. This information is followed by language family information, country or location where the language is used, a brief verbal description of the stress pattern, some representative examples and one or more descriptive sources that were used for this StressTyp entry. The database itself contains additional information, most crucially a more detailed and parametric characterization of the stress pattern, information about syllable structure etc. We refer the reader to chapter 2 for more information on the structure of StressTyp entries and chapter 12 for ways in which the database can be used for statistical and typological research. The StressTyp code is a condensed formula which characterizes the stress pattern of the languages. Below we specify the meaning of these codes.

Fixed Stress Patterns I S T A P U

Primary stress always occurs on the initial syllable. Primary stress always occurs on the second syllable. Primary stress always occurs on the third syllable. Primary stress always occurs on the antepenultimate syllable. Primary stress always occurs on the penultimate syllable. Primary stress always occurs on the final syllable.

Variable Stress Patterns I/I Place stress on the initial syllable if it is heavy (even if the second syllable is also heavy), otherwise place stress on the second syllable if it is heavy, if neither first nor second syllables are heavy, then place stress on the first syllable. I/S Place stress on the initial syllable if it is heavy (even if the second syllable is also heavy), otherwise place stress on the second syllable if it is heavy, if neither first nor second syllables are heavy, then place stress on the second syllable.

848

The StressTyp Codes

S/I Place stress on the second syllable if it is heavy (even if the first syllable is also heavy), otherwise place stress on the first syllable if it is heavy, if neither first nor second syllables are heavy, then place stress on the first syllable. S/ T Place stress on the second syllable if it is heavy (even if the third syllable is also heavy), otherwise place stress on the third syllable if it is heavy, if neither second nor third syllables are heavy, then place stress on the third syllable. U/U Place stress on the ultimate syllable if heavy (even if the penultimate syllable is also heavy), otherwise place stress on the penultimate syllable if it is heavy, if neither are heavy, place stress on the ultimate syllable. U/P Place stress on the ultimate syllable if heavy (even if the penultimate syllable is also heavy), otherwise place stress on the penultimate syllable if it is heavy, if neither are heavy, place stress on the penultimate syllable. P/U Place stress on the penultimate syllable if heavy (even if the ultimate syllable is also heavy), otherwise place stress on the ultimate syllable if it is heavy, if neither are heavy, place stress on the ultimate syllable. P/P Place stress on the penultimate syllable if heavy (even if the ultimate syllable is also heavy), otherwise place stress on the ultimate syllable if it is heavy, if neither are heavy, place stress on the penultimate syllable. Or: Place stress on the penultimate syllable if heavy (even if the antepenultimate syllable is also heavy), otherwise place stress on the antepenultimate syllable if it is heavy, if neither are heavy, place stress on the penultimate syllable. The code for this type is also P/P with the note that EM ¼ right P/A Place stress on the penultimate syllable if heavy (even if the antepenultimate syllable is also heavy), otherwise place stress on the antepenultimate syllable if it is heavy, if neither are heavy, place stress on the antepenultimate syllable. A/A Place stress on the antepenultimate syllable if heavy (even if the penultimate syllable is also heavy), otherwise place stress on the penultimate syllable if it is heavy, if neither are heavy, place stress on the antepenultimate syllable. F/F Place stress on the first heavy syllable. If there is no heavy syllable present, place stress on the first syllable. F/L Place stress on the first heavy syllable. If there is no heavy syllable present, place stress on the last syllable.

The StressTyp Codes

849

L/F Place stress on the last heavy syllable. If there is no heavy syllable present, place stress on the first syllable. L/L Place stress on the last heavy syllable. If there is no heavy syllable present, place stress on the last syllable. Other Codes and Connectives Lex The locations of either main or secondary stresses are specified in the lexicon for the majority of the words in the language. This means that stress can be phonemic, because two non-monosyllabic words that are identical in segmental make up may di¤er in stress location and meaning. NMS Stands for No Main Stress. All stresses are reported to be equally prominent. L(CNT) This is a so-called ‘‘count system’’. Primary stress is assigned to the head of the last foot in the word. Stress is assigned from left-to-right. This leads to di¤erent stress locations for words with an odd and an even number of syllables. F(CNT) This is a so-called ‘‘count system’’. Primary stress is assigned to the head of the first foot in the word. Stress is assigned from right-to-left. This leads to di¤erent stress locations for words with an odd and an even number of syllables, usually Initial stress in the even case and Second stress in the odd case. IRR is used to indicate that stress varies unpredictably within the domain. Pitch and Tone are added between parentheses to indicate interaction between pitch or tone assignment and metrical structure. ;

This connective indicates that there is some degree of variation between two (or more) patterns for main stress. The dominant pattern comes before the semicolon.

– This connective indicates that ‘‘superheavy’’ syllables are involved in the computation of stress. If such a syllable occurs in the position indicated before the hyphen, it bears stress. Otherwise a standard rule (placed after the hyphen) comes into operation. % This connective indicates a stress shift outside the bounded stress domain under special circumstances. Stress shifts to the location after the % sign under these circumstances, and stays in the bounded domain otherwise.

Subject index accent/stress acute 451, 457, 467 antepenultimate 33, 36–7, 163, 168, 183–184, 199, 207, 208, 210, 223, 253, 277, 294, 295, 296, 323, 349, 355, 357, 365, 407, 416, 437, 444, 449, 455, 457, 476, 477, 557 Celtic, pitch features 440 change (diachronic) see historical change circumflex 451, 457, 467 compound 190, 201, 221, 262, 263, 271, 274, 280, 281, 283, 338, 347, 368, 393, 462, 469, 530, 581, 595, 598 cyclic 20, 25, 30–1, 327 demarcative function 19, 223, 383, 391, 414–5 determined by lexical category 206, 278, 279, 280, 281, 298 distinctive 341, 348, 352, 353, 355 dynamic 11 final (see ultimate) fixed 33, 35, 37, 456, 457, 462, 465, 490, 493, 650–652 free (see variable) 357, 441 grave 467 initial 33–6, 56, 159, 160, 161, 162, 163, 165, 179, 181, 182, 184, 186, 191, 192, 193, 195, 196, 197, 199, 200, 201, 202, 203, 207, 211, 218, 219, 221, 223, 255, 256, 257, 262, 263, 289, 292, 293, 294, 320, 326, 346, 353, 354, 358, 364–7, 402, 437, 441, 442, 449, 454, 461, 479, 481, 486, 487, 495, 523, 537, 559, 570, 572, 575, 581, 586 left edge 56, 173, 175, 191, 192, 197, 198, 199, 201, 205, 206, 209, 211, 212, 213, 214, 215, 216, 217, 218, 219, 221, 226, 256, 262, 273, 283, 284, 289, 299, 326, 333, 354

levels of – 18, 20, 224–225 lexical accent (see lexical accent systems) ?? 169, 170, 177, 178, 179, 184, 185, 186, 189, 193, 197, 198, 201, 204, 205, 206, 210, 211, 214, 216;, 218, 219, 221, 294, 337, 450 moveable 434 musical 11 non-moveable 478 overlength 493 oxytone (see ultimate) paroxytone (see penultimate) penultimate 33–6, 64, 92, 97–8, 157, 163, 164, 166, 167, 168, 174, 176, 178, 179, 180, 183, 184, 185, 186, 188, 190, 191, 193, 198, 199, 200, 201, 203, 206, 207, 208, 209, 210, 211, 217, 218, 219, 221, 222, 223, 253, 255, 257, 263, 264, 267, 269, 270, 272, 273, 274, 276, 277, 278, 279, 281, 282, 290, 293, 294, 295, 296, 323, 325, 327, 329, 330, 333, 340, 341, 349, 352, 353, 355–7, 364–7, 394, 396, 399, 400, 402, 405–8, 414, 415–6, 437, 438, 439, 444, 454, 457, 466, 474, 475, 476, 486, 491, 524, 557, 617, 619, 621, 622 phrasal 268, 277, 286, 289, 296, 353, 389, 407, 442 prefinal (see penultimate) preparoxytone (see antepenultimate) pre-prefinal (see antepenultimate) primary (as opposed to non-primary) 18, 20, 22–3, 25, 37, 56, 156–157, 163, 164, 165, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 178, 179, 181, 183, 184, 185, 186, 187, 188, 189, 191, 192, 193, 194, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 209, 210, 211,

852

Subject index

212, 214–224, 266, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 286, 289, 293, 296, 297, 299, 328, 331, 334, 336, 348, 351, 353, 356, 358, 367, 456, 494, 660 right-edge 64, 90–2, 174, 193, 199, 211, 212, 213, 217, 218, 286, 291, 294, 298, 356, 654–655 second syllable 33–6, 255, 263, 264, 269, 270, 275, 292, 293, 294, 295, 298, 319, 320, 325, 326, 333, 336– 8, 345, 354, 356, 358, 469, 474, 475, 558, 617, 660 secondary 19–20, 29–30, 161, 167, 168, 170, 174, 175, 180, 181, 182, 184, 185, 186, 190, 191, 193, 194, 195, 196, 197, 199, 200, 201, 202, 204, 219, 224, 225, 275, 276, 277, 278, 279, 283, 284, 289, 290, 292, 293, 296, 297, 319, 323, 324–8, 331, 334, 336, 337, 342, 349–51, 353, 354–6, 358, 364–7, 437, 439, 449, 454, 465, 468, 488, 493, 494, 513, 581, 586, 587, 625 shift 470 stem-initial 353, 354, 367, 393, 398, 405, 415–6 stem-final 517 symbol 3 third syllable 33–6, 168, 202, 319, 326, 335, 353, 437 ultimate 33–6, 162, 163, 167, 168, 172, 174, 177, 178, 179, 180, 185, 188, 189, 190, 193, 200, 201, 205, 206, 207, 208, 209, 210, 211, 217, 253, 255, 257, 258, 259, 260, 265, 268, 269, 270, 271, 272, 273, 274, 276, 277, 278, 279, 280, 281, 282, 283, 286, 288, 289, 290, 291, 292, 294, 295, 296, 297, 298, 364–8, 387, 390, 438, 439, 444, 462, 466, 488, 492, 533, 535, 557, 593, 594, 598, 619, 622, 623 variability of – 29–32

variable (as opposed to fixed) 19, 33, 38–9, 650, 654–658 accent/stress system types bounded systems 33–9, 360, 456, 636 count systems 43–6, 69, 164–165, 360, 362, 636, 638, 658 duration-accent systems 13, 251, 267–268, 279 fixed accent systems 455 first/first systems 40–2, 157, 158, 170, 187, 189, 218, 434–435, 451, 452, 486, 490, 491, 537, 549, 567, 639 first/last systems 40–2, 157, 174, 175, 176, 185, 481, 489, 548, 549 last/first systems 40–2, 178, 179, 388, 389, 391, 399, 405, 454, 456, 465, 486, 491, 495, 532, 537, 549, 620, 625, 630, 631, 639, 641 last/last systems 40–2, 157, 174, 491, 536, 549, 566, 625 lexical accent systems 62, 455 non-stress accent systems 12, 315, 316–9, 320, 336, 346, 359, 362, 382–5 pitch-accent systems 11–2, 14, 17– 8, 114–6, 120–2, 251, 252, 314– 19, 325, 330, 336, 345, 382–5, 390–1, 410–6, 434, 438, 441, 457, 467, 468, 471, 474, 475, 529, 542, 544, 548, 550, 560, 578, 588 prominence-driven systems 47, 656 quantity-insensitive systems (see weight-insensitive systems) 202, 222, 322, 334, 457 quantity-sensitive systems (see weight-sensitive systems) 159, 163, 167, 170, 186, 191, 197, 199, 205, 206, 211, 214, 215–216, 217, 298, 322, 333, 336, 340, 350, 457, 475 stress-accent systems 11, 251, 283, 341, 345, 347, 352, 354, 355, 359, 361, 434, 441, 468, 470, 471, 475, 536, 560, 568

Subject index tone-accent systems 12–4, 254–255, 382–5, 390–1, 455, 481 unbounded systems (in general) 39– 42, 63, 360, 439, 486, 540, 549, 621, 622, 639, 655 weight-insensitive systems 38, 57, 314, 334, 449, 454, 457, 475, 476, 650–652, 661 weight-sensitive systems 39, 61, 314, 319, 354, 386, 388, 389, 398, 437, 438, 454, 456, 475, 482, 529, 542, 560, 561, 565, 638, 650, 654–659, 661 accent window 37, 173, 281, 339, 342–4 accent/stress -attracting properties, 321, 350, 360 accent-tone interaction 125–131, 171, 172, 174, 199, 204, 254–256, 383, 415–6 accentual cues (see also phonetic correlates of accent/stress) 3, 8– 10, 13–4, 17 ?? accentual/stress domain (see also threesyllable window) 40–1, 57, 63, 346 a‰xes accent/stress neutral 360 accent/stress attracting/sensitive 321, 350, 360 Class I 447, 463 Class II 447, 464 cyclic 463 non-cyclic 463 pre-accenting (su‰xes) 327 su‰xes 463, 466, 468 prefixes 79, 178–179, 181, 184–185, 188, 190, 191, 194, 196, 197, 205, 220–224, 266, 267, 275, 293, 294, 339, 354 allophones 264, 282 allophonic cues 5, 13 apocope 470 areal relationships 155, 217–219, 598

853

articulation preciseness of – 4, 9, 14 articulatory force 4 aspiration 5, 355, 474 Basic Accentuation Principle (BAP) 434, 452 breathy voice 481 clitics 27, 263, 272, 275, 279, 327, 328, 342, 360, 365–6, 398, 470, 497, 560 colon 488 compound stress/accent 21–2, 30, 262, 263, 271, 274, 280, 281, 283, 393, 466 consonants geminate 444 long 443 contour tone 250, 252, 253, 254, 256, 260, 285, 328, 351 contrast lexical pitch 438 couplet 257–258, 260, 261–264 culminativity 19, 36, 288, 352, 382–3, 391, 411–3 derived words 447 desinence 462 di¤usion 75–81 distinctiveness 13 distributional asymmetries 6, 8, 394– 5, 401–4, 408–9 Europe 432 extrametricality 37, 40, 59, 186, 197, 203, 280, 291–292, 319–20, 331, 336, 354, 364, 414–5, 446, 449, 468, 491, 541, 635, 656 falling contour 438 feet (also, foot) bounded 57, 175, 211 degenerate 58, 60, 322, 335, 346, 357, 659–660

854

Subject index

iambic 159, 163, 164, 168, 169, 170, 175, 176, 183, 199, 201, 202, 203, 207, 209, 211, 214, 215, 216, 217, 335, 638, 653 canonical iambic 335 monosyllabic 446 trimoraic 437 trochaic 37, 57, 64, 188, 198, 199, 203, 215, 262, 264, 286, 462, 468, 576, 581, 586, 587, 588, 638, 653 unbounded 41, 177, 211, 215, 218 focus 16, 22, 25 narrow 454 geminates/gemination 268, 444, 449, 518 genetic relationships 155, 213, 214–216 hiatus 438 historical change 190, 214–215, 388, 397, 400, 640 hybrid word-prosodic systems 116, 123–131 inherent stress-attracting properties 321, 414 instrumental evidence of stress/accent 318, 360 intensity 251, 274, 278, 279, 316–7, 319, 320, 324–5, 329, 333, 338, 342, 358 intonation 9–10, 15–8, 163, 183, 255, 257, 266, 272, 277 boundary tone 15 intonational pitch-accents 15, 17, 21–3, 100–2 iterativity 58, 178, 205, 321, 331, 345–6 language families vii–viii language profiles vii Law of Three Morae 467 lexical marking 468 lexicographic practice 35–6 loanwords 270, 272, 277, 278, 283, 288, 289, 296, 299, 326, 335, 387, 483, 535

Meeussen’s Rule 411–2 metrical theory 32, 37, 57 metrical grid 24 metrical tree 44 bottom-up/top-down construction 45 minimal pair 319, 323, 327, 334, 354 mora (see syllable) moraic trochee 475 mora-timing/timed morphology (see also a‰xes) 9–10, 155–156, 160, 169, 170, 172, 173, 176–179, 182, 186, 188, 196, 206, 214, 217, 218, 220–224, 226, 253, 254, 259, 261, 271, 280, 287, 296 nasality 297, 298, 328–9 neutralization of contrast 5–7, 12, 250, 253, 254, 255, 257, 259, 260, 261, 262, 263, 264, 266, 267, 269, 271, 285, 287, 297, 318, 329, 352 nonfinality (see extrametricality) non-peripherality (see extrametricality) obligatoriness 383, 390–1, 413 OCP 411 openness 343 phonetic correlates of accent/stress amplitude 4, 9, 155, 156, 205, 224, 225, 279, 350 duration 4, 9, 12, 14, 251, 259, 266, 267–268, 279, 296, 316–20, 325, 333, 338, 342, 344, 351, 382, 389, 392, 394, 407, 416 energy 4 pitch 4, 9–12, 155, 156, 167, 168, 170, 176, 180, 181, 184, 190, 197, 201, 203, 205, 206, 209, 224–225, 226, 251, 263, 267, 270, 274, 278, 279, 296, 314–5, 317, 320, 324– 6, 329–31, 334, 336, 338, 345–6, 351–2, 356, 382, 392, 407–8, 416 intensity 4, 251, 274, 278, 279, 316– 7, 319–20, 324–5, 329, 333, 338, 342, 358

Subject index loudness 155, 170, 181, 184, 197, 201, 224, 225 spectral tilt 4, 316 phonotactic cues 5, 8, 12–3 phonotactics 5 phrasal stress/accent 16–7, 23–6, 277, 286, 289, 296, 389, 407 phrase intonational 559 iambic 595 phonological 559 pitch-accent (see intonation and accent/stress system types) positional faithfulness 8 proclitics 438 prominence factors 47, 382–5, 656 prominence scale 343, 367 prosodic morphology 9 quantity (see syllable and accent/stress system types) Romance Accent Rule 462 Romance Extrametricality Rule 462 Romance Verb Accent Rule 463 rhythm 4, 18–20, 24–6, 29, 58, 163, 165, 167, 169, 175, 176, 182, 183, 184, 187, 190, 198, 199, 200, 203, 204, 205, 207, 211, 213, 215, 217, 220, 221, 226, 286, 520, 570, 652– 654 bidirectional 168, 213 iambic 321, 354, 595, 653, 662 iterative 58, 321, 331, 345–6 left-to-right 58, 163, 169, 176, 183, 187, 198, 203, 207, 211, 213, 320, 334, 336, 338, 349, 661, 662 phrasal 24, 353, 389, 407 right-to-left 64, 169, 213, 326, 336, 349, 661, 662 rule 447 ternary 494 trochaic 58, 322, 326, 349, 486, 653, 662

855

weight-insensitive 58 weight-sensitive 61, 386, 388, 389, 398, 587 sandhi processes 253, 256, 257, 260, 263, 264, 265 Saussure’s Law 451 schwa (see vowel) segmental distribution asymmetries 384–5, 395, 401–4, 408–9 sonorance 343 stem 442, 462, 475 stød 443, 447, 448 stress see accent/stress stress-timed/timing 562 Stress Erasure Convention 463 StressTyp v, 45, 89–90, 156–157 syllable acute 452 ballistic 258, 259, 260 bimoraic 466 closure 46, 446, 449, 483 coda 7, 296, 323, 325, 330–1, 333, 335, 340, 351, 657 complexity 7, 9 constraints 5 controlled 258, 259, 260 extrasyllabicity 635 first 456 heavy 38, 272, 273, 276, 277, 279, 284, 290, 298, 319–20, 322–3, 325, 331, 333, 336, 344, 350, 352– 5, 357, 359, 364–7, 657 light 38, 264, 276, 279, 290, 319–20, 322–3, 326, 335–6, 350, 352, 355, 357, 359, 364, 657 long accented 438 mora 12, 318, 323, 333, 346, 349– 51, 357–8, 366–8, 562 penultimate 465, 475 neo-acute 452 rhyme 12, 270, 279 superheavy 438 unaccented 8, 12, 437 unstressed 439

856

Subject index

weight (see also accent/stress system types) 38, 41–46–8, 57, 386, 388, 389, 398, 442, 449, 468, 481, 486, 493, 518, 533, 546, 556, 656–658 syllable timed/timing 21, 476, 560 syncope 470 three-syllable window 281, 343–4, 387, 434, 441, 461, 468, 469 tone 171, 172, 173, 174, 186, 195, 196, 199, 204, 206, 208, 210, 214, 216, 217, 219, 224, 225, 250, 251, 252, 252–256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 271, 271–272, 279, 283, 285, 293, 294, 296, 297, 298, 299, 315– 8, 324–5, 327–31, 336–41, 345–8, 350–2, 358, 366, 525, 562 acute 452 circumflex 452 contour 12, 250, 252, 253, 254, 256, 259, 260, 285 dragging 448 falling 448 language 12, 560, 578, 580, 586 pattern 11 perturbation 253, 254, 261 predictable (tone systems) 413 register (tone systems) 252, 253 restricted (tone systems) 250, 413 rising 481 sandhi 253, 256, 257, 260, 263, 264, 265, 580 shift 412

word 444, 447 tone-stress/accent interaction 125– 131, 383, 391, 394, 395, 401, 580 vowel deletion 162, 197, 200, 204, 225, 271, 353, 355 epenthesis 179–180, 188, 189, 207, 208, 210, 265, 396, 330–1, 438 full 5, 29, 465 harmony 19, 352 height 14, 322 long/length 7, 12, 46, 92, 319–20, 322–5, 331, 333, 340, 346, 353–5, 358, 407–8, 437, 443, 446, 449, 468, 470, 494, 527, 538, 561, 565, 570, 577, 62, 658–659 long unaccented 438 quality 466, 483, 484, 490, 532, 565 reduction 178, 225, 319, 325, 407 schwa 4–5, 8, 29, 441, 462, 465, 466, 517, 520, 556 short 446 syncope 165, 296 voiceless 320, 329, 341, 343–5, 352, 364, 366–7 words minimal 269, 291, 659–660 prosodic 28, 296, 321, 323, 335–6, 348, 350, 357–8 tense 449 unaccented 27–29

Language index to Part I ||Ani 394, 426 !Xoo 395, 426 Abau 121–123, 138, 142, 148 Abelam 119–120, 148 Abenaki Western 167, 169, 212–213, 229, 242 Abipo´n 345, 362 Abkhaz 432, 477–478, 484, 500 Abun 124–125, 139, 148 Acehnese 96, 106 Achagua 326, 376 Achumawi 185–186, 245 Adamawa-Ubangi (family) 400, 418 Adyghe 477–478, 484 Aguacatec 286, 306 Aguacateco 288 Ahtna 170–173, 212, 233, 243 Ahuelican Nahuatl 296 Ainu 510, 512, 524, 528–530, 578, 602, 610, 612 Akhvakh 481, 484 Akkadian 617, 619–621, 641–645 Akuntsu 329, 369 Alactlasch Mixtec 263, 311 Alagı¨lac 250, 290 Alamblak 119, 139, 148 Alaskan Haida 174, 212, 214, 217, 224 Alawa 65–66, 70, 73, 84 Albanian 434, 465–466, 497–498, 503, 505 Gheg 465 Old 466 Tosk 465 Aleut 159, 163, 212–213, 221, 227, 242, 485, 512 Algic (family) 170, 212, 218, 237, 242 Algonquian (subfamily) 158, 163–170, 212–214, 219, 222, 224–225, 227, 237, 241–243 Algonquin 164, 242

Alutiiq 162–163, 234, 242 see also Pacific (Gulf ) Yupik Alyawarra 60, 85 Ambonese Malay 95 Amele 40, 119, 145, 148, 658 Amharic 386, 388–389, 423, 619 Amuzgo Guerrero 258 Amuzgoan (family) 256, 258 Ancient Egyptian 618, 645 Ancient Greek 466–468 Andamanese 510, 576, 608 Andi 477–478, 480–481, 484 Apache 173, 240, 244 Jicarilla 173, 244 Western San Carlos 172–173 Apalaı´ 332 Apurina˜ 325–326, 361, 364, 372 Arabana-Wangkangurru 58–59, 83 Arabic 381, 386–389, 391, 417, 420– 423, 425, 432, 535, 565, 618–620, 625–646 Baghdadi 631 Bani-Hassan 388, 638–639, 644 Bedouin Hijazi 626, 631, 634–635 Cairene 629, 634–640 Chadian 386, 417 Classical 388–389, 625, 629, 634, 637, 639–640, 643 Damascene 388, 627, 634–636, 638, 640 Egyptian Radio 629 Egyptian Sa‘iidi 634–635 Gulf 388, 421 Lebanese 627, 632 Libyan 633, 642, 645 Maltese 432, 641 Moroccan 386, 420, 633–635, 640– 642 Negev Bedouin 628, 634–636, 638, 640–641

858

Language index to Part I

Nigerian 386, 425 Palestinian 627–628, 631, 637–638, 642 Riyadh 388, 638 Saan‘aani (San‘aani) 630, 634–636 Sudanese 633, 639 Tunisian 633, 639 Yemen Plateau 631, 634–635, 639 Aramaic 619–622, 641–642, 644, 646 Neo-Aramaic 387–388, 421, 619– 622, 644 Arapesh Mountain 119, 149 Arara 329, 332, 365 Arawa´ (family) 348–349, 361–362, 372 Arawak(an) (family) 314, 319, 321, 361–363, 372, 376 Archi 39, 477, 482–484 Arikem 330, 361 Armenian 434, 464–465, 496, 498, 506, 532, 557, 615 Asheninca 320–322, 355, 361, 364, 377 Assamese 552, 558, 562, 604, 608 Athabaskan (family) 170, 172–173, 212, 214, 217–220, 225–226, 232– 233, 235, 238, 240 Atlantic languages 397–399 Atsugewi 185–186, 212, 240, 245 Atzingo Matlatzinca 265 Au 119–120, 145, 148 Austronesian (family) 34, 38, 40, 46– 47, 87–94, 96–99, 102–108, 110– 111, 118–119, 130–131, 381, 385, 396–397, 422, 426, 547–578, 589, 647, 649, 664, 666 Autu 119, 149 Avar 477–478, 480–481, 484 Awa 121–124, 131, 133, 143, 149 Awa Pit 344–345, 359, 361, 368, 370 Awadhi-Bagheli 552, 554 Aweti 328, 362, 365 Awtuw 119, 140, 149 Ayapa Zoque 272

Aymara 353, 362, 367, 375 Ayuru 329 Ayutla Mixe 274 Ayutla Mixtec 262, 307 Azkoitia Basque 500 Baburiwa 126, 149, 152 Baghdadi Arabic 631 Bagheli see Awadhi-Bagheli Bagvalal 477, 480–481, 484, 658 Bahasa Indonesia 88, 106, 109–110 see also Malay, Indonesian Bakairi 332–333, 361, 365 Balochi 564, 567, 603 Balsas Nahuatl 296, 303 Bambara 402 Banawa´ 349, 361, 367, 370 Bani-Hassan Arabic 388, 638, 644 Baniwa 319, 324, 378 Baniwa-Curripaco 325, 363–364 Baniwa of Maroa 324 Bantik 94–95, 103 Bantu (family) 10, 52, 121, 382, 384– 385, 397, 404–416, 418–422, 424– 425, 427 Barasano 338–340, 373 Barbacoan (family) 344 Bari 340, 559 Basaa 409 Basaburua and Imoz Basque 471 Bashkir 531–532, 535, 600, 609 Basque 115, 142, 316–317, 374, 430, 432, 470–475, 500, 503, 550 Azkoitia 474, 500 Basaburua and Imoz 471 Bidasoa Valley 472 Gernika 471–472 Hondarribia 472, 475 Lekeitio 472 Northern High Navarrese 472 On˜ati 473 Roncalese 473 Sakana 473 Souletin 470–471, 473, 475 Zeberio 473

Language index to Part I Batak Toba 93, 98, 101–103, 107–108 Bawm 584, 586 Bedawi 391, 618 Bedouin Hijazi Arabic 626, 631, 634– 635 Beja 382, 391, 421, 618 Belhare 586–587, 600 Bengali 552, 554, 558–560, 562, 601, 605, 607 Berber (family) 381, 386, 389–390, 417, 425, 617–618 Berguner Romansch 450, 658 Betawi 99–100, 107–108 see also Jakartan Bhojpuri 552, 554 Biak 95, 97, 106, 109 Bidasoa Valley Basque 472 Binumarien 123, 131, 133, 143, 149 Blackfoot 168, 214, 230, 242 Bocata´ 298 Bondei 406, 415 Bora 345 Boran 345 Boro´ro 341–342, 361–362 Boruca 296–298 Brahui 568–570, 603 Brat 139, 149, 151 Brazilian Portuguese 6, 316 Breton 435–436, 438–439, 505 Bribri 296–297, 301, 340 Buglere 297 Bukiyip 119, 139, 149 Bulgarian 451, 455–457 Buol 98 Burarra 57, 59, 66, 82 Burushaski 471, 510, 550, 563, 577– 578, 599 Cabe´car 296–298, 340 Caddoan (family) 204, 213, 238, 247 Cahuilla 157, 193–195, 198, 212–213, 221, 238–239, 246 Cairene Arabic 629, 634–640 Cajonos Zapotec 270, 307

859

Cakchikel 288 Canela-Krahoˆ 342, 362, 366, 377 Caodeng Gyalrong (rGyalrong) 588, 612 Capanahua 39, 334 Capanawa 334, 361, 366 Cariban (family) 341, 331–332, 354, 376 Carijona 332 Cashibo 334, 361, 366 Castilian Spanish 115, 460 Catalan 458, 460–461 Cayuga 157, 207, 209–213, 224, 230, 248 Central Alaskan Yupik 160–163, 212, 225, 236, 242 Central Kalapuyan (family) 180 Central Khmer 590 Central Mazahua 265 Central Pame 266–267 Central Pomo 186, 245 Central Siberian Yupik 159–161, 163, 212–213, 242 Central Sierra Miwok 182–183, 212, 230, 245 Chadian Arabic 386, 417 Chadic (family) 386, 392, 617, 634–635 Chaga 416, 423 Chalcatongo Mixtec 262, 306 Cham 96 Chamorro 98, 105 Cha´nguena 297 Chantyal 582, 587, 609 Chapakuran (family) 359 Chatino (family) 253, 255, 269, 308, 310 Tataltepec 269, 308 Yaitepec 253, 255, 269 Chechen 477, 479, 503 Chehalis Lower 244 Upper 179, 244 Chemakum 176, 244 Chemehuevi 157, 193, 199, 212–213, 216, 237

860

Language index to Part I

Chepang 582, 586 Cheremis Eastern 490, 505 Cherokee 207, 210, 248 Chiankue¨shbo 335, 366 Chiapas Zoque 272, 275–276 Chibchan (family) Proto 297 Chicahuaxtla Trique 253, 264 Chichewa 406 Chichimec 253, 265 Chichimeca-Jonaz 265 Chickasaw 205, 225, 231, 248 Chicomuceltec 283 Chimakuan (family) 176, 212, 217, 244 Chimane 353 Chimila 340 Chimwiini 422 Chinantec 253–255, 259–260, 302, 308, 309–310 Comaltepec 260, 309 Lealao 253, 259–260, 309 Palantla 255 Quiotepec 260, 302, 308 Sochiapan 260, 302 Tepetotutla 260, 310 Usila 253 Chinantecan (family) 253, 256, 259 Chinese 96, 108–109, 136–137, 142, 510, 578–581, 595–596, 601, 603, 607, 613 Mandarin 578–580, 607 Chintang 586, 600 Chiquitano 341 Chiriguano 327, 362, 365, 371 Chitimacha 205, 212, 239, 248 Chiwere 201–203, 212, 216, 236, 247 Otoe 201 Chizigula 406, 415, 422 Choapan Zapotec 270, 305 Chocho-Popolocan (family) 267 Choco´ (family) 357, 361–362 Choctaw 205, 207, 212–213, 216, 225, 236, 240, 248

Cholan-Tzeltalan (family) 282, 289 Tzeltal 289, 302 Chontal Highland 282, 310 Lowland 282 Chrau 590, 593, 612 Chugach 162 Chukchi 512–514, 517, 522, 524, 603 Chumashan (family) 191, 212, 218– 219, 246 Chuvash 495, 530–532, 601–602, 607 Classical Arabic 388–389, 625, 629, 634, 637, 639–640, 643 Coast Tsimshian 179–180, 212, 217, 229, 244 Coatla´n (family) 271–272, 300, 304 Coatla´n-Loxicha Zapotec 271, 300 Coatzospan Mixtec 262–263, 302–303 Cochimı´-Yuman (family) 188, 212, 214, 245 Coeur d’Alene 176, 244 Columbian 177–179, 212, 244 Comaltepec Chinantec 260, 309 Comanche 191, 198, 212–213, 221, 228, 246 Concepcio´n Pa´palo Cuicatec 261, 300 Connacht Irish 437 Coosan (family) 180, 218, 244 Copainala´ Zoque 276 Copola Trique 252–253, 255 Corachol-Aztecan (family) 294 Cornish 435, 437, 439, 505 Cowlitz 179, 244 Cree East 168, 228, 242 Plains 168–169, 212–213, 242 Creek-Seminole 206, 212 Cuaiquer 344 Cuiba 355–356, 362, 377 Cuicatec (family) 257, 260–261, 300 Concepcio´n Pa´palo 261, 300 Cuitlatec 250, 290 Cuna 297–298, 303, 340, 362, 366, 380 Cupen˜o 193–195, 199, 212, 232, 246

Language index to Part I Cushitic (family) 386, 390–391, 424, 617–618 Czech 19, 33, 439, 451–454, 456–457 Dagbani 402, 424 Dakota 33, 36, 200–201, 203, 212, 216, 221–222, 239 Lakhota (Teton Sioux) 201, 238 Stoney 201, 203, 212, 239, 247 Waxpe´tu`na˜ Santee Dakota 200 Damana 340–341, 344, 359, 361, 366, 379 Damascene Arabic 388, 636, 638, 640 Dani Lower Grand Valley 139, 151 Danish 384, 440, 443–444, 447–448 Dari 564, 565 Daˆw 346–348, 363 De’kwana 332, 361 Delaware 164–166, 169, 212–213, 230, 243 Munsee 164–166, 169, 230, 243 Unami 164–165, 169, 243 Delta-Cross (family) 402 Deni 349 Dholuo 393, 426 Dieguen˜o 188–189, 212, 234, 246 Mesa Grande dialect of Ipai Dieguen˜o (Northern Dieguen˜o) 188 Digo 406, 415, 422 Dinka 394, 425 Diola-Fogny 398, 400, 425 Diuxi Mixtec 264, 301, 308 Doracic 297 Dorasque 297 Doutai 124–125, 130–132, 143, 149 Drung 585, 588 Dutch 6, 8, 19, 30, 32, 43, 49–50, 52, 128, 134, 140, 143, 316, 440–442, 444–449, 500–501, 651, 658, 665 East Cree Southern 168, 228 Eastern Cheremis 490, 505

861

Eastern Maya (family) 286, 288 Mamean (family) 283, 286, 288 Quichean (family) 283, 288 Eastern Ojibwa 164, 242 Nishnaabemwin 164–166, 169, 241 Eastern Pomo 186–187, 212–213, 218–219, 224–225, 235, 245 Eastern Popoloca 267, 305 Eastern Tukanoan (family) 317, 336, 338, 340, 373 Efik 403–404, 419 Egyptian Ancient 618, 645 Egyptian Radio Arabic 629 Egyptian Sa‘iidi Arabic 632, 634– 635 Ekagi 133, 140, 149 Ekari 117, 123, 132–135, 139, 149– 150 Elamite 568, 617, 646 Embera 38, 357, 361–362, 369–370 Chamı´ dialect 357, 361, 375 Jaidukama dialect 357, 362, 380 English Modern 3–7, 9–12, 16–19, 21, 23, 25–30, 48–50, 52–53, 100, 240, 316, 381, 384, 440–441, 445–447, 449, 501, 523, 581, 607–608, 651 Old 441, 446 Eskimo-Aleut (family) 159, 163, 212– 213, 221, 242, 485, 512 Esselen 190, 212, 239, 246 Estonian 485, 492–494, 499, 504, 506 Etruscan 430, 495, 497, 504 European Portuguese 381 Evenki 538–539, 541–542, 608 Faroese 440–442, 449, 462, 497 Farsi 564–565 Filomeno Mata Totonac 280, 306 Finnish 33, 140, 485, 488, 492–494, 496, 552 Fore 121–122, 125–126, 131, 143, 145, 149 Fort Ware Sekani 172, 220

862

Language index to Part I

Fufulde 398 Fuuta Jaloo 398 Gaanay 60–61, 63, 68, 661 Gaelic Munster 436 Scottish 435, 437–438, 440 Galela 118, 145, 149 Galibi 332 Garawa 66–68, 70, 72, 76, 78, 80, 82 Gaulish 435, 437 Gavia˜o 328, 376 General Central Yupik 160–161, 213 Georgian 476–477, 497, 617 German 6, 440–442, 444–447, 449, 459, 499, 501, 506, 509, 642 Gernika Basque 471 Gheg 465 Ghodoberi 480 Giryama 406, 414–415, 419 Gokana 403–404 Golin 14, 125–126, 131, 139, 150 Gorontalo 98 Gothic 440–442, 501 Grebo 400–401, 421, 424 Greek Ancient 466–468 Modern 432, 434, 441, 461, 464, 466–469, 498, 500, 503, 507 Greenlandic 163, 224, 230, 238, 242 West 163, 230, 238 Guahibo 355–356, 361–362, 377 Guahiboan (family) 355 Guaicurian (family) 250, 290 Guajiro 319–320, 360–361, 364, 374 Guato´ 341, 377 Guatuso 296–297, 340 Guayabero 355, 362, 379 Guaymı´ 296–297 Guaymian (family) 296 Guelavı´a Zapotec 270, 304 Guerrero Amuzgo 258 Guevea de Humboldt Zapotec 271, 306 Guichicovi 272–273 Gujarati 551, 553, 555–556, 562, 608

Gulf Arabic 388, 421 Gunbarlang 65–67 Gur (family) 402, 424 Gureng gureng 59, 60 Gyalrong (rGyalrong) 588, 612 Caodeng 588, 612 Haida 173–174, 212–214, 217, 224– 225, 230, 234, 244 Alaskan 174, 212, 214, 217, 224 Masset 174, 212–213, 217, 225, 230 Halang 590, 592 Hanis Coos 180, 218 Harakmbut-Katukina (family) 313 Hare Slavey 243 Hatam 116, 124, 144, 150 Hausa 392, 419, 424, 427 Hebrew Modern 387–388, 419, 619–620, 623, 625, 641–642, 645–646 Tiberian 619, 623, 641, 645 Herero 406, 423 Hibena 413 Hidatsa 203, 212, 235, 238, 247 High Mari 489 High Runyankore 415 Highland Chontal 282, 310 Hindi 551–552, 554, 556–557, 562 Hittite 469–470, 503, 506, 615 Hixkaryana 331, 361, 365, 371 Hmong-Mien 551, 595–596 Hoca´k see Winnebago Hokan (superfamily) 185, 245, 290 Hondarribia Basque 472 Hooper Bay-Chevak 160 Hopi 157, 195, 199, 212–213, 219, 224, 233, 246 Hua 131, 141, 154 Huajuapacan Mixtec 256 Huajuapan Mixtec 263, 308 Huaraz 353, 361 Huariapano 336, 361, 366, 377 Huastec 283–285, 299, 301 Huastecan (family) 282–283 Huautla Mazatec 254–255, 268

Language index to Part I Huave 250, 253–255, 290, 292, 302, 305, 307 San Mateco 253 Huehuetla Tepehua 276–277, 282, 299, 305 Huehuetla Totonac 279 Huichol 253, 255, 294, 304 Huitoto 345–346 Huitoto (family) 345–346, 362 Huitoto-Ocaina 345–346, 362 Hungarian 19, 36, 484–485, 487–488, 496, 499, 505–506 Hunzib 477, 482 Hup(da) 346, 348, 363, 372 Hurrian 470, 617, 646 Huva see Hua Iatmul 116, 146, 150 Iau 124–125, 130, 138, 150 Ibibio 403–404, 417, 426 Icelandic 19, 440–442, 449, 461, 497 Ijoid (family) 404 Ika 340 Ikpeng 332 Imyan Tehit 142, 150, 153 Indonesian 32, 52–53, 88–89, 101– 103, 106–107, 109–110, 138–139, 141, 143–146, 426, 649, 666 see also Malay Ingarico´ 332, 371 Ingush 477, 479 Inuit-Inupiaq (subfamily) 159, 163, 242 Ipai Dieguen˜o Mesa Grande 188 Iquito 317, 357–358, 361, 363, 368, 376 Iraqw 390–391, 424, 651 Irish Connacht 437 Old 437 Iroquoian (family) 157, 207, 209–214, 222, 224, 228, 235, 248 Isthmian (family) 296–297

863

Isthmus 253, 255, 270, 272, 296, 306– 308 Isthmus Zapotec 255, 270, 306–308 Italian 21, 52, 458–461, 463–464 Itunyoso Trique 264, 301 Itza´ 284 Ixil 286 Jabuti 341 Jacalteco 289 Jahai 593, 601 Jakartan 99, 101–102 see also Betawi Jaltapec 272 Jamina´wa 335 Jamul Tipay 188–189, 235 Japanese 10–12, 17, 29, 43, 50, 52, 115, 135–136, 315–317, 346, 373, 510, 530, 543, 547–548, 550, 578, 589, 600–601, 604, 606–607, 609– 610 Tokyo 10–11, 29, 43, 115, 346, 543, 548 Jaqaru 353, 362, 367, 373 Jarawara 349, 361, 367, 371 Javanese 88, 100–102, 110 Jeˆ (family) 341–342, 361–362 Jicaltepec Mixtec 263, 300 Jicaque 250, 290 Jicarilla Apache 244 Jilipian Pame 266–267 Jingulu (Djingili) 65–68, 81–82 Jita 382, 411–414, 419 Jonaz 265, 305 Chichimeca-Jonaz 265 Ju|hoansi 395, 423 Junı´n 352 Juruna 328, 363 Kadiwe´u 345–346, 361–362, 367, 378 Kainji (family) 404, 419 Kakua 346–347, 363 Kalami 553, 555, 560, 563 Kalapuyan 180, 212, 218, 244 Central 180

864

Language index to Part I

Kamano 121, 127, 131, 140, 142, 150 Kamasau 119, 145, 150 Kamayura´ 326, 362, 365 Kampan 321 Kanamarı´ 313 Kanjobalan (family) 283, 288 Kanjobalan-Chujean (family) 283 Kannada 569, 571, 611, 612 Kapauku 133, 146, 149–150 Kapo´n 332 Karaja´ 341–342, 362, 366 Karelian 485, 492, 494 Kari’nya 332, 361, 365 Karihona 332 Kariri 341 Karitiana 330–331, 361, 365, 379 Karo (South America) 329–330, 362– 363, 365, 372 Karok 185, 228, 245 Kashaya 186–187, 212–213, 219, 221, 225, 228, 237, 245 Kashmiri 553, 560, 613 Katawixı´ 313 Katukina 313, 335, 362, 366, 368, 378 do rio Bia´ 313 Katxuyana 332, 361 Kaugel 131, 139, 150 Kaure 125–127, 130, 139, 151 Kawaiisu 191–193, 199, 212, 241, 246 Kaxinawa´ 335, 362, 366 Kera 392, 419 Kesu? 99, 109 Ket 510, 512, 524–526, 612 Kewa 121–122, 132, 141, 151 Kham 582, 586, 613 Khanty 484–485, 487–488, 496 Kharia 552, 573–575, 609–610 Khasi 572, 589, 591–592, 594, 606, 610 Khmer 96, 511, 550, 572–573, 576, 578, 589–592, 595, 599, 602, 609– 611 Central 590 Khmu 591–592

Khoisan (family) 381, 384–385, 394– 395, 404, 409, 420, 423, 426 Kinande 406, 424 Kirufiji 406 Kishambaa 406 Kitanemuk 195, 246 Kitha˜ulhu 352 Kitja 61, 84 Kitsai 204, 247 Kituba 406 Klamath 183–184, 212–213, 218–219, 223–224, 227, 245 Koasati 206–207, 212, 224, 233, 248 Kodagu 569, 571 Kogui 340 Koi 569, 582 Kolami 568, 570, 603, 612 Komi-Permyak 484, 488–489 Konkow 181, 212, 218, 223, 245 Korean 510, 530, 542–543, 545, 550, 578, 601–603, 605–607, 610 Koreguaje 340 Koromfe 402, 425 Koryak 513, 520–523, 608, 613 Kpelle 401 Kru (family) 400, 423 Kubeo 317, 336, 337–340, 363, 366, 376 Kuikuro 332–333, 362, 365 Kukuya 51, 409 Kuot 119, 143, 151 Kupang Malay 95 Kurdish 564, 567, 608, 615 Kurukh 570, 604 Kutenai 200, 212–213, 217–218, 230 Kuyawi 346–347, 362 Kwak’wala 157, 174, 176, 212, 215, 218, 244 Kwakiutl 174, 227, 231, 244 Kwami 392, 422 Lacondo´n 284 Lahu 585, 587, 608 Lak 477–478, 482, 497, 503, 521, 599 Lakhota 201, 238

Language index to Part I Lakondeˆ 350–351, 361, 363, 367, 379 Lampung 100, 110 Lango 393, 424 Lao 597–598, 603 Latin 281, 369, 376, 380, 407, 441, 458–459, 461–462, 464, 474, 506, 636, 640–641 Latundeˆ 350–351, 361, 363, 367, 379 Latvian 450, 456–457 Lavukaleve 119, 146, 151 Lealao Chinantec 253, 259–260, 309 Lebanese Arabic 627, 632 Lekeitio Basque 472 Lencan (family) 293 Lezgi 477, 482 Lhasa Tibetan 578 Libyan Arabic 633, 642, 645 Lillooet 176, 178, 212, 241, 244 Lingala 405–406, 408, 416, 420 Lithuanian 450–451, 456–458, 497, 563 Livonian 485, 493–494 Lou 102, 105 Low Mari 490 Lower Chehalis 244 Lower Grand Valley Dani 139, 151 Lowland Chontal 282 Luisen˜o 193–195, 199, 212, 246 Lushootseed 176, 179, 212, 244 Northern 179, 212 Luvale 406, 421 Luvian 469–470, 503 Lycian 469–470, 503 Lydian 469–470, 503 Macedonian 33, 35–36, 433, 450, 455– 457, 616 Macro-Jeˆ (family) 314, 341–342, 361– 362, 377 Magdalenian (family) 296 Maibrat see Maybrat Maidu 181, 212, 218–219, 225, 239, 245 Mountain 181, 212–213

865

Maiduan (family) 181, 212–213, 223, 245 Maipurean (family) 314 Mairasi 125, 129, 144, 151 Maithili 552, 555 Makah 157, 175–176, 212, 215, 244 Makonde 406–407 Maku´ 346–348, 362–363, 375, 377 Makushi 332, 361, 365 Malagasy 381, 385, 396–397, 422 Malak-malak 66, 69, 76, 82 Malay 46, 50, 88, 95–96, 99, 103, 105–110, 593 Ambonese 95 Jakartan 99 Kupang 95 Manado 95, 103, 108, 110 North Moluccan 95, 110 Pattani 46, 50, 96, 106 Malayalam 39, 569–570, 572, 599 Maleseet-Passamaquoddy 212–213 Maltese Arabic 432, 641 Malto 561, 602, 612 Mam 48, 286–287, 299, 301, 311 Mamaindeˆ 317, 350, 361, 363, 367, 371–372 Mamean (family) 283, 286, 288 Manado Malay 95, 103, 108, 110 Mandan 203, 212–213, 216, 236, 247 Mandarin Chinese 579–580, 607 Mande (family) 401–402, 404, 419, 427 Mangarayi 65–66, 74, 84 Mangbetu 393 Manggarai 100 Mansi 484–485, 487, 524 Mantjiltjara 57–62, 64–65, 84 Manx 435–438, 499, 505 Mapuche 356–357, 368, 379 Maranunggu 60, 66, 85 Marathi 553, 560, 562, 609 Maratiton 250, 290 Mari 485, 489–492, 495, 501 High 489 Low 490

866

Language index to Part I

Maricopa 189, 212, 230, 246 Masset Haida 174, 212–213, 217, 225, 230 Matis 334 Matlatzinca 253, 265, 299, 301 Atzingo 265 de San Francisco 301 Matlatzincan (family) 265 Matses 334, 362, 366, 372 Maung 65–67, 73–74, 82 Mawe´ 328, 362, 365, 378 Maxakali 341 Mayan (family) 48, 249, 282–284, 286, 288, 299, 301–303, 306 Cholan-Tzeltalan (family) 282, 289 Tzeltal 289, 302 Eastern Maya (family) 286, 288 Mamean (family) 283, 286, 288 Quichean (family) 283, 288 Huastecan (family) 282–283 Huastec 283–285, 299, 301 Kanjobalan-Chujean (family) 283 Western Maya (family) 288 Kanjobalan (family) 283, 288 Yucatecan (family) 283–284, 286 Yucatec 254, 283–286, 299, 302– 303 Maybrat 139, 149, 151 Mayo 197–199, 212, 224, 246, 293– 294 Mayoruna 334 Mazahua 265–266, 308 Central 265 Mazahuan (family) 265–266 Mazatec 254–255, 268, 305, 308 Huautla 254–255, 268 Proto 268, 305 Soyaltepec 254, 268, 308 Mazatecan (family) 267–268 Mbaya´ 345 Me’phaa see Tlapanec Meah see Meyah Menominee 164–165, 167, 169, 212– 213, 233, 236, 242

Mesa Grande dialect of Ipai Dieguen˜o 188 Metzontla 267, 310 Meyah 43, 137, 141, 152 Mezquital Otomi 253, 266, 310 Miguelenho 359, 361–362 Minangkabau 100 Minica 345 Minnesota Ojibwa 164–165 Minto 171 Misantla Totonac 279–280, 306 Mı´skito 289–290 Misumalpan (family) 249, 289–290 Mitlazapotec 253 Miwok 182–183, 212–213, 218, 228, 230, 245 Central Sierra 182–183, 212, 230, 245 Northern Sierra 182–183, 212–213, 228, 245 Southern Sierra 182–183, 212–213, 228, 245 Mixe(an) (family) 249, 272–276, 298 Ayutla 274 Proto 272–274 Mixe-Zoque (family) 272, 276 Mixtec (family) 253–254, 256, 260– 264, 300–304, 306–308, 311 Alactlasch 263, 311 Ayutla 262, 274, 307 Chalcatongo 262, 302, 306 Coatzospan 262–263, 302–303 Diuxi 264, 301, 308 Huajuapacan 256 Huajuapan 263, 308 Jicaltepec 263, 300 Mixtepec 263, 271, 307–308 Molinos 263, 304 Pinotepa Nacional 263 Proto 261 San Antonio Atatlauhca 253 San Miguel el Grande 253, 261 Silacayoapan 263, 307 Mixtepec Mixtec 263 Miztecan (family) 253

Language index to Part I Mocama 207, 212–213 Mocovı´ 345, 362 Modern English 3–7, 9–12, 16–19, 21, 23, 25–30, 48–50, 52–53, 100, 240, 316, 381, 384, 440–441, 445– 447, 449, 501, 523, 581, 607–608, 651 Modern Greek 432, 434, 441, 461, 464, 466–469, 498, 500, 503, 507 Modern Hebrew 387, 419, 623, 625, 641–642 Mohawk 7, 157, 207–208, 210–212, 222–223, 248 Moken 96, 107 Moklen 96, 107 Moksha 485, 490, 495 Molinos Mixtec 263, 304 Mon 572, 591, 593, 596, 600, 610 Monde´ 328 Mongolian 388, 512, 530, 536–538, 542, 578, 601, 609, 612–613 Mono 191, 193, 199, 202, 212, 246 Montagnais 168–169, 212, 219, 235, 242 Mopa´n 284 Mordvin 485, 490, 495 Moroccan Arabic 386, 420, 633, 642 Mororene 98, 105 Mosete´n 353–354, 362, 367, 378 Mosetenan (family) 353 Mountain Arapesh 119, 149 Mountain Maidu 181, 212–213 Move (Papua New Guinea) 125, 127– 129, 131, 152, 154 Move (South America) 296 Movere 296 Mpur 123–124, 143, 152 Muinane 345 Mundari 573, 575 Munduruku´ 329, 370–371, 377 Munsee Delaware 164–166, 243 Munster Gaelic 436 Mura 343 Mura (family) 343, 361–362 Murik 119, 138, 152

867

Murrinh-patha 66, 71 Murui 345 Muskogean (family) 205–207, 212– 213, 216, 222, 236, 248 Muskogee 206 Muysca 340 Nade¨b 346–347, 362 Na-Dene 173 Nahuatl 157, 198–199, 212, 227–228, 239–240, 294–296, 299–300, 303, 309 Ahuelican 296 Balsas 296, 303 Puebla 296 Tetelcingo 198, 240, 294–295, 299 Nakara 66–68, 72, 77 Nambikwara (family) 313, 317, 350, 352, 361, 363 Namwanga 406, 418 Nanti 321–323, 361–362, 364, 371 Naolan (family) 247, 250, 290 Natchez 207, 212, 248 Navajo 172, 219, 235, 244 Ndebele 406 Negev Bedouin Arabic 641 Nenets 484–486, 495, 501, 505 Neo-Aramaic 387–388, 421, 619–622, 644 Nepali 552, 555, 562, 568, 587 Nez Perce 183–184, 218, 227–228, 245 Nga¨be´re 296 Ngalakan 65–66, 74, 84 Ngalkbun 65–66, 74–75, 84 Nganasan 484–486, 495, 500 Ngankikurrungkurr 66, 69, 71, 83 Ngarinyin 81–82 Ngarndji 65–67, 82 Nguni (family) 406, 416 Niger-Congo (family) 8, 381, 384– 385, 397, 404, 409, 418, 427 Nilo-Saharan (family) 381, 385, 393, 418 Nilotic (family) 393–394, 419 Ningil 119, 143, 152

868

Language index to Part I

Nipode Huitoto 345 Nisenan 181, 212, 218, 223, 229, 245 Nivkh 510, 512, 524, 527, 604, 611 Noon 399–400, 405, 426 Nootka 157, 175–176, 212–213, 215, 218, 223, 239, 244 Tsishaath 175–176, 212–213, 215, 239 Nor 119, 152 Norse 440–442 Old 440–441 North Moluccan Malay 95, 110 Northeastern Pomo 186, 221, 245 Northern Bizkaian Basque 316 Northern Dieguen˜o 188 Mesa Grande 188 Northern High Navarrese Basque 472 Northern Lushootseed 179, 212 Northern Nambikwara 317, 350, 372 Northern Pame 255, 266–267, 300 Northern Pomo 224, 245 Northern Popoloca 267, 309 Northern Qiang 585, 588 Northern Sierra Miwok 182–183, 212–213, 228, 245 Northern Tepehuan 196, 212, 221, 224, 241, 246, 254–255 Northwestern Ojibwa 164 Norton Sound 160–161, 213 Norwegian 440, 442–444, 447–448, 494, 502, 504 Nukak 346 Nunggubuyu 65–66, 72, 83 Nunivak 160, 162 Nyakyusa 406, 419 Nyigina 81, 84 Nyooro-Tooro 407 Obokuitai 125–126, 130, 145, 152 Ocaina 345–346, 362 Ocuilteco 265, 307 Ofaye´ 341 Ojibwa 158, 164–166, 169, 212–213, 222, 225, 237, 240, 242

Eastern 164, 242 Nishnaabemwin 164–166, 169, 241 Minnesota 164–165, 169, 222 Northwestern 164 Severn 164, 240, 242 Old Albanian 466 Old Church Slavonic 451, 455 Old English 441, 446 Old Irish 437 Old Norse 440–441 Old Persian 567, 611–612 Oluta Popoluca 272–273 Omotic (family) 386, 390–391, 617– 618 On˜ati Basque 473 Oneida 207–208, 210–212, 235, 248 Onondaga 207, 209, 248 Orejo´n 340, 379 Orok 538–539 Osetin 563–564 Otoe 201 Oto-Manguean (family) 249, 256–259, 267–269, 271–272 Otomi 253, 266, 299–300, 303, 308, 310 Mezquital 253, 266, 310 Temoayan 266, 299 Tenango 266, 300 Otomian (family) 265–266 Otopamean (family) 253, 256, 265, 299 Proto 265, 299 Pa’disua 34, 37, 152 Pacaraos Quechua 352 Pacific (Gulf ) Yupik 162–163, 212– 213, 242 see also Alutiiq Pagu 118, 147, 152 Paiute 191–193, 199, 212–213, 238, 246 Southern 191–193, 199, 212–213, 238 Paiwan 93 Palaic 469–470, 503

Language index to Part I Palaihnihan (family) 185, 212, 245 Palantla Chinantec 255 Palestinian Arabic 627–628, 631, 637–638, 642 Palisua 118, 152 Pame 253, 255, 266–267, 300, 303 Central 266–267 Jilipian 266–267 Northern 255, 266–267, 300 Pamean (family) 265–267 Panamint 191, 198, 212–213, 229, 246 Panare 332 Pano 313, 333, 361–362, 368, 370–371 Panoan (family) 314, 335, 373 Pano-Takanan (family) 333 Papago 196, 238 Papantla Totonac 278 Paraguayan Guarani 327, 365 Pashto 564, 566–567, 603 Passamaquoddy 164, 166–167, 169, 212–213, 231, 235, 240, 242 Pattani Malay 46, 50, 96, 106 Paumari 348–349, 362, 367, 371 Pawnee 204, 237, 247 Paya 296–297, 304 Pech 296–297, 304 Pemon 332, 361, 365 Penutian (superfamily) 179, 183–185, 212–213, 223, 244–245 Persian 535, 564–565, 567–568, 611– 613, 617 Old 567, 611–612 Phoenician 619, 623, 643, 646 Phrygian 433–434, 464, 470, 497 Pilaga´ 345, 362 Pinotepa Nacional Mixtec 263 Pipil 296, 300 Piraha˜ 343, 361–362, 367, 603 Plains Cree 168–169, 212–213, 242 Plateau Penutian (family) 183–184, 212–213, 223, 245 Platoid (family) 404, 419 Poconchi 283 Pogolo 407 Pokomchı´ 283

869

Polabian 451, 453–454, 456–457 Polish 19, 33, 36, 439, 451, 453–454, 456–457 Pomoan (family) 157, 186–187, 212– 213, 218–219, 221, 235, 245 Central Pomo 186, 245 Eastern Pomo 186–187, 212–213, 218–219, 224–225, 235, 245 Northeastern Pomo 186, 221, 245 Northern Pomo 224, 245 Southeastern Pomo 157, 186–187, 212, 221, 245 Southern Pomo 157, 186–187, 212, 221, 245 Southwestern Pomo 186 Ponapeic 100 Popoloca 267–268, 305, 309–311 Eastern 267, 305 Metzontla 267 Northern 267, 309 Tlacoyalco 268, 309 Western 267, 311 Popolocan (family) 253, 256, 267, 272 Popoluca Oluta 272–273 Sayula 272–273 Sierra 276, 302 Portuguese 6, 49, 131, 316, 371, 380– 381, 432, 458, 461–462, 659 Brazilian 6, 316 European 381 Pota 296–297 Potawatomi 164–165, 242 Proto-Arawa´ 349 Proto-Chibchan 297 Proto-Huitoto-Ocaina 346 Proto-Mazatec 268, 305 Proto-Mixean 273–274 Proto-Mixtec 261 Proto-Otopamean 265 Proto-Zoquean 274 Provenc¸al 458, 460 Puebla Nahuatl 296 Puinave 346–347, 363, 373 Pulaar 398–399, 401, 424

870

Language index to Part I

Fufulde 398 Fuuta Jaloo 398 Puluwat 100, 106 Pumi 585, 588 Southern 585, 588 Punjabi 556, 558, 562–563 Pure´pecha 292 Qiang 585, 588, 607 Northern 585, 588 Quechua 313, 352–353, 361–362, 367–368 Quechuan 47 Quiche´ 288 Quichean (family) 283, 288 Quileute 176, 244 Quiotepec Chinantec 260, 302, 308 Rama 296–297 Ramarama 329, 362–363 Ratahan 94, 103, 107 Rembarrnga 66, 74, 77, 84 rGyalrong see Gyalrong Riyadh Arabic 388, 638 Romani 552, 554 Romansch 458, 460, 658 Berguner 450, 658 Roncalese Basque 473 Rumanian 458–459, 461 Runyankore 407, 415, 421, 425 High 415 Russian 43, 128, 142, 435, 450–452, 455–457, 502, 512, 514, 531–532, 535, 554, 557, 651 Saami 485, 492, 494 Saanich 178, 212 Sabaneˆ 350, 352, 361, 367, 369 Sacapultec 288 Safwa 10–13, 121, 146, 310 Sahaptin (subfamily) 184, 185, 232, 238, 245 Sahu 118, 146, 152 Pa’disua, Palisua 152 Sakana Basque 473

Salcha 171, 173, 212 Salishan (family) 176, 179, 212, 217, 220, 224–225, 244 Samal 94 San Andre´s Chicahuastla Trique 253 San Antonio Atatlauhca Mixtec 253 San Bartolo Tzotzil 253, 283 San Carlos 172–173 San Francisco Matlatzinca 301 San Francisco Ozolotepec Zapotec 271, 305 San Juan Copola Trique 253, 255 San Juan Mixtepec Zapotec 271, 307 San Mateco Huave 253 San Miguel Chimalapa Zoque 272 San Miguel el Grande Mixtec 253, 261 Saan‘aani (San‘aani) Arabic 630, 636 Sangir 95 Sanskrit 551, 553, 555, 560, 607 Santa Maria Chimalapa Zoque 272 Santali 573, 575, 609 Sarare´ 352, 370 Sardinian 458–459, 461 Sayna´wa 335, 362, 366, 371 Sayula Popoluca 272–273 Scottish Gaelic 435, 437–438, 440 Sedang 590, 592 Sekani 172–173, 212, 220, 225, 243 Fort Ware 172, 220 Sekoya 340 Selayarese 99, 105 Selkup 484–486, 495, 500, 502 Semelai 589, 593, 607 Semitic (family) 386–389, 399, 420, 423–425, 568, 615, 617–621, 623, 625, 641–644 Southern (family) 386, 388–389 Seneca 207, 209–213, 228, 235, 248 Sentani 118, 140, 153 Serbo-Croatian 115, 135, 452, 455– 457, 501 Seri 189, 212, 235, 246, 250, 290–291, 299, 306 Severn Ojibwa 164, 242 Shanenawa 335, 362, 366, 370

Language index to Part I Shilha 389–390 Shilluk 394, 425 Shimakonde 423 Shina 553, 560, 563, 610 Shipibo 333–334, 361, 366, 372, 375 Shona 407 Shoshone 191–192, 198, 212–213, 221, 229, 235, 246 Sierra Miwok 182–183, 212–213, 228, 230, 245 Central 182–183, 212, 230, 245 Northern 182–183, 212–213, 228, 245 Southern 182–183, 212–213, 228, 245 Sierra Popoluca 276, 302 Sikaritai 40, 47, 125, 127, 130, 143, 153 Silacayoapan Mixtec 263, 307 Sindhi 553, 555 Siona 340, 380 Siouan-Catawban (family) 200 Sioux 201, 203, 247 Teton 201 see also Lakhota Sipacapen˜o 288 Slave 171–173, 212, 237, 499 Slavey 171–172, 233, 243 Hare 243 South 171–172, 233 Slavonic 451, 455, 498 Old Church 451, 455 Sliammon 176, 178, 212, 218, 244 Slovak 451, 453–454, 456–457 Slovenian 455 Slovincian 453–457 Sm’algyax 179–180, 229 Sochiapan Chinantec 260, 302 Somali 382, 390–391, 421 Songhay 394, 420, 424 Sora 573, 575 Sorbian 451, 453–454, 456–457 Soteapan Zoque 272 Sotho-Tswana 407 Sougb 123–124, 144, 153

871

Souletin Basque 470, 473 South Arabian 426, 619–620, 625, 643, 645–646 South Slavey 171–172, 233 Southeastern Pomo 157, 186–187, 212, 221, 245 Southeastern Tepehuan 196–197, 212, 221, 224, 241, 293, 305 Southern East Cree 168, 228 Southern Nambikwara 350, 352 Southern Paiute 191–193, 199, 212– 213, 238 Southern Pomo 157, 186–187, 212, 221, 245 Southern Pumi 585, 588 Southern Semitic (family) 386, 388– 389 Southern Sierra Miwok 182–183, 212–213, 228, 245 Southern Uto-Aztecan 293, 295, 299 Southern Waikuru´an 345 Southwestern Pomo 186 Soyaltepec Mazatec 254, 268, 308 Subtiaba-Tlapanecan (family) 256, 268 Spanish 21, 45, 115, 131, 270, 272, 278, 281, 283, 288–289, 296, 299, 335, 458, 460–461, 464 Castilian 115, 460 Spokane 178–179, 212, 227–228 Squamish 176, 178–179, 212, 229, 244 Stoney 201, 203, 212, 239, 247 Sudanese Arabic 633, 639 Sumerian 617, 620, 643, 645 Supyire 402, 418 Suruwaha´ 349, 361, 367, 379 Swahili 407, 418–419 Swati 406 Swedish 7, 12, 115, 135, 440, 442–444, 447–448, 472, 497 Taba 91 Tabaru 118, 140, 153 Tabla 118, 139, 153

872

Language index to Part I

Tachelhit (Tashlhiyt, Tashllhiyt) 389– 390 Tahltan 172, 226, 243 Taita 407 Tajik 564–565 Takelma 180, 217, 225, 238, 244 Takelman (family) 180, 212, 223, 244 Tamazight 389–390, 417 Tamil 569, 571–572, 599–600, 606 Tanacross 172, 233, 243 Tanah Merah 118, 153 Tanana 171–172, 240, 243 Minto 171 Salcha 171, 173, 212 Tapachulteco 272 Tapiete 327, 362, 365, 373 Tarascan (family) 250, 290, 292 Tarasco 292 Tariana 324–325, 361, 364, 368 Tarma 352, 368 Tashlhiyt, Tashllhiyt see Tachelhit Tataltepec Chatino 269, 308 Tatar 489, 492, 531, 535–536, 600– 601 Tatuyo 340 Tausug 94 Teco 247, 286 Tehit 140, 142, 150, 153 Telefol 121–123, 141, 153 Telugu 568–570, 606 Temoayan Otomi 266, 299 Tenango Otomi 266, 300 Tepehua 276–277, 282, 299, 305, 310 Huehuetla 276–277, 282, 299, 305 Tlachichilco 276–277, 310 Totonaco-Tepehua 276 Tepehuan (family) 196–199, 212–213, 221, 224, 241, 246, 254–255, 293, 305 Northern 196, 212, 221, 224, 241, 246, 254–255 Southeastern 196–197, 212, 221, 224, 241, 293, 305 Tepetotutla Chinantec 260, 310

Tequistlatecan (family) 249, 282 Teribe 296–297, 307–308 Tetelcingo Nahuatl 240, 294, 299 Teton Sioux 201 see also Lakhota Texistepec Zoque 272 Texmelucan Zapotec 252 Thai 96, 136, 578, 587, 589, 597–598, 611 Tiberian Hebrew 619, 623, 641, 645 Tibetan 578–579, 581, 583, 586, 588, 600 Lhasa 578, 586 Tidore 118, 146, 153 Tiene 409, 421 Tigrinya 386, 388–389, 619 Tillamook 178, 218, 229, 244 Tilquiapanz Zapotec 270 Timucua 207, 212–213, 231, 248 Mocama 207, 212–213 Timucuan (family) 207, 212–213, 248 Tiriyo´ 331–332, 361, 365, 376 Tiwi 64–66, 84 Tlachichilco Tepehua 276–277, 310 Tlacoyalco Popoloca 268, 309 Tlahuiltepec 272 Tlapanec 252–255, 269 Tlauica (Tlawika) 265–266 Tlingit 173, 217, 234, 243 Toba (South America) 345, 362 Toba Batak 93, 98, 101–103, 107–108 Tobelo 118, 142, 153 Tocharian A 469 Tocharian B 469 Tohono O’odham 196–198, 212–213, 232 Tokyo Japanese 10–11, 29, 43, 115, 346, 543, 548 Tol 250, 290 Tonkawa 203–204, 212–213, 225, 233, 247 Toraja 99, 109 Kesu? 99, 109 Tosk 465

Language index to Part I Totonac 276, 278–280, 300, 306 Filomeno Mata 280, 306 Huehuetla 276–277, 279, 282, 299, 305, 309 Misantla 279–280, 306 Papantla 278, 305 Upper Necaxa 278, 300 Totonacan (family) 249, 276, 281, 298, 310 Totonaco-Tepehua 276 Totontepec 272–273, 309 Trique (family) 252–253, 255–257, 260, 264–265, 301, 303, 305 Chicahuaxtla 253, 264 Copola 252–253, 255 Itunyoso 264, 301 San Andre´s Chicahuastla 253 San Juan Copola 253, 255 Tsakhur 482–484 Tsimshian 179–180, 212, 217, 229, 244 Coast 179–180, 212, 217, 229, 244 Tsimshianic (family) 179, 217, 244 Tsishaath Nootka 175–176, 212–213, 215, 239 Tsonga 407 Tswana 407 Sotho 407 Tu¨batulabal 193, 199, 212–213, 221, 241, 246 Tukano 313, 336, 340, 363, 378 Yepaˆ-Masa 336, 378 Tukanoan (family) 314, 317, 336, 338, 340, 373 Tule-Kaweah 183, 245 Wikchamni 183, 230 Tumbuka 407 Tu¨mpisa Panamint 191, 198, 212–213 Tunebo 340–341, 363, 366, 374 Tunica 205, 212–213, 231, 248 Tunisian Arabic 633, 639 Tuparı´ 329 Tupian 314, 326, 331, 376 Tupi-Guarani (family) 313, 326–330, 361–363, 378

873

Turkish 33, 36, 388, 432, 464, 530– 531, 533–536, 568, 600–602, 604– 607, 611 Tuscarora 210–212, 241, 248 Tuyuka 340 Txunhua˜-djapa´ 313 Tzeltal 289, 302 Tzotzil 253, 283, 289, 303 San Bartolo 253, 283 Tzutujil 288, 301 Udmurt 484, 488, 495, 498 Ukrainian 456–457 Ulithian 100, 109 Umbugarla 65–66, 72, 82 Umutina 342–344, 361–362, 366, 379 Una 120–121, 317 Unami Delaware 164–165, 243 Upper Chehalis 179, 244 Upper Necaxa Totonac 278, 300 Urartian 465, 470, 617, 646 Urubu-Kaapor 326–327, 362, 365, 375 Usarufa 122, 125–127, 131, 133, 138, 153 Usila Chinantec 253 Uspantec 253, 283 Utian (family) 182–183, 212–213, 245 Uto-Aztecan (family) 157, 191, 193, 195, 198–199, 212–216, 219, 221, 224–225, 227–228, 232, 234–236, 238–240, 246, 250, 293–295, 299– 300 Southern 293, 295, 299 Uwa 341, 366 Uzbek 530–531, 535, 600 Valley Yokuts 183, 245 Yawelmani 183, 212, 227 Vanimo 123, 145, 154 Venda 407 Viceita 296 Vietnamese 578, 592, 594–595, 610, 612 Votic 485, 493

874

Language index to Part I

Waanji 66, 68, 70, 72, 84 Wa¤a 122–123, 126, 132–133, 137 Wahgi 124–125, 130–131, 143–144, 154 Waikuru´(an) (family) 345, 361–362 Waimiri-Atroari 332 Waiwai 332, 361, 373 Wakashan (family) 157, 174–176, 212–215, 217–218, 223, 230, 244 Wanano 340, 380 Wa˜nso¨ho¨t 346 Wappo 190, 237, 240, 246 Waray 65–66, 68, 77 Wardaman 65–66, 70–71, 84 Warekena 323–324, 362, 364, 368 Wari 359, 361–362, 372 Washoe 190, 212, 246 Waunana 357, 361, 372 Waxpe´tu`wa˜ Santee Dakota 200 Wayana 332 Welsh 435–436, 439–400, 505–506, 552 West Greenlandic 163, 230, 238 West Mande (family) 401, 427 Western Abenaki 167, 169, 212–213, 229, 242 Western Apache 173, 244 San Carlos 172–173 Western Maya (family) 288 Kanjobalan (family) 283, 288 Western Popoloca 267, 311 Western Tukanoan 340 Wetan 100, 107 Wichita 204, 213, 225, 238, 247 Wikchamni 183, 230 Winnebago 33–34, 36, 201–203, 212– 213, 224, 231, 236, 239, 247, 656 Wintu 180–181, 212, 218, 237, 245 Wintuan (family) 180, 212–213, 218, 245 Witotoan (family) 345, 369 Witsuwit’en 172–173, 212 Wiyot 170, 212, 240, 243 Wolaitta 390–391, 417

Wolof 398, 421 Wulna 66, 73 Xhosa 406, 416 Xincan (family) 250, 290, 293, 299 Xipaya 328, 363, 378 Yagaria 121, 125, 127, 140, 145, 150, 152, 154 Hua, Huva 131, 141, 154 Move 125, 127–129, 131, 152, 154 Yaitepec Chatino 253, 255, 269 Yakima Sahaptin 184, 232 Yana 187, 212, 218–219, 238, 245 Yanomami 358, 363, 378 Sanuma´ (dialect) 358, 370 Xamatauteri (dialect) 358, 370 Yanyuwa 65–68, 70–71, 77 Yaqui 197–199, 212–213, 224, 229, 233, 246, 651 Yaruro 356, 362, 368, 377 Yatheˆ 341, 362, 366 Yavitero 326, 362, 364, 377 Yawelmani 183, 212, 227 Yazva 488 Ye’kwana 332, 361 Yele 119, 141, 154 Yemen Plateau Arabic 631 Yepaˆ-Masa 336, 378 Yiddish 440–441, 624 Yil 119, 143, 154 Yimas 119, 140, 154 Yindjibarndi 47, 62–63, 85 Yokuts 183, 212, 236, 245 Yokutsan (family) 183, 212, 245 Yolngu (family) 72, 77 Yucatec 254, 283–286, 299, 302–303 Yucatecan (family) 283–284, 286 Yuchi 207, 241, 248 Yuhup 346–347, 363, 377 Yukaghir 430, 433, 484–485, 510, 512, 524, 526–527, 608–609 Yuki 190, 212, 224, 246 Yukian (family) 190, 212, 246 Yukpa 332–333

Language index to Part I Yukulta 62–63, 68, 83 Yupik 159–163, 212–213, 221, 225, 233, 234 Central Alaskan 160–163, 212, 225, 236, 242 General Central 160–161, 213 Hooper Bay-Chevak 160 Norton Sound 160–161, 213 Nunivak 160, 162 Central Siberian 159–161, 163, 212–213, 242 Pacific (Gulf ) 163, 212 see also Alutiiq Chugach 162 Yurakare´ 354–355, 368, 372 Yurok 170, 227, 238, 243 Zande 400, 405, 426 Zaparoan (family) 317, 357 Zapotec (family) 252, 255, 257, 270– 271, 300, 304–309 Cajonos 270, 307 Choapan 259, 270, 305 Coatla´n-Loxicha 271, 300 Guelavı´a 270, 304

875

Guevea de Humboldt 271, 306 Isthmus 253, 255, 270, 272, 296, 306–308 San Francisco Ozolotepec 271, 305 San Juan Mixtepec 271, 307 Texmelucan 252 Tilquiapanz 270 Zoogocho 270, 309 Zapotecan (family) 253–254, 256, 269, 270 Zeberio Basque 473 Zoogocho Zapotec 270, 309 Zoque(an) (family) 249, 272–276, 298, 311 Ayapa 272 Chiapas 272, 275–276, 288–289 Copainala´ 276 Mixe-Zoque 272, 276 Proto 274 San Miguel Chimalapa 272 Santa Maria Chimalapa 272 Soteapan 272 Texistepec 272 Zulu 406 Zuni 199–200, 212–213, 219, 236, 247

Language index to Part II Abkhaz 793 Aceh Atjeh; Atjehnese; Achehnese, Achinese 691 Aguacateco 771 Aklanon; Aklan 691 Alamblak 735 Alawa 669 Alune; Sapalewa; Patasiwa, Alfoeren 692 Alyawarr; Iliaura; Alyawarra 669 Ambrym, Southeast 692 Amele 737 Apoze 692 Apurina˜ 775 Arabic, Bani Hassan 833 Arabic, Bedouin Hijazi 833 Arabic, Beirut 833 Arabic, Classical 834 Arabic, Damascene 834 Arabic, Egyptian Radio 834 Arabic, Egyptian Spoken; Cairene Arabic 835 Arabic, Gulf 835 Arabic, Lebanese 836 Arabic, Libyan spoken; Cyrenaican Bedouin 836 Arabic, Negev Bedouin 836 Arabic, South Levantine Spoken; Palestinian Arabic 837 Aralle-Tabulahan 693 Aramaic 837 Archi; Archin 793 Armenian 796 Arosi 693 Arrarnta, Western; Aranda, Western 669 Asmat, Central 738 Assyrian Neo Aramaic 838 Au 736 Aukaans; Ndyuka 786 Avar; Dagestani 793

Awadhi-Bagheli 797 Awtuw; Autu 735 Aymara 777 Baadi; Bardi; Badimaya 669 Bagirmi 844 Bagundji 670 Bagvalal; Bagulal; Bagval; Bagvalin 794 Balantak; Kosian 693 Bali; Balinese 693 Bambam; Pitu Ulunna Salu; Pitu Ulunna Binanga 694 Banawa´ 775 Banggai; Aki 694 Banggarla; Parnkalla 670 Bantik 695 Basaburua and Imoz 788 Bashkir; Basquort 825 Basque, Bidasoa Valley 789 Basque, Gernica 788 Basque, Hondarribia 790 Basque, Lekeitio 790 Basque, Northern High Navarrese 789 Basque, On˜ati 791 Basque, Roncalese 791 Basque, Sakana 791 Basque, Souletin 792 Basque, Zeberio 792 Bawm 830 Beja; Bedawi; Bischari 838 Bengali 797 Betawi; Jakarta Malay; Batavi, Batawi, Melayu Jakarte 695 Bhojpuri 797 Bhumij; Mundari 827 Biak; Numfor; Nufoor; Mafoorsch 695 Bicolano, central 695 Bima; Bimanese 696

Language index to Part II Breton 797 Bugis; Buginese; ugi; Boegineesch, Ugi, De’, Rappang Buginese 696 Bukiyip; Arapesh, Mountain 736 Buli 697 Burarra; Barera 670 Buru; Boeroe; Buruese 697 Burushaski, Yasin 832 Cahuilla, Desert and Mountain 763 Campa, Pajonal Ashe´ninca; Atiri 776 Canela-Krahoˆ 780 Capanahua 782 Carib; Kalihna 777 Catalan-Valencian-Balear 798 Cavinen˜a 783 Cayubaba; Cayuvava 786 Cayuga 753 Cebuano 697 Chamorro 698 Chatino, Yaitepec 773 Chepang 830 Chinanteco, Lealao 773 Chinese, Mandarin 831 Chocktaw 755 Chukot; Chuckchee 829 Chuvash 825 Comanche 763 Cornish 798 Cree, plains 745 Cupen˜o 764 Czech 799 Da’a; Bunggu 698 Dakota; Sioux 761 Danish 799 Dari 800 Dehu; Lifu 698 Dhurga; Thaua 671 Dieri; Diyari 671 Diola-Fogny 841 Djingili; Tjingili 671 Dobel; Kobro’or; Kobroor; Doibel 699 Doutai; Tolitai; Taori-So 732

877

Dutch 800 Dyirbal 672 Ekari; Ekagi; Kapauku 738 Emae; Mae 699 Embera Saija; Epera Pede´e 779 Embera-Chami 779 Enggano 699 English 800 Ese Ejja; Chama 783 Estonian 818 Evenki; Tungus 825 Faroese 801 Farsi, Western; Persian 801 Fijian 699 Finnish 819 Fordata; Larat 700 Fore 738 French 801 Gaalpu 672 Gaelic, Irish 802 Gaelic, Munster 802 Gaelic, Scottish 802 Galela 733 Garawa 672 Gayardilt; Kayardild 673 Gayo; Gajo 700 Georgian 794 German 803 Ghodoberi 794 Gidabal 673 Gilyak; Nivkh 832 Gooniyandi; Gunian; Konejandi 673 Gorowa 838 Grebo 842 Greek, Modern 803 Greenlandic, west 748 Guarani 784 Guguyimidjir; Koko Imudji 674 Gujarati 803 Gunwinggu; Mayali 674 Halang; Koyong 827

878

Language index to Part II

Hanis Coos 757 Hanuno´o 700 Hawaiian 701 Hebrew, Modern 839 Hebrew, Tiberian 839 Hindi 804 Hixkarya´na 778 Hoca´k; Winnebago 762 Hopi 764 Huasteco 771 Huitoto, Nipode; Witoto; Huitoto, Miunane 785 Hungarian 819 Hunzib 795 Iaai; Iai; Yai 701 Icelandic 804 Indonesian; Bahasa Indonesia 701 Inga 782 Ingush 795 Iraqw 839 Irarutu; Irahutu 702 Italian 805 Jacalteco 771 Jaqaru 777 Jaru; Djaru 674 Javanese; Jawa 702 Juat 675 Juwalarai; Yuwaalaraay 675 Kaiwa´ 784 Kalami; Kalam Kohistani 805 Kalapuya 757 Kalkutung 675 Kambera 703 Kara 703 Karelian 819 Karo Batak 704 Karok 751 Kashaya; Southwestern Pomo 751 Kaure 739 Kawaiisu 765 Kei; Kai; Evav 704 Kenuzi-Dongola 844

Kera 839 Ketengan 739 Kewa 739 Khasi 828 Khmer, Central; Cambodian 828 Khmu’ 828 Kilivila 704 Kisar; Meher; Loikera 705 Kitsai 747 Klamath-Modoc 757 Koasati 755 Kobon 740 Koi; Koya 829 Kokata; Gugada 675 Kola; Warilau; Kulaha; Marlasi 705 Komi-Permyak 820 Kongo 842 Konjo; Kondjo; Konyo; Tiro; Pegunungan 705 Koromfe 842 Kuanua; Tolai 706 Kuku-Yalanji 676 Kulamanen 706 Kunama 844 Kunjen 676 Kutenai 768 Kuuku Ya’u 676 Kwaio 706 Kwakiutl; Kwakw’ala 768 Lamaholot; Solor; Solorese 707 Lampung 707 Lamu-lamu; Lamalama 677 Lango; Lwo 845 Laragia; Larakia 677 Larike 707 Latin 805 Latvian 806 Lauje; Laudje; Tinombo; Ampibabo 708 Lavukaleve 731 Ledo; Palu; Paloesch 708 Lenakel 708 Leti 709 Lezgi; Lezgian; Kiurintsy 795

Language index to Part II Lithuanian 806 Liv; Livonian 820 Lower Grand Valley Dani 740 Luisen˜o 765 Luvale 843 Ma’anyan; Maanyak Dayak; Ma’anjan; Siang 709 Macedonian 807 Macushi 778 Mai Brat; Maybrat 733 Maidu, Mountain 758 Mairasi 741 Maithili 807 Makah; Kwe-Nee-Chee-Aht; Kweedishchaaht 768 Makassar 709 Malagasy 710 Malay 710 Malayalam 830 Maltese 840 Mam 771 Mamasa 710 Manam 711 Mangarayi 677 Manggarai 711 Mansi; Vogul 820 Mantjiltjara; Mandjildjara 678 Manx 807 Maori 712 Mapuche; Araucanian; Aucan 775 Maranunggu 678 Margany 678 Mari, High; Mountain (Western) Cheremis 821 Mari, Low; Meadow (Eastern) Cheremis 821 Maricopa 752 Marind 741 Martuyhunira; Martuthunira 679 Masset Haida; Northern Haida 756 Mataco; Wichı´ Lhamte´s 781 Mathimathi 679 Maung 679 Mayapi; Mayi; Mayi-Yapi 680

879

Mbabaram 680 Meah; Mansibaber 731 Menomini 745 Mentawai 712 Mentu; Mentuh Tapuh 712 Meso Grande Dieguen˜o 752 Minangkabau; Minang, Padang 712 Mı´skito 773 Miwok, Central Sierra 758 Miwok, Northern Sierra 759 Miwok, Southern Sierra 759 Mocama 769 Mohawk, Akwesasne 754 Moksha; Mordvin; Mordva 822 Mongolian; Khalkha 826 Mongondow; Bolaang Mongondow; Mongondou; Minahassa 713 Mpakwathi; Anguthimri 680 Mullukmulluk; MalakMalak 681 Muna; Wuna; Mounan 713 Munsee; Delaware 746 Mu´ra-Piraha˜; Piraha˜ 781 Murik; Nor 735 Murle 845 Muruwari; Murawari 681 Muskogee; Creek 756 Nahuatl, Tetelcingo 765 Nambas, Big 713 Nambiquara, Southwest; Nambikua´ra, Southern 781 Napu; Pekurehua 714 Narrinyeri; Jabulajabula 681 Nenets, Tundra 822 Nengone 714 Nepali; Gurkhali; Gorkali 808 Nez Perce 759 Ngada; Nga’da; Ngadha; Nad’a; Bajava; Badjava 714 Ngalakan 681 Ngalkbun; Dalabon; Boun 682 Ngarinyin; Ungarinjin 682 Nias; Batu 715 Ningil 737 Norton Sound 748

880

Language index to Part II

Norwegian 808 Nuaulu; Patakai; Fatakai 715 Nunggubuyu 683 Nyambo; Kinyambo 843 Nyawaygi 683 Obokuitai; Obogwitai; Baburiwa 732 Old English 808 Oloh Mangtangai 715 Oneida 754 Ono 741 Orok; Oroc 826 Osetin; Ossete; Ossetic 809 Oto; Chiwere 762 Pa’disua; Palisua 733 Paama; Paamese 716 Padoe; South Mori; Padoe´; Alalao 716 Pagu; Pago; Pagoe 734 Paiute, Southern 766 Paiwan 716 Pakaa´snovos; Wari 779 Pamona; Bare’e; Baree; Poso; Taa; Wana 716 Pashto 809 Passamaquoddy; Malecite 746 Paulohi 717 Paumarı´ 776 Pintupi-Luritja 683 Piro 776 Pitjantjatjara 684 Pitta pitta; Bidhbidha 684 Pokomchı´; Poconchi 772 Polabian 809 Polish 809 Pomo, Eastern 752 Portuguese 810 Portuguese, Brazilian 810 Provenc¸al; Occitan 811 Pulopetak 717 Puluwatese; Puluwat 717 Pure´pecha; Tarasco; Tarascan 774 Quechan; Yuma; Kwitchyan 753

Quechua 783 Quiche´ 772 Quichua, Highland, Imbabura; Otavalo Quichua 783 Rama 779 Rantepao; Kesu’ 718 Rapanui; Easter Island; Pascuense 718 Ratahan; Bentenan; Pasan 718 Ritarungo 684 Romani, North Russian 811 Romansch, Berguner 812 Roro; Waima 719 Rote; Rotti; Rotinese; Roti 719 Rotuman 719 Ruija 823 Rumanian 812 Russian 813 Saho 840 Sakao, Northern 720 Salish, Southern Puget Sound 760 Sama Baangingi, Bajan 720 Sangir; Sangihe´; Sangirese; Sangi 720 Sanskrit 813 Sanuma´ 785 Sardinian, Campidanese 814 Sawai; Weda; Were 721 Sedang 829 Selaru; Salaru 721 Selayar; Salayar; Selayarese 721 Selepet 742 Seneca 754 Senoufo, Supyire 843 Sentani 742 Serbo-Croatian 814 Seri 774 Shipibo-Conibo 782 Shiriana; Ninam? 780 Shoshone, Panamint; Koso 766 Sikaritai; Aikwakai 732 Sikule; Sichule 722 Sindhi 814 Sio 722

Language index to Part II Siroi 742 Slave 757 Slovak 815 Slovenian; Slovene 815 Slovincian 815 Sorbian 816 Spanish, Castilian 816 Spokane 761 Squamish 761 Stoney Dakota 761 Sunda; Sundanese; Priangan 722 Suriname Arawak 776 Swahili 843 Swedish 817 Taba; East Makian 723 Tabaru; Tobaru 734 Tabla; Tanh Merah 743 Tacana 784 Tachelhit; Shilha 840 Tagalog 723 Tahitian 723 Tajik 817 Talaud; Talaut; Talodda 724 Tamazight; Central Atlas 841 Tanna, Southwest 724 Tarangan, West; Trangan, West 725 Tawala 725 Taz Selkup 823 Tehuelche 780 Tepehuan, Southeastern 767 Thaypan; Kuku-Thaypan 685 Thurawal 685 Tibetan; Lhasa 831 Tigak 725 Timor; Atoni; Timorese; Timol; Dawan; Ambenu; Vaikenu 725 Tiruray 726 Tiwi 685 Toba-Batak 726 Tobelo 734 Tol; Jicaque 774 Tondano 726 Tongan 727

881

Tontemboan; Tompakewa; Tountemboan; Pakewa 727 Trumai 786 Tsakhur; Tsaxur; Caxur 796 Tsimshian, Coast; Sm’algyax 760 Tu¨batulabal 767 Tuamotuan; Pa’umotuan 728 Tukang Besi 728 Tunica 751 Tupi 784 Turkish; Osmanli 826 Tzotzil, Zinacanteco 772 Ulithian 728 Uma; Pipikoro; Uma Aria; Oema 728 Unami; Delaware 747 Uradhi 685 Urubu´-Kaapor 785 Usan 743 Uzbek, Northern 827 Vod; Votic 823 Waalubal 686 Wahgi 743 Walmajarri; Walmatjari 686 Wambaya 687 Wambon 744 Wangaybuwan-Ngiyambaa 687 Waorani; Auca; Sabela 787 Warao; Guarao 787 Wardaman 687 Warlpiri; Walbiri; Wailbry 688 Warrgamay 688 Welsh; Cymraeg; Cymric 818 Weri 744 Wetan; Wetang 729 Wichita 747 Wiyot 747 Woisika 744 Wolio; Baubau; Buton; Butung; Butonese; Boetonees 729 Wongkumara; Wankumara 688

882

Language index to Part II

Yamdena; Jamdena; Jamden 730 Yana 753 Yanyuwa; Anyula; Janjula 689 Yapese 730 Yaqui 768 Yawelmani 760 Yazva 824 Yele; Yeletnye 731 Yidiny 689 Yil 737 Yimas 736 Yindjibarndi; Jindjibandi 690

Yir Yoront 690 Yuchi 769 Yukaghir 831 Yupik, Chevak 749 Yupik, General Central 749 Yupik, Pacific Gulf; Alutiiq; Chugach 749 Yupik, St. Lawrence Island 750 Zoque, Copainala´ 773 Zuni; zun˜i 770