Reduplication in Indigenous Languages of South America [1 ed.] 9789004272415, 9789004272408

Citation preview

Reduplication in Indigenous Languages of South America

Brill’s Studies in the Indigenous Languages of the Americas Series Editors David Beck (University of Alberta) Mily Crevels (Leiden University) Hein van der Voort (Museu Paraense Emílio Goeldi) Roberto Zavala (CIESAS-Sureste)

Editorial Board Peter Bakker, Aarhus University – Nora England, University of Texas, Austin Ana Fernández Garay, Universidad Nacional de La Pampa – Michael Fortescue, University of Copenhagen – Victor Golla, Humboldt State University – Pieter Muysken, Radboud University Nijmegen – Enrique Palancar, CNRS – Keren Rice, University of Toronto – Frank Seifart, Max Planck Institute for Evolutionary Anthropology – Leo Wetzels, CNRS/Sorbonne-Nouvelle, VU Amsterdam

volume 7

The titles published in this series are listed at brill.com/bsila

Reduplication in Indigenous Languages of South America Edited by

Gale Goodwin Gómez Hein van der Voort

leiden | boston

Library of Congress Cataloging-in-Publication Data Reduplication in indigenous languages of South America / Edited by Gale Goodwin Gomez, Hein van der Voort. pages cm. – (Brill's studies in the indigenous languages of the Americas ; Volume 7) Includes bibliographical references and index. ISBN 978-90-04-27240-8 ((hardback) : alk. paper) – ISBN 978-90-04-27241-5 (e-book) 1. Indians of South America–Languages. 2. South America–Languages. 3. Language and culture–South America. I. Gomez, Gale Goodwin, editor of compilation. II. Voort, Hein van der, editor of compilation. PM5008.R43 2014 498–dc23 2014006239

This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering Latin, ipa, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.com/brill-typeface. issn 1876-5580 isbn 978 90 04 27240 8 (hardback) isbn 978 90 04 27241 5 (e-book) Copyright 2014 by Koninklijke Brill nv, Leiden, The Netherlands. Koninklijke Brill nv incorporates the imprints Brill, Brill Nijhoff, Global Oriental and Hotei Publishing. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill nv provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, ma 01923, usa. Fees are subject to change. This book is printed on acid-free paper.

Contents 1 2 3 4

5 6 7 8

9 10 11 12

13 14 15

Reduplication in South America: An Introduction 1 Gale Goodwin Gómez and Hein van der Voort Reduplication in Mapuzungun: Form and Function 17 Fernando Zúñiga and Antonio Edmundo Díaz-Fernández Reduplication in Andean Languages 39 Katja Hannß and Pieter Muysken Four Types of Reduplication in the Cha’palaa Language of Ecuador 77 Simeon Floyd Reduplication in Nheengatu 115 Aline da Cruz Reduplication in Hup (Northwest Amazonia) 143 Patience Epps Reduplication in the Yanomae Language of Northern Brazil 161 Gale Goodwin Gómez Reduplication as a Tool for Morphological and Phonological Analysis in Awetí 185 Sebastian Drude Reduplication and Ideophones in Trumai 217 Raquel Guirardello-Damian Reduplication and Verbal Number in Mẽbengokre 247 Andrés Pablo Salanova Forms and Functions of Reduplication in Tupian Languages 273 Wolf Dietrich The Interaction of Reduplication with Word Classes and Transitivity in Cavineña 313 Antoine Guillaume Reduplication in Movima: A Prosodic Morphology Approach 343 Katharina Haude When Vowel Deletion Blurs Reduplication in Mojeño Trinitario 375 Françoise Rose Reduplication in Karitiana (Tupi) 401 Luciana R. Storto

vi 16

contents

Is Reduplication an Areal Feature of the Guaporé-Mamoré Region? 427 Hein van der Voort Index of Names 463 Index of Languages 467 Index of Subjects 471

chapter 1

Reduplication in South America: An Introduction Gale Goodwin Gómez and Hein van der Voort

Reduplication is a phenomenon that occurs in the majority of the world’s languages, which explains why Pott’s (1862) seminal monograph on reduplication bears the title Doubling (reduplication, gemination) as one of the most important building means of language. Although definitions of reduplication vary, it is generally considered a morphological process that involves repetition of words or parts of words as a derivational—and sometimes inflectional—strategy. In many languages it is a marginal phenomenon. In European languages it tends to be limited to expressions such as English bye-bye, hush-hush, so-so, wishywashy, riff-raff, etc. Other languages, however, may exhibit elaborate patterns of reduplication, having many different functions. The Austronesian languages are well-known for reduplication, such as Malay, where plural is indicated by doubling the noun, e.g. anak ‘child’ vs. anak anak ‘children.’ Different types of reduplication may have different grammatical functions, as in Tagalog sulat ‘write’ vs. su~sulat ‘will write’ (future) and mag-sulat~sulat ‘write intermittently’ (distributive). In the last decade there has been an increased interest in reduplication among linguists. This has resulted in special conferences on reduplication (Graz 2002, 2007, Brussels 2012), many articles, and several dedicated volumes (e.g. Ammann and Urdze ed. 2007, Hurch ed. 2005, 2009, Hurch and Mattes eds. 2009, Inkelas and Zoll 2005, Kouwenberg ed. 2003, Michaud and Morgenstern eds. 2007, Stolz et al. 2011) in which descriptive, historical and theoretical issues have been raised and intriguing reduplication phenomena from a diverse number of languages have been discussed. South American indigenous languages, however, have been underrepresented in these developments. This situation is unfortunate both for indigenous linguistics of South America and for general and typological linguistics. The relatively few studies of reduplication in certain specific South American languages, such as Araújo (2008), Borges (2008), Bruno (2003), Dixon and Vogel (1996), Everett and Seki (1986), Fargetti (1998), García-Medall (2003), Gomes (2007), Goodwin Gómez (2009), Meira (2000), Rose (2005), Seki (1984), van Gijn (to appear), and van der Voort (2003), both confirm universally attested patterns and show unusual phenomena that have not been recorded elsewhere or discussed in the general literature. Clearly, South American languages have an important contribution to

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_002

2

goodwin gómez and van der voort

make to the general study of linguistics, but a better and more detailed understanding of these languages is required. In July 2009, we organized a symposium, Reduplication in Amazonian Languages, at the 53rd International Congress of Americanists in Mexico City. The stimulating results motivated us to put out a call for submissions to a general volume on reduplication in indigenous languages of South America. In the present volume, we have brought together fifteen descriptive and theoretical articles by scholars of South American languages. These contributors were referred to Moravcsik (1978), Rubino (2005; 2011) and the Graz Reduplication Project website (http://reduplication.uni-graz.at/) as the basic sources on reduplication for basic concepts and terminology. In order to focus our efforts as much as possible on comparable phenomena, we defined reduplication in a relatively narrow but generally accepted way: the repetition of morphemes or of parts of morphemes by which a new morpheme with a new, related meaning is created, or by which a grammatical function is systematically expressed. Not all repetition, however, is reduplication. A distinctive characteristic of reduplication is that it does not entail repetition of semantic content. Rather, the meaning of a reduplicated form is different from that of its components. This essential part of the definition excludes mere repetition of words or phrases as in very, very good, suppletive repetition of synonyms as in dazed and confused, recursive application of morphemes as in great-great-grandfather, or argument agreement as in Nós fala-mos português. ‘We speak Portuguese.’ This and other criteria for the distinction between repetition and reduplication are discussed extensively in, e.g., Gil (2005) and Stolz (2007a). With regard to terminology used by the authors in this volume, we have suggested that the ‘original,’ or ‘source’ component in a reduplication construction be identified as the base and the ‘copy’ component as the reduplicant (elsewhere also reduplicand). For segmentation and glossing of bound forms we suggested following the Leipzig glossing convention (Comrie et al. 2008) of separating the base and reduplicant in a morphological reduplication construction with a tilde (~). However, we decided not to follow the Leipzig practice of giving a semantically specific gloss for the reduplicant. Rather, if the reduplicant represents a copy of the entire base, it is glossed identically; in case the reduplicant represents only part of the base, it is glossed as ‘red.’ The reason for this is that we did not want to obscure the special reduplicative nature of the construction by non-literal, interpretive glossing. An exception to this practice is made by Haude (this volume) with regard to a specific complex subset of reduplicative constructions in Movima, where such glossing would hamper transparency of the examples.

reduplication in south america: an introduction

3

From a formal point of view, two basic types of reduplication can be distinguished: full reduplication and partial reduplication. In full reduplication, also known as total reduplication, an entire word, root, or stem is repeated, as in Yanomae noma ‘die’ vs. noma~noma a ‘death,’ kroke ‘grey’ vs. kroke~kroke a ‘cloud,’ wehe- ‘dry’ vs. wehe~wehe-ha ‘on dry land’ (Goodwin Gómez 2009). Partial reduplication is attested in several different formal types. Often, the unit of partial reduplication is the syllable, as in Emerillon su?u ‘to bite’ vs. su~su?u ‘to gnaw’ (Rose 2005), or the mora, as in Mosetén -tom ‘comitative’ vs -to~tom ‘with several’ (Sakel 2004, 79). In fact, the mora may be the more usual unit of partial reduplication, being indistinguishable from the syllable in case of a CV pattern. Partial reduplication can also involve multiple syllables/moras, as in Baure -averoč- ‘to go far’ vs. -avero~veroč- ‘to go very far’ (Danielsen 2007, 227). Another factor that may determine the shape of the reduplicant is prosody: in some languages, such as Movima (Haude 2006, 85), the reduplicant corresponds to a foot. Productive reduplication of bound morphemes has so far been convincingly attested only in Kwaza, as in nũri-xa-re (full-2-int) ‘Are you full?’ vs. nũri-xa~xay-hỹ-re (full-2~2-nom-int) ‘Were you full?’ (van der Voort 2003, 82). Depending on the language or linguistic subsystem in question, reduplicated syllables can be prefixed, suffixed or infixed. The relative position of the reduplicant, i.e., the directionality of the reduplicative process, may even have functional significance, as appears from various languages discussed in the present volume. In addition to the distinction of full and partial reduplication, three other phenomena concerning reduplication are distinguished and encountered in some of the languages discussed in the present volume: simple reduplication, complex reduplication and automatic reduplication (cf. Rubino 2005). Simple reduplication refers to the repetition of form without additional phonological or morphological changes. Complex reduplication involves further phonological changes, such as a fixed high tone vowel /ə́ / in Mundurukú ipák ‘It’s red.’ > ipak~pə́k ‘It’s not so red.’ (Picanço 2005, 382) and an additional rV syllable in Wari’ cat ‘break (sg)’ > cara~cat ‘break (pl)’ (Everett and Kern 1997, 378). In theoretically-informed literature this is also referred to as reduplication with fixed segmentism (cf. Alderete et al. 1999). Automatic reduplication does not by itself change the meaning of the resulting construction but occurs obligatorily in combination with an existing specific grammatical morpheme.1

1 Note that some authors found this definition problematic, since it may be hard to determine that the involved reduplication does not contribute in any way to the meaning of the resulting construction.

4

goodwin gómez and van der voort

As an example, the Cavineña habitual morpheme -ni- occurs often—but not exclusively—in combination with reduplication: kwa- ‘go’ > kwa~kwa-ni- ‘go habitually’ (Guillaume this volume). From a functional point of view, many different types of reduplication have been attested. Reduplication is used to express a diverse range of notions like plural, distributive, collective, size, completion, inchoative, progressive, transitive, reciprocal, nominalization and even such opposite notions as intensification and attenuation. All these functions of reduplication are also attested in South America. One of the terms that recurs in several articles in the present volume is pluractionality, which refers to verbal number (see, e.g., Corbett 2000) and can express lexical or grammatical aspectual notions such as repetitive, distributive, frequentative, and other types of event (or event-internal phase) plurality as well as participant plurality (which is different from argument agreement, which represents nominal number). In the literature on reduplication attention is often given to questions of iconicity. Many instances of reduplication are clearly iconic, e.g., the repeated, multiplied or augmented form of a reduplication construction symbolizing a corresponding repetitive, plural or augmentative meaning. Other meanings, such as attenuative and diminutive, are at times characterized as non-iconic or counter-iconic, although an iconic interpretation can be argued for just as well. Apart from dispersed remarks in some contributions, not much weight is given to this issue in the present volume. We would nevertheless like to refer the interested reader to some very useful general discussions of iconicity in reduplication by Abraham (2005), Fischer (2011) and Stolz (2007b). Furthermore, van Gijn (to appear) deals specifically with iconicity and reduplication in the central Bolivian language isolate Yurakaré. Finally, as also noted by Cruz (this volume) and Rose (this volume), Queixalos’ (2002) concept distensitivité, which is in this respect perhaps translatable as ‘unboundedness,’ might represent a common ground for—among other things—various seemingly divergent semantic functions of reduplication. Many languages have unproductive types of reduplication that cannot be related to non-reduplicated forms. This is also known as inherent or lexical reduplication and is seen especially in onomatopoeic, or in other ways symbolic forms, as in Yanomae tukutukumu ‘to beat (as a heart),’ kirakiramo ‘parrot,’ etc. (Goodwin Gómez 2009). Since these often highly iconic constructions fall outside of the definition of reduplication as a productive morphological process, they are not the focus of attention in the present volume and are dealt with only marginally, if at all. During the editing process, the issue of the nature of reduplication as a morphological process came up. Is it derivational or inflectional? The answers

reduplication in south america: an introduction

5

depend on one’s criteria for a distinction between derivation and inflection. The most broadly accepted definitions involve a combination of semantic and formal criteria, such as semantic effect, paradigmaticity, transpositionality, obligatoriness and other criteria. The distinction is not absolute, relevant or possible in all languages, and derivation and inflection are perhaps better regarded as parts of a continuum (Bybee 1985). Since reduplication tends to be rather different formally from canonical inflection, scholars of reduplication sometimes define the morphological nature of reduplication on the basis of function alone: if reduplication marks, e.g., plural it is inflectional; if it marks augmentative it is derivational. This strategy shifts the problem to decisions about which functions are universally derivational and which are universally inflectional. Furthermore, such a strategy may presume artificial distinctions in a particular language, where a single reduplication strategy expresses a function that is translated by different functions in the metalanguage. The issue has not been resolved in the Graz database, where instances of reduplication are characterized as derivational or inflectional in an ad hoc manner, not depending on consistently applied criteria but on the particular choice of the individual contributors of the data. The issue has not been resolved in the present volume either, but we have requested that the authors provide explicit criteria when characterizing (certain instances of) reduplication as inflectional or as derivational. A number of universals and universal tendencies have been formulated with respect to reduplication. Moravcsik (1978, 315) has hypothesized for example that there is “no reduplication pattern where the number of repetitions is freely chosen from the set of all numbers.” However, it appears that some supposed universals are falsified by the facts from South America. The often cited claim that languages with partial reduplication also have full reduplication is contradicted by Trinitario (Rose, this volume), which appears to have only partial reduplication. Moravcsik (1978, 315) also stated that no reduplication patterns exist that are based on any properties other than phonological and phonotactic, such as syllable number, consonantality-vowelhood, or linear position. However, in Kwaza, reduplication based on morphological units is attested (van der Voort 2003, this volume) and thus contradicts Moravcsik’s generally accepted universal. In the history of linguistics, South American languages have sometimes provided counterexamples to theoretical claims and typological universals. In this manner, the increased participation of scholars of South American languages in the general linguistic debate contributes significantly to the development of linguistic theory. Conversely, developments in theoretical, typological and historical linguistics contribute to a better understanding of South American

6

goodwin gómez and van der voort

languages and their history and relationships. Moreover, the raising of new general linguistic issues also requires further investigation of South American languages, which presupposes their documentation and which stimulates their preservation. The study of specific topics across South American languages, such as reduplication, may also help answer areal linguistic questions and, consequently, reinforce the value of the linguistic heritage of the continent. In spite of the theoretical importance of South American languages in general and of reduplication in particular, the present volume is predominantly oriented towards description. This is due to the fact that the majority of the languages discussed here are still in an initial phase with regard to documentation and description and, in some cases, almost no published descriptive studies exist. One of the assets of this volume is that it contributes both to the phenomenology of reduplication and its delimitation and to the description of under-documented and endangered languages. Several authors have taken our questions to the field, which helped them to conduct a better focused survey with regard to the issue of reduplication than they might otherwise have done. The volume’s contents are intended to represent reduplication in South America through as well-balanced a linguistic sample as possible, from geographic, areal, typological and genetic perspectives. Admittedly, in this volume southern Amazonia is somewhat over-represented geographically and the Arawakan languages are proportionally under-represented genetically. However, the typological diversity of the continent is well represented as it includes isolating, configurational, agglutinating, fusional and polysynthetic languages. Furthermore, most of the continent’s major linguistic areas (i.e. the Amazon Basin, the Central Plains, the Pacific Coast and the Andean region) are represented, and the volume’s genetic diversity is guaranteed by articles representing 14 language families, 12 language isolates and at least one mixed language.

Contents of the Volume The order of presentation of the specific articles in this volume is based on geographical region, rather than language family. The first two articles (by Zúñiga and Díaz-Fernández and by Hannß and Muysken) focus on Andean languages, and the third article (by Floyd) deals with Cha’palaa, a language of northwestern Ecuador that historically was in contact with the Andes through Ecuadorian Quechua. The remaining twelve articles encompass a variety of

reduplication in south america: an introduction

7

languages, representing a wide range of linguistic families and isolates, all found in the Amazon Basin and central Brazil. The language described in the first paper is Mapuzungun, which is spoken by the Mapuche people of Argentina and Chile. In “Reduplication in Mapuzungun: Form and function,” Zúñiga and Díaz-Fernández review previous literature and examine specific examples (from both current and early sources) to assess the formal patterns and functions of reduplication. They provide evidence for two basic functions: word formation (which is no longer productive) and a grammatical function that principally affects verb stems, although some cases involving adjective stems, adverbs, indefinite pronouns, and question words are described. Reduplication in Mapuzungun may indicate plurality of participants and express meanings of intensity, continuity and the repetition of the action conveyed by the verb. “Reduplication in Andean languages” by Katja Hannß and Pieter Muysken surveys a broad sample of Andean languages, comparing formal and functional features of reduplication. The authors suggest that “reduplication patterns may be linked to the typological organisation of a language” and dispute the claim (in the World Atlas of Language Structures online) that Aymara and Jaqaru exhibit productive use of partial reduplication. The paper’s major points are supported by specific examples from the following languages: Quechua, Aymara and its close relative Jaqaru, Pukina, Kallawaya, and Uru as well as from Cholón, Leko, and Shuar. Results of the analysis are represented graphically using NeighborNet figures to reflect degrees of areal and typological relatedness. Simeon Floyd’s paper on “Four types of reduplication in the Cha’palaa language of Ecuador” shifts the focus back to a single language. In Cha’palaa four types of reduplication are shown to occur in different word classes and grammatical constructions. While not a type of true reduplication, ideophones are included in Floyd’s inventory, which also includes what he terms adjunct and predicate reduplication and a word derivation process based on reduplication. Notable in Cha’palaa is the high functional load of reduplication when compared to a closely-related language, Tsafiki, and to the neighboring, but unrelated, Equadorian Quechua. Nheengatu is the modern descendant of Tupinambá, the language frequently cited in the early chronicles by European explorers in Brazil. In “Reduplication in Nheengatu” Aline da Cruz provides an in-depth look at the phonological, morphological, and semantic characteristics of reduplication in this language. Particular focus is placed on the process of reduplication with specific types of verbs and on the resulting semantic property of pluractionality. In addition, examples are provided to illustrate a relation between the use of

8

goodwin gómez and van der voort

reduplication and object omission in certain syntactic contexts and to show how reduplication may help distinguish reciprocal from reflexive constructions. Da Cruz also points out the historical impact of reduplication, specifically in the emergence of two modern dialects: the Negro-Içana and Middle Negro varieties. Moving on to languages of the northwest Amazon, Patience Epps looks at reduplication from a typological perspective in Hup, a Nadahup (Makú) language. In “Reduplication in Hup (northwest Amazonia)” Epps shows that the process of reduplication is restricted to “a small set of words whose phonotactics are intermediate between those exhibited by canonical monomorphemic and multimorphemic words,” and she claims that reduplication in Hup is always, in some sense, iconic. Reduplication of verb roots expresses repetition, intensity, iterativity, and continuity (classic functions of the process crosslinguistically). Likewise, most examples of reduplication in nouns are limited to inherent reduplication and occur in names for fauna and a few “other entities that are characterized by punctuated repetition or duration … [such as] musical instruments, waves, and forest clearings.” As a typologically unusual function in Hup, however, Epps shows that reduplication may be used to mark attributive modifiers in certain nominal compounds. Epps concludes her article with the suggestion that repetition and reduplication in Hup may be diachronically related. “Reduplication in the Yanomae language of northern Brazil” describes the semantic and grammatical functions of one of the major Yanomami languages. Gale Goodwin Gómez presents examples in Yanomae where the process of reduplication expresses iterativity or increased quantity, intensity, or duration of the semantic content expressed by particular noun or verb stems. In addition, reduplication of specific verb stems creates nouns, which are identified by accompanying nominal suffixes and clitics. The author also discusses examples of inherently reduplicative structures, ideophones, and onomatopoeia, which while not classifiable as true reduplication nonetheless are widely represented in the lexicons of Amazonian languages, especially in animal and plant names. Goodwin Gómez’s claim that reduplication is inherently iconic concurs with similar sentiments expressed in P. Epps’ paper. In “Reduplication as a tool for morphological and phonological analysis in Awetí” Sebastian Drude describes the process of reduplication in verb stems in great detail. The semantic effect of reduplication differs according to verb class. Separate sections are devoted to the discussion of examples involving reduplication in active verbs and in stative verbs. In addition, Drude maintains that reduplication “can contribute to the understanding of phonological and morphological units and processes,” which he illustrates, for example, by using

reduplication in south america: an introduction

9

reduplication in Awetí to confirm the abstract morpheme-final archiphonemes /P, T, K/ and to strengthen the case for distinguishing inherently oral vowels from neutral vowels. The author also attempts to use reduplication as a criterion to distinguish between inflectional and derivational affixes, but he concludes that this strategy is not always sufficient. Raquel Guirardello-Damian summarizes the formal aspects of the process of reduplication in the isolate Trumai in her article, “Reduplication and ideophones in Trumai,” with the intention of identifying the underlying principles governing the process. The examples reflect careful detail to the phonological aspects of reduplication. Verbs and adjectives are the word classes that most commonly undergo reduplication, although a limited number of examples are also given of reduplication involving nouns, adverbs, and numerals. The semantic aspects attested for reduplication in Trumai include intensity, repetition or repeated action, and number or quantity. Guirardello-Damian also provides some interesting examples involving inherent reduplication and a comprehensive discussion of ideophones, which she compares with reduplication. “Reduplication and verbal number in Mẽbengokre” by Andrés Pablo Salanova examines a Jê language of north-central Brazil. The principal focus of the paper is reduplication as affixation on verb stems that indicates iteration of an event or participant plurality. Salanova discusses in detail the canonical shape of verbs and also clarifies the position of reduplication within the language’s broader verbal morphology. The author concludes with a comparison of the characteristics of reduplication in Mẽbengokre to reduplication in Kaingang, another Jê language. Wolf Dietrich’s article, “Forms and functions of reduplication in Tupi languages” seeks to provide an overview of reduplication across several TupiGuarani languages and other languages of the Tupian stock. The author presents examples from a variety of historical as well as contemporary sources with the aim of distinguishing event-internal plurality in the case of monosyllabic reduplication from event-external plurality of disyllabic reduplication. Dietrich claims that in most modern Tupian languages this distinction has merged into a single function, which expresses “iterative or distributive plurality of the process, quantifier, or expressed quality of the subject or the object of the verb.” In “The interaction of reduplication with word classes and transitivity in Cavineña,” the reader’s attention shifts from Brazil to the Bolivian Amazon. Antoine Guillaume describes a broad inventory of reduplication processes in Cavineña, a Tacanan language. Most commonly affecting verbs, nouns and adjectives, reduplication in Cavineña is presented in terms of thirteen patterns, including simple and automatic and full and partial processes. As the author

10

goodwin gómez and van der voort

notes, “The most interesting aspect of reduplication in this language is probably its syntactic uses, as a word class-changing device and as a valence-changing process.” Katharina Haude in “Reduplication in Movima: A prosodic morphology approach” describes another Bolivian language with an extensive range of reduplication processes, which are based on prosodic categories rather than (parts of) morphemes. Reduplication is regressive and determined by the morphological or lexical base or the syntactic context. On verbs reduplication marks direct, inverse or middle voice; whereas, reduplication of nouns may indicate inalienable possession or derive possessive predicates or mark predicate nominals in subordinate clauses. Reduplication in Movima, unlike reduplication in many languages, is characterized more by its grammatical functions than iconicity and semantic functions. Verbal reduplication in the Trinitario dialect of Mojeño, an Arawak language of the Bolivian lowlands, is re-examined by Françoise Rose in her article, “When vowel deletion blurs reduplication in Mojeño Trinitario.” The paper is a revision of the analysis of Trinitario reduplication patterns originally appearing in Gill (1957). This current analysis incorporates new data and reveals “the greater complexity in the morphology-prosody interface.” It examines verbal reduplication at the semantic, morphological and phonological levels, providing new analyses for two reduplication patterns: one that involves consonant doubling and the other the deletion of the root-final vowel. Of particular interest to typologists is Rose’s claim that, contrary to the widely accepted hypothesis that “all languages that have partial reduplication also have total reduplication” (Moravcsik 1978, 328, Rubino 2005, 12), Trinitario has only partial but not full reduplication. “Reduplication in Karitiana (Tupi)” by Luciana R. Storto focuses on verbal reduplication and the reduplication of a quantifying affix to adjectival roots. The reduplication of both verbs and adjectives carries an iconic interpretation of augmentation. Storto points out that although full reduplication of some verb roots expresses plurality of the event (pluractionality), the same function is indicated in other verbs by a suppletive plural form. An affix -Vra intensifies the quantification of adjectival roots to which it is attached. Ideophonic phrases used to indicate a durative event or the repetition of an activity are distinguished from reduplication, and what might appear to be nominal reduplication is identified as symbolic, where the repeated noun root indicates a typically repetitive movement or characteristic sound associated with the noun. To conclude the volume, Hein van der Voort surveys a diverse array of language families and isolates found in the Guaporé-Mamoré region along the

reduplication in south america: an introduction

11

border between Brazil and Bolivia. The functions and characteristics of productive reduplication in individual languages are described and compared. In his article, “Is reduplication an areal feature of the Guaporé-Mamoré region?” van der Voort highlights several grammatical traits that are shared among these languages and suggest the possibility of a linguistic area, and he considers whether reduplication might be an areal feature of the region.

Map The languages featured in the present volume are located on the map on page 12. They are indicated by numbers that are explained in the subsequent table. Twelve articles focus on reduplicative phenomena in individual languages, indicated on the map by a number, e.g., 1, 2, etc. Two articles discuss reduplication phenomena across languages in specific regions. These are marked on the map by a letter plus number combination, e.g., A1 for the Andean region and G1 for the Guaporé-Mamoré region. One article discusses reduplication across languages of the Tupi family, marked by, e.g., T1.

Acknowledgements Now that the articles have appeared, we can reveal the identity of and express our gratitude to the reviewers of the articles: Mily Crevels, Sebastian Drude, Pattie Epps, Sidi Facundes, Simeon Floyd, Vilacy Galucio, Ana Fernández Garay, Rik van Gijn, Katharina Haude, Ben Hermans, Lev Michael, Denny Moore, Helder Perri Ferreira, Walkiria Neiva Praça, Francesc Queixalós, Eduardo Ribeiro, Françoise Rose, Frank Seifart and Pilar Valenzuela. Furthermore, we thank the editors at Brill and Brill’s anonymous reviewers, as well as Laurie Meijers of TAT Zetwerk. We also thank the following colleagues for their valuable contributions through personal communication about the languages of their expertise: Swintha Danielsen on Baure, Liliana Fernández Fabian on Ashaninka, Suzi Lima on Yudja. We are furthermore very grateful to Willem Doelman who made the map for our introduction. And finally we are indebted to our authors, who have patiently endured the delays and the critical comments from various reviewers. We hope they will be pleased with the final product.

12

map 1

goodwin gómez and van der voort

Location of the languages featured in the present volume. (© Willem Doelman)

reduplication in south america: an introduction table 1

13

Languages featured in this book

1 2 3 4 5 6 7 8 9 10 11 12

Mojeño Trinitario (Arawak) Cha’palaa (Barbacoa) Mẽbengokre (Jê) Hup (Nadahup) Cavineña (Tacana) Aweti (Tupi) Karitiana (Tupi) Nheengatu (Tupi) Yanomae (Yanomami) Mapuzungun (isolate) Movima (isolate) Trumai (isolate)

A1 A2 A3 A4 A5 A6 A7 A8 A9

Aymara (Aymara) Jaqaru (Aymara) Shuar (Jivaro) Quechua (Quechua) Uru (Uru-Chipaya) Kallawaya (mixed) Cholón (isolate) Leko (isolate) Pukina (isolate)

G1 G2 G3 G4 G5 G6 G7 G8 G9 G10

Baure (Arawak) More (Chapacura) Oro Win (Chapacura) Wari’ (Chapacura) Arikapu (Macro-Jê) Djeoromitxi (Macro-Jê) Rikbaktsa (Macro-Jê) Mosetén (Mosetén) Latundê (Nambikwara) Mamaindê (Nambikwara)

G11 Sabanê (Nambikwara) G12 Southern Nambikwara (Nambikwara) G13 Ese Ejja (Pano-Tacana) G14 Mekens (Tupi) G15 Aikanã (isolate) G16 Cayuvava (isolate) G17 Itonama (isolate) G18 Kanoê (isolate) G19 Kwaza (isolate) G20 Yurakaré (isolate) T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15 T16 T17 T18 T19 T20 T21

Juruna (Juruna) Munduruku (Munduruku) Makurap (Tupari) Tupari (Tupari) Anambé (Tupi-Guarani) Araweté (Tupi-Guarani) Asurini of Tocantins (Tupi-Guarani) Avá-Canoeiro (Tupi-Guarani) Chiriguano (Tupi-Guarani) Guajá (Tupi-Guarani) Guajajara (Tupi-Guarani) Ka’apor (Tupi-Guarani) Mbyá (Tupi-Guarani) Paraguayan Guarani (Tupi-Guarani) Parakanã (Tupi-Guarani) Parintintin (Tupi-Guarani) Siriono (Tupi-Guarani) Tapirapé (Tupi-Guarani) Tupinambá (Tupi-Guarani) Yuki (Tupi-Guarani) Kokama (Tupi-Guarani/mixed?)

14

goodwin gómez and van der voort

References Abraham, Werner. 2005. Intensity and diminution triggered by reduplicating morphology: Janus-faced iconicity. In Hurch (ed.), 547–568. Alderete, John, Jill Becknam, Laura Benua, Amalia Gnanadesikan, John McCarthy, and Suzanne Urbanczyk. 1999. Reduplication with fixed segmentism. Linguistic Inquiry 30/3: 327–364. Ammann, Andreas, and Aina Urdze (eds.). 2007. Wiederholung, Paralellismus, Reduplikation: Strategien der multiplen Strukturanwendung. [Diversitas Linguarum 16]. Bochum: Universitätsverlag Dr. N. Brockmeyer. Araújo, Leopoldina. 2008. Reduplicação e ênfase no Parkatêjê. In Topicalizando MacroJê, ed. Stella Telles and Aldir Santos de Paula, 215–227. Recife: Nectar. Borges, Mônica Veloso. 2008. Reduplicação verbal em Avá-Canoeiro (Tupi-Guarani): uma língua brasileira ameaçada de extinção. UniverSOS. Revista de Lenguas Indígenas y Universos Culturales 5: 233–243. Bruno, Ana Carla. 2003. Reduplicação em Waimiri Atroari. Amérindia 28: 83–94. Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam/Philadelphia: John Benjamins. Comrie, Bernard, Martin Haspelmath, and Balthasar Bickel. 2008. The Leipzig glossing rules: Conventions for interlinear morpheme-by-morpheme glosses. http://www .eva.mpg.de/lingua/resources/glossing-rules.php (accessed March 22, 2013). Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press. Danielsen, Swintha. 2007. Baure: An Arawak language of Bolivia. [Indigenous Languages of Latin America (ILLA) 6]. Leiden: CNWS Publications. Dixon, R.M.W., and Allan R. Vogel. 1996. Reduplication in Jarawara. Languages of the World 10: 24–31. Everett, Daniel, and Barbara Kern. 1997. Wari’: The Pacaas Novos language of Western Brazil. London: Routledge. Everett, Daniel, and Lucy Seki. 1986. Reduplication and CV skeleta in Kamaiurá. Linguistic Inquiry 16/2: 326–330. Fargetti, Cristina Martins 1998. Re-reduplicação em Jurúna. Actas III Jornadas de Lingüística Aborigen 199–205. Buenos Aires. Fischer, Olga. 2011. Cognitive iconic grounding of reduplication in language. In Semblance and signification, ed. Pascal Michelucci, Olga Fischer and Christina Ljungberg, 55–81. Amsterdam/Philadelphia: John Benjamins. García-Medall, Joaquín. 2003. Sobre reduplicación morfológica en lenguas amerindias. In I Simposio Antonio Tovar sobre lenguas amerindias, ed. Emilio Ridruejo and Mara Fuertes, 25–66. Tordesillas: Casa del Tratado/Instituto Interuniversitario de Estudios de Iberoamérica y Portugal. Gijn, Rik van. to appear. Reduplication in Yurakaré. In Word formation in South Ameri-

reduplication in south america: an introduction

15

can languages, ed. Katja Hannss, Swintha Danielsen and Fernando Zúñiga. Amsterdam/Philadelphia: John Benjamins. Gil, David. 2005. From repetition to reduplication in Riau Indonesian. In Hurch (ed.), 31–64. Gill, Wayne. 1957. Trinitario Grammar. Manuscript. San Lorenzo de Mojos: Misión Nuevas Tribus. Gomes, Dioney Moreira. 2007. Reduplicação verbal em Mundurukú. In Línguas e culturas Tupí, I, ed. Aryon Dall’Igna Rodrigues and Ana Suelly Arruda Câmara Cabral, 391–396. Campinas: Curt Nimuendajú/Brasília: LALI, UnB. Goodwin Gómez, Gale. 2009. Reduplication, ideophones and onomatopoeic repetition in the Yanomami languages. In Hurch (ed.), 21–38. Haude, Katharina. 2006. A Grammar of Movima. PhD diss., Radboud Universiteit. http://webdoc.ubn.ru.nl/mono/h/haude_k/gramofmo.pdf (accessed March 22, 2013). Hurch, Bernhard. 2005ff. Graz Database on Reduplication. http://reduplication.uni-graz .at/redup/ (accessed March 22, 2013). Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Hurch, Bernhard, with Veronika Mattes and Ursula Stangel, ed. 2009. Reduplication: Diachrony and productivity, a reprise. Special issue of Grazer Linguistische Studien, 71. Inkelas, Sharon, and Cheryl Zoll. 2005. Reduplication: Doubling in morphology. Cambridge: Cambridge University Press. Kouwenberg, Silvia, ed. 2003. Twice as meaningful: Reduplication in pidgins, creoles and other contact languages. [Westminster Creolistics Series 8]. London: Battlebridge Publications. Meira, Sérgio. 2000. Reduplication in Tiriyó (Cariban). [Languages of the World 17]. München: Lincom Europa. Michaud, Alexis, and Aliyah Morgenstern, ed. 2007. La réduplication. Special issue of Faits de langues, 29. Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of human language, vol. 3, ed. Joseph H. Greenberg, 297–334. Stanford: Stanford University Press. Picanço, Gessiane. Lobato. 2005. Mundurukú: Phonetics, Phonology, Synchrony, Diachrony. PhD diss, University of British Columbia. http://www.etnolinguistica.org/tese :picanco_2005 (accessed March 22, 2013). Pott, August Friedrich. 1862. Doppelung (Reduplikation, Gemination) als eines der wichtigsten Bildungsmittel der Sprache, beleuchtet aus Sprachen aller Welttheile. Lemgo/ Detmold: Meyer. Queixalós, Francesc. 2002. Sur la distensivité. In La pluralité, ed. Jacques François, 55–71. [Mémoires de la Société de Linguistique de Paris XII]. Leuven: Peeters.

16

goodwin gómez and van der voort

Rose, Françoise. 2005. Reduplication in Tupi-Guaraní languages: going into opposite directions. In Hurch (ed.), 351–368. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. . 2011 [2005]. Reduplication. In The world atlas of language structures online, ed. Matthew Dryer and Martin Haspelmath, chapter 27. Munich: Max Planck Digital Library. http://wals.info/chapter/27 (accessed March 22, 2013). Sakel, Jeanette. 2004. A grammar of Mosetén. [Mouton Grammar Library 33]. Berlin/ New York: Mouton de Gruyter. Seki, Lucy. 1984. A reduplicação em Kamaiurá e Tupinambá. Anais do VIII Encontro Nacional de Lingüística, 49–56. Rio de Janeiro: PUC. Stolz, Thomas. 2007a. Das ist doch keine Reduplikation! Über falsche Freunde bei der Suche nach richtigen Beispielen. In Ammann and Urdze (eds.), 47–80. . 2007b. Re: duplication: Iconic vs counter-iconic principles (and their areal correlates). In Europe and the Mediterranean as linguistic areas: Convergencies from a historical and typological perspective, ed. Paolo Ramat and Elisa Roma, 107–132. Amsterdam/Philadelphia: John Benjamins. Stolz, Thomas, Cornelia Stroh and Aina Urdze. 2011. Total reduplication: The areal linguistics of a potential universal. [Studia Typologica 8]. Berlin: Akademie Verlag. Voort, Hein van der. 2003. Reduplication of person markers in Kwaza. Acta Linguistica Hafniensia 35: 65–94.

chapter 2

Reduplication in Mapuzungun: Form and Function Fernando Zúñiga and Antonio Edmundo Díaz-Fernández

Mapuzungun has reduplicative structures based on elements other than verb stems that are of very limited productivity. With verb stems, however, several formal patterns can be distinguished, which consist of the repetition of the lexical verb stem plus the addition of an apparently grammaticalized version of one of three verb roots or a zero morpheme. The previous literature has attempted to identify form/function correlations for these more or less productive verbal reduplicative patterns, and the present paper contributes to the discussion by surveying older studies and exploring several cases that suggest that such form/function correspondences are substantially less straightforward than a casual observer might think.

1

Introduction

In this paper we present an overview of reduplication in Mapuzungun, the language of the Mapuche people who live in what is now Argentina and Chile. The language shows different degrees of vitality in the areas where it is still spoken or at least known; it seems to be more vital in Chile than in Argentina. It seems to be in a terminal stage in many rural communities in Argentina, especially in the provinces of La Pampa, Río Negro and Chubut. Since most of the Mapuche population lives on the western side of the Andean ridge, the majority of the native speakers are found in Chile. The language, also called Mapudungun, Mapuchezungun, Mapuchedungun, Chezungun and Chedungun (cf. Díaz-Fernández 2006, Zúñiga 2006, 44) includes several mutually intelligible dialects; Chesumun or Williche, an obsolescent variety found in southern Chile, is markedly different from the other dialects. Mapuzungun has been considered a language isolate, although some authors tried to relate it to different families. Among the attempts to establish a genetic relationship for this language we briefly mention the following ones. Stark (1970) tried to relate Mapuzungun to the Mayan family. Hamp (1971) followed Stark’s hypothesis and included the Uru-Chipaya group. Loos (1973) thought there were similarities between this language and Proto-Panoan. For Grimes (1978, 48), it is “possibly Penutian.” Key devoted two main papers to prove a genetic relationship with

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_003

18

zúñiga and díaz-fernández

“Araucanian” (i.e. the varieties of Mapuzungun) and the Pano-Tacanan family (Key 1978 a, 1978 b). Finally, Croese (1999) proposed a new relationship for the Mapuche language, this time with the Arawak family; this hypothesis was further explored by Díaz-Fernández (2011). The main sources on which this work is based are Augusta’s dictionary (1916), which we checked thoroughly to get as complete a list of possible cases of reduplication as possible, a corpus recorded by the first author in Mapuche communities in the province of Chubut, Argentina, from 1987 up to the present, and the second author’s field notes from his work in the province of Araucanía, Chile, from 1998 up to the present. Other sources we consulted are Díaz-Fernández (2003 vol. 2), Fernández Garay (2001), Harmelink (1996), Hernández Sallés et al. (1997), Smeets (2008), and Vúletin (1987). It is important to mention that several reduplicated items recorded by Augusta (1916) are obsolescent on both sides of the Andean ridge. Some of these items are still current in certain areas while others were unknown at least to the vernacular speakers who were interviewed in Mapuche communities in Chubut, Argentina. The remainder of section 1 sketches the basics of Mapuzungun phonology. After outlining a general terminological and analytic introduction to reduplication (section 2), we briefly consider some formal aspects of reduplication in Mapuzungun (i.e. morphological make-up of the reduplicated units and the lexical classes involved in reduplication) in section 3. Verbal reduplication is addressed in section 4, and section 5 closes the paper with some conclusions. The appendix briefly deals with inherent reduplication and noun reduplication. Compared to other languages of the Southern Cone, Mapuzungun has a relatively simple phonology.1 There are six vocalic phonemes, viz. u (/u/), e (/e/), o (/o/), a (/a/), i (/i/), and ü (/ɨ/), the latter of which occurs as epenthetic segment in order to avoid tautosyllabic consonant clusters. There are also three glides, viz. y (/j/), w (/w/), and g (/ɰ/), which can appear word-initially before the homorganic vowels i, u, and ü. In addition, word-final /o/ can be realized as [u] (except with monosyllables like ko ‘water’). The consonantal phonemes are shown in Table 1 below. The palato-alveolar and retroflex obstruents are affricates; the others are simple plosives. Liquids are approximants; all except

1 Mapuzungun examples are written here by means of the so-called Alfabeto Mapuche Unificado (AMU) ‘Unified Mapuche Alphabet,’ which was the product of the Encuentro para la Unificación del Alfabeto Mapuche (Meeting for the Unification of the Mapuche Alphabet), held in Temuco, Chile, in May 1986. The original AMU has d instead of z for the interdental fricative.

19

reduplication in mapuzungun: form and function table 1

Mapudungun consonantal phonemes

Inter- Dento- Palato-alveolar/ Labial dental alveolar palatal Retroflex Velar nasal m /m/ obstruent p /p/ fricative f /f/ liquid

ṉ /n̟ / ṯ /t̟/ z /θ/ l ̱ /l̟/

n /n/ t /t/ s /s/ l /l/

ñ /ɲ/ ch /ʧ/ (sh) /ʃ/ ll /ʎ/

tr /tʂ/

ng /ŋ/ k /k/

r /ɻ/

r are lateral. The phonemic status of sh /ʃ/ is disputable and not accepted by all scholars. Well-formed syllables in the language show the form (C1)V(C2), where C1 can be any consonant or glide except g, V can be any vowel, and C2 can be any nonobstruent consonant or glide. Words can be mono- or plurisyllabic, and stress rules make crucial reference to syllable type. Roughly, closed syllables attract stress in disyllabic words, e.g. rúka ‘house,’ máñke ‘condor,’ kurám ‘egg,’ tralkán ‘thunder.’ Words with more than three syllables usually have secondary stress, e.g. àchawáll ‘hen, rooster,’ but in verbs some affixes attract stress. Most native verb and noun roots are monosyllabic or disyllabic, but final words, especially verb forms, can have many more syllables due to extensive suffixation (and, although less extensively, compounding); most trisyllabic words are Quechua (e.g. achawall) or Spanish (e.g. kuchillo ‘knife’) loanwords. See Zúñiga (2006, 63 f.) for more details.

2

Reduplication

Reduplication is a morphological process that involves the repetition of a linguistic unit such as a phoneme, a morpheme or a lexeme (cf. Dubois et al. 1998, 524). According to the morphological make-up of these constructions, two major types can be distinguished: (a) complete or full reduplication and (b) partial reduplication (cf. Dubois 1998, 524; Rubino, 2005, 11;). For Rubino, the former type involves the repetition of a lexical base, be it a root, a stem, or a word, while the latter type involves the repetition of a segment (vowel lengthening, consonant gemination), a syllable, or even a “nearly complete copy of a base” (2005, 11). Rubino also characterizes reduplication as either (a) simple (when a unit is repeated without segmental changes), (b) complex (when a given unit is not simply copied but changed in some way), or (c)

20

zúñiga and díaz-fernández

automatic (when reduplication is obligatory in combination with some other morphological process and does not add meaning by itself to the construction as a whole) (cf. Rubino 2005, 11–16, 18). As far as the functional dimension is concerned, Rubino’s overview (2005, 19–22) identifies a number of possible meanings of reduplicated forms from a cross-linguistic perspective, viz. number (plurality, collectivity, distributivity), aspectuality (continuation, iteration, completion, inchoativity), and several other categories (tense, intensity, transitivity, conditionality, reciprocity, etc.). Working on reduplication in Kashmiri, Koul (2005, 151) says that “reduplication is mainly used for emphasis, plurality, generality, intensity, or to show continuation of an act.” Sandvig (1987, 148–151), who focuses on Mapuzungun, states that the functions of reduplication in verbs are to indicate continuity, repetition or intensity of the verbal event and distributivity or individualized plurality.2 As we will see in what follows, Mapuzungun features both full and partial reduplication patterns of the simple type. Whereas word formation by reduplication does not seem to be productive in present-day Mapuzungun, inflectional reduplication appears to be used with verb stems in order to express some, but not all, categories identified by Rubino (2005).

3

Formal Aspects of Reduplicative Morphology in Mapuzungun

1:aspectMapuzungun full reduplication mainly involves verb stems (cf. also Malvestitti 2003, 268–269, Díaz-Fernández 2009), which are typically disyllabic roots that may or may not be followed by different verb suffixes. Other elements involved in reduplicative patterns include adjective stems, adverbs, indefinite pronouns, and question words. We deal with these in turn in what follows. 3.1 Verb Stems Reduplicated verbal stems constitute the largest group, viz. 172 items in Augusta (1916). We deal with instances of this processes in detail in section 4; examples (1) and (2) illustrate two different patterns found, viz. mere stem reduplication and stem reduplication plus tu- ‘take’ respectively.3

2 Individualized plurality implies that the action focuses on each of the participants in the event and that each of them experiences the event individually. 3 There are several different -tu’s in Mapuzungun, and we have glossed them differently here because it is not entirely clear whether they have different origins or they are all special(ized)

reduplication in mapuzungun: form and function

(1)

a. fay-ün ferment-nfin ‘ferment (v.)’

b. fay~fay-ün ferment~ferment-nfin ‘start sweating’ (Augusta 1916)

(2)

a. ira-n split-nfin ‘split (v.)’

b. ira~ira-tu-n split~split-take-nfin ‘chop up (wood)’ (Augusta 1916)

21

Even though, as mentioned in the introduction, most native verb roots are monosyllabic or disyllabic, it is verb stems that are reduplicated, and there is no prosodic restriction precluding trisyllabic or tetrasyllabic elements from undergoing reduplication. For instance, stems with incorporated nominals like küchalongko- ‘wash one’s hair’ (kücha- ‘wash,’ longko ‘head, hair’) and denominal stems like kofketu- ‘eat bread’ (kofke ‘bread,’ -tu ‘vbzr’) can also be reduplicated. So can Spanish loan roots like fakuna- ‘vaccinate’ (< Sp. vacunar ‘vaccinate’) and kompone- ‘repair’ (< Sp. componer ‘repair’). 3.2 Adjective Stems Very few adjective stems in our corpus seem to occur productively in reduplication. The most frequent example is ka ‘other, another, different’, which appears marked for nonsingular number and reduplicated with this inflection in (3): (3) ka-ke~ka-ke wingka kontupa-e-n-ew. other-nsg~other-nsg white.person come.to.see-inv-1sg.pat-3.agt ‘Different (kinds of) white people come to see me.’ Furthermore, the resulting form may be turned into a verb by taking aspectual and modality affixes: (4) Füta Chaw4 ka-ke~ka-ke-küno-y ñi pu püñeñ great father other-nsg~other-nsg-res-real.3 3.poss pl child ‘The Great Father set different kinds (of people) among his children.’

grammaticalized instances of the verb root tu- ‘take.’ In addition to the adverbializer and the verb root (the latter of which occurs on most reduplicated forms discussed in section 4), there is a verbalizer -tu and a telicizer -tu. 4 Füta Chaw ‘Great Father’ is a theonym that shows a monotheist idea, perhaps this idea was developed under Catholic influence, although it is not the only divinity whom the Mapuche address in rituals.

22

zúñiga and díaz-fernández

3.3 Adverbs (and Nouns) Reduplicated adverbial stems are seldom found; more often than not, adverbs undergoing reduplication consist of nouns or adjectives (like pülka ‘dirty, blemished with crap’ in example (5)) plus the adverbializing suffix -tu: (5)

pülka~pülka-tu zungu-y. dirty~dirty-advzr speak-real.3 ‘He speaks inefficiently.’

Such an adverb may in turn be used as a verb by taking the appropriate morphology: (6)

yoz pülka~pülka-tu-a-n ñi wingkazungun. more dirty~dirty-advzr-fut-real.1sg 1sg.poss Spanish.language ‘I will speak Spanish (lit. my Whiteman’s language) more inefficiently.’

In (7), the adjective/adverb root denotes the idea of tying up, entangling, and (metaphorically) getting entangled with speech, and no adverbializing morphology is needed. Something similar happens in (8), based on the adjective/noun ketro ‘stammerer, stutterer; mute; blunt’: (7) kelta~kelta5 zungu-y. entangle~entangle speak-real.3 ‘He speaks inefficiently.’ (8) ketro~ketro zungu-y. blunt~blunt speak-real.3 ‘He speaks inefficiently.’ In contrast, the adjective/adverb wichu ‘apart’ does not appear reduplicated as a separate word, but it occurs used as a verbal stem with the same kind of aspectual morphology we already saw in (4) above to form wichukünon ‘set apart people or things, one separated from another.’

5 Cf. keltañmaw ‘small metallic pieces to join parts of a Mapuche jewel’ and keltañmawün ‘get the horse entangled with its own reins.’

reduplication in mapuzungun: form and function

23

(9) Füta Chaw wichu~wichu-küno-fi-∅6 mapu ñi great father apart~apart-res-3 pat-real.3 land 3.poss mule-a-m ñi pu püñeñ inhabit-fut-nfin 3.poss pl child ‘The Great Father set apart lands for his children to inhabit.’ Lastly, observe in (10) that the noun trokiñ ‘division, part, portion, section’ can take the nonsingular suffix -ke (normally used with adjectives only, like in example (3) above) and be used as a verb as well, in a clearly distributive sense: (10) trokiñ-ke~trokiñ-ke-küno-y ñi pu püñeñ. part-nsg~part-nsg-res-real.3 3.poss pl child ‘He set apart (different classes of) his children.’ It is not entirely clear at this point how the interplay between nonsingular -ke and resultative -küno works in examples like (4), (9), and (10) (cf. also examples (12) and (13) further down). Some nouns occur as reduplicated stems without either nonsingular -ke or the resultative -küno but follow one of the normal verbal patterns instead, like the one illustrated in (11a and 11b) (chiwüz- as a verb root and chiwüzchiwüz as noun are unattested): (11) a. chiwüz owl ‘owl’

b. chiwüz~chiwüz-nge7-n owl~owl-vblz-nfin ‘rotate continuously’ (Augusta 1916)

These may be the remnants of an erstwhile more productive and perhaps regular way of expressing distributivity, but more research is needed here.

6 Although verbal reduplication is typically built on reduplicated verb stems plus the verbs ye- ‘carry,’ tu- ‘take,’ or nge- ‘be’ (cf. section 4), forms with a resultative marker -küno (probably a grammaticalized version of the verb root küno- ‘leave’) can occasionally be found as well. 7 As already mentioned for tu (footnote 4), there are also several nge’s in Mapuzungun. We have glossed them differently depending on their function, viz. as verbalizer -nge (example 15) or as the light verb ‘be, exist’ in most reduplicated forms in section 4 and examples like nge-la-y dungu (be-neg-real.3 matter) ‘there is no (bad) news.’ Lastly, the passive suffix is also -nge (not illustrated in this paper).

24

zúñiga and díaz-fernández

3.4 Pronouns Kiñe-ke ‘some,’ based on kiñe ‘(any)one,’ is one of the few words belonging to this part of speech that can undergo reduplication: (12) re kiñe-ke~kiñe-ke müle-y exclusively one-nsg~one-nsg be-real.3 ‘There are scarcely any.’ As with the other parts of speech outlined above, kiñe-ke ‘some’ as well as kis(h)u ‘he, himself / she, herself / they, themselves; own’ can also be used as verb stems and both of them undergo reduplication (13) and (14): (13) kiñe-ke~kiñe-ke-le-y one-nsg~one-nsg-be-real.3 ‘To be one here, another there, and another farther (i.e. all scattered)’ (14) Füta Chaw kishu~kishu-küno-y ñi pu püñeñ great father themselves~themselves-res-real.3 3.poss pl child ‘The Great Father left his children separated by themselves (i.e. different kinds of children).’ 3.5 Question Words There is only one question word found reduplicated, viz. chumte ‘how much’: (15) chumte~chumte ñi kim-ün fey ta how.much~how.much 1sg.poss know-nfin that disc kimeltu-a-fi-ñ teach-fut-3.pat-real.1sg ‘I will teach him what I know.’

4

Verbal Reduplication

As the following examples show, bare verb root reduplication can be attested in Mapuzungun: (16) a. ina-n follow-nfin ‘follow’

b. ina~ina-n-chi pu che follow~follow-nfin-attr pl person ‘subordinate people’

reduplication in mapuzungun: form and function

25

However, verb root reduplication is often combined with the addition of a suffixed grammaticalized verb root, as will be shown in this section. Different authors have dealt with function-form correspondences in the context of verbal reduplication in different ways, and section 4.1 surveys the accounts given by Salas (2006), Harmelink (1996) and Smeets (2008). Section 4.2 deals with a number of questions raised by these authors’ accounts in the light of the results of our own research. 4.1 Previous Studies of Verbal Reduplication Salas (2006, 174) addresses what he calls temas iterativos ‘iterative stems’ in the following terms. They are formed by reduplicating the root (or stem, as we shall see further down) and suffixing -nge ‘be,’ (17b), -tu ‘take’ (18b), or -ye ‘carry’ (19b): (17) a. pi-n say-nfin ‘say’

b. pi~pi-nge-n say~say-be-nfin ‘say repeatedly’

(18) a. rüngkü-n jump-nfin ‘jump’

b. rüngkü~rüngkü-tu-n jump~jump-take-nfin ‘bounce’

(19) a. küpal-ün bring-nfin ‘bring’

b. küpal~küpal-ye-n bring~bring-carry-nfin ‘be continually bringing, bring many times’

Stems involving reduplication plus -nge ‘be’ are reported to be basically intransitive, whereas those taking -tu ‘take’ can be stems that have been previously transitivized, as can those taking -ye ‘carry’ (in example (19a) above, küpa-l- is such a transitivized stem: küpa- ‘come’ + -l ‘caus’). Salas further notes that the meaning of ye- reduplicative forms “oscillates between duration and iteration” (p. 174). Harmelink (1996, 245–246) identifies the same three formal patterns as Salas but characterizes their functional yield in somewhat different terms: those taking -nge ( fane- ‘weigh (itr.), be heavy,’ tuku- ‘put (on),’ fem- ‘do so,’ and winguz- ‘drag’) indicate intensity or currently performed action, i.e. progressive aspect; those taking -tu (mawüṉ - ‘rain,’ kurüf- ‘(wind) blows,’ küzaw- ‘work,’ and rüngkü- ‘jump’) mean ‘try, test’; lastly, those taking -ye (kücha- ‘wash,’ tuku- ‘put (on),’ küpal- ‘bring,’ and ye- ‘carry’ itself) are distributive. Some of Harmelink’s examples are given below:

26

zúñiga and díaz-fernández

(20) a. mawüṉ -ün rain-nfin ‘rain (v.)’

b. mawüṉ ~mawüṉ -nge-y. rain~rain-be-real.3 ‘It is raining.’

(21) a. tuku-n put-nfin ‘put (on)’

b. tuku~tuku-tu-fi-ñ kamisa. put~put-take-3.pat-real.1sg shirt ‘I am trying the shirt on.’

(22) a. kücha-n wash-nfin ‘wash’

b. kücha~kücha-ye-fi wash~wash-carry-3.pat ‘She/he is doing the dishes.’

ti art

ral ̱i. plate

Smeets (2008, 304–307) identifies four “stem formatives” instead of only three: -nge, -tu, -ye, and -∅. Reduplicated verb forms taking -nge are intransitive but may be based on intransitive or transitive roots, and they are said to denote “a lasting situation in which the event takes place repeatedly and with intensity”: (23) a. wirar-ün shout-nfin ‘shout’

b. wirar~wirar-nge-n shout~shout-be-nfin ‘shout constantly’ (Smeets 2008, 305)

In contrast, forms taking -tu retain the valency of their root (i.e., they are either intransitive or transitive) and denote “a non-serious event and [indicate] that the action is performed for fun, in a playful way, or that the subject is pretending or just attempting to perform the action,” or “are clearly onomoatopoeic.” (24) kinging~kinging-tu-n whine~whine-take-nfin ‘whine like a dog’ (Smeets 2008, 304) Reduplicated verb forms taking -ye are transitive and transitive-based, and they denote “a lasting situation in which the event takes place repeatedly and with intensity”: (25) a. neng-ün move-nfin ‘move (itr.)’

b. neng-üm~neng-üm-ye-n move-caus~move-caus-carry-nfin ‘move (tr.) constantly’ (Smeets 2008, 306)

reduplication in mapuzungun: form and function

27

Significantly, Smeets observes that nge-forms and ye-forms stand in paradigmatic opposition, cf. intransitive af- ‘end, finish’ in (26a, 26b) and its causative, i.e. transitive, counterpart af-üm- > apüm- ‘finish up’ in (27a, 27b):8 (26) a. af-ün end-nfin ‘end (itr.)’

b. af~af-nge-n end~end-be-nfin ‘end (itr.) bit by bit’

(27) a. ap-üm-ün end-caus-nfin ‘end (tr.), finish (tr.)’

b. ap-üm~ap-üm-ye-n end-caus~end-caus-carry-nfin ‘finish up (tr.) bit by bit’

Finally, forms taking the “stem formative” -∅ are intransitive, but Smeets notes (pp. 306–307) that some of such reduplicated forms may denote postures of manifold subjects resulting from an earlier event (e.g. trana~trana- ‘be all over the floor,’ from trana- ‘fall’) and yet others denote actions that are performed quickly (e.g. püra~püra- ‘go (quickly) upstairs,’ from püra- ‘ascend’). In all fairness, neither Salas’ passage nor Harmelink’s outline claim to be comprehensive analyses of verbal reduplication in the language, and so it is perhaps only natural that there are some problems with their accounts. First, both state that the reduplicated units are only verbal roots, but this is clearly wrong because other grammatical categories can be reduplicated as well, as we have seen in section 3. Their own examples show that stems consisting of roots and affixes can also be reduplicated: as mentioned above, küpal- ‘bring’ is bimorphemic (from küpa- ‘come’ and the causative suffix -l). The same is true of Smeets’ account, which only mentions roots but includes one example showing the alternation between the root af- ‘end (itr.)’ and the bimorphemic stem af-üm- > ap-üm- ‘finish up (tr.),’ cited in (27b) above. By a similar token, the reduplicated form of feypi- ‘say’ appears not to be *fey-pi~pi-nge but feypi~feypi-nge- (from the demonstrative fey and pi- ‘say’).9,10

8

9 10

When the verbal root af- takes the causative suffix, it changes into ap- (thus, the causative stem is apüm- instead of *afüm-). Roots ending in f and g are subject to such morphophonemic changes ( f > p, g > k) when causativized with -m. There are two verbs for ‘say’ in Mapuzungun, viz. feypin, which precedes the direct quotation if there is one, and pin, which follows it. Smeets (2008, 307) explicitly mentions ( fey-)pi~pi-nge- ‘say (itr.) constantly’ and ( fey-)pi~ pi-ye- ‘say constantly (tr.)’ as grammatical, but without giving actual examples. All our language consultants agree on the ungrammaticality of such forms.

28

zúñiga and díaz-fernández

Second, from what Harmelink explicitly says, it would seem that the second verb stem (-nge, -tu or -ye) is chosen on grounds not related to transitivity but to the specific meanings to be attained, many of which might be relevant for a considerable number of verbs. The default seems to be that different stems, typically based on different roots, take different second verbs when reduplicated. Smeets’ analysis suggests that the choice between -nge and -ye is basically related to a transitivity opposition, whereas the choice between those two morphemes and -tu is related to the semantic opposition between ‘for fun, playfully’ on the one hand and ‘durative with repetition and intensity’ on the other. 4.2 Problems and Recalcitrant Issues Elicitation sessions conducted by the second author of the present paper tend to confirm the intuitions advanced by Salas, Harmelink and Smeets to some extent, but the emerging picture is more complex than their accounts suggest. First, even though the informants’ grammaticality judgments are heterogeneous and somewhat variable when faced with different reduplicated forms built on the same stems, native speakers seem to have no difficulties whatsoever in identifying the usual second verb employed with a particular reduplicated verbal stem independent of context. Many lexical verbs, however, can indeed occur reduplicated with more than one of the grammaticalized verbs tu- ‘take,’ nge- ‘be,’ and ye- ‘carry.’ Second, the nge/ye-opposition seems to be less robust than Smeets’ data suggest (i.e., the opposition appears to be limited to some verbs only). The verbs explicitly mentioned or discussed by Smeets as allowing the paradigmatic nge/ye-opposition are af- ‘end (itr.)’ vis-à-vis apüm- ‘finish (tr.)’ and pi‘say.’ Our consultants confirmed this, but any attempts to elicit further verbs that might behave as shown in (26a, b) and (27a, b) above failed. It seems that, at least in the speech of our informants, reduplicated forms with -ye are somewhat less productive in contrast with those built with -tu. Our results basically confirm Harmelink’s account: some sort of distributivity seems to be the basic meaning of reduplication with -ye (example (28), based on katrü‘cut, chop’), even though other cases are apparently better covered by the notion of intensity (example (29), based on küllkaw- ‘move (a liquid) to and fro (itr.)’): (28) katrü~katrü-ye-y mamüll. cut~cut-carry-real.3 wood ‘She/he chopped different kinds of wood.’

reduplication in mapuzungun: form and function

29

(29) küllkaw~küllkaw-tu-ye-n move~move-take-carry-nfin ‘strongly shake (a liquid within a bottle) (tr.)’ (Augusta 1916, 85) Third, the notion of intensity seems to be more robustly connected to reduplication with -nge, as follows from to a number of Augusta’s (1916) dictionary entries like the following, based on lüykü- ‘leak’ and ngüllfü-w- ‘bend (itr.)’ respectively (the latter verb stem includes the reflexive/anticausative suffix -w):11 (30) l ̱üykü~l ̱üyku-nge-n leak~leak-be-nfin ‘leak intensely, leak a lot’ (Augusta 1916, 118) (31) ngüllfü-w~ngüllfü-w-nge-n bend-acaus~bend-acaus-be-nfin ‘bend, sway strongly’ (Augusta 1916, 58) The semantics in example (32), based on kewlu-n, a verbal noun meaning ‘blaze, flare’ (the verb is unattested) seems to be analogous to the one in (30) and (31). By contrast, (33), based on pinüf- ‘fly (typically inanimate objects that have no wings),’ illustrates the notion of plurality of participants: (32) kewlu~kewlu-nge-y kütral. flare~flare-be-real.3 fire ‘The fire blazes.’ (Augusta 1916, 223; from Panguipulli in southern Chile) (33) pinüf~pinüf-nge-n fly~fly-be-nfin ‘fly (many objects) (itr.)’ (Augusta 1916, 183) In addition, our consultants invariably emphasized the durative or progressive meaning of the reduplicated forms built with -nge. With some intransitive verbs such as lef- ‘run,’ ñam- ‘get lost,’ ruka- ‘build a house,’ and weyel‘swim,’ as well as with the derived ditransitive wül-el- (give/hand-appl) ‘give to,’ for instance, the progressive value is involved, and this was also confirmed

11

The intransitive counterpart of transitive ngüllf- ~ ngüllfü- ‘bend, shake’ is either ngülluor the derived anticausative/reflexive ngüllfü-w-.

30

zúñiga and díaz-fernández

by the vernacular speakers’ translation into Spanish of the examples using the [estar+gerund] construction plus some intensifying adverbial. When pressed to describe the difference between these reduplicated forms and simple Mapuzungun progressives with the suffixes -meke or -(kü)le, consultants would usually add ‘right now’ or ‘at this very moment’ to their original characterization of the reduplicated forms (cf. lef-meke-i = lef-küle-i ‘s/he is running’ vis-à-vis lef~lefnge-i ‘s/he is running (right now)’; see Zúñiga 2001 for more on the Mapuzungun progressives). Only sporadically would they add ‘intensely,’ ‘strongly,’ or the like. A related but different notion of ‘sustained or continued event’ can be identified in some entries in Augusta (1916): (34) chiwüz~chiwüz-nge-n turn~turn-be-nfin ‘turn/rotate continually’ (Augusta 1916, 24) (35) mülchong~mülchong-nge-n tilt~tilt-be-nfin ‘continually nod, shake one’s head’ (Augusta 1916, 136) Fourth, as noted by Harmelink, the reduplication with -tu is associated with the meanings ‘not seriously,’ ‘for fun’ and ‘playfully,’ as with virtually all the verbs mentioned in the examples. For the following example, our consultants agreed that the reading ‘not well’ was prominent: (36) katrü~katrü-tu-y mamüll. cut~cut-take-real.3 wood ‘She/he chopped firewood, but s/he did not do it well.’ With the derived ditransitive wül-el- (give/hand-appl-) ‘give to,’ both ‘playful’ and iterative readings are possible in an example like (37): (37) mamüll mew wül-el~wül-el-tu-fi-ñ. wood ppos give/hand-appl~give/hand-appl-take-3pat-real.1sg ‘I hit him/her playfully/repeatedly with a stick.’ (FZ’s field notes) Such a finding directly confirms Salas’ account in terms of temas iterativos ‘iterative stems,’ but note that it is semantically/pragmatically compatible with Harmelink’s and Smeets’ analysis in terms of playfulness as well. Augusta (1916) also records iterative stems:

reduplication in mapuzungun: form and function

31

(38) chapash~chapash-tu-e-n-ew. stroke~stroke-take-inv-1sg.real-3agt ‘She/he caressed me repeatedly.’ (Augusta 1916, 17) In fact, some additional forms given by Augusta (1916) lead to the conclusion that the form-function correspondences are even less straightforward than what the data presented hitherto suggest. Reduplication with -nge is also found with an (additional?) iterative meaning (39), and both reduplication with -tu and the reduplication with -ye are also found meaning ‘intensely, eagerly’ (40): (39) trültriw~trültriw-nge-n scream~scream-be-nfin ‘scream loudly and repeatedly’ (Augusta 1916, 232)12 (40) a. zewma~zewma-tu-n b. zewma~zewma-ye-n do~do-take-nfin do~do-carry-nfin ‘do (something) eagerly’ ‘do (something) eagerly’ (Augusta 1916, 30) In addition, possibly lexicalized versions of intensity (41) and distributive motion (42) appear with reduplication with -tu, and plurality of participants is found even with reduplication without -tu (i.e., with Smeets’s ∅ “stem formative”) (43): (41) a. koṉ a~koṉ a-tu-n b. koṉ a~koṉ a-tu-l-ün warrior~warrior-take-nfin warrior-warrior-take-appl-nfin ‘become daring/courageous’ ‘harangue (somebody) before battle’ (Augusta 1916, 93) (cf. koṉ a ‘young warrior’; koṉ a- as a verb is unattested) (42) mütrüm~mütrüm-tu-yaw-ün call~call-take-peramb-nfin ‘go about calling (somebody) here and there’ (Augusta 1916, 139) (43) anü~anü-n sit.down~sit.down-nfin ‘sit down (many people, or one after the other)’ (Augusta 1916, 9) 12

Trültriw- ‘scream’ refers to the specific scream of the southern lapwing (Vanellus chilensis). In the variety of the language once spoken in El Chalía, an Aonik’enk-Mapuche community, speakers call this bird tültiw; its name is tregül in other varieties of Mapuzungun.

32

zúñiga and díaz-fernández

Lastly, with some stems reduplication without a grammaticalized verb has an arguably inchoative yield: (44) wa~wa-n maize~maize-nfin ‘maize begin to bloom’ (Augusta 1916, 246) Other cases of reduplication without a grammaticalized verb are semantically akin to those indicating ‘intensity’ (45a) or iterativity (45b): (45) a. apo~apo-n fill-fill-nfin ‘fill too much (itr.)’

b. zungu~zungu-me-n speak~speak-loc-nfin ‘speak repeatedly’

We will summarize and discuss possible origins of verbal reduplication below, as part of the conclusion.

5

Inherent Reduplication

Most of the cases of reduplication found in our corpus can be classified as inherent reduplication, and it is rather frequent in the language. Most of the bases that show inherent reduplication in Mapuzungun are disyllabic (e.g. petran+petran13 ‘mildew’), with some few cases consisting of only one syllable, e.g. ki+ki ‘greater grison (Galictis vittata)’ and kaw+kaw ‘gull,’ but units consisting of more syllables are common, e.g. küchü+küchü ‘Harris’s hawk (Parabuteo unicinctus)’ and llamke+llamke ‘butterfly.’ All three admissible syllable structures in Mapuzungun are found in this morphological process, viz. [V], [CV] and [CVC]. Regarding word structure, both full and partial reduplication patterns are found. More often than not, the reduplicated element forms a nominal or a verbal stem that may be combined with further morphemes, which can be either affixes or roots, e.g.: (46) kuy+kuy-zewü bridge-mouse (cf. kuykuy ‘bridge’) ‘crossbeam below the top of the roof’ 13

Since this is inherent reduplication, we have used the + symbol, instead of a tilde ~, to indicate its submorphemic structure.

reduplication in mapuzungun: form and function

33

(47) witra+witra-l-we stand.up+stand.up-caus-nmlz ‘vertical beams of the Mapuche loom’ (48) shilli+shilli-tu-n chirp+chirp-take-nfin ‘chirp (of the diuca-finch (Diuca diuca))’ To be sure, the fact that we are unable to identify the stems from which the inherently reduplicated forms may have derived does not imply that these words were not cases of true reduplication at some point in the past; the reduplicated units may have been independent forms that disappeared in earlier stages of the language. Given that Mapuzungun has been written only scantily and comparatively recently, there may very well be undocumented diachronic changes that are responsible for the present state of affairs. Be that as it may, reduplication of nominal bases cannot be considered a productive lexeme-deriving process in present-day Mapuzungun (cf. Smeets’ 2008, 119 statement to the same effect). The Chubut corpus also includes certain items involving repeated kinship expressions like pal ̱u ‘paternal aunt,’ with varying grammatical roles: in one case the noun functions as agent of the verb (49), and in the others it appears to be an adverbial expression (50). In both cases the semantic yield of such a repetition is far from clear: (49) pal ̱u~pal ̱u(-tu) kaltu-lel-e-n-ew paternal.aunt~paternal.aunt-advzr put.wool-ben-inv-1.sg pat-3.agt ‘My (paternal) aunt covered me with wool (lit. put me some wool).’ (50) amu-a-n ñi pal ̱u~pal ̱u(-tu) mu go-fut-real.1.sg 1sg.poss paternal.aunt-paternal.aunt(-advzr) ppos kal ̱-me-tu-lu put.wool-loc-vblzr-nfin ‘I will go to my paternal aunt’s (place) for her to cover me with wool.’ Perhaps these cases are merely stylistic variations (involving an emphatic or affective component) and therefore not bona fide cases of inherent reduplication at all (Díaz-Fernández 1998).

34 6

zúñiga and díaz-fernández

Conclusion

The purportedly productive verbal reduplication in the literature can be summarized in Table 2 below, together with our own findings: table 2

Form-function correspondences of Mapuzungun reduplicative verbal patterns RED plus -tu ‘take’

Salas iterative Harmelink ‘try, test’ Smeets

‘playfully’ (i)tr.→(i)tr. (onomato-poetic) this study ‘playfully,’ iter (intens)

RED plus -nge ‘be’ RED plus -ye ‘carry’

RED

iterative intens, progressive intens+dur+iter (i)tr.→itr.

iterative, durative – distributive – intens+dur+iter tr.→tr.

(onomatopoeic)

intens, prog (iter)

distributive (intens)

onomatopoeic, inchoative

In the absence of both a better documented history of the language and known relatives, it is rather speculative to hypothesize how the verbal reduplication system might have originally been. Even though not much suggests that former neat form-function correspondences were disrupted in the course of time, whether due to contact with Spanish or not, it is conceivable that the different formal patterns had particular functional correlates associated with them. Reduplication plus -tu may have originally symbolized iterativity with an implicature of playfulness, that became lexicalized in some cases. Analogously, reduplication plus -nge must originally have symbolized iterativity and acquired the meaning ‘intensely,’ with a progressive implicature. Reduplication plus -ye is a good candidate for the distributive meaning, with the intensifying yield as an implicature as well. Lastly, the reduplicated stem without an accompanying verb (-tu, -nge, -ye) towards the end of the complex might be an older pattern still present in some frozen forms. The inchoative meaning is probably an implicature, just like playfulness. Mapuzungun reduplication shows a variety of formal patterns and two basic functions, (a) word-forming function that was presumably productive in the past as many examples of inherent reduplication may suggest and (b) grammatical function which is apparently fairly productive in the present-day language, as the cases dealt with above show this. Reduplication may have been (semi-) productively used in order to form new lexemes from nominal, adjectival or

reduplication in mapuzungun: form and function

35

verbal stems; there are units built upon elements that have no independent existence, at least presently. Even though Smeets correctly says that “reduplication is not a productive process in nominal morphology” (2008, 119) and not many lexemes are formed by means of reduplication in the obsolescent varieties at the beginning of the 21st century, evidence found in the lexicon of the language suggests that it may have been productive in the not-too-distant past. Nevertheless, as in many other indigenous languages of the Americas, reduplication in Mapuzungun is used to express meanings related to intensity, continuation, number of participants and repetition of the event depicted by verb roots/stems, according to an interplay of form and function that further research will help us elucidate.

Acknowledgments Except for bibliographical examples, the corpus on which this survey stands was recorded in the following Mapuche communities: Villarrica, Chile and Lago Rosario and Nahuelpán, Argentina. We thank the following native speakers of Mapuzungun for their cooperation: Leonel Lienlaf (Alepúe/Villarrica), Sergio Nahuelpán and Lorenzo Quilaqueo (Nahuelpán) and Carolina Castro, Rosario Cayecul and Mercedes Nahuelpán (Lago Rosario). We are grateful to the editors of the present volume, as well as to Ana Fernández Garay, for their numerous and valuable comments on a previous version of this paper. All errors and misconceptions remain our own.

References Augusta, Fray Félix José de. 1916. Diccionario araucano: Mapuche-Español, EspañolMapuche. Santiago: Imprenta Univeristaria. Croese, Robert A. 1999. Evidencias léxicas y gramaticales para una posible filiación del mapudungun con la familia Arawak. In Actas del VIII Congreso Internacional de la Asociación de Lingüística y Filología de la América Latina, ed. Elena M. Rojas Mayer, 383–386. San Miguel de Tucumán: ALFAL & Facultad de Filosofía y Letras, Universidad Nacional de Tucumán. Díaz-Fernández, Antonio. 1998. Una aproximación al análisis del epew. In Lingüística y literatura mapuche: Aproximaciones desde ambos lados de los Andes, ed. Lucía Golluscio and Yosuke Kuramochi, 79–86. Temuco/Buenos Aires: Universidad Católica de Temuco y Universidad de Buenos Aires. . 2003. Descripción del mapuzungun hablado en comunidades del departa-

36

zúñiga and díaz-fernández

mento Futaleufú, provincia del Chubut: Lago Rosario-Sierra Colorada y Nahuelpán. PhD diss., Universidad Nacional del Sur, Argentina. . 2006. Glosónimos aplicados a la lengua en el mapuche. In Anclajes 10, ed. José Javier Maristany, 95–111. Santa Rosa: Facultad de Ciencias Humanas, Universidad Nacional de La Pampa. . 2009. La reduplicación en mapuzungun. In Literatura—lingüística. Investigaciones en la Patagonia IV, ed. Guillermo Cegna and Alejandra Cornide (compiladores). Trelew: Instituto de Investigaciones Lingüísticas y Literarias de la Patagonia, Universidad Nacional de La Patagonia. CD-rom edition. . 2011. Relaciones genéticas del mapuzungun. Aportes para su ubicación dentro del stock Ecuatorial. In Investigaciones sobre lenguas indígenas sudamericanas, ed. Ana Fernández Garay and Antonio Díaz-Fernández, 67–113. Santa Rosa: EdUNLPam. Dubois, Françoise, Mathée Giacomo, Louis Guespin, Christiane Marcellesi, Jean-Baptiste Marcellesi, and Jean-Pierre Mével. 1998. Diccionario de lingüística. Madrid: Alianza Editorial. Fernández Garay, Ana V. 2001. Ranquel-Español / Español-Ranquel. Diccionario de una variedad mapuche de La Pampa (Argentina). [Indigenous Languages of Latin America (ILLA) 2]. Leiden: CNWS Publications. Grimes, Barbara, ed. 1978. Ethnologue. Huntington Beach: Wycliffe Bible Translators, Inc. Hamp, Eric. 1971. On Mayan-Araucanian Comparative Phonology. International Journal of American Linguistics 37/3: 156–159. Harmelink, Bryan. 1996. Manual de aprendizaje del idioma mapuche: Aspectos morfológicos y sintácticos. Temuco: Ediciones Universidad de La Frontera. Hernández Sallés, Arturo, Nelly Ramos Pizarro, and Carlos Cárcamo Luna. 1997. Diccionario ilustrado Mapudungun Español English. Santiago de Chile: Pehuén Editores. Key, Marie Ritchie. 1978a. Lingüística Comparativa Araucana. In Vicus Cuadernos: Lingüística II, 45–56. Amsterdam: John Benjamins. . 1978b. Araucanian Genetic Relationships. International Journal of American Linguistics 44/4: 280–293. Koul, Omkar N. 2005. Studies in Kasmiri linguistics. Delhi: Indian Institute of Language Studies. Loos, Eugene E. 1973. Algunas implicaciones de la reconstrucción de un fragmento de la gramática Proto-Pano. In Estudios Panos II, ed. Mary Ruth Wise, 263–282. Serie Lingüística Peruana. Yarinacocha, Perú: Instituto Lingüístico de Verano. Malvestitti, Marisa. 2003. La variedad mapuche de la Línea Sur. Aspectos lingüísticos y dialectológicos. Santa Rosa: Instituto de Análisis Semiótico del Discurso. CD edition. Rubino, Carl. 2005. Reduplication: forms, function and distribution. In Studies in reduplication, ed. Bernhard Hurch with Veronika Mattes, 11–29. Berlin/New York: Mouton de Gruyter.

reduplication in mapuzungun: form and function

37

Salas, Adalberto. 2006. El mapuche o araucano. 2nd edition. Santiago: Centro de Estudios Públicos. Sandvig, Timothy. 1987. La reduplicación en Mapudungun. Actas de Lengua y Literatura Mapuche 1: 143–156. Temuco: Universidad de La Frontera. Smeets, Ineke. 2008. A grammar of Mapuche. Berlin/New York: Mouton de Gruyter. Stark, Louise. 1970. Mayan affinities with Araucanian. Papers from the VI Regional Meeting of the Chicago Linguistic Society 57–69. Vúletin, Alberto. 1987. Neuquén: toponomástico, fauna, flora, riquezas naturales, guía turística. Neuquén: Siringa Libros. Zúñiga, Fernando. 2001. Dos progresivos y dos resultativos en mapudungun, LIAMES 1: 63–75. . 2006. Mapudungun: el habla mapuche. Santiago de Chile: Centro de Estudios Públicos.

Abbreviations 1 3 acaus advzr agt appl art attr ben caus disc distr dur fut hab intens iter itr. inv loc

first person third person anticausative adverbializer agent applicative article attributivizer benefactive causative discursive particle distributive durative future tense habtitual action intensifier iterative intransitive verb inverse verbal locative

neg nfin nmlz nsg pat peramb pl poss ppos prog rdp real rep res sg tel tr. val vblz

negation nonfinite verb form nominalizer nonsingular patient perambulative plural possessive postposition progressive reduplication realis mode repetitive resultative singular telicizer transitive verb validator verbalizer

chapter 3

Reduplication in Andean Languages Katja Hannß and Pieter Muysken

The goal of the paper is to discuss typological similarities and differences in an areally defined sample of languages, the Andean languages, with respect to reduplication. After giving a definition of reduplication and presenting and justifying our sample, we first compare their formal features, then compare their functional features, and finally discuss degrees of similarity and differences using an areal and typological approach. In our final section some NeighborNet examples are presented in support of this.

1

Introduction

For the following description of reduplication strategies in Andean languages, we apply the definition of reduplication as proposed by the Graz Reduplication Project at http://reduplication.uni-graz.at/:1 a reduplicative construction is a set of at least two linguistic forms F and F’ in a paradigmatic, i.e. non-suppletive morphological relation in which F’ contains a segment or a sequence of segments, which is derived from a non-recursive repetition of (a part of) F. Reduplication exists if a specific grammatical form makes systematic use of reduplicative constructions. As the overall goal of the paper is to discuss typological similarities and differences in a specific sample of languages with respect to reduplication, we will begin our article with an overview of formal reduplication features in the languages under consideration. In section 2.1, we will discuss full reduplication of lexical roots, thereby also addressing the behavior of grammatical markers in reduplication, before turning to instances of partial reduplication in section 2.2. In the following section 3, we will present functional features of reduplication, starting with a discussion of iconic features of reduplication (see section

1 This particular definition is used in all contributions to this volume and, in accordance with that, we, too, adopt it for the present paper.

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_004

40

hannß and muysken

3.1), before proceeding to some less iconic functions in section 3.2. We will conclude our paper with a comparison of both formal and functional reduplication features and we will discuss language clusters that arise from this typological approach to reduplication (see section 4). Apart from proposing language clusters based on typological similarities and differences with regard to reduplication features, one of the more specific outcomes of the article is the suggestion that reduplication patterns may be linked to the typological organization of a language (see section 2.2). Another outcome is the claim that neither Aymara nor closely related Jaqaru make productive use of partial reduplication, as is proposed in the World Atlas of Language Structures online (WALS). Instead, it is suggested that both languages have only full reduplication (see section 2.2). Finally, we argue that reduplication in the Andean languages is not the default mechanism to convey a certain meaning, which, instead, is affixation, but that reduplication often bears a particular connotation and is used to express e.g. a special type of plural. The linguistic data for the present paper is mainly taken from the literature. Among the main works are Adelaar (1977) on Tarma Quechua and Hoggarth (2004) on the Cuzco variety, Hardman’s work (1966; 1974; 1983; 1986) on Aymara and Jaqaru as well as the research of Adelaar and van de Kerke on Pukina as published in 2009 and via the webpage of Leiden University (see http:// www.unileiden.net/ore/). For Cholón, we relied mainly on the dissertation of Alexander-Bakkerus (2005), while for Leko we referred to van de Kerke (2009). Information on Shuar was provided by Beuchat and Rivet (1909) and Gnerre (1999). Finally, the data for Kallawaya and Uru were taken from the authors’ own research (see Hannß 2008, on Uru; Hannß and Muysken, in prep., on Kallawaya). General works on Andean and Amerindian languages include Adelaar and Muysken (2004) and Lenguas de Bolivia, edited by Crevels and Muysken (2009).2 We are aware of the fact that a fuller account of the members of the Quechua language family is ideally needed since the family harbors more variation than we can show here. Furthermore, a number of other languages from the area have not been treated here, notably Mochica, Atacameño and Mapuche (on Mapuche, see Zúñiga and Díaz-Fernández, this volume). However, we hope that the survey presented here is already interesting in its own right.

2 Please note that this enumeration is by no means exhaustive and does not cover every single work used for our database. For a complete overview, please see the works in the list of references below.

reduplication in andean languages

41

1.1 The Language Sample Our sample includes the highland languages Quechua, Aymara and its close relative Jaqaru, Pukina, Kallawaya, and Uru. Languages of the eastern slopes and the Andean foothills are represented by Cholón, Leko, and Shuar. The sample is based mainly on typological and geographic criteria. First, the highland languages are (or, in the case of Pukina and Uru, were) spoken in close vicinity to each other or even in one and the same region. One proposal is, thus, to examine these languages for signs of areal features. However, the highland languages, and in particular Quechua, exerted considerable influence on the neighbouring languages spoken at the foothills of the Andes. Another question which motivated the sample is how much influence, exemplified by reduplication, Quechua had on the languages of the eastern slopes and whether languages of the Andean foothills resemble the highland languages with respect to reduplication or whether distinct features prevail. Therefore, we also include in our sample Leko and Shuar, two languages which are known for long and intensive contact with the Quechua-speaking region. Finally, according to Torero (2002), Cholón does not form part of the central Andean cluster, and in this paper we will examine whether this claim also holds for reduplication strategies. In the following, we will give a brief introduction to each of the sample languages. The Quechua language family is divided into branches I and II. Varieties of the Quechua I branch are spoken in the central highlands of Peru, while Quechua II varieties are found in the highlands of Ecuador and southern Peru as well as on the neighbouring Bolivian Altiplano. Quechua is agglutinative and suffixing; the nominal and especially the verbal morphology is complex, and all Quechua varieties have a considerable number of verbal suffixes expressing tense, aspect, and mood. Subjects (and in part objects, too) are obligatorily cross-referenced on the verb; the constituent order is SOV, although Quechua allows some variation, and modifiers precede their head (for a comprehensive overview of the Quechua language family, see Adelaar and Muysken 2004, 179 ff.). Aymara and Jaqaru belong to the Aymara language family, which is divided into a northern and a southern branch. The northern branch is represented by Jaqaru and Cauqui (also Kawki), which are on the verge of extinction (see Adelaar and Muysken 2004, 171). The southern branch is represented by Aymara proper, which is spoken mainly in the highland provinces of Bolivia, while Jaqaru is spoken in the community of Tupe, southeast of the Peruvian capital, Lima (see Cerrón-Palomino and Carvajal Carvajal 2009, 169, Hardman 1983, 23). Jaqaru is, therefore, isolated from the main part of the Aymara-speaking region and, due to its prolonged contact with surrounding Quechua, shows many

42

hannß and muysken

influences from Quechua, particularly that of Cuzqueño Quechua (IIC) (see Hardman 1983, 28). In contrast to Quechua, Aymara has less internal dialectal variation, although further research on this topic is required. Aymara is an agglutinative language which uses only suffixes; the system of verbal directional markers, in particular, is complex. Word order is, likewise, modifier-head, and the constituent order is mainly SOV. In contrast to Quechua, Aymara has an elaborate system of morphophonemic rules (see Adelaar and Muysken 2004, 274f.). The phonological systems of Cuzqueño Quechua and Aymara resemble each other considerably, the latter has had some influence on Southern Quechua, including Cuzqueño Quechua. Both are marked by a series of aspirated and glottalized stops, with the glottal stop having phonemic value. Furthermore, both have a three-vowel system, where /i/, /a/, and /u/ have phonemic value, while [e] and [o] are non-phonemic. The syllable structure of Cuzqueño Quechua and Aymara is marked by an alternating pattern of consonants and vowels, with a heavy, closed syllable of the CVC-type being the largest syllable. Within a syllable, no sequence of vowels or consonants occurs. Both have penultimate stress. However, while morphophonemics do not play a role in Quechua, Aymara uses final vowel suppression as a means to mark syntactic relationships, most notably a direct object (see also section 2.2) (for an overview of the phonological systems of Cuzqueño Quechua and Aymara, see Cerrón-Palomino and Carvajal Carvajal 2009, 174ff., on Aymara, and Plaza Martínez 2009, 221ff., on Quechua). With respect to the latter, Aymara resembles its sister language Jaqaru, which also has vowel suppression, although in a more complex and less clear way than Aymara (Adelaar and Muysken 2004, 302f.). Apart from that, Jaqaru also has aspirated and glottalised stops and three phonemic vowels, /i/, /a/, and /u/. The latter are distinguished for length, a feature that is non-phonemic in Aymara. Finally, Jaqaru has an additional alveo-palatal stop [ty], which can be glottalized and aspirated and which is not present in Aymara (for an overview of Jaqaru phonology, see Adelaar and Muysken 2004, 301 ff.). The phonological system of Tarma Quechua differs in some points from that of Cuzqueño Quechua, as they belong to different branches of the Quechua language family. Tarma Quechua has no aspiration or glottalization of stops (which is a feature found solely in the Cuzqueño variety), but instead shows voiced stops. Another difference is that Tarma Quechua has five vowels, /i/, /e/, /a/, /u/, and /o/, with vowel length being phonemic. However, syllable structure and stress falling onto the penultimate syllable are identical to Cuzqueño Quechua (for an overview of Tarma Quechua phonology, see Adelaar 1977, 31ff.).

reduplication in andean languages

43

The Pukina language was widespread in the Andean area in pre-colonial and early colonial times but became extinct in the 19th century (see Adelaar and van de Kerke 2009, 125). The only Pukina data of any substance is a sample of Christian prayers and sermons (de Oré 1607). It therefore does not come as a surprise that statements about the Pukina phonological system remain somewhat preliminary. Thus, the number of vowels is under debate: while Torero (2002, 413) suggests five vowels, /i/, /e/, /a/, /u/, and /o/, the phonemic status of /u/ and /o/ is not entirely certain, according to Adelaar and van de Kerke (2009, 128). However, vowel length appears to have been distinctive in Pukina. There is no evidence for glottalized or aspirated stops, as they are found in Aymara and Cuzqueño Quechua, but Pukina possesses a word-internal glottal stop (Adelaar and van de Kerke 2009, 130). Given that one has to rely on written sources only, whose spelling, in addition, is not overly consistent, statements concerning morphophonemic processes are precarious. However, Adelaar and van de Kerke (2009, 129) suggest that Pukina has regularly voicing or “softening” of occlusives, resulting in e.g. /G/ and also displays processes of sandhi. The rules and in particular the function of such morphophonemic processes are not entirely clear. In contrast to Aymara and Quechua, Pukina allows word-initial consonant clusters, often beginning with /s/. There is evidence that in earlier stages Kallawaya (see below) had similar consonant clusters, which were reduced to simple stops by quechuization (see below). Details of Pukina syllable structures and stress patterns have to remain elusive. Adelaar and van de Kerke (2009, 126) suggest a genetic link with the Arawakan languages of the Bolivian lowlands but also point out that Pukina, if it ever had an origin in the lowlands, must have long since adjusted to the Andean language model. One feature, however, reminiscent of Arawakan languages is possessive pronominal markers, which precede their head noun but can be separated from it by an adjective (see Adelaar and Muysken 2004, 352). Although the Pukina data, as reflected in the material of Oré (1607), contains some suffixes from Aymara and Quechua, the majority of the grammatical markers appear to be genuinely Pukina. Kallawaya is a mixed, secret language of northwestern Bolivia (Charazani region), spoken only by professional herbalists. The main body of the Kallawaya lexicon is assumed to come from Pukina (see Stark 1972, 199), while other lexifier languages of Kallawaya are Quechua and Aymara.3 The gram-

3 According to our present state of knowledge, about 60% of the Kallawaya lexicon is provided by Pukina, while Quechua and Aymara each contribute about 8%. The remainder of the

44

hannß and muysken

mar of Kallawaya is largely provided by Ayacuchano or Cuzqueño Quechua (both classified as IIC). The genetic affiliation of Kallawaya is still uncertain (compare e.g. Matras 2009, 288, on the genetic affiliation of mixed languages in general). The phonological system of Kallawaya is closer to Quechua than to Pukina, since Kallawaya has series of glottalized and aspirated stops, just as Cuzqueño Quechua (see above). However, Kallawaya shares with Pukina a fivevowel system, where vowel length is phonemic (see e.g. Muysken 1994, 202). Furthermore, the older sources on Kallawaya suggest that the language also had word-initial consonant clusters (see Pukina above), which, however, got lost due to increased quechuization of Kallawaya. By quechuization of Kallawaya we refer to the influence Quechua had on this language (but also on others, such as e.g. Leko, see below), particularly with respect to the phonological system. Quechuization thus mainly entails the reduction of word-initial consonant clusters, as these do not exist in Quechua as well as a reduction of the vowel system to three phonemic vowels, /i/, /a/, and /u/, whereas [e] and/or [o] loose the phonemic value they might have had. Furthermore, quechuization also describes a tendency to have syllables of the CV-type, i.e. Quechua influenced the syllable structure of Kallawaya to the effect that it resembles Quechua more than it does Pukina. Finally, the stress in Kallawaya is on the penultimate syllable, thus resembling Quechua (for an overview of the Kallawaya phonological system, see e.g. Muysken 2009, 153 f.). The lexicon of Kallawaya is considerably reduced (see also section 3.2). Uru, also referred to as Uchumataqu (see Hannß 2008), together with closely related Chipaya forms the isolated Uru-Chipaya language family. Uru had its homeland around Lake Titicaca but became extinct around 1950. Between 1894 and about 1952, ethnographic and linguistic research was conducted on Uru, and it is from this time that the data on which our observations are based comes. Uru is an agglutinative, predominantly suffixing language, but, at the time of documentation, it had one productive prefix. Other, possibly productive prefixes were already lexicalized when Uru was documented. In its nominal system Uru resembles Quechua and Aymara, but the Uru verbal system appears to be less elaborate, and it is, in particular, the verbal inflection system that differs notably from that of Quechua and Aymara. An important means of person reference in Uru are pronominal clitics, which set the language clearly apart

Kallawaya lexicon consists possibly of lowland languages, such as Tacana (see Muysken 1997, 427) as well as coined expressions. An investigation of the etymological composition of the Kallawaya lexicon has recently been conducted by Hannß (see Hannß and Muysken, in prep.).

reduplication in andean languages

45

from Quechua and Aymara. Otherwise, Uru parallels the dominant Andean languages in that its constituent order is SOV and modifiers usually precede their head. A similar influence of Aymara (and maybe also Quechua) is apparent in the phonological system of Uru. Like Aymara, Uru has aspirated and glottalized stops but in addition shows phonemic labialized /tw/ and /kw/ as well as /tk/ (Hannß 2008, 124). Another difference is that Uru has a five-vowel system where vowel length is phonemic, herein being closer to Pukina and Kallawaya than to Aymara. Uru allows consonant clustering within a syllable and also has a tendency towards word-initial consonant clusters, thus resulting in syllables of the CVCC and CCVC-type. Uru displays a number of morphophonemic processes, such as elision and metathesis, but these are “context-sensitive” (Clark and Yallop 1995, 85) and do not have morpho-syntactic functions. Vowel alternation in Uru is quite frequent, often affecting /i/ and /u/ (see section 2.1; see also Adelaar and Muysken 2004, 198, on Quechua). Stress falls mainly on the penultimate syllable (for an overview of Uru phonology, see Hannß 2008, 57 ff.). Cholón is a now extinct isolate (or, if you count the closely related Hibito, also extinct, a small family) formerly spoken in the foothills of northern Peru, with its center being around the Huallaga River (see Alexander-Bakkerus 2005, 33). There are many loans, including morphological elements, from Quechua, a language with which there clearly have been intensive contacts. The main source for the language is Pedro de la Mata’s Latinate grammar manuscript (1748), edited and analysed by Alexander-Bakkerus (2005). A large part of Cholón morphology is suffixal and agglutinative, but cross reference for person is prefixal. Cholón tends to be head-final. In describing Cholón phonology one encounters the same problems as with Pukina, i.e. there are (mostly) only written texts and spelling is not entirely coherent. Therefore, a five-vowel system of Cholón can be only tentatively proposed (Alexander-Bakkerus 2005, 56, 72), the same applies for vowel length, which may be phonemic (Alexander-Bakkerus 2005, 72). As a consequence of this insecure database, Alexander-Bakkerus (2005, 111) suggests a “minimum inventory of native consonant phonemes,” which equals by and large those of the present-day dominant Andean languages Quechua and Aymara. Vowel suppression is also frequent in Cholón, although it does not seem to mark morpho-syntactic relationships as in Aymara and thus appears to be rather “context-sensitive” in the sense of Uru (see above). However, with regard to other features, Cholón differs from the Andean languages, notably by the existence of vowel harmony and relational nouns. Furthermore, Cholón hardly has any consonant clusters, which is typical of at least Pukina, Kallawaya (in an

46

hannß and muysken

earlier stage), and Uru. The latter leads to a Cholón CVC-syllable structure; stress falls on the ultimate syllable (for an overview of Cholón phonology, see Alexander-Bakkerus 2005, 51ff.). Shuar is part of the Jivaroan language family and has about 25,000–30,000 speakers; the family is spoken in the Andean foothills and adjacent Amazonian region in the border area between Peru and Ecuador. The Shuar people have a long tradition of contact with the highlands, and the language has a number of loans from Quechua. Although ethnographically well described, there are hardly any linguistic descriptions of the Shuar language (for an overview, see Adelaar and Muysken 2004, 432ff.). Shuar has a fairly simple consonant system, with three stops (/p/, /t/, /k/), and three fricatives (/s/, /š/ and glottal /h/) (Turner 1958; Adelaar and Muysken 2004, 434). There are two affricates, /č/ and /c/, and three nasals, including a velar one. There is a vibrant /r/ and the two glides /w/ and /y/. There are four vowels, /i/, /u/, a mid-central /ı/, and /a/. Some sources suggest a vowel length contrast. Leko is a moribund isolate spoken in the Andean foothills of northern Bolivia (in Apolo and the region of the Mapiri and Coroico Rivers; see van de Kerke 2009, 288). It has been extensively influenced by Quechua. Like Cholón, Leko morphology is suffixal and agglutinative, but cross reference for person is prefixal. Leko tends to be head-final. This Quechua influence is possibly also reflected in the phonological system of Leko. Like Cuzqueño Quechua, Leko has series of aspirated and glottalized stops, the syllable structure is of the CVC-type, stress falls on the penultimate syllable, and morphophonemic processes are rare. These features are shared with Cuzqueño Quechua, but note that van de Kerke does not explicitly relate these Leko features to Quechua. However, in other aspects Leko differs from the highland languages. Apart from the glottalized and aspirated stops, it also has voiced bilabial and dental stops /b/ and /d/, and, furthermore, van de Kerke (2009, 290) suggests that Leko formerly had a labialized velar sound /gw/. The vowel inventory of Leko is also more elaborate, having six phonemic vowels. It is possible that Leko phonology mirrors its geographical position between the highlands and the Andes’ eastern slopes and neighbouring lowlands (for an overview, see van de Kerke 2009).

2

Formal Properties of Reduplication

In the following section, we will provide an outline of the formal properties of reduplication in the Andean languages. We will first delineate full redupli-

47

reduplication in andean languages

cation of lexical roots (section 2.1) before turning to instances of partial reduplication in section 2.2, where, among others, the position of the reduplicant with respect to the morphological type of the given language will be discussed briefly. 2.1 Full Reduplication of Lexical Roots All languages discussed here have full reduplication of nouns, adjectives and verbs. However, it must be mentioned that almost all languages also have other ways of expressing those functions that are conveyed by full reduplication. Mostly, these other ways are the default mechanism. That is, the meaning of example (3) from Aymara is usually expressed by the nominal suffix -rara, as e.g. in wawa-rara ‘with various children’ (Huayhua Pari 2001, 240), where wawa means ‘child,’ while -rara indicates the increase in quantity and also carries a distributional connotation. In other cases, adverbs or pronouns express what in our examples is achieved by reduplication. In Cholón, the indefinite pronoun mek ‘all’ can precede incha ‘thing,’ as in mek incha ‘all the things’ (Alexander-Bakkerus 2005, 173; see also section 3.1), thus yielding a plural reading (see also example (1) below). Therefore, when dealing with reduplication in Andean languages, one has to bear in mind that reduplication is but one of several ways to express a certain meaning and often, reduplication has a certain connotation. That reduplication, both full and partial, is often not the first and only way to express something is furthermore reflected in the fact that all languages discussed in the present paper show reduplication, although to different degrees. Thus, in Uru, adjectives tend to show mainly partial reduplication (see section 2.2), while in Aymara and closely related Jaqaru verbal reduplication is extremely rare. In Kallawaya, reduplication is primarily a phenomenon of the lexicon but hardly occurs in phrases. Except for lexicalized expressions (see section 3.1), reduplicated elements occur only once. Below are two examples of full reduplication of nouns (example 1) and verbs (example 2); example (2) is one of the few examples of verbal reduplication in Aymara. (1)

incha~incha thing~thing ‘things’

(2) qurumi~qurumi roll~roll ‘while/in rolling’

Cholón (Alexander-Bakkerus 2005, 191) Aymara (Porterie-Gutiérrez 1988, 305)

48

hannß and muysken

Example (3) presents yet another instance of reduplication from Aymara, in this case full reduplication of the noun aruma ‘night,’ where reduplication yields a reading of ‘several nights’ or ‘from night to night.’ (3) aruma~aruma night~night ‘several nights; from night to night’

Aymara (Huayhua Pari 2001, 287)

Example (4) demonstrates full reduplication of an adjectival root from Cuzqueño Quechua. In Cuzqueño Quechua (but also in other varieties), adjectives can function as nouns, as indicated in example (4) by the use of the nominal direct object marker -ta. Note that the notion of ‘very’ is usually expressed in Quechua by an adverb, such as e.g. ancha ‘very.’ Reduplication often carries an emphatic connotation (see Hoggarth 2004, 88). (4) allin~allin-ta good~good-do ‘very good’

Quechua (Hoggarth 2004, 162)

In Quechua grammatical marking can play a role in reduplication. According to Hoggarth (2004, 95, 99), in Cuzqueño Quechua the ablative -manta and the locative marker -pi can be used in plural expressions formed by reduplication. The resulting forms containing one of these markers then have a meaning of ‘from among X’ and ‘in among X,’ respectively, as in sach'a~sach'api ‘(in) among the trees’ (Hoggarth 2004, 99). Here, the new, compositional sense of the markers -pi and -manta ‘in/from among’ results from reduplication. Although most languages under discussion here have a set of suffixes which regularly appear in full reduplication and whose sequence may be fixed, the languages do not have what Rubino (2005, 18) describes as “automatic reduplication.” Each of the suffixes mentioned above can occur outside reduplications and without their counterparts, i.e. without the other suffix present in reduplications, and none of the suffixes triggers reduplication in itself. That is, it is the concept to be expressed rather than the suffix that causes reduplication. However, grammatical markers do not only affect the meaning of the reduplicated lexical material, they themselves can be subject to reduplication. A particularly interesting example of the reduplication of grammatical material comes from Uru, where the possessive marker -chi is regularly attached to body parts (5) and characteristics (6).

49

reduplication in andean languages

(5) ana kar[a]-chi~chi Uru neg hand-pos~pos ‘one-armed/ one-handed (lit.: not arm-/hand-having)’4 (Polo 1901, 472) (6) tuk[i]-chi~chi old-pos~pos ‘grandfather (lit.: old-/age-having)’ (Polo 1901, 464) Furthermore, reduplicated possessive marker -chi occurs systematically on numerals, where it is used to form numerals higher than ten (compare Vellard 1951, 17ff.). It has been suggested (see Hannß 2008, 148 f.) that the reduplication of possessive -chi marks inalienable possession, although the meaning originally associated with the reduplication of -chi might have been an intensifying one. Should this assumption be correct, then Uru reduplication of -chi is one of those rare cases (with respect to Andean languages) where reduplication appears to be the only way to express a certain concept. Here, we would like to point to a peculiarity of Kallawaya and Uru reduplicated forms, which concerns a vowel change from /u/ to /i/ in the reduplicated element, thus resulting in complex reduplication. In example (7b), the reduplicant jiphu is derived from its base juphu ‘orifice,’ but jiphu itself is not a lexeme of Kallawaya. Please compare (7a) and (7b). (7) a. juphu~juphu orifice~orifice ‘sieve’ b. jiphu~juphu red~orifice ‘rough’

Kallawaya (Oblitas Poblete 1968, 86)

(Oblitas Poblete 1968, 52)

A similar vowel change can be observed for Uru, as demonstrated in examples (8a) and (8b).5

4 The sources provide both ‘hand’ and ‘arm’ as the meaning of kara. 5 Although examples (8a) and (8b) are instances of partial reduplication, they will be dealt with here due to the vowel change in the reduplicant. For a more detailed discussion of partial reduplication in Andean languages, please see section 2.2.

50

hannß and muysken

(8) a. chu~chuñi red~good ‘very good’ b. chi~chuñi red~good ‘very good’

Uru (Vellard 1951, 10)

(Vellard 1967, 33)

Although formally both processes are identical, their functions appear to be different in Kallawaya and Uru, as is suggested by the translation provided for examples (7) and (8). In (7b), the meaning of the complex reduplication involving vowel change is different from the simple reduplication without vowel change (see example 7a), while for (8a) and (8b) the meaning remains the same. For Kallawaya, we propose that the vowel change is related to the lexical manipulations we find in the Kallawaya lexicon, among which vowel change is one of the most common means (see Hannß and Muysken, in prep.). Usually, vowel change in Kallawaya, and other secret languages, is interpreted as a technique to disguise meaning in order to keep the secret language unintelligible to outsiders (see e.g. Matras 2009, 292). Here, in contrast, the vowel change does not appear to be used as a means of disguise but rather serves to create new lexical items with which the reduced Kallawaya lexicon can be expanded (see also section 3.2). Thus, the expression jiphu juphu ‘rough’ has a meaning quite different from that of juphu juphu ‘sieve.’ For Uru, the purpose of the vowel change is less clear, but lexical manipulations as an explanation can probably be discarded. The vowel change in Uru may be related to what is observed for Quechua, where the sequence of e.g. reflexive -ku and directional -mu becomes -ka-mu, instead of *-ku-mu (see Cusihuamán 1976, 211). That is, Quechua avoids the clustering of two morphemes with the same vowel. If this is also the case in Uru, this is at best a tendency but no morphophonemic rule, as we also find instances such as (8a),6 where the same vowel is maintained in both the reduplicant and the base. Complex reduplication involving phoneme change is attested among the world’s languages, as in e.g. Indonesian belat~belit ‘underhanded,’ based on belat ‘screen’ (Rubino 2005, 16, quoting Macdonald and Soenjono 1967, 54). Note, however, that complex reduplication appears to be exceptional among the Andean languages and may be explained for Uru with its status as an isolate, while for Kallawaya, complex reduplication may be attributed to its reduced lexicon.

6 Please see also example (17).

51

reduplication in andean languages

Before concluding the discussion of full reduplication, we would like to point to an interesting case from Tarma Quechua (Quechua IB), where the fixed sequence of nominalizer -y and verbal subordinator -r expresses pretence. Although example (9) is a repetitive construction and as such not within the scope of the definition of reduplication used here, we would like to mention it briefly as a relevant part of Tarma Quechua. Repetitive constructions such as example (9) occur in Tarma Quechua subordination and thus form part of the syntactic processes in this Quechua variety (see Adelaar 1977, 158). (9) tinka-y tinka-r snap-nmlz snap-sub ‘as with a snap of the finger’7

Quechua (Adelaar 1977, 159)

2.2 Partial Reduplication Evidence for partial reduplication in the languages under consideration here is less clear than it is for full lexical reduplication. However, Kallawaya, Uru and Cholón show partial reduplication, whereas for Quechua, Aymara, Jaqaru, Leko and Shuar this feature cannot be attested. Whether Pukina had partial reduplication is doubtful (for a more detailed discussion, see further below). From the data available on Leko (see e.g. van de Kerke 2009) and Shuar (see e.g. Rouby et al. 1983), it appears that these languages make use only of full lexical reduplication. Quechua has partial reduplication only in lexicalized forms, as in rapa-papa-y ‘to flutter’ or in ata-tá-w ‘horrible! disgusting!’8 (Dedenbach et al. 2002, 159, 160), a feature also observed for Aymara (see also section 3.1). As these forms do not fall under the Graz definition of reduplication as applied here, they will not be considered any further (but see section 3.1 for an overview). The cases of Aymara and Jaqaru require a closer look. It is claimed in the World Atlas of Language Structures Online9 that both languages have productive full and partial reduplication. However, the evidence for productive partial reduplication in Aymara and Jaqaru is insufficient. Although Hardman (1986,

7 A free translation of the entire sentence containing this reduplication is provided by Adelaar (ibid.): ‘They are walking extremely fast (as if flipped away with the finger) in such a way that their bellies seem to be shaken out.’ (the italicized part marks the translated reduplication, the authors). 8 Again, the separate elements do not have a meaning of their own; the repetition is iconic here. 9 http://wals.info/feature/27A?s=20&z2=3000&z3=2999&z1=2998&tg_format=map&v1= cd00&v2=cf6f&v3=cfff (20/05/2011). The claim is based on Hardman (1966) and Hardman de Bautista (1974).

52

hannß and muysken

269) provides examples such as (10, Aymara), and (11, Jaqaru), where in both cases the final vowel of the first component is deleted, she herself does not analyse them as partial reduplication but, in the case of example (10), suggests an interpretation as full reduplication (ibid.). (10) mark~marka-tan=wa town~town-4=decl ‘We are many communities.’

Aymara (Hardman 1986, 269 f.)

(11) ut~uta Jaqaru house~house ‘place where there are many houses’ (Hardman 1983, 184) We adopt Hardman’s interpretation of these instances as full reduplication. Final vowel deletion in Aymara and Jaqaru does not operate solely in reduplication but occurs in many other, non-reduplicative morphological processes as well, e.g. in the marking of direct objects (see Hardman 1986, 67, on Aymara) or in non-reduplicative nominal compounding, as in (12, Jaqaru). There, the final vowel of wajra ‘horn’ is likewise deleted, but the construction is clearly nonreduplicative. (12) wajr-qucxa horn-lake ‘lake in the shape of a horn’

Jaqaru (Hardman 1983, 140)

The deletion of a final vowel, therefore, is not bound to reduplication, and we suggest that instances such as (10) and (11) be analysed as full reduplication. If this line of argumentation is taken, however, neither Aymara nor Jaqaru can be said to have productive partial reduplication. Although the evidence for partial reduplication in Kallawaya and Cholón is not extensive, we propose that these languages have partial reduplication. In Kallawaya, the final syllable of the lexical base is copied, resulting in a CV-structure, which follows the base. The function of reduplication in (13) remains uncertain. (13) okha~kha-na sell~red-inf ‘to sell’

Kallawaya (Soria Lens 1951, 33)

53

reduplication in andean languages

A particularly interesting example is provided in (14). The form eke~ke-y ‘crash, collision’ is based on uke- ‘to crash, collide,’ where the reduplication shows vowel change from /u/ to /e/. As such a vowel change is attested for at least one other instance (see example 7b), we may assume a similar process in (14). Here, reduplication may serve intensification. This is noteworthy as Kallawaya tends to follow its grammar-providing language Quechua by expressing intense by means of adverbs. The notion of ‘very expensive,’ for instance, is thus phrased as ancha latapi (Girault 1989, 110), where ancha means ‘very’ and latapi ‘expensive’ (for ancha ‘very,’ see also section 2.1). (14) eke~ke-y collide~red-nmlz ‘crash, collision’

Kallawaya (Girault 1989, 89)

An intriguing case from Cholón which requires further analysis is (15), which may involve partial reduplication, unless we analyse ma as a grammaticalized reduced form of maha. However, if the form ma is viewed as an instance of partial reduplication, then the first syllable is copied and precedes its base.10 This is in contrast to what has been suggested for Kallawaya (for a more detailed discussion, see further below). (15) ma maha-ly ints ints-restr ‘intensive’

Cholón (Alexander-Bakkerus 2005, 189)

Of the languages discussed here, Uru is the one for which partial reduplication is most clearly attested. The reduplication types can be correlated with function and word class: verbs usually reduplicate their entire stem, while nouns show both full and partial reduplication; with adjectives, we have mostly partial reduplication (for a discussion of the functions, see section 3.1). In general, partial reduplication is less frequent than full reduplication. In the former type, the reduplicant consists of the initial CV sequence and precedes the base, as in (16), which is one of the rare instances where partial reduplication occurs with verbal material. 10

That maha-ly is indeed the base, which in example (15) is preceded by the reduplicant is proven by the following clause: maha-ly ayča ulyupaŋ ‘he eats much meat’ (AlexanderBakkerus 2005, 189), where ayča means ‘meat’ and ulyupaŋ is the verbal predicate. Here, the intensifier maha-ly occurs in isolation, demonstrating that it is indeed the lexical base which in our example (15) is preceded by a copy of the first syllable.

54

hannß and muysken

(16) sa~saxk'i=chay red~run=decl ‘to run very hard’

Uru (Vellard 1951, 29)

With nouns, we find both types of reduplication; example (17) presents an instance of partial reduplication. For pronouns, only full reduplication is attested. However, all pronouns in Uru can take nominal markers and are, thus, considered to form a nominal subcategory (see Hannß 2008, 183). We may, therefore, assume that pronouns are also subject to partial reduplication but that these cases are simply not reflected in the Uru database. (17) chu~chula red~child ‘child, little boy’

Uru (Vellard 1950, 60)

As mentioned above, it is uncertain whether partial reduplication exists in Pukina. The severely limited database of Pukina provides only one example that might be interpreted as partial reduplication, but this claim is arguable. The expression hani gogoroguta ‘to heaven’ can be analysed in two possible ways. Please see (18 and 19). (18) hanigo~go-ro-guta high~red-x-all ‘to heaven’

Pukina (TP5)11

(19) hanigo goro-guta haniqu qura-guta high world- all ‘to heaven’

(TP5)

In both cases, we encounter some difficulties. If we assume an interpretation as in (18), the final syllable of hanigo ‘high’ would be reduplicated, thus representing an instance of partial reduplication, possibly with an intensifying effect. However, in this case, the alleged suffix -ro cannot be analysed, at least not on 11

The Pukina examples are taken from: http://www.unileiden.net/ore/texts.aspx. The references are adopted from the project website. The first letter refers to the text itself, the second letter indicates the language of the text (i.e. whether a text is Spanish, Quechua, or Pukina), while the number indicates the line of the respective text. TP5 then reads: text T, written in Pukina, line 5.

reduplication in andean languages

55

the basis of our current knowledge. If an interpretation as in (19) is assumed, the form goro must be analysed as actually meaning qora ‘world,’ in which case, (19) is not an instance of partial reduplication. Although no definitive judgement can (at present) be made, an analysis as suggested in (19) appears more plausible, given that in this case all elements in the expression can be analysed and taking into consideration that words are spelt inconsistently in the sources on Pukina. Should the interpretation as suggested in (19) apply, only instances of full reduplication occur in the available Pukina data. Before proceeding to an outline of the functions of reduplication, we will briefly discuss the position a reduplicated element occupies in reduplications. In Kallawaya,12 reduplicants appear to the right of the lexical base. In Cholón and Uru, on the other hand, reduplicated elements occur before the lexical base, i.e. to their left. Although the database is scanty, we tentatively suggest that the position of the reduplicant can be linked to the morphological type of the relevant language. The observation that reduplicated elements in Kallawaya follow the base corresponds with the entirely suffixing structure of the language, which is similar to that of Quechua, which has provided most of its grammar. Cholón and Uru are also overwhelmingly suffixing in nature but mark person reference by prefixes and proclitics (see Alexander-Bakkerus 2005, 213, on Cholón; Hannß 2008, 226ff., 274f., on Uru13). This may be correlated with the position of reduplicated elements to the left of the lexical base, although it must be noted that only a few of the grammatical markers in Cholón and Uru occur as prefixes and proclitics. With respect to its morphological type, Pukina is rather mixed. In its verbal system, it is almost exclusively suffixing, while in pronominal possessive marking, it applies proclitic elements (see Adelaar and van de Kerke 2009, 130f.; see also section 1.1). Should the assumption that Pukina has partial reduplication (see example 18) turn out to be correct, then Pukina partial reduplication is aligned with the verbal rather than the pronominal system. However, as it is uncertain whether Pukina makes use of partial reduplication at all, this has to remain elusive for the time being. Table 1 summarizes the formal properties of partial reduplication in the Andean languages.

12 13

As well as for Pukina, if we assume that this language also has partial reduplication (see example 18). Although the pronominal clitics of Uru can appear in pro- as well as in enclitic position (see Hannß 2008, 134), the sources suggest that there was an increasing tendency to apply them in enclitic position, which is probably due to the generally suffixing structure of Uru.

56

hannß and muysken

table 1

Formal properties of reduplication in Andean languages

Language

Partial reduplication

Position of the reduplicant Precedes the lexical base

Quechua Aymara Jaqaru Pukina Kallawaya Uru Cholón Shuar Leko

3

– – – ? × × × – –

Follows the lexical base

? × × ×

Functions of Reduplication

In what follows, we will discuss the various functions expressed by reduplication in the Andean languages. We will first sketch the more iconic functions in section 3.1, before we proceed to a discussion of the less iconic cases in section 3.2. We use the term iconicity here to indicate the parallelism or similarity between the form of a word and its meaning. 3.1 Iconic Functions Iconic functions of reduplication are found in all languages discussed here. By iconic functions we understand the expression of plurality and distribution (usually with nouns), intensification (usually with adjectives) and that of an imperfective aspect, such as e.g. future or progressive (usually with verbs). Below are examples of the expression of nominal plurality (example 20, see also section 2.1) and distribution (example 21). (20) mek incha~incha all thing~thing ‘all things’

Cholón (Alexander-Bakkerus 2005, 173)

57

reduplication in andean languages

(21) huata~huata-na year~year-loc ‘in each year’

Pukina (XP5)

In Quechua and Aymara, plurality expressed by reduplication often implies a cluster of the reduplicated items. Thus, for example, Quechua t'ika~t'ika does not only mean ‘flowers,’ but “a patch, a mass of flowers” (see Hoggarth 2004, 79, on Quechua; Huayhua Pari 2001, 249, on Aymara). Furthermore, in Quechua “[r]eduplication of a noun […] is rather like a collective noun and implies concern more with a group as a whole than with its constituent elements” (Hoggarth 2004, 79). Plurality expressed by full reduplication of nouns in Quechua and Aymara thus signals a special type of plural, while plurality, as a default, is conveyed by suffixation in both Quechua and Aymara: Quechua has the form -kuna, while in Aymara it is -naka. Thus, ‘flowers,’ without any particular emphasis or connotation, would be t'ika-kuna in Quechua (our example). Although intensification is by no means restricted to the word class of adjectives but can occur with verbal reduplication as well (compare examples 14 and 16 from Kallawaya and Uru), it most usually is expressed by adjectives. Example (22) presents an instance of adjectival reduplication from Shuar with an intensifying meaning. It must be noted, however, that intensification in Shuar is generally expressed by adverbs and suffixes (see Rouby et al. 1983, 45). (22) puéngar~puéngar-eiti good~good-cop ‘It is very good.’

Shuar (Rouby et al. 1983, 45)

The following is an instance of adjectival, intensifying reduplication from Cholón. (23) ihna-ly ihna-ly quickly-restr quickly-restr ‘overhasty’

Cholón (Alexander-Bakkerus 2005, 294)

Among the most iconic functions of verbal reduplication is the expression of an imperfective aspect, of which the notion of progressive is the most common one and attested for Quechua, Aymara, Jaqaru, Uru, Shuar and Leko. It is remarkable that although Kallawaya is among those languages in which reduplication plays an important role and expresses a broad range of functions, the notion of an imperfective aspect is apparently not expressed by reduplication in Kallawaya. The databases of Pukina and Cholón do not provide us

58

hannß and muysken

with examples of verbal reduplication expressing an imperfective aspect, and it remains uncertain whether these languages make use of this reduplicative function. Examples (24) and (25) are instances of verbal reduplication with a progressive meaning. (24) jal[a]~jal[a]-k-i fall~fall-prs-3 ‘He/she/it falls all the time.’ (25) pita-n pita-n smoke-3 smoke-3 ‘(S)he is smoking constantly.’

Jaqaru (Hardman 1983, 183) Quechua (Dedenbach et al. 2002, 159)

A similar meaning is expressed by Shuar íwijmia íwijmia ‘signalling’ (Gnerre 1999, 102). Two other instances of an imperfective aspect, which are typologically widely attested but not found in the other languages discussed here, are repetition (example 26) and the expression of future action (example 27) in Uru. It must be mentioned, however, that future tense in Uru is usually expressed by suffixation. Example (27) is, therefore, marked and probably has an emphatic connotation, although we cannot know this for sure, as (27) is the only instance of a future tense expressed by reduplication. (26) hoxk~hoxk-achu=chay go~go-incl=decl ‘We go again and again.’

Uru (Vellard 1949, 157)

(27) ik~ik-[a]cha=chay meet~meet-irr=decl ‘I will meet [someone].’

(Polo 1901, 481)

As pointed out by Rubino (2005, 12), it is expected that languages which have full and partial reduplication also make a functional distinction between these two reduplication forms. For Kallawaya, we have only two reliable instances of partial reduplication (see examples 13 and 14) and this hardly supports any firm statements on functional distinctions between full and partial reduplication. We can, however, observe that in Kallawaya partial reduplication occurs only with verbs, expressing intensification, while full reduplication is attested so far only for nouns and adjectives. The functions expressed by full reduplication are various (see Table 2 and also section 3.2) and include intensification. Although

59

reduplication in andean languages

we may, thus, not say that the expression of intensification is restricted to partial reduplication, we may say that partial reduplication can yield only an intensifying reading, particularly as we find a similar situation in Uru. There, partial reduplication of nouns, adjectives, and verbs always expresses intensity (see example 16), while fully reduplicated verbs do not express intensity but an imperfective aspect (see examples 26 and 27). Full reduplication with nouns and pronouns marks plurality, intensity, and distribution (with nouns and pronouns). Most iconic, plurality in Uru can be expressed only by full reduplication, and it is suggested (see Hannß 2008, 146) that partial reduplication is insufficient in conveying the idea of plurality and, therefore, only full reduplication renders a plural reading. The database of Cholón is far too scanty to enable remarks on a possible functional distinction between full and partial reduplication in this language, and although example (15) suggests an intensifying interpretation, it is uncertain whether this notion is due to the alleged partial reduplication in this example. The functional distribution of reduplication in Kallawaya and Uru is shown in Table 2. table 2

Language

Functional distribution of full and partial reduplication in Kallawaya and Uru

Form

Word class

Function

Kallawaya Full reduplication nouns, adjectives – associative quality – creation of new lexical item – distribution – intensification – plural Partial reduplication Uru

verbs

– intensification

Full Reduplication nouns, pronouns – plurality – intensification – distribution verbs – imperfective aspect Partial Reduplication

nouns, verbs, adjectives

– intensification

60

hannß and muysken

Finally, we would like to mention briefly some lexicalized and onomatopoeic expressions created by reduplication. As already outlined previously (see section 2.1), lexicalized and thus ‘recursive’ reduplicated expressions are excluded by the Graz definition applied here. However, these lexicalized reduplicated forms testify that in former phases of some of the languages under discussion, both full and partial reduplication were once productive to an extent that cannot be attested in present stages of the languages.14 Furthermore, as these reduplicated expressions form a great and important part of the lexicon of the Andean languages, we find it vital to mention them. Most denominations of animals and plants are built by reduplication, as e.g. chama chama ‘spider’ (Oblitas Poblete 1968, 51; Kallawaya), tširu tširu ‘mosquito’ (Lehmann 1929d, 10, Uru), or suju suju ‘reed’ (Girault 1984, 62, Kallawaya). Often, astronomical constellations are also referred to by reduplicative expressions: wara wara ‘star’ (DeLucca 1983, 687, Aymara) and kerkeru ‘Pleiades’ (Métraux 1935, 104, Uru). In Shuar, onomatopoeic reduplications also denote natural sounds, as in tápir tápir ‘the noise of a breeze’ or as in túput túput ‘the noise of intense fire and strong wind’ (Gnerre 1999, 101). Furthermore, they can also refer to bodily functions, such as turúu turúu ‘the noise of the intestines when one feels hungry’ (Gnerre 1999, 101). Other lexicalized expressions indicate a certain kind of movement, as in kut kut ‘the movement of taking something’ (Gnerre 1999, 101) and also in maepé maepé ‘jump around, move from one place to another’ (Gnerre 1999, 102). It is particularly in many varieties of Quechua but also in Aymara that interjections are often reduplicative, as e.g. alaláw! ‘the cold!’; akakáw! ‘Ouch!’ (Dedenbach et al. 2002, 211) from Quechua or its Aymara equivalent atataw! ‘Ouch!’15 (DeLucca 1983, 46). It is in these lexicalized expressions that Quechua and Aymara show partial reduplication, a feature otherwise not attested for these languages (see also section 2.2). It is likewise notable that in these cases the repeated element can occur twice, something otherwise not found in the languages under discussion here. However, it must be noted that partial reduplication in most Quechua varieties and also in Aymara is restricted to ideophones and interjections, which are not covered by the definition of reduplication here.

14 15

This applies in particular to Aymara and to most varieties of Quechua. DeLucca does not provide a literal translation, but only remarks: “Interjección que expresa dolor”—‘Interjection that expresses pain’ (DeLucca 1983, 46; translation by the authors).

61

reduplication in andean languages

3.2 Less Iconic Functions Although the more iconic functions of reduplication are found in all or almost all Quechua varieties, some have less iconic functions that are peculiar to a certain variety. Thus, for example, it is claimed by de Granda (2003) that in Argentinean Quechua spoken in Santiago del Estero (classified as IIC) adjectival reduplication has a diminutive effect.16 (28) chiri~chiri cold~cold ‘not very cold’

Quechua (de Granda 2003, 126)

As Kouwenberg and LaCharité (2005, 537ff.) note, a diminutive interpretation in reduplication can develop from a dispersive reading involving more than one instance, as in yala-yala/yelo-yelo ‘yellow-spotted, yellowish’17 (ibid.), where “more [our emphasis] means many occurrences distributed over a single surface. The real-world effect of such scattered distribution of colour is to tone down rather than intensify the colour, to diminish rather than augment it.” (Kouwenberg and LaCharité 2005, 538). According to the authors, this effect is not restricted to color terms, but whether it applies to expressions of temperature, as in example (28), remains somewhat doubtful. de Granda proposes (2003, 126) that the diminishing effect of adjectival reduplication in the Quechua of Santiago del Estero can be traced back to the substratum language Kakán (also Diaguita), which was spoken in northwestern Argentina until the 18th century (see Adelaar and Muysken 2004, 177, 407 ff.). However, de Granda (2003, 127) also notes that today the intensifying function, as found elsewhere, has increased in use. The idea that “many occurrences […] diminish rather than augment” (Kouwenberg and LaCharité 2005, 538) appears to be more suitable for the next example from Aymara. There, adjectival reduplication usually signals intensifi-

16

17

Please note that in principle a diminutive effect may also be iconic. That is, if in our example (28) the omission of linguistic material, such as e.g. the deletion of a syllable of the base, signaled a reduction in quality (here: cold) this would clearly be an instance of iconicity. However, as it is, the increase of linguistic material, i.e. the full reduplication in example (28), actually indicates a reduction in the denoted quality. As such, the use of more linguistic material to denote less on the semantic level is non-iconic. Although diminutive readings may be regarded as iconic, we do not adopt this analysis here, and, as a consequence, view reduction of a certain quality indicated by an inrease of linguistic material as non-iconic. From Jamaican Creole.

62

hannß and muysken

cation, but example (29) is an exception from that because it does not have an intensifying reading but rather a diminutive one, i.e. instead of ‘very different’ the reduplication means ‘a little different.’ It is possible that ‘little different’ here means that X differs only in some (and minor) aspects from what it is compared to, but since we lack a context for the example provided by Hardman, this has to remain speculative. (29) mayja~mayja=raki=wa different~different=agg=decl ‘It’a [sic] a little different.’18

Aymara (Hardman 1986, 270)

For Tarma Quechua (IB), Adelaar (1977, 158ff.) lists a number of non-iconic functions of reduplication: frustrated action (30), resulting condition (31), and the readiness to act (32). Notice that all three patterns in Tarma Quechua involve different marking strategies. (30) churi-n-ta tari~tari-ru-ra Quechua son-3-do find~find-rep-pst ‘He had almost found his son (but lost him again).’ (Adelaar 1977, 160) (31) rakta-y rakta-y-ta=m thick-inf thick-inf-do=decl ‘warmly (dressed)’ (Adelaar and Muysken 2004, 233)19 (32) šarku-q šarku-q-la=m ka-ya-n stand.up-ag stand.up-ag-lim=decl this-att-3 ‘He wants to get up all the time.’ (Adelaar 1977, 162) The Kallawaya example (33) can be analysed as describing an associative quality. (33) moko~moko knot~knot ‘knotty; gnarled’

18

19

Kallawaya (Oblitas Poblete 1968, 108)

Hardman (1986, 270) apparently analyses mayja as being composed of the numeral maya ‘one’ and the quantity marker -ja. However, we follow DeLucca (1983, 304) in interpreting mayja as a lexicalized expression, meaning ‘different.’ Glossing and translation from Adelaar and Muysken (2004, 233).

63

reduplication in andean languages

The expression of associative qualities20 is a common function of reduplication in Kallawaya (see example 33) and Leko. In the latter, verbal reduplication has a variety of meanings: ‘produce the sound of,’ ‘produce the quality of,’ etc. Examples are given in (34a) to (34c). (34) a. hel~hel mirror~mirror ‘reflect,’ ‘shine’

Leko

b. selel~selel balsa~balsa ‘balance’ c. wichka~wichka fish~fish ‘flap wings,’ ‘move tail’

(van de Kerke 2009, 313)

We propose that the notion of associative qualities is an extension of an intensifying or a plural reading, i.e. something that contains (many) knots is ‘knotty’ or ‘gnarled.’ In case the reduplication is based on a verb, the interpretation of an associative quality may evolve only in case the verb base is a non-terminal one, which can come to denote a permanent quality. Finally, the creation of new lexical items by reduplication is another frequent function of reduplication in Kallawaya (see also example 7a), which can be explained by its reduced lexicon and the resulting need to expand it (see also sections 1.1 and 2.1). Example (35) is another instance of the creation of a new word by reduplication. (35) ithi~ithi21 bad~bad ‘trap,’ ‘pitfall’

20

21

Kallawaya (Oblitas Poblete 1968, 101)

As an example for associatives, Rubino quotes from Yokuts (Penutian) “k'ɔhis ‘buttocks’ > k'ɔk'ɔhis ‘one with large buttocks’” (Newman 1944, in: Rubino 2005, 21). Although Rubino does not provide a definition of ‘associative quality,’ it seems justified to infer from this example that the term denotes some characteristic(s) of an entity which easily come(s) to one’s mind when talking about this person or object. Hence, we understand by ‘associative quality’ a feature that is typical and as such readily associated with the referent in question. In Oblitas Poblete spelt as ithi iti.

64

hannß and muysken

Example (36) suggests that we find a similar process in Shuar. (36) tagwa~tagwa feather~feather.crown ‘hat’

Shuar (Beuchat and Rivet 1909, 815)

With respect to the creation of new lexical items, it must be noted that although the semantic content of the lexical base is often still transparent, the item in question is lexicalized. This is the case with e.g. ithi ithi in example (35), which is based on ithi ‘bad, evil,’ but in reduplication means ‘trap’ or ‘pitfall’ (possibly in a figurative sense). Although a trap or pitfall can be understood to be evil, the process underlying the derivation cannot sufficiently be described as a mere intensification or agentivization, because the resulting new word does not simply mean ‘very bad’ or ‘somebody/-thing bad.’ Instances such as ithi ithi ‘trap; pitfall’ are thus best described as a new lexical item created by reduplication. Similar considerations apply to the Shuar example in (36): although the lexical base tagwa ‘feather crown’ still shines through in the reduplication, the resulting expression tagwa tagwa ‘hat’ is semantically abstracted from its lexical base and is not a(n) intensification, distribution, or pluralization, etc. of it. Accordingly, tagwa tagwa is best considered an entirely new lexeme. In some cases, however, the creation of a new lexical item clearly involves another function, e.g. intensification or distribution, as in juphu juphu ‘sieve’ (see also example 8a). The lexical base is juphu ‘orifice’ and juphu juphu ‘sieve’ can be understood to involve some distribution (many holes distributed over a surface), but it is not only distribution because juphu juphu does not mean ‘holey’ or ‘perforated’ but describes an instrument (which, however, is characteristically perforated). Therefore, instances such as juphu juphu can be viewed as new lexical items, in whose creation a sense of e.g. distribution is involved.

4

Discussion and Conclusions

In what follows, we will first present an overview of our findings, before we turn to a discussion and some conclusions. 4.1 Summary Overview Table 3 provides an overview of the reduplication strategies found in the Andean languages discussed here.

reduplication in andean languages table 3

65

Overview of reduplication strategies in the Andean languages

Language

Form

Function

Full reduplication

Partial reduplication

Cuzqueño Quechua22

×

–

Aymara

×

possibly not – plurality and distribution (with nouns) – diminutive effect – intensification (with adjectives) – progressive (with verbs)

Jaqaru

×

possibly not – plurality and distribution (with nouns) – intensification (with adjectives) – progressive (with verbs)

Pukina

×

possibly not – plurality and distribution (with nouns) – possibly intensification (with adjectives)

Kallawaya

×

22 23

×

– plurality and distribution (with nouns) – intensification (with adjectives) – progressive (with verbs)

Full reduplication23 Partial reduplication – distribution – intensification – intensification (with verbs) (with nouns) – creation of new lexical items – ‘associative qualities’ – plurality

Please note that we limit our overview to the Cuzqueño variety of Quechua as we do not have sufficient data on reduplication in most other Quechua varieties. For a more detailed discussion of the functional distinction between full and partial reduplication, please see section 3.1. and Table 2.

66 table 3

hannß and muysken Overview of reduplication strategies in the Andean languages (cont.)

Language

Form

Function

Full reduplication

Partial reduplication

Uru

×

×

Partial reduplication Full reduplication – intensification – imperfective (with nouns, aspect (with verbs) adjectives, verbs) – plurality (with (pro)nouns) – distribution (with (pro)nouns) – intensification (with (pro)nouns)

Cholón

×

×

Full reduplication – plurality (with nouns)

Shuar

×

–

– intensification (with adjectives) – progressive (with verbs) – creation of new lexical items

Leko

×

–

– plurality (with nouns) – intensification (with adjectives) – ‘associative qualities’ (with verbs) – progressive (with verbs)

Partial reduplication – intensification (with adjectives)

4.2 Discussion and Conclusions In our comparative analysis we will make use of the visual tool of NeighborNet representations (see Huson and Bryant 2006). The methods of NeighborNet originally developed from evolutionary biology, where “[e]volutionary relationships are usually represented using phylogenetic trees” (Huson and Bryant 2012, 4). These phylogenetic trees are also referred to as ‘split graphs.’ Although the method comes from evolutionary biology, it is nowadays also used in linguistics as well, to measure and visualize typological distances and similarities between languages. The underlying computation is based on algorithms.

reduplication in andean languages

67

The split graphs resulting from our investigation portray the typological distance and similarity of the different feature specifications that we have given for the languages in our sample. The features chosen for the visualizations are those found for the languages discussed here, i.e. every feature that was attested in at least one language of our sample was included in the feature set and checked for all the languages. Note that for our computation all features were considered equally meaningful. Depending on whether a feature is present (marked as “1”) or not (marked as “0”),24 a matrix is generated that shows the typological relations of the sample languages. Split graphs can be rooted, i.e. assuming a common ancestor for the languages in question, or unrooted. Our visualizations are unrooted, as our aim is to provide an overview of typological similarities regarding reduplication, while not assuming any common source for the reduplication features of our language sample or any other kind of genetic relationship between our sample languages. The Figures 1 to 3 visualize the typological distance of the languages discussed in the preceding sections with respect to reduplicative patterns. Figure 1 shows the results for the formal properties of the languages, while Figure 2 represents the functional characteristics. Figure 3, finally, combines both formal and functional features. The Quechua language family is represented only by the variety of Cuzqueño Quechua, as this is the one referred to most often in this paper and, furthermore, for which sufficient information on reduplication processes is available. The network visualization of the languages under discussion here shows three groups (see Figure 1), which are sorted according to the two most prominent formal features of our survey: the existence or absence of partial reduplication and the position of the reduplicant. Leko, Aymara, Jaqaru, Cuzqueño Quechua and Shuar form one great cluster, as none of these languages makes use of partial reduplication. Pukina appears at a slight distance from this Andean cluster as it is uncertain whether Pukina makes use of partial reduplication and, accordingly, whether the reduplicant follows the base (see discussion in section 2.2).

24

If it is uncertain whether a certain feature exists in a language, this can be indicated by “?” and is considered in the computation.

68

hannß and muysken

figure 1

Formal features of reduplication in the languages under consideration25

Uru, Cholón and Kallawaya are set apart by the existence of partial reduplication. Within this group, Uru and Cholón26 form one small cluster since in both languages the reduplicant precedes the lexical base (see section 2.2). Accordingly, Kallawaya appears in isolation, as it has partial reduplication, thereby differing from the great ‘Andean’ cluster, but, in contrast to Uru and Cholón, the reduplicant in Kallawaya strictly follows the lexical base. This sets Kallawaya apart from Uru and Cholón, and it is because of the use of partial reduplication and the position of the reduplicant that Kallawaya appears in isolation. When it comes to functional features, the resulting visualization is more complex and diverse (see Figure 2). Aymara and Uru occupy the same position within the network since both languages show the same functional features of reduplication. Cuzqueño Quechua and Jaqaru are likewise almost identical and appear in the vicinity of Aymara and Uru but lack the diminutive function of reduplication that has been assumed for these and are thus separated from Aymara and Uru.

25 26

Please note that for the sake of readability, Cuzqueño Quechua is represented only as ‘Quechua’ in the network visualisations. We are fully aware that the material on Uru and Cholón as well as that on Pukina represents different stages of the respective languages. For our purpose here, we pursue a synchronic approach but would like to point out that an investigation of contact-induced similarities in reduplication strategies, which also takes into consideration the diachronic nature of the data, is still needed. Such an investigation, however, is beyond the scope of the present paper.

reduplication in andean languages

figure 2

69

Functional features of reduplication in the languages under consideration

Of the Andean languages, Pukina shows only some of the most iconic functions of reduplication, namely plurality, distribution and possibly intensification, while a progressive aspect with verbal reduplication and other, less iconic features of reduplication cannot be asserted for Pukina. The language, thus, appears separated from the other Andean languages. It must be noted, however, that the position of Pukina is, at least in parts, due to the insufficient database of this language. We find a similar situation for Cholón. The use of reduplication to express plurality and intensification puts the language in the vicinity of Cuzqueño Quechua, Aymara, Jaqaru and Uru, but at the same time, Cholón is separated from these because it is uncertain whether the language expresses distribution by means of reduplication. This lack of knowledge arises from the scanty database on Cholón and this also holds for Leko and Shuar. No other functional feature can be attested for Cholón for lack of data. This places Cholón halfway between the Andean languages (with which Cholón shares the expression of plurality and intensification) and Shuar and Leko (with which it shares the uncertainty about other functional characteristics, without, however, expressing associative qualities or creating new words by reduplication).

70

hannß and muysken

What Leko has in common with Cholón is the lack of certainty whether reduplication in Leko expresses the notion of distribution, and this results in the separation of Leko from the Andean languages. Leko is further separated from these by the expression of associative qualities, something shared with Kallawaya. In contrast to the latter, however, Leko does not make use of reduplication to create new lexical items, and it, therefore, appears at some distance from Kallawaya, too. The intermediate position of Leko in the network visualization is due to this particular combination of functional reduplication features. Finally, Kallawaya and Shuar appear farthest away from any other language and are also detached from each other. In the case of Shuar, this can, in part, be attributed to the restricted database on this language, as it is uncertain whether reduplication in Shuar can express plurality, distribution, a diminutive effect or associative qualities. Particularly the lack of the first three features clearly separates Shuar from the Andean languages. On the other hand, Shuar creates new words by reduplication, a feature it shares with Kallawaya. In contrast to Kallawaya, however, Shuar does not express associative qualities by reduplication and is, therefore, separated from it, too. Kallawaya holds a similar position as Shuar in the “split graph” tree. In contrast to Cuzqueño Quechua, Aymara, Jaqaru and Uru, neither an imperfective aspect nor a diminutive effect are among the functions of reduplication in Kallawaya. At the same time, Kallawaya uses reduplication to create new lexical items, a feature shared with Shuar. However, Kallawaya also expresses associative qualities by reduplication, which is not attested in Shuar, and, therefore, Kallawaya appears (again) in isolation from the other languages on the “split graph” (see also Figure 1). It does not come as a surprise that in a network visualization combining formal and functional characteristics (see Figure 3), Cuzqueño Quechua, Pukina, and Jaqaru form one cluster, as these languages show almost identical features of reduplication. These languages are also the ones with the most iconic reduplication strategies. Aymara does not form part of this tightly knit cluster but still occurs in considerable closeness to it. Its slightly distanced position from the Andean cluster can be explained by the diminutive effect of reduplication, assumed for Aymara but not attested for Cuzqueño Quechua (see also Figure 2). As in Figure 1, Uru and Cholón form a cluster, due to the fact that both languages make use of partial reduplication, in which the reduplicant precedes the lexical base , a feature not found in any of the other languages. It is conspicuous that here formal characteristics apparently dominate over the functional ones. The situation of Leko and Shuar is comparable to that of Uru and Cholón, i.e. both languages appear in relative closeness to each other and, at the same

reduplication in andean languages

figure 3

71

Formal and functional features of the languages under consideration

time, at some distance from Aymara, Cuzqueño Quechua, Pukina, and Jaqaru. Shuar appears at even greater distance as it lacks the expression of a progressive aspect with verbal reduplication, something which is attested in Leko as well as in Aymara, Cuzqueño Quechua, and Jaqaru. However, what Leko and Shuar share with the latter is the lack of partial reduplication, which places them in greater proximity to the ‘Andean cluster’ than to Uru and Cholón, which occur almost diametrically opposed to them. Finally, Kallawaya is the one language that sticks out and appears in complete isolation from any other language discussed here (see also Figures 1 and 2). The isolation of Kallawaya is due to a number of less usual reduplication features (‘less usual’ with respect to Andean languages), such as partial reduplication, the expression of associative qualities and reduplication as a productive means to create new lexical items. Taken together, these formal and functional features set Kallawaya clearly apart from the other languages. Furthermore, the “split graph” illustrates that the mixed language status of Kallawaya (Muysken 1994, 1997) contributes to its typologically separate status among the Andean languages.

72

hannß and muysken

When both formal and functional features are taken into consideration, we may conclude the discussion with the observation that the languages discussed here can be sorted into three typologically-based groups; the languages within these groups are rather loosely connected. The largest and most tightly knit group consists of the highland languages: Cuzqueño Quechua, Pukina, Jaqaru and Aymara.27 Another group in relative closeness to the Andean cluster is formed by Leko and Shuar, both languages of the eastern slopes and the Andean foothills, while the last group consists of Uru and Cholón. This group is also the one farthest away from any other cluster. Kallawaya does not belong to any of these and appears in isolation. Although all languages discussed here make use of reduplication and although some also express less iconic functions, reduplication in the Andean languages has never developed into a fully grammaticalized means to express certain concepts. As has been mentioned repeatedly in the preceding sections, none of the languages discussed here uses reduplication as the only way to express plurality or an imperfective aspect, for example, but all languages have grammatical morphemes which usually express these concepts. If reduplication is employed, it serves a particular function and/or bears a special connotation, such as e.g. the expression of collectivity in the case of Quechua full nominal reduplication (see section 3.1). However, as a default, the languages discussed in the present paper use affixation to convey the notion of e.g. plurality, distribution, or an imperfective aspect. So, the question is: why did reduplication never become fully grammaticalized in the Andean languages? We suggest that the answer is to be found, at least in part, in the rich nominal and verbal morphology existent in all languages considered here. The morphology of these languages makes the expression of aspectual notions, size, and plurality, for example, by reduplication unnecessary because all of these concepts are conveyed by affixation. A final remark concerns Kallawaya, which does not seem to fit into this scheme (see also Figure 1) since the language shows a number of formal and functional features of reduplication not attested in other Andean languages. However, as mentioned before (see sections 1.1 and 2.1), reduplication appears to be first and foremost a phenomenon of the Kallawaya lexicon, where it mainly serves to expand the reduced lexicon of this mixed and secret language. In Kallawaya phrases, we rarely find reduplication, as Quechua grammar operates there and offers various possibilities to express such phenomena as dis-

27

Although Aymara appears at some distance from the other three languages, it can still be seen as forming part of this cluster.

reduplication in andean languages

73

tribution, intensification, or verbal aspect by suffixation, so that reduplication need not be applied in Kallawaya phrases. Thus, when the Kallawaya grammar is considered, the language behaves very much like Quechua. We would like to conclude with the observation that reduplication is not a primary means of conveying a certain concept in the Andean languages, but instead is often reserved for expressing a particular connotation or emphasis. However, the concept of reduplication is clearly present in these languages and must have been quite productive at an earlier stage, as is also evidenced by a number of expressions that are lexicalized nowadays.

Acknowledgments We would like to thank Paul Heggarty, Simon van de Kerke, Fernando Zúñiga, and two anonymous reviewers for their helpful and inspiring comments on this paper. Muysken was funded by the ERC grant Traces of Contact for this paper.

References Adelaar, Willem F.H. 1977. Tarma Quechua—grammar, texts, dictionary. Lisse: The Peter de Ridder Press. Adelaar, Willem F.H., with Pieter C. Muysken. 2004. The Languages of the Andes. Cambridge: Cambridge University Press. Adelaar, Willem F.H., and Simon van de Kerke. 2009. Puquina. In Crevels and Muysken (ed.), 125–146. Alexander-Bakkerus, Astrid. 2005. Eighteenth-Century Cholón. Utrecht: LOT. Beuchat, Henri, and Paul Rivet. 1909. La langue jibaro ou šiwora 1. Anthropos 4: 805–822; 1053–1064. Cerrón-Palomino, Rodolfo, and Juan Carvajal Carvajal. 2009. Aimara. In Crevels and Muysken (ed.), 169–214. Clark, John, and Colin Yallop. 1995. An Introduction to Phonetics and Phonology, 2nd edition. Surrey: Blackwell. Crevels, Mily, and Pieter Muysken, ed. 2009. Lenguas de Bolivia. Tomo I: Ámbito andino. La Paz: Plural editores. Cusihuamán Gutiérrez, Antonio. 1976. Gramática quechua: Cuzco-Collao. Lima: Instituto de Estudios Peruanos. Dedenbach-Salazar Sáenz, Sabine, Utta von Gleich, Roswith Hartmann, Peter Masson, and Clodoaldo Soto Ruiz. 2002. Rimaykullayki: Unterrichtsmaterialien zum Quechua Ayacuchano. Berlin: Dietrich Reimer Verlag.

74

hannß and muysken

Girault, Louis. 1984. Kallawaya, guérisseurs itinérants des Andes. Paris: Éditions de l’ORSTOM. . 1989. Kallawaya: el idioma secreto de los Incas: Diccionario. La Paz: Unicef, Panamerican Health Organization (OPS) and World Health Organization (OMS). Gnerre, Maurizio. 1999. Profílo descrittivo e storico-comparativo di una lingua amazzoica: lo shuar ( jívaro). Naples: Instituto Universitario Orientale. Granda, Germán de. 2003. Sobre un—ilusorio—proceso de transferencia sintáctica por contacto en el español andino. La reiteración adjetival elativa. In Estudios lingüisticos hispanoamericanos, ed. Kerstin Störl and Germán de Granda, 121–130. Frankfurt am Main: Peter Lang. Hannß, Katja. 2008. Uchumataqu. The lost language of the Urus of Bolivia. A grammatical description of the language as documented between 1894 and 1952. [Indigenous Languages of Latin America (ILLA) 7]. Leiden: CNWS Publications. Hannß, Katja, and Pieter Muysken. in prep. Sources of the Kallawaya Lexicon. Nijmegen. Hardman, Martha J. 1966. Jaqaru: Outline of phonological and morphological structure. The Hague: Mouton. . 1983. Jaqaru. Compendio de estructura fonológica y morfológica. Lima: Instituto de Estudios Peruanos/Instituto Indigenista Interamericano. . 1986. Structure of the Aymara Language. [without publisher, without place]. Hardman de Bautista, Martha. 1974. Outline of Aymara, phonological and grammatical structure. Gainesville: University of Florida. Hoggarth, Leslie. 2004. Contributions to Cuzco Quechua grammar. Aachen: Shaker. Huayhua Pari, Felipe. 2001. Gramática descriptiva de la lengua aimara. Lima: Negocios Arco Iris S.R.L. Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Huson, Daniel H., and David Bryant. 2006. Application of phylogenetic networks in evolutionary studies. Molecular Biology and Evolution 23/2: 254–267. . 2012. User manual for Splits Tree4 V4.12.8. www.splitstree.org (accessed November 15, 2013). Kerke, Simon van de. 2009. Leko. In Crevels and Muysken (ed.), 287–333. Kouwenberg, Silvia, and Darlene LaCharité. 2005. Less is more: Evidence from diminutive reduplication in Caribbean Creole languages. In Hurch (ed.), 533–545. Lehmann, Walter. 1929. Vokabular des Uro-Dialektes von Ts’imu bei Puno, Titicaca-See. Ts’imu bei Puno, den 26. Oktober 1929. Manuscript. Berlin: Ibero-Amerikanisches Institut. DeLucca, Manuel. 1983. Diccionario Aymara-Castellano. Castellano-Aymara. La Paz: CALA (Comisión de Alfabetización y Literatura en Aymara). Macdonald, Ross R, and Darjowidjodjo Soenjono. 1967. Indonesian reference grammar. Washington D.C.: Georgetown University Press.

reduplication in andean languages

75

Mata, Pedro de la. 1748. Arte de la Lengua Cholona. Trujillo. Manuscript. Additional 25322. London: British Library. Matras, Yaron. 2009. Language Contact. Cambridge: Cambridge University Press. Métraux, Alfred. 1935. Contribution à l’ethnographie et à la linguistique des indiens Uro d’Ancoaqui (Bolivie). Journal de la Société des Americanistes XXVII: 75–110. Muysken, Pieter. 1994. Callahuaya. In Mixed Languages. 15 Case Studies in Language Intertwining, ed. Peter Bakker and Maarten Mous, 207–211. Amsterdam: IFOTT. . 1997. Callahuaya. In Contact languages. A wider perspective, ed. Sarah G. Thomason, 427–448. Amsterdam/Philadelphia: John Benjamins. . 2009. Kallawaya. In Crevels and Muysken (ed.), 147–167. Newman, Stanley. 1944. The Yokuts language of California. New York: The Viking Fund Publications in Anthropology. Oblitas Poblete, Enrique. 1968. El idioma secreto de los incas (vocabulario castellanocallahuaya). La Paz, Bolivia [no publisher]. Oré, Luis Jeronimo de. 1607. Rituale seu Manuale Peruanum. Naples. Plaza Martínez, Pedro. 2009. Quechua. In Crevels and Muysken (ed.), 215–284. Polo, José Toribio. 1901. Los indios Urus del Perú y Bolivia. Boletín de la Sociedad Geográfica de Lima 10: 445–482. Porterie-Gutiérrez, Liliane. 1988. Étude linguistique de l’Aymara septentrional (PérouBolivie). PhD diss., Université de Paris-Sorbonne, Paris IV. Rouby, Angel, Otto Riedmayer and Luis María Callet. 1983. Shuar Chicham. Gramática Shuar. [Sucúa]: Mundo Shuar. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Soria Lens, Luis. 1951. Pequeño vocabulario callawaya. Boletín de la Sociedad Geográfica de La Paz, 32–35. Stark, Louisa R. 1972. Machaj-Juyay: secret language of the Callahuayas. Papers in Andean Linguistics 1/2: 199–218. Torero, Alfredo. 2002. Idiomas de los Andes. Lingüística e historia. Lima: IFEA and Editorial Horizonte. Turner, Glen D. 1958. Alternative phonemicizing in Jivaro. International Journal of American Linguistics 24/2: 87–94. Vellard, Jean (Jehan) A. 1949. Contribution à l’étude des Indiens Uru ou Kot’suñs. Travaux de l’Institut Français d’Études Andines I: 145–209. . 1950. Contribution à l’étude des Indiens Uru ou Kot’suñs. Travaux de l’Institut Français d’Études Andines II: 51–89. . 1951. Contribution à l’étude des Indiens Uru ou Kot’suñs. Travaux de l’Institut Français d’Études Andines III: 3–39. . 1967. Contribución al estudio de la lengua uru. Cuadernos de Lingüística Indígena 4: 1–37.

76

hannß and muysken

Zúñiga, Fernando, and Antonio E. Díaz-Fernández. this volume. Reduplication in Mapuzungun: Form and function.

Abbreviations ag agg all att cop decl dir do incl inf int ints irr lim loc

agentivizer aggregate allative attentive copula declarative directional direct object inclusive infinitive interrogative intensification irrealis limitative locative

mod neg nmlz pos prs pst red rep restr sep sub 3 4

modifier negation nominalizer possessed present tense past reduplication repetitive restrictive separative subordination third person fourth person (first plural inclusive)

chapter 4

Four Types of Reduplication in the Cha’palaa Language of Ecuador Simeon Floyd

The Cha’palaa language of northwestern Ecuador is characterized by the relatively high functional load it places on reduplication; it applies a variety of distinct types of reduplication across different word classes and construction types. This paper describes four major types of reduplication in Cha’palaa: full and partial repetition of ideophones (not a fully grammaticalized reduplication type), full reduplication of verbal elements adding iterative aspect to predicates, full reduplication of non-verbal elements to form verbal adjuncts, and partial reduplication as a derivational process for creating adjective-like attributive words. Because these types of reduplication apply to a range of different constructions with largely unrelated semantics, it appears most likely that Cha’palaa independently developed reduplication at several different times in its history. Cha’palaa illustrates how a language can come to increase the functional load placed on reduplication and how languages can organize and distinguish multiple interacting reduplication processes within their grammatical systems.

1

The Diversity of Cha’palaa Reduplication Types

The Cha’palaa language of northwestern Ecuador is characterized by the relatively high functional load it places on processes of reduplication, applying a variety of distinct types of reduplication across different word classes and construction types. This paper provides a first descriptive account of reduplication in Cha’palaa, a language which is largely undescribed except for a few limited sources (Lindskook 1964, Vittadello 1988, Tapuyo Pianchiche 2009). More sources exist for Cha’palaa’s sister languages (Moore 1961, 1979; Obando Ordóñez 1992; Dickinson 2000, 2002; Curnow 1997, 2002; Vásquez de Ruíz 1988), but reduplication in the Barbacoan language family has not been addressed directly, and one of its members is reported to have no productive reduplication at all (Awa Pit, in Curnow 1997). Cha’palaa’s historical contact language, Ecuadorian Quechua, features only one major type of full reduplication, used

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_005

78

floyd

for intensification of adjectival or adverbial modifiers (Cole 1982, Floyd 2011). In contrast, Cha’palaa features both full and partial reduplication constructions with a range of different functions. Drawing on examples from an audio/video text corpus of natural speech and elicitation exercises collected by the author in Cha’palaa-speaking communities, this paper identifies four major types of reduplication construction in Cha’palaa, in addition to a number of minor types or subtypes. These types differ in how discrete, coherent, unified and productive they are, but applying this typology as a descriptive tool provides a good way for taking stock of the full diversity of form and meaning found in Cha’palaa reduplication constructions. The four types are ideophone reduplication, reduplication within predicates, reduplication to form predicate adjuncts, and reduplication for attributive word derivation. (i) Ideophones can be fully or partially reduplicated to express the iteration or extension of events and, unlike the reduplicated elements in all other reduplication construction types, they are not limited to a single reduplicant, nor is their syntactic position strongly restricted. Under most definitions of reduplication this would be considered repetition rather than reduplication proper: (1)

tulun tulun tulun ti-we ideo ideo ideo say-decl ‘Tulun tulun tulun said (the thunder).’

(ii) Reduplication within predicates, in some ways comparable to morphological aspect inflection, marks iterative aspect on predicates through full reduplication of specific elements of the complex predicate system (the ‘coverbs’). (2) mera mera de-ke-we listen listen pl-do-decl ‘They listened repeatedly.’ (iii) Reduplication in predicate adjuncts applies full reduplication to nonpredicating elements (nominal forms and other ‘non-verbal’ elements) to form a variety of different adverbial adjunct constructions that modify predicates. One subtype gives path information for motion verbs, for example. (3) tsala tsala ji-nu beach beach go-inf ‘go along the beach’

four types of reduplication in cha’palaa

79

(iv) Reduplication for attributive word derivation applies partial reduplication to words to form a class of adjective-like attributive words that describe qualities like color and texture. These words apply a phonological template [(σ.σ.)σ.σbase-σred] to bases of two or more syllables, reduplicating the final syllable. (4) lushi~shi silver (metal)~red ‘green/blue’ These four types of reduplication can each be divided into various subtypes, and in some cases they overlap with each other to different degrees, but all four have morphosyntactic, semantic and phonological features and constraints that provide clear tests for distinguishing among them. The Pacific coast of Ecuador is not easily placed in any single larger linguistic area like Amazonia or the Andes (Aikhenvald 2007), but by any standard Cha’palaa stands out in the region for its many diverse types of productive reduplication. As noted above, in the adjacent Quechua-speaking area reduplication is less productive, as in the other Barbacoan languages. Cha’palaa’s most closely-related sister language, Tsafiki, has some productive reduplication, but it is mainly limited to ideophones (Dickinson 2002, p.c. 2010). In both Cha’palaa and Tsafiki ideophones as a class overlap with the word classes of some of the roots used in nominal and verbal reduplication constructions, so it is possible that Cha’palaa’s more grammaticalized systems have a diachronic connection to less grammaticallyconstrained types of reduplication like that seen with ideophones. Because Cha’palaa’s distinct reduplication types currently apply to several different word classes and construction types, it appears the language developed grammaticalized reduplication at several different times in its history, gradually increasing the functional load it places on reduplication processes over time. Cha’palaa features a number of morpho-phonemic processes that affect the processes of reduplication discussed in this chapter. Although several points about phonological processes are made in the text where appropriate, it may be useful to provide a small introductory sketch here. Cha’palaa has four vowels (a, e, i, u) and twenty consonant phonemes that are represented here with a practical, Spanish-based orthography. Most graphemes resemble their IPA counterparts, but these are some of the departures: Several digraphemes are used, including ch for [tʃ], ll for [ʎ], sh for [ʃ], ts for [ts], and ty, dy and ñ for palatalized t, d and n. Like in Spanish, r represents [ɾ], y represents [j], and j represents [X]. Before front vowels w has the allophone [v], and t, p, and k all have voiced allomorphs after nasals; m is an independent phoneme,

80

floyd

but n also has an allomorph m after bilabials. A phonemic glottal ' occurs only syllable-finally. All vowels can be long, and syllable shapes are generally CV or CVV, but can also be CVC and CVVC for a small set of syllable-final consonants (n, j, s, sh). Phonological reduction applies in many cases, including the deletion of intervocalic consonants and harmonization of vowels to form long vowels as well as the reduction of some final syllables (in many cases leaving a glottal stop). The agglutinating morphology of the language can form longer phonological words, with a strong tendency toward penultimate stress. In addition, there are a number of idiosyncratic irregularities in the system which cannot be fully addressed here.

2

Reduplication of Ideophones

In terms of grammatical productivity, ideophone reduplication is perhaps the least interesting type of reduplication in Cha’palaa and might not even be considered grammatical reduplication under a narrow definition, as it could be described simply as repetition. However, given the probable diachronic connections of ideophone reduplication to the other more grammaticalized types of reduplication in the language, it is worthwhile to consider them together with the other types. Cha’palaa ideophones in both simplex and reduplicated forms are relatively free from morphosyntactic restrictions; they are morphologically simple and can occur in a variety of syntactic positions, or as independent utterances, as is common for ideophones cross-linguistically (Kita 1997, Voeltz and Kilian-Hatz 2001). Example (5), from the text of a story, shows the partially-reduplicated ideophone dijtya acting as an utterance independently of the adjacent phrase; dijtya refers to the sizzling sound and motion of a burning substance: (5) dijtya~tya~tya~tya~tya dijtya~tya~tya~tya fasi ideo~red~red~red~red ideo~red~red~red~red easy jupe-tyu-wa burn-neg-past ‘Dijtyatyatyatya dijtyatyatyatya, it did not burn easily.’ Unlike all other reduplication types, ideophones can have more than one reduplicant, and multiple reduplicants iconically extend the duration, distribution or iteration of events by increments, as in (5) where the reduplication conveys the image of an array of small objects hopping and sizzling in a fire. Also in contrast to all other forms of reduplication in Cha’palaa, ideophone reduplication

four types of reduplication in cha’palaa

81

can variably be full or partial, with full reduplication generally associated with repeated events and partial reduplication associated with temporal or spatial extension of single events. Example (6) shows a partially-reduplicated ideophone from the text of a story, and (7) shows an elicitation example that was also judged acceptable with full reduplication of the same ideophone: (6) tsaa shilla~lla~lla-a man-paij-mi=ren sem ideo~red~red-foc again-descend-decl=emph ‘Like (a monkey) she went (sliding) down shillallalla (quickly, down the tree).’ (7) chi-sha shilla shilla shilla ji-we tree-loc1 ideo ideo ideo go-decl ‘In the tree (she) goes (sliding) shilla shilla shilla (in multiple movements).’ While ideophones are often associated with predicates, ideophone reduplication differs from the productive reduplication of predicate elements because ideophones are not strongly restricted in terms of the class of verbs they occur with. Ideophones tend to associate with the ‘do’ and ‘say’ classes of verbs, but they are fluid and can occur with verbs of multiple classes, unlike the ‘coverb’ elements in predicate reduplication constructions described in the next section, which are each restricted to a specific class of finite verb. Example (8), from natural speech, shows the ideophone puum, used for talking about loud, brief sounds, reduplicated adjacent to the verb bai ‘to fall,’ which is from the i ‘become’ class (motion verbs and changes-of-state); examples (9) and (10), from elicitation notes, show how the same ideophone can be used with the ke activity class and the ti speech verb class. (8)

puum puum bai-we ideo ideo fall-decl ‘It falls puum puum.’

(9)

puum puum ke-we ideo ideo do-decl ‘It does puum puum.’

(10) puum puum ti-we ideo ideo say-decl ‘It says puum puum.’

82

floyd

All other types of reduplication in Cha’palaa make an abstract semantic distinction with a binary opposition between simplex and reduplicated forms, but ideophone reduplication is simply incremental repetition. However, it is appropriate to include ideophones in this discussion because their reduplication formally resembles the two formats for grammatical reduplication in the language: full reduplication of words and partial reduplication of final syllables. Because roots in Cha’palaa tend to be underspecified for word class membership and many word classes overlap with the ideophone class, it could be that ideophones served as a model for the development of more productive reduplication. It is impossible to determine the exact diachronic relationships among the different synchronically-distinct reduplication types in Cha’palaa, but during the discussion of the morphosyntactic criteria for identifying these types it is worth remembering that they contrast in meaning and productivity with less-grammaticalized types of reduplication in the language, while sometimes resembling them in form.

3

Reduplication within Cha’palaa Complex Predicate Constructions

Reduplication within Cha’palaa complex predicate constructions marks iterative aspect on otherwise unaltered predicates, a process in some ways similar to how in many languages a verb might inflect for an aspectual value with an affix. The morphosyntax of this type of ‘predicate reduplication’ follows a rigid pattern of interaction with the other elements in the predicate: reduplicated elements always occur immediately to the left of the finite verb and agree in class with that verb. Unlike with the ideophones described above, only one reduplicated form can occur in a single predicate, but other elements of the predicate may occur to its left. All of these elements, the finite verb, the reduplicated element, and any additional morphemes, contribute to the compositional semantics of the predicate. 3.1 Complex Predicates and Verb Classifiers In order to understand reduplication within Cha’palaa’s complicated predicate system, a short sketch of the system is appropriate. Predicates in Cha’palaa discourse are frequently complex, meaning that they feature two or more roots. Put generally, these constructions feature a semantically-broad ‘generic’ verb to the right carrying finite morphology and one or more ‘coverb’ elements to the left that add information to the predicate.1 Generic verbs are a closed word 1 These terms have been used in the literature on Australian languages, some of which have

four types of reduplication in cha’palaa

83

class distinguished by their ability to act as finite predicates, while most coverbs are from a large open class of roots that are underspecified for lexical class— they can often occur in both predicates and noun phrases, and sometimes as ideophones as well. In Cha’palaa, then, ‘coverb’ is best thought of as a position within complex predicate constructions rather than a word class in its own right. This term has sometimes been used in the literature to refer to a lexical class, but in this paper it refers to the structural position of non-finite elements of complex predicates, regardless of their lexical class. In addition, a subset of the generic verbs also act as verb classifiers that group all of the verbal elements of the language into one of a set of five verb classes (marked with a set of six classifier verbs). The verb classifiers have special properties such as the ability to anaphorically reference predicates that fall into their class. When verb classifiers occur, they take the position of finite verb, and additional parts of the predicate like generic verbs must occur in non-finite coverb position to the left.2 table 1

ke/ki i ju ti na di

Verb classifiers in Cha’palaa

‘to do’ ‘to become’ ‘to be’ ‘to say’ ‘to be in a position’ ‘to come into a position’

most active verbs (the most common class) changes-of-state, motion verbs abstract existential speech verbs positionals (alternates with inchoative di) positionals (alternates with stative na)

The Cha’palaa predicate system is complicated, but for the purposes of discussing reduplication within the system there are primarily two important points to consider: (i) the reduplicated element is a non-finite ‘coverb’ to the left of the finite verb and (ii) the roots that occur in complex predicates fall into semantically-based verb classes, and when they are reduplicated they

similar complex predicate systems (see e.g. Schultze-Berndt 2000). My use of the terms is partially informed by Dickinson’s (2002) application of them to the related language Tsafiki. 2 For the sake of consistency in the glosses, I have labeled all occurrences of a root with its meaning as in simple predicates (‘do,’ ‘become,’ ‘say,’ etc.), and have not distinguished their usage as verb classifiers, generic verbs and coverbs. These distinctions are recoverable from the root’s position in the predicate construction.

84

floyd

occur with their verb classifier which takes the finite morphology (finite morphology is not applied to the reduplicated forms). Predicate reduplication in Cha’palaa can then be defined as a construction with the following basic structure: [coverb coverb verb-finite.morph]. When used instead of the simplex construction, the predicate reduplication construction marks the addition of iterative aspect, meaning the event is durative and consists of two or more repeated sub-events. 3.2 Reduplication for Iterative Aspect within the Complex Predicate Examples (11) and (12) show two comparable predicates that differ only in that (12) is reduplicated to refer to a plurality of events (not a plurality of actors; Landman 1995, McKay 2006). (11) willi ke-we. wave do-decl ‘(He/she) waved (something).’ (12) willi willi ke-we. hit hit do-decl ‘(He/she) waved (something) repeatedly.’ The constructions in (11) and (12) help to illustrate how coverb elements like willi that cannot be finite on their own depend on generic verbs like ke (‘do,’ here also a verb classifier) to carry the predicate’s finite morphology. The system is flexible, however, and generic verbs that can otherwise be finite can also sometimes occur in coverb position, in which case the appropriate verb classifier occurs as the finite verb. Here the generic verb ka is shown in both finite verb position (13) and coverb position with the classifier ke (14); singular and plural predicate options are also illustrated: (13) alla (de-)ka-we fish (pl-)grab-decl ‘He/she (or they) caught fish.’ (14) alla ka (de-)ke-we fish grab (pl-)do-decl ‘He/she (or they) caught fish.’ The two constructions in (13) and (14) do not have a sharp contrast in meaning with or without the addition of a verb classifier. However, in reduplication con-

four types of reduplication in cha’palaa

85

structions only the second pattern can be followed, as shown in (15). Since the reduplicated item cannot occur as a finite verb, the verb classifier necessarily fills this role: (15) alla ka ka (de-)ke-we fish grab grab (pl-)do-decl ‘He/she (or they) caught fish repeatedly.’ (16) *alla ka ka-we fish grab grab-decl ‘He/she caught fish repeatedly.’ The verb classifiers must agree with the class of the accompanying coverb and using an incorrect classifier results in an ungrammatical construction. The ungrammatical example in (17) uses the classifier i, which classifies changes-ofstate and motion verbs, instead of the appropriate classifier ke, which classifies most dynamic activity verbs. (17) *alla ka ka i-we fish grab grab become-decl ‘(He/she) grabbed the fish repeatedly.’ Motion verbs do fall into the i class, on the other hand, and when motion verbs are reduplicated the finite verb in reduplication constructions must agree with their class, as in example (18) from natural speech: (18) fiba-la en-ku ne ne white-col dm.prx-loc2 go.around go.around de-i-shu-juntsa-la pl-become-r.cl-dm.dst-col ‘The whites that come around here repeatedly’ Verb classifier agreement in reduplication constructions can be clearly observed in (19), in which the speaker uses two different predicate reduplication constructions, the first a motion verb with the corresponding i classifier ( ji ji in the first line), and the second a positional verb, now choosing the appropriate di positional classifier (tsu tsu in the third line).

86

floyd

(19) i-bain in ruku-ba kuwan ji ji i-yu kule-n-chi. 1-also 1poss man-com downriver go go become-ego canoe-n-ins ‘I also with my husband went, on repeated occasions, downriver in a canoe,’ pallu juu ji-tu lala mita'=tala peechulla-la-nu ya two be go-sr 1col middle=recp black.person-col-acc house pa'-tu borrow-sr ‘going between the two of us, in the middle, staying with the blacks,’ tsu tsu de-di-yu. lie lie pl-pos.inch-ego ‘we lay down (and slept) repeatedly.’ All potential finite verbs (generic verbs and their subset, verb classifiers) can also occur as coverbs and can then be productively reduplicated, like the generic verb fi ‘to eat,’ in (20): (20) uyala supu-la-nu=bain kai-lla-nu=bain ka-' fi fi foreigner female-col-acc=also child-col-acc=also grab-sr eat eat ke-la do-col ‘The foreigners, they captured and ate the women and the children repeatedly.’ In (20) the generic verb fi is classified by the verb ke ‘do,’ which takes the role of a verb classifier here. In other cases the same roots that can act as classifier verbs in some constructions can also occur as reduplicated coverbs in other constructions. When this happens, the root occurs three times: twice through reduplication and once with finite morphology. Example (21) shows this phenomenon. The Spanish borrowing gueraa ‘war,’ has been verbalized in Cha’palaa with the generic verb ke, which takes the finite morphology. When that complex verb is reduplicated for iterative aspect, however, the generic verb can no longer take finite morphology, so a third root ke occurs as a verb classifier, acting as the main verb. (21) uyala-la-nu de-tu'-ñu-ba matyu gueraa ke ke de-ke-ñu foreigner-col-acc compl-kill-dr-com so war do do pl-do-dr ‘They finished killing the foreigners, as they had repeatedly been making war on them.’

four types of reduplication in cha’palaa

87

Because of this multi-functionality of roots, it is important to distinguish lexical categories (like generic verbs, verb classifiers, and other kinds of roots) from positions in the complex predicate constructions (like finite verb, coverb, etc.) in order to understand how both are relevant to reduplication. 3.3 Reduplicated Forms and Phonological Identity Predicate reduplication in Cha’palaa applies full reduplication to phonological words of different sizes; all predicate reduplication is of full words, and monomorphemic multisyllabic roots cannot be partially reduplicated. This applies both to elements that can be finite verbs such as pipe ‘to bathe,’ in (22) and (23), and to those that cannot, such as willi, in (24): (22) kai-lla pure' pi-sha pu-na-mu, child-col many water-loc1 be.in-pos.stat-ag.n pipe pipe ke-n-de-tsu-we bathe bathe do-n-pl-lie-decl ‘The many children that are in the water are swimming repeatedly.’ (23) *pipe~pe ke-n-de-tsu-we3 bathe~red do-n-pl-lie-decl ‘They are swimming repeatedly.’ (24) *willi~lli ki-we wave~red do-decl ‘(She/he) waves repeatedly.’ Many elements like willi ‘wave,’ that occur only in coverb position but not as finite verbs bear some semantic resemblanceto ideophones in that they are non-inflecting elements that encode events with some degree of iconic sound symbolism. As parts of predicates, however, they cannot undergo partial reduplication like ideophones. The same is true for wajchu ‘blink,’ shown in (25) and (26). (25) kapuka wajchu wajchu ke-mu eye blink blink do-ag.n ‘(He) blinks (his) eyes repeatedly.’ 3 The verb tsu, ‘lie’ has grammaticalized as a progressive aspect marker. While I gloss all occurrences as ‘lie’, when it occurs as the final element of a complex predicate it is usually in an auxiliary role.

88

floyd

(26) *kapuka wajchu~chu ke-mu eye blink~red do-ag.n ‘(He) blinks (his) eyes repeatedly.’ Additional coverbs can be added to predicate reduplication constructions to the left, like wish (‘softly,’ ‘whispering’—also probably an ideophone) in (27). Predicate reduplication is often articulated as a single phonological word, depending on the speed of speech and the particular phonemes that occur at root boundaries. Predicates can begin to resemble complex agglutinative constructions, and the reduplicated material can occur tightly embedded in this morphosyntactic environment. (27) ma-lui-ñu'=mitya wish-kii~kii-ti-n-tsu-ma-a. again-climb-dr=res soft-scream~scream-say-n-lie-ag.n-foc ‘As (the jaguar) just climbed up (the tree) they repeatedly screamed softly.’ Predicate reduplication also applies to morphologically complex words, in contrast with ideophone reduplication, which applies to morphologically simple forms. Example (28) shows two cases of predicate reduplication, the first morphologically simple ( fi) and the second morphologically complex (ma-ja). (28) chachi fi fi ke-mu-aa de-ti-ña achuwa tejku-sha, person eat eat do-ag.mnlz-foc pl-say-decl hair tooth-loc1 ‘It eats people (repeatedly) they say; hair in its teeth,’ achuwa puu ma-ja ma-ja i-mi chachi hair be.in/on again-come again-come become-decl person fi-tu eat-sr ‘there is hair because it came (repeatedly) after having eaten people.’ Predicate reduplication interacts with morpheme boundaries in interesting ways. Multimorphemic words can fully reduplicate, as in (29), from natural speech, or they can be partially reduplicated, as in (30), from elicitation notes. But this ‘partial’ reduplication only applies at morpheme boundaries, since roots are not partially reduplicated. (29) dyabulu-a kayu puder ta-tu ki-n-tyu-ka, diablo-foc more power have-sr do-n-neg-dub ‘If the devil isn’t more powerful and does it,’

four types of reduplication in cha’palaa

89

lala-nu ura nuka=bain ma chi'-ka chi'-ka ki-mi. 1col-acc good where=also again pull-grab pull-grab do-decl ‘he also pulls us again repeatedly in whichever direction.’ (30) chi'-ka ka ki-mi pull-grab grab do-decl ‘He pulls and pulls repeatedly.’ While there may be a slight semantic difference between reduplication of complex words, as in (31), and single roots, as in (32), there also may be a preference for economy at play that leads speakers to opt for single roots over complex forms in specific instances. However, for morphologically simple forms such a preference would be blocked by a restriction on partially reduplicating simple roots, as shown in (33). (31) ma-fale ma-fale ki-we again-exit again-exit do-decl ‘Again he emerges (from the water) repeatedly.’ (32) ma-fale fale ki-we again-emerge emerge do-decl ‘Again he emerges (from the water) repeatedly.’ (33) *(ma-) fale~le ki-we (again-)emerge~red do-decl ‘Again he emerges (from the water) repeatedly.’ In addition, Cha’palaa features many kinds of phonological processes of syllable reduction, so disyllabic roots often have monosyllabic allomorphs that occur in certain contexts.4 For example, the root fale can reduce to faa by deleting the intervocalic consonant and harmonizing the vowels to produce a long vowel. Some approaches to reduplication emphasize morphological and semantic identity over phonological identity (Inkelas and Zoll 2005) but in Cha’palaa predicate reduplication both are required. So while both allomorphs fale and faa can reduplicate, crucially both reduplicants must be the same allomorph.

4 Moore (1962) first described these phonological processes in his comparison of Cha’palaa’s sound system with that of the related language Tsafiki.

90

floyd

(34) a. fale fale ki-we emerge emerge do-decl ‘He emerges repeatedly.’ b. faa faa ki-we emerge emerge do-decl ‘He emerges repeatedly.’ (35) a. *fale faa ki-we emerge emerge do-decl ‘He emerges repeatedly.’ b. *faa fale ki-we emerge emerge do-decl ‘He emerges repeatedly.’ In summary, in predicate reduplication iterative aspect can be marked by reduplication of all or part of the coverbal element, but at least one root must be fully reduplicated, and its reduplication must satisfy requirements of both morphological and phonological identity between the two forms. If one of these conditions and not the other is satisfied, the construction will be rejected by speakers. 3.4 Predicate Reduplication in Discourse Context Predicate reduplication interacts with other aspects of discourse structure, such as anaphoric relationships. In the final line of example (36) the semblative tsan is reduplicated in coverb position as a ke class active predicate. The semblative is able to form part of the complex predicate in this way because it refers back to the ke class predicate in the previous clause, to ‘throw a party.’ The literal meaning of the verbalized semblative in isolation, something like ‘to do like that,’ is filled in by the context—‘to throw a party.’ (36) ufeeda ke-' tsan-ke-' panda fi-' chachi offering do-sr sem-do-sr food eat-sr people wa'-di-' get.together-pos.inch-sr ‘Making offerings, eating food, getting people together,’

four types of reduplication in cha’palaa

91

naa fandangu ke-na-a=bain tsan-ke-' how party do-inf-foc=also sem-do-sr ‘how to throw a party (“fandango”), doing like that,’ tsan tsan-ke-mu-de-e-wa-ña, in abuelu=milla sem sem-do-ag.n-pl-become-past-already 1poss grandfather=dec.ref ‘they were ones who did like that repeatedly, my deceased grandfather.’ Predicate reduplication is frequent in discourse, and in any given minute of transcript it is easy to find several examples. Sometimes complex multi-clause constructions string together many successive iterative events, as in (37): (37) ma mitya mitya ke-n-tsu-mu-aa de-ti, again lean lean do-n-lie-ag.n-foc pl-say ‘It was again leaning (on the tree) repeatedly, they say,’ ishka ishka ke-' mati waashi waashi ke-' sniff sniff do-sr so scratch scratch do-sr ‘sniffing repeatedly, scratching repeatedly,’ mati nepele willi willi ke-' laa laa di-ma-a so tail wave wave do-sr take.out take.out pos.inch-ag.n-foc ‘wagging its tail repeatedly, it made it come out repeatedly,’ de-ti, juntsa aa=kela pl-say dm.dst aug=jaguar ‘they say, that big jaguar.’ Compared to other reduplication types in Cha’palaa, predicate reduplication is the most productive and pervasive. It is highly integrated into the grammar of the complex predicate system and occurs there under the specific morphosyntactic constraints outlined above. The following section will distinguish predicate reduplication from another type, reduplication to form adverbial adjunct phrases.

4

Reduplication to Form Adjunct Phrases

While ‘predicate reduplication’ occurs at the core of the verb phrase and follows a rigid pattern that unifies it as a coherent type, what I will call ‘adjunct

92

floyd

reduplication’ is more of a conglomeration of overlapping subtypes, all of which are distinguished from predicate reduplication because, while they are part of the verb phrase, they are external to the core predicate and have less rigid morphosyntax. Adjunct reduplication constructions commonly relate to the predicate as modifiers at the level of the verb phrase but do not include predicating elements themselves. This general description covers several related reduplication types, and some of the subtypes of adjunct reduplication can also modify noun phrases, but all of them are sharply distinguished from predicate reduplication in several ways, especially due to their more flexible syntax. Cha’palaa has a general preference for verb-final word order, so pre-verbal position is common for adjuncts—with or without intervening material—but post-verbal position is also a possibility. 4.1 The Syntactic Flexibility of Adjunct Reduplication These first examples represent a subclass of adjunct reduplication that reduplicates words describing different kinds of terrain that, when reduplicated, describe routes of motion verbs. They show the possibility for pre-verbal position in (38), from natural speech, as well as post-verbal position in (39), constructed in an elicitation session. (38) kai-lla feka ji-mu-la tsala tsala ji-n-de-tsu-we child-col upriver go-ag.n-col beach beach go-n-pl-lie-decl ‘The children that go upriver are going along the beach.’ (39) kai-lla feka ji-mu-la ji-n-de-tsu-we tsala tsala child-loc upriver go-ag.n-col go-n-pl-lie-decl beach beach ‘The children that go upriver are going along the beach.’ Also in contrast with predicate reduplication, modifier phrases formed through adjunct reduplication can be stacked as multiple modifiers: (40) supu-la pi pi tsala tsala pure-i-n-de-tsu-we woman-col water water beach beach many-become-n-pl-lie-decl fe='mitya upriver=res ‘The women become many (going) upstream through the water and along the beach.’ A related construction has similar properties but includes a limitative clitic (‘only’) with scope across the reduplicated words [adjunct phrase: red red=

four types of reduplication in cha’palaa

93

lim] to form adjunct phrases with a similar meaning. The limitative construction can be considered a subtype of adjunct reduplication referring mainly to paths of motion verbs. (41) kule-nu mika wi'-tu tu tu=tene. canoe-loc many/enough enter-sr earth earth=lim ‘In the canoe many enter, (going) along the ground.’ In discourse, adjuncts formed by reduplication can also modify elided predicates that are available anaphorically, unlike predicate reduplication, which always occurs with an overt finite verb. For instance, the following example could be an answer to the question: “Where/how did they walk?” (42) pi pi=ren water water=emph ‘Along the water (precisely)’ Adjunct reduplication forms only a loose class, both in terms of form and semantics, and the path constructions above group together as a subtype. Taken as a general ‘type’ of reduplication, adjunct reduplication subdivides further into several construction types, like those represented by mashti mashti or ‘machete machete,’ an instrumental construction for tasks requiring repeated machete strikes, and by pallu pallu ‘two two’ for actions done ‘two by two,’ which reduplicates a numeral. All of these subtypes can be thought of in terms of a general construction type ‘adjunct phrase’ comprised of different, more specific subtypes that are all in some sense iterative or distributional (either in time or in space, as with the path constructions). Unlike predicate reduplication, adjunct reduplication can occur with intervening material between it and the finite verb root, such as the semblative in (43) and the discourse particle ne ‘just’ in (44): (43) tsa='mityaa challa juntsa paate jayu jayu tsa-ti-n-tu=ren sem=res now dm.dst part little little sem-say-n-sr=emph ‘For that reason now, talking little by little about that’ (44) baka' baka' ne chu-na-mu de-ju spread.out spread.out just sit/live-pos.stat-ag.n pl-be ‘They were ones who just lived separately (each individually).’

94

floyd

As seen in the different constructions above, there is a general iconic principle of temporal or spatial extension and iteration to adjunct reduplication (‘two by two,’ ‘little by little’ or in the case of motion verbs: ‘along the beachand-more-beach’). In (35) maali ‘alone,’ is reduplicated to mean ‘each alone’ or ‘one at a time.’ Unlike predicate reduplication, but like ideophone reduplication, adjunct reduplication is not restricted to specific verb classes, so maali maali can occur with the i class in (45) but also with the di class in (46). (45) rega-de-i-ñu maali maali ji-de-i-ñu. spill-pl-become-inf.ev alone alone go-pl-become-inf.ev ‘They seem to have spilled out, going one by one, it seems.’ (46) maali maali chu-di-lla alone alone live-pos.inch-col ‘They lived one by one.’ Lexical categories in Cha’palaa are a complex topic, but, put generally, adjunct reduplication applies to nominals and modifiers, with loose syntactic relationships to predicates, in contrast to predicate reduplication’s tight incorporation of primarily verbal elements. 4.2 Predicate Reduplication Compared to Adjunct Reduplication Adjunct reduplication constructions are necessarily associated with a verb (if only elliptically), but they are less morphosyntactically bound to finite verbs than predicate reduplication constructions. Like in predicate reduplication, and in contrast with ideophone reduplication, adjunct reduplication can occur with complex, multi-morphemic words. (47) tu-la-n-paki tu-la-n-paki=tene ji-n-de-tsu. earth-go.up-n-cl:flat earth-go.up-n-cl:flat=lim go-n-pl-lie ‘They are going along the flat earth formations.’ When they occur together, reduplicated elements of predicate reduplication and adjunct reduplication constructions can be distinguished by their ordering. Elements that are part of the predicate occur to the immediate left of the finite verb, while adjuncts can occur further leftward or elsewhere. The opposite order is ungrammatical because the adjunct cannot intervene between elements of the predicate.

four types of reduplication in cha’palaa

95

(48) tsala tsala ji ji i-we beach beach go go become-decl ‘He went repeatedly along the beach.’ (49) *ji ji tsala tsala i-we go go beach beach become-decl ‘He went repeatedly along the beach.’ Adjunct reduplication can modify predicates without immediate leftward adjacency to them; they can be right-dislocated as in (50) or occur with intervening material as in (51). (50) ( feka) ji ji i-we tsala tsala upriver go go become-decl beach beach ‘They went repeatedly (upriver) along the beach.’ (51) tsala tsala feka ji ji i-we beach beach upriver go go become-decl ‘They went repeatedly upriver along the beach.’ Example (52) from a natural speech recording is a good illustration of the differences between adjunct and predicate reduplication. The first line includes an example of predicate reduplication; the second includes an example of adjunct reduplication, both occurring with predicates using the same generic verb ti (‘to say’), yet only the adjunct reduplication can occur with intervening material (the semblative tsan) between the pair of reduplicants and the finite verb. (52) tsa='mitya-a uñi-lla-nu=bain ajaa ajaa ti-ee-ña i-ya sem=res-foc chief-col-acc=also angry angry say-foc-decl 1-foc ‘For that reason I also scold the officials repeatedly,’ tsa'=mitya-a challa juntsa paate jayu jayu tsan-ti-n-tu=ren. sem=res-foc now dm.dst part little little sem-say-n-sr-emph ‘for that reason, now speaking little by little about that.’ Unlike predicate reduplication, adjuncts formed by reduplication have no corresponding simplex forms, since the simplex forms cannot modify predicates in the same way that their reduplicated forms can. Instead, unreduplicated forms show the properties of other word classes such as nouns. In predicates, on the other hand, the simplex forms have the same meaning as the reduplicated form

96

floyd

except for the addition of iterative aspect. Examples (53) and (54) illustrate how the simplex form of an adjuct reduplication construction for ‘go along the beach’ is ungrammatical—it could mean ‘go to the beach’ with the appropriate locative added, as in (55). The simplex form would be nominal, compared to the adjunct created in the reduplicated form, which is basically adverbial. (53) tsala tsala(=tene) ji-we beach beach(=lim) go-decl ‘(He/she) goes along the beach.’ (54) *tsala (=tene) ji-we beach go-decl ‘(He/she) goes along the beach.’ (55) tsala-sha ji-we beach-loc1 go-decl ‘(He/she) went to the beach.’ In summary, predicate and adjunct reduplication in Cha’palaa differ in that in predicate reduplication the reduplicated form contrasts with a simplex form for a semantic distinction. However, in adjunct reduplication the reduplicated forms are not opposed to simplex constructions but rather create a new phrase type with the syntactic properties of a predicate modifier. In general predicate reduplication holds up much better as a single unified type due to its regular and rigid morphosyntax, while adjunct reduplication deserves further study in order to more clearly delineate its numerous subtypes and their individual characteristics.

5

Reduplication as a Phonological Template for Attributive Word Derivation

The final type of reduplication to be discussed here differs from all the other types described so far (except for some cases of ideophone reduplication) in that it involves partial reduplication rather than full reduplication of roots. ‘Attributive reduplication’ is a process used to form words from a special class of semantically attributive words. In the sense that attributive reduplication creates words of a specific class out of words from different classes, it is a ‘derivational’ process. The derived words are in some respects similar to adjectives, as they are used for modifying noun phrases or in attributive predicates. For

four types of reduplication in cha’palaa

97

the purpose of this discussion I will call them ‘R-attributive’ words in order to contrast the word class derived through reduplication with the other kinds of attributive words in Cha’palaa. The reduplication process applies to roots and to complex words of two or more syllables, reduplicating the final syllable rightward: TEMPLATE: attributive word class = [(σ.σ.)σ.σbase~σred] The words to which this template is applied can be morphologically complex, but in many cases the derived forms are unproductive and lexicalized, including elements that are un-analyzable and that have no known corresponding simplex forms. However, enough specific R-attributive words are transparently and productively derived to show that attributive reduplication is still a productive process in Cha’palaa. 5.1 Transparency and Opacity in R-Attributive Word Derivation The following examples show cases in which it is possible to identify simplex root morphemes for the corresponding reduplicated forms and to establish a semantic connection between the two. (56) lushi > lushi~shi silver silver~red ‘money’ (lit. ‘silver metal’) ‘blue/green’ (57) pala > pala~la branch.in.two branch.in.two~red ‘division in two branches’ ‘split apart’ (trees, shoulders) (58) pujpu soft.airtight.container ‘bubble/bladder’

> pujpu~pu soft.airtight.container~red ‘inflated’ or ‘puffed up’

(59) dape piece ‘piece of something’

> dape~pe piece~red ‘cut into pieces’

In examples (56) to (59) all of the reduplicated forms have identifiable simplex forms, such as dape in (59), which in another context can be a nominal root (‘a piece’) or part of a predicate (of the ‘do’ class, meaning ‘to cut into pieces’).

98

floyd

Sometimes the simplex forms are themselves morphologically complex and transparent, or partially transparent, and it is possible to suggest likely etymologies. The word pala is associated with things that divide in two, so it may be possible to identify pa as the root ‘two,’ while it is unclear if la corresponds to one of the different la morphemes that exist in the language (including a verb ‘to go up/out’ and a collective suffix). Similarly, the word pujpu can be analyzed as an ideophone puj that generally refers to pressurized air and a verb pu ‘to exist inside’ (or possibly a reduced form of the nominal classifier for spherical objects, puka). It is used for a number of compound terms like shii=pujpu ‘(human) bladder’ and chii=pujpu ‘bubble (in water/river).’ In addition to these analyzable examples, there are also many cases of words that meet the formal requirements for membership in the R-attributive class— they are semantically attributive and feature final-syllable reduplication—but which do not have corresponding simplex forms or break down into recognizable morphemes. At present etymologies for the reduplicated forms in examples (60) to (63) are unknown, and speakers do not generally accept the simplex forms as possible words (sometimes recognizing them but calling them “incomplete”). (60) (te')wallullu ‘curly’ (*wallu) (61) tiriri ‘wadded-up’ (*tiri) (62) pidyadya ‘shiny’ (*pidya) (63) dulala ‘narrow’ (*dula) There are also cases in which multiple attributive words appear to be based on some of the same morphemes and to have some similarity of meaning, but some of the individual morphemes remain opaque. For example, (64) and (65) are phonologically similar and both refer to distinct kind of softness; nakululu means ‘soft’ as in wet and slimy or squishy, while nakususu means ‘soft’ as in dry and cottony or spongy. Because of their overlapping semantic and phonological features, these terms appear to be related diachronically, but the individual morphemes they include are difficult to identify since they are synchronically frozen forms.

four types of reduplication in cha’palaa

99

(64) nakulu~lu ??~red5 ‘soft’ (wet) (65) nakusu~su ??~red ‘soft’ (dry) A likely scenario is that for some words syllable-final reduplication has become less productive over time for many words, leaving frozen forms with no simplex counterparts. The cases in which productive reduplication is still observable provide a model for how the less productive words were derived. However, whether or not their history is transparent or opaque, all of the R-attributive words pattern together coherently as a semantic class as well as a morphosyntactic class. 5.2 R-Attributive Words as a Semantic Class and Color as a Subclass The words in the Cha’palaa attributive class derived through final-syllable reduplication are similar to adjectives in that they refer to properties that are attributed to nouns, either by modifying them or through attributive predicates. However, they do not include words for many of the most canonical subclasses of adjectives cross-linguistically. Of the four ‘core’ semantic types proposed by Dixon (2004), members of this class in Cha’palaa refer only to ‘color,’ and in some cases ‘dimension’ (big/small), but not to ‘age’ (young/old) or ‘value’ (good/bad). Many attributes not covered by words of the R-attributive class can be described by dedicated forms from other classes, either nouns, prenominal modifiers, predicate elements, or clitics like the augmentative and the dimunitive.6 Additionally, of the more ‘peripheral’ semantic types associated with adjectives in Dixon (Ibid.), Cha’palaa R-attributive words do not include ‘human propensity’ or ‘speed,’ but do include some words referring to a ‘physical property,’ with many other words for physical properties falling into other classes. 5 According to some speakers nakululu is related to a word for ‘placenta’: nakulu. This may be the historical simplex form for nakululu, and the root na, for ‘child’, may be common between both words for ‘soft.’ Speakers do not recognize a simplex form for nakususu (*nakusu). 6 In addition to the general diminutive kaa= and augmentative aa= there are at least two R-attributive words that refer to extreme small size, atyutyu and añuñu, and a word for ‘tall’ that incorporates the augementative: aabarere. There are only a few words of this class that refer to dimension, much fewer than those referring to color and texture.

100

floyd

The common semantic element to all words in the R-attributive class is an orientation to perceptual qualities, particularly of qualities perceived visually or through tactile means, since words for sound, taste and smell fall into other word classes. R-attributive words with syllable-final reduplication primarily include words for color or light quality (red, shiny, dark), texture and consistency (wrinkly, curly, sharp, soft), and form or shape (cut into pieces, wadded up, narrow, divided in two branches). The perceptual experiences highlighted by R-attributive words recall to some extent the vivid ‘depictive’ quality of ideophones (Dingemanse 2011). In addition, their form resembles the partial reduplication of ideophones in Cha’palaa, which also applies to final syllables but is not limited to a single reduplicant. Another difference between R-attributive words and ideophones is that the former can be morphologically complex, while the latter are generally thought to be morphologically simple. On the other hand, expansion of the definition of the term ‘ideophone’ to include multi-morphemic roots in constructions like these, with their expressive sound-symbolic meanings oriented around perceptual experience, might also be warranted. Of the perceptual qualities referred to by attributive words, color has far more terms than any of the others, with several dozen terms and counting. Examining the semantic domain of color is a good way to explore both the general coherency and the specific inconsistencies within the R-attributive class. Of the many color terms, a few of them are more widely and consistently recognized than many of the less frequently-heard terms.7 While some speakers were not aware of the entire inventory, all twelve speakers in a color naming survey consistently applied the terms for black, white and red to the same color chips. Interestingly, these three specific terms form a subclass that is morphologically distinguished from all the other color terms by a suffix, -ba, which is likely to be an old, partially-productive classifier, perhaps one specifically for colors. Like many Cha’palaa roots that are underspecified for word class, the roots for ‘black,’ ‘white’ and ‘red’ cannot occur as independent words, but must be combined with other elements like classifiers. Derivational reduplication

7 Based on the preliminary results of elicitation tasks focused on color terms, speakers showed a high degree of consistency with words for black, white and red, with less consistency for terms for yellow and green/blue, and with extremely high inter-speaker variability for all of the other terms (referring to specific shades of lighter or darker yellow, for example). This hierarchy of consistency resembles the implicational hierarchy for the existence of specific color terms proposed in Berlin and Kay (1969): black/white>red>yellow/blue/green>others. A more extensive study of the Cha'palaa color term system is planned.

four types of reduplication in cha’palaa

101

then applies to the complex word (consisting of the root and the classifier) to create words that belong to the R-attributive class. (66) pa-ba black-cl:color? ‘black’ (pre-nominal modifier)

> pa-ba~ba black-cl:color?~red ‘black’ (R-attributive word)

(67) fi-ba white-cl:color? ‘white’ (pre-nominal modifier)

> fi-ba~ba white-cl:color?~red ‘white’ (R-attributive word)

(68) llu-ba > llu-ba~ba red/ripe-cl:color? red/ripe-cl:color?~red ‘red/ripe’ (pre-nominal modifier) ‘red/ripe’ (R-attributive word) The primary morphosyntactic difference between the reduplicated and simplex forms of these color terms is that the simplex forms modify pre-nominally, like many of the other property-concept words in the language (ura ‘good’; llupu ‘male’; yuj ‘crazy,’ etc.), while the attributive words derived through reduplication modify post-nominally.8 (69) pa-ba ruku black-cl:color? man ‘black man’ (70) ruku pa-ba~ba man black-cl:color? ~red ‘black man’ The colors black, white and red can be expressed with both of these adjectivelike form classes in Cha’palaa, pre-verbal by simplex forms and post-verbal reduplication by forms with final-syllable reduplication. Terms for colors other than black, white and red do not modify pre-nominally, flagging these three colors as distinct. Further complicating matters, there is another term for ‘red’ in the R-attributive class whose component morphemes are opaque. 8 There is a morphosyntactic difference between the types of noun phrases formed by pre- and post-nominal modification, but for the few color words that can occur in both positions it is unclear if there is also a semantic difference, as consultants say that the meanings are the same.

102

floyd

(71) un-kala~la red-??~red ‘red’ (*unkala) The quality of redness that the root llu conveys is associated specifically with the ripeness of fruit, while unkalala appears to be more general. Speakers overwhelmingly opt for unkalala when talking about colors more abstractly but will also accept llubaba as a near-synonym. The majority of other color terms pattern more like unkalala, with opaque or partially opaque morphology and no simplex form. After white, black and red, the next two most consistently identified colors are blue/green and yellow; blue/green (lushishi) has a probable etymology, as discussed in (56), but yellow is more difficult to analyze. (72) lajkili~li ??~red ‘yellow’ (*lajkili) Beyond terms for white, black, red, yellow and green/blue, there is a large set of other color terms, many of which appear to share roots with the core colors, and often refer to lighter and darker shades of them. However, unlike the core terms, speakers often disagree as to whether any specific word refers to a lighter color, a darker color, a more intense color, a more faded color, or a blend of colors. (73) lush-kata~ta blue/green-??~red ‘dark/intense blue’ (74) laj-mele~le yellow-??~red ‘light/greenish yellow’ (75) yamura~ra ??~red ‘grey’ The first syllable of R-attributive words can act as an independent morpheme, as shown in (73) and (74), with which other material combines. The other material in the words (kata and mele) is opaque but has some degree of productivity, since both of the opposite constructions lajkatata and lushmelele are also attested (although with very little agreement among speakers about which col-

four types of reduplication in cha’palaa

103

ors they refer to). There are also cases of color terms like (75), which do not have any recognizable component morphemes but sometimes share elements with other terms, like the term pamurara, which refers to a purplish color. This sample of the internal structure of the subclass of color terms gives a good idea of the heterogeneity of the attributive word class. In this domain the terms vary in their level of productivity, segmentability, and consistency of meaning, and beyond color terms the situation is similar for the other semantic domains included in the class. 5.3 Deriving Complex Attributive Words through Reduplication There is another productive process that applies to R-attributive words that may give some clues to where their unidentifiable morphemes originate. It is possible to apply the first syllable (for most purposes a ‘root’) of color terms and other attributive words to nouns, which are then incorporated into the attributive word through final-syllable reduplication. This kind of noun incorporation is mainly limited to nouns that also act as nominal classifiers for shape-based categories, the most common of which refer to spherical objects (puka ‘fruit’), oblong objects (papa ‘pod’) and elongated objects (lura ‘stalk’). When combined with roots for qualities like color, it is possible to create complex attributive words9 that encode two qualities simultaneously, such as a color and a shape. Example (76) shows an example of this process, and (77) illustrates that the same combination is ungrammatical in simplex form. (76) un-puka~ka red-cl:sphere/fruit~red ‘red spherical object’ (77) *un-puka red-cl:sphere/fruit ‘red spherical object’ This process is quite productive, and any color can combine with any shape; some examples are shown in (78) and (79). Non-color terms can also be used, as shown in (80), which is related to the word ejkeke or ‘dried out’; ejkeke is often used with the more idiomatic meaning of ‘thin’ when applied to people. Like

9 These forms are in some ways similar to the cross-modal compound perception terms discussed in Brown (2011) in that they productively encode multiple qualities and senses within a single complex term.

104

floyd

roots for colors, roots for other qualities can be combined with classifiers to derive descriptive words productively. (78) fi-lura~ra white-cl:elongated~red ‘white-elongated’ (79) lush-papa~(p)a10 blue/green-cl:oblong~red ‘blue/green-oblong’ (80) ej-puka~ka dry-cl:sphere/fruit~red ‘dry-spherical’ The three most common classifiers are essentially nouns that have extended their usage to classification, and a number of other nouns that show classifiertype features can also undergo this productive process. Two of these include chuwa, for objects with long, vine-like shapes, and paki, for flat objects (typically coins, cookies, or flat rock formations). (81) un-chuwa~wa red-cl:string/vine~red ‘red-string/vine’ (82) ej-paki~ki dry-cl:flat~red ‘dry-flat’ While such constructions with classifiers can be productively generated to make terms that combine shape with a second attribute like color, there are limits to incorporating nouns into these constructions. When attempting to generate new terms in elicitation, some forms were judged acceptable while others were not. Consultants found it acceptable to create a term for ‘dry-leafy’ (83), ‘dry/skinny-man’ and ‘dry/skinny-woman.’ However, consultants would not accept a similar construction with ‘dog.’

10

After final syllables reduplicate other phonological processes can apply, such as the intervocalic consonant deletion seen here, which creates a final long vowel: lushpapaa.

four types of reduplication in cha’palaa

105

(83) ej-jaki~ki dry-leaf~red ‘dry-leafy’ (84) ej-ruku~ku dry-man~red ‘skinny-adult male’ (85) ej-shinpu~pu11 dry-woman~red ‘skinny-adult female’ (86) *ej-kucha~cha dry-dog~red ‘dry-canine’ or ‘skinny-canine’ Complex, multi-attribute constructions are certainly productive, but their productivity is limited, with some constructions being highly conventionalized, others being novel but acceptable, and others being unacceptable. If their component morphemes are identifiable, the meaning is generally predictable from the parts, but it is more complicated to predict exactly which morphemes can be incorporated. The general rule is that the more grammaticalized its classifier function, the more likely a noun is able to be incorporated in an R-attributive word. The classifiers, in turn, should agree with the class (usually shape) of whatever noun the R-attributive word modifies. 5.4 The Morphosyntactic Distribution of Attributive Words While attributive word reduplication can be clearly distinguished semantically and morphologically from the other types of reduplication in Cha’palaa, it is also worthwhile to briefly outline the syntactic features of this word class and to note some of the ways it interacts with other reduplication types. The two main syntactic positions in which attributive words occur are as predicate elements, as in (87) and (88), and as post-nominal modifiers, as in (89). R-attributive words can either predicate on their own in stative constructions or occur as i (‘become’) class predicates, cast as changes-of-state. 11

The orthography shows a phonemic representation; however, sometimes regular phonological processes, such as voicing and assimilation, apply. In example (85) the nasal assimilates to a bilabial before the bilabial stop, and the bilabial stop is voiced after the nasal (shimbupu). The reduplicated syllable retains the unvoiced stop.

106

floyd

(87) lushishi-we blue-decl ‘(It) is blue.’ (88) lushishi i-we. blue become-decl ‘(It) became blue.’ (89) kule lushishi kee-yu canoe blue see-ego ‘I see (a/the) blue canoe.’ While some members of the R-attributive class are productively derived and others are lexicalized and opaque, all members of the class share the same general syntactic possibilities and restrictions, and all can occur in the three construction types shown above. In addition, as a coverb, descriptive words are subject to further processes of reduplication like those described in section 3 on predicate reduplication. For example, in (90) the color term unkalala (‘red’) occurs as a simplex predicate, in (91) it occurs as a reduplicated predicate with iterative aspect, and in (92) it combines with adjunct reduplication to illustrate all three major reduplication types in the same construction. (90) unkalala i-we red become-decl ‘(It) became red.’ (91) unkalala unkalala i-we red red become-decl ‘It became red repeatedly.’ (92) kuwain kuwain unkalala unkalala i-we. slow slow red red become-decl ‘It slowly and repeatedly became red.’ I have not directly addressed the prosody of attributive words and cannot do so at length here, but it should be noted that in discourse, words of this class tend to feature final-syllable stress and vowel lengthening. In Cha’palaa penultimate stress is the general norm, although other patterns exist for some specific phonological environments. For R-attributive words both penultimate and final stress with vowel lengthening are possible, and the preferred form is

four types of reduplication in cha’palaa

107

the final-stress variant. Their exceptional prosody is another suggestive detail regarding the historical relationship of attributive words to ideophones, which tend to show uncommon phonotactics. Word-final stress of descriptive words is often accompanied by a lengthening of the vowel to create a heavy syllable. Examples (93), (94) and (95) from natural discourse illustrate some common contexts for descriptive words, many as post-nominal modifiers. (93) wara tapipií pu-min pants short.hanging be.in-hab ‘(They) usually wear short pants.’ (94) tsaa pababaá achuwa te'wallulluú sem black hair curly ‘(They’re) like black, with curly hair.’ (95) mish-puka chiyayaá juntsa-a head-cl:sphere/fruit tangled dm.dst-foc ‘(It has a) tangled head, that one.’ Example (96) shows an R-attributive word as a predicate (piichilili ‘striped’); it has no overt argument but refers back to a discourse antecedent ( jali ‘clothing’). (96) jali=bain yala-' jali traje, clothing=also 3col-poss clothing clothing ‘Clothing also, their clothing (or) traje’ wee kolor-ya pu-mu-de-e-we, piichililií-wa other color-foc be.in-ag.n-pl-become-decl striped-past ‘(they) wear a different color, (they were) striped.’ While it is impossible to fully describe the properties of R-attributive words in Cha’palaa here, these details are enough to establish that they cohere as a word class in many ways and that they contrast in both form and meaning with the other types of reduplication in the language.

108 6

floyd

Conclusions

This paper organized Cha’palaa reduplication constructions into four major types, describing each of them and some additional minor types and subtypes to the extent possible in the space available. While the facts of the language are complex and many more details could be addressed in further work,12 the general four reduplication types as outlined here can accommodate and account for the great majority of specific instances of reduplication found in the data set. Beyond simply holding up well descriptively, however, this typology of reduplication constructions also allows us to appreciate their great diversity in Cha’palaa and the extent to which reduplication can be integrated with grammatical processes. Since each of the reduplication types in the language has a unique set of morphosyntactic characteristics, makes a distinct semantic contrast, and applies to different word classes and construction types, it is likely that each type developed independently, meaning that at various times in its history Cha’palaa has increased the functional load its grammar places on reduplication. Of the four types, ideophone reduplication is the least grammaticalized, and as noted in section 2, it might be better described as repetition rather than productive reduplication. However, comparing ideophone constructions to the other reduplication types helps to highlight some possible historical connections between repetition and reduplication, as patterns of ideophone repetition provide one possible model for the basic formal property of reduplication, the doubling of all or part of a word form. The two formats for ideophone reduplication, full reduplication and partial reduplication of final syllables, formally resemble the more grammaticalized types of reduplication in the language: predicate and adjunct reduplication show full forms and attributive word derivation shows final-syllable reduplication. With ideophones full reduplication is associated with temporality and iteration while partial reduplication is associated with spatial extension and distribution. Somewhat similarly, the grammaticalized types of reduplication associate full reduplication with iterated events while partial reduplication is associated with physical qualities of objects. Since there is some degree of overlap between ideophones and the other reduplicating word classes in the language, it is possible that grammaticalized reduplication originated in analogy to the less-morphosyntactically constrained ideophone reduplication, which became “tamed” by increasing

12

Important topics that remain to be addressed include reduplication in negative constructions and in interrogative and imperative sentence types, for example.

109

four types of reduplication in cha’palaa

morphosyntactic restrictions and was extended productively to other word classes. Ultimately it may be impossible to do more than just speculate about plausible ways that Cha’palaa could have developed so much grammatical reduplication while its closest relatives and neighbors did not, but the Cha’palaa data suggests that the ideophone analogy may be at least one route to grammaticalization of reduplication, raising the question of whether similar accounts could apply to other languages as well. From a synchronic perspective it is possible to rank the current reduplication types in Cha’palaa by their degree of morphosyntactic restriction. Reduplicated ideophones can occur almost anywhere in a sentence, can act as single utterances themselves, and can repeat two or more times. Adjunct reduplication must occur in association with a predicate, but its position is highly flexible with respect to the predicate, which may even be elliptical. Predicate reduplication is part of the core predicate and must occur immediately to the left of a finite verb, without possibilities of ellipsis. Table 2 compares the relative degree of morphosyntactic restriction of the three types of full reduplication in the language.13 table 2

Morphosyntactic limits of reduplication

> More restricted>

Reduplication type

Ideophone reduplication

Adjunct reduplication

Predicate reduplication

Morphosyntactic restrictions

i. free morphosyntax ii. unrestricted number of reduplicants iii. full and partial reduplication

i. loose morphosyntax restriction at phrase level ii. only one reduplicant iii. only full reduplication

i. restricted to coverb position within the core predicate ii. only one reduplicant iii. only full reduplication

13

For partial reduplication, ideophone constructions would be similarly ranked as the least restricted, as they can repeat two or more time, while attributive words are derived by a single reduplicant.

110

floyd

This continuum of relative restrictedness also resonates with the idea of a possible grammaticalization path in which unrestricted ideophone-like reduplication became more and more incorporated into the grammar and more subject to morphosyntactic restrictions. All the types of reduplication share a common iconic semantic element of iterative or distributional meaning, but there are differences with respect to how abstractly grammatical this iconicity has become over time. Cha’palaa shows how a language can recruit these iconic properties and extend them to different grammatical functions in different ways. If at a language-internal level Cha’palaa illustrates how a single language can feature diverse structural types of reduplication applied to a broad range of grammatical functions, the level of cross-linguistic comparison can provide another way for thinking about the relative weight of reduplication in the grammars of different languages. Cha’palaa can be considered a language that places a relatively large functional load on reduplication compared to languages that rely on other kinds of morphosyntactic processes for comparable grammatical processes. Thinking in these terms yields another kind of relative ranking at the language-external level among languages that rely on grammaticalized reduplication to different degrees. The following table ranks English, three of the Barbacoan languages,14 and Cha’palaa’s two main contact languages, Spanish and Ecuadorian Quechua. table 3

Relative importance of reduplication

Load on reduplication:

Quechua (Ecuadorian) Tsafiki

More> Cha’palaa

At the far left of the continuum are languages like English and Spanish, which have little or no productive reduplication to speak of, as well as Cha’palaa’s relative Awa Pit, which is reported to have no productive reduplication. In the 14

This ranking of the Barbacoan languages is based on the available sources cited in section 1. It is possible, however, that more information could emerge to alter this ranking. For instance, while Curnow (1997) found no productive reduplication for Awa Pit in his data, a future study examining different data might reveal some form of reduplication processes like those reported for its sister languages.

four types of reduplication in cha’palaa

111

middle are languages like Ecuadorian Quechua, Cha’palaa’s closest neighboring indigenous language, which limits reduplication mainly to intensification of adjectival and adverbial modifiers, and Tsafiki, Cha’palaa’s closest relative, which features some patterns similar to those of Cha’palaa but in a less grammaticalized state. At the far right are languages like Cha’palaa, which have a wide range of reduplication types and subtypes used for a variety of grammatical functions. Both looking at language-internal continua of morphosyntactic restriction among reduplication types and at language-external continua of grammars with varying functional loads placed on reduplication show promising ways for exploring the diversity of reduplication processes in the languages of South America and beyond.

Acknowledgments Thanks to the Fulbright-Hays program (2008–2009) and National Science Foundation Dissertation Improvement Fellowship (award #0921500) for funding the dissertation research during which the data for this paper was collected in Chachi communities in Ecuador. Much thanks is also due to members of the communities of Tsejpi and Zapallo Grande who hosted me and participated in recordings and elicitation, and to Johnny Pianchiche for linguistic consultation and transcription work. Additionally, comments from the editors and reviewers were very helpful in preparing the paper.

References Aikhenvald, Alexandra. 2007. Languages of the Pacific Coast of South America. In Vanishing languages of the Pacific, ed. Osahito Miyaoka, Osamu Sakiyama and Michael Krauss, 183–205. Oxford: Oxford University Press. Berlin, Brent, and Paul Kay. 1969. Basic color terms. Their universality and evolution. Berkeley: University of California Press. Brown, Penelope. 2011. Color me bitter: Crossmodal compounding in Tzeltal perception words. The Senses and Society 6/1: 106–116. Cole, Peter. 1982. Imbabura Quechua. Amsterdam: North Holland. Curnow, Timothy. 1997. A grammar of Awa Pit (Cuaiquer): An indigenous language of south-western Colombia. PhD diss., Australian National University. . 2002. Conjunct/disjunct marking in Awa Pit. Linguistics 40/3: 611–627. Dickinson, Connie. 2000. Mirativity in Tsafiki. Studies in Language 24/2: 379–422. . 2002. Complex predicates in Tsafiki. PhD diss., University of Oregon.

112

floyd

Dingemanse, Mark. 2011. The meaning and use of ideophones in Siwu. PhD diss., Radboud University Nijmegen. Dixon, Robert M.W. 2004. Adjective classes in typological perspective. In Adjective classes: A cross-linguistic typology, ed. R.M.W. Dixon and Alexandra Aikhenvald, 1–49. Oxford: Oxford University Press. Floyd, Simeon. 2011. Re-discovering the Quechua adjective. Linguistic Typology 15: 25– 63. Inkelas, Sharon, and Cheryl Zoll. 2005. Reduplication: Doubling in morphology. Cambridge: Cambridge University Press. Kita, Sotaro. 1997. Two-dimensional semantic analysis of Japanese mimetics. Linguistics 35: 379–415. Landman, Fred. 1995. Plurality. In Handbook of contemporary semantic theory, ed. Shalom Lappin, 425–457. London: Blackwell. Lindskook, John N. 1964. Vocabulário cayapa. Quito: ILV/Ministério de Educación Pública. McKay, Thomas. 2006. Plural predication. New York: Oxford University Press. Moore, Bruce R. 1961. A statistical morpho-syntactic study of Colorado (Chibcha). International Journal of American Linguistics 27: 298–307 Moore, Bruce R. 1962. Correspondences in South Barbacoan Chibcha. In Studies in Ecuadorian Indian languages: I, ed. Benjamin F. Elson, 270–289. Norman, Oklahoma: Summer Institute of Linguistics. Moore, Bruce R. 1979. Método para aprender el idioma colorado, gramática pedagógica. Quito: Instituto Lingüistico de Verano. Obando Ordóñez, Pedro. 1992. Awa-Kwaiker: An outline grammar of a Colombian/Ecuadorian language, with a cultural sketch. PhD diss., University of Texas at Austin. Schultze-Berndt, Eva. 2000. Simple and complex verbs in Jaminjung. A study of event categorisation in an Australian language. [MPI Series in Psycholinguistics 14]. Nijmegen: Radboud University Nijmegen. Tapuyo Pianchiche, Victor. 2009. Cha’palaachi dape pilla: Diccionário de la lengua Chachi. Quito: Ministerio de Educación del Ecuador/Abya Yala. Vásquez de Ruiz, Beatriz. 1988. La predicación en guambiano. Lenguas aborígenes de Colombia. [Série Descripciones 2]. Bogotá: CCELA. Vittadello, Alberto. 1988. Cha’palaachi = El idioma cayapa. Esmeraldas, Ecuador. Guayaquil: Pontificia Universidad Católica del Ecuador, Museos del Banco Central del Ecuador. Voeltz, F.K. Erhard and Christa Kilian-Hatz, ed. 2001. Ideophones. [Typological Studies in Language 44]. Amsterdam/Philadelphia: John Benjamins.

four types of reduplication in cha’palaa

Abbreviations 1 2 3 acc ag.n aug cl col com dec.ref dm.prx dm.dst dr decl dub ego emph foc hab

first person second person third person accusative agentive nominalizer augmentative classifier collective comitative deceased referent proximal demonstrative distal demonstrative different reference declarative (non-egophoric) dubitative egophoric perspective emphatic, precision focus habitual

ideo inf inf.ev

ideophone infinitive inferential/knowledge uptake evidential lim limitative, ‘only’ loc1 directional locative loc2 general locative neg negation n nominalizer past optional past tense pl plural pos.inch inchoative positional pos.stat stative positional poss possessive q question red reduplication res resultative r.cl relativizer sem semblative sr same reference

113

chapter 5

Reduplication in Nheengatu Aline da Cruz

This paper provides the first in-depth look at reduplication in Nheengatu. Phonological analysis shows that reduplication in Nheengatu is either full or partial, has the prosodic foot as the reduplicant, and is prefixational. Like other Tupi-Guarani languages, the main use of reduplication in Nheengatu is pluractionality, and it is also used for intensification. Reduplication expressing pluractionality can occur with all types of verbs. With active intransitive verbs, reduplication indicates both repetition of the event on a single occasion or on different occasions. When reduplication occurs with transitive verbs, the NP object can be omitted. Analysis shows that reduplication expressing pluractionality forces the event to be atelic, which in combination with plural indefinite NP objects leads to object omission. Reduplication expressing intensification is restricted to stative verbs. In this latter verb class, the uses of reduplication have changed over time, leading to the emergence of new varieties of Nheengatu. In the conservative variety, reduplicated stative verbs indicate distributivity of an intensified property over multiple participants; while in the progressive variety, reduplicated stative verbs only indicate intensification of a property. Nheengatu differs from other Tupi-Guarani languages in that it combines reduplication with a morpheme of ‘reflexivity/reciprocity’ in order to partially distinguish reciprocal from reflexive constructions.

1

Introduction

This paper provides an analysis of reduplication in Nheengatu, the TupiGuarani language that descends from Tupinambá and was used as língua geral1 during the colonization of the Amazonian region. Nheengatu is currently used as native language of Baré, Baniwa and Warekena groups in the Upper Rio

1 According to Rodrigues (1996, 6), the term ‘língua geral’ refers to the group of languages spoken as first language by the descendents of Portuguese men and women, speakers of a language from the Tupi-Guarani family. The languages recognized as língua geral were also

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_006

116

da cruz

Negro, Amazonia. The topic is discussed in a broad perspective by studying (i) the phonological characteristics of reduplication, which show that there is either full or partial reduplication with the prosodic foot as the reduplicant; (ii) the morphological properties, indicating the process as prefixational; (iii) the semantic properties, characterizing its functions as expressing pluractionality and intensification. This paper presents a classification of verbs in Nheengatu, in order to discuss reduplication within each type of verb. Furthermore, it shows some examples of noteworthy uses of reduplication and discusses the interaction between reduplication and object omission. The paper also shows that the use of reduplication in Nheengatu has changed over time, leading to the emergence of new varieties of Nheengatu. More specifically, with regard to reduplication of stative verbs two varieties of Nheengatu can be observed: the conservative NegroIçana variety and the Middle Negro variety. Finally, the use of reduplication in reciprocal constructions, which allows them to be distinguished from reflexive constructions, is discussed. The majority of the data used for this study was collected during fieldwork conducted between 2007 and 2009. It was obtained from indigenous communities along the Negro and Içana Rivers in the Amazon region. The paper is organized as follows: Section 2 gives a brief introduction of the history of Nheengatu and its sociolinguistic situation. Section 3 provides a phonological description of the reduplication process. Section 4 gives an overview of the different types of verbs that can be distinguished in Nheengatu, and section 5 describes the semantic interpretation of the reduplication process when applied to each type of verb. Section 6 deals with the syntactic effects of reduplication in transitive, active intransitive, and stative intransitive verbs. Section 7 deals with the use of reduplication to indicate reciprocity. The paper concludes with an overview of the main characteristics of reduplication in Nheengatu.

2

A Brief History of Nheengatu

In the 16th century Portuguese colonizers landed on the northern coast of Brazil, where they established contact with Tupinambá people. Some colonizers fathered children with local Tupinambá women. These children, their

used as língua franca by the population formed by African descendants and other indigenous groups who were incorporated into the colonial population.

reduplication in nheengatu

117

Portuguese fathers, and their indigenous families formed a new Tupinambáspeaking society: the caboclo2 society. In 1616 the first Portuguese settlement in Amazonia was created with the foundation of Belém do Pará. It was from there that caboclo people, Portuguese colonizers and Jesuit missionaries started to occupy larger portions of the Amazon. When the caboclo people moved deeper into the Amazon region, they came into contact with speakers of many other indigenous languages. Due to this migration, Tupinambá became increasingly used for communication in the Amazon: the caboclos used it to communicate among themselves, the colonizers used it to speak with indigenous groups, and indigenous groups used it to communicate with each other if they had no language in common (see Rodrigues 1996 and Bessa Freire 2004). While Tupinambá was spread deeper into the Amazon region, it underwent changes and became known as língua geral Brasílica (see Oliveira 2008 and Lee 2005). In the 18th century the língua geral Brasílica expanded along different rivers of the Amazon and evolved into different varieties (see Bessa Freire 2004 and Felix 2002). Despite the widespread use of the language and its varieties, it had been mostly replaced by Portuguese by the early 19th century. Nowadays, Brasílica is known by the remaining speakers as Nheengatu.3 It is spoken as the main language in the Upper Rio Negro region by Baré, Baniwa and Warekena peoples, who have switched from their traditional Northern-Arawak languages to Nheengatu. UNESCO estimates that there are nearly 6,000 speakers, spread out over Brazil and Venezuela. Based on data from the public health system (Martins, n.d.), it is estimated that in the municipality of São Gabriel da Cachoeira there are nearly 4,500 Nheengatu speakers who use the language on a day to day basis (almost 1,000 Warekena and nearly 3,500 Baré and Baniwa of South Içana). The Warekena group is predominantly monolingual in Nheengatu, while the Baré and Baniwa tend to be bilingual in Nheengatu and Portuguese. Nheengatu is disappearing on other rivers of the Amazon region. In the indigenous territory of Andirá-Marau, inhabited by the Sateré-Mawé people, there are some adults who speak Nheengatu (Silva and Schwade, 2013). However, they do not use it in daily communication (Sergio Meira, personal communication, 2010; Raynice Silva, personal communication in an email on Septem2 Since there is no proper English translation, the Brazilian Portuguese term caboclo is used to refer to this society. 3 According to Rodrigues (1996), the term Nheengatu was created by Couto de Magalhães (1876). It is a compound formed by the words nheen ‘speak’ and katu ‘be good,’ which means the ‘good language.’

118

da cruz

ber 5, 2011). On the Solimões River, it is believed that the last speakers died at the end of 20th century (Ana Suelly de Arruda Câmara Cabral, personal communication, May 2008). The author is unaware of any other region where Nheengatu is still spoken.

3

Phonological Characteristics of Reduplication

To explain the process of reduplication in Nheengatu, first some brief background information on its metrical system is given, and then the phonological characteristics of reduplication are described. The section ends by providing various examples. Consider the verbs u-yuká [u.ju.ˈka] ‘he kills’ and u-yuka [u.ˈju.ka] ‘he removes.’ In the former example the verb has final stress, whereas in the latter it has prefinal stress. These examples serve to illustrate why the stress placement is considered lexical (see Taylor, 1985; and Grenand & Ferreira, 1989). Even though there is a group of words with prefinal stress, several phonological processes indicate that the metrical system follows iambic rhythm. Examples of these phonological processes are: the dropping of the unstressed last syllable (e.g., u-pawa /u-ˈpaua/ [u.ˈpa] ‘he finishes’) and the formation of diphthongs to avoid words with stress on the prefinal syllable (e.g., u-putari /upuˈtaɾi/ [u.pu.ˈtaj] ‘he wants’). From a diachronic perspective, the emergence of words with prefinal stress has two sources: (a) the reinterpretation of an old unstressed morpheme -a (a marker of argumentative case), as in the case of the verb pawa ‘finish’; or (b) the insertion of a final vowel, whose function was to avoid coda, as in the case of the verb putari ‘want.’4 The formation of the prosodic foot in Nheengatu follows the iambic rhythm. Prosodic feet are disyllabic, right-headed, and assigned from right to left. In (1) the prosodic structure of the verb u-yuká ‘he exits’ is illustrated. (1)

u-yuká ‘he kills’ syllabification u ju ka iamb ( · x) end rule final ( x) surface form [u.ju.ˈka]

4 For a more detailed analysis, see Cruz (2011, chapter 2 and 3).

reduplication in nheengatu

119

Words with prefinal stress do not take the last syllable into account for the formation of the prosodic foot. Whether a word has prefinal or final stress is unpredictable. Example (2) illustrates the prosodic structure of a word with prefinal stress. The extrametrical syllable is marked by brackets: ⟨ and ⟩. (2)

u-yuka ‘he removes’ syllabification iamb end rule final surface form

u ju ⟨ka⟩ ( · x) ( x ) [u. ˈju.ka]

Syllables can be open in which case they have a CV or V structure, or they can be closed by a glide, yielding the syllable structure (C)VG (Moore, Facundes and Pires 1993). In Nheengatu closed syllables can only occur at the end of words. In Nheengatu, words must be minimally formed by a disyllabic prosodic foot. Lexical monosyllabic words /CV1/ are realized phonetically as [CV1.'ʔV2] in which V2 is a copy of V1 and a glottal stop is inserted as onset of the final syllable. One piece of evidence in support of this analysis comes from the adaptation of loanwords from Portuguese: monosyllabic words appear in Nheengatu as disyllabic, following the [CV1.'ʔV2] structure. For instance, the Portuguese word chá /ʃa/ ['ʃa] ‘tea’ is adapted into Nheengatu as xaa /ʃa/ [ʃa.'ʔa] ‘tea.’ Another bit of evidence is that in compounds, lexical monosyllabic words appear in /CV1/ form; whereas, in isolation they appear in the [CV1.'ʔV2] form. An example of this is the word ii /i/ [i.'ʔi] ‘water,’ which appears as ipiranga /i-pi’ɾãga/ [i.pi.'ɾã.ga] ‘Negro River’ (literally, red river) in its compound form. Additional evidence comes from reduplication. Lexical monosyllabic verbs are realized with [CV1.'ʔV2] structure when they occur in isolation. However, when they are reduplicated, only the underlying syllable is part of the reduplicant. For instance, the basic form of suu ‘bite’ is phonologically represented as /su/ and realized as [su.'ʔu]. When suu ‘bite’ is reduplicated, only the underlying form /su/ participates in the reduplication, creating susu ‘bite repeatedly.’ Other examples of reduplication of lexical monosyllabic verbs are saã [sã.'ʔã] ‘fell,’ reduplicated as [sã.'sã] ‘fell repeatedly,’ and maã [mã.'ʔã] ‘see,’ reduplicated as [mã.'mã] ‘see repeatedly.’ Up until now the metrical system of the language has been described, which serves as background information for understanding reduplication in Nheengatu. Now the phonological characteristics of reduplication in this language are described.

120

da cruz

As is shown by the upcoming examples, Nheengatu has full and partial reduplication. Both types of reduplication are generated by the same process: the reduplicant is formed by copying the leftmost prosodic foot of the verbal stem. The process is prefixational: the reduplicant always occurs immediately before the leftmost prosodic foot. The prefixational nature of reduplication corresponds with the general pattern of placement of morphemes: all verbal morphology (whether derivational or inflectional) is prefixational. As is discussed in more detail later, derivational morphemes can become part of the reduplicant, whereas inflectional morphemes cannot. It should also be noted that the coda of the closed syllable is never part of the reduplicant. Therefore, the glide of a falling diphthong does not participate in the reduplication: in the case of C1V1C2V2G# sequences the reduplicant is formed as C1V1C2V2. In the remainder of this section, several examples of the reduplication process are presented and analyzed with regard to their morphological and prosodic structure. Each example contains the verbal stem used as input for reduplication, the reduplicated stem, and the inflected form of the verb in 3rd person singular. Furthermore, the examples also contain the surface form, which corresponds to its phonetic implementation, and a translation. In (3) and (4) the verbal stems yuká ‘kill’ and sereu ‘lick’ are reduplicated. These verbs consist of only two syllables, thus forming the prosodic foot. Observe that in (4) the glide of the falling diphthong is not part of the reduplicant. (3)

yuká ‘kill’ Morphological analysis Phonological analysis

Input

yuká ‘kill’

syllabification iamb end rule final

ju ka ( · x) ( x)

Reduplication yuka~yuka kill~kill

syllabification iamb end rule final

ju ka ju ka ( · x) ( · x) ( x)

Inflection

syllabification u ju ka ju ka iamb ( · x) ( · x) end rule final ( x)

u-yuka~yuka 3sg.a-kill~kill

121

reduplication in nheengatu

Morphological analysis Phonological analysis Surface form

[u.ju.ka.ju.ˈka]

Translation

‘He kills repeatedly.’

(4) sereu ‘lick’ Morphological analysis Phonological analysis Input

sereu ‘lick’

syllabification iamb end rule final

se ɾew ( · x) ( x)

Reduplication sere~sereu red~lick

syllabification iamb end rule final

se ɾe se ɾew ( · x) ( · x) ( x)

Inflection

u-sere~sereu 3sg.a-red~lick

syllabification u se ɾe se ɾew iamb ( · x) ( · x) end rule final ( x)

Surface form

[u.se.ɾe.se.ˈɾew]

Translation

‘He licks repeatedly.’

In (5) and (6) verbal stems consisting of three syllables are reduplicated, illustrating that only the prosodic foot is the reduplicant and not the complete verbal stem. As was mentioned earlier, the stress in the verbal stem can occur on the final syllable, such as in (5), or on the prefinal syllable, such as in (6). Notice that in (5) the reduplicant (tiku) appears as if it has been suffixed to the verbal stem ( yatiku) to form the reduplicated stem ( yatikutiku), whereas in example (6) the reduplicant (munu) is placed as a prefix to the verbal stem (munuka). As was stated earlier, Nheengatu obeys the following rule: the reduplicant is placed before the leftmost prosodic foot. Thus, in yatiku the reduplicant (tiku) is actually placed in front of the leftmost prosodic foot (tiku) to form yatikutiku (so directly after ya, and before the final tiku).

122 (5)

da cruz

yatiku ‘drown’ Morphological analysis Phonological analysis

Input

yatiku ‘drown’

syllabification iamb end rule final

ja ti ku ( · x) ( x)

Reduplication ya⟨tiku~⟩tiku ⟨red~⟩drown

syllabification iamb end rule final

ja ti ku ti ku ( · x) ( · x) ( x)

Inflection

syllabification u ja ti ku ti ku iamb ( · x) ( · x) ( · x) end rule final ( x)

u-ya⟨tiku~⟩tiku 3sg.a-⟨red~⟩drown

Surface form [u.ja.ti.ku.ti.ˈku] Translation

‘He drowned.’ (‘He tried to emerge [from the water] repeatedly’)

In (6) the last syllable of the stem is not taken into account in the formation of the prosodic foot and, thus, is not part of the reduplicant. As was stated earlier, the group of words whose final syllable does not participate in the foot is unpredictable. Some examples of words in this group are shown in (6), (9) and (10). (6) munuka ‘cut’ Morphological analysis Input

munuka ‘cut’

Reduplication munu~munuka red~cut

Phonological analysis syllabification iamb end rule final

mu nu ⟨ka⟩ ( · x) ( x )

syllabification iamb end rule final

mu nu mu nu ⟨ka⟩ ( · x) ( · x) ( x )

123

reduplication in nheengatu

Morphological analysis

Phonological analysis

Inflection

u-munu~munuka syllabification u mu nu mu nu ⟨ka⟩ 3sg.a-red~cut iamb ( · x) ( · x) end rule final ( x )

Surface form

[u.mu.nu.mu.ˈnu.ka]

Translation

‘He cuts repeatedly’

In (7) the verbal stem purũgita ‘talk’ consists of four syllables, thus forming two prosodic feet. This example illustrates that the reduplicant is formed by the leftmost prosodic foot, and remains a single foot. (7)

purũgita ‘talk’ Morphological analysis

Input

purũgita ‘talk’

Reduplication purũ~purũgita red~talk

Phonological analysis syllabification iamb end rule final

pu ɾũ gi ta ( · x) ( · x) ( x)

syllabification iamb end rule final

pu ɾũ pu ɾũ gi ta ( · x) ( · x) ( · x) ( x)

Inflection

u-purũ~purũgita syllabification u pu ɾũ pu ɾũ gi ta 3sg.a-red~cut iamb ( · x) ( · x) ( · x) end rule final ( x)

Surface form

[u.pu.ɾũ.pu.ɾũ.gi.ˈta]

Translation

‘He has a [very long] talk’

124

da cruz

Recall that in Nheengatu words must be minimally formed by a disyllabic prosodic foot. When a verb with a monosyllabic verbal stem is reduplicated, the underlying monosyllabic morpheme is reduplicated as such. The resulting reduplicated stem is a disyllabic sequence that satisfies the minimal word requirement. In (8) this is illustrated with suu. (8) suu ‘bite’ Morphological analysis Phonological analysis Input

su ‘bite’

syllabification su iamb no foot formed end rule final no foot formed

Reduplication su~su bite~bite

syllabification iamb end rule final

su su ( · x) ( x)

Inflection

u-su~su 3sg.a-bite~bite

syllabification iamb end rule final

u su su ( · x) ( x)

Surface form

[u.su.ˈsu]

Translation

‘He bites repeatedly.’

A verbal stem used as input in the process of reduplication can have been previously derived by the causative (mu-) or by the reflexive/reciprocal ( yu-) markers. If these morphemes occur in the prosodic foot of the stem they become part of the reduplicant. In (9) mu- ‘causative’ is not part of the prosodic foot (pini), and thus it is not reduplicated. On the other hand, in (10) mu- is part of the prosodic foot (mupa) and, consequently, it is part of the reduplicant. Since personal markers are applied post-lexically, they are not part of the stem and, consequently, are never considered for the formation of the reduplicant.5

5 See Dietrich (in this volume) for examples of languages in which personal markers are considered in the formation of the reduplicant.

125

reduplication in nheengatu

(9)

pinima ‘be colourful’ Morphological analysis

Input

pinima ‘be colourful’

Phonological analysis syllabification iamb end rule final

pi ni ⟨ma⟩ ( · x) ( x )

Derivation mu-pinima caus-be.colourful

syllabification iamb end rule final

mu pi ni ⟨ma⟩ ( · x) ( x )

Reduplication

mu-pini~pinima caus-⟨red~⟩ colourful

syllabification iamb end rule final

mu pi ni pi ni ⟨ma⟩ ( · x) ( · x) ( x )

Inflection

u-mu-pini~pinima syllabification u mu pi ni pi ni ⟨ma⟩ 3sg.a-caus-⟨red~⟩ iamb (· x) ( · x) ( · x) be.colourful end rule final ( x )

Surface form

[ũ.pi.ni.pi.ˈni.ma]

Translation ‘He writes repeatedly’

(10) paka ‘wake up’ Morphological analysis Phonological analysis Input

paka ‘wake up’

syllabification iamb end rule final

pa ⟨ka⟩ (x) (x )

Derivation mu-paka syllabification caus-wake.up iamb end rule final

mu pa ⟨ka⟩ ( · x) ( x )

126

da cruz (cont.)

Morphological analysis

Phonological analysis

Reduplication mupa~mu-paka syllabification red~caus-wake.up iamb end rule final Inflection

u-mupa~mu-paka 3sg.a-red~causwake.up

mu pa mu pa ⟨ka⟩ ( · x) ( · x) ( x )

syllabification u mu pa mu pa ⟨ka⟩ iamb ( · x) ( · x) end rule final ( x )

Surface form [ũ.bã.ˈba.ka] Translation

4

‘He wakes up [an unspecific number of persons].’

Morphosyntactic Background: Types of Verbs in Nheengatu

In this section the morphosyntactic background necessary to understand the effects of reduplication in Nheengatu is provided. Since reduplication is restricted to verbs, the types of verbs of the language, along with their inflectional properties and transitivity, are described. Nheengatu, like many other Tupi-Guarani languages, has an active-stative system, i.e., a system that divides intransitive predicates into active and stative. This system is based on semantic factors, which are not entirely transparent because of the interference of grammaticalization and lexicalization (Mithun 1991). In general, verbs that express activities and processes are categorized as active, whereas verbs that express states are categorized as stative. The system can be described as follows: in transitive verbs the subject (A)6 is marked by prefixes of the agent-like series, while the object (O) is expressed by an NP. Active intransitive verbs express their unique argument (SA) in the same way as the subject of the transitive verbs (A): by prefixes of the agent-like series

6 Following Dixon (1979): A, subject of a transitive clause; O, object of a transitive clause; SA the unique core argument of an active intransitive clause; So the unique core argument of a stative intransitive clause.

reduplication in nheengatu

figure 1

127

Types of verbs in Nheengatu

(SA = A). Stative intransitive verbs, on the other hand, do not express their unique argument (SO) in this manner (SO ≠ A). Within stative intransitive verbs two further classes can be distinguished: inflectional and non-inflectional. Non-inflectional intransitive stative verbs are prefixless, and thus their unique argument is only expressed by an NP. Inflectional intransitive stative verbs indicate their unique argument by a personal prefix of the non-agent-like series.7 The source of this division is traceable in the historical development of the language. Until at least the 19th century, all stative intransitive verbs could combine with prefixes of the non-agent like series. However, at some point in time, a group of stative intransitive verbs in Nheengatu lost their ability to combine with prefixes of the non-agent like series. The reason that non-inflectional stative verbs are still considered as verbs is because this class shares many morphosyntactic properties with other verbs: (a) they all function primarily as predicates; (b) they must take a nominalizer to function as argument; (c) they do not function as a modifier in the noun phrase, except when inside a relative clause.8 In Figure 1 below, the classification of verbs in Nheengatu is summarized. In the examples below a transitive verb (11), an active intransitive verb (12), an inflectional stative intransitive verb (13) and a non-inflectional stative intransitive verb (14) are provided to illustrate the different types of verbs. In (11) A is indicated by a personal prefix of the agent-like series, while its O is

7 The prefixes of the non-agent-like series are not restricted to verbs, but they can also occur as a genitive complement of nouns and postpositions. 8 For more details and examples, see Cruz (2011, chapter 4).

128

da cruz

lexically manifested: yepe piasa ‘a piassava tree.’ In (12) SA is also indicated by a prefix of the agent-like series. In (13) SO is indicated by a personal prefix of the non-agent-like series. In (14) SO is expressed by an NP and the verb remains prefixless. Notice that in (14) the manifestation of SO as an NP is obligatory, whereas in (11), (12) and (13) it is optional. (11) Transitive (ixe) a-munuka yepe piasa I 1sg.a-cut indf piassava.tree ‘I cut a piassava tree.’ (12) Active intransitive (ixe) a-wata I 1sg.a-walk ‘I walked.’ (13) Inflectional stative intransitive (ixe) se-kuere=wã I 1sg.na-be.tired=pfv ‘I am already tired.’ (14) Non-inflectional stative intransitive ixe puranga I be.beautiful ‘I am beautiful.’

5

Semantic Interpretation of Reduplication: The Expression of Pluractionality

Reduplication in Nheengatu is mainly used to express pluractionality, which is defined by Cusic (1981, 64) as a semantic category that expresses “the multiplicity of actions, events, occurrences, occasions and so on.” In transitive and active intransitive verbs, pluractionality indicates that the event expressed by the predicate takes place more than once. In stative intransitive verbs the expression of pluractionality is mainly interpreted as distribution of a property among participants. Additionally, to some degree it is also interpreted as an intensification of the property expressed by the verb. As an example of the semantic effect of reduplication, compare (15a) to (15b). In (15a) a sentence with the basic intransitive verb kupiri ‘clear (i.e. clear

reduplication in nheengatu

129

vegetation)’ is provided, while in (15b) the verb kupiri is partially reduplicated. Whereas (15a) does not make any reference to the number of times that the clearing event has taken place, (15b) denotes that the clearing event occurred several times. The utterances of (15) were used in a fable about a sloth that goes to the field every day to clear it. (15) a. ape ai u-kupiri conj sloth 3sg.a-clear ‘Then, the sloth cleared (vegetation).’ b. ape paa u-kupi~kupiri conj rep 3sg.a-red~clear ‘Then, they say he cleared [every day, over a long period].’ In the pair of examples in (16), a basic and a reduplicated stative verb are shown. In (16a) the basic verb puranga ‘be beautiful’ selects an NP that expresses a single entity: semanhã ‘my mother.’ The particle retã ‘intensifier’ expresses the intensification of the property. In (16b) the reduplicated stative verb purapuranga ‘be beautiful’ indicates that the property is distributed over participants. Additionally, despite the lack of the particle retã ‘intensifier,’ it also indicates the intensification of the property. (16) a. puranga retã se-manha be.beautiful ints 1sg.a-mãe ‘My mother was very beautiful.’ b. ta-ranha9 pura~puranga 3pl.na-tooth red~be.beautiful ‘Each one of their teeth was very beautiful.’ Thus far, it has been shown that Nheengatu uses reduplication to indicate multiplicity of events. In transitive verbs reduplication can be used to express that the repeated events affect multiple participants. The language can also

9 The morpheme ta- ‘3rd plural non-agent-like argument’ (used with nouns, postpositions and inflectional stative verbs) is homophonous with ta- ‘3rd plural agent-like argument,’ used to indicate the subject of active verbs. The forms differ in distribution and function. Furthermore, ta- ‘agent-like’ has the allomorphs tau- in conservative varieties and tu- in the variety spoken on the Xié River, while the non-agentive form is always ta-.

130

da cruz

indicate multiple participants through nominal plurality.10 The difference between these two strategies is that reduplication is used with an implicit plural indefinite NP object, as illustrated in (17a); whereas, the expression of nominal plurality involves a plural definite NP object, as illustrated in (17b). Incidentally, these examples also support Wood’s hypothesis (2007) that pluractionality, expressed in Nheengatu by reduplication, and nominal plurality have complementary uses. (17) a. ta-yuká~yuká 3pl.a-kill~kill ‘They killed [a large number of animals]!’ b. Davi u-yuká suu ita Davi 3sg.a-kill animal pl ‘Davi killed the animals.’

6

Reduplication per Type of Verb

In section 4 it was shown that verbs in Nheengatu can be classified into transitive, active intransitive, inflectional and non-inflectional stative intransitive types. In this section more detail is provided about what can happen when each of these types of verbs is reduplicated. In the process of reduplication, noninflectional and inflectional stative intransitive verbs behave similarly; hence, reduplication of stative intransitive verbs as a whole is presented in only one section. 6.1 Transitive Verbs Reduplicated transitive verbs indicate an event that occurs repeatedly. The event can affect a single participant over multiple occurrences. This is illustrated in (18b) where the different occurrences of sereu ‘lick’ always affect

10

In Nheengatu the form ita is a plural particle that originated through grammaticalization of the noun eta ‘many’ from Tupinambá. As a plural particle, ita can be combined with numerals (e.g. vinti uka ita ‘twenty houses’) and with the loanword mui ‘many’ (e.g. mui mira ita ‘various persons’). Moreover, ita can trigger agreement with the third person plural, as in Bare ita ti=wã=ta-kuntai Baré (Bare pl neg=pfv=3pl.a–talk ‘Barés do not talk Baré anymore’) in which ita of the NP subject Bare ita agrees with the verb ta-kuntai ‘3pl.a-talk.’ For more details, see Cruz (2011, 377–378).

reduplication in nheengatu

131

the same patient participant (expressed by the definite NP object i-noiva ‘his fiancée’). Similarly, in (19b) su~su ‘red~bite’ repeatedly affects the same patient participant (expressed by the NP whose nucleus is a demonstrative). (18) a. ape paa nhaã mikura u-sereu ae conj rep dem mucura 3sg.a-lick dem ‘Then, they say that a mucura11 licked her.’ b. u-sere~sereu paa nhaã i-noiva 3sg.a-red~lick rep dem 3sg.na-fiancée ‘They say that he repeatedly licked his fiancée.’ (19) a. yakaré u-suu inde alligator 3sg.a-bite you ‘The alligator bit you.’ b. ta-su~su-pa ae 3pl.a-bite~bite-complete dem ‘They bite it completely [by biting it several times].’ Alternatively, each occurrence of an event can affect multiple participants, as in (20b) where the reduplicated verb puem ‘grab’ indicates that a large number of fish has been caught. Likewise, in (21b) each occurrence of munuka ‘cut’ affects a different piassava tree. (20) a. u-puem pirá 3sg.a-grab fish ‘He grabs fish.’ b. u-puem~puem nhaã garape upe 3sg.a-grab~grab dem river loc ‘He takes them [a large number of fish] from that river.’ (21) a. a-munuka yepe piasa 1sg.a-cut indf piassava.tree ‘I cut a piassava tree.’

11

Mucura is a small marsupial, similar to an opposum.

132

da cruz

b. yawe paa u-munu~munuka be.like.that rep 3sg.a-red~cut ‘They say that he cut [a large number of piassava trees].’ Based on examples in the corpus, a general pattern can be described with regard to object omission and reduplication: when reduplication is used with a plural indefinite NP object, the NP object is omitted, as illustrated in (20b) and (21b). A theoretical justification for this pattern is given by Hopper & Thompson (1980). Those authors hypothesized that the presence of an NP object, or lack thereof, is related to several properties, such as individuation of the NP object and telic events. Atelic events are more likely to lead to object omission than telic events. Similarly, highly non-individuated NP objects are more likely to be omitted. On the one hand, reduplication, by expressing pluractionality, forces the event to be interpreted as atelic. For instance, in (21a) the event munuka yepe piasa ‘cut a pissava tree’ is telic, whereas its reduplicated counterpart in (21b) is atelic. On the other hand, plural indefinite NP objects are highly non-individuated. The combination of the atelic event and the plural indefinite NP has a high chance of leading to object omission. 6.2 Active Intransitive Verbs The active intransitive verbs mark their unique argument (SA) with a personal prefix of the agent-like series. The reduplication of this kind of verb indicates either a repetition of the event on a single occasion, as illustrated in (22), or a repetition of the event on different occasions, as illustrated in (23). (22) ape paa u-puka~puká nhaã taina ta-irũ conj rep 3sg.a-laugh~laugh dem child 3pl.na-com ‘Then, they say he laughed with those children.’ (23) ti=ya-pudei ya-tiri~tirika amũ tetama ta-rupi neg=1pl.a-can 1pl.a-red~move other region 3pl.na-perl ‘We cannot move [several times] to other regions.’ Reduplicated active intransitive verbs can select an NP subject which expresses either single or multiple entities. With NP subjects expressing a singular entity, the interpretation is that the action is being repeated several times, as illustrated in (24). With NP subjects expressing multiple entities, it triggers a distributive reading, as illustrated in (25). Notice that, while the multiple participants in (23) above and (25) below allow for a distributive reading, they do not necessarily realize the action simultaneously.

reduplication in nheengatu

133

(24) nhaã u-pua-puamu dem 3sg.a-red~stand.up ‘He stands up (there and then again)’ (25) taina-mi ita ta-yaxiu~xiu child-dim pl 3pl.a-red~cry ‘Each of the children cried’ 6.3 Stative Intransitive Verbs With regard to reduplication of stative intransitive verbs, two different varieties of Nheengatu can be distinguished: a conservative variety, spoken where the Negro and Içana Rivers meet (called here the ‘Negro-Içana variety’), and a progressive variety, spoken in the city of São Gabriel da Cachoeira and in the Middle Rio Negro (called here the ‘Middle Negro variety’). Section 6.3.1 describes the stative verbs of the Negro-Içana variety, and 6.3.2 those of the Middle Negro variety. In 6.3.3 the social backgrounds of the variations are briefly discussed. 6.3.1 Stative Intransitive Verbs in the Negro-Içana Variety Reduplicated stative verbs indicate distribution of a property among multiple participants, and, consequently, they have to select an NP subject that expresses multiple entities. Multiple entities can be expressed by subjects marked plural, as in (26); by subjects preceded by a quantifier muiri ‘a lot,’ as in (27); or by a mass noun, as in (28).12 Reduplicated stative verbs with an NP subject explicitly marked as singular are considered ungrammatical, as exemplified in (29). Reduplication of stative verbs also expresses intensification of the distributed property, which is exemplified in (26), (27) and (28). (26) ta-kaú~kaú paa nhaã suayara ita 3pl.na-be.drunk~be.drunk rep dem 3sg.na.brother.in.law pl ‘They say that each one of her brothers-in-law was very drunk.’ (27) Muiri kuema paa Wakurawá pitu~pituna u-su=wã kaa kiti a.lot morning rep Wakurawá red~be.dark 3sg.a-go=pfv forest all ‘They say that several mornings were dark, and Bacurau went to the forest.’ (Casasnovas, 2006, 97; my analysis) 12

The fruit yaka ‘bacaba’ (Oenocarpus bacaba) is a small palm tree fruit, similar to açaí. Since the individual fruits are clustered together, they are not counted separately. It takes a large amount of these fruit clusters to make juice. Thus, yaka is considered a mass noun.

134

da cruz

(28) pura~puranga kua yaka red~be.beautiful dem bacaba ‘Each one of the bacaba fruits was very beautiful.’ (29) *yepe kunhã pura~puranga indf woman red~be.beautiful ‘One woman is very beautiful.’ The requirement that reduplicated stative verbs have to select an NP subject that expresses multiple entities is a conservative pattern. In the 19th century Hartt (1872; 1938) documented sentences in which this pattern can be observed; some of these sentences are shown in (30), (31) and (32).13 In (30) pixúm~pixum indicates that the property of ‘being black’ is distributed among participants; similarly, in (31) puku~pukú also indicates a distributive reading. Notice that Hartt translates14 both sentences with an adverb of intensification, even though only in (31) the particle reté ‘intensifier’ is actually present. In (32) the basic verb kyá ‘be dirty’ is intensified by the particle reté, and Hartt translated this in the same manner. (30) i-pixún~pixun sesá 3sg.na-be.black~be.black 3sg.na.eye ‘His eyes are very black.’ (Hartt, 1938, 348; my analysis) (31) mokoin itá pokú~pokú reté two stone be.long~be.long ints ‘Two stones (iron) is very long.’ (Hartt, 1938, 313; my analysis) (32) yg kyá reté water be.dirty ints ‘The water is very dirty.’ (Hartt, 1938, 352; my analysis) 6.3.2 Stative Intransitive Verbs in the Middle Negro Variety It was shown in the previous section, based on data that was collected in the 19th century, that the use of reduplication to indicate distribution among

13 14

In examples (30), (31) and (32), the spelling of the author (Hartt 1938) was kept. Hartt’s work was first published in English in Hartt (1872). A more elaborate version was published in Portuguese in Hartt (1938). The sentences in (30), (31) and (32) are from the latter version.

reduplication in nheengatu

135

participants is a conservative pattern of the language. The functionality of reduplication of stative verbs has undergone changes in the Middle Negro variety of Nheengatu: the distributive reading has been lost as well as the requirement of the NP subject to indicate multiple entities. Some examples of reduplication in the Middle Negro variety are given in (33), (34) and (35). In (33a), (34a) and (35), reduplicated stative verbs select an NP subject with no plural markers. This expresses, by default, that there is one single entity. Alternatively, the NP can be marked by the plural particle ita to indicate multiple entities, which can be seen in (33b) and (34b). Notice that the construction in (33a) and (34a) is impossible in the Negro-Içana variety: in that variety, reduplicated stative verbs select an NP subject which expresses multiple entities (see examples (26), (27), (28) and (29) in the section 6.3.1 above). (33) a. kunhã puku~puku woman be.long~be.long ‘The woman is very tall.’ b. kunhã ita puku~puku woman pl be.long~be.long ‘The women are very tall.’ (34) a. apiga sasi~sasiara man red~be.sad ‘The man is quite sad.’ b. apiga ita sasi~sasiara man pl red~be.sad ‘The men were quite sad.’ (35) ape paa pituna puku~puku=wã nhaá Kurupira u-sika conj rep be.dark be.long~be long=pfv dem Kurupira 3sg.a-arrive ‘Then, they say, the night was very long, and Kurupira arrived.’ (Data from Floyd, n.d.; adapted gloss) In the Middle Negro variety multiple entities are expressed by the plural particle, and the reduplicated stative verb can select a singular or plural NP subject. Therefore, it can be concluded that in this variety reduplication is only used to express intensification of a property and has lost the ability to imply distribution of a property among participants.

136

da cruz

6.3.3 Dialectal Variation and Influences of Language Contact In order to better explain why the Negro-Içana variety is more conservative, while the Middle Negro variety underwent more changes, the sociolinguistic situation of both varieties is discussed in this section. In the Negro-Içana variety (6.3.1), Nheengatu is used in daily conversations and Portuguese is only used to maintain commercial relations with the nearest city. Reduplicated stative verbs only allow the choice of an argument that is interpreted as multiple entities. In this area reduplication of stative verbs is used to indicate distributivity of an intensified property among participants, a pattern inherited from Tupi languages. It is plausible that the sociolinguistic situation in which Nheengatu is still acquired as a first language contributes to the conservation of the use of reduplication to indicate distributivity. In the Middle Negro variety (6.3.2), Nheengatu is only spoken by adults, and it has been replaced by Portuguese for everyday conversations. It is plausible that this dominance of Portuguese has contributed to a faster change in the way Nheengatu uses reduplication. The loss of the requirement that the NP subject express multiple entities can be interpreted as a typological change in the way Nheengatu expresses the number category: reduplicated stative verbs can only indicate distributivity of a property over participants when the NP subject is explicitly marked as plural. When the NP subject is singular, reduplicated stative verbs express only intensification of a property. The requirement of a plural marker in the NP to indicate multiple entities can be associated with the intense contact with Portuguese.

7

Reduplication Expressing Reciprocity

Cross-linguistically, reduplication, as the main process through which pluractionality is expressed, is often associated with reciprocity (Nedjalkov 2007). The ability to express repetition of events makes reduplication a good strategy to express reciprocity: the event is affecting at least two participants (A affects B, and B affects A), and each occurrence of the event affects one of the two participants. Reduplication expresses pluractionality, and the prefix yu- ‘reflexive/reciprocal’ marks that the subject has to be interpreted as both agent and patient. Combined together, these morphemes form the reciprocal construction. Examples of reciprocal constructions are given in (36), (37) and (38).

reduplication in nheengatu

137

(36) ta-yu-mundu~mundu 3pl.a-r/r-give.order~give.order ‘They give orders to each other.’ (37) ya-yu-mã~mã 1pl.a-r/r-see~see ‘We were looking at each other.’ (38) taina ita ta-yu-miru~miru child pl 3pl.a-r/r-annoy~annoy ‘The kids got annoyed with each other.’ In Nheengatu the prefix yu- marks both reflexivity and reciprocity. When the stem of the verb is reduplicated, the only possible interpretation of yuis reciprocity.15 Even if the verb is not reduplicated, a verb with the marker yu- can still be interpreted as reciprocal. Consider the examples in (39) and (40). In (39) the verb ajudai ‘help’ is combined with the reflexive/reciprocal marker yu-, leaving the resulting construction ambiguous. Since ajudai is a loanword which cannot be reduplicated, it can only be interpreted as reciprocal or as reflexive within the proper context. In (40a) the verb munuka ‘cut,’ combined with yu-‘reflexive/reciprocal,’ is also ambiguous (albeit preferably reflexive). In (40b) the form yumunuka is combined with reduplication in order to avoid ambiguity. The resulting construction can only be interpreted as reciprocal. (39) ta-yu-ajudai 3pl.a-r/r-help ‘They help themselves (or They help each other).’ (40) a. ta-yu-munuka 3pl.a-r/r-cut ‘They cut themselves (or They cut each other).’ b. ta-yu-munu~munuka 3pl.a-r/r~red-cut ‘They cut each other.’

15

A similar construction is observed in the language Apalai (Cariban) (Nedjalkov 2007, 176).

138

da cruz

The use of reduplication in reciprocal constructions is not an innovation of Nheengatu. In Kamaiurá, another Tupi-Guarani language, reciprocal constructions can also be combined with reduplication (Seki 2000, 284). However, in Kamaiurá reduplication in reciprocal constructions is not necessary for making a clear distinction between reciprocal and reflexive events because this language has a specific morpheme for each of these functions: je ‘reflexive’ and jo ‘reciprocal.’ These two morphemes are inherited from Proto-Tupi-Guarani, as reconstructed by Jensen (1998). What is new is that Nheengatu has lost the cognate of *je, and the cognate of *jo has had its use expanded to indicate both ‘reflexivity’ and ‘reciprocity.’ In order to partially reestablish the distinction between reciprocal and reflexive, yu- is combined with reduplication, meaning reciprocal.

8

Conclusion

This paper describes reduplication as a prefixational process used to form reduplicated verbal stems. The phonological analysis indicates that the reduplicant is formed by the leftmost prosodic foot of the verbal stem. Prefixational reduplication with a prosodic foot as reduplicant is a common feature of TupiGuarani languages (see, for instance, Borges 2008, Rose 2008). Nheengatu differs from more conservative languages of this family because it does not have monosyllabic reduplication, a process that occurs, for instance, in Tupinambá (Rodrigues 1953), the language from which Nheengatu descends. Having introduced the phonological characteristics of reduplication in Nheengatu, the uses of reduplication with each type of verb were addressed. It was shown that reduplication mainly expresses pluractionality. Additionally, in stative verbs, it expresses intensification of properties. In the analysis of transitive verbs, it is possible to let repeated events always affect the same patient participant, which is expressed by a manifest definite NP object. Alternatively, it is also possible to let each occurrence of the repeated event affect a different patient participant. In this latter case, the expected NP object, which is plural indefinite, is omitted. The fact that object omission occurs with reduplicated transitive verbs when the expected NP object is indefinite and plural reinforces the notion that there is an interaction among pluractionality, referentiality and transitivity (see, for instance, Hopper and Thompson 1980, Cusic 1981 and Queixalós 2002). It remains to be seen whether object omission with reduplicated transitive verbs is a common feature of other Tupi languages. In active intransitive verbs reduplication is used to indicate multiplicity of occurrences of an event. And, in conjunction with a plural NP subject, it

reduplication in nheengatu

139

can trigger a distributive reading. Reduplication with stative verbs underwent changes: it used to select an NP subject without the need of a plural marker to express multiple entities. This original use of reduplication was associated with distributivity. However, this is no longer the case in a progressive variety of Nheengatu, in which reduplicated verbs can select an NP subject that can be either plural or singular. In the latter case, reduplication is interpreted only as intensification of a property. This latter pattern is likely associated with the intense contact with the Portuguese language. The final section of this paper shows that another function of reduplication in Nheengatu is to express reciprocal constructions. Reciprocal constructions are formed by combining: yu-, which expresses both reflexivity and reciprocity, and reduplication, which expresses pluractionality. The combination of these morphological processes distinguishes reciprocal from reflexive meanings.

Acknowledgments I wish to thank Leo Wetzels, Francesc Queixalós, Antoine Guillaume, Léia Silva, Gale Goodwin Gómez, Hein van der Voort and the anonymous reviewer for making comments on this paper. I am also grateful to Patricia Cabredo Hofherr for comments on my analysis of Nheengatu data.

References Bessa Freire, José Ribamar. 2004. Rio Babel—a história das línguas na Amazônia. Rio de Janeiro: Atlântica. Borges, Mônica Veloso. 2008. Reduplicação em Avá-Canoeiro (Tupi-Guarani): uma língua brasileira ameaçada de extinção. UniverSOS—Revista de Lenguas Indígenas y Universos Culturales 5: 233–243. Casasnovas, Afonso. 2006. Noções de língua geral ou Nheengatú—Gramática, lendas e vocabulário. Manaus: Editora da Universidade Federal do Amazonas. Couto de Magalhães, José Vieira. 1876. O Selvagem. Rio de Janeiro: Typographia da Reforma. Cruz, Aline da. 2011. Fonologia e Gramática do Nheengatú: A língua geral falada pelos povos Baré, Warekena e Baniwa. Utrecht: LOT. http://www.lotpublications.nl/ publish/issues/Cruz/index.html (accessed November 15, 2013). Cusic, David D. 1981. Pluractionality and Aspect. PhD diss., Stanford University. Dietrich, Wolf. this volume. Forms and functions of reduplication in Tupian languages.

140

da cruz

Dixon, Robert M.W. 1979. Ergativity. Language 55/1: 59–138. Felix, Maria Ivamete de Santana. 2002. A Língua Geral Amazônica: Contribuição para o estudo de suas variedades dialetais faladas ao longo do Rio Amazonas e seus tributários no século XIX. MA thesis, Universidade Federal do Pará. Floyd, Simeon. n.d. Nheengatu data. Unpublished document. Nijmegen. Grenand, Françoise, and Epaminondas Henrique Ferreira. 1989. Pequeno dicionário da Língua Geral. Manaus: SEDUC. Hartt, Charles Frederick. 1938. Notas sobre a língua geral ou tupi moderno do Amazonas. Anais da Biblioteca do Rio de Janeiro 54: 307–390. . 1872. Notes on the Lingoa Geral or Modern Tupí of the Amazonas. Transactions of the American Philological Association 3: 58–76. Hopper, Paul J., and Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56/2: 251–299. Jensen, Cheryl. 1998. Comparative Tupi-Guarani morphosyntax. In Handbook of Amazonian Languages, vol. 4, ed. Desmond. C. Derbyshire and Geoffrey. K. Pullum, 480– 618. Berlin/New York: Mouton de Gruyter. Lee, Kittiya. 2005. Conversing in colony: the Brasílica and the Vulgar in Portuguese America, 1500–1759. PhD diss., The Johns Hopkins University. Martins, André Luiz. n.d. Sistema de Informação da Atenção à Saúde Indígena (SIASI), 2008. Unpublished document. São Gabriel da Cachoeira. Mithun, Marianne. 1991. Active/agentive case marking and its motivation. Language 67/3: 510–546. Moore, Denny, Sidney Facundes, and Nádia Pires. 1993. Nheengatu (LGA), its history, and effects of language contact. In Survey of California and other Indian Languages 8: 93–118. Moseley, Christopher, ed. 2010. Atlas of the world’s languages in danger, 3rd ed, Paris: UNESCO Publishing. http://www.unesco.org/culture/en/endangeredlanguages/ atlas (accessed November 15, 2013). Nedjalkov, Vladimir. 2007. Encoding of the reciprocal meaning. In Reciprocal constructions, ed. Vladimir Nedjalkov, 147–208. Amsterdam/Philadelphia: John Benjamins. Oliveira, Renata Lopes Gomes de. 2008. Natureza e direções das mudanças lexicais ocorridas na Língua Geral Amazônica do século XVII. MA thesis, Universidade de Brasília. Queixalós, Francesc. 2002. Sur la distensivité. In Mémoires de la Société de Linguistique de Paris: La pluralité, 55–71. Louvain: Peeters. Rodrigues, Aryon Dall’Igna. 1996. As línguas gerais sul-americanas. PAPIA—Revista de Crioulos de Base Ibérica 4/2: 6–18. . 1953. Morfologia do verbo Tupi. Letras 1: 121–152. Rose, Françoise. 2008. Action répétitive et action répété: Aspect et pluralité verbale dans la réduplication en émérillon. Faits de Langues 28: 125–132.

141

reduplication in nheengatu

Seki, Lucy. 2000. Gramática do Kamaiurá—Língua Tupi-Guarani do Alto Xingu. Campinas & São Paulo: Editora da Unicamp & Imprensa Oficial. Silva, Raynice Geraldine Pereira da, and Michéli Carolíni de Deus Lima Schwade. 2013. Interferências fonético-fonológicas do Mawé no Nheengatu do rio Andirá (Médio Rio Amazonas). Paper presented at the II Congresso Internacional da Faculdade de Letras da UFRJ, September 2–5, in Rio de Janeiro. Taylor, Gerald. 1985. Apontamentos sobre o Nheengatu falado no Rio Negro, Brasil. Amerindia 10: 5–23. Wood, Esther. 2007. The semantic typology of pluractionality. PhD diss., UC Berkeley.

Abbreviations 1 2 3 a all caus com dem indf loc na

first person second person third person agent-like argument allative causative commitative demonstrative indefinite locative non-agent-like argument

neg neg.contr perl pfv pl red rep restr r/r sg

negation contrastive negation perlative perfective plural reduplication reportative restrictive reflexive /reciprocal singular

chapter 6

Reduplication in Hup (Northwest Amazonia) Patience Epps

This article contributes a case study of reduplication in Hup, a Nadahup (Makú) language spoken in northwest Amazonia, and considers its form, function, and diachrony in typological perspective. In Hup, the phonotactics of reduplicated forms are intermediate between those of canonical monomorphemic and multimorphemic words. Reduplication functions primarily to indicate iterativity in verbs and also serves an attributive function in a limited number of lexicalized nominal forms that are composed of a noun and an adjective or stative verb. Synchronically, repetition and reduplication in Hup are not fully distinct, lending support to the view that the two processes may be diachronically related.

1

Introduction

This article presents a case study of reduplication in Hup, a member of the small Nadahup (Makú)1 family of the northwest Amazon, with attention to both formal and functional aspects. Like many Amazonian languages, Hup was until recently almost completely undescribed and its contribution to our understanding of typological phenomena like reduplication still unknown. As this case study reveals, reduplication and related phenomena in Hup resemble in many ways those found in other languages of Amazonia and of the world, but also reveal certain intriguing variations that may inform our understanding of reduplication more generally. From the perspective of form, reduplicated words in Hup are interesting in that they belong to a small set of words whose phonotactics are intermediate between those exhibited by canonical monomorphemic and multimorphemic words. From the perspective of function and distribution, Hup reduplication is semi-productive in verbs, whereas

1 ‘Nadahup’ is preferred to ‘Makú’ because the latter name is used as an ethnic slur in the region and has also been applied to several unrelated language groups in Amazonia. ‘Nadahup’ combines the names of the four established member languages in this family (Hup, Yuhup, Dâw, Nadëb).

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_007

144

epps

only ‘inherent reduplication’ (where no simplex form can be identified) has been observed in nouns (with one possible exception). Reduplication has a further, marginal function of marking attributive modifiers in a small class of lexicalized compounds. While this attributive use of reduplication is typologically unusual, I argue that it is nevertheless iconically motivated and is thus consistent with reduplication’s other functions both in Hup and cross-linguistically. Finally, a comparison of repetition and reduplication phenomena in Hup indicates that the distinction between the two is not entirely clear-cut, which may in turn be evidence for a diachronic link between them.

2

Hup and Its Speakers

Hup is spoken in the Vaupés region, an area of northwest Amazonia that straddles the Brazil-Colombia border. It has approximately 1500 speakers and is at present fully viable, learned as a first language by virtually all Hup children. The Vaupés region is well known for the linguistic exogamy—obligatory marriage across language groups—practiced by its Tukanoan and Arawak inhabitants (see, e.g., Sorensen 1967; Jackson 1983). While Hup and other Nadahup peoples do not participate in this exogamy system, Hup speakers are nonetheless in frequent interaction with speakers of Tukanoan languages and are in general bilingual. The high degree of multilingualism of the Vaupés peoples has fostered extensive language contact, leading to considerable grammatical diffusion among the various languages (see Aikhenvald 2002 for Tariana/Tukanoan contact; Gomez-Imbert 1996 for Cubeo/Baniwa; and Epps 2007, 2008b for Nadahup/Tukanoan). Hup’s phonological system is characterized by nine vowels (of which only six are contrastive in nasal contexts), nineteen consonants (plus /p'/, which is only marginally phonemic), and contrastive tone and nasalization.2 Hup’s vowel and consonant inventories are provided in Tables 1 and 2 below.

2 In general, the orthography used here conforms to the International Phonetic Alphabet, except that y is used for the glide /j/, and j for the voiced palatal stop /ɟ/. High tone is indicated by v́ and rising tone by v̌. Although nasalization is (in general) a property of the entire morpheme in Hup, it is indicated orthographically here on the segmental level in the interest of reader-friendliness: Where at least one consonant or vowel in a syllable is nasal, all other segments (other than those transparent to nasalization) should be assumed to be nasal even if unmarked.

145

reduplication in hup (northwest amazonia) table 1

Close Mid Mid/Open

table 2

Hup vowels

Front

Central

Back

i e æ

ɨ ǝ a

u o ɔ

Hup consonants

Bilabial Dental-alveolar Palatal Velar Glottal Voiceless stops Voiced stops Glottalized stops Fricatives (voiceless) Glides Glottalized glides

p b b', (p') w w'

t d d'

c j j' ç y y'

k g g'

ʔ

h

Hup’s two contrastive tones occur only on the syllable of the word that receives lexical stress. Nasalization is contrastive on the level of the syllable, rather than the segment, and normally the entire morpheme is either wholly nasal or wholly oral. There is a strong isomorphism between the morpheme and the syllable in Hup, such that most roots consist of only one syllable. Hup syllables usually have both onsets and codas, and most are CVC; no diphthongs or consonant clusters exist. CV syllables are obligatorily lengthened to CVV when word-final; VC syllables occur only in a restricted class of suffixes, and take an epenthetic onset by geminating the coda of the preceding root where one exists. Accordingly, the minimal Hup word is monosyllabic, and obligatorily bimoraic and consonant-initial, e.g. mɔ̌y ‘house.’ For multisyllabic, monomorphemic words, primary stress normally falls predictably on the final syllable. Multimorphemic words are made up of roots, which may be compounded, plus any derivational or inflectional morphology. Compounding is a particularly productive operation for verbs, in which up to five roots may occur together. Phonological wordhood for these multimorphemic forms is defined primarily on the basis of stress patterns (which vary according to the type of compound or category of suffix involved); secondary stress is iambic.

146

epps

Noteworthy features of Hup morphosyntax include its verb-final basic constituent order, preference for dependent-marking, and nominative-accusative alignment. Hup exhibits extensive verb compounding, suffixation, and a tendency toward zero anaphora, such that arguments (especially objects) are frequently dropped. For a detailed description of Hup phonology and grammar, please see Epps (2008a).

3

Formal Aspects of Hup Reduplication

Reduplication in Hup affects only lexical roots and in most cases involves only a partial copy of the base (cf. Moravcsik 1978, 328; Rubino 2005, 12; compare the repetition phenomena discussed in section 5 below). The process appears to be best understood as involving a CVC copy of the first syllable of the base (most roots are only one syllable), in which the final consonant is left unspecified. This procedure is in keeping with Hup’s preference for CVC syllable structure, as noted above. The reduplicated material occurs at the beginning of the base.3 Evidence for an unspecified final consonant in the reduplicative template is the following. First, where the initial consonant of the base is a voiced stop or its nasal allophone, the pronunciation of the reduplicated form includes a voiceless stop as a syllable coda, homorganic with the first consonant of the base: (1)

a. bubud'- [mbup.but ̚] ‘roll into a coil’ (bud'- ‘roll into a circle’)4 b. maman'- [mãp.mãt ]̚ ‘roll around and around another object’ (man'- ‘roll around an object once’)

Similarly, where the initial consonant of the root is a voiceless stop, fricative, or glide, the reduplication produces a geminate medial consonant: (2)

a. kokot- [kok.kot ̚] ‘move in circles’ (kot- ‘move in an arc’) b. pɔpɔt- [pɔp.pɔt ̚] ‘be circular’ (pɔt ‘circle’)

3 An anonymous reviewer suggests that the restricted nature of the reduplicant coda may be due to its position in an unstressed syllable of an iambic foot; however, occurrence in an unstressed position does not result in coda restrictions for roots in e.g. verb compounds, which behave as single phonological words with respect to stress patterns. 4 I provide a simplex form wherever it is possible to do so (i.e. except in cases of inherent reduplication).

reduplication in hup (northwest amazonia)

c. yoyo- [joj.jo]

147

‘swing while suspended from above’ ( yo- ‘dangle,’ esp. from hand)

Finally, where what appears to be a reduplicated root ends in a glottal stop, this stop appears in the coda of the first repeated syllable (example 3). However, examples in my corpus involving a final glottal stop are limited to cases of ‘inherent’ reduplication, where no simplex form exists, so are not clearly reduplication according to the definition used in this volume. Nevertheless, this is the only context in which reduplication or related phenomena in Hup (as opposed to other forms of repetition) can be understood as involving a full copy of the base. This presence of the glottal stop in this environment is consistent with the cross-linguistic observation that glottal stops pattern more freely than other consonants (Macaulay and Salmons 1995). (3)

a. peʔpeʔ- [peʔpeʔ] ‘grope, pat’ b. wɨʔwɨʔ- [wɨʔwɨʔ] ‘tremble’

It should also be noted that some variation is found in stems exhibiting inherent reduplication. A few lexicalized forms do not exhibit the homorganic/geminate consonant phenomenon, and thus appear more like true monomorphic roots (even in slow speech; e.g. nunút ‘moth’). These variants suggest that inherently reduplicated words in Hup—which are by definition lexicalized—may exhibit different degrees of lexicalization. On the other hand, complex medial consonants are not necessarily copied traces of root-final consonants, as in lexical items like bobó [mbop.bó] ‘bird sp.’ (alternative pronunciation [mbobó]), and in words formed via the combination of a root plus a vowel-initial suffix (e.g. /tǒk-ót/ [tǒk.kót ̚] (belly-obl) ‘in the belly’; see Epps 2008a for further discussion). These variations are nonetheless consistent with the generalization that CVC is the preferred or default syllable structure in Hup, and that reduplicated forms in Hup maintain this more general preference.5 Reduplicated forms (including those that are inherently reduplicated) and (semi-) lexicalized former compounds form a phonologically distinct lexical class in Hup, which is intermediate between monomorphemic and multimorphemic words. As noted above, most Hup morphemes are monosyllabic, bimoraic, and consonant-initial; in morphemes with two syllables (the maxi-

5 Because of these variations, I have chosen not to represent the consonantal coda of reduplicated syllables orthographically, aside from the glottal stop, but note that true (productively) reduplicated forms should nevertheless be understood to contain a CVC copy.

148

epps

mum, with a very few exceptions), the medial consonant is usually weak, /h/ or /w/ (e.g. mɔhɔ̌y ‘deer’). Multimorphemic forms, on the other hand, typically involve adjacent or fully geminated consonants across morpheme boundaries (e.g. ʔǝg-naʔ- (drink-die) ‘be drunk’; tóg-ót [tôgŋ.ŋgót ̚] (daughter-obl) ‘with daughter’). Reduplicated forms and lexicalized former compounds, whose synchronic status is neither clearly multimorphemic nor monomorphemic, reflect this intermediate status phonologically in that they are the only forms in which medial homorganic consonant clusters (as in 1 above) occur. These medial clusters are effectively intermediate between both the single (weak) consonant typically found in monomorphemic words and the pair of distinct and fully specified consonants encountered across morpheme boundaries. Accordingly, the phonological status of reduplicated (and other lexicalized) forms in Hup suggests that phonological wordhood in this language is best understood in terms of a gradient or scalar template, rather than as a fixed set of properties (compare Ito and Mester 1995 for Japanese; Schiering et al. 2010; see Epps 2008a, 102).

4

Functional Aspects of Hup Reduplication

Reduplication in Hup occurs primarily with verbs. However, reduplicative forms also include at least a few nouns and certain attributive modifiers (occurring within a small class of semi-lexicalized compounds). Ideophones also exhibit repetition phenomena that bear some resemblance to reduplication. Despite variations in functions and productivity across word classes, reduplication is in all cases iconic, such that the complexity of the linguistic sign corresponds to a relative complexity in semantic content (Haiman 1985; Dressler 1985). In Hup, this semantic complexity reflected in reduplication generally involves repetition or durativity. 4.1 Reduplication in Verbs Reduplication in verbs is semi-productive in Hup and tends to contribute a broadly aspectual function relating to repetition or intensity. However, in many cases reduplication results in specific, non-transparent lexical meanings. In still other verb roots, no corresponding simplex form can be identified; some of these roots may in fact be onomatopoeic, while others probably were formed historically via reduplication and later underwent lexicalization and loss of the simplex form from the lexicon. The examples in (4–6) illustrate the aspectual association of reduplication with iterativity and/or a rapid succession of events (see Epps 2008a, 579–580 for further examples).

reduplication in hup (northwest amazonia)

(4)

149

a. wawat- ‘walk around, back and forth in village or field’ wat‘pass through, visit in village’ b. tãtãw- ‘hit with a stick repeatedly (esp. with multiple light repetitions)’ tãw‘hit with stick’

(5) a. yo~yo-g'ãʔ-ãy=mah yup mǐh-ĩh dangle~dangle-hang-dynm=rep that.itg tortoise-decl ‘That tortoise swung about, hanging (by his teeth).’ b. máj-ãt tɨh yo-g'ãʔ-y󰀟́ʔ-ní-b'ay-áh, basket-obl 3sg dangle-hang-tel-be-again-decl yúp tɨh=tæ̃ h=mæh-ǎn-ãh that.itg 3sg=offspring=dim-obj-decl ‘He hung (it) up again in the basket, that little child of hers.’ (6) a. d'ǔç hɨd tǝ~tǝd-d'óʔ-óy=mah timbó 3pl beat.timbó~beat.timbó-take-dynm=rep ‘They beat timbó (fish-poison vine) (rapidly).’ b. d'ǔç tɨh tǝ́d-ǝ́h, yúp d'ǔç tɨh timbó 3sg beat.timbó-decl that timbó 3sg tǝd-g'ét-ét=mah… beat.timbó-stand-obl=rep ‘He beat timbó (in a measured, forceful way) …’ Compare the following inherently reduplicated forms, where no simplex form is attested, but the semantics also relate to iterative action: (7)

a. b. c. d.

nɔnɔypapadtætæpwɨʔwɨʔ-

‘swing back and forth’ ‘moan continuously in pain’ ‘shake body (dog or animal)’ ‘tremble’

In stative verbal expressions, reduplication may be associated with many simultaneous realizations of a characteristic, rather than with repetition of an event over time (see Epps 2008a, 581); see also the examples in (1) above:

150 (8)

epps

d'id'ib'- ‘curly’ (multiple curls; e.g. curly hair) d'ib'‘curled’ (individual entity)

The reduplicated forms in (9–10) are semantically more heterogeneous but represent states or activities associated with intensity, continuity, or repetitive characteristics (see Epps 2008a, 582): (9)

a. wãwãw- ‘reeling’ wãw‘follow a winding path’ (e.g., a wire) b. hicoco- ‘be happy’ (hi- Factitive prefix) co‘rest’ c. wãwaŋ- ‘head lolling around’ waŋ‘extend neck’ d. kokotkot-

‘move or go in circles or a circle’ ‘go in an arc’

e. ʔɨʔɨdʔɨd-

‘mumble to oneself’ ‘speak’

(10) a. yúp=mah tɨh=tǒk that.itg=rep 3sg=belly hi~hit-b'uy-d'ǝh-ham-y󰀟́ʔ-ay-áh scratch~scratch-throw-send-go-tel-inch-decl ‘Then she gutted (the animal).’ (hihit- ‘cut/scratch multiple times’; ‘open an animal’s gut’) b. verdúra hæyhɔ́ hit-b'ah-wob-y󰀟́ʔ-󰀟́y vegetable middle scratch-split-place-tel-dynm ‘He cut the vegetable in two.’ (hit- ‘cut/scratch in a single line’) 4.2 Reduplication in Nouns In contrast to verbs, reduplication proper (where a simplex form can be identified) is virtually absent in Hup nouns. A limited set of frozen lexical forms involve ‘inherent’ reduplication (see Epps 2008a, 208–209 for a near-exhaustive list of such nouns in the Hup corpus), but only in one or two cases can a possible simplex root be suggested. Nevertheless, most inherently reduplicated nouns refer to entities characterized in some way by repetition or multiplicity,

reduplication in hup (northwest amazonia)

151

thus resembling the more productive aspectual function of the ‘true’ reduplication observed in verbs. The repetition encountered in Hup noun roots may be historically associated with the more productive reduplication that occurs in verbs, perhaps via analogy. Most of Hup’s inherently reduplicative nouns refer to small living creatures. In some cases, these may be associated with quick, repetitive movements (compare the iterative/repetitive aspectual function of verbal reduplication in (4– 5) above); in others, they are creatures that are commonly encountered in swarms or flocks (compare the verbal forms associated with multiple simultaneous repetitions in (1) and (8) above). The largest group of such nouns is that of insects (example 11), followed by small animals and fish (12), and birds (13): (11) a. b. c. d.

yɨy󰀟̌w b'eb'ěp nunút j'ɨj'󰀟̌b'

‘ant sp.’ ‘butterfly’ ‘moth’ ‘small fruitfly sp.’

(12) a. b. c. d. e.

b'ib'ǐb' kukúy b'ab'ǎw bǝbǝ̌d wowǒd

‘small squirrel (generic)’ ‘night monkey’ ‘snake sp.’ ‘toad sp.’ ‘mandi type (fish sp.)’

(13) a. bebé ‘small bird sp.’ b. totób' ‘black-tailed trogon (bird)’ c. mæmǽç ‘grey-winged trumpeter (bird)’ In addition to animals, inherently reduplicated nominal roots include a handful of other entities that are characterized by punctuated repetition or duration (example 14). These are musical instruments, waves, and forest clearings. The latter term may capture the idea that a clearing is a multiplicity of the ‘holes’ formed in the forest by felling individual trees. (14) a. hehěh ‘pan-flute (instrument), its music, and/or accompanying dance’ b. wowó ‘mawaco’ (small tube-shaped whistle held vertically) c. ʔaʔáb' ‘wave’ d. hohód ‘clearing in forest’ (from hǒd ‘hole’?)

152

epps

Finally, repetitive noun roots in Hup include a few terms for plants and body parts. While a few of these entities, such as ‘fish spine,’ are like those listed above in having distinguishing repetitive characteristics, this is not clearly the case for others. Some of these plant and body part names may derive from the attributive function of reduplication (see 4.3 below), involving a reduplicated form plus a nominal element, which may have been subsequently dropped. (15) a. b'ab'áʔ b. bǝbǝ́g

‘imbaúba’ (Cecropia sp., a tree) ‘cubiu’ (Solanum sesseliflorum, a plant bearing multiple orange fruits) c. wɔ̃wɔ̌ m núh ‘broad-leafed epiphyte sp.’ (possibly from wɔ̌ m ‘squirrel’ + núh ‘head’) d. yæyæ̌ w t󰀟́t ‘vine sp.’ (tɨt ‘vine’)

(16) a. b. c. d.

(tɨh=)cɨc󰀟́n g'ag'ǎw hahád pãpáŋ

‘fish spine’ ‘lymph nodes’ ‘underarm’ ‘hip’

4.3 Reduplication in Attributive Modifiers within Compounds From a typological perspective, the most unusual function of reduplication in Hup is to mark attributive modifiers in a small set of lexicalized nominal compounds. In these compounds, a noun is followed by a reduplicated adjective or stative verb root. Reduplication distinguishes the lexicalized compound from other, more productive combinations of a noun + modifier. The examples below give a comprehensive list of these reduplicative compounds from my corpus (see also Epps 2008a, 221–222); as in the case of nominal roots, reduplication in these compounds appears to be only marginally or historically productive. Reduplicated modifiers are not found in any other contexts in Hup. (17) a. cob tæ̃ ~tæ̃ ́ h b. c. d. e.

(finger small~small) ‘pinky finger’ compare cob tæ̃ h ‘(any) small finger’ cob po~pǒg (finger big~big) ‘thumb’ compare cob pǒg ‘(any) big finger’ nuh to~tǒy' (head support~support) ‘neck’ kãkãy' j'ɨ~j'󰀟̌g (fish sp. sharp~sharp) ‘mandi (fish) sp.’ tõj ya~yǎg (fish.sp. spotty~spotty) ‘jacundá (fish) sp.’

The list of nominal expressions in (18) appears to involve a similar process, but no reduplicated root can be identified:

153

reduplication in hup (northwest amazonia)

(18) a. b. c. d. e. f. g.

nuh yǝyǝ́g j'ib kãkãw yɔʔ j'æ̃ j'æ̌ ̃ p yãʔam wãwã ̌t j'ɨb'ɨh j'ũj'ũ ̌y' j'ɨb'ɨh kǝkǝ̌c cǔg pũpũ ́=mæh

(head?) (foot?) (wasp?) (jaguar?) (bat?) (bat?) (hummingbird?=dim)

‘upper neck in back’ ‘ankle bone’ ‘wasp sp.’ ‘bush dog’ ‘sheath-tailed bat’ ‘fruit bat sp.’ ‘hummingbird sp.’

While this ‘attributive’ function of reduplication in nominal compounds appears at first glance to be distinct from the uses of reduplication elsewhere in Hup, in fact it probably conforms to similar iconic principles. In these nominal compounds, the modifier is understood as an inherent quality of the referent, something absolute and therefore durative. In contrast, modifiers in productive combination with nouns are more likely to be understood as temporary or relative qualities. For example, the smallness of a person’s finger can only be gauged in relation to the set of fingers it is compared with (those of babies, children, or adults), whereas a ‘pinky’ finger is always small in relation to the other fingers on the same hand. Similarly, whether an individual fish is judged to be spotty may vary in comparison to other individual fish, whereas the characteristics of types of fish are more easily defined. The relatively fixed, durative quality signaled by reduplication in these compounds is much like the iterative or durative characteristics indicated by reduplication in other Hup word classes.6 While this particular use of reduplication found in Hup compounds appears to be cross-linguistically unusual,7 it likewise has functional parallels in other languages. In particular, the use of reduplication to signal the durativity of an attribute is found in languages like Alabama (Hardy and Montler 1988; Rubino 2005, 29), where reduplication of adjectives distinguishes a temporary (non-reduplicated) from a more permanent (reduplicated) state. In Yanomae (Goodwin Gómez, this volume), the reduplication of a verb or adjective derives a noun, which can be understood as a durable or permanent instantiation of the property in question; e.g. noma~noma a [die~die sg] ‘death’; here~here pǝ [wet~wet pl] ‘humidity.’ Other derivational functions of reduplication are attested in a variety of languages (Rubino 2005, 21; see also Moravcsik 1978, 325). 6 The lexicalized nature of these compounds may also contribute to the more fixed, durative interpretation—but it is important to note that most occurrences of simplex attributive modifiers in Hup do not involve lexicalization, whereas all occurrences of reduplicated modifiers do. 7 I have found no mention of it in the literature on reduplication, such as Rubino (2005).

154

epps

Hup’s use of reduplication on attributive modifiers in lexicalized compounds—which appears on one hand to be so unusual, but on the other is consistent with uses of reduplication both within Hup and in other languages— reminds us that an enormous amount of semantic and functional variation exists across reduplication phenomena (see Stolz 2007). Even where this variation fits into broad generalizations, a great deal of rich complexity still awaits exploration. 4.4 Ideophones Hup ideophones constitute a distinct word class of sound symbolic expressions that tend to exhibit highly repetitive forms, as is typical of ideophones crosslinguistically (e.g. Voeltz and Kilian-Hatz 2001; Nelson 2005). Hup ideophones commonly involve a multi-syllabic string made up of one repeated syllable, or in some cases a bisyllabic form; they may also be composed of a single prolonged vowel or sonorant segment, or a single syllable (see Epps 2008a, 721– 724). While the repetition found in Hup ideophones is distinct from reduplication proper,8 I consider it here because it shares certain characteristics with reduplication. Like reduplication, repetition in Hup ideophones occurs within the domain of the word and is an iconic representation of repetitive or durative motions, sounds, or states. However, it differs crucially in that no simplex counterpart to the repeated form appears to have ever existed. Furthermore, repetition in Hup ideophones differs formally from the reduplication found in other word classes: the repeated syllable need not be word-initial (as in 19); it may involve a full copy of the base, in which the coda consonant of the reduplicated syllable is fully expressed (as in 20); and the number of repetitions of the base form varies across tokens and speakers, with up to five repetitions commonly encountered. (19) wídídídí, pótʔah-an, kǝk-d'ǝh-hám-ãp, wídídídí ideo upriver-dir pull-send-go-dep ideo ‘(Going-up noise), upriver, (he) pulled them …’ (20) tɨh tɔʔɔh-wɔn-næn-y󰀟́ʔ-ay-áh, túktúktúktúktúk! 3sg run-follow-come-tel-inch-decl ideo ‘She came running after him, (sound of running)!’

8 See the criteria defined by the Graz Reduplication Project, http://reduplication.uni-graz.at.

reduplication in hup (northwest amazonia)

5

155

Reduplication and Repetition

Reduplication is normally understood as a phenomenon distinct from repetition. While reduplication applies within words, such that the meaning of the resulting word is different from that of constituent parts, repetition applies across morphemes, words, or phrases and repeats semantic content (see e.g. Gil 2005, 33). Reduplication is marginal or absent from some languages, but repetition is probably universal; it is among the “playful and esthetic” functions of language (Sherzer 1995). Nevertheless, the formal and functional similarities between the two phenomena suggest that a diachronic relationship may exist between them, at least in certain cases. This is supported by evidence from particular languages, such as Riau Indonesian (Gil 2005; Hurch 2005, 1), although few studies have addressed the issue directly. Because a diachronic link presupposes a transitional stage between repetition and reduplication, evidence to support it would include synchronically intermediate structures. As I argue below, we find in Hup a clear set of such formally and functionally intermediate examples. With respect to function, the repetition of verbs and adverbs in Hup resembles the productively iterative use of reduplicated verbs described above. This is illustrated by the repeated adverb ‘inside’ in (21), yielding the meaning ‘one inside the other’: (21) yúp ʔæ̌ y-g'od ʔæ̌ y-g'od ʔæ̌ y-g'od that.itg together-inside together-inside together-inside ʔæ̌ y-g'od=mah, kotów=teg, tɨh cud-cák-áh together-inside=rep dance.staff=stick 3sg be.inside-climb-decl ‘One inside the other, they had stacked their dance-staffs.’ The formal similarities between reduplication and repetition are particularly striking. While an inflectional suffix is normally required on all verbal predicates, repeated verbs and verb phrases in Hup may occur as bare stems—the only context in which they may appear uninflected outside of verbal compounds.9 However, they must be followed by the ‘summarizing’ inflected verb ni- (‘be, exist,’ copula), as in (22). The construction thus has the status of a

9 Imperative verb forms in Hup may also appear as bare verb stems but require the addition of a particular tone value.

156

epps

single finite predicate, much like a serial verb construction—though without the verb-root compounding that characterizes Hup’s serial verbs in all other contexts (see Epps 2008a, 824–825). (22) h󰀟́d-ǎn wæd-nɔ̌ ʔ, h󰀟́d-ǎn wæd-nɔ̌ ʔ, 3pl-obj eat-give 3pl-obj eat-give ni-yóʔ p󰀟́d=mah tɨh way-yɨʔ-p󰀟́d-󰀟́h be-seq dist=rep 3sg go.out-tel-dist-decl ‘(He) would give them food, (always) give them food, having done thus he would go out again, it’s said.’ Where the repeated verb is a single root (as in 23), the verb + ni construction arguably yields a single grammatical word. Moreover, primary stress within the unit falls on the final root (ni), thus conforming to the pattern found in verb compounds generally in Hup, which are considered to be single phonological words. (23) nút d'oʔ-cud-yóʔ, núp pɔ̌ t bɨʔ-yóʔ, j'󰀟̌p̃ j'󰀟̌p̃ j'󰀟̌p̃ here take-be.inside-seq this circle make-seq tie tie tie ní-ĩy hɨd d'ǝh-d'ǝh-hám-b'ay-áh10 be-dynm 3pl send-send-go-again-decl ‘Having put (the cord) in here, having made this (string) circle, with a wrap-wrap-wrap they would send off (the toy top).’ Constructions like that in (23), in particular, appear to represent a middle ground between straightforward repetition and reduplication. Both phonologically and morphosyntactically, they are at the same time less tightly integrated than is required by reduplication proper in this language, and more integrated than would be expected in the simple repetition of words or phrases. Diachronically, it is plausible that constructions like that in (23) were reduced through frequent use to yield reduplicative verb forms like those discussed in section 4.1 above.

10

The repetition of the verb root d'ǝh- in this example is probably not a case of reduplication, but rather a reapplication of this root motivated by the partial lexicalization of the compound d'ǝh-ham-.

reduplication in hup (northwest amazonia)

6

157

Conclusion

Reduplication in Hup is consistent with cross-linguistically attested patterns, particularly with respect to the functional iconicity common across its various realizations. However, it also has a number of intriguing characteristics that can inform our understanding of reduplication and associated phenomena more broadly. With respect to form, the behavior of word-medial consonants in reduplicated Hup words suggests that they have a phonologically intermediate status between mono- and multi-morphemic words, thus providing support for a gradient notion of phonological wordhood in this language. With respect to function, Hup’s use of reduplication on attributive adjectives in lexicalized compounds is typologically unusual, but nevertheless conforms to familiar iconic principles associating reduplication with durativity. Finally, the fact that certain Hup structures share properties of both repetition and reduplication provides evidence for a synchronic middle ground, which in turn may derive from a diachronic link between these two phenomena. In sum, Hup—like many other understudied languages of Amazonia—contributes a rich set of insights to shape a more fine-grained understanding of linguistic phenomena like reduplication.

Acknowledgments Information on Hup (Hupda, Jupde) was obtained via fieldwork on the Rio Tiquié, Brazil, 2000–2004. I am grateful to my Hup hosts, friends, and language teachers, to the Instituto Socioambiental, FOIRN, and the Museu Goeldi in Brazil, and to Fulbright-Hays, the National Science Foundation (BCS0111550), and the Max Planck Institute for Evolutionary Anthropology, Leipzig for supporting this research. Thanks also go to the organizers and participants in the symposium ‘Reduplication in Amazonian Languages’ (53rd International Congress of Americanists, Mexico City, July 2009) for their helpful comments, and to Hein van der Voort, Gale Goodwin Gómez, and an anonymous reviewer for further suggestions on this paper. This chapter builds on and refines the view of Hup reduplication provided in Epps (2008a).

References Aikhenvald, Alexandra. 2002. Language contact in Amazonia. Oxford/New York: Oxford University Press.

158

epps

Dressler, Wolfgang. 1985. Morphonology: The dynamics of derivation. Ann Arbor: Karoma. Epps, Patience. 2007. The Vaupés melting pot: Tucanoan influence on Hup. In Grammars in contact: A cross-linguistic typology, ed. Alexandra Aikhenvald and R.M.W. Dixon, 267–289. Oxford: Oxford University Press. . 2008a. A grammar of Hup. [Mouton Grammar Library 43]. Berlin/New York: Mouton de Gruyter. . 2008b. Grammatical borrowing in Hup. In Grammatical borrowing: A crosslinguistic survey, ed. Yaron Matras and Jeanette Sakel, 551–565. Berlin/New York: Mouton de Gruyter. Gil, David. 2005. From repetition to reduplication in Riau Indonesian. In Hurch (ed.), 31–64. Gomez-Imbert, Elsa. 1996. When animals become ‘rounded’ and ‘feminine’: Conceptual categories and linguistic classification in a multilingual setting. In Rethinking linguistic relativity, ed. John Gumperz and Stephen Levinson, 438–469. Cambridge: Cambridge University Press. Goodwin Gómez, Gale. this volume. Reduplication in the Yanomae language of northern Brazil. Haiman, John. 1985. Iconicity in syntax. Amsterdam/Philadelphia: John Benjamins. Hardy, Heather K., and Timothy Montler. 1988. Imperfective gemination in Alabama. International Journal of American Linguistics 54: 399–475 Hurch, Bernhard. 2005. Introduction. In Hurch (ed.), 1–10. Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology, 28]. Berlin/New York: Mouton de Gruyter. Ito, Junko, and Armin Mester. 1995. The core-periphery structure of the lexicon and constraints on reranking. In University of Massachusetts occasional papers in linguistics 18: Papers in optimality theory, ed. Jill Beckman, Laura Walsh Dickey, and Suzanne Urbanczyk, 181–210. Amherst: University of Massachusetts. Jackson, Jean. 1983. The fish people: Linguistic exogamy and Tukanoan identity in northwest Amazonia. Cambridge: Cambridge University Press. Macaulay, Monica, and Joseph C. Salmons. 1995. The phonology of glottalization in Mixtec. International Journal of American Linguistics 61: 38–61. Moravcsik, Edith A. 1978. Reduplicative constructions. In Universals of human language, ed. Joseph H. Greenberg, Vol. 3, 297–334. Stanford: Stanford University Press. Nelson, Nicole. 2005. Wrong side reduplication is epiphenomenal: Evidence from Yoruba. In Hurch (ed.), 135–160. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Schiering, Rene, Balthasar Bickel, and Kristine Hildebrandt. 2010. The prosodic word is not universal, but emergent. Journal of Linguistics 46: 657–709.

reduplication in hup (northwest amazonia)

159

Sherzer, Joel. 1995. Reduplication in four languages: Where formal grammar and speech play/verbal art meet. In Salsa III (Proceedings of the Third Annual Symposium about Language and Society—Austin), ed. Risako Ide, Rebecca Parker, and Yukako Sunaoshi, 257–265. Austin: University of Texas, Department of Linguistics. Sorensen, Arthur P. 1967. Multilingualism in the northwest Amazon. American Anthropologist 69: 670–684. Stolz, Thomas. 2007. Re: duplication. Iconic vs counter-iconic principles (and their areal correlates). In Europe and the Mediterranean as linguistic areas. Convergencies from a historical and typological perspective, ed. Paolo Ramat and Elisa Roma, 317– 350. Amsterdam/Philadelphia: John Benjamins. Voeltz, F.K. Erhard, and Christa Kilian-Hatz. 2001. Introduction. In Ideophones, ed. F.K. Erhard Voeltz and Christa Kilian-Hatz, 1–8. Amsterdam/Philadelphia: John Benjamins.

Abbreviations decl dep dir dist dynm ideo inch itg

Declarative Dependent Directional Distributive Dynamic Ideophone Inchoative Intangible (demonstrative)

obj obl pl red rep seq sg tel

Object Oblique Plural Reduplication Reported evidential Sequential Singular Telic

chapter 7

Reduplication in the Yanomae Language of Northern Brazil Gale Goodwin Gómez

The morphological process of reduplication is common in Yanomae, a member of the Yanomami language family. It has both semantic and grammatical functions. As is typical of this process in languages throughout the world, reduplication in Yanomae indicates some aspect of what Stolz calls “semantic augmentation” (2007, 320), which is the intensification of the meaning of a particular object, action or state. Specifically, the result of the reduplication of a noun or verb stem in Yanomae might indicate iterativity or increased quantity, intensity, or duration of a substance, object, action, movement, position or state. Reduplication in Yanomae also has a category-changing function, whereby nouns are derived from verb stems. This paper provides examples of both full and partial reduplication, although partial reduplication is much less common.

1

Introduction

This paper describes reduplication in Yanomae, one of four major linguistic subgroups that comprise the Yanomami language family; the other subgroups are Yanomamɨ, Sanumá, and Ninam (also called Yanam). This paper maintains the original categorization of the subgroups made by Ernesto Migliazza (1972). A somewhat different division of the subgroups was proposed by Henri Ramirez (1994) in his doctoral dissertation, and a description of Ramirez’s view of the languages in the Yanomami family can be found in Goodwin Gómez (2009b). An overview of the occurrence of reduplication in the languages of the Yanomami family was presented at the Graz Reduplication Conference in 2007 (Goodwin Gómez, 2009a). The present paper expands the description of one language, Yanomae, and focuses on the dialect spoken in the community of Watorikɨ located in the state of Amazonas in northern Brazil.1 Yanomae is the

1 The linguistic data in the present study were collected during personal fieldwork by the author and in collaboration with Bruce Albert. Specific examples are from the manuscript

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_008

162

goodwin gómez

language spoken by the greatest number of Yanomami living in Brazil, approximately 9,123 speakers or about 47.5% of the total 19,000 Brazilian Yanomami, according to the 2011 Brazilian National Health Foundation (FUNASA) census. A complete description of Yanomae has yet to be written, and many aspects of the language are still poorly understood.2 Like other Yanomami languages, Yanomae has SOV word order and a highly complex verbal morphology that is characterized by one or more verb roots and numerous suffixes. Verbs, nouns, pronouns, and a large inventory of particles (including most notably nominal classifiers, quantifiers, and postpositions) form the basic word classes; adjectival concepts are expressed by stative verbs. Nouns and verbs are distinguished by their ability to occur with specific affixes and particles. Nouns occur with nominal classifiers, quantifiers, postpositions, and the ergative suffix, while verbs occur with verbal suffixes, such as tense and aspect. Generally, nouns do not occur as bare stems but are usually accompanied by additional nominal morphology. Consequently, the data do not show that reduplication alone can create nouns out of verbs, but there is nevertheless convincing evidence that this is, in fact, the case since verbs nominalized by reduplication do not behave any differently from genuine nouns. Similarly, the category of verbs (except for stative verbs) is defined by the (verbal) suffixes that accompany verb stems. The ergative case is marked by the suffix -nɨ on agents and instruments, and there is no grammatical gender. Yanomae has thirteen consonants: /p/, /t/, /k/, /th/, /h/, /hw/, /s/, /ʃ/, /r/, /m/, /n/, /w/ and /j/and seven vowels: /a/, /e/, /i/, /o/, /u/, /ə/, and /ɨ/. Throughout the paper standard IPA symbols are used, with two exceptions: the retention of the symbol /x/, which is common usage in Brazilian linguistics, for the voiceless (alveo) palatal fricative /ʃ/, and /y/ for the palatal glide /j/. The basic syllable structure is CV. Stress is not significant. The structure of a “word” is not easily defined in Yanomami languages, but the present paper will consider a root with its accompanying derivational and inflectional elements as constituting a word (Goodwin Gómez 2009a, 2).

of a thematic lexicon of Yanomae that is in preparation (Albert and Goodwin Gómez, n.d.), and the botanical identification of specific plants derives from collaboration with William Milliken (Milliken and Albert with Goodwin Gómez 1999; Albert and Milliken with Goodwin Gómez 2010). 2 Helder Perri Ferreira has written a master’s thesis (2009) on the nominal classifier system of Yanomama, a closely related dialect spoken in Papiu (Brazil), and he is currently in doctoral studies at Radboud University (Nijmegen), where he is expected to complete a description of that dialect of the language.

reduplication in the yanomae language of northern brazil

2

163

Formal Aspects of Yanomae Reduplication

The main focus of this paper is what is generally accepted as “true” reduplication, which is defined on the Graz Reduplication Project website (http:// reduplication.uni-graz.at/). According to this definition reduplication is a morphological process that involves copying, partially or in full, a simplex form (or base) for the purpose of creating a new lexeme that has a different meaning or that serves a grammatical function. Also found in Yanomae is a large number of forms that appear to have reduplicative structure but cannot be identified with a simplex form. Two terms that are commonly used to refer to such forms are inherent reduplication and lexical reduplication. In Yanomae such lexemes are especially common as ideophones and animal and plant names of ideophonic or onomatopoeic origin. Kilian-Hatz defines an ideophone as a “vivid representation of an idea in sound. A word, often onomatopoeic, which describes a predicate, qualificative or adverb in respect to manner, colour, sound, smell, action, state or intensity” (2006, 509). Inherently reduplicated lexemes lack simplex forms and distinct semantic or grammatical functions, and lexical reduplication is defined in Hurch and Mattes as “lexemes with reduplicative structure which do not (in a synchronically transparent way) correspond to a simplex form” (2009, 316, footnote 16). While these forms may contain repeated syllables, their status as examples of true reduplication is unclear. Thus, these phenomena will be described at the end of the paper in section 5. Yanomae has both full and partial reduplication, although full forms predominate. Full forms generally consist of disyllabic verbal roots, although an occasional verbal stem, composed of a monosyllabic root plus a suffix, may serve as the simplex form for the reduplicant copy. The process of reduplication in Yanomae has two functions: a category-changing function, whereby nouns are derived from verbs, and an iconic function, whereby “reduplication is used to express quantitative or qualitative augmentations” (Hurch and Mattes 2009, 304) and which Stolz generalizes as “semantic augmentation” (2007, 320). Iconic reduplication in Yanomae represents the intensification of the meaning of a word denoting a substance, object, action, movement, position or state. Specifically, the result of reduplication might indicate iterativity or an increase in quantity, intensity, or duration. Nouns in Yanomae are always accompanied by a classifier, a quantifier (i.e. sg, pl) or are inalienably possessed or specified in some way (i.e. (11a) wakǝ proke ‘fire empty’ = ‘(it’s an) empty fire’). Likewise, verbs (except for stative verbs) are defined in terms of the suffixes that accompany them. Examples such as (11a) show both noun and verb without any suffixes or clitics, but this type appears to be the only case when this occurs—where the bare noun is specified

164

goodwin gómez

by a bare (stative) verb. The nominal status of reduplicated roots is confirmed by their co-occurrence with clitics and suffixes associated with nouns. Clitics that accompany nouns include classifiers, directionals, locatives, and quantifiers (sg, pl, agg). For example, the derived noun herehere ‘wetness’ in (1) co-occurs with the plural pə, to create herehere pə ‘humidity’ and the aggregate/collective kɨkɨ accompanies naxinaxi ‘sourness’ in (2) to produce naxinaxi kɨkɨ ‘honeycomb.’ Example (3a) shows a stative verb oxe ‘young’ followed by the indefinite pronoun thə and the plural pə to produce the nominal phrase oxe thə pə ‘youngsters’ while in (3b) the same stative verb is reduplicated as oxeoxe and accompanied by the locative ha to describe a geographical location, ‘in a low growth forest.’ The process of reduplication may be used to express very different, often metaphorical, meanings from the literal meaning associated with the simplex form. (1)

here~here=pə wet~wet=pl ‘humidity [i.e. from rain]’

(2) naxi~naxi=kɨkɨ sour~sour=agg ‘honeycomb (of) sour (tasting honey)’ (3) a. oxe=thə=pə young=indef=pl ‘youngsters, children’ (lit. ‘young ones’) b. oxe~oxe=ha young~young=loc ‘in a low growth forest’ Furthermore, the nominal status of certain reduplicated roots, such as oxeoxe in example (3c), is supported by the ability of this new form to take the verbalizing suffix -mu, which creates intransitive verbs from nouns, such as the kinship term hwĩĩ ‘father’ in example (4). Like other nouns in Yanomae, nominals derived through the process of reduplication (3c) can be “de-nominalized” by the addition of the verbalizing suffix -mu. (3) c. ya=oxe~oxe-mu-xoa I=young~young-vbzr-still ‘I still walk with a limp.’ (lit. ‘I still [move] as a youngster.’)

reduplication in the yanomae language of northern brazil

(4)

165

ya=e=hwĩĩ-mu I=poss=father-vbzr ‘I treat him as my father.’

The instrumental suffix -nɨ is another indicator of nominal status, and, as such, it may -co-occur with a reduplicated stem as in example (5), where a quantity of “muddiness” is thrown on a child. (5) ãhĩ~ãhĩ=pə-nɨ3 oxe=ya=thə=pahu-pra-re-ma muddy~muddy=pl-instr young=I=indef=throw.on-punc-tel-past ‘I threw mud on a child.’ Their co-occurrence with nominal suffixes as well as with clitics is further evidence that reduplicated stems function fully as nouns. The preceding examples (1–3, 5) show full reduplication; however, partial reduplication can be seen in Yanomae, although it is much less frequent. It is not obvious and maybe not important which form in instances of full reduplication is the original and which is the copy. This is also unclear in some cases of partial reduplication where the structure of the root consists of two identical syllables; however, example (6) provides a transparent case that suggests an analysis in which the reduplicant is the last CV syllable of the disyllabic root. Described in greater detail in section 4 on iconic reduplication, partial reduplication conveys an iconic meaning of iterativity in the sense of repeated motion or repeated position, as in example (6). (6) hwetɨ~tɨ-a-ɨ symmetrical~red-theme-dyn ‘(be) one after the other (in a symmetical position)’

3

Nominalizing Reduplication

The most commonly occurring function of reduplication in Yanomae is word class derivation of nouns from verbs. In most cases the corresponding forms involve full reduplication whereby a verb root or stem is copied immediately adjacent to the original form. It is not clear which of the two forms is the copy,

3 The ergative (erg) and instrumental (instr) suffixes are homophones, but it is not clear whether they are, or were historically, the same morpheme.

166

goodwin gómez

and the precise location (to the right or left of the base) does not appear to have any particular relevance. However, the derivation of nouns from verbs may also involve partial reduplication, and this is illustrated in section 4.2 with the transitive verb roro ‘appear’ in examples (30b) and (30c). Most roots that undergo reduplication have two syllables so a suffix or thematic vowel may be added to a monosyllabic base before it undergoes reduplication. The resulting reduplicated form may be accompanied by a nominal suffix or clitic, and this cooccurrence is, in fact, an important indicator in distinguishing between nouns and verbs. Ramirez (1999) distinguishes six classes of verbs in Yanomae: stative, intransitive, positional, semelfactive, consonantal and transitive. The verb classes are defined by semantic criteria as well as specific suffixes which may or may not occur with verb stems of that class. Stative, positional and intransitive verbs were all found to undergo nominalizing reduplication, although examples with stative verbs were the most common. Examples of iconic reduplication appear to include verbs from all classes. 3.1 Reduplication of Stative (Adjectival) Verbs Stative verbs in Yanomae are distinguished from intransitive verbs by the fact that they may occur alone, without any suffix, when they express a current state. They may occur with the dynamic suffix -ɨ only when the meaning expressed is a “process” rather than a state. Normal intransitive verbs always require a verbal suffix of some type. Stative verbs are sometimes referred to as adjectival verbs because they express the meanings attributed to adjectives in other languages (such as English). Furthermore, because they may occur as simple roots (i.e. without any suffix), stative verbs resemble adjectives, as in sentence (7). (7) thiwǝ totihi woman good/beautiful ‘(She’s a/there’s a) good/beautiful woman.’ The simplex form kõi ‘(be) hairy’ is a stative verb root, whether or not the phrase kasikɨ kõi (8a) is analyzed as a noun compound or as a simple sentence. When this verb root is reduplicated, however, it becomes a noun (8b) as indicated by its co-occurrence with the nominal classifer pəka ‘orifice.’ The resulting noun kõi~kõi ‘hair’ corresponds to other body part nouns that may occur with the same classifier pəka ‘orifice,’ as in examples (9a) and (9b).

reduplication in the yanomae language of northern brazil

167

(8) a. kasi-kɨ=kõi lip-pl=hairy ‘[It’s a] moustache.’ (lit. ‘Lips [are] hairy.’) (8) b. kõi~kõi=pəka hairy~hairy=cl.orifice ‘[It’s a] hair follicle.’ (9) a. hũka=pəka nose=cl.orifice ‘[It’s a] nostril.’ b. yǝmaka=pəka ear=cl.orifice ‘[It’s an] auditory canal or earlobe piercing.’ Certain stative verb roots that are transformed into nouns by reduplication may accept locative =ha and directional =hamɨ enclitics. Expressions describing places, such as a dirty place (10b), a clearing (11b) and (12b), and dry land (13b and 13c), commonly include reduplicated roots preceding these locative and directional enclitics. In examples where the locative =ha occurs, the directional =hamɨ could be substituted, given the proper (meaningful) context. (10) a. wa=hõxa-mo-rayu wa=xami-mahi you=rub.off-refl-tel you=dirty-very ‘Rub yourself off; you’re very dirty!’ b. xami~xami=ha dirty~dirty=loc ‘in a dirty place’ (11) a. wakə proke fire empty ‘[It’s] an empty fire [i.e. with nothing cooking].’ b. proke~proke=ha empty~empty=loc ‘in a [man-made] clearing’

168

goodwin gómez

(12) a. hutu=a=roxi garden=sg=cleared.deforested ‘cleared, deforested garden plot (intended or ready for planting)’ b. roxi~roxi=ha cleared.deforested~cleared.deforested=loc ‘in a clearing, deforested area’ (13) a. koraha ya=rõxi=wehe ha wa-rɨ-nɨ rõxi=hõko-ke-ma banana I=cl=dry seq eat-top-seq cl=get.stuck-asp-past ‘After I ate a cooked (lit. ‘dry’) banana, it got stuck (in my throat).’ b. wehe~wehe=ha dry~dry=loc ‘on dry land’ c. wehe~wehe=hamɨ dry~dry=dir ‘toward (in the direction of) dry land’ Likewise, color terms (also in the class of stative verbs) may be nominalized by reduplication, as shown in examples (14) and (15). The resulting words appear to be lexicalized since their meanings are not predictable from the meanings of the component morphemes even though the referents are characterized by (or, in other words, contain a substantial quantity or intensity of) the particular color. (14) uxi~uxi=kɨkɨ black~black=agg ‘black fungal rhizomorphs’ (15) kroke~kroke=a gray~gray=sg ‘a cloud’ 3.2 Reduplication of Intransitive and Positional Verbs Intransitive verbs in Yanomae have roots that must be accompanied by a suffix and, unlike positional verbs, may be directly attached to the dynamic suffix -ɨ. An intransitive verb describes an action or a process undergone by the subject or patient of the sentence. Like stative verbs, certain intransitive verb roots

reduplication in the yanomae language of northern brazil

169

undergo the same process of reduplication to produce corresponding nouns. For example, the intransitive verb, ĩxĩ- ‘(become) burnt,’ which is transitivized by the addition of the causative and dynamic suffixes in (16a), is transformed into a noun in (16b) by reduplication and accompanied by the nominal aggregate clitic kɨkɨ. (16) a. ĩxĩ-ma-ɨ burnt-caus-dyn ‘burn’ (lit. ‘cause to become burnt’) b. ĩxĩ~ĩxĩ=kɨkɨ burnt~burnt=agg ‘gunpowder’ As was the case with stative verb roots, in the process of reduplication a full copy of the intransitive verb root appears immediately adjacent to the base as in examples (17) and (18). Nominal clitics, such as the singular a in (17) and the locative ha in (18), accompany the derived noun. (17) noma~noma=a die~die=sg ‘death’ (18) hemo~hemo=ha ascend~ascend=loc ‘in an elevated place (on a hill or mountain)’ Sentence (19) presents a slightly different case, whereby the intransitive verb ke- ‘fall’ is reduplicated along with its associated thematic vowel -a, so that the form which is copied is the disyllabic stem ke-a. (19) hemeyo hura=ya-kɨ ha koa-rɨ-nɨ ya=pihi-kaẽ medicine malaria=I-pl seq drink-top-seq I=think-comit ke-a~ke-a-nɨ kiri-hi fall-theme~fall-theme-instr fearful-X ‘After having drunk malaria medicines, I was (overcome) with a frightening sensation of falling.’

170

goodwin gómez

Similarly, the roots of certain positional verbs, such as wətə ‘low’ in example (20a), may be reduplicated to form nouns and accept nominal clitics, as wətəwətə=ha ‘in a lowland area’ in example (20b). (20) a. he=wətə head=low ‘(with) head bowed’ b. wətə~wətə=ha low~low=loc ‘in a lowland area’ 3.3 Reduplication of Transitive Verbs Certain transitive verb roots may also undergo reduplication. In sentence (21) the transitive verb koni- ‘mix or stir’ occurs as both a verb root koni and a derived noun konikoni pə ‘mixture.’ Notice that the subject (agent) xapiri thə pə ‘some shamans’ of the transitive verb koaɨhe ‘they drink’ is marked for the ergative case -nɨ, thus, signaling the transitive nature of the sentence. (21) xapiri=thə=pə-nɨ koni~koni=pə koa-ɨ-he shaman=indef=pl-erg mix~mix=pl drink-dyn-3pl yuri=puu=u=pə=koni-i fish=juice=cl.liquid=pl=mix-dyn ‘Some shamans drink the mixture (with) fish broth mixed in.’

4

Iconic Reduplication

In addition to the grammatical function of deriving nouns from verbs, reduplication in Yanomae also serves an iconic function. In this case, a noun or verb stem may be reduplicated to indicate iterativity or increased quantity, intensity, or duration of an object, action, movement or state. The meanings expressed by iconic reduplication at the morphological level in Yanomae, as in other Yanomami languages (Goodwin Gómez 2009a), fall clearly within the broad cross-cultural categories described in a variety of studies on reduplication, such as Rubino (2005), Gil (2005), and Stolz (2007). The augmentative meaning contributed by the process of reduplication in Yanomae permeates the wide range of examples, whether the base is a noun or a verb, transitive or intransitive. Iconic reduplication is exemplified by both full and partial forms, although, as was the case in nominalizing reduplica-

reduplication in the yanomae language of northern brazil

171

tion, the reduplicant is more often a complete rather than a partial copy of the base. 4.1 Increased Quantity or Intensity Both noun and verb roots may undergo reduplication to express an increase in quantity or intensity of the meaning conveyed. Two sets of examples (22) and (23) show full reduplication of noun roots to express increased quantity. What may be understood in (22a) and (22b) as a solid bodily secretion, xãhe, may have its quantity increased through reduplication and be applied metaphorically to the contents of a river bottom in (22c) where ‘a layer of rotting leaves’ is interpreted as the ‘river’s secretions.’ The noun root in (23a), 󰀟̃rã ‘scaffolding,’ may be pluralized, but when this noun root is reduplicated and accompanied by the aggregate marker, the meaning ‘bridge’ reflects a greater quantity of the scaffolding in a specific arrangement. (22) a. hũkakɨ=xãhe nose=secretion ‘(dried) nasal secretions’ b. yəmaka=kɨ=xãhe ear=pl=secretion ‘earwax’ c. xãhe~xãhe=pə secretion~secretion=pl ‘layer of rotting leaves on a river bottom’ (23) a. 󰀟̃rã=kɨ scaffold=pl ‘wooden scaffold or raised platform’ b. 󰀟̃rã~󰀟̃rã=kɨkɨ scaffold~scaffold=agg ‘a wooden bridge’ (lit. ‘an aggregate of multiple scaffoldings’) An increase in intensity is illustrated by the reduplication of the stative verb root oxe ‘young’ in (24) and its co-occurrence with an intensifier, mahi. It is not difficult to see the semantic connection between the meaning of the simplex root ‘young’ and the reduplicated form ‘very weak.’ This is similar to the meaning conveyed in example (3c).

172

goodwin gómez

(24) ya=uku=oxe~oxe=mahi I=joints=young~young=very ‘My joints (in my legs) [are] very weak.’ (lit. ‘My joints [are] very young.’) The reduplicated expression in (24) retains its status as a stative verb oxe~oxe ‘young~young,’ but earlier in this paper in example (3b) and (3c), the same reduplicated expression is shown in other contexts as nouns with the locative ha and the verbalizer -mu, resulting in different but metaphorically relatable meanings. 4.2 Iterativity and Increased Duration Reduplicated constructions may be used to convey meanings of iterativity and an increase in the duration of an action or movement, such as rubbing or shaking. Not unexpectedly, such reduplicated verbs may co-occur with an intensifying suffix (mahi ‘very’), as in example (27b), or the durative suffix -ti, as in examples (25b) and (26). These iconic meanings may be expressed by either full or partial reduplication of the verb stems. (25) a. hwei thə-nɨ warasi wa=pə=riə-ri this indef-erg sore you=pl=rub-imp ‘Rub your sores with this!’ b. wa=riə~riə-ma-mo-ti you=rub~rub-caus-refl-dur ‘Rub (this) on yourself repeatedly over time!’ (26) wa=thə=raẽ~raẽ-ma-ti-nomai ya=mi-o-pihi-o you=indef=shake~shake-caus-dur-proh I=fall.asleep-mid-want-mid ‘Stop causing it to shake continuously, I want to fall asleep!’ (27) a. thuə=thə=pə=heri təhə sitipasi woman=indef-pl=sing when shell =pə=rẽkẽ~rẽkẽ-ma-ɨ-he =pl=shake.up.and.down~shake.up.and.down-caus-dyn-3pl ‘When women sing, they make the shells (on their loincloths) shake up and down (continuously).’

reduplication in the yanomae language of northern brazil

173

b. ipa wana=a=rẽkẽ~rẽkẽ-mahi-tiko 1sg. poss arrowpoint.case=sg=shake.up.and.down~ shake.up.and.down-very-unfortunately ‘Unfortunately, my arrowpoint case keeps shaking up and down.’ The process of reduplication in example (28b) is similar to example (19), in which the verb root ke- ‘fall’ is monosyllabic. It requires that a thematic vowel accompany the root so that the entire (now disyllabic) verb stem mi-a- can undergo the copying process. In addition, this is a case of reduplication of a compound verb mɨsi mi, and the preposed element mɨsi is not copied. Note that mɨsi=mi- in (28a) and mo=hwetu-mu in (29a) are fixed combinations in which the meanings of all individual components are not transparent. Examples (28a) and (28b) appear to derive from mi- ‘fall asleep’ (see example (26)), and reduplication suggests an iterative or punctual reading of ‘closing one’s eyes repeatedly or momentarily’ as the semantic basis for the gloss ‘blink, wink.’ (28) a. mɨsi=miX=fall.asleep ‘close one’s eyes’ b. mɨsi=mi-a~mi-a-mu X=fall.asleep-theme~fall.asleep-theme -vbzr ‘blink, wink’ Notice that the form hwetu- in (29a) and (29b) is an allomorph of hwetɨ(as in example (6), which has been copied below as example (29c) for the reader’s convenience), which results from assimilation to the vowel [u] in the verbalizing suffix -mu. Such assimilation to an adjacent vowel can also be seen in example (32b), where the classifier kɨkɨ (homophonous with and possibly derived from the aggregate kɨkɨ) assimilates to the final vowel of the preceding root in oru kuku ‘snake.’ (29) a. mo=hwetu-mu X=symmetrical-vbzr ‘move around/fidget from side to side (in a hammock)’ b. mo=hwetu~hwetu-mu X=symmetrical~symmetrical-vbzr ‘fidget from side to side continuously (in a hammock)’

174

goodwin gómez

c. hwetɨ~tɨ-a-ɨ symmetrical~red-theme-dyn ‘(be) one after the other (in a symmetical position)’ Partial reduplication is much less frequent in Yanomae. Nevertheless, the repeated appearance of the cayman’s head in (30b) and the turtle’s head in (30c) are conveyed by reduplication. In keeping with the analysis of example (29c), the reduplicant in both (30b) and (30c) is the considered to be the last CV syllable (ro) of the disyllabic root (roro). Iterativity in the sense of repeated motion in (30b) and (30c) and repeated or sequential position in example (29c) are conveyed by partial reduplication. It is interesting to note that examples (30b) and (30c) also involve denominalization by the addition of the verbalizing suffix -mu as first illustrated in example (3c) in section 2. The transitive verb root roro ‘appear’ is partially reduplicated to create a noun which is then changed to a verb which has the added meaning of repeated action, in addition to the category change. This verb represents a case in which both functions of reduplication occur simultaneously. (30) a. miha kawahi wa=kɨ=he=roro-a-ɨ-taa-ɨ there electric.eel you=pl=head=appear-theme-dyn-see-dyn ‘Look there, the electric eel shows its head (on the surface of the water)!’ b. iowa he=roro~ro-mu cayman head=appear~red-vbzr ‘The cayman’s head appears (and disappears) repeatedly (on the water’s surface).’ c. waẽmõramohi he=roro~ro-mu turtle head=appear~red-vbzr ‘The turtle’s head appears (and disappears) repeatedly (from its shell).’ In the next two sets of examples, it is more complicated to explain what appears to be partial reduplication; these examples, involving polysyllabic roots, are exceptional in the data. In the first case (31), the root ãk󰀟̃kã- ‘squeeze’ undergoes partial reduplication to indicate intensity of the action of ‘squeezing.’ It is not obvious, however, how the final reduplicated form ãk󰀟̃k󰀟̃ãɨ is obtained A possible explanation of this process might be that in this case partial reduplication involves copying the second syllable so that the underlying reduplicated stem would be: ãk󰀟̃~k󰀟̃-kã-, and that the final syllable of the root -kã is subsequently

reduplication in the yanomae language of northern brazil

175

deleted. The resulting form ãk󰀟̃~k󰀟̃-ɨ would then trigger the addition of a thematic vowel (to which the nasalization of the preceding vowel spreads) between the almost identical vowels of the copy and the aspectual suffix, resulting in the final form that is found in example (31b). Another possibility is the deletion of the initial k- of the final syllable of ãk󰀟̃~k󰀟̃-kã- for phonological reasons (i.e. ease of articulation). The aspectual -ɨ suffix is never nasalized and has the phonetic value of a voiced velar fricative [ɣ]. (31) a. paxara=ãhũ=hehã-ɨ təhə ãhũ=a=ãk󰀟̃kã-ɨ bundle.funerary=cl.plant=tie-dyn when cl.plant=sg=squeeze-dyn ‘When a funerary bundle is tied (with cord), it’s squeezed tightly.’ b. ya=thə=ãk󰀟̃~k󰀟̃-ã-ɨ I=indef=squeeze~red-theme-dyn ‘I’m squeezing it forcefully!’ Example (32b) appears to involve partial reduplication, if compared with example (32a). However, the overt base form keta ‘stretch’ in (32b) requires further explanation. The process might be considered full reduplication if the simplex aketa has a disyllabic allomorph keta, which is the base that is reduplicated in (32b). More likely, the reduplicant copy keta in (32b) represents a partial copy of aketa, whose initial unstressed vowel a- was suppressed or deleted because of the occurrence of an identical vowel a- immediately preceding it: /áka akéta/ → [áka kéta]. The correctness of this assumption needs to be checked with a native speaker. (32) a. yanomae=thə=marixi-o təhə thə=aketa-mo-rayu person=indef=sleepy-mid when indef=stretch-refl-tel ‘When a person is sleepy, s/he stretches out (her/his arms, legs).’ b. oru=kuku aka=keta~keta-mo-rayo-ma snake=cl tongue=stretch~stretch-refl-tel-past ‘The snake stuck out its tongue repeatedly.’ In a transitional area between true reduplication and ideophonic (onomatopoeic) repetition is a curious set of verb roots in Yanomae that resemble a sort of echo-word formation across lexemes.4 While not equivalent to such

4 See Stolz (2008) for a detailed examination of true echo-word formation.

176

goodwin gómez

echo words in English as hurdy-gurdy or hoity-toity, the similarity in the forms hẽrẽ, kẽẽrẽ, and xẽẽrẽ seems more than coincidental. The original, simplex root hẽrẽ ‘lung’ undergoes full reduplication in example (33b), and with the addition of the verbalizing suffix -mu, the result is a verb that, when occurring with the additional verb kohipə ‘strong,’ has the intensified meaning of ‘breathe deeply.’ The verbal forms in (34) and (35) resemble other repeated ideophones which are common in words related to bodily activities and functions, such as tuku.tuku-mu ‘beat (as a heart)’ and yati.yati-mu ‘tremble/shake (from fever or weakness).’ The verbs in examples (33) through (35) convey specific types of breathing and reflect in their onomatopoeia the unique quality of ‘irregular’ and ‘hissing’ breaths, respectively. However, when examined in the context of examples (33a) and (33b), a clear connection emerges. The sounds in the individual forms kẽẽrẽ (34) and xẽẽrẽ (35) resemble in an echoic fashion the noun root hẽrẽ ‘lung’ and its related form, hẽrẽke ‘inhale.’ One could imagine that the ideophones kẽẽrẽ and xẽẽrẽ derived from hẽrẽ by lengthening the first vowel and changing the initial consonant. In their repeated forms, the initial velar stop in (34) signals a break in the flow of breath indicated by the glottal fricative in (33b), and the palatal fricative ([x] = [ʃ]) in (35) audibly creates the ‘hissing’ breath. (33) a. wa=hẽrẽ-ke-ki you=lung-foc-imper ‘Inhale!’ b. a=hẽrẽ~hẽrẽ-mu-kohipə sg=lung~lung-vbzr-strong ‘He breathes deeply.’ (34) kẽẽrẽ.kẽẽrẽ-mu breathe.ideo-vbzr ‘breathe irregularly’ (35) parɨkɨ xẽẽrẽ.xẽẽrẽ-mu chest breathe.ideo-vbzr ‘breathe with a hissing, creaking (sound in the) chest’

reduplication in the yanomae language of northern brazil

5

177

Unclear Cases: Lexicalization, Ideophonic Repetition, Onomatopoeia

Examples of so-called lexical reduplication and inherently reduplicated lexemes lack simplex forms and, thus, the semantic or grammatical functions of these forms are not obvious. This section is dedicated to the presentation of forms that are not strictly considered reduplication according to the definition that is applied in the current volume but that represent a significant part of the lexicon in Yanomae and are often confused with true reduplicated forms. Moreover, simplex forms may exist in some cases but have just not yet been recorded. Example (36) reveals a different problem, where an apparently reduplicated form and its simplex base have the same meaning. Although the noun yarɨyarɨ a ‘lowlands’ in (36) appears to be the result of reduplication, a synonymous simplex form yarɨ a also exists. (36) yarɨ.yarɨ=a ha mapra-rɨ-nɨ yama=kɨ=hemo-rayo-ma lowlands=sg seq finish-top-seq 1=pl=ascend-tel-past ‘After the lowlands ended, we arrived at a higher place.’ Because no difference in meaning or function has been identified between the two forms, the duplicated one has not been written with a tilde, since it may not be a true case of reduplication. Perhaps, the duplicated form is modeled analogously after other geographic terms, such as hemohemoha ‘in an elevated place’ and weheweheha ‘on dry land,’ that do represent true reduplication. The use of ideophones and ideophonic repetition is quite common in Yanomae, as illustrated by examples (37) and (38). In these sentences the verb stems are formed by adding the verbalizing suffix -mu to a noun, which is composed of a repeated syllable that produces, as Kilian-Hatz so aptly states, a “vivid representation of an idea in sound” (2006, 509). Example (37) clearly conveys the tinkling tĩrĩ.tĩrĩ of the shells on the women’s loincloths as they dance, and example (38) copies the sound of a wood post cracking kreno.kreno. Ideophonic repetition has also been a source for neologisims in Yanomae, such as metal tools: kɨrɨkɨrɨ a ‘a handsaw’ and porepore a ‘a hoe.’ Additional examples have been described in Goodwin Gómez (2009a) for Yanomae as well as other languages in the Yanomami family. (37) thuə=thə=pə=heri təhə sitipasi=pə=tĩrĩ.tĩrĩ-mu woman=indef=pl=sing when shell=pl=jingle.ideo-vbzr ‘When women sing, the shells (on their loincloths) jingle.’

178

goodwin gómez

(38) yano=nahi=xatio=ihe-o təhə house=cl.post=supporting=weak-mid when nahi=kreno.kreno-mu cl.post=cracking.ideo-vbzr ‘When a supporting house post is weak, it makes a cracking noise.’ The other topic that may involve ideophonic repetition or onomatopoeia (especially for animal names, for example) but that may not be overtly sound symbolic is lexical or inherent reduplication. It merits a brief mention in this description of Yanomae reduplication because there are many such words with apparent reduplicative structure in the lexicon. In these lexemes, the reduplicative forms, unlike true reduplication, cannot be associated with an individual (or simplex) base form and do not show a corresponding change that can be identified with a grammatical function or semantic distinction. As has been documented for other Amazonian languages, these inherently reduplicative lexemes are especially common in the names of animals and plants.5 Examples (39) through (48) illustrate the wide range of referents for the names that include inherently reduplicative roots: (39) thoo.thoo-ma=a ideo.ideo-form= sg ‘giant spiny tree rat’ Echymis grandis (40) kuu.kuu=moxi ideo.ideo=cl6 ‘night monkey’ Aotus sp. (41) hrae.hrae-ma=a ideo.ideo-form= sg ‘small frog’ Otophryne robusta (Boulenger) (42) krii.krii=usikɨ ideo.ideo=cl ‘dragonfly’

5 Berlin (1992) devotes an entire chapter (Chapter 6) to the non-arbitrariness of ethnobiological nomenclature. 6 The exact meaning of many of these classifiers remains to be determined.

reduplication in the yanomae language of northern brazil

179

(43) posi.posi-ma=namo ideo.ideo-form=cl.sharp ‘small owl’ Otus sp. (44) wisa.wisa-ma=si ideo.ideo-form=cl ‘Paradise Tanager’ Tangara chilensis (45) pixi.pixi-ma=aka ideo.ideo-form= cl.tongue ‘hermits and hummingbirds’ (46) kira.kira=mo ideo.ideo=cl ‘Red-fan Parrot’ Deroptyus a. accipitrinus (47) xiri.xiri=ãthe ideo.ideo=cl.fish.poison.vine ‘liana used for fish poison’ Banisteriopsis lucida (Richard) Small (48) hato.hato=koxihi ideo.ideo=cl.tree (sp.) ‘forest tree’ Hymenaea courbaril L. One exception is the plant name in example (49), however, which has an identifiable simplex form, rasi ‘flat,’ which characterizes the physical appearance of this vine. Moreover, this appears to involve the category-changing function of true reduplication since the reduplication of the adjectival verb rasi results in a noun rasi~rasi that is followed by a nominal classifer thotho ‘cl.liana.’ (49) rasi~rasi=thotho flat~flat=cl.liana ‘flat-shaped liana’ Philodendron surinamense (Miq.) Engl. Each of these roots is followed by either the singulative or a nominal classifier and in some cases, such as (39), (41), (43), (44) and (45), the nominal roots require the addition of a formative suffix -ma, which precedes the classifier. It is also interesting to note that the range of classifers is diverse, although only some, such as namo ‘sharp, pointed,’ aka ‘tongue,’ ãthe ‘fish poison vine,’ and thotho ‘liana,’ are identifiable or associated with consistent groupings. With

180

goodwin gómez

some of the animal names, especially for birds and frogs, an onomatopoeic origin for their roots seems likely, and native speakers have suggested as much. Nevertheless, further consultation with native speakers would be necessary to confirm any hypotheses about the possible origins for these inherently reduplicative roots. Perhaps, like rasirasi thotho (49), there are other, as yet unidentified, simplex forms that can be discovered by additional, focused field research to provide a better understanding of the origins of these ethnobiological and botanical names.

6

Conclusion

Reduplication in Yanomae fulfills two functions: a grammatical categorychanging function and a semantic, iconic function. The iconic function is characterized by an immediately adjacent full or partial copying of a noun or verb stem. The grammatical function has been observed in the derivation of nouns from stative, positional, intransitive, and transitive verbs. A full copy of the verb stem occurs immediately adjacent to the original base, and, like other nouns, the resulting form may be accompanied by nominal classifiers as well as clitics expressing location, direction, or plurality and suffixes marking ergativity or denominalization. Semantically, Yanomae illustrates several classic, iconic functions of reduplication: increased quantity of a substance or increased intensity, iterativity or duration of the action, movement, position, or state expressed by a verb stem. All of these semantic “increases” constitute some form of augmentation, which corresponds to the physical duplication of the form. Likewise, copying a stative verb stem to create a noun can be seen as moving from a quality (adjective) to a substance (noun), which represents a substantial (or increased) quantity of that quality. Thus, augmentation is the iconic meaning par excellence of the process of reduplication, and in Yanomae it could be argued that both the grammatical category change and the semantic change represent iconic functions. In other words, all reduplication in Yanomae might be considered iconic, even when its function is grammatical (category changing from verb to noun). If one can conceive of a noun as representing an intense degree or amount— one might call it the “embodiment”—of a quality or state (expressed in Yanomae by an adjectival verb), such as ‘gray’ (kroke) or ‘wet’ (here), it is not difficult to see the augmentation of ‘grayness’ and ‘wetness’ as meaning ‘cloud’ (krokekroke a) and ‘humidity’ (herehere pə) in the derived Yanomae nominals in (15) and (1), respectively. Stolz’s suggestion that “reduplication at least intuitively invites the interpretation of conveying more meaning than the simple

reduplication in the yanomae language of northern brazil

181

form” (2007, 322) supports the inherent iconicity of the process, which has been demonstrated by the examples presented here from the Yanomae language of northern Brazil.

Acknowledgements I gratefully acknowledge financial support from the American Philosophical Society for a Franklin Research Grant, the Alice Cozzi Heritage Language Foundation, the Rhode Island College Faculty Research and Faculty Development Funds for multiple grants, and the Fulbright Commission for a Fulbright-Hays Scholar Lecture/Research Grant. The research on which the material for this article is based was facilitated by their generous support. I am grateful to numerous colleagues at the Pro-Yanomami Commission (CCPY) and its more recent incarnation as the Instituto Socioambiental (ISA) in Boa Vista, who provided assistance in Brazil, and, in particular, to Bruce Albert, for his collaboration. I would also like to thank the anonymous reviewer for helpful suggestions, the organizers and participants of the Graz Reduplication Conference in 2007 for sharing their knowledge and enthusiasm about reduplication studies, and Hein van der Voort for his detailed comments and encouragement. Above all, I am indebted to the Yanomami of the community of Watorikɨ for their patience, friendship, and warm hospitality over many years. All errors are my own.

References Albert, Bruce, and Gale Goodwin Gómez. n.d. Léxico themático da língua Yanomae. Manuscript. Albert, Bruce, and William Milliken, with Gale Goodwin Gómez. 2010. URIHI A: A terra-floresta Yanomami. São Paulo: Instituto Socioambiental. Berlin, Brent. 1992. Ethnobiological classification: Principles of categorization of plants and animals in tradional societies. Princeton, NJ: Princeton University Press. Gil, David. 2005. From repetition to reduplication in Riau Indonesian. In Hurch (ed.), 31–64. Goodwin Gómez, Gale. 2009a. Reduplication, ideophones, and onomatopoeic repetition in the Yanomami languages. Grazer Linguistische Studien 71: 1–18. . 2009b. Review of three works by Henri Ramirez. Iniciação à língua Yanomamɨ; Hapa Të Pë Rë Kuonowei: mitologia Yanomamɨ. Texto de Leitura—II; and Le parler Yanomamɨ des Xamatauteri. International Journal of American Linguistics 75/3: 433– 439.

182

goodwin gómez

Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Hurch, Bernhard, and Veronika Mattes. 2009. Typology of reduplication: The Graz Database. In The use of databases in cross-linguistic research, ed. Martin Everaert, Simon Musgrave and Alexis Dimitriadis, 301–327. [Empirical Approaches to Language Typology 41]. Berlin/New York: Mouton de Gruyter. Kilian-Hatz, Christa. 2006. Ideophones. In Encyclopedia of language and linguistics, 2nd ed., Vol. 5, ed. Keith Brown, 508–512. Amsterdam: Elsevier. Migliazza, Ernesto. 1972. Yanomama grammar and intelligibility. PhD diss., Indiana University. Milliken, William, and Bruce Albert, with Gale Goodwin Gomez. 1999. Yanomami: A forest people. London: Royal Botanic Gardens, Kew. Perri Ferreira, Helder. 2009. Los clasificadores nominales del Yanomama de Papiu (Brasil). MA thesis, Centro de Investigaciones y Estudios Superiores en Anthropologia Social (CIESAS), Mexico. Ramirez, Henri. 1994. Le parler Yanomamɨ des Xamatauteri. PhD diss., Université de Provence, France. . 1999. A Prática do Yanomami. Boa Vista, Roraima, Brasil: CCPY. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Stolz, Thomas. 2007. Re: duplication: Iconic vs counter-iconic principles (and their areal correlates). In Europe and the Mediterranean as linguistic areas: Convergencies from a historical and typological perspective, ed. Paolo Ramat and Elisa Roma, 317– 350. Amsterdam/Philadelphia: John Benjamins. . 2008. Total reduplication vs. echo-word formation in language contact situations. In Language contact and contact languages, ed. Peter Siemund & Noemi Kintana, 107–134. Amsterdam/Philadelphia: John Benjamins.

Abbreviations agg caus cl dur dyn erg foc form indef

aggregate/collective causative classifier durative dynamic ergative focalizer stem formative indefinite pronoun

ideo inten instr mid nomzr past pl poss proh

ideophone intensifier instrumental middle voice nominalizer past plural possessive prohibitive

reduplication in the yanomae language of northern brazil

refl seq sg tel theme

reflexive sequential singulative, unitizer telic thematic vowel

vbzr x 1 2 3

verbalizer unidentified segment 1st person 2nd person 3rd person

183

chapter 8

Reduplication as a Tool for Morphological and Phonological Analysis in Awetí Sebastian Drude

Verbal reduplication in Awetí is always full reduplication of the stem, quite independently from its CV structure. There are no different formal reduplication types, but there is some variation in the semantic effects according to the verb class (transitive, active intransitive, stative). The reduplicated active verbs with consonant-final stems show an additional unstressed -e, which also occurs in some simple verbs and in certain forms of stative verbs. Reduplication provides evidence that this -e and a related -zã are separate morphs, different from inflectional affixes and the stem. Reduplication also contributes to deciding several phonological questions; most importantly, the abstract morphemefinal archiphonemes /P, T, K/ are confirmed. Also, the analysis of inherently oral vowels different from neutral vowels, be it in complementary distribution, has been strengthened. Although reduplication is not always a clear criterion to distinguish derivation from inflection, it can reveal the ‘underlying’ form of derivational affixes.

1

Introduction

This study describes reduplication in verb forms of Awetí, a Brazilian indigenous language spoken by a small group living in three villages in the region of the headwaters of the Xingu River in Mato Grosso, central Brazil. Awetí does not belong to but is the closest relative of the well-known Tupí-Guaraní subfamily, the largest well-established branch of the Tupí family. Mawé, Awetí and Tupí-Guaraní together constitute a major branch of Tupí, “Mawetí-Guaraní” (Rodrigues and Dietrich 1997; Drude 2006; Drude and Meira, forthcoming). The study of reduplication can contribute to the understanding of phonological and morphological units and processes. This is shown by discussing reduplication in connection with key issues in Awetí: nasalization, abstract morpheme-final phonemes, and the identification of the stem and of morpheme boundaries. To a lesser degree, reduplication can also be used for the identification of derivation vs. inflection.

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_009

186

drude

Sections 2 and 3 serve as a basis for the paper. Section 2 summarizes the phonological system of Awetí. Section 3 gives some basic morphological properties and presents some diagnostic affixes used in the remaining sections. Then section 4 gives a short general overview of verbal reduplication in Awetí. Section 5 describes the basic patterns of reduplication found with active (transitive and intransitive) verbs. The members of a small particular subclass of these active verbs end in an unstressed syllable; these are treated in section 6. Section 7 uses evidence from reduplication for a closer look at certain (morpho-)phonological alternations at the beginning and end of verbal stems. Reduplication with stative verbs is treated in section 8. Finally, section 9 discusses the domain of reduplication if derivational affixes occur. The most important findings are summarized in the conclusion in section 10.

2

Phonemes and Phonotactics of Awetí

Awetí has six oral vowels /i, ɨ, u, e, a, o/ and their nasal counterparts /ĩ, 󰀟̃, ũ, ẽ, ã, õ/. In most non-final syllables one of six additional ‘neutral’ phonemes /i̱, ɨ̱, u̱ , e̱, a̱ , o̱/ occurs, marked by (non-IPA) underlining. These are phonologically unspecified for orality or nasality and acquire these features phonetically due to nasal harmony. Besides the basic consonantal phonemes /p, t, k, kw, ʔ, m, n, ŋ, j, w, ts, z, h, l, r/,1 a small group of more abstract underspecified phonemes is postulated due to neutralizations. In particular, the contrast between the simple stops and their nasal counterparts (p:m, t:n, k:ŋ) is neutralized in the coda, giving rise to the archiphonemes /P, T, K/, which harmonize with the nasality or orality of the preceding vowel (see section 7). The abstract (underspecified) archiphoneme /P/, for instance, contains only the property “bilabial.” Depending on the environment, it is realized after oral vowels as an unreleased [p ̚] (before pause or stops) or as a ‘lenited’ [β] (before vowels—/T/ and /K/ are lenited to to [ɾ] and [ɣ], respectively), and as [m] after nasal vowels or before nasal consonants. At the beginning of certain (in particular, modal and nominalizing) suffixes, there is an abstract consonantal phoneme /°/.2 After [m, n, ŋ] (that is, after /P, T,

1 The status of /kw, h/ as basic consonantal phonological units is debatable. The phonetic realization is made explicit in all examples. 2 As is common for more abstract phonemes, there is no IPA symbol for this segment. We use a small raised circle. The segment contains only the properties ‘oral’ and ‘stop/plosive.’

187

reduplication as a tool for analysis in awetí

K/ following nasal vowels), it is realized phonetically as a homorganic stop [p, t, k] and analogically as [t] after [ j] or as [p] after [w]. If /°/ occurs after [p, t, k] (that is, after /P, T, K/ following oral vowels), it inhibits their lenition, following the pattern /t+úP+°u/ > [ˈt+up+pu] > [ˈtupu] ‘to stay,’ cf. (8). The abstract patterns in A shows the main positions for the basic disyllabic as well as for monosyllabic and trisyllabic stems, where brackets indicate that consonants are optional in all slots. The patterns B to H below refer to these abstract patterns. A.

[C1] V1 [C2] V2 [C3] — [C1] V2 [C3] — [C1] V1 [C2] V1 [C2] V2 [C3]

The phonemes distribute differently over these positions; for instance /i̱, ɨ̱, u̱ , e̱, a̱ , o̱/ occur only in V1, and in C3 only /j, w, P, T, K/ occur; the latter three occur only here. The picture can be complicated, if rarely, by occurrence of glides between positions of the patterns A. Awetí words are usually stressed on the last syllable of the stem (Drude 2011b). There are, however, a few paroxytonic stems with an additional final unstressed syllable (see section 6). Only a few suffixes, mostly derivational ones, attract lexical stress. In the phonological representations in this study, stress is marked by an acute accent over the vowel of the stressed syllable.

3

Relevant Morphological Facts and Terms

As far as verb forms are concerned, Awetí is a moderately agglutinating language. Typical verb forms of simple verbs contain three or four, those of derived verbs maximally seven morphs. For the ease of the reader, the morphs of verb forms in the text are separated by a middle dot ‘·’ (e.g., ‘toK·tóK·e̱·tu’). Generally, only one derivational affix is present in any given verb form. The principal morphological positions or ‘slots’ of an Awetí verb form are as follows: person

voice

causative

STEM

derivative

aspect

mood

Awetí verb forms contain at most one person prefix referring either to the subject or to the object.3 The voice prefixes include te̱- ‘reflexive’ and to̱ - ‘reciprocal’;

3 If not explicitly indicated otherwise, all person prefixes in transitive verbs in the examples refer to the subject.

188

drude

the same slot may be occupied by incorporated nouns instead. Before an active intransitive stem, a derivational prefix may occur. With other stems, derivational affixes immediately follow it (cf. section 9). Many verb forms take one of the two aspect suffixes -ju or -zo̱ ko (see Tables 3 and 4 below). The perfective forms are unmarked. Finally, many moods, most of them used for subordination, are marked by a suffix occurring at the very end of the main verb. One of these, -tu (see Table 1 below), is frequently used in this paper. Also the negative suffix -(ɨ)ka appears in the final slot. By the notion ‘stem’ (of a word) I here understand a morph or sequence of morphs which carries the lexical meaning of the word and which serves as a basis for adding one or several inflectional affixes (yielding complete inflected forms of the word), and/or as a basis for derivation or composition. The result of these latter is again a stem: the stem of a complex word. I maintain the traditional distinction between derivation (yielding new stems with a different lexical meaning) and inflection (marking functional categories of different forms of the same lexical word), although there may be cases which are difficult to assign. For the purposes of this paper, I focus on certain inflected forms of verbs which have, in particular, certain affixes that are useful for identifying morpheme boundaries. These forms are: the absolute form (Tables 1 and 2) which is the citation form used to refer to the verb itself; the imperfective forms (Table 3), meaning: ‘to do … as a habit,’ or ‘to be about to do …’; and the progressive forms (Table 4), general meaning: ‘to be doing …,’ with stative verbs possibly also ‘to become ….’ table 1

Suffixes and their allomorphs in absolute verb forms: -tu/-°u ‘abs,’ -zã/-ã4

Stem ends in:

Vowel

Consonant

Verb type: Active Stative

-tu -zã·tu

-°u -ã·tu

4 The element -zã/-ã, with empty lexical/functional semantics, is glossed zã. See below section 6.

reduplication as a tool for analysis in awetí table 2

Prefixes and their allomorphs in absolute verb forms: t- ‘abs,’ 5 n-/nã-6

Stem begins with:

Vowel

Consonant

Verb type: Transitive active Intransitive active & stative

nt-

nã–7

table 3

Allomorphs of the imperfective suffix -zo̱ ko/-o̱ ko/-e̱·zo̱ ko ‘ipfv’

Stem ends in:

Vowel

/T/8

Other consonants

Verb type: Active Stative

-zo̱ ko -zo̱ ko

-zo̱ ko -e̱·zo̱ ko

-o̱ ko -e̱·zo̱ ko

table 4

189

Allomorphs of the progressive suffix -ju/-°e̱ju/-e̱·ju ‘prog’

Stem ends in:

Vowel

/T, j/

Other consonants

Verb type: Active Stative

-ju -ju

-ju -e̱· ju

-°e̱ju -e̱· ju

5 The elements t- and -tu/-°u in intransitive verbs can be seen as parts of a circumfix ‘abs.’ 6 In absolute forms, this 3rd person object prefix could also be glossed as ‘abs’ and/or be analyzed as part of a circumfix. Note that Awetí women use t- or i̱- instead. See Drude (2002). 7 There is no prefix in absolute forms of consonant-initial intransitive verbs, but a stem-initial /p/ changes to /m/. 8 After oral vowels, /T/ is phonetically deleted before -ʐ̱ oko. After nasal vowels, /T/ surfaces as [n] which apparently behaves as any other consonant: the suffix seems to be just -o̱ ko. However, phonologically we postulate /…T-z…/ in both cases. The apparent idiosyncrasies arise on the phonetic level by the rules of solving sequences of homorganic consonants: […tʐ…] > […ʐ…]; but […nʐ…] > […n…].

190

drude

Henceforth we refer to the suffixes above as -tu, -zo̱ ko and -ju, respectively. In the text, verbs are cited by their absolute form, with the affixes in Tables 1 and 2. For the finite forms, the examples show mostly third person forms. The third person prefixes, glossed ‘3,’ are o̱ - (resyllabified as [w] before certain vowels) for active intransitive verbs, we̱j- for transitive verbs, and for stative verbs i̱(before consonants) or t- (before vowels). The patterns of reduplication are the same with other person categories. Note that singular and plural are not distinguished in third person Awetí verb forms. In the free translations, we usually use ‘he’ or ‘him’ (‘it’ for inanimate entities), but in fact all third person pronouns could be used: ‘he/she/it/they’ or ‘him/her/it/them.’ All forms presented here are unmarked for tense, that is, they belong to the non-future category. In the free translations, in particular of perfective forms (forms unmarked for aspect), we usually choose past tense because this reflects the perfective meaning and is the unmarked translation given by Awetí speakers.

4

General Properties and Semantics of Reduplication in Awetí

In Awetí, reduplication as a productive regular morphological process occurs only with verbs.9 Generally, the input for reduplication is a verb stem, not including inflectional affixes.10 In many cases, the input may consist of a stem formed by derivation (that is, the derivational affix, such as a valence changing affix, is reduplicated together with the original simple stem), but this does not work with all affixes or with all stems (see section 10). Reduplication has been assumed to be suffixing or ‘postfixing’ in the analysis for all Tupian languages we are aware of (Jensen 1998; Rodrigues 1953; Seki 2000). But, in fact, as reduplication in Awetí is always full reduplication, it is almost impossible to decide which of the two tokens would be the ‘base’ and which the ‘reduplicant.’ If only some but not all morphs of a stem are copied, these are often the first morphs, which could be taken as evidence for prefixing reduplication in Awetí. Generally, we sympathize with ‘projection– linearization’ approaches to reduplication which do not distinguish between a

9 10

Reduplicated forms of postpositions and idiophones are not discussed in this paper. In contrast, Rose (2007; 2005) observes that in Tupí-Guaraní languages (parts of) person prefixes can be included in the base and reduplicant in order to complete a two-syllabic reduplication pattern.

reduplication as a tool for analysis in awetí

191

base and a reduplicant (affix).11 Thus we speak generally rather of the ‘left hand copy’ and the ‘right hand copy’ of the ‘input’ (usually the stem of the original, non-reduplicated verb) instead of the ‘reduplicant’ and the ‘base.’ The formal properties of reduplication in Awetí are summarized below. They are all extensively described and exemplified in the following sections. For quick orientation, some examples (only stems, not entire inflected word forms) are also given. 4.1 Formal Properties of Reduplication in Awetí a. Stems with final stress are copied entirely, independently of their syllable structure (e.g. tó > to~tó; ekó > e̱ko~ekó; mo̱ tó > mo̱ to~mo̱ tó; see section 5); b. Active verbs with stems ending in a consonant show a final additional unstressed -e̱ in the reduplicated form (e.g. úT > uT~úT-e̱; mɨ̱zũ ́K > mɨ̱zũK~ mɨ̱zũ ́K-e̱; see section 5); this does not hold for stative verbs (e.g. t󰀟́P > tɨP~t󰀟́P; pot󰀟́j > potɨj~pot󰀟́j; see section 8); c. Stems with a final unstressed syllable do not copy the final parts -e̱ or -ã (after consonants; e.g. tóK-e̱ > toK~tóK-e̱; kã ́K-ã > kãK~kã ́K-ã) or the final part -zã (after vowels; e.g. kɨrʔã ́-zã > kɨrʔã~kɨrʔã ́-zã); see section 6. In the terms given in the previous section, although highly productive12 and formally regular, reduplication in Awetí is clearly not an inflectional but a derivational (word-formation) process. In other words, reduplication is used to obtain new lexical items (verbs) with meanings which are systematically related to the meaning of the original simple verb (with the non-reduplicated stem). Semantically, different effects of reduplication exist, but these are not correlated with formal distinctions. In particular, there is no semantic distinction between monosyllabic versus di- or multisyllabic reduplication. This is because there is only total/full, no partial reduplication in Awetí, independent of the number of syllables. There are, however, differences between lexical verb classes (transitive vs. active intransitive vs. stative verbs), which in turn reflect semantic differences (stative verbs mostly denote properties). The main lexical-semantic effects of reduplication are those listed below, ordered (impressionistically) by importance or frequency. The most frequent effects are related to pluractionality. Often several of these effects occur in combination [e.g. (2)], or the meaning of the resulting verb varies [e.g. (1), (3)],

11 12

See, for instance, Raimy (2000), Halle (2008), Reiss & Simpson (2009). We have not found any example of a verb which would not permit reduplication.

192

drude

applying one or another semantic effect. At least one example each is indicated in the following list: 4.2 Semantic Effects of Reduplication in Awetí a. The action/event happens repeatedly (several times, simultaneously or in sequence) (3) (15) (24); b. There are several subjects of the action/event, simultaneously or not (in particular with intransitive verbs) (11) (15) (17); c. With transitive verbs, the action/event happens with respect to several objects (5) (31). Sometimes there is an additional distributive meaning component (5) (28). In the case of ditransitive verbs, plurality may concern the recipient/benefactive (3); d. With stative verbs (property concepts), the property does not hold fully (attenuation) but rather ‘more or less’ (26); e. The action/event involves several different directions (mostly with verbs of motion) (2) (11) (22); f. Several other somewhat less frequent meaning components are partly lexicalized: lack of control (2) (10);—‘little-by-little’ (after 18) (35);—inchoative (‘is beginning to …’) (2) (35);—lack of reason or purpose (6). We now turn to describe and illustrate reduplication in more detail, starting with active verbs.

5

Reduplication with Active Verbs: Basic Patterns

As there is no Awetí verb with a stem consisting of only one phoneme. One of the simplest patterns for verb stems is C1V2.13 As for all cases of reduplication with vowel-final stems with stress on the last syllable, the reduplicated verb has just two complete copies of the input, the original stem. Thus, pattern B holds for a C1V2 stem. B.

C1V2

>

C1V2~C1V2

Consider for instance the verb tó·tu14 ‘to go, to leave’ (stem: tó) and its reduplicated counterpart to·tó·tu which has two meanings: ‘to flee’ (many, typically

13 14

The indices refer to the patterns A in section 2. For the pattern V2C3, see (8), below. For the absolute forms of verbs used as citation forms see section 3.

reduplication as a tool for analysis in awetí

193

in different directions) and ‘to wander’ (as a small child). We show the simple (third person) perfective form in (1a,c) and the absolute citation form in (1b,d) for both verbs. In all examples we show a phonetic representation in the first line and a phonological representation in the second line.15 The morphs are glossed in the third line, and free translations are given in the fourth and possibly a fifth line. (1)

a. [ɔˈtɔ] b. [ˈtɔtu] c. [ɔtɔˈtɔ] d. [tɔˈtɔtu] /o̱ -tó/ /tó-tu/ /o̱ -to~tó/ /to~tó-tu/ 3-go go-abs 3-go~go go~go-abs ‘He went.’ ‘to go’ ‘They fled.’ ‘to flee (many)/ ‘(The child) wandered.’ to wander’

Similarly, pattern C holds for disyllabic stems with final stress and beginning and ending in a vowel. Again, the whole stem is copied without further modifications. C.

V1C2V2

>

V1C2V2~V1C2V2

Consider the verb t·e̱kó·tu ‘to walk’ and its reduplicated counterpart t·e̱ko·e̱kó·tu with the meaning ‘to begin to walk’ (typically stumbling, trying again and again, as a toddler, also ‘walking around without a clear direction’), exemplified in (2). (2)

a. wɛˈkɔ b. wɛˌkɔɛˈkɔ o̱ -ekó o̱ -e̱ko~ekó 3-walk 3-walk~walk ‘He walked.’ ‘(The child) began to walk (stumbling).’

Also consonant-initial and vowel-final disyllabic stems with final stress are completely reduplicated, see pattern D. D.

15

C1V1C2V2

>

C1V1C2V2~C1V1C2V2

The phonological representation indicates morpheme boundaries by hyphens or, between the two copies in reduplicated forms, the tilde ‘~’. In (1), we indicate the phonetic/phonological status by brackets and slashes, respectively. We will not do so in the rest of this study. Awetí forms in italics in the main text are always phonological if not marked by brackets.

194

drude

For instance, consider the transitive verb nã·mo̱ tó·tu ‘to give’ and its reduplicated counterpart nã·mo̱ to·mo̱ tó·tu meaning ‘to give repeatedly’ or ‘to give to several people, to distribute.’ We only show the finite forms for these words in (3), adding a further verb derived from the reduplicated verb nã·mo̱ to·mo̱ tó·tu by means of the completive suffix -wã. (3)

a. ˌw̃ ɛ̃ɲmɔˈtɔ b. ˌw̃ ɛ̃ɲmɔˌtɔmɔˈtɔ c. ˌw̃ ɛ̃ɲmɔˌtɔmɔtɔˈwã we̱j-mo̱ tó we̱j-mo̱ to~mo̱ tó we̱j-mo̱to~mo̱to-wã ́ 3-give 3-give~give 3-give~give-compl ‘He gave it.’ ‘He gave repeatedly.’ ‘He gave it all, repeatedly.’ ‘He gave to ‘He gave it all to several people.’ several people’

In the case of transitive verbs the stem of which starts with a vowel, forms with subject-agreement exhibit an additional /t/ between the person prefix and the stem proper. In principle, three analyses are possible for this /t/: (a) it is part of the stem (that is, the stem has an allomorph with an additional segment /t/ which occurs after subject person prefixes); (b) it is part of the prefix (that is, the subject prefixes have allomorphs with a final segment /t/ occurring immediately before vowel-initial stems); (c) it is a morph by itself, for instance a generic object marker.16 The three possibilities are illustrated ́ ‘to in (4), together with an object-centered form of the same verb n•a̱ tĩ•tu tie.’ (4)

a. wɛ̃ɲtãˈntĩ b. kãɲãˈntĩ we̱j-ta̱ tĩ ́ we̱jt-a̱ tĩ ́ we̱j-t-a̱ tĩ ́ ka̱ j-a̱ tĩ ́ 3.subj-tie 3.subj-tie 3.subj-obj?-tie 1.incl.obj-tie ‘He tied it.’ ‘[Someone] tied you and me.’

An argument against hypothesis (a) is provided by reduplication data: the /t/ is not reduplicated, which suggests that it is not part of the stem. See the ́ ‘to tie several, or at several places’ forms of the reduplicated verb n•a̱ tĩ•a̱ tĩ•tu in (5).17

16 17

It is quite possible that the /t/ is at least historically related to the third person prefix of stative verbs (and of nouns, in the female variety of Awetí, cf. Drude 2002). The morpheme breaks in (5a) follow analysis (b). Still, the other options are not completely refuted, not even (a); see footnote 28.

reduplication as a tool for analysis in awetí

(5)

195

a. ˌwɛ̃ɲtãˌntĩãˈntĩ b. ˌkãɲãˌntĩãˈntĩ we̱jt-a̱ tĩ~a̱ tĩ ́ ka̱ j-a̱ tĩ~a̱ tĩ ́ 3.subj-tie~tie 1.incl.obj-tie~tie ‘He tied them all.’ ‘[Someone] tied everyone of us (including you).’ ‘He tied it all over.’ ‘[Someone] tied us at many places.’

The pattern of reduplication identified so far, see patterns B, C and D, changes slightly when the simple stem ends in a consonant. Also in this case the entire original simple stem is copied, but in addition there is now a final unstressed -e̱ in the reduplicated form, as illustrated by the simple finite forms of the verb táK·°u ‘to cry’ and their reduplicated counterpart taK·táK·e̱·tu ‘to cry often (for no reason)’ in (6). They exemplify the pattern E, below, for a consonant-initial monosyllabic stem. Importantly, the -e̱ does not show up between the two copies of the stem, where it would phonotactically make sense (avoiding consonant encounters, as happens generally in Awetí morphology). So this -e̱ is evidently not a phonetic or phonological phenomenon. E.

(6)

C1V2C3

>

C1V2C3~C1V2C3-e̱

a. ɔˈtak̚ b. ɔtak̚ˈtaɣɛ o̱ -táK o̱ -taK~táK-e̱ 3-cry 3-cry~cry-tv ‘He cried.’ ‘He cried often (for no reason).’

We gloss this final -e̱ as ‘thematic vowel’ (“tv”).18 It does not seem to have a proper meaning, neither lexical nor functional, being comparable to thematic vowels or composition junctures in other languages. Still, it seems to have morphological status, being part of the stem derived by reduplication. Note, for instance, that the -e̱ is not omitted before inflectional suffixes, which appear in their post-vowel allomorph. Consider for example the imperfective forms with the suffix -zo̱ ko (compare Table 3, above), given in (7).19

18 19

Despite the label, we assume thematic vowels to be morphs, similar to ‘epenthetic interfixes’ in composition (cf. the German Fugenmorphem, or the French e muet). Note that the second possible semantic interpretation of the imperfective, ‘to be about to …,’ is not possible, or much less common, with reduplicated verbs.

196 (7)

drude

a. ɔˈtaɣɔkɔ b. ɔtak̚ˈtaɣɛʐɔkɔ o̱ -táK-o̱ ko o̱ -taK~táK-e̱-zo̱ko 3-cry-ipfv 3-cry~cry-tv-ipfv ‘He used to cry.’ ‘He used to cry frequently (for no reason).’ ‘He is about to cry.’

Also before the mood suffixes, the morph -e̱ is obligatory in the reduplicated verbs. This is illustrated by the two citation forms given in (8). Again, the absolute suffix -tu/-°u appears in its post-vowel allomorph -tu in the reduplicated verb. (8)

a. ˈtaku b. tak̚ˈtaɣɛtu táK-°u taK~táK-e̱-tu cry-abs cry~cry-tv-abs ‘to cry’ ‘to cry repeatedly (for no reason)’

The thematic vowel -e̱ is different from the /°e̱/ which occurs at the beginning of the allomorph -°e̱ju of the aspect suffix -ju, see Table 4. The latter contains the initial abstract consonant /°/ which inhibits lenition of final stops and attracts word accent. Both are illustrated in (9a), in contrast to (9b)20 [and also to (6b), (7b) and (8b)]. (9)

a. ɔtaˈkɛju b. ɔtak̚ˈtaɣɛju o̱ -taK-°éj̱ u o̱ -taK~táK-e̱-ju 3-cry-prog 3-cry~cry-tv-prog ‘He was crying.’ ‘He was crying repeatedly (for no reason).’

On the other hand, the /°e̱/ in -°e̱ju does not occur with a stem ending in /T/, where the progressive suffix appears in its shorter form -ju, phonetically fusing with the stem (by regular deletion of the first segment of a sequence of two homorganic consonants). Compare the perfective and progressive forms of the verbs téT·°u [ˈtɛtu] ‘to sleep’ and teT·téT·e̱·tu [tɛˈtɛɾɛtu] ‘to snooze repeatedly (for instances),’ in (10b and d). (10) a. ɔˈtɛt ̚ o̱ -téT 3-sleep ‘He slept.’ 20

b. ɔtɛˈtɛɾɛ o̱ -teT~téT-e̱ 3-sleep~sleep-tv ‘He snoozed repeatedly.’

(9b) is hypothetical; it has not been elicited and might be odd for semantic reasons.

reduplication as a tool for analysis in awetí

197

c. ɔˈtɛju d. ɔtɛˈtɛɾɛju o̱ -téT-ju o̱ -teT~téT-e̱-ju 3-sleep-prog 3-sleep~sleep-tv-prog ‘He was sleeping.’ ‘He was snoozing repeatedly.’ The pattern for vowel-initial stems also shows the thematic vowel. In the case of a monosyllabic consonant-final stem with pattern V2C3, the pattern F holds. F.

V2C3

>

V2C3~V2C3-e̱

This pattern is illustrated below with forms of the verb t·úT·°u ‘to come’ and t·uT·úT·e̱·tu ‘to gather’ (that is, for several to come, from different directions): in (11) with perfective forms, in (12) with imperfective forms, and in (13) for the absolute forms. [Here, V2=u and C3=T, compare patterns A above.] (11) a. ɔˈut ̚ o̱ -úT 3-come ‘He came.’

b. ɔuˈɾuɾɛ o̱ -uT~úT-e̱ 3-come~come-tv ‘They gathered.’

(12) a. ɔˈuʐɔkɔ b. ɔuˈɾuɾɛʐɔkɔ o̱ -úT-zo̱ ko o̱ -uT~úT-e̱-zo̱ ko 3-come-ipfv 3-come~come-tv-ipfv ‘He used to come.’ ‘They used to gather.’ ‘He is about to come.’ (13) a. ˈtutu t-úT-°u abs-come-abs ‘to come’

b. tuˈɾuɾɛtu t-uT~úT-e̱-tu abs-come~come-tv-abs ‘to gather’

Further patterns could be added, such as the patterns G or still others for stems with three or more syllables. Two of them will occasionally be illustrated below, see the corresponding examples. G.

a. C1V1C2V2C3 > C1V1C2V2C3~C1V1C2V2C3-e̱ (20) b. C1V1V2C3 > C1V1V2C3~C1V1V2C3-e̱ (22) c. V1C2V2C3 > V1C2V2C3~V1C2V2C3-e̱

198

drude

Interestingly, there is a small group of simple (not reduplicated) verbs which also have the thematic vowel -e̱, or an unstressed final element -(z)ã. These are discussed in the next section.

6

Reduplication with Active Verbs with a Final Unstressed Syllable

With a small group of non-reduplicated paroxytonic verbs it is not immediately clear whether the final -e̱ or -(z)ã is a part of the stem or not. Consider the verb [ˈtɔɣɛtu] ‘to tear’ (unidimensional objects such as threads, lines, ropes, etc.). Two characteristic forms and my analysis are given in (14). (14) a. ɔˈtɔɣɛ b. ɔˈtɔɣɛju o̱ -tóK-e̱ o̱ -tóK-e̱-ju 3-tear-tv 3-tear-tv-prog ‘It tore.’ ‘It was tearing.’ Words like [ˈtɔɣɛtu] are phonologically suspicious. Generally all Awetí words have lexical stress on the last syllable of the stem; exceptions involve known morphological processes. Also, the phonetic segment [ɣ] generally occurs as an allophone of /K/, which only occurs morpheme-finally. So is the phonological form tóɣe̱-tu or tóK-e̱-tu? Without the data from reduplication, we would not have a compelling reason for assuming a morphological boundary before the final /e̱/, because there is no form of the verb [ˈtɔɣɛtu] without this /e̱/. In the reduplicated counterpart [tɔk̚ˈtɔɣɛtu] ‘to tear (many unidimensional objects, and/or several times, at several places),’ the suspicious /e̱/ is not included in the left hand copy, see (15a). The [ɣ] in the right hand copy corresponds to [k̚] in the left hand copy, pointing indeed at an underlying phoneme /K/ (cf. section 2). (15) a. ɔtɔk̚ˈtɔɣɛ b. ˈtɔɣɛtu o̱ -toK~tóK-e̱ tóK-e̱-tu 3-tear~tear-tv tear-tv-abs ‘(The rope) tore at several places.’ ‘to tear (unidimensional)’ ‘They (e.g., many ropes) tore.’ We conclude that the /e̱/ is not part of the input of the reduplication process (nor of the ‘base,’ in terms of reduplication as affixation); the part of the original verb to be reduplicated consists of only the ‘proper’ stem. Therefore, we can assume a morphological border before the /e̱/ not only in the complex word

reduplication as a tool for analysis in awetí

199

toK·tóK·e̱·tu but also in the simple verb tóK·e̱·tu. This indicates that the final -e̱ is, again, a thematic vowel, already in the non-reduplicated word, see (14) and (15b).21 Thus, the pattern H holds for reduplication in this case and in similar cases. H.

C1V2C3-e̱ >

C1V2C3~C1V2C3-e̱

Although the thematic vowel -e̱ seems to be more common with monosyllabic stems, the regular pattern H holds analogously with more syllables. For example to̱ róK·e̱·tu ‘to tear’ (of bidimensional objects, that is, of cloth, nets, paper, etc.) has its reduplicated counterpart to̱ roK·to̱ róK·e̱·tu ‘to tear (many bidimensional objects, and/or several times, at several places).’ There is a second even smaller class of active verbs that have another final unstressed vowel, namely an /ã/ (intrinsically nasal) with similar properties as the /e̱/ discussed above. For phonological reasons a morphological boundary before that -ã is even more obvious than in the case of the thematic vowel -e̱: the nasality of the -ã usually does not spread to the proper stem, as would be expected if it were the last vowel of the stem.22 Also, as I will show below, there is a variant of -ã after vowels, -zã, where the /z/ (a segment that cannot occur in position C3) is also not included in the reduplication. This is the same -(z)ã of Table 1, above. Consider first the example of the intransitive verb kã ́K·ã·tu [kˈãŋãntu] ‘to dry’ in (16) and its reduplicated counterpart kãK·kã ́K·ã·tu [kãŋkˈãŋãntu] ‘(for several) to dry’ in (17).23 The morph -(z)ã does not have a clearly identifiable meaning although formerly it might have been a derivational suffix that derived verbs from nouns (however, the relation of kã ́K·ã·tu ‘to dry’ to the noun kã ́K [kãŋ] ‘bone’ is hard to establish). For lack of a clear semantic/functional label, this morph is here glossed just as ‘zã.’

21

22

23

In other words, the complete stem of all forms of tóK·e̱·tu consists of two morphs, the ‘proper’ stem tóK and the thematic vowel -e̱, similar to many inflected verb forms in Romance languages. Admittedly, this could be accounted for by postulating nasalization rules which refer to lexical stress, as they are known in Tupí-Guaraní languages with nasal harmony. In many cases, the stem is also nasal, which can be explained by lexicalization, where the morphological boundary tends to disappear. The stem kãK of this verb is intrinsically nasal. In section 8 on stative verbs there will be examples that illustrate that the nasality of the morph zã does not spread to an oral stem.

200

drude

(16) a. ɔ̃ ˈŋkãŋã o̱ -kã ́K-ã 3-dry-zã ‘It dried.’

b. ɔ̃ ˈŋkãŋãɲũ o̱ -kã ́K-ã-ju 3-dry-zã-prog ‘It was drying.’

(17) a. ɔ̃ ŋkãŋˈkãŋã b. kãŋˈkãŋãntu o̱ -kãK~kã ́K-ã kãK~kã ́K-ã-tu 3-dry~dry-zã dry~dry-zã-abs ‘Several dried.’ ‘(for several) to dry’ For a vowel-final stem with an additional morph -(z)ã, see the example with the monosyllabic stem põ ‘to be loaded, filled up (as of a canoe)’ and its reduplicated counterpart põ·põ meaning ‘(for many) to be loaded, filled up’ in (18).24 Despite their semantics, these are active verbs. (18) a. ɔ̃ ˈmpɔ̃ʐã b. ɔ̃ mpɔ̃ ˈmpɔ̃ʐã ́ o̱ -põ -zã o̱ -põ~põ -́ zã 3-be.full-zã 3-be.full~be.full-zã ‘It was full.’ ‘Several (e.g., boats) were filled up.’ A disyllabic example is kɨrʔã ́·zã·tu ‘to gain weight/be fat’ and the reduplicated kɨrʔã·kɨrʔã ́·zã·tu ‘to (begin to) gain weight (little by little)’ (as after a serious illness). Again, as the previous example, this is an active verb whose semantics are like that of a stative verb, denoting a property. (This may be related to the occurrence of -(z)ã.) There is also a disyllabic consonant-final example, ʔatũ ́K·ã·tu ‘to burn down’ and ʔatũK·ʔatũ ́K·ã·tu ‘(for several) to burn down.’ In sum, verbs with the final unstressed morph -(z)ã reduplicate only the proper stem, without -(z)ã, as formulated in pattern I. I.

24

STEM-(z)ã

>

STEM~STEM-(z)ã

The citation forms of these verbs, mõ ·́ zã·tu and mõ·mõ ·́ zã·tu, will be explained in the following section.

reduplication as a tool for analysis in awetí

7

201

Reduplication and the /p/–/m/ Allomorphy and Final Consonants

Consider the verb mɨ̱zũK·°u ‘to stamp one’s foot.’ All inflected forms contain the stem pɨ̱zũK, an allomorph which begins with /p/, as shown in (19a). (19) a. ɔ̃ mp󰀟̃ˈʐũŋ b. m󰀟̃ˈʐũŋku o̱ -pɨ̱zũK mɨ̱zũK-°u 3-stamp.foot stamp.foot-abs ‘He stamped his foot.’ ‘to stamp one’s foot’ In fact, showing an /m/ instead of a /p/ in the absolute form is a phenomenon not restricted to verbs (and also known from other Tupí languages) (Jensen 1998; Seki 2000). For instance, the nouns /m󰀟́/ [ˈmɨ] ‘foot’ and /mo̱ r󰀟́wyt/ [mɔˈɾɨwɨt]̚ ‘custom, culture’ both show a stem- initial /p/ in the inflected (possessed) forms, e.g. /i̱-p󰀟́/ [iˈpɨ] ‘my foot’ or /o̱zo̱ -po̱ r󰀟́wyt/ [ɔʐɔpɔˈɾɨwɨt]̚ ‘our (excl) culture.’25 The following example (20), with a finite and the absolute form of the reduplicated verb mɨ̱zũK·mɨ̱zũ ́K·e̱·tu ‘to stamp one’s foot repeatedly’ (often also by a group of people, as this activity is part of several rituals), shows that the process of changing the /p/ to /m/ is not just, say, a phonetic phenomenon which occurs word-initially. It shows also that the initial /m/ is not a prefix but an integral part of the stem, which happens to have two different allomorphs, one for all inflected forms and one for the absolute form. For building the absolute form of the reduplicated verb, reduplication takes the simple absolute stem allomorph mɨ̱zũK as its input. (20) a. ɔ̃ mp󰀟̃ˌʐũŋp󰀟̃ˈʐũŋɛ b. m󰀟̃ˌʐũŋm󰀟̃ˈʐũŋɛtu ́ o̱ -pɨ̱zũK~pɨ̱zũ K-e̱ mɨ̱zũK~mɨ̱zũ ́K-e̱-tu 3-stamp.foot~stamp.foot-tv stamp.foot~stamp.foot-tv-abs ‘He stamped his foot repeatedly.’ ‘to stamp one’s foot repeatedly’ The p/m alternation is independent of the orality or nasality of the stem in question. The next example (21) shows máP·°u ‘to end, to be over, to be used up,’ an intransitive verb with an oral stem, and its reduplicated counterpart maP·máP·e̱·tu ‘to end (of several).’

25

Obviously, pɨ/mɨ ‘foot’ is also the first element in the lexicalized verb mɨ̱zũK·°u in (19).

202

drude

(21) a. ɔˈpap̚ b. ɔpaˈpaβɛ o̱ -páP o̱ -paP~páP-e̱ 3-end 3-end~end-tv ‘It ended, it is used up.’ ‘Several ended, e.g. a people became extinct.’ c. ˈmapu d. maˈmaβetu máP-°u maP~máP-e̱-tu end-abs end~end-tv-abs ‘to end, to be used up’ ‘to end (plural subject)’ There are several interesting facts to be observed in (21). For one thing, it can again be seen that sequences of (homorganic) consonants at the morpheme boundary are phonetically reduced (21b&d). More importantly, in a form like (21d) all vowels are phonetically oral, showing that inherently oral vowels V2 are not nasalized even when in an unstressed position and before a phonetically nasal consonant. This is an important argument for inherently oral vowels. (Apparently obvious alternative analyses would assume only nasal and neutral vowels where the neutral vowels would be phonetically realized as oral “by default.”) Consider next mo̱ áT·°u ‘to shoot’ (e.g. an arrow), an intransitive verb where the stem-final consonant C3 /T/ is not homorganic with the initial p/m, and its reduplicated counterpart mo̱ aT·mo̱ áT·e̱·tu ‘to shoot (several)’ (e.g. arrows, typically in different directions, firing indiscriminately) in (22). (22) a. ɔˈpwat ̚ o̱ -po̱ áT 3-shoot ‘He shot [an arrow].’

b. ɔpwatˈ̚ pwaɾɛ o̱ -po̱ aT~po̱áT-e̱ 3-shoot~shoot-tv ‘He shot several [arrows].’

c. ˈmwatu d. mwanˈmwaɾɛtu mo̱ áT-°u mo̱ aT~mo̱áT-e̱-tu shoot-abs shoot~shoot-tv-abs ‘to shoot (e.g., an arrow)’ ‘to shoot (several arrows, in all directions)’ This is, again, a strong argument in favor of the abstract final segment /T/ as an archiphoneme (unspecified for nasality or orality) as postulated in section 1. /T/ is phonetically not deleted in the reduplicated verb, at least not in slow speech (this is also reflected by the orthography: ⟨opwatpware⟩, ⟨mwan-

reduplication as a tool for analysis in awetí

203

mwaretu⟩). Its phonetic outcome varies—before the nasal consonant /m/, it is realized as [n], as part of the rules of nasal harmony (the outcome depends not only on the preceding vowel, but also on the following consonants). That means that [n] and [t] are not just in complementary distribution among different morphemes ([n] after nasal vowels, [t] after oral vowels) but actually vary in different occurrences of the same morpheme. The choice of the oral or nasal (or lenited, cf. (26), below) allophone of the final segments /P, T, K/, is thus a phonetic process which takes the phonological forms (with absolute /m/-initial or general /p/-initial allomorphs) as its input. This behavior of /T/ is also a good argument that phonetically nasal vowels in position V2 [such as [ũ] in [p󰀟̃ˈʐũŋ] in (19)] are intrinsically nasal (/pɨ̱zũK/) even before morpheme-final nasal segments [m, n, ŋ]. Alternative analyses could try to avoid abstract archiphonemes /P, T, K/ and propose phonetic nasalization of vowels V2 before intrinsically nasal final consonants C3, which would result in forms like /o̱ -pɨ̱zu̱ ŋ/ (instead of /o̱ -pɨ̱zũK/) and oral counterparts /po̱ á̱t/ (instead of /po̱ áT/). But, in such an analysis, even if it could be explained why an intrinsically oral final /t/ would be nasalized in forms like (22d), why does nasality then not also spread to the preceding vowel [not even in (21d)]? The following examples show the same phenomenon for verbs with a final /K/ and again a final /T/, but in both cases with a thematic vowel -e̱ already in the simple verb: móK·e̱·tu ‘to burst’ and moK·móK·e̱·tu ‘(for several) to burst (and/or at several places),’ in (23), and móT·e̱·tu ‘to jump’ and moT·móT·e̱·tu ‘to bounce, to jump repeatedly,’ in (24). (23) a. ɔˈpɔɣɛ o̱ -póK-e̱ 3-burst-tv ‘It burst.’

b. ɔpɔk̚ˈpɔɣɛ o̱ -poK~póK-e̱ 3-burst~burst-tv ‘Many burst,’ ‘It burst several times, at different places.’

c. ˈmɔɣɛtu d. mɔŋˈmɔɣɛtu móK-e̱-tu moK~móK-e̱-tu burst-tv-abs burst~burst-tv-abs ‘to burst’ ‘(for several) to burst,’ ‘to burst (in several places)’ (24) a. ɔˈpɔɾɛ b. ɔpɔtˈ̚ pɔɾɛ o̱ -pó̱ T-e̱ o̱ -poT~póT-e̱ 3-jump-tv 3-jump~jump-tv ‘He jumped.’ ‘He bounced (jumped around).’

204

drude

c. ˈmɔɾɛtu d. mɔnˈmɔɾɛtu móT-e̱-tu moT~móT-e̱-tu jump-tv-abs jump~jump-tv-abs ‘to jump’ ‘to bounce (to jump repeatedly)’ In the preceding three sections 5, 6, and 7 we demonstrated the patterns of reduplication for active verbs, focussing on intransitive active verbs. The next section briefly discusses reduplication with another major class of verbs: stative verbs.

8

Reduplication with Stative Verbs

Stative verbs in Awetí (as in other Tupí languages) mainly express concepts which in many other languages are expressed by adjectives. They differ formally from active verbs by several morphological properties, especially with regard to person prefixes (Drude 2008). They also show an obligatory thematic vowel -e̱ before the aspectual sufixes. In the absence of aspectual suffixes, the element -(z)ã occurs before most modal suffixes (described as allomorphy of the suffixes in Drude 2011b). Most relevant for this study is the fact that stative verbs also show a different behavior with respect to the thematic vowel -e̱ in the case of reduplication. These phenomena will be demonstrated and discussed in this section. A typical stative verb in Awetí is t󰀟́P·ã·tu ‘to be many’ (the absolute form is built with the element -(z)a. If the suffix -tu occurs after aspect suffixes, -(z)ã does not appear). We show the simple third person indicative perfective, imperfective and progressive forms. As can be seen, forms of stative verbs with aspect suffixes show the thematic vowel -e̱. The prefix for 3rd person is i̱- on a consonant-initial stative stem. (25) a. iˈtɨp̚ b. iˈtɨβɛʐɔkɔ c. itɨˈβɛju i̱-t󰀟́P i̱-t󰀟́P-e̱-zo̱ ko i̱-tɨP-é̱-ju26 3-be.many 3-be.many-tv-ipfv 3-be.many-tv-prog ‘They are many.’ ‘They are about to ‘They are becoming be many.’ more.’

26

Interestingly, the word accent in the progressive forms seems to fall on the syllable with the thematic vowel -e̱, or at least it varies between this syllable and the (expected) preceding syllable.

reduplication as a tool for analysis in awetí

205

Reduplication with stative verbs differs from reduplication with active verbs not only semantically (having an attenuative effect) but also in that the perfective forms (unmarked for aspect) do not receive the thematic vowel -e̱. Besides this, the reduplicated stems behave just like other stems of stative verbs. We show in (26) a perfective finite form and the absolute form of the reduplicated verb tɨP·t󰀟́P·ã·tu ‘to be relatively (i.e., more or less, not very) many.’ (26) a. itɨp̚ˈtɨp̚ b. tɨp̚ˈtɨβãntu i̱-tɨP~t󰀟́P tɨP~t󰀟́P-ã-tu 3-be.many~be.many be.many~be.many-zã-abs ‘They are relatively many.’ ‘to be relatively many’ Also the verb mo̱ tɨj·mo̱ t󰀟́j·ã·tu ‘to be a little heavy,’ reduplicated from mo̱ t󰀟́j·ã·tu ‘to be heavy’ (p/m alternation), does not have the -e characteristic for active reduplicated verbs: the simple perfective 3rd person forms are i̱-po̱ t󰀟́j and i̱po̱ tɨj~po̱ t󰀟́j, respectively. The same holds for the corresponding forms of ta̱ ʔóK·ã·tu ‘to be angry,’ a stative verb with a final /K/, and its reduplicated counterpart ta̱ ʔoK·ta̱ ʔóK·ã·tu with, according to some speakers, the (rather active) meaning ‘to quarrel (without good reason), to squabble’ instead of or in addition to ‘to be somewhat angry.’

9

The Morphological Domain of Reduplication

So far we have seen reduplication applied only to a morphologically simple (‘underlying’ or ‘proper’) stem. No inflectional affixes are included in reduplication in Awetí, not even the thematic vowel -e̱ or the morph -zã, both arguably part of the complete stem of the verb. This poses the question of the morphological domain of reduplication—the behavior of complex (derived and composed) stems with respect to reduplication. Are there any morphs that can be reduplicated together with the stem? And what does this possibly reveal about the nature of these morphs? The causative prefixes mo̱ - and ((z)e̱)z(o̱ )- are examples of derivative affixes which are reduplicated together with the original stem.27 Both derive transitive from active intransitive verbs. Consider (27) which involves the simple verb kúj·e̱·tu ‘to fall down.’

27

In other words, the complete stem of the derived (transitive) verb is reduplicated, not only the stem of the underlying intransitive verb.

206

drude

(27) a. ɔˈkujɛ b. wɛ̃ɲmɔˈkujɛ c. wɛ̃ɲmɔˌkujmɔˈkujɛ o̱ -kúj-e̱ we̱j-mo̱ -kúj-e̱ we̱j-mo̱ -kuj~mo̱ -kúj-e̱ 3-fall-tv 3-caus-fall-tv 3-caus-fall~caus-fall-tv ‘He fell down.’ ‘He dropped ‘He let several (things) fall (something).’ down.’ The concomitative-causativizer ((z)e̱)z(o̱ )- is the Awetí affix with the most intriguing allomorphic variation known so far. Its allomorphs vary depending on the segment (consonant or vowel) that follows it and also according to the preceding person prefix (subject-marking or object-marking), as summarized in Table 5, below. table 5

Allomorphs of ((z)e̱)z(o̱ )-

Following segment: Consonant Vowel Person prefix: Subject marking Object marking

zo̱ e̱zo̱ -

ze̱ze̱z-

If the meaning of the original intransitive verb is ‘X does A,’ then the meaning of the derived transitive verb is ‘X does A and causes Y to participate in the same action A (together with X).’ In reduplication, the prefix ((z)e̱)z(o̱ )- is copied with the root. Interestingly, the right hand copy of the reduplication always contains the allomorph used for object-marking prefixes, even in subject-centred forms. The form of the prefix in the left hand copy depends on the prefix type, as in the simple verb. This is illustrated in (28) for the verb n·e̱zo̱ ·tó·tu ‘to take [(sth.) with oneself],’ derived from the very common Awetí verb tó·tu ‘to go/leave’ with the consonant-initial stem tó [cf. (1), above]. (28) a. wɛjʐɔˈtɔ b. itɛʐɔˈtɔ we̱j-zo̱ -tó i̱t-e̱zo̱ -tó 3-cocaus-go 1.obj-cocaus-go ‘Hei took it with himselfi’ ‘Hei took me with himselfi’ c. wɛjʐɔˌtɔɛʐɔˈtɔ we̱j-zo̱ -to~e̱zo̱ -tó 3-cocaus-go~cocaus-go ‘Hei took each of many things with himselfi’

reduplication as a tool for analysis in awetí

207

It seems odd that the reduplicated form is not just *we̱j·zo̱ ·to·zo̱ ·tó, simply copying the prefix allomorph zo̱ - as it occurs in the non-reduplicated verb, together with the original stem tó. In terms of a sequence of processes, one might suggest that the allomorph e̱zo̱ - is the ‘default’ or ‘underlying’ one in this case, from which zo̱ - is derived by elision of the initial /e̱/. The choice of the allomorph in the left hand copy is then something which would happen ‘later’ than reduplication. The same happens with vowel-initial stems: the right hand copy in the reduplicated form presents the allomorph of the prefix which usually occurs after object-marking prefixes. Compare the examples in (29), completely analogous to the forms in (28), but now with forms of n·e̱z·úT°u ‘to bring,’ derived from the simple verb t·úT·°u [ˈtutu] ‘to come’ (stem: úT). (29) a. wɛjʐɛˈʐut ̚ we̱j-ze̱z-úT 3-cocaus-come ‘Hei brought (something) [with himi].’

b. itɛˈʐut ̚ i̱t-e̱z-úT 1.obj-cocaus-come ‘Hei brought me [with himi]’

c. wɛjʐɛˌʐuɾɛˈʐuɾɛ we̱j-ze̱z-uT~e̱z-úT-e̱ 3-cocaus-come~cocaus-come-tv ‘Hei brought each of many things [with himi].’ In this case, the right hand copy of the prefix is e̱z-, an allomorph which is shorter, not longer, than ze̱z-, the allomorph occurring in the left hand copy (as expected after a subject-marking prefix). So instead of elision of an initial /e̱/, we would have epenthesis of an initial /z/ due to the preceding person prefix, in the left hand copy, and in the not reduplicated verb. However these facts might be analyzed (e.g. in terms of underlying forms, reduplication as affixation to the left or right, or in projection terms etc.), it seems clear that reduplication in Awetí is described best in morphological rather than phonological terms. Interestingly, the reduplication process ‘has access’ to the underlying morphemes so that, in the reduplicated verb, allomorphs may surface which are not present in the simple verb.28

28

This fact weakens the argument against the hypothesis (a) (presented earlier in the discussion of example (4)) that the /t/ between the subject person prefix and the proper stem of a transitive verb may be part of the stem. The allomorph of the stem without the

208

drude

So far, complex verbs formed by derivation reduplicate their whole stem, including a derivational prefix. The same holds for most derivational suffixes. For instance, transitive verbs (with a meaning ‘X does A to Y’), may be causativized by a suffix -°u̱ káT. The derived verb has a meaning ‘Z makes/lets X do A to Y.’ Interestingly, this suffix attracts word accent, which is evidence that it is part of the new stem of the derived bi-transitive verb. Indeed, even this disyllabic suffix is consistently fully reduplicated with the original stem. See the examples in (30). (30) a. wɛjˈtup̚ we̱j-túP 3-see ‘He saw [something].’

b. wɛjˌtupuˈkat ̚ we̱j-túP-°u̱ káT 3-see-caus ‘He made/let [someone] see [something].’

c. wɛjtupuˌkatupuˈkaɾɛ we̱j-tuP-°u̱ kaT~túP-°u̱ káT-e 3-see-caus~see-caus-tv ‘He made/let [someone] see [each of many things].’ This reduplication of original stem plus -°u̱ káT is independent of the number of syllables of the root. In the case of nã·mi̱ʔĩK·°u ‘to appear (to someone),’ a simple verb with a disyllabic stem, the stem of the reduplicated verb with the meaning ‘to show each of several things to someone’ is mi̱ʔĩK-°u̱ kaT~mi̱ʔĩK-°u̱ káT-e̱ [mĩʔĩŋkuˌkanmĩʔĩŋkuˈkaɾɛ]. For stative verbs (semantics: ‘X has [property] P,’ for instance lo̱ lé·zã·tu ‘to be bad/ugly’), there are two derivational suffixes; both carry the lexical accent of the resulting word. One, the resultative -ʔáT, derives intransitive active verbs (‘X acquires P’). This suffix is, again, included in reduplication. As the derived verb lo̱ le·ʔáT·°u ‘to become bad/ugly’ is an active verb, the final thematic vowel -e̱ is present in the reduplicated verb lo̱ le·ʔaT·lo̱ le·ʔáT·e̱·tu ‘(several things/everything) to become bad.’ Finally, the causative suffix -ká derives transitive verbs from stative verbs (‘X causes Y to acquire property P’). However, unexpectedly, this suffix is not reduplicated with the stem. Compare the examples in (31).

/t/ could be the “underlying” allomorph used for reduplication, and the left hand copy could exhibit the allomorph determined by the environment (following a subject person prefix).

209

reduplication as a tool for analysis in awetí

(31) a. wɛjlɔlɛˈka b. wɛjlɔˌlɛlɔlɛˈka we̱j-lo̱ le-ká we̱j-lo̱ le~lo̱ le-ká 3-bad-caus 3-bad~bad-caus ‘He spoiled it.’ ‘He spoiled several things, each of them.’ However, the semantics of lo̱ le·lo̱ le·ká·tu ‘to spoil (each of) several things’ (plural objects) is that of reduplication of the transitive verb lo̱ le·ká·tu ‘to spoil,’ not that of causativation of the reduplicated stative verb lo̱ le·lo̱ lé·zã·tu ‘to be more or less bad/ugly,’ which would yield the meaning ‘to make something more or less bad.’ So, semantics follow the expected pattern observed with the other derivational affixes; but form and semantics go different ways, as illustrated in Table 6.29 table 6

(a) formal and (b) semantic derivations combining -ka and reduplication

(a)

(b)

lo̱ lé

redupl.→ lo̱ le·lo̱ le

‘bad’

↓caus↓

↓caus↓ ↓caus↓

↓caus↓ lo̱ le·lo̱ le·ká

lo̱ le·ká

redupl.→ red(‘bad’)

redupl.→ lo̱ le·ka·lo̱ le·ká

caus(red(‘bad’)) caus(‘bad’) redupl.→ red(caus(‘bad’))

The odd semantics of lo̱ le·lo̱ le·ká·tu can be seen as an argument that -ká is a regular derivational affix which happens to show a formal (but not semantic) irregularity not to be copied in reduplication. Indeed, besides this fact there is nothing that would point to another analysis of -ká than that of a derivational affix; it is certainly not an inflectional affix.30 The semantic effect is perfectly

29

30

We restrict the presentation of the formal aspects to the stems, and indicate the semantics in functional notation: red(‘bad’) = ‘to be more or less bad’—caus(red(‘bad’)) = ‘to make something more or less bad’—caus(‘bad’) = ‘to spoil’—red(caus(‘bad’)) = ‘to spoil many things.’ The only uncommon feature of -ká (and -ʔáT) is that it does not have a vowel-initial allomorph for use with consonant-final stems (differently from the negation suffix -ka/-ɨka). This suggests rather a lexical origin of -ká (an older verb stem used in compositions?).

210

drude

analogous to that of mo̱ - and -°u̱ káT; it clearly concerns the lexical meaning, not functional categories. Also, -ká attracts word accent, as do -°u̱ káT and -ʔáT. Consequently, reduplication in Awetí formally does not apply to all derivational affixes, but only to most.31 I have been unable so far to identify formal or semantic criteria that would determine which affix is reduplicated and which is not. This means, however, that reduplication unfortunately is not a completely reliable formal criterion for distinguishing inflectional from derivational affixes. As a case in point, consider the two prefixes te̱- ‘reflexive’ and to̱ - ‘reciprocal,’ occurring with transitive stems. They show some properties of derivation. Most importantly, their semantic effect could be described in terms of manipulating the lexical meaning (‘X acts on X himself’ and ‘X acts on Y and Y acts on X’). Then, verb forms with these prefixes take the same person prefixes as intransitive verbs do. These prefixes are illustrated in (32) with the transitive ́ ‘to hurt, to combat, to kill.’ Note that the original verb shows the verb nã·k󰀟̃j·tu transitive third person prefix we̱j- while the forms with te̱- and to̱ - contain the third person prefix o̱ - which otherwise occurs with active intransitive verbs. (32) a. w̃ ɛ̃ɲˈk󰀟̃ɲ b. ɔ̃ ntɛ̃ˈŋk󰀟̃ɲ c. ɔ̃ ntɔ̃ ˈŋk󰀟̃ɲ we̱j-k󰀟̃j́ o̱ -te̱-k󰀟̃j́ o̱ -to̱ -k󰀟̃j́ 3-hurt 3-refl-hurt 3-recp-hurt ‘He hurt someone.’ ‘He hurt himself.’ ‘They hurt one another.,’ ‘They fought.’ The behavior of these prefixes could allow for an analysis as derivational prefixes—they would derive active intransitive verbs from transitive verbs. (In particular, the meaning ‘to combat’ is the most common one with reciprocal forms.) Nonetheless, the analysis so far (e.g. Drude 2008) considers these prefixes to be inflectional. Accordingly, the forms with these prefixes belong to categories ‘reflexive’ and ‘reciprocal,’ and these are names of functional ‘voice’ (or genus verbi) categories. These forms are part of the paradigm of the transitive verb. Reasons for this analysis are that te̱- and to̱ - are highly regular and almost never show idiosyncrasies or effects of lexicalization.32 The lexical semantic effect is 31

32

Another derivational suffix which does not (or not always) reduplicate is -wã ‘completive,’ see (3). Reduplicated verbs with the derivational suffix -túT ‘to want to …’ have not been elicited. The case of ‘to combat’ as a default meaning of the reciprocal form is an atypical exception;

reduplication as a tool for analysis in awetí

211

always transparent and concerns only arguments which are already present in the original lexical meaning. It is comparable, for instance, to the semantics of the passive in Germanic languages. Finally, their interaction with the syntactic arguments can explain the change in person prefixes. Like the passive, they reduce the valency. Interestingly, the forms with te̱- and to̱ - show variation with respect to reduplication: with some stems the prefixes are reduplicated (present in both copies), but with other stems they are not. ́ Consider again the reduplicated forms related to the simple verb nã·k󰀟̃j·°u ‘to hurt, to combat, to kill.’ In this case, the reflexive and the reciprocal prefixes are copied with the original stem k󰀟̃j,́ as seems to be the case with most if not all monosyllabic stems. (33) a. ɔ̃ ntɛ̃ˌŋk󰀟̃ɲtɛ̃ˈŋk󰀟̃ɲɛ̃ b. ɔ̃ ntɔ̃ ˌŋk󰀟̃ɲtɔ̃ ˈŋk󰀟̃ɲɛ̃ ́ ́ o̱ -te̱-k󰀟̃j~te̱-k󰀟̃j-e̱ o̱ -to̱ -k󰀟̃j~to̱ -k󰀟̃j-e̱ 3-refl-kill~rfl-kill-tv 3-recp-kill~recp-kill-tv ‘He hurt himself several times.’ ‘They fought again and again.’ If the forms with te̱- and to̱ - are inflected forms of the simple verb, the two ́ ‘to hurt forms in (33) are inflectional forms of the reduplicated verb nã·k󰀟̃j·k󰀟̃j·°u several times, to combat several times, to kill several beings.’ True, this is the only known case where an inflectional affix is reduplicated, and this seems odd at first sight. (The analysis may also have to face certain problems of a semantic nature.) There are other verbs where the prefixes in question cannot be copied in the left hand copy. For instance, the reflexive form ‘he burned himself several times, at several places,’ based on the transitive verb nã·ʔa̱ p󰀟́·tu ‘to burn,’ is o̱ ·te̱·ʔa̱ pɨ·ʔa̱ p󰀟́, and not *o̱ -te̱-ʔa̱ pɨ~te̱-ʔa̱ p󰀟́, which would be analogous to o̱ -te̱-k󰀟̃j~te̱-k󰀟̃j.́ This variation generally seems to be correlated with the number of syllables, but there are exceptions. The exact conditions for allowing or requiring (or not) reduplication of reflexive/reciprocal prefixes together with the original stem have not yet been identified. They may be partially lexicalized (idiosyncratically), and it seems that there is also some variation from speaker to speaker.

still, also the non-reciprocal forms can mean ‘to attack.’—Note that these prefixes are the only known way to express reflexivity/reciprocity in Awetí; there are no independent pronominal reflexive forms or other reflexive morphemes. The same prefixes also occur with prepositions.

212

drude

Thus, the fact that forms with te̱- and to̱ - share properties of both derivation and inflection is mirrored by the variable behavior of these prefixes with respect to reduplication. A similar situation holds for incorporation. Usually, incorporation derives intransitive verbs from transitive ones by inserting before the verbal stem a noun which refers to the object of the action. This is not a very productive process, often with idiosyncratic (lexicalized) meanings. Only few nominal elements can be incorporated, for instance body parts such as po ‘hand,’ or mo̱ ʔáT ‘person, human being.’ For example, the latter is used to derive the intransitive verb mo̱ ʔáT·a̱ ʔó·tu ‘to swear’ from the transitive verb n·a̱ ʔó·tu ‘to curse.’ When reduplicated, the incorporated part is not copied with the root, as is shown in (34). (34) a. wɛjaˈʔɔ b. wɛjaˌʔɔaˈʔɔ c. ɔ̃ mɔˌʔaɾaˌʔɔaˈʔɔ we̱j-a̱ ʔó we̱j-a̱ ʔo~a̱ ʔó o̱ -mo̱ʔaT-a̱ ʔo~a̱ ʔó 3-curse 3-curse~curse 3-person-curse~curse ‘He cursed.’ ‘He cursed a lot, ‘He swore a lot, about/at several people.’ several people.’ The form *o̱ -mo̱ ʔáT-a̱ ʔo~mo̱ ʔáT-a̱ ʔó, which theoretically could exist along with (34c), is ungrammatical. On the other hand, there are cases where the incorporated part is reduplicated. For instance, some intransitive verbs are based on the transitive verb nã·ʔú·tu ‘to ingest (eat, drink, inhale, …)’: kaT·ʔú·tu ‘to eat, to have a meal’ (cf. káT ‘thing, spirit’), ʔɨ·ʔú·tu ‘to drink, to satisfy one’s thirst’ (cf. ʔ󰀟́ ‘water, liquid’), pe·ʔú·tu ‘to smoke, to inhale tobacco’ (cf. pé ‘tobacco’). In contrast to several other verbs with incorporation, with these verbs, the nominal element is reduplicated together with the root. See (35) for forms of nã·ʔú·tu and kaT·ʔú·tu. The third form of the reduplicated verb kaT·ʔu·kaT·ʔú·tu shows its rather specific inchoative meaning (or, at least, that is the default reading), which suggests lexicalization. (35) a. wɛjˈʔu b. ɔkaɾˈʔu c. ɔkaɾˌʔukaɾˈʔu we̱j-ʔú o̱ -kaT-ʔú o̱ -kaT-ʔu~kaT-ʔú 3-inɡest 3-thing-ingest 3-thing-ingest~thing-ingest ‘He ate/drunk/ ‘He had a meal.’ ‘He started little by little smoked.’ to eat again.’ There are two main possible explanations for this divergent behavior of these verbs: (a) They are rather old (also reconstructed for Tupí-Guaraní, see e.g.

reduplication as a tool for analysis in awetí

213

Mello 2000) and hence possibly lexicalized, so that presently the stem is not analyzed any more, although morphologically transparent. (b) The verbal root ʔú ‘ingest’ is monosyllabic, and so are the incorporated nouns. In sum, the morphological domain of reduplication is the stem. In the case of complex stems, generally the whole (derived or composed) stem is reduplicated, but there are exceptions. Some, but not all of these can be explained by lexicalization or number of syllables. On the other hand, the inflectional voice affixes te̱- and to̱ - are sometimes reduplicated together with the stem, which coincides with the fact that they are in some respects similar to derivational affixes.

10

Conclusion

Besides describing the facts of reduplication in Awetí, this study showed that reduplication may provide a useful criterion for answering phonological and morphological questions. The major findings based on reduplication data presented here were the following: – The analysis of morpheme-final non-glide consonants as abstract archiphonemes due to the neutralization of the oral–nasal contrast has been confirmed (otherwise oral segments may be realized as nasal allophones before nasal consonants). – The analysis of inherently oral vowels in position V2 [see patterns A] has been strengthened (these vowels are phonetically oral even between nasal consonants). – The word-initial alternation of /p/ (inflected forms) and /m/ (absolute forms) is a morphological, not a phonological one (the /m/ is present in both copies). – The unstressed final elements -e̱ and -(z)ã that appear in certain simple verbs are morphs that do not belong to the proper stem. (They are never reduplicated, and -e̱ occurs in many reduplicated verbs.) Interestingly, the same morphs appear in certain inflected forms of stative verbs. With respect to reduplication itself, the following two facts may be the most interesting: – In reduplicated verbs, shorter and longer allomorphs that are not present in the simple verb, may occur without a phonological explanation. See examples (28) and (29).

214

drude

– When combining derivation and reduplication, form and semantics may diverge with respect to the ordering of applying the respective transformations. See example (31) and Table 6.

Acknowledgments The general knowledge of Awetí underlying this study is a result of more than ten years of work on Awetí, including a total of around fourteen months of fieldwork. From 2001 to 2005, the project was supported by the Volkswagen Foundation within the DOBES (Documentation of Endangered Languages) program, focusing on documenting the language and aspects of the culture. From 2008 on, my research has been supported by a Dilthey-fellowship which again allowed me to do fieldwork and to invite speakers to the Museu Goeldi in Belém. Reduplication was studied, and the specific data used in this study was elicited during a field trip in April/May 2009. I am very grateful for the support by the funding agencies, the Museu Goeldi and my Awetí teachers. I also owe thanks to several reviewers of earlier versions of this paper, in particular to an anonymous reviewer and the editors. Remaining shortcomings are mine alone.

References Drude, Sebastian. 2002. Fala masculina e feminina em Awetí. In Línguas indígenas brasileiras: Fonologia, gramática e história 1, ed. Ana Suelly Arruda Câmara Cabral and Aryon Dall’Igna Rodrigues, 177–190. Belém: Ed. UFPA. . 2006. On the position of the Awetí language in the Tupí family. In Guaraní y “Mawetí-Tupí-Guaraní.” Estudios históricos y descriptivos sobre una familia lingüística de América del Sur, ed. Wolf Dietrich and Haralambos Symeonidis, 47–68. [Regionalwissenschaft Lateinamerika 11]. Berlin & Münster: LIT Verlag. . 2008. Tense, aspect and mood in Awetí verb paradigms: analytic and synthetic forms. In A world of many voices: lessons from endangered languages, ed. David K. Harrison, David Rood and Arienne Dwyer, 67–109. Amsterdam/Philadelphia: John Benjamins. . 2011a. Nominalization and subordination in Awetí. Amérindia 35 [Special issue La structure des langues amazoniennes II, ed. Francesc Queixalós and Leo Wetzels,] 189–218. . 2011b. Word accent in Awetí and its manifestation. Amérindia 35 [Special issue La structure des langues amazoniennes II, ed. Francesc Queixalós and Leo Wetzels,] 7–40.

reduplication as a tool for analysis in awetí

215

Halle, Morris. 2008. Reduplication. In Foundational issues in linguistic theory. Essays in honor of Jean-Roger Vergnaud, ed. Robert Freidin, Carlos Peregrín Otero, Maria Luisa Zubizarreta and Jean-Roger Vergnaud, 325–357. [Current Studies in Linguistics 45]. Cambridge, MA: MIT Press. Jensen, Cheryl Joyce S. 1998. Comparative Tupí-Guaraní morphosyntax. In Handbook of Amazonian languages 4, ed. Desmond C. Derbyshire and Geoffrey K. Pullum, 487–618. Berlin/New York: Mouton de Gruyter. Meira, Sérgio, and Sebastian Drude. forthcoming. A preliminary reconstruction of Proto-Mawetí-Guaraní segmental phonology. Boletim do Museu Paraense Emílio Goeldi. Ciências Humanas. [Special issue Variação em línguas Tupi: Genealogia, mudança linguística e tipologia, ed. Wolf Dietrich and Sebastian Drude.] Mello, Augusto A.S. 2000. Estudo histórico da família Tupí-Guaraní: aspectos fonológicos e lexicais. PhD diss., Universidade Federal de Santa Catarina. Raimy, Eric. 2000. The phonology and morphology of reduplication. [Studies in generative grammar 52]. Berlin/New York: Mouton de Gruyter. Reiss, Charles, and Marc Simpson. 2009. Reduplication as projection. Concordia University. http://0branch.com/project/redup (accessed September 2010). Rodrigues, Aryon Dall’Igna. 1953. Morfologia do verbo Tupí. Letras 1: 121–152. Rodrigues, Aryon Dall’Igna, and Wolf Dietrich. 1997. On the linguistic relationship between Mawé and Tupí-Guaraní. Diachronica XIV/2: 265–304. Rose, Françoise. 2005. Reduplication in Tupi-Guarani languages: going into opposite directions. In Studies on reduplication, ed. Bernhard Hurch with Veronika Mattes, 351–368. Berlin/New York: Mouton de Gruyter. . 2007. Action répétitive et action répétée: aspect et pluralité verbale dans la réduplication en émérillon. Faits de Langues 29, La reduplication: 125–143. Seki, Lucy. 2000. Gramática do Kamaiurá, língua Tupi-Guarani do Alto Xingu. Campinas: Editora da Unicamp.

Abbreviations 1 3 abs agnr caus cocaus compl excl incl

first person third person absolute agent nominalizer causative concomitative-causative completive (plural) exclusive (plural) inclusive

ipfv obj prog recp refl result sub subj tv

imperfective object progressive reciprocal reflexive resultative subjunctive subject thematic vowel

chapter 9

Reduplication and Ideophones in Trumai Raquel Guirardello-Damian

Trumai, a genetically isolated indigenous language spoken in Brazil, exhibits several instances of reduplication, which can be grouped in two main types: full or partial. The cases of partial reduplication involve the copying of one or two light syllables, depending on the size of the stem, with the reduplicated material prefixed to it. In the cases of full reduplication, the configuration of the stem does not seem to allow partial reduplication to take place; thus, the stem as a whole is copied. This pattern accounts for the majority of the occurrences of reduplication found in the language, but there are some exceptional cases that require a separate treatment. Reduplication in Trumai occurs mainly with verbs and adjectives, although it can also be observed with some nouns (or in the composition of nominal compounds). It is used to express a range of meanings, such as intensity or repetition of action. There are instances of partial reduplication in nouns that do not present a non-reduplicated counterpart; for these cases, the semantic effect is not clear. The present paper aims to investigate the instances of reduplication attested in the language, analyzing mainly their form (although some observations about their function are also made). In addition, it explores Trumai ideophones, which also involve repetition of syllables but in a different manner from what happens in reduplication.

1

Introduction

1.1 Preliminaries The focus of the present paper is to explore the phenomenon of reduplication in Trumai. The language exhibits many cases of reduplicated forms, occurring in words of different sizes, different syllabic configurations, and belonging to various word classes. The intention of the study is to analyze the principles that govern reduplication in this language. It also explores Trumai ideophones, which are other elements that exhibit repetition of syllables. As we will see, ideophones present some similarities to reduplicated forms, although they seem to have a different nature.

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_010

218

guirardello-damian

Section 1.2 offers a brief description of the Trumai language, presenting some relevant typological features. Section 2 investigates reduplication, focusing on its formal aspects, and relates them to general claims in the linguistic literature. The section is subdivided according to word class: verbs and adjectives (2.1 and 2.2), nouns (2.3), other classes (2.4). For each class, there is a brief description of the semantic effects produced by the reduplication. In 2.5, there is an overview of these effects, providing some generalizations and analyzing the place that reduplication has in the structure of the language (i.e., how it relates to other areas of the grammar). Section 3 is dedicated to instances of inherent reduplication attested in the language, which represent interesting cases. Section 4 outlines the study of ideophones, describing and examining their shapes, meanings and uses. Finally, section 5 summarizes the facts observed in Trumai and further analyzes them taking into consideration theoretical and typological issues. 1.2 The Language: General Features The Trumai people constitute a Brazilian indigenous group who live in the Xingu Indigenous Reserve, which is also inhabited by other ethnic groups. The Trumai population is concentrated mainly in three villages: Três Lagoas (from the former village known as Terra Preta), Boa Esperança, and Steinen. There are also families living in other places within the reserve or in nearby cities; these families maintain contact with their relatives in the villages and often visit them.1 Genetically speaking, the language is considered an isolate, i.e., despite some studies, its genetic affiliations are still unclear. The basic word order is SOV, with possible variations depending on pragmatic factors. There is not much inflectional morphology, and words usually consist of a single morpheme. The language has four open classes: nouns, verbs, adjectives, and adverbs. The sound system contains six vowels and twenty-three consonants:

1 The number of people living in the villages varies from year to year since the families have been moving around in the last few years. The most recent official count was 64 individuals (Unifesp, 2010). The number of speakers is smaller: there are 50 people who can speak Trumai, with different degrees of knowledge about the language. They are also bilingual in Portuguese (the official language of Brazil). The other members of the community speak other Xinguan languages and/or Portuguese.

reduplication and ideophones in trumai

219

a) Vowels: /i/, /e/, /ɨ/, /a/, /u/, /o/ b) Consonants: stops: /p/, /t̪ /, /t/, /d/, /k/, /ʔ/ ejective stops: /t̪’/, /t’/, /k’/ affricate: /ts/ ejective affricate: /ts’/ fricatives: /f/, /s/, /ʃ/, /x/, /h/ nasals: /m/, /n/ liquids: /l/, /ɬ/, /ɾ/ glides: /w/, /j/ The consonants /ts/, /ts’/, /t̪’/, /t’/ and /k’/ are all analyzed as single phonemes rather than consonant clusters. Trumai seems to be the only language in the Xingu reserve to have ejective consonants since other languages in the region show no evidence of such sounds (Emmerich 1980; Fargetti 1992; Seki 2000; Dourado 2001). Stress is fixed, always falling on the final syllable of the word. With regard to syllable patterns, there are four types: CV, CVC, V and VC. The most frequent ones are CV and CVC. As examples, we have:2 (1)

a. lo ‘a kind of bird’ b. ʃu.da ‘make’

CV

(2)

a. hid ‘arrow’ b. lat' ‘meat’ c. pits' ‘foot’

CVC

(3)

a. a.o b. o.ke

‘father’ V ‘medicinal herb’

(4)

a. ap b. el c. oɬ

‘collect’ ‘lover’ ‘sleep’

VC

2 Symbols of the International Phonetic Alphabet (I.P.A.) are employed for the data in this article. The Trumai words are presented in their phonological form.

220 2

guirardello-damian

Reduplication

Reduplication in Trumai occurs mainly with verbs and adjectives, although there are also some attested cases with nouns, adverbs, and numerals. It can be subdivided in two types: full or partial. Full reduplication is the default, occurring with a high number of lexical items. Partial reduplication is more restricted and only occurs under certain conditions. Some examples are: partial (5) a. pen b. pe~pen (6)

full (7)

(8)

‘vomit’ ‘vomit many times’

a. pumat ‘scream’ b. puma~pumat ‘scream many times/a lot’

a. sapku b. sapku sapku

‘swing’ ‘swing a lot’

a. make b. make make

‘bite’ ‘chew (lit. bite many times)’

The cases of partial reduplication involve the copying of one or two light syllables (i.e., syllables with one mora3). The onset and nucleus are copied, but not the coda, and the reduplicated material is prefixed to the stem. The number of syllables copied depends on the size of the stem: if it is monosyllabic, one syllable is copied (9). If it has two or more syllables, two syllables are copied, and this is the maximum allowed (10–11). one light syllable copied (9) a. pi~piɾ ‘sweep many times/a lot’ b. ɬa~ɬat ‘laugh many times/a lot’ c. xo~xom ‘suck many times’

3 For the relationship between number of moras and syllable weight, cf. Marantz (1982).

reduplication and ideophones in trumai

221

two light syllables copied (10) a. uja~ujaɾ ‘stick a lot’ b. kɨt̪ɨ~kɨt̪ɨw ‘rub many times/a lot’ c. ole~olem ‘cook a lot’ (11) kaɾa~kaɾain ‘scratch many times’ In which environment does partial reduplication take place? I addressed this question in previous works (Guirardello 1999, Guirardello-Damian 2005), but further investigation of the issue revealed that the state of affairs is actually more complex than what I originally described. Various factors interact and influence the occurrence of partial reduplication: – – – –

word class; size of the stem (i.e., number of syllables); whether the stem contains a heavy syllable (i.e., a syllable with a coda); position of the heavy syllable in the word (whether it is final or not).

Let us examine these points in more detail. 2.1 Verbs and Adjectives Reduplication in adjectives usually indicates intensity (12), but sometimes it can also indicate a repeated attribute, i.e., an entity has a feature that is repeated (13). In verbs, it is used to express actions performed with intensity or repeated various times (14). In some cases, it can indicate that the action is performed by many individuals, although such use is not very common (15c). All these possible uses (intensity, repetition, number) confirm a tendency already observed in other languages, in which reduplication is employed for similar reasons (Rubino 2005b, Adelaar 2000, Meira in prep., Kemmer 1996). (12) a. pet̪'ek ‘sticky’ b. pet̪'e~pet̪'ek ‘very sticky’ (13) ʃaʃxo t̪oɾe t̪oɾe drawing white white ‘white stripes’ (the feature—white line—is repeated) (14) a. lat' ‘lie’ b. la~lat' ‘lie many times/a lot’

222

guirardello-damian

(15) a. aʃ󰀟kida-n ale. jump-3Abs hearsay ‘He jumped.’ b. wan aʃ󰀟kida-n ale. pl jump-3Abs hearsay ‘They jumped.’4 c. wan aʃ󰀟kida aʃ󰀟kida-n ale. pl jump jump-3abs hearsay ‘Lots of them jumped.’ (lit. lots, they jumped) For monosyllabic verbs, partial reduplication occurs with stems that exhibit a CVC pattern (16). However, this only occurs for particular lexical items, since there are also CVC stems that undergo total reduplication (17). For other syllable types (CV, VC), total reduplication is the preferred process (18–19).5 (16) a. la~lat' ‘lie a lot’ b. wa~wal ‘sing a lot’ c. ka~kan ‘have sex many times’ (17) a. kis kis ‘steal many times’ b. dal dal ‘search many times’ c. lan lan ‘scratch many times’ (18) a. fa fa b. ʃo ʃo

‘hit many times’ ‘strike many times’

(19) a. ap ap b. oɬ oɬ

‘collect many times’ ‘sleep a lot’

With regard to monosyllabic adjectives, there is only one attested case of partial reduplication, which occurs with a CVC stem (20). In contrast, there are many instances of full reduplication of adjectives (21).

4 Trumai has two morphemes that indicate plurality: a ‘dual’ and wan ‘plural.’ The language also has quantifiers, such as aʔdɨ ‘many (countable),’ pɨx ‘much (non-countable),’ jupun ‘all,’ etc. 5 There are no stems consisting only of V that undergo reduplication.

223

reduplication and ideophones in trumai

(20) t̪a~t̪aɾ

‘very pot-bellied’

(21) a. kod kod ‘very rounded’ b. pom pom ‘very tasty’ c. mox mox ‘very swollen’ The situation is different for disyllabic verbs and adjectives. This is where partial reduplication occurs in a productive way. When the final syllable of the stem is heavy and the previous one is light, partial reduplication takes place: (22) a. b. c. d.

pupa~pupan laxo~laxod faʃ󰀟~faʃ󰀟ʃ pudi~pudits'

‘roll up many times’ ‘sniff many times/a lot’ ‘feel anger a lot’ ‘like a lot’

(23) a. b. c. d.

paɾa~paɾaw xuɾa~xuɾaɾ tsiɾi~tsiɾiw at̪u~at̪uk

‘very mixed’ ‘very wrinkled’ ‘very spotty’ ‘very long’

Partial reduplication does not occur when the stem has a different configuration: when both syllables are heavy (24) or when the final syllable is light (regardless of the configuration of the previous syllable (25–26)), the stem as a whole is copied, producing full reduplication. (24) a. watkan watkan ‘cry a lot’ b. keftat' keftat' ‘very difficult’ (25) a. enɨ enɨ b. miɾo miɾo

(wat.kan) (kef.tat’)

‘very dirty’ (e.nɨ) ‘chop wood’ (lit. cut cut)6 (mi.ɾo)

6 One could argue that miɾo miɾo is not an instance of full, but rather partial reduplication: since partial reduplication in Trumai involves the copying of light syllables, and given that miɾo has two light syllables, both would be copied, resulting in miɾo miɾo. However, there are also examples, such as lakida lakida ‘stand up many times,’ in which three syllables are copied. Such a case cannot be treated as partial reduplication, given that in this process the maximum of duplicated syllables allowed is two. Thus, in order to treat them uniformly, both miɾo miɾo and lakida lakida are classified as instances of full reduplication. This allows us to

224

guirardello-damian

(26) a. homte homte b. maxke maxke

‘shine a lot’ ‘very expensive’

(hom.te) (max.ke)

Verbs with three or more syllables follow the same pattern: partial reduplication takes place when the final syllable is heavy and the previous ones are light (27). Full reduplication occurs under other conditions: the final and one of the previous syllables are heavy (28) or the final syllable is light (29). However, similarly to what happens with monosyllabic verbs, partial reduplication is not productive here. There are trisyllabic verbs that present the right configuration, and yet they only allow full reduplication (30). In the case of adjectives, there are no attested cases of partial reduplication with words with three or more syllables. Only instances of full reduplication are observed (31). (27) a. ɨɾɨ~ɨɾɨɾɨw b. wake~waket̪an

‘slip many times’ ‘get ready many times’

(28) ankeft̪an ankeft̪an

‘breathe many times’

(29) a. b. c. d.

‘forget many times’ ‘plant many times’ ‘imitate many times’ ‘jump many times’

falpuʃu falpuʃu falomle falomle kefane kefane aʃ󰀟kida aʃ󰀟kida

(30) a. piɾiɾiw piɾiɾiw b. midoxos midoxos

‘roast many times’ ‘call many times’

(31) a. xeɾeɾe xeɾeɾe ‘very wet’ b. miɬikt̪e miɬikt̪e ‘very different’ c. wakadima wakadima ‘very happy’ All these facts can be summarized in Table 1:

propose similar rules for disyllabic and trisyllabic stems that contain only light syllables (i.e., it makes the rules simpler).

225

reduplication and ideophones in trumai table 1

Size of the stem

Reduplication in verbs and adjectives

Partial

Monosyl- Syllable labic pattern

Type of reduplication Full

CVC ka~kan

Syllable pattern

CV fa fa VC ap ap Some CVC stems lan lan

Productivity –

Disyllabic Syllable pattern

Light-Heavy (L-H) Syllable pupa~pupan pattern

Productivity + Trisyllabic Syllable (or more) pattern

Productivity CV and VC: + CVC: – Other than L-H watkan watkan enɨ enɨ homte homte

Productivity +

Light-Light-Heavy Syllable (L-L-H) pattern ɨɾɨ~ɨɾɨɾɨw

Other than L-L-H ankeft̪an ankeft̪an falpuʃu falpuʃu falomle falomle kefane kefane Some L-L-H stems piɾiɾiw piɾiɾiw

Productivity –

Productivity Other than L-L-H: + LLH: –

If we consider disyllabic stems and some of the trisyllabic ones in order to understand how partial reduplication operates, we can see that the last syllable

226

guirardello-damian

of the word is a key element in the process of partial reduplication.7 If it is not heavy, partial reduplication is already blocked. If the final syllable is heavy, then the status of the previous syllables will matter—if they are light, the reduplication takes place, otherwise not. 2.2 Verbs and Adjectives: Exceptions The rules described so far account for the majority of the occurrences of reduplication with verbs and adjectives. However, there are some exceptional cases that require a separate treatment. 2.2.1 Exceptions—1 Three attested cases of reduplication present a peculiar pattern: (32) a. kot̪'kan b. kot̪'ka~kot̪'kan

‘bring together’ ‘bring together many times/many things’

(33) a. jɨkpɨt b. jɨkpɨ~jɨkpɨt

‘disgusting’ ‘very disgusting’

(34) a. waʃoxman ‘bent’ b. waʃoxma~waʃoxman ‘very bent’ According to the general rule that applies to verbs and adjectives, these stems should undergo full reduplication since they do not present the syllable pattern L-H or L-L-H: (35) a. kot̪'.kan H-H b. jɨk.pɨt H-H c. wa.ʃox.man L-H-H However, this is not what happens. Instead, they undergo some sort of partial reduplication, involving the copying of one heavy syllable and one light syllable (or, in the case of waʃoxman, three syllables: light-heavy-light). The result is that the final consonant of the reduplicant is truncated.

7 As mentioned before, the last syllable is also the stressed one. Maybe the fact that this syllable bears stress makes it more prominent and somehow relevant for the process of reduplication. It would be interesting to investigate whether in other languages the weight of the stressed syllable is also a relevant issue for reduplication.

reduplication and ideophones in trumai

(36) a. kot̪'.ka CVC.CV b. jɨk.pɨ CVC. CV

227

+ kot̪'kan

+ jɨkpɨt

c. wa.ʃox.ma + waʃoxman CV. CVC. CV These are the only stems that present such patterns. It is not clear why they behave differently from the other verbs and adjectives. They do not seem to be borrowings, and although jɨkpɨt may possibly be an onomatopoeia, which could account for its atypical behavior,8 the same cannot be said for the other two stems. 2.2.2 Exceptions—2 Adjectives whose configuration is V-CV-CV usually undergo full reduplication since they present the syllable pattern L-L-L rather than L-L-H. For instance: (37) a. at'afa ‘loose’ b. at'afa at'afa ‘very loose’ (38) a. aɾiki b. aɾiki aɾiki

‘strong’ ‘very strong’

However, there is one adjective (39a) and one adverb (40a) whose configuration is also V-CV-CV but whose behavior is atypical. In their process of reduplication, the initial vowel is omitted, i.e., the stem is treated as if it were dissyllabic (CV-CV): (39) a. at̪'ɨxɨ ‘sharp’ b. at̪'ɨxɨ t̪'ɨxɨ ‘very sharp’ (40) a. aloke ‘soon’ b. aloke loke ‘very soon’

8 About this issue, see section 3.2.

228

guirardello-damian

The adjective ataxa ‘hard’ represents another interesting case. In the speech of older speakers, as in (41) below, this adjective follows the same pattern observed in (39–40). Young speakers pronounce the adjective as taxa (42), but when it occurs reduplicated inside a nominal compound (43), they pronounce it as ataxa, like the older speakers (i.e., maybe because the compound is a fixed form9). (41) a. ataxa ‘hard’ b. ataxa taxa ‘very hard’ (42) a. taxa b. taxa taxa

‘hard’ ‘very hard’

(43) home ataxa taxa beetle very.hard ‘species of beetle with a hard skin’ A similar state of affairs is observed with the adjective atoxo ‘bloated.’ Most speakers pronounce it as toxo (44), but when this stem occurs reduplicated inside of a noun formed via nominalization (45), the form is atoxo and follows the same pattern of reduplication of examples (39–40). (44) toxo ha ʃ󰀟. bloated 1 cop ‘I am bloated/full.’ (45) atoxo toxo-k bloated bloated-nzr ‘kind of fruit’ (lit. the one that is very bloated/full)’ What is interesting about these words (at̪'ɨxɨ, aloke, ataxa, atoxo) is that they exhibit a different type of reduplication: the initial vowel is ignored during the copying, i.e., it is opaque to the reduplication process. This pattern is observed only with these four stems (39–45). Other adjectives with the same syllable

9 For compounds, I adopt the definition proposed by Mathews (1997, 66): “a word formed from two or more units that are themselves words.” Example (43) is an instance of this since it is internally formed by the units home (beetle) + ataxa (hard), which are themselves words.

reduplication and ideophones in trumai

229

configuration (V-CV-CV) do not show such behavior, for example, the adjective ariki presented earlier in example (38): (46) a. aɾiki ‘strong' b. *aɾiki ɾiki 2.3 Nouns Reduplication with nouns is used to express large quantities of something. There are only a few attested cases. The majority of the examples involve full copying of the stem (47–49), but there are also instances of partial reduplication (50), following the same general rule that applies to verbs and adjectives (i.e., copying one or two light syllables with stems that present the syllable pattern L-H or L-L-H). (47) a. kut̪u b. kut̪u kut̪u

‘lump’ ‘many lumps’

(48) a. olo b. olo olo

‘fishing spear’ ‘many fishing spears’

(49) a. okomane ‘knot’ b. okomane okomane ‘many knots’ (50) a. ameʃ b. ame~ameʃ10

‘lightning’ ‘many lightning flashes (during a thunderstorm)’

When one tries to elicit reduplicated nouns, speakers reject the data (51–53), which shows that the use of reduplication with this word class is limited. Instead, the language has other devices for indicating number and large quantities; see section 2.5 below. (51) a. *kiki kiki b. *axo~axos

10

‘many men’ ‘many children’

In this example, the reduplication can express the sense of extra speed as well. The lightning flashes come one after the other; thus, there is also a sense that they are quite fast.

230

guirardello-damian

(52) a. *t̪ahu t̪ahu ‘many knives’ b. *muɾi~muɾiɾ ‘many baskets’ (53) a. *asulu asulu ‘many pigeons’ b. *aku~akuts ‘many agutis’ c. *kojo~kojoʃ ‘many bats’ 2.4 Other Classes Besides verbs, adjectives and nouns, reduplication can be found in some numerals and adverbs. It does not seem to be productive in these cases. With the numeral mihin ‘one’ (54), partial reduplication produces a word that functions as a quantifier with the sense of ‘some’ (in opposition to yupun ‘all’). Full reduplication of the numeral huʃ ‘two’ is observed inside of nouns formed via nominalization (55). Other numerals do not present reduplicated forms. (54) mihi~mihin ‘some’ (55) a. huʃ huʃ-t̪'a-k two two-result.of-nzr ‘twins’ (lit. the ones made of two) b. huʃ huʃ-ke two two-nzr ‘two bananas fused together’ Reduplication with adverbs has a more predictable semantic effect, indicating intensity. All the attested cases involve full reduplication (except for aloke ‘soon,’ as mentioned in section 2.2.2). As examples, we have: (56) a. komet̪ani ‘slowly’ b. komet̪ani komet̪ani ‘very slowly’ (57) a. tsia b. tsia tsia

‘well’ ‘really well’

2.5 Semantic and Grammatical Aspects As we have seen, reduplication in Trumai produces different semantic effects depending on the word class. They can be summarized as follows:

reduplication and ideophones in trumai

231

– intensity (for verbs, adjectives, adverbs) – number (for nouns, numerals, sometimes for verbs) – repetition (for verbs, sometimes for adjectives) How marginal or central a position does reduplication occupy in the language? For nouns and numerals, the use of reduplication is very marginal. In fact, the reduplicated forms could be considered as lexicalizations, i.e., only a few nouns/numerals exhibit reduplicative patterns because they are lexicalized, not because a productive process is taking place (it might have been productive in past times but not any longer now). In order to express ‘number’ for nouns, the language rather employs other strategies: the use of pluralizers—a ‘dual’ and wan ‘plural’—and the use of quantifiers such as aʔdɨ ‘many (countable),’ pɨx ‘much (non-countable),’ jupun ‘all,’ etc. For example: (58) a. kiki wan man pl ‘the men’ b. aʔdɨ kiki wan many man pl ‘many men’ c. jupun kiki wan all man pl ‘all men’ For adverbs, the use of reduplication is also limited. To express the idea of intensity, speakers usually employ jumane, the particle of intensity: (59) aleitak jumane lingeringly very ‘very lingeringly’ With adjectives, the use of reduplication is more frequent. It is often observed with adjectives occurring inside of nominal compounds or nominalizations. For example: (60) ik'ada tsiɾi~tsiɾiw-ke leaf red~spotty-nzr ‘kind of plant’ (lit. the one that has a very spotty leaf)

232

guirardello-damian

(61) homa~homat'-ke red~red-nzr ‘kind of piranha’ (lit. the one that is very red) It can also occur with adjectives that are the predicate of a sentence (62). However, in this environment its use is not as common as the use of the particle jumane, which seems to be the preferred way of expressing intensity. The difference between them seems to be a matter of degree or emphasis—the reduplication indicates a slightly more intense attribute or a more emphatic speech (63).11 (62) enɨ enɨ hi ʃ󰀟 dirty dirty 2 cop ‘You are very dirty.’ (63) a. kod jumane ‘very rounded' b. kod kod ‘really rounded’ It is with verbs that reduplication has a more central role. It is frequent in both environments—verbs occurring inside of nominal compounds/nominalizations (64) or verbs as the predicate of a sentence (65). It complements the role of jumane, offering semantic nuances that are not conveyed by the intensity particle—jumane expresses intensity only, while reduplication can also express the idea of repeated action. However, while in some languages verb reduplication can also express certain types of aspect, this is not the case in

11

This difference is also reflected in the Portuguese translation provided by the consultants who are bilingual. In Portuguese, there are various ways of expressing intensity for adjectives and adverbs. One can employ adverbs (muito ‘very,’ bem ‘well/very’), but it is also possible to use the diminutive (in Romance languages such as Spanish and Portuguese, the diminutive can be used as an intensifier, cf. Taylor 1995), although this effect does not work for all adjectives. The diminutive can indicate a slightly stronger quality or a more emphatic speech. For instance, for the adjective igual ‘similar,’ one can express intensity by saying bem igual ‘very similar’ or igualzinho ‘really really similar.’ It is interesting to observe that when the Trumai consultants employ jumane to modify an adjective, they tend to translate it to Portuguese using the adverbs muito or bem, and when they employ reduplication, they tend to translate it using the diminutive (or they combine the adverb bem with the diminutive, to indicate even more emphasis). Thus kod jumane is translated in Portuguese as bem redondo ‘very rounded,’ while kod kod is translated as redondinho / bem redondinho ‘really rounded.’

reduplication and ideophones in trumai

233

Trumai; aspectual information is indicated through the use of auxiliaries, not via reduplication. (64) jaw kɨtɨ~kɨtɨw-kewʃa person red-rub-nzr.instr ‘sponge’ (lit. the thing for one to rub) (65) tsi-ɬe de puma~pumat le hen. 3poss-mother already red-scream hearsay then ‘Then his mother screamed several times.’ In sum, reduplication in Trumai has a more significant role with regard to verbs, less with adjectives (although its presence is still strong) and even less with adverbs. For nouns or numerals, its role is minor and limited to a few forms that are most likely already lexicalized.

3

Further Issues Concerning Reduplication

3.1 Cases of Inherent Reduplication Like other languages of the world, Trumai has several instances of reduplicated forms that do not have corresponding non-reduplicated (simplex) forms. If non-reduplicated counterparts ever existed in the language, they seem to have been lost because the current speakers are not able to identify them. All the attested cases are nouns, and they obey the same rules described earlier for productive reduplication: copying of one/two light syllables for stems with the syllable pattern L-H or L-L-H. Some examples are: (66) a. t̪ama~t̪amain ‘ingá (a kind of fruit)’ b. *t̪amain (67) a. waɾu waɾu b. *waru

‘mangaba (a kind of fruit)’

(68) a. t̪aime t̪aime b. *t̪aime

‘mosquito larva’

3.2 Other Cases of Inherent Reduplication There are some nouns in the language that involve repetition, but this is not due to reduplication, strictly speaking. They are actually onomatopoeias and

234

guirardello-damian

the repetition is rather an attempt to imitate the sound of the object/animal referred to (69). Some of these nouns do not follow the stress patterns of Trumai but have the penultimate syllable as the stressed one, instead of the last one (70). (69) huɾi huɾi

‘whiner’ (an instrument that produces a loud sound)

(70) a. t̪eɾu t̪eɾu ‘a kind of bird (Portuguese name: quero-quero)’ b. piɾu piɾu ‘a kind of bird’ This brings us to the next section, which analyzes the ideophones found in Trumai.

4

Ideophones

4.1 General Characteristics Similar to other Brazilian languages (Seki 2000, Gabas and van der Auwera 2004, van der Meer 1983), Trumai makes extensive use of ideophones. Ideophones also involve repetition of forms, although in a different manner from what happens in reduplication. In this section, the main facts about Trumai ideophones are presented. In section 5, we have a comparision between ideophones and the types of reduplication described in section 2. Ideophones are words employed to present a vivid representation of a particular sensory perception. According to Voeltz and Kilian-Hatz (2001, 3), they possess “a special dramaturgic function,” simulating “an event, an emotion, a perception through language.” They also present a “tendency for iconicity and sound-symbolic behavior,” i.e., they often have an onomatopoeic nature. However, this does not mean that any word that is onomatopoeic is automatically an ideophone. Apart from the phonological shape, other features are also relevant for identifying an ideophone: the sort of meaning conveyed by the word; its environment of occurrence; its syntactic behavior; its internal configuration, etc. In Trumai, ideophones are often found in narratives, although they can occur in other kinds of texts as well (for example, procedural texts). They are used to express an action or movement. Some forms have a consistent meaning; they are not a mere imitation of a sound—that is, although they mimic it, they are also conventionalized forms. It is interesting to observe that some of them are similar to the ones found in Kamaiurá, another Xinguan

reduplication and ideophones in trumai

235

language. This might be due to the influence of Kamaiurá on Trumai culture.12 For instance: Trumai Kamaiurá (71) tsiw tsiw tsiw tsiwk tsiwk tsiwk ‘cut’ (72) tɨk tɨk tɨk tɨk tɨk tɨk ‘go (by walking)’ 4.2 Classification Even though the Trumai ideophones share some behaviors with adverbs and verbs, they seem to constitute a class of their own. Like adverbs, the ideophones occur at the margins of a sentence (beginning or end), as an adjunct.13 However, they do not occur in other syntactic environments observed with adverbs, such as second position or other positions inside the clause (cf. Guirardello-Damian 2005, 239–242). They do not exhibit other characteristics of adverbs either; for instance, they cannot be modified by the particle of intensity ( jumane). Although the ideophones indicate the same sort of meaning that verbs usually express, they do not present the basic features exhibited by verbs: they do not receive the third person absolutive enclitic (-n/-e), they cannot be modified by auxiliaries of aspect,14 and they cannot be modified by the particles of negation (t̪ak) and imperative (wana, waki). One could wonder if the ideophones could be considered a subclass of verbs that are defective with regard to finiteness. Such analysis cannot be supported, for several reasons. First, the language has no verbs of this kind. Any verb in a main clause is finite. In certain subordinate clauses, the verb is non-finite, but in this case there are clear indicators of subordination, and the verb is treated as a possessed noun. This does not happen with the forms classified as ideophones.

12

13

14

The Trumai people were influenced by other Upper-Xinguan groups, having assimilated several features of the common culture that exists in the region. The stronger impact came from the Kamaiurá people, with whom the Trumai had closer links via intertribal marriages and alliances (about this issue, cf. Quain and Murphy 1955, Guirardello-Damian 2011). The influence of Kamaiurá is visible in the lexicon of Trumai (presence of many borrowings) and in the mythic narratives. In this regard, Trumai ideophones match the scenario observed in other languages. As Gabas and van der Auwera (2004, 401) point out, ideophones “tend to occupy a position at the ‘skirts’ of the sentence.” Tense is not marked in the verb, but it is rather expressed at the sentence level by adverbs or focus/tense particles. In the absence of those particles, tense is understood from context.

236

guirardello-damian

Second, verbs never occur in the same conditions that ideophones do. Let us examine an example containing an ideophone at the end of the sentence: (73) wan kaʔʃ󰀟-n ale hen tɨk tɨk pl walk-3abs hearsay then go.walking go.walking tɨk. go.walking(ideoph) ‘Then they went walking.’ A verb would never be found in a similar constellation, i.e., after the adverb hen, without the third person absolutive enclitic and the pluralizer (which is necessary here since the subject is plural). See the contrast between the example above and this one: (74) a. wan sone-kma-n hen wan sa-n. pl drink-perf-3abs then pl dance-3abs ‘Then they finish drinking and they dance.’ b. *wan sone-kma-n hen sa. pl drink-perf-3abs then dance As we can see in (74a), each verb has a third person absolutive marker and a pluralizer. If in (73) we employed a verb instead of an ideophone, the same would happen: (75) wan kaʔʃ󰀟-n ale hen wan wal-e. pl walk-3abs hearsay then pl sing-3abs ‘Then they went walking and they sang.’ Let us examine another example with an ideophone, now occurring at the beginning of the sentence: (76) tsuruk wan pumu-n ale. enter(ideoph) pl enter-3abs hearsay ‘They entered.’ Again, we would not find a verb in a similar context. If a verb occurs in initial position, there is always extra morphology. If it is the main verb of the sentence, it is modified by the particle ke, which appears when the absolutive NP does not

reduplication and ideophones in trumai

237

immediately precede the verb (77). If it is the verb of a subordinate clause, there is a subordination marker (78). (77) tɨʃ󰀟 ke kain hai-ts di ji. scarify ke foc/tens 1-erg woman art ‘I scarify the woman.’ (78) sa-tke-n-es tete-n hen. dance-des-3abs-sub paint-3abs then ‘When he wants to dance, he gets painted.’ As we can see, what happens with verbs is quite different from what we observe with the ideophones. Thus, there are no reasons to believe that the ideophones are verbs or a verbal subclass. To do so would be to force them to be like verbs when they are clearly displaying different characteristics. And to consider them to be verbs that are defective for finiteness would be to force the language to have a subclass that it otherwise does not possess. Therefore, the best analysis for the ideophones is to treat them as a separate class. 4.3 Uses Syntactically, Trumai ideophones have two main uses, similar to what Seki (2000) describes for Kamaiurá: I. “As an element with the same meaning of a verb present in the sentence” (Seki 2000, 105)—in this use, the ideophone basically reinforces what the verb is expressing. An example is (73) above. Other examples are: (79) asima kawa-n ale hen tsom. dive go-3abs hearsay then dive(ideoph) ‘Then he went to dive.’ (80) tsok ato tɨtʃ󰀟-n ale. pull(ideoph) arm pull-3abs hearsay ‘She pulled out her arm.’ II. As an element “with a meaning not expressed by the verb in the sentence” (Seki 2000, 105)—in this use, the ideophone refers to an extra action. For instance:

238

guirardello-damian

(81) pelasawak odaʃ-ki aʃ󰀟kida-n ale tɨkɨ. tapir forehead-dat jump-3abs hearsay grab(ideoph) ‘He jumped on the tapir’s forehead and grabbed it.’ (82) tsiw tsiw tsiw tsiw wan katnon-e. cut cut cut cut(ideoph) pl work-3abs ‘They worked, cutting weeds.’ When the syntactic construction does not have a verb, the ideophone alone conveys information about the main action: (83) tɨk tɨk tɨk tɨk tɨk ateɬxu mala-ki15 hen. walk walk walk walk walk(ideoph) big.river border-dat then ‘Then they went walking to the border of the river.’ (84) anenewte ita tu tu tu tu tu. woods allat run run run run run(ideoph) ‘He ran towards the woods.’ There is also another context of use: the ideophone is not integrated in the sentence, but it is rather juxtaposed to it, as an utterance in itself.16 In (85) there is a direct speech utterance by a character of the story, and the ideophone is used. It is clearly not part of the direct speech utterance (there is also a pause between the utterance and the ideophone). In (86) the ideophone is employed and next comes a discursive connector introducing the following sentence. The presence of the connector indicates that the ideophone is outside of the sentence (there is a pause between them as well). (85) “aj ʃ󰀟 ha xuj ine lots.” tu tu tu tu. already foc/tens 1 escape 3 ablat run run run run(ideoph) ‘“I already escaped from him” (the boy said). Then he ran.’ (86) tɨkɨ. ina hen dama dama-n ale kud-ea-ɬ. grab(ideoph) disc.con then pull pull-3abs hearsay hair-3poss-dat ‘He grabbed him. Then he pulled his hair.’ 15 16

There are in Trumai three dative markers: -ɬ, -ki and -s. They are selected depending on the type of noun occurring as the head of the NP. This is also observed in other languages. According to Gabas and van der Auwera (2004, 401), ideophones can also be “uttered separately, in an exclusive intonation unit.”

reduplication and ideophones in trumai

239

4.4 Shape In terms of internal configuration, Trumai ideophones can be subdivided into two types (again, similar to what is observed in Kamaiurá; cf. Seki 2000): – ideophones constituted by a single form (e.g., tsom). They express a punctual action; – ideophones constituted by repeated forms (e.g., pɨw pɨw pɨw). Usually the form is repeated three times, but sometimes it can be more (six or seven times). They express a durative action. Some forms always appear repeated and do not seem to have a non-repeated equivalent (e.g., tɨk tɨk tɨk ‘go (walking),’ which does not have the counterpart tɨk with the same meaning). However, there are also cases with both types of forms, i.e., simple form = punctual action / repeated form = durative action. For instance: (87) a. tsiw ‘cut (with a single movement)' b. tsiw tsiw tsiw ‘cut (step-by-step, continuously)’ Overall, the forms exhibit the same sounds and syllabic patterns found in the language, but given their onomatopoeic nature, they also allow the occurrence of sounds that do not exist in Trumai; for example, voiced stops such as /b/. Such a characteristic—i.e., having special sounds that do not fit within the sound system of the language—is often observed in ideophones in languages of the world (Gabas and van der Auwera 2004, 401). Tables 2 and 3 show instances of Trumai ideophones that indicate punctual and non-punctual actions. For some of these, there are slight variations in the shape of the word. table 2

Examples of Trumai ideophones—punctual action

Trumai form

Meaning

tɨkɨ tsok; tsuk tsokɨ tsom; tom buɾuɾuɾu tsuɾuk

grab pull out shoot (with arrow) dive fall in a hole enter (a house, a place)

240

guirardello-damian

table 3

Examples of Trumai ideophones—Non-punctual action

Trumai form

Meaning

pow pow pow; pɨw pɨw pɨw kuru kuru kuru tɨk tɨk tɨk tsiw tsiw tsiw tu tu tu hufa hufa hufa tewken tewken tewken kɨw kɨw kɨw puɾuk puɾuk puɾuk kɨɾɨk kɨɾɨk kɨɾɨk17

row drink go walking cut run practice shamanic activity swing a hammock eat stomp (group of people/big animal) climb a tree

There are also forms that are attested in texts, but they occur with less frequency (cf. table 4 below). For them, it is not easy to determine if they are already conventionalized or if they still undergo fluctuation in their use (with regard to their form or the meaning they convey). In fact, we can propose for Trumai the same that Gabas and van der Auwera (2004, 402) state for the Karo language: the ideophones form an open class, and the speakers can create new forms depending on their discursive needs. table 4

Ideophones—Some unclear cases

Trumai form

Meaning

tof pookal tuk

hit (with a single blow) throw something pierce

Non-punctual pok pok pok tan tan tan; tax tax tax bo bo bo hon hon hon

put things in a container crush (in a mortar) flap the wings burn something

Punctual

17

Aweti, another Upper-Xinguan language, has an ideophone with the same form and meaning as this one (Sabine Reiter, personal communication).

reduplication and ideophones in trumai

5

241

Reduplication and Ideophones: General Considerations

In terms of typology, Trumai reduplication fits many of the patterns described in the linguistic literature. Moravcsik (1978, 328) proposes that “languages that have partial reduplication also have total reduplication.” This is true for Trumai, which exhibits both types, with the full one being more widespread. According to Rubino (2005b, 114), for partial reduplication, although the reduplicated material may occur in various positions, it “is most often found at the beginning of a base.” This again holds for Trumai—the copied material is prefixed to the base. With regard to semantics, reduplication in Trumai fits the possible ranges described by Rubino (2005a, 19–22), indicating intensity, repeated action (iterative), action performed by many individuals, or a large quantity of something. The meaning will vary depending on the class of the words that undergo reduplication; they can be verbs, adjectives, nouns, adverbs, and even numerals. In this respect, Trumai also matches the situation observed in other languages of the world—as Rubino (2005a, 21) demonstrates, reduplication can affect a variety of word classes. However, in Trumai the importance of reduplication varies from class to class: it plays a significant role for verbs, less so for adjectives and adverbs, and it has only a minor role for nouns or numerals (cf. section 2.5). With reference to the phonological aspects, the reduplication process in Trumai can be summarized as follow: – Full reduplication would be the default, occurring with a large number of lexical items. Partial reduplication is more restricted and only occurs under certain conditions. – Partial reduplication involves the copying of light (monomoraic) syllables. – One light syllable (for monosyllabic stems) or two light syllables (for stems with two syllables or more) are copied, while coda consonants are not. – Conditions under which partial reduplication takes place: – Monosyllabic stems have a CVC pattern. – Stems with two syllables or more have the following configuration: the last syllable is heavy, the previous syllable(s) is/are light. – Disyllabic stems represent the optimal environment for occurrence; here the process is productive. – With monosyllabic and trisyllabic stems, partial reduplication is not productive. It varies across lexical items (some undergo it; others undergo only full reduplication).

242

guirardello-damian

Compared to other South-American languages, Trumai reduplication presents certain similarities to Tupi-Guarani languages (Rose 2005), where verbal stem reduplication also involves copying of one or two monomoraic syllables, with the coda consonant being excluded. However, there are differences as well: in Trumai, the initial syllables of the stem are copied, while in Tupi-Guarani languages, the final syllables are affected. In this regard, Trumai is similar to Emerillon, which is an exceptional case in the Tupi-Guarani family. Another important difference is that in Tupi-Guarani languages, the same verbal stem can have one or two syllables copied (see Dietrich, this volume). What influences the number of copies is the meaning that one wishes to express: if one syllable is duplicated, a sense of ‘successive action’ is conveyed (e.g., swallow objects, one after the other). If two syllables are copied, the reduplication process expresses the idea of a ‘frequentative/iterative action’ (e.g., swallow objects frequently). In Trumai, only a fixed number of syllables in a verbal stem can be duplicated. What influences the copying is the size of the stem, rather than the meaning. Let us now consider the ideophones. As seen in section 4, Trumai ideophones also show iteration of linguistic material in which repeated forms express durative events, in contrast with punctual actions. For some languages, the term ‘reduplication’ is employed for repetitive patterns observed in ideophones, e.g., Nuckolls (2001, 276), with regard to Pastaza Quechua. Could we say the same for Trumai, i.e., that the patterns observed in the ideophones follow the same principles observed in the reduplication process, or is there a distinction? Comparing repetition in ideophones with the cases of reduplication described in section 2, various differences should be noted: – Among the ideophones, not a single attested case of “partial reduplication” exists. There is only full repetition of the form. For example, the ideophone for ‘stomp’ is puruk puruk puruk, not puru~puruk. – As seen in section 2, when full reduplication occurs, the form is repeated twice. With ideophones, the repetition can occur three times or even more, which is often excluded from definitions of reduplication. – Reduplication is regulated by phonological rules that take into account the syllabic configuration of the word. The repetition that occurs in ideophones does not seem to be motivated by phonological rules, but rather by sound symbolism (i.e., it is an attempt to imitate the sound of the action that is being described). – Another important difference between reduplication and ideophones in Trumai relates to the semantics. Iteration of linguistic material in ideo-

reduplication and ideophones in trumai

243

phones can express an aspectual difference (punctual versus durative event). Reduplication, in constrast, conveys other sorts of meaning (intensity, number, repeated action). – Trumai reduplication shows differences in relation to reduplication in Kamaiurá, which follows the patterns of Tupi-Guarani languages described by Rose (2005) as mentioned earlier in this section. In contrast, Trumai ideophones are quite parallel to those in Kamaiurá (in terms of shape, use, syntactic environment, etc.). In other words, while the reduplication process of Trumai originates in the language, the ideophones (or at least several of them) might be borrowings, and, if this is the case, they might be following phonological patterns that originally came from the source language, not Trumai. Therefore, the repetitions that occur in Trumai ideophones are not instances of reduplication. They do not represent the same process. It might be that in the past they were related; that is, it could be that the language originally had a set of ideophones that followed the same principles that occur in reduplication, but they may have changed due to the influence of Kamaiurá and other Upper-Xinguan languages. The difficulty is to identify the extent to which this external influence took place. Another possible analysis involves the idea that ideophones by nature have a different constitution (due to iconicity and sound symbolism) and, therefore, do not follow the phonological rules that operate in reduplication. In this case, what happens in Trumai could also occur in other languages, i.e., it would not be due to external influence but rather due to universal characteristics of ideophones. This is an issue that needs to be further explored. In any case, the repetition observed in the ideophones of Trumai is not the same as the reduplication that occurs with other word classes.18 In future studies, it would be interesting to contrast reduplication in this language (which, as mentioned earlier, is classified genetically as a linguistic

18

An alternative analysis suggested to me was that the ideophones could be seen as a subclass of reduplicated forms that follow less the phonological structure of the language. They would be reduplications that were created recently, being closer to iconicity and less to the formal structure of the language. However, it is not evident what exactly the advantage of such an analysis would be. If the ideophones belonged to the class of reduplicated forms, they would be so peripheral that one could hardly justify their inclusion in it. There are no real motivations to bring reduplicated forms and ideophones together. Apart from the fact that both allow the iteration of linguistic material, they have nothing in common (not even the semantics). Thus, a unified treatment does not seem called for.

244

guirardello-damian

isolate) with reduplication in other Brazilian linguistic families. The study by Rose (2005) provides data for an initial comparison with Tupi-Guarani languages. It would now be useful to compare Trumai with other languages of the Tupi stock, as well as with the Gê, Carib and Arawak stocks, which are those present in the Xingu area. Finally, regarding ideophones, a detailed comparison with Kamaiurá and other Upper-Xinguan languages will be necessary in order to evaluate how much they have contributed to the patterns currently found in Trumai. This will help us gain a better understanding about the nature of this word class and how it fits with the rest of the language.

Acknowledgments I would like to thank Gale Goodwin Gómez and Hein van der Voort for their careful and helpful comments. I would also like to thank the Trumai consultants, in particular Macarea Trumai, who contributed more directly to the present study.

References Adelaar, K. Alexander. 2000. Siraya reduplication. Oceanic Linguistics 39/2: 33–52. Dietrich, Wolf. this volume. Forms and functions of reduplication in Tupian languages. Dourado, Luciana. 2001. Aspectos morfossintáticos da língua panará (Jê). PhD diss., Universidade Estadual de Campinas. Emmerich, Charlotte. 1980. A fonologia segmental da língua Txikão: um exercício de análise. Rio de Janeiro: Departamento de Antropologia, Museu Nacional/UFRJ. Fargetti, Cristina M. 1992. Análise fonológica da língua juruna. MA thesis, Universidade Estadual de Campinas. Gabas Jr., Nilson, and Johan van der Auwera. 2004. Ideophones in Karo. In Language, culture and mind, ed. Michel Achard and Suzanne Kemmer, 397–413. Stanford: CSLI Publications. Guirardello, Raquel. 1999. A reference grammar of Trumai. PhD diss., Rice University. Guirardello-Damian, Raquel. 2005. Fonologia, classes de palavras e tipos de predicado em Trumai. Boletim do Museu Paraense Emílio Goeldi, Série Ciências Humanas 1/2: 193–306. . 2011. Léxico comparativo: Explorando aspectos da história Trumai. In Alto Xingu: Uma sociedade multilíngue, ed. Bruna Franchetto, 113–152. Rio de Janeiro: Editora do Museu do Índio/Funai.

reduplication and ideophones in trumai

245

Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Kemmer, Suzanne. 1996. Collective and distributive marking, or: where unity meets multiplicity. In The twenty-third LACUS Forum 1996, ed. Alan K. Melby, 231–249. Chapel Hill, NC: Linguistics Association of Canada and the United States. Marantz, Alec. 1982. ReReduplication. Linguistic Inquiry 13: 435–482. Matthews, Peter H. 1997. The concise Oxford dictionary of linguistics. Oxford: Oxford University Press. Meira, Sérgio. In prep. A grammar of Tiriyó. Berlin/New York: Mouton de Gruyter. Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of human language, vol. 3, ed. Joseph Greenberg, 297–334. Stanford: Stanford University Press. Nuckolls, Janis B. 2001. Ideophones in Pastaza Quechua. In Ideophones, ed. F.K. Erhard Voeltz and Christa Kilian-Hatz, 271–286. Amsterdam/Philadelphia: John Benjamins. Quain, Buel, and Robert F. Murphy. 1995. The Trumai Indians of Central Brazil. Seattle: University of Washington Press. Rose, Françoise. 2005. Reduplication in Tupi-Guarani languages: Going into opposite directions. In Hurch (ed.), 351–368. Rubino, Carl. 2005a. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Rubino, Carl. 2005b. Reduplication. In The world atlas of language structures, ed. Martin Haspelmath, Matthew S. Dryer, David Gil, and Bernard Comrie, 114–117. Oxford: Oxford University Press. Seki, Lucy. 2000. Gramática do Kamaiurá, língua Tupi-Guarani do Alto Xingu. Campinas: Editora da Universidade Estadual de Campinas. Taylor, John R. 1995. Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon Paperbacks. Van der Meer, Tine H. 1983. Ideofones e palavras onomatopaicas em Suruí. Unesp— Estudos Lingüísticos 7: 10–25. Voeltz, F.K. Erhard, and Christa Kilian-Hatz, eds. 2001. Ideophones. Amsterdam/Philadelphia: John Benjamins.

Abbreviations abla abs allat art cop dat

ablative absolutive allative article copula dative

des dis.con erg foc/tens ideoph nzr

desiderative discursive connector ergative particle of focus + tense ideophone nominalizer

246 nzr.instr perf pl

guirardello-damian

instrument nominalizer poss (“the tool for”) red perfective sub plural

possessive reduplicated element subordinator

chapter 10

Reduplication and Verbal Number in Mẽbengokre Andrés Pablo Salanova

In this paper I examine data from the lexicon of Mẽbengokre, a Jê language spoken in the eastern reaches of the Amazon basin, in central Brazil, and survey the reduplicative processes found in the language. Two separate types of reduplication are identified: one (relatively uninteresting) process associated with onomatopoeia, found in several names for plant and animal species, and one where reduplication is one among several prefixes indicating the plural in verbs, a category which may both indicate the plurality of one of the verb’s arguments, or the repetition of the eventuality itself. I finish by comparing the facts of verbal reduplication in Mẽbengokre to what is found in Kaingang, a southern Jê language where the phenomenon has been described more fully.

1

Introduction

Mẽbengokre is a Jê language spoken in north-central Brazil by two indigenous nations, the Xikrin and the Kayapó, numbering over ten thousand individuals in total. It is closely related to Apinayé (whose phonology is described by Oliveira 2005, Burgess and Ham 1968 and Callow 1962), Suyá (described by dos Santos 1997), Timbira (whose phonology is described by Popjes and Popjes 1986, and Alves 1999), and Panará (described by Dourado 2001). Other languages in the family include Xavante, Xerente, Kaingang (described in Wiesemann 1972 and Cavalcante 1987), and Xokleng. Though the latter are quite distant from Mẽbeng, all languages in the family are strikingly similar in their overall morphosyntactic structure. There are several non-concatenative morphological phenomena in Mẽbengokre, most notably initial and final truncation, discussed in Salanova (2004; 2011), and reduplication. Reduplication itself is somewhat limited in scope and productivity in Mẽbengokre. Nevertheless, I will show that reduplication has a well-delimited function in the language and that interesting comparative observations may be made with other languages of the family. More solid conclusions await further research. Even though at the present time there exist preliminary grammatical sketches for most of the languages of the family, the lexicon has practically not been studied at all in any of them.

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_011

248 2

salanova

Outline of Mẽbengokre Phonology

In this section, we present a brief outline of Mẽbengokre phonology. A fuller treatment may be found in Stout and Thomson (1974) and Salanova (2001). The consonant and vowel inventories of Mẽbengokre are as follows: table 1

Mẽbengokre consonants

Labials

Alveolars

Palatals

Velars

Glottal

p b m w

t d n ɾ

ʧ ʤ ɲ j

k g ŋ

ʔ

Voiceless obstruents Voiced obstruents Nasal sonorants Oral sonorants

table 2

Mẽbengokre oral vowels

Front Unrounded high mid-high mid-low low

table 3

i e ɛ

ɯ ɤ ʌ a

u o ɔ

Mẽbengokre nasal vowels

Front Unrounded high mid low

Back Unrounded Rounded

ĩ ẽ

Back Unrounded Rounded ɯ̃ ʌ̃ ã

ũ õ

The symmetry apparent in the consonant and vowel inventories is deceptive. In actual fact, the consonants /d/ and / ʧ/ are vanishingly rare, and the phonemic status of the vowels /ũ/ and /ã/ is not very clear. Finally, /ɲ/, /ʤ/ and /j/

reduplication and verbal number in mẽbengokre

249

are in near complementary distribution, and the three phonemes most likely derive from a single one historically, even if synchronically they seem to be distinct. For information on the latter, as well as for some historical evidence for a phoneme /h/, I invite the reader to consult Salanova (2011). Other than in this section, in this paper I give the Mẽbengokre examples in the most common orthographic convention used to write the language, where r = /ɾ/, ’ = /ʔ/, nh = /ɲ/, ng = /ŋ/, x = /ʧ/, and dj = /ʤ/; ê = /e/, e = /ɛ/, ô = /o/, o = /ɔ/, y = /ɯ/, à = /ʌ/, ỳ = /ɤ/, ỹ = /ɯ̃ /, and ã = /ʌ̃, ã/, with other orthographic symbols being identical to the IPA symbols used in the table above to represent the phonemes of the language. Syllables in Mẽbengokre can be of the form (C)(C)(C)V(C). Stressed syllables may only be onsetless if the onset consonant is lost to a process of onset apheresis that reflects third person inflection. This proces is described in detail in Salanova (2011). Onset consonants display interesting co-occurrence restrictions: consonants may be classed into three groups according to their main articulator, as in table 4. No two consonants using the same articulator may be simultaneously present in an onset. That means that combinations such as */nɾ/, */pw/, */tj/ are excluded.1 table 4

Articulator and sonority classes of Mẽbengokre consonants

noncontinuants rhotic approximants

Labial

Laminal

Dorsal

pbm

tn ʧʤɲ ɾ j

k g* ŋ

w

The vertical dimension in table 4 reflects how Mẽbengokre classifies segments according to their sonority. Segments in a complex onset need to follow a sequence of strictly increasing sonority, with noncontinuants being the least sonorous, and approximants the most. Putting the two constraints together, one has as possible three-consonant onsets only /ŋɾw/ and /kɾw/, and as possible two-consonant onsets the following: /pj/, /mj/, /bj/, /pɾ/, /mɾ/, /bɾ/, /tw/, /nw/, /ʧw/, /ʤw/, /ɲw/, /kj/, /ŋj/, /kɾ/, /ŋɾ/, /kw/, /ŋw/, /gw/. Onset restrictions are not simply static generalizations. Various fixes occur to prevent invalid sequences that result from resyllabification. 1 /d/ and /ʔ/ do not occur in complex onsets, while /g/ only occurs with /w/.

250

salanova

Stress falls on the final syllable of a lexeme. An epenthetic vowel that normally follows word-final /ɾ/, /w/ and /j/ is unstressed and is not represented in the orthography. Various unstressed elements, such as adpositions and determiners, may encliticize to lexemes without shifting the stress. The suffix -re ‘diminutive,’ which appears in several names of animals below, is also unstressed. There are only a handful of allophonic processes in Mẽbengokre. The most important of these affects coda consonants. When the coda is underlyingly a voiceless obstruent or a nasal,2 this segment’s nasality and voicing is determined almost completely by context, with possibly only a hint of a voice or nasal contour keeping them apart in a __C context and nothing at all in a __N context.3 table 5

VC: ṼC: VN: ṼN:

Coda alternations of noncontinuant consonants

/ɾɔp/ jaguar /ɲõt/ sleep /atɔm/ dirty /kõm/ drink

__C

__N

[ɾɔp ̚ kɾa] jaguar’s cub [ɲõt ̚ ket] doesn’t sleep [a'tɔbp ̚ ket] not dirty [kõmp ̚ ket] doesn’t drink

[ɾɔm nɔ] jaguar’s eye [ɲõn mʌ̃] is about to sleep [a'tɔm mʌ̃] about to be dirty [kõm mʌ̃] is about to drink

It is likely that neutralization between obstruents and nasals in the coda is in fact complete among nonpalatal consonants, where no minimal pairs exist. The reasoning sustaining such a contrast in the orthography is the following: The consonant that emerges when a coda consonant is resyllabified with a following onsetless syllable is pronounced, like all onset consonants, without audible contours, and its pronunciation is claimed to reflect the underlying consonant (R. Thomson, p.c., 1997). Some lexical items such as mjên ‘husband’ are such that this resyllabified consonant may be pronounced with variable voicing and nasalization, as in (1), contrasting with other lexical items such as 2 Voiced obstruents cannot occupy coda position. 3 Note that Stout and Thomson (1974) claim that the contrast is maintained in all contexts. We do not find that the data confirm this.

reduplication and verbal number in mẽbengokre

251

krit ‘pet,’ where the consonant is always a voiceless stop, as in (2). However, I believe that this difference might be determined by the preceding vowel’s height, rather than being lexically determined. In other cases it might also hinge on the consonant’s place of articulation (for instance, only /k/ occurs among the velar consonants in the coda). (1)

['mje.nɔ.'ba]—['mje.dɔ.'ba]—['mje.tɔ.'ba] mjên o ba 3.husband with go.around ‘She is going around with her husband.’

(2) ['kri.tɔ.'ba] (others not possible) krit o ba pet with go.around ‘S/he is with his/her pet.’ The facts are somewhat different in the case of coda palatals, where at least one minimal pair, for which we nevertheless do not have very good evidence, exists: /mɛ ʧ/ ‘good’ vs. /mɛɲ/ ‘honey,’ illustrated in (3) and (4). Both final / ʧ/ and final /ɲ/ alternate with [j] rather than with contour segments, and the choice reflects emphasis (with [ʧ] and [ɲ] being more emphatic) and is otherwise in free variation. (3) ['mɤ.ja.ɔ.'mɛ.ʧɔ.ba]—['mɤ.ja.ɔ.'mɛ.jɔ.ba] mỳja o mex o ba things P good P go.around ‘He was/is going around fixing things.’ (4) ['mɛ.jɔ.'ba]—['mɛ.ɲɔ.'ba] (*['mɛ.ʧɔ.'ba] is not possible) menh o ba honey P go.around ‘He was/is going around with honey.’ On the other hand, neutralization to [j] or [j̃], dependent solely on the nasality of the preceding vowel, is complete in palatals whenever they are followed by a noncontinuant consonant:4

4 Glosses to the words used in the table are as follows: /mɛʧ/ ‘good,’ /ket/ ‘not,’ /mʌ̃/ ‘in order to,’ /kumrẽʧ/ ‘real,’ /mɛɲ/ ‘honey,’ /ujarẽɲ/ ‘him telling a story.’

252 table 6

salanova Coda alternation with palatal consonants

VC: /mɛʧ/ good ṼC: /kumrẽʧ/ first VN: /mɛɲ/ honey ṼN: /ujarẽɲ/ narrate

__C

__N

[mɛj ket] not good [kumrẽj̃ ket] not first [mɛj ket] not honey [ujarẽj̃ ket] doesn’t narrate

[mɛj mʌ̃] is about to be good [kumrẽj̃ mʌ̃] about to be first [mɛj mʌ̃] for the honey [ujarẽj̃ mʌ̃] is about to narrate

Vowel epenthesis occurs after coda /ɾ/. The epenthetic vowel is normally a copy of the preceding one, but for /a/ and some instances of /ɔ/ and /ʌ/ it is [i]. In the Xikrin dialect, epenthesis after /ɾ/ is limited to final position; elsewhere, coda /ɾ/ neutralizes to /t/ and undergoes the same allophony as voiceless obstruents in that position, as shown in (5): (5) /bʌɾ/ ‘tree’ (['bʌ.ɾi] in isolation in both dialects) + /prʌ/ ‘coal’: [bʌt.prʌ] ‘charcoal’ (Xikrin) ['bʌ.ɾi.'prʌ] ‘charcoal’ (Kayapó) The alternations that we have described in this section are morphophonologized in some reduplicated forms, but often in an unexpected manner. See in particular the discussion under (13).

3

Reduplicative Phenomena in Mẽbengokre

The data for this study come from my own field notes, collected on several trips to Kayapó and Xikrin villages over a period of 15 years, and from the examination of a draft Mẽbengokre-Portuguese lexicon, which was particularly useful for surveying and translating plant and animal names.5 In order to have an

5 This consists of a Toolbox database gathering all published and unpublished lexical data on Mẽbengokre, some of it corrected and augmented by the author and by Maria Amélia Reis Silva in the field. The data include materials by Pe. Antonio Sala, Horace Banner, Earl Trapp,

reduplication and verbal number in mẽbengokre

253

unbiased preliminary overview of reduplicative phenomena in Mẽbengokre, I start by looking at all lexical items where at least a CV sequence is repeated, whether there is additional evidence for morphology or not, and even if a base is not independently attested in my notes. As will be seen, a decent case can be made for the derived nature of the reduplicated forms that express verbal number, even if reduplication in this domain is not as robust as elsewhere in the language family. In terms of form, one may identify C(C)V, C(C)VC, and CVCV reduplicants:6 (6)

gogo kaprêprêk karõrõ keket rôrôk tatak mrômrôtire ràràr rara 'ã'ãre

‘to shake, make noise’ ‘to give a beating’ ‘to snore, roar’ ‘to laugh’ ‘to fall (plural)’ ‘to clap, type’ ‘tadpole’ ‘orange (color)’ ‘box-shaped woven basket with lid’ ‘seagull (?)’

(7)

prõrprõt tertet dujduj porpot wãnhwãnh

‘to float up and down’ ‘to tremble from being cold’ ‘pied lapwing (Hoploxypterus cayanus)’ ‘pauraque (Nyctidromus)’ ‘show-off’

(8)

kôtêkôtê ‘blue-crowned motmot (Momotus momota)’ kurakura ‘acará-boi (fish of the Chichlidae sp.)’ muremure ‘grey-winged trumpeter (Psophia crepitans)’

Terence Turner, Joan Bamberger, Pe. Renato Trevisan, Ruth Thomson, Isabelle Giannini, Darrell Posey, and Vanessa Lea, and were collected and organized by the latter, with whom the author and Reis Silva collaborated between 1998 and 2000. 6 Mẽbengokre has no grammatical gender. In the translations, I use ‘he,’ ‘she’ or ‘it’ indistinctly. Likewise, no morphological distinction is made in the language between the present and the past, and my translations are often erratic in this regard. Note however that the contrast betweeen nominal (or non-finite) and verbal (or finite) forms of verbs is consistently rendered in the translation as a difference between present or simple past (for finite verbs) and present perfect or resultative (for non-finite verbs).

254

salanova

The last set has an unusual trochaic prosody and always designates animal species, strongly suggesting onomatopoeia in a very narrow sense. One can establish two broad semantic classes in the data, with just a few words left out: live species, and verbs of repeated action. The former seem to have peculiar phonology even when the reduplicant is smaller: 'ã'ãre ‘seagull,’ for instance, can be pronounced with an independent stress on each 'ã, something which can never befall, e.g., a verb such as rôrôk ‘to fall (pl).’ Most importantly, however, one can never identify a base in the lexicon from which the forms in the first semantic class would be formed via reduplication. While in reduplicated forms denoting repeated action there are a few cases where the base is not found synchronically, reduplication is paradigmatic in a majority of cases. For this reason, I make a division between two processes: onomatopoetic repetition, which I will not discuss further, and reduplication proper, normally associated with a reiterative sense, which will be addressed in what remains of the paper.7 The latter group may be further subdivided into those forms for which a base may be readily identified synchronically, (9), and those for which it cannot, (10). (9)

totyktyk kyjkyj krãkrãk tatak kryjkryj

‘to strike repeatedly’ ‘to make many scratches’ ‘to swallow’ ‘to clap, type’ ‘to scratch’

totyk kyj tokrãk tak djàkruj

‘to strike’ ‘a scratch or cut’ ‘to swallow at once’ ‘to hit’ ‘walk by dragging one’s feet’

(10) kypkyp ‘to chew’ gogo ‘to shake, make noise’ ngàrngàt ‘make hissing sound’ I contend that no thick line should be drawn a priori between these two sets. Later in the paper, I do make a distinction which I justify a posteriori. Furthermore, I will not pay undue attention to morphophonology. As I will argue later in the paper, reduplication in Mẽbengokre is the expression of a 7 I will exclude from consideration the handful of words that are in neither of these semantic categories: ràràr ‘orange-colored,’wãnhwãnh ‘show-off,’ rara ‘box-shaped basket.’ I believe this exclusion to be justified, given that, like for the onomatopoeias, no base may be identified in the lexicon, precluding us from calling this process reduplication. Nothing of substance hinges on this exclusion. Below I will address the more serious question of whether some of the ‘verbs of repeated action’ should not be themselves considered onomatopoeias.

reduplication and verbal number in mẽbengokre

255

derivational process, rather than of inflection. As in other cases of derivation, productivity is not complete, some of the roots on which actual forms are based are not in use, and morphophonological and semantic idiosyncrasy can often be observed. This having been said, please note that the substitution of part of the stem which one observes in cases such as tokrãk is indeed contemplated in the discussion of the classifier prefixes below. As for the phonology, once obvious onomatopoeia is excluded, one may state that the reduplication that is left fits a C(C)V template or C(C)VC template. Furthermore, it is apparent that the reduplicant is a prefix or an infix inserted before the last, stressed, syllable (cf. kaprêprêk, karõrõ). In fact, for reduplicated forms with a C(C)V template, analyses proposing affixation to the right of the base are equivalent to those proposing affixation to the left. A comparison of cases of C(C)VC and C(C)V reduplication shows that the bigger picture is simpler, however, if one assumes that the reduplicant is a prefix or infix appearing before the stressed syllable of the base:8 (11) C(C)V reduplication Base Affixation before the base go → go~go ket → ke~ket kaprêk → ka⟨prê⟩prêk

Affixation after the base go~go ke⟨ke⟩t,*ket~ket kaprê⟨prê⟩k,*kaprêk~prêk

(12) C(C)VC reduplication Base Affixation before the base kyj → kyj~kyj totyk → to⟨tyk⟩tyk

Affixation after the base kyj~kyj,*ky⟨ky( j)⟩j totyk~tyk,*toty⟨ty(k)⟩k

If one chooses the analysis where the reduplicants are attached after the base, one would have to stipulate that C(C)VC reduplicants are always suffixes, whereas C(C)V reduplicants are infixes whenever the base ends with a consonant. On the other hand, if one assumes that reduplicants are attached before the base no such stipulation is needed: the reduplicant is always attached before the stressed syllable. The same stipulations are required if one draws a distinction between partial and full reduplication based on the accidental fact that the base is small enough to be identical to the reduplicant.

8 In some of these examples (i.e., gogo, keket) the base is putative, as it is not attested synchronically.

256

salanova

Further down, I will argue that the conclusion that reduplicative morphemes are prefixes or pre-stressed-syllable infixes fits in well with what is known about Mẽbengokre verbal morphology more generally. Yet another possibility is that reduplication always follows a C(C)VC template, and morphophonological readjustment rules result in coda consonants getting dropped in certain environments. In this view, coda consonants get dropped in non-final syllables only, regardless of whether the non-final syllable is the reduplicant (i.e., when the reduplicant is a prefix) or the stem itself (i.e., when the reduplicant is a suffix): (13) Fixes are necessary Base Reduplication Fixes ket → ket~ket → ke~ket kaprêk → ka-prêk~prêk → ka-prê~prêk Fixes are not necessary go → go~go kyj → kyj~kyj totyk → to-tyk~tyk This approach is also interesting to deal with cases such as tertet ‘shiver’ or prõrprõt ‘float up and down,’ where one part of the reduplicant ends in -t while the other ends in -r (-r is in all these cases followed by an epenthetic vowel: [tɛɾɛ'tɛt], [pɾõɾõ'pɾõt], but this is due to a relatively superficial phonological rule). As seen above, /r/ and /t/ alternate in certain contexts in the Xikrin dialect of Mẽbengokre, and it might thus be argued that in the reduplicated forms found in (13) there is full C(C)VC reduplication, with the coda C becoming /r/ or /t/ as an effect of a context-sensitive phonological rule. However, note that the alternation between /r/ and /t/ in these examples seems to work in exactly the opposite way that it works in the regular phonology, described in section 2, where /r/ becomes /t/ in pre-obstruent position (cf. Xikrin par ‘foot’ + kà ‘skin’ = [patkʌ] ‘shoe,’ bàr ‘wood’ + prà ‘coal’ = [bʌtprʌ] ‘charred wood’). Though intuitively pleasing, the approach where reduplicants are uniformly C(C)VC and get pared down by the phonology is liable to a serious objection, which might have already been apparent in examining the second data set in (13). It is impossible to predict from the phonological environment just when a consonant will be dropped or not. Compare kypkyp ‘to chew’ with tatak ‘to clap, type.’ In both of these, C(C)VC reduplication would juxtapose two voiceless stops. In one case, the heterosyllabic stop sequence is fixed by dropping the first stop (taktak → tatak). This simplification is attested elsewhere in the

reduplication and verbal number in mẽbengokre

257

language: the masculine personal name formative Bep, for instance, becomes Be in names such as Bekĩnhre. In the case of the reduplicated kypkyp, however, the simplification does not take place, and there is nothing to explain this other than lexical idiosyncrasy.9 Regardless of the solution one chooses, therefore, it has to be specified for each item of the lexicon that undergoes reduplication whether the reduplicant will be of the form C(C)V or C(C)VC, whether it is done by saying that there are two different affixes, or by a phonological rule that is applied to a lexicallyconditioned subset of forms. This contrast between stems that get reduplicated with C(C)V and those that get reduplicated with C(C)VC could in principle form the basis of a subdivision of reduplicated forms into internally coherent classes,10 but this is a question that will have to be the object of further research. So far, I have offered the reader a broad preliminary characterization of one type of reduplication found in Mẽbengokre: a C(C)V-or C(C)VC-derivational prefix appearing in verbs, vaguely denoting repeated action. Several objections may be made to this characterization: (1) this is not a fully productive process of reduplication, (2) I might not have been fully persuasive in arguing that it is not onomatopoeia, and (3) the semantics of “repeated action” is somewhat elusive. These are all potentially valid objections that I address to varying degrees in this paper. The first objection was summarily pushed aside above, when I classified reduplication in Mẽbengokre as derivational. The point is further developed in section 4, where I put reduplication in the broader context of verbal morphology in Mẽbengokre. The second and third objection are addressed in section 5, where I discuss the category of verbal number in Mẽbengokre, and sketch an approach where verbal onomatopoeia and grammatical pluractionality are extremes in a continuum.

9

10

The dropping of coda consonants before obstruents is also unsystematic in the case of personal names. Next to the proper name Bekĩnhre, there is also the proper name Bepkra. There is nothing obvious in the phonological environment that would explain the maintenance of /p/ in the latter. This idiosyncrasy constitutes a very interesting topic that I will not pursue further in this paper. For example, if one happened to find that there is a correlation between reduplicating with a C(C)VC template and greater productivity or semantic transparence, one could in good conscience talk about two separate reduplication rules, one of which might not be synchronically active.

258 4

salanova

MebengokreVerbal Morphology and Quasi-Morphology

For reasons of space, I restrict my discussion of verbal morphology in Mẽbengokre to one small aspect of derivation, namely what I will soon call the classifier prefixes of transitive verbs. Given the fact that such morphology is intricately tied with verbal roots, some arguments need to be advanced in favor of establishing the category of these prefixes as separate morphemes. A discussion homologous to what follows is presented by Oliveira (2005, 116–128), about the closely-related language Apinayé. I begin by pointing out that all Mẽbengokre verbs, except for obviously complex fixed expressions, are either mono-or bisyllabic. Furthermore, the distribution of monosyllables versus bisyllables is not random: while most underived intransitive verbs are monosyllables, the majority of transitives are bisyllables:11 (14) a. to ‘dance,’ tẽ ‘go,’ rê ‘swim,’ ngre ‘sing,’ bôx ‘arrive,’ ty ‘die,’ tỹm ‘fall’ b. jadjà ‘put on,’ nhĩpêx ‘make,’ kunõ ‘paint red,’ kabi ‘taste,’ pynê ‘catch’ In fact, the pre-stressed syllable of transitive verbs is chosen among a small set of syllables: ja-, ka-, ku-,12 nhĩ- and py-, with a handful of others being marginally present. This fact was observed in the earliest descriptions of Mẽbengokre verbs (Trapp n.d.). On the other hand, the final, stressed syllable of a transitive verb can be whatever the maximum syllable permitted in the language is, i.e., CCCVC, though the final C is normally dropped in the finite form. Let us call the template that describes the verb stem a sesquisyllable, borrowing a term that is commonly used in the description of some Southeast Asian languages (cf. Matisoff 1973, 86) to name stems that are composed of one full syllable preceded by one CV syllable chosen from a small set. One can make a preliminary generalization that underived intransitive verbs in Mẽbengokre are monosyllables, while underived transitive verbs are sesquisyllables.

11

12

There are a few exceptions to this generalization. The most important of these comes up in example (16) and surrounding discussion. A summary is given in (35), but a fuller discussion is out of the question here for reasons of space. The reader should trust that the exceptions do not go against the grain of my analysis. For a more complete overview of Mẽbengokre verbal morphology, see Salanova (2007). This ku-should not be confused with the third person marker ku-‘3ac.’

reduplication and verbal number in mẽbengokre

259

I contend that such sesquisyllables are morphologically complex, with the stressed syllable constituting the root, and the restricted set of possible initial syllables constituting various prefixes indicating transitivity. The best type of evidence for this would be to find pairs of intransitive and transitive verbs that differ only by the presence of the prefixes in the latter. Second best would be to find sets of verbs that share a base and differ in the prefix that they use. This might further give us a hint about any semantic contribution that the prefixes make to the verb stem, in addition to marking transitivity. As it turns out, both types of evidence may be found. Of the ‘short syllables’ that make up the transitive verb’s sesquisyllabic template, ka-is the best candidate for a transitivizer or marker of transitivity. Anticipating my analysis, we will call this marker a classifier,13 and gloss it cl below. (15) a. b. c. d.

tur kwỳr djàkôr djàkêj

‘urinate’ ‘defecate’ ‘breathe’ ‘scratch (intr.)’

ka-tur ka-kwỳr ka-kôr ka-kêj

‘urinate on’ ‘defecate on’ ‘blow on’ ‘scratch (tr.)’

For examples (15c,d), an unprefixed base is not attested. In fact, and counter to my generalization about the form of verbs in Mẽbengokre, in these examples we have a prefix djà-that marks intransitivity, just as ka-and others mark transitive verbs. The reason why I claimed above that intransitive verbs are unprefixed despite examples such as (15c,d) is that the prefix djà-is also one of the allomorphs of the antipassive. When an antipassive, djà-attaches outside a transitive verb’s prefixes, like other valency-reducing affixes such as the anticausative. The antipassive and the anticausative morphemes are shown, respectively, in the following examples; what I consider to be the transitivity prefixes are underlined. The syncope rule in (16b) is not idiosyncratic to this example: (16) a. aj-kamẽ, bi-kamẽnh anticaus-pull.away.v, anticaus-pull.away.n ‘to move away.’ (from kamẽnh ‘to pull away’) b. a-ptà, djà-ptàr antipass-protect.v, antipass-protect.n ‘to block the way.’ (from pytàr ‘to protect’)

13

Oliveira (2005) uses the term formative to designate what we here call classifier.

260

salanova

In these examples, the antipassive and the anticausative are shown to be clearly distinct from the classifiers that I discuss here, as they appear outside the classifiers, rather than in paradigmatic opposition with them. What examples such as (15c,d) show is that, in addition to being an antipassive, djà- can be simply a classifier that marks intransitivity, in paradigmatic opposition with ka-. The prefix djà- as a classifier may be seen again in the following examples: (17) a. ropre nẽ a-kê dog nfut cl-scratch.v ‘The dog scratches (at stuff).’ b. ropre djà-kêj kêt dog cl-scratch.n neg ‘The dog does not scratch (at stuff).’ (18) a. ropre nẽ pyka ka-kê dog nfut earth cl-scratch.v ‘The dog scratches the earth.’ b. ropre kute pyka ka-kêj kêt dog 3erg earth cl-scratch.n neg ‘The dog does not scratch the earth.’ (19) a. ba i-djà-pêj 1nom 1-cl-work.n ‘I work.’ b. ba mỳja ku-pêj 1nom something cl-work ‘I work on something.’ (Port.: ‘mexo com alguma coisa.’) I conclude that Mẽbengokre has at least one classifier, ka-, that marks transitivity, and at least one classifier, djà-, that marks dynamic (unergative) intransitives, in paradigmatic opposition to ka-. I now move on to examine the second form of evidence which I anticipated at the beginning of the section, namely the paradigmatic opposition between the different transitive cls. This will allow us to get an idea about the prefixes’ semantic contribution, a necessary step before I may present the reduplicative prefixes as simply a special

reduplication and verbal number in mẽbengokre

261

case of cl, and the justification behind not choosing to consider ka- to be an applicative. In example (19b) and in example (16b), I have already introduced some examples where the transitivity prefix present in a stem is not ka-but some other one. In what follows, I take a brief look at the prefixes ja-, ku-, nhĩ -, pyand the occasional other prefix (e.g., dju-, nhỹm- and nhõ- in (21) below) found in pre-root position. The following examples show that these prefixes fulfill the same function as ka-:14 (20) ngrà tom

‘dry’ (adj.) ‘covered with beeswax’ prãr ‘to be empty handed’ djàpêj ‘to work’ djàkôr ‘to breathe’ djàptôr ‘to spit’

kungrà ‘to wipe, scrub’ kutom ‘to apply beeswax to the head’ kuprãr ‘to empty out, raze’ nhĩpêx ‘to make’ jakôr ‘to smoke (tr.)’ kutôr ‘to spit on’

Furthermore, the following examples show that a single root may become a transitive stem by receiving any one of various different cl prefixes: (21) djàpêj ‘to work’ nhãnh ‘to bite’ bjêr

‘to drag along’

pôt

‘open’

(base unattested)

(base unattested)

14

kupêx nhĩpêx kamjãnh djumjãnh jabjêr djupjêr japôk nhõpôk kapôj kungrĩ kangrĩ jangrĩ nhỹmngrĩ djàmjỳr kamjỳr pymjỳr

‘to work on’ ‘to make’ ‘to chew’ ‘to chew on grains’ ‘to bring back game’ ‘to carry on the back’ ‘to make holes on’ (tr.) ‘to hollow out’ ‘to make an incision’ ‘to arrange in a bundle’ ‘to put together, pack up’ ‘to cover by surrounding’ ‘to cover a load with leaves’ ‘to poke the ground with a hoe’ ‘to poke with a needle’ ‘to stick into the ground’

From now on, I will gloss over the morphophonological irregularities found in the data (cf., for instance, the /p/ in djàptôr in (20)). These are very idiosyncratic, and I have very little to say about them.

262

salanova

The nuances in meaning between the various stems are subtle and quite idiosyncratic. One may broadly characterize the cl prefixes as encoding certain properties of the object, and of how the latter is affected, with the root carrying the core of the description of the event. The details of this are beyond the scope of this paper, except for one trait: verbal number. The opposition among different cl prefixes attached to the same verbal root often reflects the number of the object, or whether an action is singular or repeated: (22) a. ku-ôj cl-burn.n ‘light a fire (i.e., a bonfire)’ b. ja-ôj cl-burn.n ‘light many small fires (i.e., people’s cigarettes)’ (23) a. ka-kôr cl-blow.n ‘blow (e.g., on a trumpet)’ b. ja-kôr cl-blow.n ‘smoke (i.e., take repeated short breaths from a pipe)’ In other cases, instead of an opposition between different cl prefixes, one sees the prefix present only in the plural form:15 (24) a. nhôr hang.n ‘hang (i.e., a large piece of meat)’ b. ja-nhôr cl-hang.n ‘hang (i.e., beads on a string)’

15

The transitive verb nhôr is another exception to my generalization on the form of verbs. A brief explanation is given in (35d).

reduplication and verbal number in mẽbengokre

263

As I stated above, ‘repeated action’ is part of the meaning of one set of reduplicated forms identified in (6). The reduplicated forms below denote repeated punctual eventualities: (25) prêk tãk tak tok mok

‘to startle, snap’ ‘to jump up’ ‘to strike’ ‘to startle’ ‘to open’

kaprêprêk tãtãk tatak totok momok

‘to slap repeatedly’ ‘to beat (i.e., the heart), be sore’ ‘to clap, type’ ‘to make repeated noise’ ‘to crack open repeatedly’

In my view, the reduplicative prefix in these examples is doing exactly the same as the cl prefix ja- in (24b). One finds particularly clear examples of reduplicative prefixes in paradigmatic opposition with other prefixes in the subclass of verbs which for simplicity I call ideophonic verbs. These verbs have a superficial resemblance with ideophones, which in Mẽbengokre may be the main predicate of a clause, in that both ideophonic verbs and true ideophones use the same cl prefixes to- and e-. In true ideophones, the root is clearly expressive, may be repeated several times, is prosodically separated from the prefix, may contain sounds that are not part of the phonological inventory of the language, and is often, though not necessarily, followed by the verb anẽ, anhỹr ‘to do thus; say.’ The following is an example: (26) õkôt bê to tuk tuk tuk anẽ his.chest on cl ideoph ideoph ideoph do.thus.v ‘He went tuk, tuk, tuk [hit him] on his chest.’ The ideophonic verbs described in this section use the same prefixes as the ideophones, but otherwise behave like regular verbs. The cl prefixes employed are to- if they are transitive verbs, or e- if they are intransitive verbs: (27) a. toyk ‘to slide (tr.)’ b. eyk ‘to slip (intr.)’ (28) a. toy b. ey

‘to drag along the floor (tr.)’ ‘to drag oneself along the floor (intr.)’

(29) a. totyk ‘to snap (tr.)’ b. etyk ‘to snap (intr.)’

264

salanova

The relevance of ideophonic verbs to the discussion is that they often reduplicate to indicate plurality of an event. When they do, the reduplicant prefix normally supplants the cl prefix, and the verb is transitive by default; this can be observed in the following examples, where reduplication fulfills a function identical to what one sees with ja- in examples (22b) and (23b): (30) a. kyj scratch ‘a scratch (n)’ b. to-kyj cl-scratch ‘to make a scratch on a surface (tr. v)’ c. kyj~kyj red~scratch ‘to make many scratches on a surface (tr. v)’ (31) a. to-krãk cl-swallow ‘to swallow (tr. v)’ b. krã~krãk red~swallow ‘to swallow repeatedly or lengthily (tr. v)’ I conclude that red and cl are in paradigmatic opposition as markers of transitivity, with the former being linked to plural action, and the latter to singular action. In other words, red is just a subtype of cl that indicates plurality.16

16

The reader might ask about the form totyktyk, presented in (9). This, like kaprêprêk, also presented earlier, is one of a handful of cases where reduplication seems to occur concurrently with the cl prefixes. I do not know what to say about these at this point.

reduplication and verbal number in mẽbengokre

5

265

Pluractionality versus Onomatopoeia

A problem found in many examples of reduplication is that one does not readily find a base out of which the reduplicated stem is formed. In some cases, the absence is plausibly an artifact of historical erosion, while in others one might wonder whether the reduplication is not of the onomatopoetic type, imitating a sound, rather than a marker of repeated action. This is the dilemma that I will address in this section. Consider first the following examples, all of which include some sense of repeated action: (32) krakrak ‘to break up, munch grains’ titik ‘to beat repeatedly, type’ pepek ‘to droop repeatedly’ These examples could be considered to employ reduplication as a marker of verbal plural, as was the case with the verbs examined so far, but they could equally well be thought of as imitating an event that includes repetition at some level of perception, without necessarily remitting to a notion of verbal plural. A more extreme example is found in verbs that refer to noise-generating events, and which are transparently onomatopoetic: (33) 'à'àk gogo kangãngã kangêngê karõrõ keket krikrit ngjengjek rãrãk tertet xorxot

‘to shake a rattle’ ‘to make noise’ ‘to moan, complain, buzz’ ‘to wave a rattle’ ‘to snore, roar’ ‘to laugh’ ‘to make noise (peccary, snake)’ ‘to buzz, squeak’ ‘to roar, thunder’ ‘to shiver’ ‘to vibrate, oscillate’

It would be difficult to claim that these are anything but onomatopoeias. To argue that ‘to laugh’ or ‘to tremble’ is understood by Mẽbengokre speakers as a repeated semelfactive event seems quite far-fetched, as it would imply a granularity that seems quite alien to the way humans perceive such events. Rather, one would rather claim that, e.g., keket imitates the sound of laughter, and tertet the sound of clattering teeth. In all of these examples, the fact that

266

salanova

we find no base from which the reduplicated verbs might be derived is an important confirmation that the pluralizing reduplication is not at work in forms such as (33) and possibly (32). However, there are cases where a base cannot be identified, but which do not denote noise-generating activities, thus casting doubt on them being onomatopoeias: (34) rôrôk ‘to fall (plural)’ ngrengrer ‘to raze, destroy’ prõrprõt krôkrôk nhĩ 'ĩk

‘to float up and down in the water’ ‘to shake on branches (plural)’ ‘to hang (plural)’

(possible base ngrẽk ‘to stir’) (possible base prõt ‘to run, flow, boil’)

To put this puzzle behind us, I will draw a line between those cases where a base for the reduplicated form can be identified and those cases where it cannot. Though this choice puts non-onomatopoetic examples like rôrôk, krôkrôk and possibly titik on the wrong side of the line, most of the verbal plurals do have an identifiable base, and most of the cases where a base cannot be identified do not involve a repeated action or plurality of participants in any obvious sense.17 The base of reduplicated words such as rôrôk and titik would be, in my account, the victims of historical attrition, and might be found to exist in related languages. Words such as keket and tertet, on the other hand, possibly never had a base that displays no reduplication.

6

Synopsis of Reduplication in Mẽbengokre Verbs

I am now in a position to summarize my thoughts about where reduplication fits into Mẽbengokre verbal morphology.

17

In fact, I’ve “put the puzzle behind us” with a simple stipulation that might not be fully satisfying from a theoretical point of view, and omitted further discussion of verbal plurals for lack of space. I invite the reader interested in the theory of pluractionality to consult Cusic (1981). In line with what this author concludes for several native languages of North America, the verbal plural in Mẽbengokre may subsume the pluralization of complex events, as well as the pluralization of punctual events that can be seen as forming a single complex eventuality. Given this, it seems possible to argue that there is a continuum between (25) and at least some of (33), with the latter being in the more granular extreme of the continuum.

reduplication and verbal number in mẽbengokre

267

Semantically, reduplication expresses iteration of an event or plurality of one of the participants. Formally, reduplication is a C(C)V-prefix attached to verbal roots. Like Marantz (1982) and much subsequent work in prosodic morphology, I mean this quite literally: the affix is a phonological skeleton, and its exact form is determined by phonological copying: there is no distinct morphological process of reduplication, only affixation. This is useful here as it allows me to identify the reduplicative morpheme with the other verbal prefixes: the reduplicant both occupies the same slot as ka-, ja-, etc., and has a similar C(C)V(C) shape. Furthermore, the other prefixes often also encode event iteration or participant plurality, though they do not do so exclusively, and are associated with transitive or unergative intransitive verbs, features that are also characteristic of reduplication, with very few exceptions. To conclude, let me flesh out the generalization about the canonical form of verbs, with which I set out at the beginning of section 4: (35)

Canonical form of Mẽbengokre verbs: a. Mẽbengokre verbs are either simple roots (monosyllables) or roots plus a cl prefix (sesquisyllables). b. The cl prefixes are: i. ja-, ka-, ku-, nhĩ -, py-, to-, C(C)V- and C(C)VC- for transitives; ii. djà-, dju-, e-, C(C)V-and C(C)VC-for unergative intransitives. iii.Meanwhile, to-and e- are cl prefixes used exclusively with ideophonic verbs. c. The cl prefixes are associated with the function marking transitivity, number and object properties on verbs. d. Unprefixed verbs include unaccusative intransitives and a special class of transitives which can be identified by the use of the accusative form of the third person object marker, ku- (3ac).

There are scarcely any verbs that escape this generalization. Two that I can think of are tak ‘hit’ and 'ôk ‘paint,’ which are transitive but bear no prefix, and inflect for third person with ∅ rather than ku-. There are also a few other possible cl prefixes that appear on a small number of transitive verbs. I did not list these above because they do not appear on enough verbs for me to make any meaningful claims. These isolated facts should serve as leads into newer inquiries. As for myself, I wish to end my paper with a comparison of the Mẽbengokre facts to those found in another Jê language, Kaingang, as I believe that the function of the cl prefixes of Mẽbengokre will only be fully understood when examined comparatively.

268 7

salanova

Reduplication and Number in Kaingang

Cavalcante (1987, 58 et ss.) describes the formation of verbal plurals in Kaingang. This morphological category is claimed to indicate the number of the absolutive argument (cf. also Urban 1985), though D’Angelis (2004) shows that it may also indicate repeated action with singular participants. The majority of verbs in Kaingang do not contrast a singular and a plural form. Of the approximately 200 that do in Cavalcante’s data, just over half do so by reduplication. In all the Kaingang examples, the first column gives the singular form of the verb, while the second column has the plural:18 (36) kõm mrãn nẽm gàg gunh jành

kõmkõm mrãnmrãn nẽmnẽm gàggàg gunhgunh jànhjành

‘to dig’ ‘to wet’ ‘to cut with scissors’ ‘to roast’ ‘to stick into the ground’ ‘to urinate’

(37) kanhin kãmun jẽmĩ wĩrĩn

kanhinnhin kãmunmun jẽmĩmĩ wĩrĩrĩn

‘to play’ ‘to measure’ ‘to feel with the hands’ ‘to turn around, to surround’

(38) gren gringren ‘to dance’ mranh mrynhmranh ‘to break’ (39) kar mrô nã nãn nê nĩ non nũr

18

kankar mrôgmrô nãgnã nãgnãn nêgnê nĩgnĩ nugnon nũgnũr

‘ready’ ‘to bathe’ ‘to lie down’ ‘to float’ ‘to hide in the earth’ ‘to sit down’ ‘to open’ ‘to sleep’

I have converted Cavalcante’s transcriptions to the orthography for simplicity, as nothing hinges on the phonological detail. The letters m, n, nh and g represent nasal or partially denasalized stops (i.e., g → [ŋ, gŋ, ŋg, gŋg], according to context); other symbols have approximately the same value as in Mẽbengokre.

reduplication and verbal number in mẽbengokre

269

The basic pattern is of a CVC reduplicant that is either prefixed (cf. (36)) or infixed before the stressed syllable of the base (cf. (37)). In some cases (cf. (38)) there is vowel raising on the reduplicant, something which incidentally serves to show that the process in question is in fact prefixation rather than suffixation. Yet another variant of the reduplicant has fixed segmentism, either replacing a segment of the base (as in (kankar), (nugnon)) or filling in in cases where the base does not have a coda (as in nĩgnĩ, mrôgmrô). Reduplication sits side by side with other morphological devices as a marker of the verbal plural. With monosyllabic stems, the most common device other than reduplication is prefixation, of either a CV or a CVC prefix (cf. (40)). With stems that are already bisyllabic, one finds various forms of substitution of the first syllable, with the most common being the augmentation of a CV first syllable to CVC, by the infixation of g (cf. (41)): (40) fã rũm grĩn tu sĩn ter

kygfã yogrũm kugrĩn, tugrĩn pãgtu kãsĩn kãgten

‘to cry’ ‘to shake, move’ ‘to wind up’ ‘to carry (long object)’ ‘to make small’ ‘to die’

(41) kafãn fãnãn kagje peju pafãm kãgmĩ tãnfyn

kygfãn fãgnãn kygje pigju pygfãm kugmĩ jãnfyn

‘to husk corn’ ‘to spend almost all’ ‘to tie a knot’ ‘to steal’ ‘to nurse’ ‘to grab, hold’ ‘to raise’

To sum up, reduplication and prefixation together account for the expression of plural in all but ten of the monosyllabic stems that mark it; prefix “substitution” accounts for the expression of the plural in three quarters of bisyllabic stems that mark it, while most of the rest reduplicate and become trisyllabic. Prefixes that indicate the plural come from a closed set (in Cavalcante’s analysis, ky-, kã-, ko-, ku-, pã-, tu- and jo-, but the elements in this set may also be augmented by a -g-). This is as far into Kaingang as I need to go for the purpose at hand. This quick incursion only intends to show that it would be plausible to advance the hypothesis that “reiterative” reduplication in Mẽbengokre verbs is a reflex of what one sees more pervasively or robustly here. Like in Mẽbengokre, verbs in

270

salanova

Kaingang are either monosyllabic or sesquisyllabic. Also like in Mẽbengokre, but perhaps even more clearly and productively, number marking affects the short syllable, either by replacing it, or by adding one where there is none.19 In both languages, the morpheme that is added may or may not be reduplicative. I have not attempted a systematic comparison in search of cognates or more quirky structural similarities, since the data are still relatively patchy.

8

Conclusions

Even though the data on which I can base my conclusions is incomplete, I hope to have shown the reader that there is a well delimited subclass of reduplicative phenomena in Mẽbengokre that fits in with the morphological template of the Mẽbengokre verb. I also hope to have shown more generally that Mẽbengokre verbs have structure beyond what is easily segmentable. Further in support of this analysis, I noted that it makes the marking of verbal number in Mẽbengokre and Kaingang virtually identical: both are expressed through prefixation, reduplication, and substitution of affixes. A research program stems from this overview, which I hope will be taken up by other researchers working in the language family: namely, to understand the interesting morphological sub-regularities found in the verbal lexicon of the Jê languages, and to investigate the subtleties in verbal meaning that lurk under the deceptively simple morphology of these languages.

Acknowledgements My warmest thanks to my main language consultants, Bep Kamrêk Kayapó and Ikrô Kayapó, for their patience in teaching me their language. I would also like to thank María Luisa Rivero, the editors of the volume and an anonymous reviewer for helpful comments on a previous draft of this paper.

19

Kaingang also has a small set of verbs where the number opposition is marked by suppletion of the stem.

reduplication and verbal number in mẽbengokre

271

References Alves, Flávia de Castro. 1999. Aspectos fonológicos do Apãnjekra (Jê). MA thesis, Universidade de São Paulo. Burgess, Eunice, and Patricia Ham. 1968. Multilevel conditioning of phoneme variants in Apinayé. Linguistics 41: 5–18. Callow, John. 1962. The Apinayé language. PhD diss, London University. Cavalcante, Marita Pôrto. 1987. Fonologia e morfologia da língua Kaingang: o dialeto de São Paulo comparado com o do Paraná. PhD diss, Universidade Estadual de Campinas. Cusic, David D. 1981. Verbal plurality and aspect. PhD diss, Stanford University. D’Angelis, Wilmar da Rocha. 2004. Concordância verbal de número em Kaingáng: algumas pistas. LIAMES 4: 71–81. Dourado, Luciana. 2001. Aspectos morfossintáticos da língua Panará (Jê). PhD diss, Universidade Estadual de Campinas, Campinas. Marantz, Alec. 1982. Re: Reduplication. Linguistic Inquiry 13: 435–482. Matisoff, James. 1973. Tonogenesis in Southeast Asia. In Consonant types and tone, Southern California Occasional Papers in Linguistics, 71–95. Linguistics department, University of Southern California. Oliveira, Christiane Cunha de. 2005. The language of the Apinajé people of Central Brazil. PhD diss, University of Oregon. Popjes, Jack, and Jo Popjes. 1986. Canela-Krahô. In Handbook of Amazonian languages, ed. Desmond Derbyshire and Geoffrey Pullum, volume 1, 128–199. Berlin: Mouton de Gruyter. Salanova, Andrés Pablo. 2001. A nasalidade em Mẽbengokre e Apinayé: o limite do vozeamento soante. MA thesis, Universidade Estadual de Campinas. . 2004. Subtractive truncation in Mẽbengokre. Manuscript. Massachusetts Institute of Technology. . 2007. Nominalizations and aspect. PhD diss, Massachusetts Institute of Technology. . 2011. A flexão de terceira pessoa nas línguas Jê. LIAMES 11: 75–114. Santos, Ludoviko dos. 1997. Descrição da morfossintaxe da língua Suyá/Kĩsêdjê (Jê). PhD diss, Universidade Federal de Santa Catarina, Florianópolis. Stout, Mickey, and Ruth Thomson. 1974. Fonêmica Txukahamẽi (Kayapó). Série Linguística 3: 153–176. Trapp, Earl. n.d. List of Kayapó verbs. Unpublished manuscript. Urban, Greg. 1985. Ergativity and accusativity in Shokleng (Gê). International Journal of American Linguistics 51: 164–187. Wiesemann, Ursula. 1972. Die phonologische und grammatische Struktur der KaingangSprache. [Janua Linguarum, Series Practica]. The Hague: Mouton.

272

salanova

Abbreviations 1 2 3 ac anticaus antipass cl erg ideoph

first person (exclusive) second person third person accusative case anticausative antipassive verbal “classifier” ergative case ideophone

n neg nfut nom red v

nominal (non-finite) form of the verb negation non-future nominative case reduplicant verbal (finite) form of the verb

Affixes are separated by hyphens (-), except for partial reduplicative affixes, which are separated by tildes (~). Infixes are surrounded by angled brackets (⟨⟩). Morpheme boundaries are not indicated where they are irrelevant.

chapter 11

Forms and Functions of Reduplication in Tupian Languages Wolf Dietrich

Reduplication in some Tupi-Guaraní languages has been described by Françoise Rose (2005). In this article I will give an overview of reduplication phenomena in other Tupi-Guarani languages and in some languages of other families of the Tupian stock. As has been outlined by Rose (2005), I will argue that the distinction between event-internal plurality in the case of monosyllabic reduplication and event-external plurality in that of disyllabic reduplication may be confirmed in some cases but has merged into one general function in most of the modern languages. This single function is iterative or distributive plurality of the process, quantifier, or expressed quality of the subject or the object of the verb.

1

Introduction

The aim of the following contribution is to give an overview of reduplication phenomena in some languages of the Tupi-Guarani family (henceforth TG) and the Tupi linguistic stock.1 Building on the foundation of Rose’s (2005) seminal paper, which includes semantic and phonological interpretations of reduplication in various Tupi-Guarani languages (Emerillon, Wayãpi, Kamayurá, Tupinambá, Urubu-Kaapor, Chiriguano, and Guarayo), the goal of this article is to present new data recently obtained from other languages of this extensive linguistic stock and to examine the existing theoretical issues. Reduplication will be studied here in the following Tupi-Guarani languages: Mbyá, Paraguayan Guarani, Chiriguano, Yuki, Siriono, Parakanã, Avá-Canoeiro, Tapirapé, Asurini of the Tocantins River, Anambé, Araweté, Parintintin, Guajá, 1 Tupi-Guarani is one of the ten families of the Tupi stock. With the exception of Tupi-Guarani, all of the families are limited to the Amazon basin. Only some of the almost 40 Tupi-Guarani languages are spoken today north and south of it, i.e., Wayãpi and Emerillon in French Guiana; and Mbyá, Kaiwá, Avá-Nhandeva and Paraguayan Guarani in the River Plate basin; Chiriguano and Tapiete in the Chaco area.

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_012 .

274

dietrich

Ka’apor, Kokama, Nheengatu and in some languages of other families of Tupian stock: Makurap, Tuparí (Tupari family), Juruna (Juruna family), and Munduruku (Munduruku family). As far as I am aware, there is no information available in existing linguistic descriptions about reduplication in such TupiGuarani languages as Kaiwá, Avá-Nhandeva, Xetá, Aché, Kayabí, Suruí, Apiaká, Amondava, Tembé and Zo’e.

2

The Nature of Reduplication

2.1 Reduplication and Lexical Repetition Reduplication, as a word-formative device, is a morphological process. However, it is determined not by morpheme boundaries since it involves the repetition of parts of the utterance independently from the morphological segmentation, i.e. independently from the boundary between grammatical morphemes and lexical stems. It is iconic in the sense that its meaning of ‘multiplicity’ and ‘repetition’ is based on its morphology (cf. Rose 2005, 352). I shall not consider here total iconicity as it is evidenced in onomatopoeic lexemes of all languages, for instance Paraguayan Guarani pururũ ‘to be crisp, crackling (bread, pastry)’ or perere ‘to flutter (bird, hen).’ We shall not consider either the very common phenomenon of lexical repetition which is often used for emphasis (e.g. German ein weiter, weiter Weg ‘a long, long way,’ Italian una donna brutta brutta ‘a very ugly woman,’ piano piano siamo arrivati fino quassù ‘little by little we arrived up there,’ Hungarian a haja volt fekete fekete ‘his hair was deep black’). The primary objective of this paper is to describe reduplication as a grammatical process of marking aspectual functions of verbal and nominal predicates in Tupi languages. In addition to this, we sometimes observe other functions of reduplication, such as plurality in any kind of arguments. 2.2 Functions of Reduplication A dichotomy between monosyllabic and disyllabic reduplication has been established for several TG languages (Rose 2005, 352–357). Monosyllabic reduplication means the repetition of one syllable, either by anticipation (see ka~ kaʁaj in example (6)), or by repetition of the final syllable of the base (see je'ẽgã~gã in example (24)). Disyllabic reduplication refers to the repetition of the last two syllables of a base which itself has at least two syllables (see oguata~guata in example (2)). The dichotomy conveys the distinction between successive and frequentative actions, respectively. However, as Rose shows, the distinction between event-internal (successive actions or successive stages of one verbal action) and event-external (actions repeated on different occasions,

forms and functions of reduplication in tupian languages

275

such as habituality) repetitions of actions (see Cusic 1981) is much more adequate for the description of those languages than the merely formal distinction between monosyllabic and disyllabic reduplication. As she points out (Rose 2005, 356–357), this distinction is not always applicable since the meaning of reduplicated verbs depends first of all on context and on the lexical meaning of the verb. Even disyllabic reduplications may convey the meaning of eventinternal plurality. In any case, this is a matter of verbal aspect. The general function of reduplication is repetition of the action, state or quality expressed by the verb or nominal predicate. According to the context, reduplication may mean repetition on one single occasion so that the action is completed by successive movements or is intensified by repeated or protracted actions (event-internal plurality), or it means repetition on various occasions and may refer to an attitude or (habitual) practice (event-external plurality). Generally, this is called the iterative or frequentative aspect (see Comrie 1976, 24–32 and 42–44). The distinction between monosyllabic and disyllabic reduplication has been observed for Tupinambá and Emerillon (Rose 2005). In the following description it will be established for Avá-Canoeiro, Asurini of the Tocantins River, Anambé, Parintintin, Guajá, Kokama, and Juruna. It does not exist in many other languages: Mbyá, Paraguayan Guarani, Chiriguano, Tapiete, Guarayo, Yuki, Siriono, Parakanã, Tapirapé, Araweté, Kamayurá, Wayãpi, Ka’apor and some languages of other families of the Tupian stock: Makurap, Tuparí (Tuparí family), and Munduruku (Munduruku family). I, therefore, will consider the distinction only where it occurs. 2.3 Anticipation and Repetition of Syllables The term and the process of reduplication have been known in Indo-European languages since the studies of Ancient Greek and Latin in Antiquity. In Greek and Latin it was used in the formation of the perfect tense (cf. Ancient Greek kaléo ‘I shout’—ké~klēka ‘I have shouted,’ Latin tango ‘I touch’—te~tigi ‘I (have) touched’). Reduplication in these languages was made by anticipating the initial consonant plus a stressed /e/ vowel, the whole resulting in a stressed syllable preceding the root. With regard to TG and Tupi languages we, therefore, might think it could be important to check which kind of reduplication we are confronted with: anticipated or repeated syllables or both. Rose (2005, 363–364) offers a convincing solution for her data (Tupinambá, Emerillon, and Wayãpi). According to her phonological analysis, the copy, independently from morphological boundaries, generally precedes the base. However, a look at the data I present in this paper makes it clear that the picture is not straightforward for all TG languages. Both of the copying processes may be found in languages described here. Redu-

276

dietrich

plication means anticipation of the last one or two syllables, without the possible coda consonant in some of the TG languages. In accordance with Rose (2005, 363–364) this is an argument for anticipatory reduplication in Parakanã, Avá-Canoeiro (see ka~kaʁaj in example (6)), Asurini of the Tocantins River (see the analysis of pá~páng in example (15)), Araweté, Parintintin, Ka’apor, Tapirapé, Guajá (see ma~marõ in example (36)), Kokama, and Makurap. On the other hand, different kinds of phonological modifications in the copy suggest the repetition of syllables in Mbyá, Paraguayan Guarani, Guarayo, Yuki, Siriono, Anambé, Juruna, Munduruku. Monosyllabic reduplication very often consists of the repetition of the final syllable of a longer root (see je'ẽgã~gã in example (24)). The phenomenon of the dropping of possible coda consonants or approximants in reduplication processes is unknown in languages of other Tupi families besides Tupi-Guarani. Thus, we find secondary final consonants in reduplicated forms in Makurap (see top~tow-a in example (98)), Tupari, Juruna, and Munduruku (see ʧokʧok~ʧo-ŋ in example (108)). 2.4 Patterns of Reduplication Reduplication is not a morphological device such as suffixing or prefixing. Though rule-based, it is basically a ludic process which enlarges the base in order to “expand” its lexical meaning. The process is rule-based because the “expansion” of the lexical meaning is expressed as grammatical morphemes, expressing different kinds of quantification or “plurality”: plurality of actions (successive, frequentative or habitual) or plurality of arguments involved in the action (see also 2.2 and 4.1.7, after example (59)). In the sense of ludic creation, reduplication has to do not only with morphology, but also with prosody and rhythm. Some authors consider reduplication as belonging to morphophonology (Rose 2005; Vallejos Yopán 2010); others adopt a metric theoretical approach (e.g. Fargetti 2007). In my view, reduplication is a ludic strategy of persuasive speech, a morphophonemic device consisting of retardation and release, especially in the case of anticipatory reduplication.2 Anticipatory and repeated reduplication form a unitary process. The meaning that results from it is iconic in the sense that Rose describes (2005, 352): the added meaning of intensity with a verbal action, performed as an intense or progressive action (event-internal plurality) or a distributive, iterative or 2 This insight results from recurring discussions with my wife, Marta Dietrich, on the subject of reduplication. She gave the final touch to my first ideas about this intriguing phenomenon. On the other hand, as a native speaker of Hungarian she also provided me with the examples given in footnote 17.

forms and functions of reduplication in tupian languages

277

habitual action (event-external plurality), is expressed by formal repetition. The existing data make clear that the semantic effect of reduplication results from the combination, generally, of a reduced form of the root—possibly without its coda consonant or approximant or without its full nasality—plus the full lexical root, which may be completed by grammatical suffixes of tense, aspect, number, evidentiality, and so on. The pattern of this kind of reduplication is the double form, a climax which goes from a reduced form to the full form. An alternate reduplication process consists of the full form followed by a reduced form, generally monosyllabic. The order of full and reduced forms is less important than the pattern itself. The pattern, of course, may consist both of two equal or two unequal forms. Each language has developed its own traditions, but, beyond them, speakers may create their own possibilities of making new distinctions, even by forming double reduplications (for triple forms, see example (46c)).3 2.5 Phonology, Syllable Structure, and Stress The phonological systems of all the Tupi-Guarani languages studied in this chapter are very similar. Generally, they have six oral vowels /i, ɨ, u, e, o, a/ and their nasal counterparts /ĩ, 󰀰͊, ũ, ẽ, õ, ã/. Some languages have only five oral vowels: Parakanã lacks /u/ and every kind of nasal vocalic phonemes. Araweté has oral and nasal vowels but lacks /u/ and /ã/. Anambé has the six standard oral vowels but lacks /õ/ in the nasal series. Asurini of the Tocantins River does not have nasal vowels and has no /u/ in the oral series. Kokama does not have nasal vowels either and has no /o/ in the oral series. Tapirapé lacks /u/ and /ũ/. Nheengatu does not have the central vowels /ɨ/, /󰀰͊/, just as Ka’apor does not have /󰀰͊/. Yuki has a system of five vowels, without /ɨ/, but all of them short and long, oral and nasal; the whole system has 20 vocalic phonemes. Within the four Tupi languages from other families studied here (Makurap and Tuparí from the Tuparí family, Juruna from Juruna family, and Munduruku

3 Examples taken from modern linguistic studies will generally be given in the spelling used by the author. In other cases, early grammatical descriptions, dictionaries made by non-linguists, and so on I will use a unified phonological spelling. Early missionary descriptions do not have glosses. Therefore, glossing which is not from modern linguistic studies is based on my own analyses. The only exceptions are ⟨y⟩ for /ɨ/ and ⟨ỹ⟩ for /͠ɨ/ in a number of cases, and ⟨’⟩, which is widely used for /ʔ/., Tense in English glosses of unmarked TG verb forms will be arbitrarily present or past because TG languages do not distinguish present from past; they all have a single unmarked form which includes English past and present. I claim that the first meaning of this “non-future tense” is past, but we may translate it sometimes by present tense for pragmatic reasons (see Dietrich 2010, 69–71).

278

dietrich

from Munduruku family) both Makurap and Munduruku have nearly the same system of five oral and five nasal vowels, with a supplementary /ũ/ (and no /u/) in Makurap. The systems of segmental vowels of Tuparí and Juruna are characterized by the opposition of quantity, short and long vowels. In both languages this occurs only for oral vowels. The nasal vowels are /ĩ, 󰀰͊, ẽ, õ, ã/ in Tuparí and /ĩ, 󰀰͊, ũ, ẽ, ã/ in Juruna. Juruna and Munduruku are tone languages. The systems of consonant phonemes of nearly all the Tupi and Tupi-Guarani languages studied are very similar, too. All of them have voiceless stops, /p, t, k/, most of them labialized occlusives, such as /kw/and /gw/ (except Nheengatu, Kokama, Tuparí, Makurap, Juruna, Munduruku). Voiced stops such as /b, d, g/ are rare; they only occur in Nheengatu, Yuki and the four languages outside the Tupi-Guarani family. Most of the languages have the glottal stop /ʔ/ (except Chiriguano, Siriono, Avá-Canoeiro, Guajá, Nheengatu, Kokama, Makurap). They all have the nasal consonants /m, n, ƞ/, the approximants /j/and /w/, and a flap /ɾ/. With the exception of Avá-Canoeiro and Makurap, they all have the fricative /h/. Tupi-Guarani languages generally distinguish between two kinds of sibilants: fricative /s/ and the affricate /ʧ/, also occurring as fricative /ʃ/. Some Tupi-Guarani languages, however, do not have /s/ but only the affricate /ʧ/ (Parakanã, Araweté, Anambé, Avá-Canoeiro, Tapirapé, Parintintin, Guajá); Asurini of the Tocantins River has /s/ and no affricate /ʧ/. Kokama has /ʦ/, /ʧ/ but no /s/; whereas Makurap has /ʦ/ and /ʤ/ but no /s/. Juruna and Munduruku have more sibilants: Juruna has five (/s/, /z/, /ʃ/, /ʧ/, /ʤ/), Munduruku four (s/, /ʃ/, /ʧ/, /ʤ/). Syllable structure in all Tupi-Guarani and Tupi languages studied is V, CV, or CVC. In languages such as Juruna we may distinguish between light syllables with one mora (V, CV) and heavy syllables with two morae (CVV, VC, or CVC). Generally, stress in isolated words of two and more syllables is on the last syllable in the Tupi and Tupi-Guarani languages discussed here, with the exception of Chiriguano, Siriono, Yuki, Avá-Canoeiro, and Kokama, where it is on the penultimate syllable.

3

Early Descriptions of TG Languages

Reduplication has been described in an early missionary grammar of Tupinambá by Anchieta (1595) and in grammars of Paraguayan Guarani (henceforth PG) by Montoya (1993 [1640]) and Restivo (1724).4 These three works each

4 Both authors came to the conclusion that there was no monosyllabic reduplication in Ancient

forms and functions of reduplication in tupian languages

279

include an entire chapter on reduplication. In his chapter on “the repetition of verbs” Anchieta (1595, 52–54) distinguishes monosyllabic and disyllabic reduplication. The first one (oro-sem ‘we go out’—oro-se~sem ‘we go out one after another’) means successive actions or successive movement of one extended action, the second one (a-se~asem ‘I went out many times,’ my segmentation and spelling) expresses repeated actions. Ancient Guarani shows very similar facts, with the difference that there is no monosyllabic reduplication. According to Montoya (1993, chap. XII, 119–120 and 241–243, my translation and my spelling, on the basis of modern Guarani): the repetition of verbs and nominal predicates is very common and belongs to the good and elegant use of the language. It means frequency, or to do something in a high degree, or successively, or (action that comes off) on every side, or for elegance and grace, as for instance akaru karu guitekóvo ‘I am eating all the time’ and ajerure rure ‘I asked for it all the time,’ and always the last two syllables of the verb or noun are repeated, and whenever it is disyllabic it is repeated entirely, as in … apó apó ‘I am hopping’ [Spanish original ando saltando]. Montoya (1993, 119–120) gives more details explaining that whenever the lexeme ends in a falling diphthong [Spanish “contractos”] ending in -i or -u, first the -i or -u is dropped and then it appears after the repetition of the last two syllables, as in a-hyvykói ‘am digging’ → a-hyvyko~vykói ‘I am going on digging,’ and che péu ‘I am having pus’ → che pe~péu ‘I am having a lot of pus.’ Verbs ending in a consonant first drop the final consonant and then present the last syllable or syllables in their full form, as in a-mokong ‘I swallow’ → a-moko~kong ‘I am continuously swallowing.’ It is evident that the dropping of the coda consonant before the reduplication is due to euphonic reasons. Simple repetition would lead to undesirable consonant clusters like [ŋgk] in *a-mokong~kong or [nm] in Parakanã *o-mokon~mokón. Rose (2005, 358) gives evidence for this, demonstrating, by her data from Emerillon, that “the reduplication process in TG languages creates reduplicants consisting only of monomoraic syllables.” Giannecchini (1896, 81–82) gives the following rules for Chiriguano (my translation and spelling):

Guarani, as Jensen presumed (Jensen 1989, 119). Just like in modern Paraguayan Guarani, there was only disyllabic reduplication.

280

dietrich

The Chiriguano used to frequently repeat the words in order to differentiate meanings. You must pay attention to the fact that 1.) disyllabic words are repeated completely, e.g. mõkwi-mõkwi ‘two at a time,’ óke-óke ‘he is sleeping peacefully’ …, but if the word has more than two syllables, you will repeat only the last two ones, as in ainũpa-nũpa ‘I hit and went on hitting him,’ ajembóe-mbóe ‘I learned and learned,’ i-kávi-kávi ‘it was excellent,’ and so on.5 One of his more extensive examples is analyzed in (1): (1)

ó-pa che-asója u-mondóro~ndóro 3-be.complete 1sg-poncho 3-tear.up~red ‘He tore into pieces my whole poncho.’ (event-internal plurality)

There is no modern description of Guarayo. Reduplication in this language is described only by Hoeller (1932, 139–140, my analysis and spelling in the examples (2)–(5)), “Generally, the last two syllables of a word are repeated: this kind of repetition—which is very common and a peculiarity of this language— expresses that the action occurs often, continuously, gradually, to a high degree, or discontinuously, only sometimes” (my translation). (2) o-guata~guata jou-reko-i 3-walk~walk 3-be-ev ‘They are walking continually.’ (3) a-mõa~mõa é che-pya pype 1sg-think~think apart 1sg-heart within ‘I am meditating on it in my heart.’ (4) o-jo-pỹro~pỹro voi i-a 3-rec-help~help immediately 3-fruit ‘Its fruit is ripening continuously.’ (lit. ‘Its fruit is helping one another continuously.’)

5 Chiriguano mõkwi, in some dialects mõkoi, means ‘two,’ ó-ke ‘3-sleep’ means ‘he/she sleeps’ or ‘is sleeping’ or ‘they sleep’ or ‘are sleeping.’ Ai-nũpa ‘1sg-hit’ means ‘I hit’; a-jembóe ‘1sglearn’ means ‘I learn’ or ‘I learned.’ The nominal expression i-kávi ‘3-good’ means ‘it is/was good.’

forms and functions of reduplication in tupian languages

281

(5) morõchi~rõchi white~red ‘more or less whitened’ Example (5) makes clear that this kind of reduplication corresponds to the repetition of the last two syllables and that the meaning of reduplication in non-predicative lexemes may be different from the event-internal or eventexternal meaning of predicates. We also find this kind of semantic weakening by repetition in PG vai~vai ‘tolerably well,’ whereas the simple vai means ‘bad’.

4

Reduplication in Modern Tupi-Guarani and Tupi Languages

4.1 Distinction of Monosyllabic and Disyllabic Reduplication 4.1.1 Avá-Canoeiro The examples given by Borges (2006, 163–165) do not allow a clear distinction between monosyllabic and disyllabic reduplication. They show anticipation of the first syllable with reduplication to the left as in (6) and (7). (6) a-ka~kaʁaj tʃi-po 1sg-red~scratch my-hand ‘I scratched my hand repeatedly.’ (7) e-kwa~kwaʁ jawaʁa 2sg-red~beat dog ‘You thrashed the dog.’ (8) o-ɨʁ~ɨʁ 3-be.born~be.born ‘(The moon) was rising more and more.’ All the studies of reduplication in TG languages consider the problem of dropped final consonants in the copy. With regard to this issue, the observation made by Rose (2005, 358) is applicable here (see (8) and (12)) and to all TG languages but not to languages of other Tupi families (see sections 4.2.7.-10.): “We suggest that the absence of final consonant in the reduplicant may be explained in terms of moras: the reduplication process in Tupi-Guarani languages creates reduplicants consisting only of monomoraic syllables” (Rose 2005, 358). Consequently, the possible syllable structure of a copy is CV (see (6) or VC (see (8)). The following examples of disyllabic reduplication do not convey a

282

dietrich

different function from monosyllabic ones, i.e. they do not refer to eventexternal plurality but to an intensified action (examples (9) and (10)) or to ‘diffusion’ (pluralization of objects involved or of events, examples (11) and (12)): (9) o-kɨti~kɨti i-kwã 3-cut~cut 3-finger ‘He deeply cut his finger.’ (10) tata-ɨɾa Iawi o-apɨ~apɨ fire-bee Iawi 3-burn~burn ‘The fire-bee burnt Iawi a lot.’ (11) o-jika~jika 3-break.into.pieces~break.into.pieces ‘It broke into many pieces.’ (12) a-puka~pukaj 1sg-red~shout ‘I shouted very much.’ 4.1.2 Asurini Reduplication in Asurini of the Tocantins River is described in Cabral and Rodrigues (2003, xxi–xxii), with a distinction between monosyllabic and disyllabic reduplication. The first one means ‘actions realized successively’ (examples (13b), (14b), (16b) and (18b)). Disyllabic reduplication means ‘action is frequently or repeatedly realized.’ It is expressed by the repetition of the last two syllables (examples (20), and (21)), or, in one case, the first two syllables (example (15)). In examples (13) and (14) the contrast between monosyllabic and disyllabic reduplication is impossible because of the monosyllabic base. Even though both are monosyllabic, the semantic interpretation of (13b) and (14b) is frequentative repetition, according to the context given in the source (Cabral and Rodrigues 2003, xxii):6

6 Examples (13a), (13b), and (21) refer to hó ‘go’ as mentioned in Cabral and Rodrigues (2003, xxii). However the same root is represented as há in the dictionary part (Cabral and Rodrigues 2003, 84) and in several other sources of Asurini of the Tocantins River. In spite of this, I give the examples just as they are represented in the introduction of Cabral and Rodrigues’ dictionary (2003, xxii).

forms and functions of reduplication in tupian languages

283

(13) a. pe-hó you-go ‘You (pl) went.’ b. pe-hó~hó you-go~go ‘You (pl) went one after another.’ (14) a. o-páng 3-wake.up ‘They woke up.’ b. o-pá~páng 3-red~wake.up ‘They woke up one after another.’ (15) apa~a-páng red~1-wake.up ‘I woke up many times.’ In the following examples we contrast monosyllabic with disyllabic (in this case, full) reduplication, respectively, illustrating successive meaning in (16b) and frequentative meaning in (17): (16) a. a-soká 1sg-kill ‘I killed him/them.’ b. a-soká~ká 1sg-kill~red ‘I killed them one after another.’ (17) a-soká~soká 1sg-kill~kill ‘I frequently killed them.’ The verb sewyt ‘to return’ (example (18)) has different allomorphs depending on the phonological context.7 7 The underlying Asurini form sewyt has the allomorph sewyn before alveolar stops such as /t/,

284

dietrich

(18) a. o-sewý 3-return ‘They returned.’ b. o-sewý~wýn 3-return~red ‘They returned one after another.’ The fact that all grammatical suffixes (TAM, number, negation, locative) and particles (evidentials, for instance) are excluded from reduplication, as also attested for Emerillon by Rose (2005, 358), is evidenced by example (19). Prefixes or parts of prefixes, on the other hand, are possibly included in the reduplicants (see examples (21) and (33)): (19) n-o-sewý~wýr-ihi neg-3-return~red-neg ‘They did not return one after another.’ (20) o-sewý~sewýn 3-return~red ‘They returned many times.’ In example (21) reduplication is evidenced as a morphological process, which happens independently from morpheme boundaries. The disyllabic reduplicant follows the base, similar to (20): (21) ere-ho~reho 2sg-go~red ‘You (sg) frequently go.’ 4.1.3 Anambé Anambé is another language where there is monosyllabic and disyllabic reduplication of the final syllables of predicates (see Julião 2005, 193–195). Examples (22) and (23) refer to the stem mukũ ‘to swallow’ that is mentioned also in other grammars of Tupi-Guarani languages (see section 3 with regard to Montoya and

sewyr before a following vowel (see (19)), and sewy in word-internal position. Apparently, the reduplicant is based on the form with the final consonant, sewyn, which is shortened to -wyn. Cabral and Rodrigues (2003, xxi) did not fully contextualize sewyn in (20) and -wyn in (18b).

forms and functions of reduplication in tupian languages

285

see examples (60) and (75) below). Monosyllabic reduplication means ‘successive actions on a single occasion,’ disyllabic reduplication means ‘frequentative action’: (22) pirã ũ-muku~kũ fish 3-swallow~red ‘He swallows fish one after another.’ (23) pirã ũ-muku~mukũ fish 3-swallow~swallow ‘He swallows fish repeatedly.’ (event-external plurality) In examples (22) and (23) we see another phonological effect of reduplication processes which may consist of suppressing the nasality of the root in its first occurrence (muku instead of mukũ). As in the case of dropped coda consonants or approximants, the original shape of the base is preserved in the reduplicant. However, the following examples involving the verbal roots je'ẽgã ‘sing’ and nupỹ ‘beat,’ show reduplication without any suppressing of nasality: (24) kurumĩ ũ-je'ẽgã~gã boy 3-sing~red ‘The boy was warbling.’ (event-internal plurality) (25) membɨ ũ-nupỹ~nupỹ offspring 3-beat~beat ‘She beat her son many times.’ (event-external plurality) (26) ka'i ũ-upo~upo monkey 3-jump~jump ‘The monkey is hopping.’ (event-internal or event-external plurality) In example (26) the meaning depends on the context. It may refer to “hopping” as an event-internal plurality (to hop means ‘to move by a series of quick leaps’) or to a (event-external) habit (‘the monkey was hopping all the time’). 4.1.4 Araweté Solano (2009, 343–348), following Rodrigues (1953), describes reduplication in Araweté as a word-formation process to express ‘multiple realization of an event (successively or at the same time)’ in the case of monosyllabic redu-

286

dietrich

plication and ‘frequentative’ or ‘intensive aspect’ whenever there is disyllabic reduplication. However, although she gives lots of examples of dissyllabic reduplication, neither of the examples she provides as evidence for monosyllabic reduplication (Solano 2009, 343) is convincing. One of them is obviously disyllabic (her example 994), the other one does not correspond to the formal definition of reduplication, which includes that the copy may not be at a distance, separated from the root by several different morphemes, but should precede or follow immediately. Nevertheless, the examples of disyllabic reduplication involve both event-internal and event-external meanings. Examples (29) and (30b) show that the reduplicant generally follows the base, so we claim the existence of this order also in (27) and (28), where it is not obvious: (27) u-maʔẽ~maʔẽ u-ẽ he r-e 3-look~look 3-sit 1sg rel-loc ‘He is looking at me all the time.’ (28) a-nupĩ~nupĩ ku he iwi 1sg-beat~beat foc 1sg soil ‘I beat the soil several times.’ (29) ne r-uri~r-uri you.sg rel-gladness~rel-gladness ‘You are very very glad,’ Example (30) once more shows reduplication processes independent from morphological boundaries (see also example (15) and (21)). In (30) I provide a consistent morphological segmentation of Solano’s example (2009, 347). (30) uru-jĩ~rujĩ ku ure we. excl-run~red foc we.excl ‘We run very much.’ 4.1.5 Parintintin In Parintintin (see Pease 2007, 18–19), there is reduplication in all word classes, even in adverbs (see (32)). We find monosyllabic and disyllabic reduplication, possibly with the same semantic distinction as in some preceding languages. This distinction, however, may depend more on contextual conditions than on formal differences.

forms and functions of reduplication in tupian languages

287

(31) ga-ho~ga-ho-i he-go~he-go-top ‘He went and went.’ (seems to be event-internal plurality) (32) o-mano-g͠weri~g͠werĩ 3-die-about.to~about.to ‘(He/she/it) was about to die.’ (event-internal plurality to express intensity) (33) n-a-ɲa~naɲan-ahɨ-uhú-i neg-1sg-run~red-forcefully-int-neg ‘I’m not going to run and run forcefully.’ (34) o-ɲi-mo-pẽ~pẽ 3-refl-caus-break~break ‘He broke (all his bones).’ (35) o-mbogwɨ~gwɨ8 3-pick~red ‘picking’ It is difficult to determine whether the reduplicant precedes or follows the base. However, coda consonants are apparently dropped before a following consonantal onset, as in (33) where the order is base-reduplicant. Consequently, in examples (34)–(35) one could assume that the order is base-reduplicant as well 4.1.6 Guajá For Guajá, a TG language spoken on the border of the Brazilian states of Maranhão, Pará, and Tocantins, Magalhães (2007, 221–223) establishes three different types of reduplication: 1) monosyllabic reduplication of the first syllable expressing ‘quantity and intensity’ (36)–(39), 2) monosyllabic reduplication of the last syllable, meaning ‘successive actions’ (40), and 3) disyllabic reduplication of the last two syllables, expressing ‘iterative aspect’ (41)–(42):

8 In Betts (2012, 90), the isolated form -mbo'gwyr ‘pick mucuim parasites’ is listed with a final -r.

288

dietrich

(36) ma'awá iraparakwỹ ∅-ma~marõ who liana.species 3-red~cut ‘Who has cut so many lianas?’ (event-internal plurality) (37) a-wɨ~wɨ'ú 1sg-red~drink.water ‘I drank much water.’ (event-internal plurality, quantity) (38) ʧanĩ-a i-wa~waré cat-cas refl-red~lick ‘The cat licked itself extensively.’ (event-internal plurality) (39) a-me~me'ẽ tʃi h-aká-pa 1sg-red~look.around impf 3-seek-ger ‘I carefully looked around seeking him.’ (event-internal plurality) (40) ∅-japi~pí haraká i-pɨ-pe 3-throw~red caus 3-foot-ins ‘He kicked (the ball) several times.’ (successive actions at one opportunity) (41) taký-a irá ∅-marõ~marõ iká anỹ knife-cas tree 3-cut~cut mvt conn9 ‘And the knife went on cutting the tree.’ (42) a-me'ẽ~me'ẽ iká pé wý-ripi 1sg-look.around~look.around mvt way border-loc ‘I went on looking around along the road.’ Whenever the root is monosyllabic, each of the above mentioned meanings may be expressed by reduplication. This means that a possible functional distinction can only be made by means of a distinct morphology, for instance monosyllabic versus disyllabic reduplication. In other cases the meaning depends on the context, as in (43): 9 Guajá is one of those Tupi-Guarani languages which conserved the use of the case-marker -a and its allomorph -∅, by means of which arguments (subjects and objects) are distinguished from unmarked nouns in adverbial phrases and possessed nouns (as in (42): pé wy-ripi, lit. ‘along the border of the way,’ i.e. ‘along the road’). Differing from Magalhães (2007), I do not note here possible zero case markers.

forms and functions of reduplication in tupian languages

289

(43) a-kwa~kwa ha-r-ipá-rehe 1sg-fasten~fasten 1sg-cont-house-loc ‘I tied (the thatching) onto my house one after another.’ (event-internal plurality)/ ‘I tied (the thatching) onto my house well.’/ ‘I kept on tying (the thatching) onto my house.’ I assume that in the proper context the meaning could also be ‘I used to tie it up each time again’ (event-external plurality). 4.1.7 Kokama Vallejos Yopán (2010) gives information about reduplication in Kokama in three sections of her impressive thesis: first in a chapter on morphophonemics (2010, 132–134), then in the following one on morphology (2010, 153–154), and, finally, in a chapter on verb phrase devices (2010, 369–372). On the basis of morphophonemics, she distinguishes initial from internal reduplication. This distinction, however, is not fully convincing. What here is called “initial reduplication” should better be described in terms of our sections 2.3 and 2.4, in other words, within the concept of anticipated copy (part of the root) followed by the full form of the root: (44) chikari ‘look for’ → chika~chikari ‘keep on looking for, continously’ tsapuki ‘call’ → tsapu~tsapuki ‘keep on calling’ kupetaka ‘limp’ → kupe~kupetaka ‘hobble, be lame’ A reduplicated initial element may not be described as “repeated” (Vallejos Yopán 2010, 133), but it is anticipated. Some details relate to the form of the anticipated syllable or syllables. When the first syllable is bimoraic and the second one monomoraic, only the first syllable is anticipated, as in majnani ‘take care’ → maj~majnani ‘protect constantly,’ as if the weight of the first heavy syllable was experienced as a double one. “In three-syllable words whose first syllable contains only a vowel, the reduplicated portion is the first vowel and the onset of the second syllable” (Vallejos Yopán 2010, 133), as in the case of (45). (45) arɨwa ‘be on top’ → ar~arɨwa ‘be in a pile, one on top of the other’ ɨyɨ ‘to grill’ → ɨy~ɨyɨ ‘grill something for a while’ The second morphophonemic kind of reduplication is “internal” or “infixal reduplication” (Vallejos Yopán 2010, 134). This means that the second syllable

290

dietrich

of the base is anticipated in the reduplicant and then repeated as if it were the onset of the base. In disyllabic bases (see example (50)), the reduplicant is formed by the first syllable of the base, cut from its onset. In the following examples (46)–(50) the stems are ipama ‘stand up,’ erura ‘bring,’ kakɨrɨka ‘move,’ yaparari ‘sink,’ and yuti ‘stay.’ Segmentation is problematic because of the cleavage of the root whose component parts do not have a separate lexical meaning. Part of the base, its first syllable or the onset consonant of the first syllable, remains unanalyzable in the segmentation. I assume that the reduplicant precedes the base, just like in (44): (46) i⟨pa~⟩pama red~stand.up ‘stand up for a while’ (47) e⟨ru~⟩rura red~bring ‘carry’ (48) ka⟨kɨ~⟩kɨrɨka red~move ‘move repeatedly, like earthquake’ (49) ya⟨pa~⟩parari red~sink ‘keep sinking, disappearing from surface’ (50) y⟨ut~⟩uti red~stay ‘stay for a long time, remain’ About example (50) Vallejos Yopán (2010, 134) notes that “in two syllable words, the nucleus of the first syllable and the onset consonant of the second syllable are repeated.” The analysis of examples (46–50) is unsatisfying, since what we call here “infixing” does not describe well the nature of these reduplication processes. They are different from the infixing of normal grammatical morphemes. Reduplication, here even more than in other cases, is a morphological procedure which contains elements of ludic creation. In an example such as (49) we see the phenomenon of stylistic retardation (see also above 2.4.): The speaker makes a first approach ( yapa-) and then continues uttering nearly the full form,

forms and functions of reduplication in tupian languages

291

but, in a playful manner, omits the beginning, already mentioned in the first part of the utterance. In this analysis we do not have infixing, but a complex reduplicant and a base which is completed by the first part of the reduplicant. Both kinds of analysis are acceptable in my eyes. As for the meanings of Kokama reduplications, “there is no correlation between types of reduplication and semantic functions” (Vallejos Yopán 2010, 369). Most of the reduplications in (44)–(47) and (49)–(50) refer to eventinternal plurality, analyzed by the author as aspectual meanings, such as continuity, intensification, and distributive. Event-external plurality (iterative aspect) is found in kupe~kupetaka (44) and ka⟨kɨ~⟩kɨrɨka (48). Other iterative examples ((51) and (52)) are made on the basis of tsakamɨka ‘(to) cross’ and michiku ‘wrinkle.’ (51) tsaka~tsakamɨka red~cross ‘intertwine, interweave’ (52) michi~michiku-ka red~wrinkle-rei ‘wrinkle, fold (many times)’ 4.2 Languages with Only One Kind of Reduplication All the TG and Tupi languages that will be mentioned in this section present only one kind of reduplication, without any formal distinction between eventinternal and event-external plurality. If there is a difference of meaning, it depends on the context. In TG languages we can observe the general rule of dropping the coda consonant or approximant of the root in the anticipated copy (see above 2.3 and 3). In other Tupi language families this rule does not exist. In all languages the meaning is event-internal or event-external plurality of the expressed action or quality, depending on the context. 4.2.1 Mbyá In Mbyá (Dooley 2006, I, 8–9), just as in other TG languages, we find disyllabic reduplication in all word classes. In non-verbal expressions it generally refers to distributive concepts (referents appearing successively), for instance peteĩ ‘one’ vs. peteĩ~teĩ ‘one by one,’ mokõ~mokõi ‘two by two,’ or: (53) o-vaẽ-va'e~va'e 3-arrive-nom~nom ‘anyone who may arrive’

292

dietrich

Being nominalized, (53) is an example of non-verbal reduplication, and it has the same distributive meaning. Disyllabic verbal reduplication generally expresses continuity or iterative aspect (Dooley 2006, I, 8–9): (54) o-veve~veve 3-fly~fly ‘He/she/it flew and flew.’ (55) o-ike~ike 3-enter~enter ‘They kept on entering.’ (56) p-eka~eka-∅ 2pl.imp-search~search-obj ‘Search and search [it]!’ Mbyá does not have final consonants but may have postvocalic glides, such as -[j]. As the canonical form of reduplicants is monomoraic, the rule also applies in the case of final glides (see mỹi ‘stir’ in (58)). In (57) the phonological effect is the same, but in this case the final -[j] is part of the negative morpheme nd … i, which does not enter into the reduplication process. Example (57) shows that the reduplicant, a disyllabic element copied from the base, follows the latter. We could claim this order also in examples (53– 56). (57) nd-a'eve~'eve-i neg-be.fine~red-neg ‘It was getting worse and worse.’ (event-internal plurality, according to the context) In example (58), however, that order cannot be proven. As an infix, the reduplicant neither precedes nor follows the base: (58) o-mỹ⟨~omỹ⟩i 3-stir⟨~red⟩ ‘He/she moved and moved.’ (event-internal and event-external plurality possible)

forms and functions of reduplication in tupian languages

293

4.2.2 Paraguayan Guarani Reduplication in modern Paraguayan Guarani10 exists in all word classes. Repetition of the word may have an intensifying meaning, just as in other languages. In monosyllabic (example (63)) and disyllabic roots (examples (60–62) and (64–65)) the entire root is repeated, in roots with more syllables (examples (59, 66)) only the last two syllables: (59) i-pire apenõ~penõ11 3-skin cracked~red ‘his very cracked skin’ (60) o-mokõ gua'u~gua'u la pohã 3-swallow simulated~simulated art medicine ‘He ceremoniously feigns swallowing the medicine.’ With verbal and nominal predicates, reduplication serves to express eventinternal plurality, as in: (61) o-joso~joso avati 3-crush~crush corn ‘She is grinding corn.’ (62) o-jara~jara y pe o-ytá-va 3-get.out~get.out water dem 3-swim-nom ‘That swimmer is pushing water aside.’

10

11

Examples are from Graciela Chamorro (p.c.), who is a native speaker from Concepción/Paraguay. She is a specialist in indigenous theology and now has the chair of Anthropology at the University of Dourados, Mato Grosso do Sul, Brazil. She has worked a lot with Caiová and Nhandeva Indians of that area. There is no information about reduplication in the existing grammars of Paraguayan Guarani (see, for example, Canese and Acosta Alcaraz 2001). Guarani is the second official language of Paraguay since 1992 and of the Argentine province of Corrientes since 2004. Both have an established orthography where ⟨y⟩ represents /ɨ/, ⟨ku⟩ /kw/, ⟨gu⟩ /gw/, and ⟨v⟩ /β/, for instance. Loans from Spanish reflect much of their Spanish pronunciation and spelling (here katelláno [kate'ʎano] from Spanish castellano, see (64)).

294

dietrich

(63) o-guy~guy pe ára 3-come.down~come.down dem day ‘The weather has become ugly today.’ Or it refers to event-external plurality, as in (64) o-mbo-jo-para~para guarani katelláno-ndie 3-caus-rec-mix~mix Guarani Spanish-with ‘He is always mixing up Guarani with Spanish.’ (65) o-guata~guatá-ma i-memby 3-walk~walk-already 3-child ‘Her baby just begins walking.’ (66) a-hovasa~vasa o-ho-mboyve 1sg-bless~red 3-go-before ‘I used to bless him before he left.’ In other contexts, the last example may also be understood as an event-internal plurality: ‘Before he left, I blessed him hastily (by several movements of making the sign of the cross).’ 4.2.3 Chiriguano Modern Chiriguano, or Chaco Guarani, is a Guarani dialect that has developed from a Paraguayan Guarani dialect by migrations in the 15th and 16th centuries. Hence, it is small wonder that Chaco Guarani shows the same kind of reduplication processes as Paraguayan Guarani (Schuchard 1979, 58–59). As already stated by Giannecchini (1896, 81–82, see above, section 3), reduplication is made by the repetition of the last two syllables of the base. Suffixes expressing tense or aspectual auxiliaries follow the copy. In non-predicative expressions, especially with numbers, reduplication expresses distributive numbers, such as mopëti~pëti12 ‘one by one,’pajandepo~ndepo ‘ten by ten.’ In predicates, reduplication means event-internal plurality; particularly, it has an intensive meaning (see example (70) and Rose 2005, 356). In (67b), (68) and (70) there are reduplicated verbs, whereas in (69) we have an example of a reduplicated nominal predicate: 12

In the standard orthography of modern Chaco Guarani vocalic nasality is represented by two points (dieresis) over the vowels (ä, ë, ï, ö, ü, ÿ). My spelling and segmentation in some cases differs from Schuchard.

forms and functions of reduplication in tupian languages

295

(67) a. o-mbopu öke-pe 3-knock door-loc ‘He knocked at the door.’ b. o-mbopu~mbopu jepi öke-pe 3-knock~knock always door-loc ‘He was always knocking at the door.’ (68) o-tyky~tyky ö-i 3-drop~drop 3-be ‘It was dropping and dropping.’ (69) ara h-yapu~apu sky 3-thunder~red ‘It was thundering and thundering.’ (70) oi-nüpa~nüpá-ta 3-beat~beat-fut ‘He will beat him strongly.’ 4.2.4 Siriono and Yuki Reduplication in Siriono, a TG language of northern Bolivia, is mentioned specifically by Schermair (1949, 277–278). According to the author, the repetition of the last two syllables of the respective verb or of the monosyllabic verb itself conveys the meaning of intensive or frequentative verbal action, depending on the context, such as in the following examples: (71) a-tjere~tjere 1sg-turn~turn ‘I am often/continuously turning about.’ (72) a-mugata~gata 1sg-open~red ‘I am opening it, trying several times, gradually.’ (73) kisa rakwa~rakwa te tĩ se-we hammock rough~rough int to.be 1sg-dat ‘The hammock is very uncomfortable for me.’

296

dietrich

(74) e-rasi~rasi te 3-ill~ill int ‘He/she is sickly/ falls ill at every moment,’ or ‘He/she is dangerously ill.’ (75) t-a-tea~tea rãã itt-1sg-see~see imp ‘I won’t miss seeing him!’ (event-internal plurality including various possible efforts) Information about reduplication in Yuki, also a TG language of northern Bolivia, may be found in Villafañe (2004, 111–112). According to the author, reduplication means duplication of the last syllable of the verbal root and devoicing of the initial consonant of the reduplicant, as in (76): (76) ∅-jibõ~põ 3sg-hunt~red ‘He is hunting tirelessly.’ (event-internal plurality) (77) ∅-sikjo~kjo 3sg-break~red ‘He broke it once more.’ (event-external plurality) 4.2.5 Parakanã Reduplication in Parakanã (see Silva 1999, 63–64) consists of the anticipation of the monosyllabic or disyllabic root. The rule of monomoraic syllables in the copy (see section 4.1.1) applies. Though the author translates example (78) by an expression that suggests successive and therefore event-internal plurality, the existing data do not allow for establishing a regular distinction between monosyllabic and disyllabic reduplication with two different functions. (78) o-ké~ké 3-enter~enter ‘They went in one after another.’ (event-internal plurality) (79) o-moko~mokón moʔáŋ-a 3-red~swallow medicine-cas ‘He swallowed the medicine several times.’ (event-external plurality)

forms and functions of reduplication in tupian languages

297

(80) a-nopó~nopó 1sg-beat~beat ‘I beat it repeatedly.’ (event-external plurality) 4.2.6 Tapirapé Reduplication in Tapirapé is attested by Almeida et al. (1983, 47–48) and studied in more detail by Praça (2007, 52–54). It seems that the semantic distinction between event-internal and event-external plurality depends on the context. Moreover, as is stated by both authors, reduplication does not only express intensification, iterative or distributive aspect, but also quantification, i.e. plurality of referents (subjects or objects), as in (81). Examples (81) and (82) are from Almeida et al. (1983, 48, my spelling and analysis): (81) a-pyy~pyyk 3-red~beat ‘Many of them were beating him.’ (82) a-pãpỹ~pãpyŋ 3-red~tremble ‘He was trembling very much.’ Example (82) shows an interesting nasalization in the anticipatory reduplicant form. Whereas pãpyŋ seems to have a non-nasal vowel in its last syllable, the preceding copy lacks the nasal coda consonant, but has a nasalized vowel. The following examples (83)–(87) are taken from Praça (2007, 53–54). The first one (83) expresses quantification of referents by reduplication of a nominal root. The following examples (84) and (85) refer to successive and repeated events. The nominal predicate in (86) expresses intensification. Some of the reduplicated forms, (83) and (87), are arguments (subjects), others, (84)–(86), are verbal and nominal predicates. On the basis of examples (81)–(82) I assume that the reduplicants generally precede the base. (83) ' ywyrã-r-e i-xe'eg-i wyrã~wyrã-∅ tree-rel-loc 3-speak-ind bird~bird-cas ‘The birds are talking in the tree.’ (84) xe-∅-yj-a a-ko~koj 1sg-rel-tooth-cas 3-red~come.out ‘My teeth came out one by one.’

298

dietrich

(85) ie-∅ ã-nopỹ~nopỹ i-reka-wo maj-a I-cas 1sg-beat~beat 3-be-ger snake-cas ‘I hit the snake several times.’ (86) Marãxe'i i-kywe~kywer-i Marãxe’i-cas 3-red~be.lean-dim ‘Marãxe’i is so very lean.’ In the case of the first reduplication in example (87), however, the reduplicant is infixed; whereas, an alternative analysis, karoka~roka, would have the reduplicant follow the base: (87) ka⟨roka~⟩roka-mõ-wãr-a pa ke mĩ a-pa~par rõ'õ ⟨red~⟩afternoon-loc-nom-cas ev dub hab 3-red~leave ev ‘It seems that the afternoon group is leaving.’ 4.2.7 Guajajara and Ka’apor The limited data of Guajajara (see Bendor-Samuel 1972, 101) show the type of reduplication we have discussed already. The final consonant is dropped in the anticipating copy: (88) u-pɨhɨ~pɨhɨk 3-red~grasp ‘He grasped it repeatedly.’ Reduplication in Ka’apor is attested and briefly described in Kakumasu (1986, 389). Three examples of the expression of “iterative aspect” were found: (89) a'e kairariʃã ∅-kutu~kutuk i-ndo ky ' y there Kairarixã 3-red~pierce 3-send now cmpl ‘Right there Kairarixã poked it repeatedly.’ (90) ita je ∅-kyty~kytyk ' y rock ev 3- red~rub cmpl ‘It is said that the rocks rubbed and rubbed.’ (91) pe riki h-ukwen ∅-tuka~tuka tã tã tã tã aja riki and ev 3-door 3-hit~hit bang bang bang bang thus ev ‘And so he knocked and knocked on the door. Bang, bang, bang, bang, thus (he knocked).’

forms and functions of reduplication in tupian languages

299

4.2.8 Juruna Reduplication phenomena in Juruna, one of the two existing languages of the Juruna family, are studied by Fargetti (2007, 173–183) and Lima (2008, 58–64).13 As a consequence of its inherent emphasizing effect, reduplication in Juruna has two functions depending on context: one of them is plural (in subjects or objects, see (92b) and (93)), the other is ‘iterative action’ (in (94) and (95b)). We find the order base followed by a partial copy in (92) through (94) and base followed by a full copy in (95). One of the characteristics of Juruna is that independent personal pronouns, which are used as subjects, have allomorphs not only when they are used as clitics (unstressed object pronouns), but also when they are determined by locative suffixes (see ulu-bé in (93)): (92) a. una bɨd󰀟́-tu I fall-vb ‘I fell.’ b. ulu'udí bɨd󰀟́~dɨ~tu we.excl fall~red~vb ‘We fell.’ (93) anɨ ulu-djida~ida-ku ulu-bé He 13-beat~red-vb 13-loc ‘He beat us.’ (94) anɨ ulu-djida~ida-k-a~daka ulu-bé He 13-beat~red-vb-past~red 13-loc ‘He beat us repeatedly.’ (event-external plurality) Example (94) evidences once more (see comments following example (50)) the ludic character of reduplication in Tupi languages. The double reduplication expressing the repeated actions of beating is not fully analyzable in terms of morphemes. The first reduplication is formed by a base and a reduplicant, followed by two suffixes, -k- ‘verbalizer’ and -a ‘past tense.’ The second reduplicant,

13

Lima (2008) does not give context-embedded examples, but only a list of reduplicated forms without a specific meaning, such as imahua ‘to lose’ vs. imahuahua, probably ‘to lose several times,’ izaku ‘to see’ vs izakazaka, probably ‘to see several times.’ Therefore I give examples from Fargetti (2007).

300

dietrich

which intensifies the frequentative meaning of the whole verbal phrase, consists of the last syllable of the preceding reduplicant and the syllable formed by the two suffixes. (95) a. anɨ hí apɨ i-zak-ú he ev jaguar 3-see~rea ‘It is said that he saw the jaguar.’ b. esí du apɨ i- zák-á~zaka14 you.pl always jaguar 3-see-unct~red ‘You always will see jaguars.’ The following example shows how the speakers of Juruna take advantage of reduplication devices, distinguishing event-internal from event-external plurality by means of one or two combined monosyllabic reduplications. This does not mean that there is a well-established morphological distinction in the language, but rather that the speakers make ad hoc use of the iconic efficacy of reduplication processes. Independently from their monosyllabic or disyllabic form, the meaning they wish to give to the utterance results from the possible expressive distinctions (monosyllabic versus disyllabic reduplication or two monosyllabic reduplications joined together): (96) a. apɨ ul-atʃu ulu-bé dog 13-bite 13-loc ‘The dog bit us.’ b. apɨ ul-atʃu~ʧu ulu-bé dog 13-bite~red 13-loc ‘The dog bit us strongly’ (event-internal plurality). c. apɨ ul-aʧu~ʧu~ʧu ulu-bé dog 13-bite~red~red 13-loc ‘The dog bit us repeatedly, many times’ (event-external plurality).

14

Fargetti (2007, 180) does not give this segmentation of the verbal forms (-ú for certainty mood, -á for uncertainty mood (see Fargetti 2007, 159–161 for what she calls “modo realis” and “modo irrealis”), nor does she reflect the semantic difference in her translation of this example.

forms and functions of reduplication in tupian languages

301

4.2.9 Tupari An interesting description of reduplication in word formation processes of Tupari is available in Rodrigues and Caspar (1957, sec. 3.3.3.2.5) and repeated in Alves (2004, 83). Roots of one or two syllables are repeated completely, whereas roots of three syllables only allow the reduplication of the last two syllables. Transitive verbs in Tupari are formed from nominal, verbal, or “attributive” roots by reduplicating them and adding the derivational verbalizing suffix -ka. Rodrigues does not describe grammatical reduplication but word formation processes by reduplication. The suffix -ka derives verbs from nouns that denote objects and qualities, for example: apiri ‘medicine’ → apiri-ka ‘to give medicine,’ no ‘other’ → no-ka ‘to change,’ etc. We cannot give a detailed description of Tupari here but will briefly mention other formations with the verbalizer -ka, which show reduplication of the basic root. It is not clear whether this has an intensifying or frequentative meaning, since it is lexicalized. From ey ‘blood’ we find ey~ey-ka ‘to stain with blood,’ but the simple *ey-ka does not exist (see Alves 2004, 170–171). The same is true for ym ‘dirty’ → ym~ym-ka ‘to soil, dirty’.15 On the other hand, we find õpo and õpo~õpo-ka with the same meaning ‘to beat, slay’ (see Alves 2004, 229; Rodrigues and Caspar 1957, section 3.3.3.2.5). The few examples given by Alves (2004, 83) originate from Rodrigues and Caspar (1957, section 3.3.3.2.5). Ideophonic formations are pak~pak-ka ‘to strike,’ apaŋ~apaŋ-ka ‘to whip, ˜ ∅-ka ˜ ˜ being a nasal rounded mid front vowel). strike,’ ∅‘to groan’ (/∅/ 4.2.10 Makurap Reduplication in Makurap (Tuparí family) is described by Braga (2005, 55–58). It expresses iterative aspect (event-internal and event external plurality). There are no restrictions on consonant clusters (see (97), (99), and (100)), but there are phenomena of weakening of final consonants followed by a suffix with vocalic onset (top~tow-a instead of *top~top-a, see (98)). (97) a. ameŋko pok-ŋ-a on jaguar kill-ev-impf I ‘I killed the jaguar.’

15

In Rodrigues’ spelling (1957, sec. 3.3.3.2.5) it is öm ‘dirty’ and eö ‘blood.’ As in other cases, ⟨y⟩ represents /ɨ/ and ⟨ỹ⟩ nasal /󰀰͠/.

302

dietrich

b. ameŋko pok~pok-ŋ-a on jaguar kill~kill-ev-impf I ‘I killed the jaguars (one after another).’ (98) Adalberto televisão top~tow-a Adalberto television look.at~red-impf ‘Adalberto watches a lot of TV.’ (99) tʃitʃip ɲã put~put girl pl tall~tall ‘The girls are tall.’ (100) warire-i weande tʃitʃip pin~pin-ŋ-a cat-dim much girl scratch~scratch-ev-impf ‘The cats scratched the girl.’16 4.2.11 Munduruku Reduplication in Munduruku is described by Picanço in terms of phonological processes (Picanço 2005, 374–385) and by Gomes in terms of derivational morphology (Gomes 2006, 55–61; 2007). I will first give examples from Gomes because he has more context-embedded ones. According to Gomes (2006, 55; 2007, 391), there are three different functions expressed by verbal reduplication in Munduruku: 1) duration (progressive and iterative), 2) intensification, and 3) pluralization. These shades of meaning, however, depend on contextual configurations. In (101) we can observe the phonological adaptation of the Munduruku imperfective suffix -m to a preceding reduplicant. In (101) the imperfective suffix appears as a velar nasal [ŋ]; whereas, in (102) the basic allomorph -m remains unchanged when adjacent to a preceding underlying [p]. All the reduplicants are monosyllabic. (101) aʤok~ʤo-ŋ tu ẽn bathe~red-impf q you ‘Are you taking a bath?’ (event-internal plurality; possible progressive meaning)

16

Braga’s French translations do not take into account the reduplications. I suggest translations that reproduce the iterative function of (99) and (100), such as “The girls are tall without exception” (event-internal-plurality) in (99). In (100) it could be the plural (“scratched by cats”) or iterative “The cats scratched the girl various times.”

forms and functions of reduplication in tupian languages

303

(102) kobe d-uʤu-kap~ka-m õn katʃoero-daŋ boat rel-ccom-pass~red-impf I waterfall-at ‘I was passing with the boat by the waterfall.’ (event-internal plurality) (103) kujaʤe õn ʧu-m ʤe-koj~koj ma Tapereba-be tomorrow I go-impf refl-row~row myself Tapereba-toward ‘Tomorrow I’ll be rowing to Tapereba.’ (event-internal plurality) (104) wujʤujũ mu-ajpan ŋu ʧe-ku wamõat tajʧi 'e~'e-m people caus-grow.up neg that.one-? shaman wife say~say-impf ‘“He will not bring up children, this one,” the wife of the shaman used to say.’ (105) aĩhĩ o-mu-kũʤo~ʤo-m osũnũj mẽn-ku mother 1sg-obj-listen~red-impf aux thus-? ‘My mother was teaching me this way all the time.’ (event-external plurality) (106) a. axima i-ku fish 3-delicious ‘Fish is delicious.’ b. axima i-ku~ku fish 3-delicious~delicious ‘Fish is very delicious.’ (event-internal plurality, intensity) (107) kopo-ju o-ʤo-wuj~wuj glass-pl 1sg-3obj-wash~wash ‘I washed several glasses.’ (event-internal plurality) In the following examples the roots presumably represent inherent reduplication. In (108) the coda consonant of the reduplicant is dropped before the following suffix: (108) Diego bibodo o-mu-ʧokʧok~ʧoŋ Diego song 1sg-caus-be.happy~red ‘I became very enthusiastic about Diego’s song.’

304

dietrich

(109) ʤe-wajwaj~waj ip o'e refl-laugh~red they aux ‘They laughed very much.’ In (110) the root oʧõ ‘cough’ probably is the result of onomatopoeia. The reduplicated form oʧõ~oʧõ means plurality in the sense of a descriptive term for ‘influenza’: (110) oʧõ~oʧõ o'-j-aoka-ka bekit~kit cough~cough 3–3obj-kill~red child~red ‘Influenza killed several children.’ (event-external plurality, distributive) Reduplication in Munduruku has various forms: monosyllabic reduplication (dakat~kat ‘to cut several times’ or ‘keep (on) cutting’; nomuwã~wã ‘to shout several times’ or ‘keep on shouting’) does not have a meaning that is different from disyllabic reduplication (ʤeorok~orok ‘to hunt all the time’ or ‘go on hunting’) or trisyllabic reduplication (ʤepidowat~pidowat ‘to breathe, gasp all the time’ or ‘keep on breathing, gasping a long time’). Picanço (2005, 382–385) mentions an additional type of reduplication. There are two kinds of reduplication with “fixed segmentism” (i.e. complex reduplication). The first one occurs with the fixed vowel /ǝ́ / and triggers a semantic weakening of the base (with pak ‘red’: i-pak~pə́k ‘It is not so red’; with boŋ ‘big’: yo-boŋ~bə́ŋ ‘it is not so big’). The second one, “reduplication with fixed vowel /e/,” however, has an intensifying meaning which may be interpreted as ‘event-internal plurality’: yo-boŋ~beŋ ‘it becomes bigger,’ ‘it has the property of expanding.’ (111) w-e-wiap~ep 1sg-poss-fan~red ‘I have a fan.’ In example (111) the event-internal plurality, appearing in an existential construction, gives special prominence to the extensive relation between the existence of the fan and the possessor, which focuses on continual possession.17

17

The last two examples of Munduruku resemble reduplication processes in Hungarian, a Finno-Ugric language characterized by vowel harmony (high vs. low and unrounded vs. rounded vowels). In popular and poetic style we find verbs like hebeg-habog ‘to stammer around,’ from hebeg ‘to stutter,’ zireg-zörög ‘to clatter, rattle around,’ from zörög ‘to clatter,

forms and functions of reduplication in tupian languages

305

4.2.12 Nheengatu Nheengatu is the modern form of Brazilian Língua Geral, a language derived from Tupinambá in the XVIIth century, spoken by Portuguese settlers and mestizos along the Brazilian coast in colonial times and spread to the Amazon Basin by the expansion of the Portuguese during the XVIIth century. Today Nheengatu is spoken in the upper Rio Negro region and adjacent Colombian and Venezuelan areas. It is spoken not only by a minority of non-Indians but also by indigenous people who gave up their former Arawakan languages (Baré, Warekena, Baniwa). Reduplication in the Nheengatu dialects spoken by Baré, Warekena, and Baniwa Indians in the upper Rio Negro area is described by Cruz (2011, 308– 316; this volume). In spite of the “derived” character of Nheengatu, reduplication processes in this language retain many of the devices we know from Tupinambá, first of all anticipating reduplicants. The reduplicant generally is disyllabic, but there is only one function, depending on context (iterative, distributive, or intensive). I simplify Cruz’ spelling and glossing somewhat, but maintain her ⟨y⟩ for /j/: (112) ti-a-pudei a-mbeu~beu ind-arã neg-1sg-can 1sg-red~tell you-dat ‘I cannot tell you any more.’ (113) si re-yere~yereu ta-yakau inde if 2sg-red~turn.back 3pl-blame you ‘If you keep on turning back, they will blame you.’

5

Conclusion

Reduplication is a widespread and very common phenomenon in TG and Tupi languages, already documented in the early missionary grammars of Tupinambá and Guarani. The most intriguing phenomenon presented here is the overlapping of morphology and rhythmic elements such as ludic retardation. The formal structure of reduplication is based on the partial or full copying of

rattle,’ or sirren-surran ‘to whisk, slip away repeatedly,’ from surran ‘to whisk, slip away.’ Some of these formations are onomatopoeic. The reduplicant precedes or follows the base. In some cases the onset of the preceding reduplicant is dropped, see izeg-mozog ‘to move to and fro frequently,’ from mozog ‘to move.’

306

dietrich

the root, which expresses intensity. The copy very often is anticipatory and, as it were, it announces the base. Alternatively, the base is echoed by a full or partial copy. The effect is intensity (any kind of plural, verbal or nominal, successive or iterative aspect). The reduplicated form does not always allow a clear morphological analysis. The reduplicant tends to be based on the whole form, not on its morphological components. Its morphophonemic shape may exceed its morphological structure, although, as a whole, the reduplication process is often a kind of derivational morphology. The distinction between ‘event-internal’ and ‘event-external’ plurality is very useful for a better understanding of what is going on in the case of successive actions, which take place on one occasion. It helps to understand the concept of ‘diffusion’ and ‘plurality’ in nominal expressions and to distinguish it from habitual or frequent events on different occasions. However, as far as the Tupi stock is concerned, it is only attested unambiguously in three languages of the Tupi-Guarani family: Asuriní of Tocantins, Anambé and Emerillon. Nevertheless, even in these languages the opposition of ‘successive’ versus ‘repeated actions’ does not always correspond with the formal distinction between monosyllabic versus disyllabic reduplication. These meanings depend more on context than on the differentiation of forms. Table 1, which includes the results given by Rose (2005), shows a characteristic geographical distribution of forms and functions of reduplication in Tupian languages; there is a central range of languages characterized by the order base + reduplicant and the distinction of two functions (event-internal and event-external plurality): Parintintin, Asurini of the Tocantins River, Guajá, and Anambé. Parallel with this northern central range is a southern central range formed by languages which, while showing the order reduplicant + base, only have one function: Kokama, Parakanã, Avá-Canoeiro, Guajajara, Ka’apor, and Tapirapé. Guajá belongs to both ranges; Tupari is not well documented. At the northern and southern borders we have groups of languages which show the order base + reduplicant and one function: Makurap, Juruna, Munduruku, and Wayãpi on the one hand, Araweté, Awetí, Kamaiurá, Siriono, Yuki, Guarayo, Chiriguano, Guaraní, and Mbyá on the other. For results for Aweti, see Drude (this volume); for Kamaiurá, see Seki (2000, 322).

307

forms and functions of reduplication in tupian languages table 1

Schematic overview of reduplication in Tupian languages

RED-base Base-RED Two functions One function Juruna Makurap Munduruku Wayãpi Guajá Nheengatu

+ + + + + +

Anambé Asurini of Tocantins Guajá Parintintin

+ + + + + +

Emerillon Tupinambá

+ +

Avá-Canoeiro Guajajara Ka’apor Kokama Parakanã Tapirapé Tupari

+ + + + + + +

Araweté Awetí Chiriguano Guaraní Guarayo Kamaiurá Mbyá Siriono Yuki

+ + + +

+ + ? ? + ? + + + + + + +

+ + + + + + + + +

+ + + + + + + + +

308

dietrich

Acknowledgments I am extremely grateful to Hein van der Voort, Gale Goodwin Gómez and an anonymous reviewer for their helpful comments on a previous draft of this paper, as well as to William Hoye (University of Münster) for his care to improve its English style. I am, of course, responsible for any remaining mistakes or improper argumentations.

References Almeida, António, Irmãzinhas de Jesus, and Luíz Gouvêa de Paula. 1983. A língua Tapirapé, Rio de Janeiro: Biblioteca Reprográfica Xerox. Alves, Poliana Maria. 2004. O léxico do Tuparí: Proposta de um dicionário bilingüe. PhD diss., Universidade Estadual Paulista, Araraquara. Anchieta, P. Ioseph de. 1595. Arte de grammatica da lingoa mais vsada na costa do Brasil. Coimbra: Antonio de Mariz. Bendor-Samuel, David. 1972. Hierarchical structures in Guajajara. Norman: SIL, University of Oklahoma. Betts, La Vera. 2012. Kagwahiva dictionary. Anápolis: SIL. http://www.sil.org/americas/ brasil/publcns/dictgram/KHDict.pdf (accessed 22 November 2013). Borges, Mônica. Veloso. 2006. Aspectos fonológicos e morfossintáticos da língua AváCanoeiro. PhD diss., UNICAMP. http://www.etnolinguistica.org/tese:borges_2006 (accessed 22 November 2013). Braga, Alzerinda de Oliveira. 2005. Aspects morphosyntaxiques de la langue makurap/tupi. PhD diss., Université de Toulouse-Le Mirail. Cabral, Ana Suelly Arruda Câmara, and Aryon Dall’Igna Rodrigues. 2003. Dicionário da língua Asuriní do Tocantins. Belém: UFPA/IFNOPAP / Brasília: UnB/IL/LALI. Canese, Natalia Krivoshein de, and Feliciano Acosta Alcaraz. 2001. Gramática Guaraní, Asunción: Instituto Superior de Lenguas, Universidad Nacional de Asunción. Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Cruz, Aline da. 2011. Fonologia e gramática do Nheengatú: A língua geral falada pelos povos Baré, Warekena e Baniwa. Utrecht: LOT. Cruz, Aline da. this volume. Reduplication in Nheengatu. Cusic, David D. 1981. Verbal plurality and aspect. PhD diss., Stanford University. Dietrich, Wolf. 2010. Tiempo, aspecto y evidencialidad en guaraní. LIAMES 10: 67– 83. Dooley, Robert A. 2006, Léxico Guarani: Dialeto Mbyá. Introdução, esboço gramatical, léxico. Cuiabá: SIL. (“Introduction and grammatical sketch”; II: “Lexicon”).

forms and functions of reduplication in tupian languages

309

Drude, Sebastian. this volume. Reduplication as a tool for morphological and phonological analysis in Awetí. Fargetti, Cristina Martins. 2007. Estudo fonológico e morfossintático da língua Juruna. [Lincom Studies in Native American Linguistics, 58]. Munich: Lincom Europa. Giannecchini, Doroteo. 1896. Reglas elementares de la lengua chiriguana para uso de los RR. PP. misioneros franciscanos. Lucca: Tipografia Arciv. S. Paolino. Gomes, Dioney Moreira. 2006. Estudo morfológico e sintático da língua Mundurukú (Tupí). PhD diss., Universidade de Brasíla. http://www.etnolinguistica.org/tese :gomes_2006 (accessed 22 November 2013). Gomes, Dioney Moreira. 2007. Reduplicação verbal em Mundurukú. In Línguas e culturas Tupí, ed. Ana Suelly Cabral and Aryon Rodrigues, 391–396. Brasília: LALI/ UnB. Hoeller, Fray Alfredo. 1932. Grammatik der Guarayo-Sprache. Guarayos (Bolivia)/Hall (Tirol, Austria): Missionsprokura der P.P. Franziskaner. Jensen, Cheryl Joyce S. 1989. O desenvolvimento histórico da língua Wayampi. Campinas: Editora da UNICAMP. Julião, Maria Risolêta Silva. 2005. Aspects morphosyntaxiques de l’anambé. PhD diss., Université de Toulouse-Le Mirail. Kakumasu, James. 1986. Urubu-Kaapor. In Handbook of Amazonian languages, ed. Desmond C. Derbyshire and Geoffrey K. Pullum, vol. 1, 326–403. Berlin/New York/ Amsterdam: Mouton de Gruyter. Lima, Suzi. Oliveira de. 2008. A estrutura argumental dos verbos na língua Juruna (Yudja). Da formação dos verbos para a análise das estruturas sintáticas. MA thesis, Universidade de São Paulo. http://www.etnolinguistica.org/tese:lima_2008 (accessed 22 November 2013). Magalhães, Marina Maria Silva. 2007. Aspectos fonológicos e morfossintáticos da língua Guajá. MA thesis, Universidade de Brasília. http://www.etnolinguistica.org/ tese:magalhaes_2007 (accessed 22 November 2013). Montoya, Antonio Ruiz de. 1993. Arte de la lengua Guaraní (1640), edición facsimilar con introducción y notas por Bartomeu Melià, S.I. Asunción: CEPAG. Pease, Helen. 2007. Parintintin grammar. Porto Velho: SIL Brasil. http://www.sil.org/ americas/brasil/publcns/dictgram/PNGram.pdf (accessed 22 November 2013). Picanço, Gessiane Lobato. 2005. Mundurukú: Phonetics, phonology, synchrony, diachrony. PhD diss., University of British Columbia, Vancouver. http://www .etnolinguistica.org/tese:picanco_2005 (accessed 22 November 2013). Praça, W. Neiva. 2007. Morfossintaxe da língua Tapirapé (Família Tupí-Guaraní). PhD diss., Universidade de Brasília. http://www.etnolinguistica.org/tese:praca_2007 (accessed 22 November 2013). Restivo, Paulo. 1724. Linguae Guaraní Grammatica Hispanice a Reverendo Patre Jesuita Paulo Restivo secundum libros Antonii Ruiz de Montoya, Simonis Bandini aliorumque

310

dietrich

adjecto Particularum lexico …, Santa María la Mayor. Re-edition by Christian Friedrich Seybold, Stuttgart: Kohlhammer. 1892. Newly edited by Silvio M. Liuzzi, Arte de la lengua guaraní (1724), at http://www.celia.crns.fr/FichExt/Paleographies/ Guarani/Guarani_intro.htm (accessed 22 November 2013). Rodrigues, Arion Dall’Igna. 1953. Morfologia do verbo tupi, Letras (Curitiba) 1: 121– 152. Rodrigues, Aryon Dall’Igna and Franz Caspar. 1957. Zur Sprache der Tuparí-Indianer (Westbrasilien). I. Einführung (Franz Caspar, 13 p.). II. Die Struktur der Tuparí-Sprache (Aryon D. Rodrigues, 81 p.). ms. Digitized and published online by Willem F.H. Adelaar and Hélène B. Brijnen: http://www.hum.leiden.edu/lucl/research/ natterer-caspar.jsp (accessed January 2012). Rose, Françoise. 2005. Reduplication in Tupi-Guarani languages: Going into opposite directions. In Studies on reduplication, ed. Bernhard Hurch, 351–368. Berlin/New York: Mouton de Gruyter. Schermair, P. Fr. Anselmo. 1949. Gramática de la lengua sirionó. La Paz: A. Gamarra. Schuchard, Bárbara. 1979. Ñande ñë. Gramática guaraní para castellano hablantes. Santa Cruz de la Sierra: APCOB—CEBIAE. Seki, Lucy. 2000. Gramática do Kamaiurá. Língua Tupi-Guarani do Alto Xingu. Campinas: Unicamp/São Paulo: Imprensa Oficial. Silva, Auristéa Caetana Souza e. 1999. Aspectos da referência alternada em Parakanã. MA thesis, Universidade Federal do Pará. http://www.etnolinguistica.org/tese:silva _1999 (accessed 22 November 2013). Solano, Eliete de Jesus Bararuá. 2009. Descrição gramatical da língua Araweté. PhD diss., Universidade de Brasília. http://www.etnolinguistica.org/tese:solano_2009 (accessed 22 November 2013). Vallejos Yopán, Rosa. 2010. A grammar of Kokama-Kokamilla. PhD diss., University of Oregon, Eugene. http://www.etnolinguistica.org/tese:vallejos_2010 (accessed 22 November 2013). Villafañe, Lucrecia. 2004. Gramática Yuki: lengua tupí-guaraní de Bolivia, Tucumán: Ediciones del Rectorado, Universidad Nacional de Tucumán.

Abbreviations 1sg 1pl 2sg 2pl 3

first person singular first person plural second person singular second person plural third person (singular or plural)

13 aux cas

caus

first person exclusive auxiliary case marker (argument case, distinguishes arguments from adverbials) causative

forms and functions of reduplication in tupian languages

ccom cmpl conn dat dim dub ev excl foc fut ger hab imp impf ind ins int irr itt

causative-comitative loc completive mvt connective particle neg dative nom diminutive obj dubitative past evidential pg exclusive (person marker or poss pronoun) q focus rea future tense rec gerund red habitual aspect rei imperative rel imperfective rep indicative tg instrumental top intensifier unct irrealis mood vb intentional mood

locative movement negation nominalizer object past tense Paraguayan Guarani possessive question marker realis mood reciprocal voice reduplication reiterative relational inflection reported speech Tupi-Guarani topic uncertainty mood verbalizer

311

chapter 12

The Interaction of Reduplication with Word Classes and Transitivity in Cavineña Antoine Guillaume

This paper discusses the reduplication processes that are at work in Cavineña, a Tacanan language from Amazonian Bolivia. This language presents a wide range of different reduplication processes which have the following main characteristics: they can be simple or automatic, full or partial; they apply mostly to verbs, nouns and adjectives; they have different degrees of productivity; they are phonologically rather easy to describe. In addition to being pervasive, the morphological processes of reduplication in Cavineña are noteworthy for having several syntactic effects, an aspect of reduplication rarely discussed in general typological works on reduplication; these tend to focus on phonological and semantic characteristics. A particularly interesting syntactic function of reduplication in Cavineña is valence reduction with antipassive effect, which occurs when reduplication applies to transitive verbal roots.

1

Introduction

Cavineña is a language from the small Tacanan family, which also consists of Araona, Ese Ejja, Reyesano and Tacana. It is spoken by 1,000~1,200 people in numerous communities scattered throughout the lowlands of northern Bolivia, a region corresponding to the southwestern border of the Amazon basin. Cavineña displays a wide range of different reduplication processes,1 distinguished according to the following formal parameters: the formal realization of reduplication (simple or automatic,2 full or partial), the word class of the base 1 A few of the patterns discussed in this paper correspond to what is sometimes called ‘inherent reduplication,’ in that they involve forms for which there is no unreduplicated base. Strictly speaking, according to the Graz Reduplication Project’s definition (see http://reduplication .uni-graz.at/), these patterns do not qualify as ‘true reduplication.’ See Epps (this volume) and Goodwin Gómez (this volume) for a discussion of similar patterns in Hup and Yanomae, respectively. 2 “Simple” reduplication means the straightforward reduplication process that only involves

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_013

314

guillaume

and/or the reduplicated word (verb, adjective, noun, quantifier, postposition), and the transitivity and inflectional status of the base and/or the reduplicated word when verbs are involved. As summarized in Table 2 (section 4), these parameters can combine in 13 possible ways. These reduplication patterns were first described in different sections of my grammar of Cavineña (Guillaume 2008), which is based on primary data (fieldnotes and texts) that I collected over 15 months of fieldwork between 1996 and 2003 in the town of Riberalta and in two traditional communities (Galilea and Misión Cavinas), and complemented by secondary data published by SIL missionaries Elisabeth Camp and Millicent Liccardi. The present work, which is based on the same material, has the goal of discussing all the reduplication patterns in one place, in order to gain a better understanding of reduplication as a general process in Cavineña. It is also intended to provide a more in-depth characterization of certain reduplication processes and to draw attention to the syntactic function of reduplication in this language. The paper is organized as follows: Section 2 provides the phonological and morphosyntactic backgrounds necessary to understand the effects of reduplication in Cavineña, with information on word classes, transitivity, predicate structure, and grammatical functions. Section 3 presents an overview of each of the 13 reduplication patterns, listing the main parameters that are used to distinguish one from the other. Section 4 provides a detailed presentation of the most interesting patterns, namely the four applying to verbs. The article ends with a conclusion and a table that summarize the main characteristics of reduplication in Cavineña.

2

Phonological and Morphosyntactic Background

Cavineña has a relatively simple phonological system. Most phonemes do not display any significant allophonic variations. The Cavineña vowel phonemes are i, e (with allophones [e] and [ɛ] in free variation), a and ʊ (written u, with allophones [ʊ] and [o] in free variation). The Cavineña consonant phonemes are p, b, t, d, c (voiceless alveo-palatal stop, written ty), ɟ (voiced alveo-palatal stop, written dy), k, kw, ts (voiceless alveolar affricate), ʨ (voiceless alveopalatal affricate, written ch), s, ɕ (voiceless alveo-palatal fricative, written sh), the repetition of material that is already present in the unreduplicated base. “Automatic” reduplication, a term taken from Rubino (2005), means a reduplication process that is “obligatory in combination with another affix, and which does not add meaning by itself to the overall construction” (Rubino 2005, 18).

reduplication, word class and transitivity in cavineña

315

h (written j), ɺ (alveolar lateral flap, written r), ʎ (alveo-palatal liquid, written ry), m, n, ɲ (written ny), w (with allophones [w] before a and [ß̞ ] before i and e) and j (written y). The syllable structure is (C)V. The phonological word is minimally disyllabic. As a result monosyllabic roots used without an additional affix or clitic must take an epenthetic suffix -u, as in bá-u ‘seeepen’ and bé-u ‘bring-epen.’ As we will see below, this happens in one of the reduplication processes. The phonological word is the domain of application of the accentual system, realized by a non-contrastive pitch accent which functions as follows: (1) the first syllable of a phonological word receives a high pitch, (2) the final two syllables receive a mid pitch (only the last syllable if it is a disyllabic word), and (3) the high pitch of the first syllable extends rightwards to any syllable(s) preceding the last two syllables (e.g., béta ‘two,’ mátuja ‘yacare caiman,’ jútákiju ‘therefore,’ wésá-tána-tsu ‘lift-passss,’ etc.). The word classes of the language are listed in Table 1. They have been identified according to semantic and morphosyntactic criteria (see discussion in Guillaume 2008, chapter 4). table 1

Cavineña word classes (Guillaume 2008, 62)

Type Lexical: open

Class

Verbs Predicative adjectives Nouns Grammatical: closed Verb modifiers Attributive adjectives Number markers Quantifiers Postpositions Pronouns Demonstratives Content question words Particles Subordinate clause markers Coordinators Semi-open Interjections Onomatopoeias

316

guillaume

The morphological structure of verbs is highly elaborate, with a maximum extension of up to 11 structural slots. However, it is strongly agglutinative and very few morphemes display allomorphic variants. By contrast, the other word classes (nouns, predicative adjectives, pronouns and demonstratives) take very few affixes or no affixes at all. Cavineña verbs have strict transitivity values: they are either transitive or intransitive. Apart from a handful of exceptions, there are no ambitransitive verbs. This rigid lexical transitivity is reflected at various levels of morphosyntactic organization. There is consistent ergative case marking on the subject argument of transitive verbs. There are special allomorphs of verbal affixes which agree in transitivity with the verb root or verb stem they attach to (e.g., the completive suffix is -tere on intransitive roots and -tirya on transitive roots). In auxiliary constructions, speakers must make a choice between an intransitive and a transitive auxiliary, depending on the transitivity of the predicate. In fact, the language requires special verbal morphemes (passive, reflexive/reciprocal, antipassive) or constructions in order to change transitivity. Verbs differ according to whether they are inflecting or non-inflecting. Inflecting verbs directly carry the (verbal) morphology and form one-word (simplex) predicates. A simplex predicate minimally consists of a verb root and a TAM (Tense, Aspect, Mood) inflection, as shown in (1a-b). However, such predicates can also be polysynthetic, involving an incorporated (body part) noun and numerous non-inflectional affixes, principally suffixes, between the root and the TAM inflection. These non-inflectional suffixes may express aspect, manner, modality, posture, motion, valence-change, etc. (1)

a. Intransitive clause Jutakiju iba diru-kware. therefore jaguar go-rem.past ‘Therefore the jaguar went away.’ (Tavo Mayo 1977, 28) b. Transitive clause Iba=ra=tu3 iye-chine takure. jaguar=erg=3sg kill-rec.past chicken ‘The jaguar killed the chicken.’ (elicited)

3 Cavineña has two types of enclitics: clause level enclitics, which occur in second position in a clause, e.g. =tu ‘3sg’, and phrase level enclitics, which attach to the last word of a phrase, e.g. =ra ‘ERG’, which precedes =tu in this same example.

reduplication, word class and transitivity in cavineña

317

Non-inflecting verbs cannot carry (verbal) morphology. They require a generic auxiliary (light verb) to form a two-word (complex) predicate. The auxiliary is either ju- ‘be,’ if the non-inflecting verb is intransitive, or a- ‘affect’ if it is transitive, as in (2a-b). (2) a. Intransitive complex predicate Santiago=tu katewa ju-wa. Santiago=3sg hide be-perf ‘Santiago hid.’ (elicited) b. Transitive complex predicate Epuna=ra endya a-kware [peya ekwita]. woman=erg say.yes.to affect-rem.past other person ‘The woman said yes to another man.’ (elicited) Cavineña has two classes of adjectives: predicative adjectives, which function as a copula complement in copula clauses, as in (3), and attributive adjectives, which function as a modifier to the head of a NP. (3)

Ari-da ju-kware aja. big-asf be-rem.past capuchin.monkey ‘The capuchin monkey was big.’4

Predicative adjectives are further subdivided according to morphological criteria. There is a subclass of bound forms, called da-adjectives, which must take an affix, as in (3). There is also a subclass of independent forms which do not require an affix. Predicative adjectives are syntactically similar to intransitive non-inflecting verbs. The copula used with them is ju- ‘be,’ which is also used as a generic auxiliary for non-inflecting verbs (discussed above). Despite this similarity, predicative adjectives and intransitive verbs do differ in a number of ways, which suggests that they are better analyzed as different word classes (see Guillaume 2008, 161–162). Nouns represent the third open lexical class. The noun class consists of three subclasses which differ according to semantic and morphological properties. There is a subclass of bound nouns that refer to parts of a whole (of the body or other entities) and that obligatorily take a dummy prefix e- (e.g., e-biti

4 When no indication of the source of an example is provided, the example comes from my own textual, conversational, or participant observation corpus.

318

guillaume

‘npf-skin’), a subclass of bound nouns which refer to kinship relations and that are obligatorily marked for their ‘possessor’ (e.g., e-tata=ke [3-father=3] ‘his father’), and a class of grammatically independent nouns that refer to any other nominal concept (e.g., iba ‘jaguar’). Cavineña encodes the grammatical functions of the arguments (S, A, O)5 by way of a case-marking system combined with a system of personal clitics in the second position in the clause. As we can see in (1a,b) and (2a,b), both systems operate according to an ergative pattern: S and O arguments are unmarked, while A receives a special ergative case enclitic (or suffix in the case of person markers). The core arguments are not obligatorily expressed. When they are expressed, their position in the clause is grammatically free and does not participate in the disambiguation of the grammatical functions.

3

Reduplication in Cavineña: An Overview

This section contains an overview of the 13 reduplication processes found in Cavineña. This section is subdivided into three subsections based on the distinction between full (3.1.) and partial (3.2.) reduplication, and between simple (3.1, 3.2) and automatic (3.3) reduplication (see footnote 2). The reduplication processes that involve verbs, which are the most productive, are discussed only briefly in this section but are covered in full detail in section 4. 3.1 Simple Full Reduplication Depending on the word classes involved and the resulting semantic effects, simple full reduplication has been sorted into nine different types. There are three full reduplication processes that involve verbs as the base. The first two produce a derived verb as the output while the third derives a noun. The two processes that derive verbs differ in the way they affect the transitivity or the inflectional properties of the base. The first process (numbered I in Table 2) only applies to transitive verbs and derives intransitive verbs with an antipassive sense, denoting a culturally-codified activity.

5 S = subject of an intransitive verb, A = subject of a transitive verb, O = object of a transitive verb.

reduplication, word class and transitivity in cavineña

(4)

Verb (transitive)

Verb (intransitive)

arabajubaeriijiitushasusutakatyautsauwa-

ara~arabaju~bajuba-u6~ba-ueri~eriiji~ijiitusha~itushasusu~susutaka~takatya-u6~tya-uutsa~utsauwa~uwaetc.

‘eat O’ ‘toast O’ ‘see O’ ‘grind O’ ‘drink O’ ‘push O’ ‘suck O’ ‘peel O’ ‘give O’ ‘wash O’ ‘plant O’

319

‘eat, have a meal’ ‘toast’ ‘see’ ‘grind’ ‘drink, have a drink’ ‘push’ ‘suck’ (e.g., a baby) ‘peel’ ‘give’ ‘wash, do the laundry’ ‘plant’

The second process (II) applies both to intransitive and transitive verbs. It derives a new verb which retains its original valency but becomes noninflecting. The reduplicated verb can only be used in a complex predicate with a following generic auxiliary for carrying the verbal affixes. The meaning conveyed by this process is the multiple repetition of the event expressed by the verb. (5)

Verb (inflecting)

Verb (non-inflecting)

jekekekuetikwinanatsajajaikwayaiwaramiritsaje-

je-u6~je-u ( ju-) keke~keke ( ju-) kueti~kueti ( ju-) kwinana~kwinana ( ju-) tsajaja~tsajaja ( ju-) ikwaya~ikwaya (a-) iwara~iwara (a-) miri~miri (a-) tsaje~tsaje (a-) etc.

‘come’ ‘shout’ ‘pass’ ‘emerge’ ‘run’ ‘miss O’ ‘call O’ ‘strike O’ ‘cut O’

‘come repeatedly’ ‘shout repeatedly’ ‘pass repeatedly’ ‘emerge repeatedly’ ‘run repeatedly’ ‘miss O repeatedly’ ‘call O repeatedly’ ‘strike O repeatedly’ ‘cut O repeatedly’

6 u is an epenthetic vowel used in full reduplication when the reduplicant is monosyllabic (see section 2).

320

guillaume

These two processes have a high degree of productivity. They will be discussed at greater length in section 4. The third full reduplication process (III) that applies to a verb base derives an independent noun as the output. This mechanism is not productive, accounting for only six words in the available corpus. The resulting meanings are not fully predictable, although the nouns all refer to entities which involve in some way the multiple repetition of the action of the verb they are derived from. This reduplication pattern will not be further discussed in this paper. (6)

Verbs (inflecting)

Noun (independent)

jerutsabusa kujapuruwene-

jeru~jeru tsa~tsa busa~busa kuja~kuja puru~puru wene~wene

‘sing’ ‘laugh’ ‘open O’ ‘blow air into O’ ‘dig O’ ‘draw/write O’

‘song’ ‘cacaré bird’ ‘glow-worm’ ‘balloon’ ‘ibis-like wading bird’ ‘letter’

There is an additional full reduplication process (IV) that applies to the subclass of bound predicative adjectives. Recall that predicative adjectives are used as a complement in copula clauses (not as a modifier in a NP) and that there is a bound subclass and an independent subclass of predicative adjectives. The reduplicated form remains a predicative adjective,7 but it becomes independent (it does not require an affix). The meaning conveyed by this process is the spatial or temporal distribution of the property expressed by the adjectival base. In some cases, the meaning is attenuative.8 (7)

Predicative adjective (bound)

Predicative adjective (independent)

apuariiwijawa-

apu~apu ari~ari iwi~iwi jawa~jawa

‘dark’ ‘big’ ‘smelly’ ‘yellow’

‘dark in different places’ ‘bigger and bigger’ ‘a bit smelly’ ‘yellowish’

7 There is one exception, involving the bound predicative adjective uyu- ‘muddy,’ whose full reduplication yields an (independent) noun, uyu~uyu ‘mud,’ and not an independent predicative adjective. 8 Note that no examples are available of this reduplication process applied to mono- and trisyllabic roots.

reduplication, word class and transitivity in cavineña

Predicative adjective (bound)

Predicative adjective (independent)

jujikasanakapudesawaseweweka-

juji~juji kasa~kasa naka~naka pude~pude sawa~sawa sewe~sewe weka~weka etc.

‘fat’ ‘strong’ ‘wet’ ‘red’ ‘green/blue’ ‘black’ ‘bright’

321

‘chubby’ ‘a bit strong’ ‘a bit wet, with little wet spots’ ‘reddish’ ‘greenish/bluish’ ‘blackish’ ‘bright in different places’

This process, although also highly productive, will not be further discussed in this paper.9 The next two full reduplication patterns apply to nouns. Recall that there are three subclasses of nouns: bound e-nouns that express part-whole relations, bound nouns that express kinship relations and independent nouns. The first full reduplication process (V) can apply to both bound e-nouns and independent nouns, turning them into independent predicative adjectives that express the notion of ‘having many X’ (X being the referent of the noun expressed by the nominal base).10 (8)

Noun -kwija -nawa -tiri -tsaru -wachi benu buje chiwe

9 10 11

Predicative adjective (independent) ‘thorn’ ‘down’ ‘root’ ‘hair’ ‘foot’ ‘bend’ ‘stain’ ‘nigua’11

kwija~kwija nawa~nawa tiri~tiri tsaru~tsaru wachi~wachi benu~benu buje~buje chiwe~chiwe

‘having many thorns’ ‘having a lot of down’ ‘having many roots’ ‘having a lot of hair’ ‘having many feet’ ‘having many bends’ ‘having many stains’ ‘having many niguas’

For more details, see Guillaume (2008, chapter 11, in particular section 11.2.2). Note that no examples are available of this reduplication process applied to mono- and trisyllabic roots. Type of parasitic flea (‘sand flea,’ ‘chigoe’ or ‘jigger,’ Tunga penetrans). The female burrows into the exposed skin of people’s and animals’ feet, and remains there while eggs develop, sometimes causing intense irritation.

322

guillaume

Noun

Predicative adjective (independent)

jibi ‘wrinkle’ kani ‘hole’ kwesa ‘facial hair’

jibi~jibi ‘having many wrinkles’ kani~kani ‘having many holes’ kwesa~kwesa ‘having a lot of facial hair’ etc.

This mechanism is very productive. Like the previous reduplication process, it will not be further discussed in this paper.12 The second full reduplication pattern (VI) that applies to nouns does not yield adjectives but nouns. This process is not productive. In the available corpus, it only applies to three nouns, which belong to the independent subclass, turning them into new independent nouns. The meaning expressed by this reduplication process is slightly different from one noun to the other. It seems to express ‘something made out of many X’ in the case of ‘alphabet’ (and perhaps also in the case of ‘picture, photo’) and ‘something of less importance than X’ in the case of ‘idol’ (X being the referent of the noun expressed by the unreduplicated nominal base). (9)

Noun (independent)

Noun (independent)

kweya ‘form, spirit’ nusa ‘small mark in a repetitive pattern’ Yusu ‘God (< Spanish Dios)’

kweya~kweya ‘picture, photo’ nusa~nusa ‘alphabet’ yusu~yusu

‘idol’

The seventh full reduplication pattern (VII) is not productive synchronically either. It belongs to the domain of onomatopoeia and ideophones. The reduplicated forms are nouns which most often refer to animals or entities with salient sonorous or rhythmic characteristics. This pattern corresponds to inherent reduplication in that no unreduplicated base can be identified. Synchronically, according to the Graz Reduplication Project’s definition, this pattern is not ‘true’ reduplication, for the reason that it does not involve “a set of at least two linguistic forms F and F’ in a paradigmatic morphological relation.” For this reason, the reduplicated forms listed below are not segmented. Although not productive, the pattern is attested in a fair number of items in the available corpus (about 30) and the items show semantic regularities (onomatopoeia and

12

For more details, see Guillaume (2008, chapter 11, in particular section 11.2.2).

reduplication, word class and transitivity in cavineña

323

ideophones). These facts suggest that this reduplication process might have been productive at an earlier stage. Some of these words might have originally been derived from verbs, through reduplication process III, or from nouns, through the reduplication process VI. (10) No base Noun (independent) – – – – – – – – – – – – – – – – – – – – –

biribiri chichi dudu dukwadukwa emeyaemeya jaajaa jabirijabiri kaekae kawakawa kumukumu kuyukuyu pupu sheishei shiwishiwi shupushupu tasatasa wanuwanu witukuwituku ribariba tiritiri warawara

‘parakeet sp.’ ‘meat’ ‘beam’ ‘titi monkey’ ‘drawing, photo’ ‘guan sp. (turkey-like bird)’ ‘kite’ ‘macaw’ ‘yellow-billed tern’ ‘drum’ ‘granulated catfish’ ‘owl’ ‘nocturnal cicada’ ‘small duck sp.’ ‘serepapa grande cichlid (fish)’ ‘black ant sp.’ ‘bee sp.’ ‘small bird sp.’ ‘toasted corn flour’ ‘type of traditional dance’ ‘forest coca’

The last two full reduplication processes (VIII and IX) each involve a closed grammatical class: quantifiers and postpositions. In both cases, only one member of the class is attested in a reduplicated form. (11) Quantifier peadya ‘one, a’

Quantifier peadya~peadya ‘some’

324

guillaume

(12) Postposition

Postposition

patya

patya~patya

‘in the middle of’

‘in the middle of’ (apparently without change of meaning)

3.2 Simple Final -CV Reduplication We will now turn to partial reduplication processes. There are two distinct patterns of simple13 final -CV reduplication in Cavineña, both of which are synchronically non-productive. In most instances, no base exists independent of the reduplicated form, in which case we are dealing with inherent (rather than ‘pure’) reduplication. These two processes are included in this paper for three reasons: (1) they are attested in an important number of items; (2) the reduplicated words show semantic regularities, and (3) in at least some cases it is possible to find an independent unreduplicated base. The first final -CV reduplication process (X) is attested in both intransitive and transitive verbs (13). In three instances of intransitive verbs, an independent unreduplicated base can be identified, as a bound predicative adjective (13a). In these forms, final -CV reduplication, therefore, has the function of verbalizing a predicative adjective. The resulting meaning is inchoative. In the rest of the (intransitive and transitive) verbs, no independent unreduplicated base can be identified (13b). The semantics of these verbs are very often onomatopoeic or ideophonic in character, expressing an action or entity with sonorous or rhythmic characteristics. (13) a. Predicative adjective (bound) apujawawekab. No base – –

13 14

‘dark’ ‘yellow’ ‘bright’14

Verb (inflecting) apu~pu- ‘darken’ jawa~wa- ‘ripen’ weka~ka- ‘be at dawn’ Verb (inflecting) bukukupakaka-

See footnote 2. Alternatively, the base might as well be the noun weka ‘light.’

‘move’ ‘fall’

reduplication, word class and transitivity in cavineña

No base

Verb (inflecting)

–15 – – – – – – – – – –

katsatsapetutusukururutadadatararatsajajawarerebajejeitatamakakapiruruetc.

325

‘drizzle’ ‘boil’ ‘drip’ ‘shiver’ ‘snore’ ‘run’ ‘turn’ ‘prepare O’ ‘shake O’ ‘hug O’ ‘roll O up’

The second final -CV reduplication process (XI) is attested in nouns (14). In only one case can an independent unreduplicated based be identified, as a transitive inflecting verb (14a). In this case, reduplication, therefore, has the function of turning a verb into a noun. This derived noun refers to an entity that involves the multiple repetition of the action of the verb it is derived from, a function which recalls what happens with full reduplication process (III). In all the other nouns, no independent unreduplicated base can be identified (14a). The semantics of these forms are very often onomatopoeic or ideophonic in character. (14) a. Transitive verb watsub. No base – – – – –

15

‘twist O’

Noun (independent) watsu~tsu ‘whirlpool’ Noun (independent) bijiji bururu bututu japipi mejiji

‘gray-fronted dove’ ‘toad’ ‘madidi ant’ ‘butterfly’ ‘beach’

A possible, although very speculative, unreduplicated base for katsatsa- might be the transitive inflecting verb katsa- ‘beat O, whip O.’

326

guillaume

No base

Noun (independent)

– – – – – – – – –

muwawa pabejerere pisusu sududu sururu tabubu tiriri tyakariri wiriri etc.

‘grass’ ‘pan for toasting’ ‘iguana’ ‘capybara (rodent)’ ‘waterfall’ ‘hurricane’ ‘spider’ ‘gecko’ ‘small partridge sp.’

Note that in all the available examples, the reduplicated syllable begins with a consonant; no examples are available of a final syllabic -V reduplication. 3.3 Automatic Reduplication Processes Automatic reduplication involves reduplication that is “obligatory in combination with another affix, and which does not add meaning by itself to the overall construction” (Rubino 2005, 18). I have identified two such mechanisms in Cavineña. The first involves full reduplication, whereas the second involves final -CV reduplication. They both apply to verbs and indicate the multiple repetition of the verb event. The first automatic reduplication process (XII) combines the full reduplication of the verb root, which can be either intransitive or transitive, with the suffix -ni. As we observe below, this marker corresponds to an allomorph of the morpheme -(ne)ni that expresses, among other things, a habitual sense. The resulting reduplicated verb has the same valency as the base. The meaning expressed is habitual repetition of the event implied by the verb stem. (15) Verb (inflecting) betsakatikwapaabubetidujuiwara-

‘swim’ ‘fight’ ‘go’ ‘cry’ ‘carry O’ ‘bring O’ ‘transport O’ ‘call O’

Verb (inflecting) betsa~betsa-nikati~kati-nikwa~kwa-nipa~pa-niabu~abu-nibeti~beti-niduju~duju-niiwara~iwara-ni-

‘swim repeatedly’ ‘fight repeatedly’ ‘go repeatedly’ ‘cry repeatedly’ ‘carry O repeatedly’ ‘bring O repeatedly’ ‘transport O repeatedly’ ‘call O repeatedly’

reduplication, word class and transitivity in cavineña

Verb (inflecting)

Verb (inflecting)

jiru- ‘smell O’ sare- ‘search O’ tupu- ‘cut O’

jiru~jiru-nisare~sare-nitubu~tubu-nietc.

327

‘smell O repeatedly’ ‘search O repeatedly’ ‘cut O repeatedly’

The second automatic reduplication process (XIII) combines reduplication of the final -CV syllable of a transitive verb root with the suffix -sha ‘causative.’ The resulting reduplicated verb is also transitive. The meaning expressed is multiple repetition of the event in space or with respect to the O argument.16 (16) Verb (inflecting) berejikwikarukatsakwerepudipukusabusakwasheditaketedi-

‘pierce O’ ‘cut O’ ‘bite O’ ‘strike O’ ‘chop O’ ‘pound O’ ‘crack O’ ‘grasp O’ ‘pierce O’ ‘smash O’ ‘hack O’ ‘rub O’

Verb (inflecting) bere~re-shajikwi~kwi-shakaru~ru-shakatsa~tsa-shakwere~re-shapudi~di-shapuku~ku-shasabu~bu-shasakwa~kwa-shashedi~di-shatake~ke-shatedi~di-shaetc.

‘pierce O repeatedly’ ‘cut O repeatedly’ ‘bite O repeatedly’ ‘strike O repeatedly’ ‘chop O repeatedly’ ‘pound O repeatedly’ ‘crack O repeatedly’ ‘grasp O repeatedly’ ‘pierce O repeatedly’ ‘smash O repeatedly’ ‘hack O repeatedly’ ‘rub O repeatedly’

Both processes appear to be fairly productive, as they are attested frequently in my text corpus. They are further discussed in sections 3.3. and 3.4.

4

Productive Verbal Reduplication Patterns

I will now provide a more detailed discussion of the productive mechanisms of reduplication applying to verbs. I discuss antipassive simple full reduplication

16

Note that no examples are available of this reduplication process applied to mono- and trisyllabic roots.

328

guillaume

in section 4.1, auxiliary-triggering simple full reduplication in section 4.2, automatic full reduplication in section 4.3, and automatic final -CV reduplication in section 4.4. 4.1 Antipassive Full Reduplication As mentioned in section 3.1, there is a full reduplication process (I) that derives intransitive verbs with an antipassive sense from transitive verbs. The antipassive characteristics of the process are illustrated with the sentences in (17a-b) and (18a-b). As we can see, the A of the underived verb becomes the S of the derived verb, while the O disappears. (17) a. Era takure ara-ya. 1sg.erg chicken eat-impfv ‘I am eating chicken.’ (elicited) b. Ara~ara-ya ike. eat~eat-impfv 1sg.abs ‘I am eating/having a meal.’ (elicited) (18) a. Metajudya=tuke=∅ [una siri=kwana] utsa-wekaka-ya. tomorrow=3sg.abs=1sg clothes old=pl wash-at.dawn-impfv ‘Tomorrow I will wash the clothes early in the morning.’ (elicited) b. Utsa~utsa-ti-wa=mike? wash~wash-go-perf=2sg.abs ‘Did you go and do your laundry?’ Three pieces of evidence suggest that the verb is detransitivized in the reduplication process. First, the subject of the reduplicated verb has an absolutive form. Compare the ergative form of the 1SG independent pronoun in (17a) with the absolutive form in (17b), and observe the absolutive form of the 2SG bound pronoun in (18b). Second, if a verbal affix sensitive to transitivity (see section 1) is used on a reduplicated verb, only the intransitive allomorph of this affix is possible. As illustrated in (19), only the intransitive allomorph -neti of the posture suffix ‘stand’ can occur on the reduplicated verb ara~ara- (the transitive allomorph is -nitya). (19) Juan ara~ara-neti-ya. (*ara~ara-nitya-ya) Juan eat~eat-stand-impfv ‘Juan is eating standing.’ (elicited)

reduplication, word class and transitivity in cavineña

329

Third, it is not possible to express the notional object, whether as a core argument or as an oblique. This property classifies this antipassive type as belonging to the category of “patientless” antipassives (Dixon and Aikhenvald 2000), sometimes also referred to as antipassives with an “implicit” patient (Polinsky 2008). Semantically, the reduplicated verb denotes a culturally-codified activity, i.e., an activity that people or animals regularly repeat in the same way with respect to the same types of entities (patients). This is the case with the activities of ‘having a meal’ and ‘doing the laundry,’ as illustrated in (17b) and (18b), and other types of activities illustrated in the examples given in (4) above, like ‘toasting (manioc flour or corn),’ ‘grinding (corn),’ ‘peeling (rice or corn),’ etc. In (20), we observe a text excerpt with a sequence of two sentences that contain the verb iji- ‘drink.’ In the first, the fully reduplicated form iji~iji- ‘have a drink’ (therefore intransitive) occurs, while in the second, a transitive nonreduplicated form iji- ‘drink O’ occurs with an overtly expressed patient, tupari ‘chicha.’ (20) Tuwa ekana iji~iji-ya. Amena tupari=tuna iji-ya. there 3pl.abs drink~drink-impfv bm chicha=3pl drink-impfv ‘There they would have drinks. They would drink chicha.’ This kind of reduplication process is not fully productive. For example, during elicitation, a number of reduplicated transitive verbs were rejected by speakers, such as *chiri~chiri- ‘steal~steal,’ *kwere~kwere- ‘cut~cut’ and *rikwa~ rikwa- ‘bark~bark.’ This restriction of application presumably has to do with the semantic specificities of this kind of reduplication: it can probably only be applied to verbs that refer to events that can potentially constitute a culturallyrecognized activity in the Cavineña society. And, probably, ‘stealing,’ ‘cutting’ and ‘barking’ are not considered activities in that sense. As a result, it is quite difficult to predict which verb will accept this type of reduplication, in particular for outsiders. For example, it is probably not obvious to anybody why tya-u~tya-u-, based on tya- ‘give,’ and ba-u~ba-u-, based on ba- ‘see,’ are possible, as opposed to the three ungrammatical forms mentioned above (‘steal~steal,’ ‘cut~cut’ and ‘bark~bark’). But as shown by the translations in (21) and (22) and the explanations provided to me by a native speaker, these reduplicated forms can be used in very specific contexts. These contexts represent culturallycodified activities: people building the roof of a traditional house and hens searching for a place to lay eggs.

330

guillaume

(21) Tya-u17~tya-u-ya. give-epen~give-epen-impfv ‘He is passing on palm leaves.’ (as they are making the roof of the house) (22) Takure=tu ba-u18~ba-u-ya. chicken=3sg see-epen~see-epen-impfv ‘The hen is looking for a place to lay eggs.’ The kind of reduplication process discussed here is also found in at least two other Tacanan languages, Ese Ejja (Vuillermet 2012a,b) and Tacana (Guillaume fieldnotes). However, cross-linguistically, it appears to be a rare phenomenon. To my knowledge, it is only attested in some Oceanic languages, such as Boumaa Fijian (Dixon 1988, 48–49), Kokota (Palmer 1999, 160–161) and Mokilese (Harrison 1973, 415, cited by Moravcsik 1978).19 From the point of view of Amazonian languages, this reduplication process is also interesting. It has been claimed that the languages in this part of the world lack antipassive constructions (Aikhenvald, 2012, 235), an assertion that was corroborated by a survey conducted by Polinsky (2008). Functionally speaking, however, there does not appear to be anything exceptional to having a reduplication process that expresses (or develops into) an antipassive mechanism. First, it is known that there is a strong relationship between the use of the antipassive and imperfective-types of aspect, such as the habitual, durative, iterative, etc. (Creissels 2006, 92; Polinsky 2008). In certain languages, derived imperfective forms occasionally have a detransitivizing antipassive effect when applied to transitive verbs.20 Second, there are attested cases of reduplication with iterative semantics that strongly disfavor the overt expression of an object NP of transitive verbs. Such is the case, for example, in the Tupi-Guarani language Nheengatu, when the repetition of an event affects multiple patient participants (Cruz, this volume). It is my impression that the scarcity of attested cases of reduplication with valence-changing syntax might result from the fact that general studies in

17 18 19

20

See footnote 6. See footnote 6. Akan, a West-African language of the Kwa family, is sometimes cited in the literature for having a reduplication process with a detransitivizing antipassive-like function (see Moravcsik 1978). However, according to Appah (2011), this process has a very restricted productivity. This is true, for instance, in the continuative-intensive derivational marker -(a)dar of the Caucasian language Avar (Charachidzé 1981, cited by Creissels 2006, 87).

reduplication, word class and transitivity in cavineña

331

reduplication have not seriously considered the possibility that reduplication can have such functions. 4.2 Auxiliary-Triggering Full Reduplication with Distributive Semantics Cavineña displays a second type of simple full reduplication that applies to verbs (II). As mentioned in section 3.1, the verbal base may be either intransitive or transitive, and retains its valency after reduplication has taken place. However, the reduplicated verb becomes non-inflectional and a generic auxiliary is required to host the verbal inflectional morphology. The meaning expressed by this reduplication process is the multiple repetition of the verb event within short and regular intervals, with exactly the same participants/referents and in the same spatio-temporal settings. These formal and semantic characteristics are illustrated in (23a-b), with the non-reduplicated and reduplicated forms of the intransitive verb keke- ‘shout.’ (23) Intransitive verb a. Tudya [ekwe e-mama] keke-kware. then 1sg.gen 1-mother shout-rem.past ‘Then my mother shouted.’ (Tavo Mayo 1977, 27) b. Tudya keke~keke ju-kware. then shout~shout be-rem.past ‘So he was yelling and yelling.’ (Tavo Mayo 1977, 53) As we see in (23b), the tense suffix -kware is carried by the intransitive generic auxiliary ju- ‘be.’ If the verbal base is transitive, the generic auxiliary is a- ‘affect,’ as shown in (24a-b) with the non-reduplicated and reduplicated forms of the transitive verb iwara- ‘call O.’ (24) Transitive verb a. Cursillo=ishu=tura=ekwana iwara-kware. course=purp=3sg.erg=1pl call-rem.past ‘She called us for a (training) course.’ b. Misionero=kwana=ra=ekwana iwara~iwara a-kware. missionary=pl=erg=1pl call~call affect-rem.past ‘The missionaries kept calling us every minute.’ (elicited)

332

guillaume

In addition to the use of a transitive auxiliary, the subject of the reduplicated verb in (24b) is marked by the ergative enclitic. This provides further evidence that the clause is still transitive. The following pair of examples with the transitive verb katsa- ‘beat O’ shows the semantic difference between the auxiliary-triggering reduplication, expressing multiple repetition in (25), and the antipassive full reduplication, expressing an activity in (26).21 (25) [Kameweti ari-da=kwana=ra]=tu e-wane=kwana=ke [dutya be.jealous big-asf=pl=erg=3sg 3-wife=pl=3 all wekaka=dya] katsa~katsa a-ya. day=foc beat~beat affect-impfv ‘Jealous men beat their wives every day.’ (Camp and Liccardi 1989, 17) (26) Katsa~katsa-ya=tu e-puna utsa~utsa-ya=ke. beat~beat-impfv=3sg npf-woman wash~wash-impfv=lig ‘The woman is beating (clothes) while she is washing (clothes).’ (elicited) Unlike the antipassive reduplication process, which can only be applied to verbal roots, the auxiliary-triggering reduplication process permits certain verbal suffixes to be reduplicated together with the verb root. These affixes must not belong to the class of TAM inflections, which can only be hosted by the generic auxiliary. The following examples show the reduplication of the intransitive verb root neti- ‘stand’ together with the aspectual suffix -baka ‘short.time’ (27a), and the intransitive verb root ju- ‘be’ together with the aspectual suffix -chinepe ‘all.day’ (27b). (27) Reduplication including affixes a. [Neti-baka~neti-baka ju-ya majaka] ekatse stand-short.time~stand-short.time be-impfv even.though 3dl diru-chine. go-rec.past ‘Even though (the oxen) kept stopping (lit. standing) for short periods all along the way, at least they kept going.’

21

Note that this sentence also includes the antipassive reduplication of the transitive verb utsa- ‘wash O,’ here within a relative clause.

reduplication, word class and transitivity in cavineña

333

b. E-sewena-tere=tuna amena res-become.black-comp=3pl bm [ijeti=ju ju-chinepe~ju-chinepe ju-ya=tibu]. sun=loc be-all.day~be-all.day be-impfv=reason ‘They have become very dark because they are always in the sun all day long.’ (Camp and Liccardi 1989, 46) Full reduplication can apply to complex predicates involving a non-inflecting verb and an auxiliary, such as nereda (a-) ‘scold O’, illustrated in (28a). When this happens, only the non-inflecting verb is reduplicated and no additional auxiliary is needed, as can be seen in (28b): (28) a. Nereda=tura=∅ a-kware [era jadya kwatsabi scold=3sg.erg=1sg affect-rem.past 1sg.erg thus tell.story.to a-wa=ju]. affect-perf=ds ‘She scolded me when I told her so.’ b. Era peya nereda~nereda a-kware. 1sg.erg other scold~scold affect-rem.past ‘I was scolding the other one over and over again.’ Unlike the antipassive reduplication process, the auxiliary-triggering reduplication process seems to be fully productive. The reiterative full reduplication is not the only mechanism that has an auxiliary-triggering effect in Cavineña. An auxiliary-triggering effect is also observed with a fairly wide range of other morphological and syntactic operations that are applied to verbs/predicates; for an overview see Guillaume (2008, ch. 10). 4.3 Automatic Full Reduplication We will now turn to the first automatic reduplication process (XII), that obligatorily includes some additional morphological material.22 This process combines the full reduplicant of an intransitive or transitive verbal root with the suffix -ni. It results in a verb with the same valency and inflectional characteristics as the non-reduplicated base. The meaning expressed is habitual repetition of the verbal event over a long duration. This process is illustrated in (29) with intransitive verbs and in (30) with transitive verbs. 22

See footnote 2.

334

guillaume

(29) Intransitive verbs a. … epuna peyainime. Pa~pa-ni-kware [tuke jeteke]. woman sad cry~cry-random-rem.past 3sg looking.for ‘(After her husband had left,) the woman was very sad. She kept crying all the time longing (lit. looking) for him.’ b. Apu~apu=ju=tuna kwa~kwa-ni-kware. dark~dark=loc=3pl go~go-random-rem.past ‘They used to go (hunting) when it was still dark.’ (30) Transitive verbs a. Ari-da=ke=kamadya=tunaja=tu big-asf=lig=only=3pl.dat=3sg duju~duju-ni-kware. transport~transport-random-rem.past ‘They would only carry the big ones (fishes).’ b. [Ekwe e-tata=ra ekwe e-mama=ra jadya] [tumeke 1sg.gen 1-father=erg 1sg.gen 1-mother=erg and that kwatsabiji] ekwana kweja~kweja-ni-kware. story 1pl inform~inform-random-rem.past ‘My father and my mother used to relate that story to us.’ (Tavo Mayo 1977, 7) The suffix -ni exists independently of the reduplication process in Cavineña as an allomorph of a verbal morpheme -(ne)ni which expresses, among other meanings, a habitual sense. The allomorph -ni is used with polysyllabic stems, as in (31a), and the allomorph -neni with monosyllabic stems, as in (31b).23 (31) a. Tumepatya=ekwana misionero=kwana=ra Tumichucua=ju at.that.time=1pl missionary=pl=erg Tumichucua=loc iwara-ni-kware … call-random-rem.past ‘At that time, in Tumichucua, the missionaries would call us (every year) …’

23

The suffix -(ne)ni can also encode the random distribution of the verbal event across different locations; for more details see Guillaume (2008, 206–208).

reduplication, word class and transitivity in cavineña

335

b. Je-neni-ya=mike! come-random-impfv=2sg ‘You always come (to Riberalta)!’ Since both verbal full reduplication and the habitual suffix -ni exist independently in the language, and since they both express multiple repetition when used independently, one may wish to analyze the full reduplication + -ni process as compositional. That is, the habitual suffix would apply after verbal reduplication and add some additional nuance to the meaning expressed by reduplication. Such an analysis is incorrect though, for two reasons that relate to the full reduplication pattern without suffixation. First, as shown in section 3.1, when simple full reduplication occurs, it yields either an intransitive inflecting verb (with an antipassive effect) or a non-inflecting verb (which takes a generic auxiliary). Neither of these effects are observed when full reduplication combines with -ni. Second, recall that full reduplication of monosyllabic roots requires an epenthetic vowel -u (see footnote 6). When we observe the fully reduplicated form of the verb root pa- ‘cry’ combined with the suffix -ni in (29a), we observe no epenthetic vowel. This suggests that, at least from a synchronic point of view, the morphological process consisting of reduplication and the suffix -ni behaves differently. Yet, it is intriguing how such a situation came about historically. I do not have any compelling hypotheses for this at the present time. I have not investigated the range of verbs that undergo this reduplication process, but it is relatively frequent in many of my texts. This suggests that it is fairly productive. 4.4 Automatic Final -CV Reduplication The second automatic reduplication process (XIII) combines the final -CV syllable of a typically transitive verb root24 with a suffix -sha, as in (32). (32) a. Tedi~di-sha-kware=tunara=ike. rub~red-caus-rem.past=3pl.erg=1sg.abs ‘They rubbed me all over.’

24

During elicitation with one informant, I obtained examples of this reduplication process with the following three intransitive verbs: ani- ‘sit,’ nubi- ‘enter,’ and tsura- ‘go up.’ More work is needed to confirm whether these forms are generally accepted by other informants and whether they are used in spontaneous texts.

336

guillaume

b. Tudya=tuna [espere jiruru] raru~ru-sha-ya then=3pl stream at.edge.of cut~red-caus-impfv kuchiru=tsewe. machete=assoc ‘They cut (the barbasco lianas) into little pieces with a machete at the banks of the stream.’ (Tavo Mayo 1977, 14) c. …[piya=tsewe salon=tsewe jadya mare~re-sha-tsu] arrow=assoc rifle=assoc and shoot.at~red-caus-ss tsajaja-sha-chine. run-caus-rec.past ‘(The Araona people) shot at them repeatedly with arrows and rifles and scared them away (lit. made them run).’ As we can see, the resulting verb is still transitive and inflecting. The meaning expressed by the derivational process is the repetition of the verb event and its distribution across space or with respect to the O argument referent. Unlike with the auxiliary-triggering reduplication process (section 4.2), the focus is not on the regularity of recurrence of the repeated event, but on the importance of its spatial scope, either over the place where it occurs or with respect to the O referent. For example, we can compare katsa~katsa (a-) ‘beat O repeatedly,’ illustrated in (25), with katsa~tsa-sha- ‘beat O (all over)’ which is roughly translated as ‘beat O on the whole body’ by native speakers. An example with the perception verb peta- ‘look at O’ is given below: (33) Era=tu peta~ta-sha-chine [tura 1sg.erg=3sg look.at~red-caus-rec.past 3sg.erg be-chine=ke] … bring-rec.past=lig ‘I’ve looked at everything he (the merchant) brought (one by one, to be sure of its quality).’ (elicited) The suffix -sha exists independently in the language, as a causative marker for intransitive verbs,25 illustrated in (34). (34) a. Akwi=tu riwi-ya. tree=3sg fall-impfv ‘The tree is going to fall down.’ (Camp and Liccardi 1989, 86) 25

Transitive verbs take a different marker, -mere; see Guillaume (2008, 292–297).

reduplication, word class and transitivity in cavineña

337

b. Bari=ra riwi-sha-kware Eduardo. giant.anteater=erg fall-caus-rem.past Eduardo ‘The giant anteater made Eduardo fall (by striking him).’ (elicited) Since both partial reduplication and a suffix -sha exist independently in Cavineña, a compositional analysis is compelling. Under such an analysis, the suffix -sha would be the causative suffix which would apply after the reduplication of the verbal root. Both would contribute some part of the resulting meaning. However, such an analysis must be rejected, at least from a synchronic point of view, on the basis that the process of final -CV reduplication is not productive at all (section 3.2). From a diachronic point of view, similar to the full reduplication + -ni process discussed above, it is intriguing how the collocation of the final -CV reduplication + -sha arose. I do not have any compelling hypotheses for this at the present time. One could imagine that at some point in the history, final -CV reduplication could have had a detransitivizing effect. However, unless this effect had a passive function, we can not see how the original A argument of underived transitive bases remained in A function after the causative -sha had been applied (e.g., tedi- ‘rub O’ → *tedi~di- ‘be rubbed (all over)’ → *tedi~di-sha- ‘cause O to be rubbed (all over) = rub O (all over)’). Whereas reduplication processes with detransitivizing antipassive functions are attested and functionally motivated (see section 4.1), reduplication processes with detransitivizing passive functions are, to my knowledge, unattested, and hard to account for in functional terms. This idea suggests that one has to look for a different kind of origin. One wonders whether the final -CV reduplication + -sha marker could have arisen in a construction where the morpheme sha- was an independent transitive verb ‘make’ and the reduplicated verb was a deverbal noun functioning as the O argument of this verb.26 A slightly different possibility might be that the process of final -CV syllable reduplication would have had an auxiliary-triggering effect, akin to the type instantiated by full reduplication in present-day Cavineña (see section 4.2). As such, the reduplicated verb would still have been used predicatively but only jointly with an auxiliary, in this case, sha-. Provided that sha- were transitive, the construction would have been transitive. Unfortunately, I currently have no proof that the morpheme sha was a verb or an auxiliary. The productivity of the reduplication process discussed here is uncertain. This process is attested in about 30 verb roots in my text corpus, and more systematic research on the topic remains to be done.

26

I thank Denis Creissels for this suggestion.

338 5

guillaume

Conclusions

In this paper, I have given an overview of all the reduplication processes that are attested in Cavineña. They are summarized in Table 2. The following conclusions can be made concerning reduplication in this language. – Cavineña exhibits both full and partial reduplication, but only full reduplication is productive, a fact congruent with Moravcsik’s (1978, 328) crosslinguistic proposal that full reduplication is more basic than partial reduplication. – In Cavineña, most simple reduplication processes and those that are productive alter the morphosyntactic properties of the base, whether in terms of transitivity, inflectional characteristics, or word class. The antipassive effect for one type of full reduplication is noteworthy, as this phenomenon seems to be rather rare cross-linguistically. – Productive reduplication processes tend to apply to verbs or predicative adjectives, rather than nouns. This fact is congruent with the highly elaborate morphological structure of verbal words in this language and the very limited morphology of nouns. – The formal and phonological characteristics of all the reduplication processes are relatively easy to describe. This fact accords well with the straightforward phonological and morphophonological structure of this language, which has very little allophony or allomorphy, simple syllable structure, and highly transparent agglutinative morphological structures. There are only two formal issues that arise in these processes. The first concerns the addition of an epenthetic vowel u to monosyllabic bases in certain contexts. The explanation for this phenomenon has to do with a disyllabic minimality constraint on phonological words. The second complication is observed in “automatic” reduplication processes, where the reduplicated form must combine with an additional suffix. The most interesting aspect of reduplication in this language is probably its syntactic uses, as a word class-changing device and as a valence-changing process. The syntax of reduplication is largely neglected in general studies on the topic. It is hoped that this paper will raise interest among linguists to look at the possible syntactic effects of reduplication.

Simple

Base word

Semantics

having many X

independent word class pred. adjective changing

noun

noun

noun

noun

none

N/A

none

spatial or temporal distribution

word class changing bound pred. independent adjective class adjective pred. adjective changing

-kwija kwija~kwija

jerujeru~jeru apuapu~apu

kuetikueti~kueti ( ju-)

araara~ara-

Examples

consisting of many X nusa nusa~nusa onomatopoeia and biribiri ideophones

multiple repetition

noun

multiple repetition

infl. verb

verb class changing

non-infl. verb

detransitivizaculturally-codified tion (antipassive) activity

Syntactic effect

infl. verb

intrans. infl. verb

Reduplicated word

The 13 reduplication processes in Cavineña

full trans. infl. verb

table 2

‘eat O’ ‘eat, have a meal’ ‘pass’ ‘pass repeatedly’ ‘sing’ ‘song’ ‘dark’ ‘dark in diff. places’ ‘thorn’ ‘hav. many thorns’ ‘mark’ ‘alphabet’ ‘parakeet sp.’

VII

VI

V

IV

III

II

I

Type

reduplication, word class and transitivity in cavineña

339

Simple

Automatic

verb

noun

trans. infl. verb none

trans. verb

verb

noun

inf. verb

infl. verb

none

partial trans. verb

full

quantifier

Reduplicated word

postposition postposition

quantifier

partial adjective

full

Base word

none

none

word class changing N/A

word class changing N/A

none

none

Syntactic effect

multiple repetition

regular/habitual repetition

onomatopoeia or ideophones

onomatopoeia or ideophones multiple repetition

inchoative

none

distributive

Semantics

jikwijikwi~jikwi-sha-

betsabetsa~betsa-ni-

watsuwatsu~tsujapipi

apuapu~pupakaka-

peadya peadya~peadya patya patya~patya

Examples

‘swim’ ‘swim repeatedly’ ‘cut O’ ‘cut O repeatedly’

‘butterfly’

‘twist O’

‘dark’ ‘darken’ ‘fall’

‘one, a’ ‘some’ ‘in middle of’ ‘in middle of’

XIII

XII

XI

X

IX

VIII

Type

340 guillaume

reduplication, word class and transitivity in cavineña

341

Acknowledgments The study reported in this paper was presented at the 53rd International Congress of Americanists, Universidad Iberoamericana Ciudad de México, Mexico, July 19–24, 2009, within the symposium “Reduplication in the Amazonian languages” organized by Gale Goodwin Gómez and Hein van der Voort. I am grateful for the comments I received during the symposium, as well as for those by Denis Creissels, Aline Da Cruz, Christian DiCanio, Gale Goodwin Gómez, Françoise Rose, Hein van der Voort and an anonymous reviewer on earlier versions of this paper.

References Aikhenvald, Alexandra Y. 2012. The languages of the Amazon. Oxford: Oxford University Press. Appah, Clement Kwamina Insaidoo. 2011. Reduplication as a detransivizing strategy in Akan: a preliminary observation. Paper presented at the 42nd Annual Conference on African Linguistics, University of Maryland, June 10–12, in College Park, MD, USA. Camp, Elizabeth L., and Millicent R. Liccardi. 1989. Diccionario cavineña-castellano castellano-cavineña con bosquejo de la gramática cavineña. Dallas: Summer Institute of Linguistics. Charachidzé, Georges. 1981. Grammaire de la langue avar. Paris: Jean Favard. Creissels, Denis. 2006. Syntaxe générale, une introduction typologique, Vol. 2, La Phrase. Paris: Hermès—Lavoisier. Cruz, Aline da. this volume. Reduplication in Nheengatu. Dixon, R.M.W. 1988. A grammar of Boumaa Fijian. Chicago: University of Chicago Press. Dixon, R.M.W., and Alexandra Y. Aikhenvald. 2000. Introduction. In Changing valency. Case studies in transitivity, ed. R.M.W. Dixon and Alexandra Y. Aikhenvald, 1–29. Cambridge: Cambridge University Press. Epps, Patience. this volume. Reduplication in Hup (northwest Amazonia). Guillaume, Antoine. 2008. A grammar of Cavineña. [Mouton Grammar Library 44]. Berlin/New York: Mouton de Gruyter. Goodwin Gómez, Gale. this volume. Reduplication in the Yanomae language of northern Brazil. Harrison, Sheldon P. 1973. Reduplication in Micronesian languages. Oceanic Linguistics 12/1–2: 407–454. Moravcsik, Edith A. 1978. Reduplicative constructions. In Universals of human language, Vol. 3, ed. Joseph H. Greenberg, 297–334. Stanford: Stanford University Press,

342

guillaume

Palmer, Bill. 1999. A grammar of the Kokota language, Santa Isabel, Solomon Islands. PhD diss., University of Sydney. Polinsky, Maria. 2008. Antipassive constructions. In The world atlas of language structures online, ed. Martin Haspelmath, Matthew S. Dryer, David Gil, and Bernard Comrie. Munich: Max Planck Digital Library, chapter 108. http://wals.info/feature/108 (accessed 22 January 2010). Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Studies on reduplication, ed. Bernhard Hurch with Veronika Mattes, 11–29. Berlin/New York: Mouton de Gruyter. Tavo Mayo, Victor. 1977. Yahua buricuare que. La tierra se partió. Riberalta: Instituto Lingüístico de Verano. Vuillermet, Marine. 2012a. Ese Ejja. In Lenguas de Bolivia, Tomo II: Amazonía, ed. Mily Crevels and Pieter Muysken, 71–114. La Paz: Plural editores. . 2012b. A Grammar of Ese Ejja, a Takanan language of the Bolivian Amazon. PhD diss., Université Lumière Lyon 2.

Abbreviations [] 1, 2, 3 a abs abs asf assoc bm caus comp contr dat dl ds epen erg foc gen hort impfv

multiple-word constituent first, second, third person transitive subject absolutive case Absolutive (dummy) adjective suffix Associative boundary marker Causative Completive Contrastive Dative Dual different subject Epenthesis Ergative Focus genitive case Hortative Imperfective

lig loc neg npf o pass perf pl purp rec.past random red rem.past rep res restr s sg ss

Ligature locative case Negative (dummy) noun prefix transitive object Passive Perfect Plural Purpose recent past random distribution Reduplication remote.past Reportative Resultative Restrictive intransitive subject Singular same subject

chapter 13

Reduplication in Movima: A Prosodic Morphology Approach Katharina Haude

Reduplication in Movima (isolate, lowland Bolivia) covers a remarkable range of grammatical functions. Different reduplication processes indicate morphosyntactic categories such as direct, inverse and middle voice on verbal bases; on nominal bases, reduplication serves to form subordinate predicates, inalienably possessed nouns, and possessive predicates. In contrast, the crosslinguistically typical functions of reduplication, such as emphasis or plurality, are not or are only marginally present. Movima reduplication is based not on morphemes but on prosodic categories (moras and feet). With all these properties, it is best analyzed in terms of prosodic morphology, where reduplicative morphemes are considered phonologically underspecified affixes.

1

Introduction

1.1 Some Grammatical Characteristics of Movima1 Movima is a linguistic isolate still spoken by several hundred elderly speakers in the Bolivian Amazon area, mainly in and around Santa Ana del Yacuma in the department of Beni. The language was first investigated by the SIL linguists Robert and Judith Judy (see e.g. Judy and Judy 1967). Movima clause structure is predicate initial. The linear order of the two core arguments in a transitive clause depends on the position of the nominal referents in a referential hierarchy, the argument denoting the higher-ranking referent occurring in first position after the predicate. The semantic roles of the two core arguments are indicated by direct/inverse morphology on the verb (see e.g. Haude 2006, Haude 2009). Movima morphology is mainly agglutinating and analytic (one morpheme—one meaning). However, to a large extent it is also prosodically based. For instance, there is an infix ⟨kak⟩ irrealis, which is inserted not inside the root but after the first iambic foot

1 For detailed information on Movima phonology and grammar, see Haude (2006).

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_014

344

haude

of the word, independently of the word’s complexity; similarly, the direct suffix -na has a base-internal allomorph -a- that is used whenever the root of a morphologically complex verb is monosyllabic and ends in a consonant. In addition, there are several reduplication processes, to be discussed in this paper. Compounding and noun incorporation are frequent, generally involving classifier-like bound elements. Tense, mood, and aspect are expressed by particles. Number and natural gender are indicated by referential elements (articles, personal pronouns, demonstratives), which additionally indicate presence, absence, position, and ongoing vs. ceased existence of the referent. There are only two case categories: core (morphologically unmarked) and oblique (marked by the prefix n-). The phoneme inventory of Movima contains 19 consonants (represented orthographically but with phonetic specification when not self-explanatory): p, t, k (realized as [pʔ͡m], [tʔ͡n], and [ʔ], respectively, in coda position), ' [ʔ], kw [kw], b [ɓ], d [ɗ], ch [tʃ], v [ß], s, j [h], ɬ, m, n, l, r [ɾ], w, y [j], and y' [jʔ]. The five vowel phonemes are i, e, a, o, and u. Syllable structure is CVC or CV(ː). Stress generally falls on the penultimate syllable of the word, except when the word ends in a glottal(ized) consonant, which attracts stress to the last syllable. A penultimate open syllable is usually lengthened; the major exception is words that end in a glottal stop. A lexical root must minimally consist of a heavy syllable, i.e., CVC or CVː. An independent content word must consist of minimally two syllables constituting three moras, as will be shown below. A lexical root can only occur independently if it fulfills these criteria; otherwise, it must be morphologically or phonologically augmented, and, as will be seen, reduplication can serve as a way to achieve this. 1.2 Reduplication in Movima: An Overview Movima has four reduplication processes, three word-initial ones and a wordinternal one. Word-initial reduplication includes monomoraic, bimoraic and foot reduplication; word-internal reduplication is monomoraic. All reduplication processes are regressive, i.e., the copy precedes the source. The different reduplication processes cover a large range of grammatical functions. In the verbal domain, word-initial monomoraic reduplication marks direct voice (Section 4.1.1); bimoraic reduplication marks inverse and middle voice (4.1.1, 4.1.2). Word-initial foot reduplication only occurs on nouns, from which possessive predicates are derived (4.2.3). Word-internal reduplication has two functions on nominal bases: on the one hand, it marks nouns as inalienably possessed (4.2.1), and, on the other hand, it derives subordinate predicate nominals (4.2.2); it is also a means to mark inverse and middle voice on some verbal bases (4.1.1, 4.1.2). The different functions of one and the same

reduplication in movima: a prosodic morphology approach

345

reduplication process are easily identified on the basis of the morphological, lexical and syntactic environment. All reduplication processes are also found in places where their grammatical function is not obvious (4.3). While in some cases, they can be identified as creating prosodically well-formed words, in others it is not clear whether their function is grammatical or rather phonological. In neither of these cases does reduplication seem to be fully productive at the present stage of the language. This paper is structured as follows. Section 2 presents general facts of the prosody of Movima words. In section 3, the four reduplication processes are illustrated. Section 4 discusses the functions of reduplication. The conclusions are summed up in section 5.

2

The Structure of Movima Words

2.1 Syllable Structure For a better understanding of Movima reduplication, a few words have to be said about the prosodic structure of Movima words and syllables. There are two syllable types in Movima: heavy (H) and light (L). Heavy syllables are either closed (CVC) or open and long (CVː); light syllables are short and open (CV). There are no consonant clusters. When the word-initial consonant is a glottal stop (see Haude 2006, 38–42), it is not orthographically represented, as in (1); the glottal stop is represented only when forming the onset of an element that occurs inside a word, e.g. in compounding, incorporation or when preceded by reduplication (see e.g. (17) and (19) below). In (2) this is illustrated for compounds. (1)

i:may eɬ'i o:ra

['ʔiːmaj] ['ʔɛɬʔi] ['ʔɔːɾa]

‘night’ ‘name’ ‘hour’

(2)

pek-'i:may [pɛʔ'ʔiːmaj] ‘all night’ (all-night) kem-'eɬ ['kɛmʔɛɬ] ‘my surname’ (inform-name) eɬ-'o:ra [ʔɛɬ'ʔɔːɾa] ‘what time’ (name-hour)

2.2 Prosodic Words Movima content words (nouns, adjectives, verbs) are minimally disyllabic and consist of at least three moras. In (3), some typical examples are given. When a content word is disyllabic, its first syllable is generally heavy (i.e., two of the

346

haude

minimally three moras occur on the first syllable); hence, if the first syllable is open (CV), its vowel is usually lengthened, as can be seen in (3) with to:mi ‘water’ and bi:jaw ‘old.’ (3)

bayɬim alra to:mi bi:jaw baba:kwa kori:di

‘garden, field’ ‘my friend’ ‘water’ ‘old’ ‘fruit’ ‘stick’

There is a closed class of about seven nouns that are phonologically defective in that they only consist of two light syllables (CVCV): (4)

ma'a pa'a je'e ve'e ko'o chi'i ke'e

‘mother’ ‘father’ ‘state of being’ ‘fire’ ‘tree, firewood’ ‘excrement’ ‘sister’ (respectful address)

These nouns have in common that they contain a glottal stop, which is followed by a vowel identical to the one preceding the glottal stop. They show special behavior when other morphemes are added (see Haude 2006, 197, and 4.3.4 below). 2.3 Stress and Vowel Length Stress normally occurs on the penultimate syllable of the word. In (5), it can be seen that when suffixes are added to a word, stress shifts to the right: (5)

a. bo:ve b. bove:-mo

['ɓɔ:ßɛ] [ɓɔ'ßɛːmɔ]

‘fan of straw’ ‘jasayé (-mo ‘clf.basket’) (type of basket)’ c. bove-mo:-ba [ɓɔßɛ'mɔːɓa] ‘little basket’ (-ba ‘clf.round’) d. bove-mo-to:da [ɓɔßɛmɔ'tɔːɗa] ‘broken jasayé’ (-toda ‘clf.piece’)

The glottal stop attracts stress. This means that words ending in /k/ (phonetically realized as a glottal stop) or in /t/ or /p/ (glottalized in coda position, see 1.1) are stressed on the last syllable:

reduplication in movima: a prosodic morphology approach

(6)

347

chu:jat [tʃuː'hat͡ʔn] ‘motacú (palm tree)’ ku:dup [kuː'ɗup͡ʔm] ‘flea’ pa:kona:nak [paːkɔnaː'naʔa] ‘fox’

The lengthening rules in Movima are, in general, as follows: i) when the penultimate syllable of a word is open, the vowel is lengthened; ii) all other syllables are short. This can be observed in (5) above. There are quite a few deviations from the lengthening rule, however. For instance, most disyllabic words ending in the simple glottal stop (which is often followed by a release vowel identical to the preceding vowel) have a short penultimate syllable: (7)

merek tochik karak bitok

[mɛ'ɾɛʔɛ] [tɔ'tʃiʔi] [ka'ɾaʔa] [ɓi'tɔʔɔ]

‘big’ ‘small’ ‘macaw’ ‘old person’ (affective)

There are also some trisyllabic words where lengthening of the penultimate syllable does not occur. Examples are given in (8). Note that despite the absence of lengthening, the words in (7) and (8) conform to the minimality criterion in that they consist of three moras (light plus heavy syllable and three light syllables, respectively). (8)

chinaɬa awaro juyeni yonali

[tʃi'naɬa] [ʔa'waɾɔ] [hu'jɛni] [jɔ'nali]

‘manioc’ ‘parrot’ ‘person’ ‘caiman’

On the other hand, there are also words that have a long syllable where this would not be expected. A case in point is Spanish loan words, which are adapted to the Movima stress rules. When the penultimate syllable is stressed, it is also lengthened when open, as shown in (9). When, however, a Spanish loan originally carries stress on a syllable other than the penultimate (e.g. Spanish máquina ‘machine,’ música ‘music’ in (10)), the penultimate syllable is stressed, but remains short, while the originally stressed syllable is lengthened. (9)

pola:ta [pɔ'laːta] ‘money’ sani:ya [sa'niːja] ‘melon’ polisi:ya [pɔli'siːja] ‘police’

(Sp. plata ‘silver, money’) (Sp. sandía ‘melon’) (Sp. policía ‘police’)

348

haude

(10) ma:kina [maː'kina] ‘machine’ (Sp. máquina) mu:sika [muː'sika] ‘music’ (Sp. música) ele:siya [ʔeleː'sija] ‘church’ (Sp. iglesia)2 In line with the general lengthening rules, disyllabic loans with the structure CVCV always have a long first syllable, like native words. However, unlike typical native words of the same structure (e.g. bo:ve in (5a) above), all loans with the structure CVCV retain the long vowel also when further suffixes are added. This contradicts the second lengthening rule, according to which the lengthening would automatically be lost. Examples (11)–(13) illustrate this: (11) a. wa:ka ['waːka] ‘cow’ b. wa:ka-wandi [ˌwaːka'wanɗi] ‘ranch’

(-wandi ‘ins.house’)

(12) a. pa:ko ['paːkɔ] ‘dog’3 b. pa:ko-na:nak [ˌpaːkɔnaː'naʔa] ‘fox’

(-na:nak ‘pseu’)4

(13) a. mo:to b. mo:to-to:da

['mɔːtɔ] [ˌmɔːtɔ'tɔːɗa]

‘motorbike’ ‘wrecked (-toda ‘piece’) motorbike’

The lengthening properties of Movima words are relevant for some of the reduplication processes discussed in the following section.

3

Forms of Reduplication

The prosodic units relevant for reduplication in Movima are moras and feet (see McCarthy and Prince 1995). The reduplication processes that occur in Movima are listed in (14). All these processes are regressive, i.e., the copy precedes the source. None of the processes can occur repeatedly within one word,

2 Since Movima has no diphthongs, Spanish diphthongs (like /ia/ in iglesia) are reanalyzed as disyllabic (/ija/). 3 The etymology of the noun pa:ko ‘dog,’ which occurs in this or similar forms in other native Bolivian languages as well, is not clear. Movima treats it like a Spanish loan both phonologically (lexical length of the first syllable) and morphologically (see 4.3.5 below). 4 The suffix -na:nak [naː'naʔ] is exceptional in that it has a lengthened penultimate syllable despite the final glottal stop, which is possibly due to a fossilized reduplication (⟨na:~⟩nak); this is irrelevant for the phenomenon illustrated here, however.

reduplication in movima: a prosodic morphology approach

349

but combinations of two different processes do occur. In (14), the reduplication processes are characterized by the prosodic properties of the copy: µ = mora, L = light syllable (CV), H = heavy syllable (CVC or CVː). As will be seen, this does not say anything about the structure of the base to which the reduplication applies; for instance, monomoraic reduplication can apply to a bimoraic unit, of which only the first CV segment is copied.5 (14)

Movima reduplication processes a. b. c. d.

initial monomoraic reduplication (µ~) initial bimoraic reduplication (µµ~) initial foot reduplication (LH~, H~, or LL~) internal monomoraic reduplication (⟨µ~⟩)

The following subsections describe each reduplication process in turn. In this section, and in line with the practice of the present volume, the copied part is glossed as ‘red,’ independently of its formal or functional properties. In the subsequent sections, where the different functions of reduplication are explained, the reduplicated elements will be treated in terms of prosodic morphology, i.e., as templatic affixes with a proper grammatical meaning. 3.1 Initial Monomoraic Reduplication (µ~) The addition of a word-initial monomoraic segment through reduplication is illustrated in (15)–(17). (Regarding (17), recall that a word-initial glottal stop is not represented in the practical orthography.) (15) be~bet-kwa red~hide-abs ‘leather, hide’ (16) ju~ju:-wa=∅ red~scold-nmlz=1sg ‘my scolding (you/her/him/it/them)’

5 In this article, reduplication is not discussed in terms of “full” and “partial.” Either of these phenomena is accidental in Movima, where the rules for reduplication are entirely prosodic. Hence, when a root or word is copied as a whole (as e.g. in (16–17) or (26) below), this is only because its structure happens to correspond to the reduplication template, not because there existed some rule for full reduplication.

350

haude

(17) a~'am-wa=∅ red~enter-nmlz=1sg [ʔa'ʔamwa] ‘my putting (it) into something’ Initial monomoraic reduplication occurs productively on verb roots, as in (16) and (17), where it marks direct voice (see 4.1.1). Non-productively it is also found with monosyllabic noun roots, as in (15) (see 4.3.4). 3.2 Initial Bimoraic Reduplication (µµ~) Bimoraic reduplication can have several forms: when the base permits it, the copy consists of a single heavy syllable, i.e., either CVC, as in (18), or CVː, as in (20); when the base commences with a light syllable (CV), the copy includes the additional CV segment of the following syllable, as in (21). Note that the latter type is not frequent: in the domain where bimoraic reduplication is most productive, namely the marking of inverse and middle voice (see 4.1.1), bases commencing with a light syllable normally undergo word-internal reduplication (see 3.4). (18) sal~sal-wa=∅ red~look.for-nmlz=1sg ‘my being looked for (by you/her/him/it/them)’ (19) am~'am-wa=∅ red~enter-nmlz=1sg [ʔam'ʔamwa] ‘my being put into (something)’ (20) ju:~ju:-wa=∅ red~scold-nmlz=1sg ‘my being scolded (by you/her/him/it/them)’ (21) puru~purut-wa=∅ red~kiss-nmlz=1sg ‘my being kissed (by you/her/him/it/them)’ 3.3 Initial Foot Reduplication Foot reduplication copies the first iambic foot of the base and can have the shape LH, H, or LL (see Kager 1995, 397). Which of these shapes the reduplicated element has, depends on the structure offered by the base. When the base

reduplication in movima: a prosodic morphology approach

351

starts in a LH segment, i.e., when a maximal iambic foot is available, this entire segment is copied, as is illustrated in (22) with the word chorankwanto ‘hat.’ (22) choran~chorankwanto red~hat ‘to have/wear a hat’ When no maximal iambic foot is available, i.e., when the base commences with a heavy or with two light syllables, the reduplication process is formally identical to bimoraic reduplication. (As will be shown in 4.2.3 below, however, iambic foot reduplication can be identified by its function of marking predicative possession.) For a heavy first syllable consider (23) (CVC) and (24) (CVː). (23) maj~majni red~offspring ‘to have children’ (24) wa:~wa:ka red~cow ‘to have cattle’ The foot reduplication of a base commencing in two light syllables (e.g. chinaɬa ‘manioc’) is shown in (25). When comparing (25) with (22), it can be seen that foot reduplication copies a prosodic unit already present in the base: it does not just copy the first CVCVC sequence to create a maximal iambic foot (which in the case of (25), would result in *chinaɬ~chinaɬa). (25) china~chinaɬa red~manioc ‘to have manioc’ The base in (24) above, wa:ka, is a loan from Spanish (vaca ‘cow’). As was already mentioned, in disyllabic loans with an open first syllable, the first syllable is inherently long, i.e., it remains long also when the word is augmented (see (11)–(13) above). It therefore counts as heavy and can constitute an iambic foot. By contrast, on native words with an open first syllable, the first syllable is lengthened only when occurring in penultimate position; when additional elements are attached, the lengthening is dropped, as can be observed in (5)

352

haude

above. Hence, the first syllable of a native word with the structure CV(:)CV not being inherently long, iambic foot reduplication copies the entire word. This is illustrated in (26) with the word ro(:)ya ‘house.’6 (26) roya~ro:ya red~house ‘to have a house’ Also the phonologically defective nouns (see (4) above), whose structure is CVCV without any lengthening, are entirely reduplicated, as in (27) and (28). (Sometimes the penultimate syllable of the resulting word is lengthened, but the data are not consistent on this.) (27) ko'o~ko'o red~tree ‘to have trees’ (28) ma'a~ma'a red~mother ‘to have a mother’ 3.4 Internal Monomoraic Reduplication (⟨µ~⟩) Internal monomoraic reduplication involves the copying of the final CV segment of the base. This, too, is a regressive reduplication process, i.e., the copy precedes the source. The regressive character of the reduplication is only apparent when the base ends in a consonant, like ɬokot- ‘boil’ in (29), since on vowelfinal bases, copy and source are identical, as shown in (30). For the sake of comparison, the (a)-examples below show the roots in their non-reduplicating form, while the (b)-examples illustrate the reduplication.7

6 In elicitation, a speaker once accepted my proposal to reduplicate only the first syllable of a native word (ro:~ro:ya ‘to have a house’). However, this was never provided spontaneously and never confirmed by text data. By contrast, the full reduplication of a Spanish loan (*waka~wa:ka ‘to have cattle’) is always considered ungrammatical. 7 In (29b), the reduplication only involves the root, not the suffix -na direct that appears in (29a). In fact, the reduplicative element is the middle voice marker (see 4.1.2), which is incompatible with the direct suffix. In (30b), the reduplication involves a suffix, -lo ‘clf.liquid.’ This, however, does not mean that the reduplication is based on the morphological structure (root + suffix) of the base. Rather, the process applies to the entire base irrespective of its internal morphological structure.

reduplication in movima: a prosodic morphology approach

(29) a. loy iɬ ɬokot-na=∅ itn 1 boil-dr=1sg ‘I’ll boil it.’ (30) a. jayna mat-lo dsc undo-clf.liquid ‘It has already melted.’

353

b. jayna ɬo⟨ko:~⟩kot is to:mi dsc boil⟨red~⟩ art.pl water ‘The water is already boiling.’ b. jayna mat-⟨lo:~⟩lo dsc undo⟨red~⟩clf.liquid ‘It melts already.’

Internal reduplication has many different functions, depending on the lexical base to which it applies. Most importantly, it marks inverse (4.1.1) and middle voice (4.1.2) on verbs, inalienable possession on nouns (4.2.1), and it derives subordinate predicate nominals from nouns (4.2.2). Internal reduplication can co-occur with iambic foot reduplication, as illustrated in (31) (see 4.2.3 for this and other examples). (31) n-as maj~maj⟨ni~⟩ni='ne obl-art.n red~offspring⟨red~⟩=3f ‘when she has children’

4

The Functions of Reduplication

The four reduplication processes described above have various grammatical meanings. The reduplication processes, the bases to which they apply productively, and the meanings they have with these bases are listed in Table 1. (Less productive occurrences and less clearly identifiable functions will be discussed in Section 4.3 below.) Given their clearly identifiable functions, the reduplication processes are best analyzed in terms of prosodic morphology, which considers reduplication as “a special case of ordinary affixal morphology, where the affixes are phonologically underspecified, receiving their full phonetic expression by copying adjacent segments” (Broselow and McCarthy 1983, 25; see also Marantz 1994). Hence, the reduplicative morphemes are represented as affixes, the tilde identifying them as reduplicated elements; furthermore, word-internal reduplication is signaled by angle brackets, which are conventionally used for infixation. In the table above and the discussion to follow, the glosses are now used to indicate the meanings of these affixes; only when no clear meaning can (as yet) be identified, are the reduplicated elements glossed as red, in the same way as in the previous section.8 8

All but some elicited single-word examples stem from spontaneous texts.

354

haude

table 1

Forms and functions of Movima reduplication

Reduplicant Base µ~ µµ~ µµ~ ⟨µ~⟩ ⟨µ~⟩ ⟨µ~⟩ ⟨µ~⟩ foot~

monosyllabic verb root with suffix monosyllabic verb root with suffix monosyllabic verb root disyllabic verb root or complex base with suffix disyllabic verb root or complex base noun noun noun

Function

Gloss

direct voice

dr~

inverse voice middle voice

inv~ md~

inverse voice

⟨inv~⟩

middle voice inalienable possession subordinate predicate predicative possession

⟨md~⟩ ⟨inal~⟩ ⟨nmlz.n~⟩ poss~

4.1 Reduplication on Verbs 4.1.1 Marking Bivalent Verbs: µ~ direct, µµ~ inverse Movima bivalent verbs, i.e., verbs that form the predicate of a transitive clause, are morphologically marked for either direct or inverse voice (see Haude 2006, 323–328). Direct marking indicates that the first post-verbal argument, attached to the predicate through ‘internal cliticization’ (marked by =), is the actor, while the second post-verbal argument, which is either not attached to the predicate or attached to it by ‘external cliticization’ (marked by --,9 is the undergoer; inverse marking indicates the reversed situation.10

9

10

Internal cliticization creates a new prosodic word, which carries stress on the penultimate syllable. Furthermore, for consonant-final hosts it involves the attachment of the linking vowel -a. This process applies to pronouns as well as to determiners (articles or demonstratives). In contrast, the only morphophonemic effect of external cliticization is the resyllabification with a preceding consonant (see Haude 2006, 97–103). The terms ‘direct’ and ‘inverse’ reflect the fact that the first postverbal argument outranks the second postverbal argument in a referential hierarchy (1 > 2 > 3, 3rd given > 3rd new; see Haude 2009); since a high-ranking referent is a more prototypical actor and a low-ranking one a more prototypical undergoer, the construction that depicts the prototypical constellation is conventionally labelled ‘direct,’ while the construction in which the constellation is reversed is labelled ‘inverse.’

reduplication in movima: a prosodic morphology approach

355

On affirmative main-clause predicates with monosyllabic verb roots, the direct marker is the suffix -na, as in (32a), and the inverse marker is the suffix -kay, as in (32b). (32) a. sal-na=n--as look.for-dr=2--3n.ab ‘You look for it.’ b. sal-kay-a=n--as look.for-inv-lv=2--3n.ab ‘It looks for you.’ When a bivalent verb is combined with further suffixes (e.g. nominalizing -wa, causative -poj, benefactive -kwa, among others; see Haude 2006, 360 ff.), direct and inverse marking can be carried out through reduplication. The examples here illustrate this with the suffix -wa, which derives action nominals, i.e., predicates of complement and adverbial clauses (see Haude 2006, 467ff. and Haude 2011) as well as of negated main clauses. On verbs that consist of a monosyllabic root and an additional suffix, the direct suffix -na is either retained, as in (33a), or it is replaced by word-initial monomoraic reduplication, as in (33b). According to the data and to the speakers, both variants are entirely equivalent. (33) a. n-as sal-na-wa=n--as obl-art.n look.for-dr-nmlz=2--3n.ab ‘when you look for it’ (lit. ‘at your looking for it’) b. n-as sa~sal-wa=n--as obl-art.n dr~look.for-nmlz=2--3n.ab ‘when you look for it’ (lit. ‘at your looking for it’) Reduplicative direct marking only occurs on verbs with monosyllabic roots. When the verb has a longer root, as in (34), or a morphologically complex base, as in (35), the marker -na is retained: (34) n-os purut-na:-wa=∅--us obl-art.n.pst kiss-dr-nmlz=1sg--3m.ab ‘when I kissed him’

356

haude

(35) ka: n-as pek-a-ɬaba-poj-na-wa=i i'neɬ prcl obl-art.n lift-dr-clf.earth-caus-dr-nmlz=3pl art.f:1 ma:ma=n mother.of=2 ‘so that they won’t make our Mother (i.e., Holy Anne) carry earth’ Inverse marking on a verb with a monosyllabic root and an additional affix is obligatorily carried out by bimoraic reduplication, as illustrated in (36); unlike the direct marker, the inverse suffix -kay is never retained. Since monosyllabic verb roots always consist of a heavy syllable (according to the condition that lexical roots be minimally bimoraic), bimoraic reduplication copies the entire root. However, this does not mean that this is a case of root reduplication, as will become apparent in the context of pluri-syllabic roots; also, like Movima reduplication in general, inverse-marking reduplication is based on prosodic and not on morphological rules. (36) n-as sal~sal-wa=n--as obl-art.n inv~look.for-nmlz=2--3n.ab ‘when it looks for you’ (lit. ‘at your being looked for (by) it’) The following text example illustrates the direct- and inverse-marking function of mono- and bimoraic reduplication, respectively, in natural discourse. It shows the two processes with one and the same verb root, yey- ‘want/like.’ The monomoraic reduplication (µ~) in (37a) and (37c) indicates the direct voice, while the bimoraic reduplication (µµ~) in (37b) (causing gemination here, i.e., an audible lengthening of the glide) indicates the inverse voice. (37) a. kas ye~yey-wa=∅ as ney-ni:-wa=∅, majni=∅, neg dr~want-nmlz=1sg art.n here-prc-nmlz=1sg offspring=1sg ‘I don’t want to be here, my child, …’ b. bo toɬ rey kas yey~yey-wa=∅ kinedé=s kwe:ya reas very mod neg inv~want-nmlz=1sg dem.nstd.f=det woman ‘… because that woman (sitting over there) doesn’t like me at all …’ c. che rey inɬa jema' kas ye~yey-wa=∅ and mod pro.1sg also neg dr~want-nmlz=1sg ‘… and I don’t like (her) either.’

reduplication in movima: a prosodic morphology approach

357

On verbs with a multisyllabic base that take additional suffixes, the inverse is marked by word-internal monomoraic reduplication (⟨µ~⟩). Example (38b) shows a verb with a disyllabic root (purut-) whose last CV-element (/ru/) is reduplicated, and (39b) shows the internal reduplication with a morphologically complex base (netawakapoj-). (38) a. purut-kay-a=n kiss-inv-lv=2 ‘(He) kisses you.’ b. n-as pu⟨ru~⟩rut-wa=n obl-art.n kiss⟨~inv⟩-nmlz=2 ‘when (he) kisses you’ (39) a. net-a-waka-poj-kay=∅ drive-dr-cow-caus-inv=1sg ‘(He) makes me drive cattle.’ b. n-os net-a-waka-⟨po~⟩poj-wa=∅ obl-art.n.pst drive-dr-cow-⟨inv~⟩caus-nmlz=1sg ‘when (he) made me drive cattle’ There seems to be a relationship, perhaps based on as yet unknown prosodic rules, between word-initial bimoraic and word-internal reduplication. When the base has a disyllabic root with a light first syllable (like purut- ‘kiss’), inverse marking can also be carried out by initial bimoraic reduplication (resulting in CVCV~). The form in (40), which was volunteered during elicitation, is semantically similar to (38b) above. Examples (41a) and (41b) are yet another illustration of the equivalence of the word form with the internal and the bimoraic initial reduplication. (40) n-os puru~purut-wa=∅--us obl-art.n.pst inv~kiss-nmlz=1sg--3m.ab ‘when he kissed me’ (41) a. n-os de⟨wa~⟩waj-wa=∅--us obl-art.n.pst see⟨inv~⟩-nmlz=1sg--3m.ab ‘when he saw me’

358

haude

b. n-os dewa~dewaj-wa=∅--us obl-art.n.pst inv~see-nmlz=1sg--3m.ab ‘when he saw me’ Inverse marking by a LL-segment, as in (40) and (41b), is less favored than internal reduplication, however: it does not occur in the text corpus, and the forms in the examples are not accepted by all speakers. Still, these data allow an interesting hypothesis. If bimoraic word-initial reduplication can occur not only with monosyllabic but also with longer roots, as in (40) and (41b) above, it can be assumed that historically, bimoraic reduplication was the principal means for forming an inverse base for further suffixation on all verb roots, and that internal reduplication on bases with longer roots is a later development. In the case of disyllabic roots, such as purut- ‘kiss’ in (38) and (40) or dewaj‘see’ in (41) above, phonological reduction of the third syllable may have led to an interpretation as internal reduplication. This possible development is illustrated in (42a–c), with (42b) showing the analysis of the hypothetical intermediate stage with the truncated verb root. (42) a. n-os puru~purut-wa=∅--us obl-art.n.pst inv~kiss-nmlz=1sg--3m.ab b. n-os puru~rut-wa=∅--us obl-art.n.pst inv~trc.kiss-nmlz=1sg--3m.ab c. n-os pu⟨ru~⟩rut-wa=∅--us obl-art.n.pst kiss⟨inv~⟩-nmlz=1sg--3m.ab ‘when I was kissed by him’ Subsequently, the internal reduplication may have been extended to more complex bases, like the one in (39). The three reduplicative allomorphs involved in the direct and inverse marking of morphologically augmented verbal bases are summed up in Table 2 (VBσ = monosyllabic verb base, VBσσ = disyllabic verb base). For the sake of completeness, note that reduplicative direct and inverse marking only occurs with verbs that are marked for direct voice by the suffix -na. On verbs that are marked for direct voice by the base-internal allomorph -a- or ⟨a⟩, reduplication is no option. Here, the direct morpheme is the same as in the underived form, as shown in (43).

reduplication in movima: a prosodic morphology approach table 2

359

Reduplicative direct and inverse marking

Reduplicative Suffix allomorph In environment Gloss Examples -na -kay -kay

µ~ μμ~ ⟨µ~⟩

__VBσ-suffix dr __VBσ(σ)-suffix inv VBσσ … __-suffix inv

(33), (37a,c) (36), (37b), (40), (41b), (42a) (38), (39), (41a), (42c)

(43) a. joy-a-ɬe=n--us b. n-as joy-a-ɬe-wa=n--us go-dr-co=2--3m.ab obl-art.n go-dr-co-nmlz=2--3m.ab ‘You take him with you.’ ‘when you take him with you’ By contrast, the inverse form of these verbs is completely unmarked for voice when combined with an additional suffix, as illustrated in (44b) (see Haude 2011, 160–162). (44) a. joy-ɬe-kay-a=n--us b. n-as joy-ɬe-wa=n--us go-co-inv-lv=2--3m.ab obl-art.n go-co-nmlz=2--3m.ab ‘He takes you with him.’ ‘when he takes you with him’ 4.1.2 Middle Voice: µµ~ and ⟨µ~⟩ Bimoraic reduplication is also found on verbal bases that, unlike the inverse verbs described in 4.1.1, are not combined with an additional suffix. Here again, we see an affinity between initial bimoraic reduplication and internal reduplication, already observed in the previous section. Initial bimoraic reduplication occurs with monosyllabic bases, as in (45), and internal reduplication with longer bases, as in (46). (45) tos~tos md~peel ‘to peel (spontaneously, e.g. skin)’ (46) ɬo⟨ko:~⟩kot boil⟨md~⟩ ‘to boil’ Here, the two reduplication processes are allomorphs of one morpheme, which is best characterized as a marker of middle voice: verbs with this form denote

360

haude

a dynamic event whose single participant is usually in some way affected by it, as can be seen in the above examples. Reduplicative middle marking also occurs on more complex bases, consisting of a verb root and one or more affixed or incorporated elements. This is illustrated in (47) (repeated from (30)): (47) a. jayna mat-lo b. jayna mat-⟨lo:~⟩lo dsc undo-clf.liquid dsc undo-⟨red~⟩clf.liquid ‘It has already melted.’ ‘It melts already.’ Middle marking cannot be confused with inverse marking, carried out by the same reduplicative morpheme, because reduplicative inverse marking only occurs on verbs that are derived by further morphemes, e.g. by the nominalizing suffix -wa (see 4.1.1). On such derived verbs, the middle marking is dropped, as can be seen in the nominalized form in (48) (see also Haude 2006, 357–365 and Haude 2011). (48) ka'de as jayna pat-wa=is, jayna pat~pat nokopa until art.n dsc sprout-nmlz=pl.ab dsc md~sprout like.this jayna dsc ‘When it sprouts (lit.: ‘until its sprouting’), it sprouts like this.’ 4.2 Reduplication on Nouns 4.2.1 Inalienable Possession: ⟨µ~⟩ Internal reduplication marks inalienable possession on nouns, as illustrated by the following examples, involving the nouns kaldo ‘soup,’ chakɬa ‘pole,’ and wa:ka ‘cow.’ The possessor-denoting constituent (in the case of a noun phrase, the article) is attached through internal cliticization (see footnote 9 above). (49) n-os dej-na-wa=y'ɬi is kal⟨do~⟩do=is obl-art.n.pst cook-dr-nmlz=1pl art.n.pl soup⟨inal~⟩=art.pl chara:ye sugar.cane ‘when we cooked the juice of the sugar cane’ (50) daya' di:re n-os chak⟨ɬa~⟩ɬa=os ro:ya dur.nstd leaned obl-art.n.pst pole⟨inal~⟩=art.n.pst house ‘(I) was sitting against the pole of the house.’

reduplication in movima: a prosodic morphology approach

361

(51) bo yey-na=∅ kos wa:⟨ka~⟩ka=i reas want-dr=1sg art.n.pst cow⟨inal~⟩=3pl ‘… because I want their meat (i.e., the meat that is on the bones).’ 4.2.2 Subordinate Predicate Nominals: ⟨µ~⟩ Internal reduplication also derives subordinate predicate nominals (see Haude 2011; subordinate verbal predicates are marked by the suffix -wa, see 4.1.1 above). The subordinate predicate is obligatorily possessed (marked by internal cliticization). Example (52) illustrates this for a noun; in (53) it can be seen that adjectives, which are noun-like in many respects (see Haude 2006, 112–119), behave in the same way. (52) n-os tolkos⟨ya:~⟩ya=∅ obl-art.n.pst girl⟨nmlz.n~⟩=1sg ‘when I was a girl’ (lit. ‘at my former being a girl’) (53) n-os di:ra to⟨chi~⟩chik-a=as Tirinra obl-art.n.pst still small⟨nmlz.n~⟩-lv=art.n Trinidad ‘when (the town) Trinidad was still small’ Since both nominalization and adnominal inalienable possession are marked by internal reduplication, homophonous forms can occur. In the following examples, consider the word bi⟨ ja~⟩jaw-a=is, which is a possessed noun in (54) and a subordinate predicate nominal in (55). However, the context is clear enough that no ambiguities occur.11 Subordinate predicate nominals, for instance, tend to occur in adverbial clauses (i.e., oblique-marked NPs), as can be seen in (55). (54) jayna tet-cheɬ–is, eney, is jayna bi⟨ ja~⟩jaw-a=is dsc scare-r/r–3pl.ab (filler) art.pl dsc old⟨inal~⟩-lv=3pl.ab ‘Then they, er, their parents (lit. ‘their old ones’) got scared.’ (55) jayna n-os bi⟨ ja~⟩jaw-a=is, jayna n-asko dsc obl-art.n old⟨nmlz.n~⟩-lv=3pl.ab dsc obl-pro.n.ab don-wa=is hate-nmlz=3pl.ab ‘Then when they got older, that was when they hated each other.’ 11

Note that the placement of the particle jayna discontinuous outside the NP, as in (54), or within the NP, as in (55), does not distinguish different kinds of NPs.

362

haude

4.2.3 Predicative Possession: Iambic Foot Reduplication Reduplication of the first iambic foot (see Section 3.3) only occurs on nouns, from which it derives intransitive possessive predicates. Reduplication in this function is possibly unique cross-linguistically (L. Stassen, p.c.), but it is highly productive in Movima.12 Example (56) illustrates two instances of the reduplication of the first two light syllables of the base, which in the second case (kweya~) leads to the reduplication of the entire disyllabic word. (56) iti~'itila:kwa jayna is tolkosya, che is itila:kwa jayna poss~man dsc art.pl girl and art.pl man dsc kweya~kwe:ya jema' poss~woman too ‘The girls already had husbands, and the men, too, had wives already.’ In (57), we first see two instances of CVC-reduplication (kor~, pul~) and then the reduplication of an initial LH segment (choran~).13 (57) kos rey buka' kor~korba:ta, kos rey buka' art.n.ab mod dur.mov poss~necktie art.n.ab mod dur.mov pul~pulse:ra, che kos rey choran~chorankwanto […], poss~watch and art.n.ab mod poss~hat kas asko-niwa mowi:maj neg pro.n.ab-nmlz Movima ‘Someone (who) wears a necktie, someone (who) wears a watch, and someone (who) wears a hat […], that one is not a Movima (Indian).’ Example (58) illustrates, again, the reduplication of a LL-segment (pola~) but also, in the second reduplicated form, the copying of a long first syllable (wa:~) of a disyllabic Spanish loan. As was illustrated in (11)–(13) above, the first syllable of disyllabic loans is never shortened and, therefore, counts as inherently heavy.

12

13

There is an alternative construction containing a demonstrative predicate that encodes gender, number, and temporal information of the possessed item (see Haude 2006, 296– 297). Note that possessive predicates, like all Movima content words, can be preceded by an article, resulting in a headless relative clause (see Haude 2006, 298–300).

reduplication in movima: a prosodic morphology approach

363

(58) is karaya:na di' pola~pola:ta, di' wa:~wa:ka art.pl rich.person rel poss~money rel poss~cow ‘(the) rich people who have money, who have cattle’ Possessive predicates can also be formed from nouns marked as inalienable by internal reduplication (see 4.2.1 above), as illustrated in (59) and (60) (containing H- and LL-reduplication, respectively). The resulting predicate characterizes an entity as inalienably possessing another entity, like bones that still have some meat on them or bees which produce their own honey. (59) vel-na=∅ kis nun-'i di' wa:~wa:⟨ka:~⟩ka, look.at-dr=1sg art.pl.ab bone-d rel poss~cow⟨inal~⟩ di' ney, wa:~wa:⟨ka:~⟩ka, isko ɬ way-na=∅ rel here poss~cow⟨inal~⟩ pro.pl.ab 1 take-dr=1sg ‘I look at the bones that have (their) meat, that (are) like this, that have (their) meat, those I take (when buying bones at the butcher’s).’ (60) is chuydi di' chara~charaye-⟨lo:~⟩lo art.pl bee rel poss~syrup-⟨inal~⟩clf.liquid ‘bees that have (their) honey’ Possessive predicates can also be turned into subordinate predicate nominals through internal reduplication (see 4.2.2), as illustrated in (61).14 (61) a. ulkwat maj~majni pro.2sg poss~offspring ‘You have children.’ b. kas jayaw-ɬe=as jeya='ne n-as neg good-neg=art.n state.of=3f obl-art.n maj~maj⟨ni~⟩ni='ne poss~offspring⟨nmlz.n~⟩=3f ‘Her state is not good for having children (i.e., it is difficult for her to give birth).’

14

The formation of a subordinate predicate through reduplication is an exclusive property of nouns and adjectives. Therefore, the fact that subordinate possessive predicates require reduplication shows that morphologically, they are nouns (or adjectives) rather than verbs.

364

haude

4.3 Less Productive Functions of Reduplication Apart from the productive and clearly identifiable functions of reduplication described above, reduplication is also found in environments where its function is less clear and where it is less productive. Many of these cases seem to serve prosodic well-formedness, but some of the often-cited functions of reduplication, like the marking of emphasis, are included here as well. 4.3.1 µµ~ on Transitive Roots in Compounds and with -ni Before certain elements, transitive verb roots are entirely reduplicated.15 Since the cases found so far only involve bimoraic roots (of the structure H or LL), this can be considered a case of bimoraic reduplication. This reduplication is found when the verb root forms a compound together with a bound nominal element, such as a classifier (62) or a truncated noun (63) (see 4.3.5): (62) is~'is-ra red~roast-clf.meat ‘roasted meat’ (63) dan~dan-so red~chew-trc.chicha ‘chewed chicha’ Reduplication is also found when a verb root is combined with the suffix -ni ‘prc,’ which marks intransitive verbs and adjectives (see Haude 2006, 493–495), as in (64) and (65). These derived verbs denote atelic actions, hinting at a possible relationship with the reduplicative middle marker (4.1.2); however, the evidence is not clear since the atelic reading may also be due to the suffix -ni or to the combination of the two morphemes.16

15

16

A reviewer suggested that this might be thought of as being a case of ‘automatic reduplication,’ defined as “reduplication that is obligatory in combination with another affix, and which does not add meaning to the overall construction; the affix and reduplicated matter together are monomorphemic” (Rubino 2005, 114). However, I would not analyze these cases in that way, firstly, because the suffixes with which the reduplication co-occurs can also occur with unreduplicated verb roots and, secondly, because it is not entirely clear that the reduplication does not add any meaning to the overall construction. On nouns, the suffix -ni ‘prc’ indicates the process of ‘becoming/turning into X’ (e.g. rulrul-ni [jaguar-prc] ‘turn into a jaguar’); on adjectives, it does not normally add anything

reduplication in movima: a prosodic morphology approach

365

(64) sal~sal-ni red~look.for-prc ‘to go about searching’ (65) aya~'aya:-ni--i red~wait-prc--3pl ‘They wait and wait.’ Both these types of reduplicated verbs, with a classifier and with the suffix -ni, are infrequent and of limited productivity. 4.3.2 Emphasis As may be expected from the cross-linguistically common iconic character of reduplication, also in Movima, reduplication can be used to mark emphasis. However, according to the data so far, this does not seem to be very productive, and it cannot be connected to one single reduplication process. In (66), the entire word (of the structure LH) is reduplicated; in (67), the copy only consists of a bimoraic segment (LL). (66) paluy~pa:luy red~cold ‘very cold’ (67) dewa~dewaj-na=∅ red~see-dr=1sg ‘I see it well.’ Emphatic reduplication seems to be most productive for verbs containing the irrealis infix ⟨(k)ak⟩, which in negated clauses means ‘there is not’ (see Haude 2006, 438–442). Here, the reduplicated entity is the iambic foot. For a verb starting with two light syllables, as in (68), only these are reduplicated; for a verb starting with a LH sequence, as in (69), this entire sequence is reduplicated, as can be seen from comparing the simple irrealis verb in (69a) and the reduplicated form in (69b).17 Again, this shows that only the prosodic but not the morphemic structure of the base is relevant for reduplication.

17

to the basic meaning (e.g. ta:doy, tadoy-ni ‘sweet’). It is a synchronically unsegmentable element of several atelic intransitive verbs (e.g. ilo:ni ‘walk’). While the irrealis affix is inserted according to prosodic properties of the base and there-

366

haude

(68) a. kas ona-kak-ra:-na=∅ b. kas ona~'ona-kak-ra:-na=∅ neg know-irr-clf.ntr-dr neg red~know-irr-clf.ntr-dr =1sg =1sg ‘I don’t know anything.’ ‘I know absolutely nothing.’ (69) a. kas sal-ak-na=∅ b. kas salak~sal-ak-na=∅ neg look.for-irr-dr=1sg neg red~look.for-irr-dr=1sg ‘I don’t look for anything.’ ‘I look for absolutely nothing.’ 4.3.3 Property-Denoting Words Some property-denoting intransitive verbs contain a reduplicated element. Reduplication does not seem to be a productive device here: the words seem to be lexicalized and are not always fully analyzable. The process involved seems to be iambic foot reduplication. The verb root in (70) has a LH structure and is reduplicated entirely. Of the verb roots in (71) and (72), which have the shape LL, the long, first syllable is apparently counted as heavy since it is copied entirely.18 (70) dewaj~de:waj red~see ‘to be visible’ (71) de:~de:ye red~lie? ‘to be visible’ (72) o:~'o:wa red~? ‘to be audible’ 4.3.4 Reduplication for Prosodic Well-Formedness Reduplication enables certain monosyllabic noun roots to occur alone or as bases for further derivation (see Haude 2006, 195). Here, the reduplication does not have any grammatical function, and most probably it originates historically from some prosodic constraint on content words. It is not a productive process

18

fore clearly an infix (see Haude 2006, 78–82), it can be represented as a suffix in cases like these, where it happens to occur between two segmentable morphemes. There is also independent evidence of the inherent length of the first syllable of some words; see Haude (2006, 56–58).

reduplication in movima: a prosodic morphology approach

367

but is lexically determined: only some noun roots are reduplicated; others take the dummy element -'i when occurring independently; yet others, when consisting of a LL pattern, simply undergo lengthening of the initial syllable (e.g. to:mi ‘water’ in (3)). Most of the monosyllabic roots that undergo reduplication denote inalienably possessed entities, mostly parts of wholes (see Haude 2006, 246 f. for more details). Consider the root duk- ‘back’ in (73). In (73a), the root is reduplicated before being marked as possessed. The attachment of the linking vowel -a creates a complete prosodic word. When no overt possessor is attached (i.e., when the possessor is the first person singular), the resulting word */duduk/ would not correspond to the prosodic constraint that the first two moras of a disyllabic word fall on the first syllable. It is therefore reduplicated once more, as shown in (73b), so that the word becomes trisyllabic.19 (73) a. as du~duk-a=n b. as du~⟨du:~⟩duk=∅ art.n red~back-lv=2 art.n red~⟨red~⟩back=1sg ‘your back’ ‘my back’ If the reduplicated elements here were interpreted as marking inalienable possession, this could only account for one of them, while the other one would not have a grammatical function. However, since inalienable possession is not marked on all body-part terms (see Haude 2006, 249–256), no clear case can be made for possessive marking here. Therefore, the hypothesis is that the word-initial reduplication in (73a) creates the prosodically well-formed base for the cliticization of a possessive marker, which is then the base for the further internal reduplication in (73b). Most monosyllabic roots are also reduplicated when combined with a suffix, e.g. the absolute-state suffix -kwa (see the list in Haude 2006, 553–555). The addition of the suffix must be a later process since the suffixation itself would form a prosodic well-formed word and no further augmentation would be needed. Most of these roots denote inalienably possessed items, as in (74), although there are exceptions, as in (75); the suffix -kwa indicates non- or alienable possession. (74) a. be~bet-a=a red~hide-lv=3n ‘its hide’

19

b. be~bet-kwa red~hide-abs ‘hide, leather’

Note that another expected strategy would be the lengthening of the first syllable.

368

haude

(75) a. la~lak-kwa red~piranha-abs ‘piranha’

b. po~poy-kwa red~animal-abs ‘animal’

Since the reduplication occurs on both possessed and unpossessed bases, its function on these roots is not an indicator of possession or any other grammatical category but is probably rather a prosodic device. Still, it is not a productive process that automatically applies to noun roots: there are also words with -kwa whose monosyllabic roots are not reduplicated, as illustrated in (76). (In (76a), the root is lengthened to comply with the mora requirement.) Thus, words like those in (74) and (75) are probably the lexicalized result of a prosodic requirement that was active at an earlier stage of the language. (76) a. si:ɬ-a=a b. siɬ-kwa hole-lv=3n hole-abs ‘its cave (of an animal)’ ‘hole, cave’

c. *si~siɬ-kwa red~hole-abs (‘hole, cave’)

Another case where reduplication does not have any other function than that of forming a well-formed word involves two ‘defective’ nouns (see (4) above): the words for ‘mother’ and ‘father’ undergo reduplication before they can be marked for possession.20 In (77) they are given with the presential article (i'nes for feminine and us for masculine) and the cliticized pronouns of second and third person. The base form only occurs with a first-person possessor. (77) i'nes ma'a ‘my mother’ i'nes ma:ma=n ‘your mother’ i'nes ma:ma=us ‘his mother’ (78) us us us

pa'a pa:pa=n pa:pa=us

‘my father’ ‘your father’ ‘his father’

4.3.5 Truncated Loans Many nouns undergo truncation when participating in compounding or incorporation (Haude 2006, 212ff.): only the final element of the noun is used in these

20

Other defective nouns are augmented with the syllable -ya replacing the AV-segment and the lengthening of the first syllable (e.g. jeya=n ‘your state of being’).

reduplication in movima: a prosodic morphology approach

369

processes. The truncation of most native nouns involves only the last syllable of the noun, as illustrated in (79). (79) cha:do ‘plate’ → loj-a:-do ‘wash plates’ (wash-dr-trc.plate) mo'incho ‘chivé’ → il-a:-cho ‘spread chivé’ (spread-dr-trc.chivé) pokso ‘chicha’ → kwajtak-so ‘maize beer’ (maize-trc.chicha) Spanish loans behave differently: when they are multisyllabic, the truncation preserves the last two syllables, as illustrated in (80). (80) sapa:to ‘shoe’ → dok-a-pa:to ‘put shoes on’ (put_on-dr-trc.shoe) aro:so ‘rice’ → duk-a-ro:so ‘grind rice’ (grind-dr-trc.rice) asu:ka ‘sugar’ → tavoj-su:ka ‘white sugar’ (white-trc.sugar) Strikingly, as first noted by Grinevald (2002), with disyllabic loans, the requirement of truncating two syllables of a loan leads to reduplication of the last syllable, as illustrated in (81). It should be noted, however, that this process is not productive anymore. It involves loans that were probably borrowed rather early, possibly along with objects that were brought by the Spanish-speaking missionaries. More recent loans (e.g. pi:la ‘battery,’ elade:ra ‘refrigerator’) are not truncated but incorporated as entire words (see Haude 2006, 216). (81) si:ya ‘chair’ → sotak-ya:ya ‘one chair’ (one-trc.chair) me:sa ‘table’ → bay-a-sa:sa ‘to knock on the table’ (knock-dr-trc.table) pa:ko ‘dog’ → sot-ko:ko ‘another dog’ (other-trc.dog)

5

Conclusion

To sum up, Movima has several productive reduplication processes: wordinitial monomoraic, word-initial bimoraic, word-internal monomoraic, and word-initial foot reduplication. These different types express different grammatical functions, which include the following: direct and inverse voice marking on verbs, middle marking on particular verb bases, inalienable possession on nouns, marking of predicate nominals in subordinate clauses, and derivation of possessive predicates. In addition, reduplication serves prosodic functions, e.g. it can enable a root to occur alone or as a base for further derivation, a process which does not seem to be fully productive anymore. Likewise, emphasis can be expressed by

370

haude

reduplication, but this is not productive either and seems to be a marginal function. At some earlier stage in the history of the language, reduplication was presumably used as a device to mark Spanish borrowings in compounds or incorporating verbs. It is striking that there is hardly any direct evidence of iconicity (reflected by markedness and weight/length of an expression) in Movima reduplication, a phenomenon that is often found in other languages where reduplication marks habitual, iterative or durative aspect on verbs or plural number on nouns (see e.g. Rubino 2005, 115). Furthermore, it might be asked why Movima expresses the aforementioned grammatical functions by reduplication instead of by ‘ordinary affixes’ (i.e., segmental material), and why particular functions are expressed by particular forms of reduplication. This question cannot be answered here, but some tentative hints can be given. For instance, the foot reduplication marking possessive predicates (4.2.3) results in a word describing a state, that of possessing something. An iconic interpretation would be, firstly, that the durative aspect of this state is iconically reflected by the reduplication, and secondly, that the state of possessing something might be considered more complex than that of being something (which accounts for the contrasting non-reduplicated form of simple nouns, which designate an entity).21 An iconically-based explanation might also be proposed for the reduplication of subordinate predicate nominals, which denote states (see Haude 2011), for the middle-marking reduplication (4.1.2), and for the reduplication of verb roots in compounds with classifiers or with the suffix -ni (4.3.1), all of which create atelic verbs; however, this account has its limitations since not all atelic verbs contain reduplication. It would be more complicated to explain the direct- and inverse-marking function of reduplication in nominalized verbs, where reduplication replaces the suffixes -na and -kay, respectively. While it does not distinguish between a more and a less marked member of an opposition, maybe the reduplication is related here to the changed rhythmic pattern of verbs with further suffixes: the monomoraic reduplication of direct verbs creates a base with the pattern LH (/sa.sal-/ ‘looking for X,’ (33b)) instead of HL (/sal.na-/, (32a)), and a word-initial light syllable may seem prosodically more convenient for further suffixation (cf. the occurrence of word-initial monomoraic reduplication on noun roots, 4.3.4). The bimoraic reduplication marking inverse verbs (/sal.sal-/ ‘being looked for,’ 21

This point is inspired by Stolz (2007), who suggests that less clearly iconic functions of reduplication may still be understood as iconic in a looser sense; thus, reduplication may simply be a way of indicating the conceptually more marked category in an opposition, even where there is no obvious iconic motivation deriving from the meaning.

reduplication in movima: a prosodic morphology approach

371

(36)) is a device which retains the prosodic pattern HH of the underived verb (/sal.kaj/, (32b)), thereby creating the same prosodic contrast with the direct form (note, however, that the attachment of the linking vowel on the underived verb creates a light syllable). An account in terms of markedness, however, which would consider the direct form as less marked than the inverse, cannot be used as an explanation since on certain verbal bases, the inverse is unmarked (see (45b) above), whereas the direct form is always overtly marked without exception. Thus, while some traditional explanations of reduplication may come to mind, there seems to be no satisfactory explanation of the extensive and multifunctional use Movima makes of reduplication. Rather, Movima seems to confirm the view expressed by Marantz (1994, 3487) that “[a]s far as morphological function is concerned […], reduplication is indistinguishable from other forms of affixation”; we are simply dealing here with a language that makes more use of prosodic morphology than others, employing metrics rather than substantial morphology as a word-forming device. As can be shown, due to the different morphological environments in which they appear, the reduplicative affixes of Movima mark grammatical functions in the same non-ambiguous way in which this is done by phonologically specified affixes. That the correspondence between form and function can be blurred by homophony is well known also in the case of such ‘normal’ affixes. The answer to the question why Movima makes so much more use of reduplication as a grammatical device than other languages probably lies in the role metrics play in the word formation of this language, a topic on which more research is needed.

Acknowledgments I thank Hein van der Voort, Gale Goodwin Gómez, and two anonymous reviewers for their helpful comments on a previous version of this paper. The article was prepared within the Movima documentation project funded by the Volkswagen Foundation (DOBES).

References Broselow, Ellen, and John J. McCarthy. 1983. A theory of internal reduplication. The Linguistic Review 3: 25–88. Goldsmith, John A., ed. 1995. The handbook of phonological theory. Oxford: Blackwell. Grinevald, Colette. 2002. Nominal classification in Movima. In Current studies on South

372

haude

American languages, ed. Mily Crevels, Simon van de Kerke, Sérgio Meira and Hein van der Voort, 216–239. [Indigenous Languages of Latin America (ILLA) 3]. Leiden: CNWS. Haude, Katharina. 2006. A grammar of Movima. PhD diss., Radboud University Nijmegen. http://webdoc.ubn.ru.nl/mono/h/haude_k/gramofmo.pdf (accessed 27 November 2013). . 2009. Hierarchical alignment in Movima. International Journal of American Linguistics 75/4: 513–532. . 2011. Referring to states and events: Subordination in Movima. In Subordination in South American languages, ed. Rik van Gijn, Katharina Haude and Pieter Muysken, 141–168. Amsterdam/Philadelphia: John Benjamins. Judy, Robert, and Judith Judy. 1967. Movima. In Bolivian Indian grammars, Vol. 2, ed. Esther Matteson, 353–408. Oklahoma: SIL Publications. Kager, René. 1995. The metrical theory of word stress. In Goldsmith (ed.), 367–443. Marantz, Alec. 1994. Reduplication. In The Encyclopedia of language and linguistics, ed. Ronald E. Asher and J.M.Y. Simpson, 3486–3487. Oxford: Pergamon Press. McCarthy, John, and Alan S. Prince. 1995. Prosodic morphology. In Goldsmith (ed.), 318–366. Rubino, Carl. 2005. Reduplication. In The world atlas of language structures, ed. Martin Haspelmath, Matthew S. Dryer, David Gil and Bernard Comrie, 114–117. Oxford: Oxford University Press. Stolz, Thomas. 2007. Re: duplication. Iconic vs counter-iconic principles (and their areal correlates). In Europe and the Mediterranean as linguistic areas. Convergencies from a historical and typological perspective, ed. Paolo Ramat and Elisa Roma, 317– 350. Amsterdam/Philadelphia: John Benjamins.

Symbols and Abbreviations ~ = ⟨⟩ 1 2 3 ab abs art caus

reduplication internal cliticization external cliticization infixation first person second person third person absential absolute state article causative

clf co d det dem dr dsc dur f inal ins

classifier co-participant dummy affix determiner demonstrative direct voice discontinuous durative feminine inalienable instrumental

reduplication in movima: a prosodic morphology approach

irr itn lv m md mod mov n neg nmlz nmlz.n nstd ntr obl

irrealis intentional linking vowel masculine middle voice modal moving neuter negation nominalization nominalization of noun nonstanding neutral oblique

pl poss prc prcl pro pseu pst reas red rel res r/r sg trc

plural possessive process preclusive free pronoun pseudo past reason reduplication relativizer resultative reflexive/reciprocal singular truncated element

373

chapter 14

When Vowel Deletion Blurs Reduplication in Mojeño Trinitario Françoise Rose

This paper discusses the process of verbal reduplication in the Trinitario dialect of Mojeño, an Arawak language spoken in the Bolivian lowlands. Trinitario displays partial reduplication of the verb root. Reduplication of the final syllable of the root expresses repetition or attenuation of the event expressed by the verb. This paper revises the analysis of Trinitario reduplication patterns as presented in Gill (1957), in light of additional first-hand data and the recent identification of a prosodic vowel deletion rule. It points to a phonologicallyconditioned allomorph of the basic reduplication pattern and to a second reduplication pattern implying a double copy of the final syllable of the root. The major interest of the Trinitario data is to demonstrate the great complexity of the morphology-prosody interface. The regular processes of vowel deletion and compensatory lengthening indeed affect the base only after the copy is derived, which results in a surface pattern where the copy and the base do not look alike.

1

Introduction

This paper discusses the process of reduplication in the Trinitario dialect of Mojeño, an Arawak language spoken in the Bolivian lowlands. Trinitario displays partial reduplication of the verb root, expressing repetition or attenuation of the event expressed by the verb. The Mojeño language (also called Moxo) consists of 4 dialects: Trinitario, Ignaciano, Loretano and Javeriano. The total Mojo population is over 32,000, with around 3,220 speakers of Trinitario and Ignaciano (Crevels and Muysken 2009). Trinitario and Ignaciano are endangered dialects, while Javeriano is moribund with approximately five speakers with varying levels of proficiency. Loretano is extinct. There is a grammar of the Mojeño language as spoken in the 17th century, which is a very valuable tool (Marbán 1701). There is no linguistic description of Javeriano and Loretano. Both Ignaciano and Trinitario have been previously described (Olza Zubiri, et al. 2002 and Gill 1957, respectively). These

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_015

376

rose

two dialects are, in general, very similar except for the pervasive vowel deletion process affecting Trinitario (Rose 2008; 2011; to appear). In this paper, I revise the analysis of Trinitario reduplication patterns as presented in Gill (1957, 117–118), in light of the prosodic vowel deletion rule. The analysis presented in this paper is based on both the author’s field data and examples given in Gill’s presentation of reduplication. The field data on Trinitario was collected by the author in Trinidad and San Lorenzo de Moxos between 2005 and 2010. The resulting database consists of 30 texts consisting of approximately 2,000 sentences. Several hundred elicited sentences were added to this database. This data and Gill’s work generally concur on major patterns but complement each other as far as minor patterns are concerned. This will be discussed later. All examples, mine and Gill’s, are phonologically transcribed using IPA.1 After some background information on Trinitario (section 2), the functions of reduplication are presented (section 3). The remainder of the paper analyzes the details of the phonological patterns of reduplication (section 4). It first shows the basic pattern of reduplication. The influence of reduplication on stem morphology, the effect of vowel deletion, and minimality constraints are then investigated. Finally, a second pattern of reduplication is presented, where double-copy occurs. The analysis presented in this paper allows for a revision of Gill’s analysis. The analysis suggested in this paper is more coherent with the overall system of the language insofar as it provides a single analysis of both stem and affix allomorphy and multiple reduplication processes. Section 5 concludes with a discussion of the ramifications of this analysis for theoretical debates concerning the morphology-prosody interface and for the typology of reduplication.

2

Background Information on Mojeño Trinitario

2.1 Phonology The phoneme inventory of Mojeño Trinitario is presented in the following two tables.

1 My corpus is transcribed in the official alphabet (Fabricano et al. 2003), which differs from the one used by Gill. Since this paper deals with phonology, IPA is used.

when vowel deletion blurs reduplication in mojeño trinitario table 1

377

Consonants

Bilabial Alveolar Palatal Velar Glottal stops nasals fricatives affricates approximants and tap

table 2

p m β w

t n s ts ɾ

(c) ɲ ç tʃ j

k

ʔ

x

Vowels

Front vowels Central vowels Back vowels high i, i: mid e, e: low

ə͡e, ə͡e: a, a:

u, u: o, o:

Trinitario syllable structure is either of the (C)(C)V(C) type or of the (C)(C)V: type. Complex onsets and null onsets are only found word-initially. Heavy syllables (with a coda or a long vowel) are only found word-internally. Stress normally falls on the penultimate syllable. Only a minority of CVCVCV roots are stressed on the antepenultimate syllable. Most instances of suffixation do not alter the principle of stress placement, that is to say stress automatically shifts to the right, to the penultimate syllable. Stress placement will be discussed in sections 2.3 and 4.1. Roots generally consist of at least two syllables, to which prefixes and suffixes are added. Very few grammatical words consist of only one syllable (see section 4.4 for a discussion of minimality requirements). 2.2 Vowel Deletion In this paper, “word” refers to the prosodic unit which serves as the domain of stress, vowel deletion, and morphophonological rules. In Trinitario, vowel deletion is quasi-systematic on stems of three syllables or more.2 It generally

2 Some irregularly-stressed roots resist vowel deletion. The initial vowel of disyllabic roots is not deleted when the root is word-initial.

378

rose

affects the first vowel of a prosodic word, and then every odd-numbered vowel to the right, except the final one. If the first syllable of a word is a member of a subset of person prefixes (ma-, ɲi-, ta- or na-), the vowel of this syllable is maintained and is transparent to the vowel deletion process. Vowel deletion then affects every odd-numbered syllable to its right, except the final syllable.3 CV1.CV2.CV3.CV4.CVFIN ↓ ↓ ∅ ∅ vowel deletion C CV2 C CV4 CVFIN CCV2 C.CV4 .CVFIN figure 1

resyllabification

Rhythmic vowel deletion

This process is probably the result of historically rhythmic stress on every even-numbered syllable. In the synchronic system, stress normally falls on the penultimate syllable of the word. There are a few lexical exceptions to this rule, and certain suffixes lie outside the stress domain, i.e. they are extrametrical. As a result of vowel deletion, almost all roots and affixes have two forms. According to their position within the prosodic word, they lose different vowels.4 Examples are given in Table 3. table 3

Vowel deletion in Trinitario

Underlying representation5 Surface representation Translation pokure su-pokure nu-tʃokojo nu-ko-tʃokojo ti-ko-xuma ti-a-koxuma

'pkure 'spokre 'ntʃokjo nkotʃ 'kojo 'tkoxma tak'xuma

canoe her canoe I am close I got close he/she/it is sick may he/she/it be sick

3 When ma- is prefixed to a root with initial /e/ or /i/, this sequence is realized as /mwe/. 4 Gill does not mention vowel deletion. In his work, there are two different classes of roots calling for different allomorphs of the suffixes. Once suffixed, roots change class and take different suffix allomorphs. Moreover, prefixes themselves also condition the root allomorphy. 5 V represents an unspecified underlying vowel.

when vowel deletion blurs reduplication in mojeño trinitario

379

In certain contexts, vowel deletion conditions compensatory vowel lengthening. Certain consonants, such as /r/, are not permitted in coda position. Vowel deletion strands the underlying consonant preceding this vowel. If this consonant is not permitted in coda position, it is deleted and the preceding vowel is lengthened. This process is schematized in figure 2 and exemplified with the sequence /riro/ realized /ri:/ in /poriropa/ ‘needle’ (segments in parentheses are not directly relevant here). Table 4 illustrates cases where a vowel following /r/ is dropped. This shows that /r/ is deleted and a process of compensatory lengthening occurs.

1) V2 → ∅ 2) r → ∅ V1 → V1:

CV1rV2 (po)riro(pa) CV1r (p)rir(pa) vowel deletion consonant deletion CV: (p)ri:(pa) compensatory lengthening

figure 2

Compensatory lengthening in Trinitario

table 4

Compensatory lengthening in Trinitario

Underlying representation Surface representation Translation poriropa su-poriropa nu-xuru-ko-riʔi nu-korVto-ko-riʔi

'pri:pa spo:'ropa nxu:'ko: ʔi nko:tok'riʔi

needle her needle I grow I grab

It is important to note that vowel deletion and long vowels are a recent development of Trinitario. They are not found in 17th century Mojeño (Marbán 1701) nor in the Ignaciano dialect (Olza Zubiri et al., 2002). 2.3 The Morphological Structure of the Verb Mojeño Trinitario is a polysynthetic language. The verb morphology is rich both at the derivational level (stem-internal) and at the inflectional level (stemexternal). As a consequence, the verb structure is morphologically complex (Figure 3), with many prefixes and suffixes. The root itself generally consists of two or three syllables.

380

rose

S-irr- cau-root~red-clf/n-plurac-act/cau/apl -irr-O/mid-pl-tam-rpt-d figure 3

Structure of the Mojeño verb

Reduplication occurs in the slot indicated in figure 3 above. The stem, delimited by the central box in figure 3, contains one prefix slot and several suffix slots. The presence of reduplication affects two of these suffix slots, clf and act, in a manner that will be specified later on. These two suffix slots are now presented. The active suffix (-ko ~ -ʧo ~ -ʔo) is used to mark active stems. Table 5 illustrates the behavior of the active suffix with respect to the two different classes of active stems. Active verbs of the major class always carry this suffix, e.g. xaɲo-ko. Active verbs of the minor class (all with root-final /o/, e.g. xikpo) obligatorily take the active suffix only when carrying a stem-internal suffix, like the -ri. Otherwise, when carrying no suffix or a stem-external suffix (such as -nu, 1st singular object), they do not take the active suffix. table 5

The active suffix on the two classes of active stems

Root

With externalWith no suffix stem suffix

xaɲo ‘watch’

n-'xaɲo-ko ‘I watch’

xikpo n-'xikpo ‘answer’ ‘I answer’

With internalstem suffix

c-xaɲo-k(o)-nu c-xaɲo-ri-ko ‘he/she/it watches me’ ‘he/she/it always watches’ c-xikpo-nu c-xikpo-ri-ko ‘he/she/it answers me’ ‘he/she/it always answers’

Second, some classifiers are found within the verb stem. On stative or active intransitive verbs, they refer to the subject, as shown in (1) and (2) respectively. On transitive verbs, they refer to the object as shown in (3). (1)

eto 'cxitumo.6 eto ti-'xitu-mo pro3nh 3-thin-clf:thin.flexible ‘It is thin (for example, some fabric).’

6 The examples in this paper are organized in four lines: the first line represents the phonological output, the second one reflects its parsing into morphemes given in their underlying

when vowel deletion blurs reduplication in mojeño trinitario

381

(2) tewarasko ti-ewara-si-ko 3-break-clf:round-act ‘It broke (for example, a round stone).’ (3) njerepareko 'motexi nu-jere-pa-re-ko 'motexi 1sg-carry-clf:grain-plurac-act earth ‘I am carrying earth.’ Stem-internal suffixes (red, classifiers and pluractional) share several properties. First, they make the active suffix obligatory on the active verbs (Compare (4) and (5)). Second, stress normally falls on the penultimate syllable of the Trinitario word (6). However, red and classifiers are normally excluded from the stress domain. When the penultimate syllable of the word is a classifier or red, stress therefore falls on the antepenultimate (7).7 Stress can fall on a classifier (9) only on irregular stems with antepenultimate stress such as (8). Third, stem-internal suffixes are not affected by vowel deletion, as illustrated in (10) where the /e/ or the classifier -me is expected to be deleted.8 (4) 'nomo nu-omo 1sg-carry ‘I am carrying it.’ (5) nomxiko nu-omo-xi-ko 1sg-carry-clf:bulk-act ‘I am carrying it (vegetal, clothing …).’

form, the third one represents the gloss, and the last one represents the free translation. The pervasive vowel deletion process (section 4.3) and the high number of morphophonological rules make the line with underlying forms necessary to recognize morphemes from one example to the other. 7 Stress is not systematically noted in this paper. 8 Only classifiers with an /i/ vowel are prone to vowel deletion.

382

rose

(6) 'cu:na ti-urina 3-be.beautiful ‘It is beautiful.’ (7) cu:'napaʔi ti-urina-paʔi 3-be.beautiful-clf:ground ‘The ground is clean.’ (8) n' juwako nu-juwa-ko 1-grind-act ‘I grind.’ (9) njuwa'meko nu-juwa-me-ko 1-grind-clf:flat.hard-act ‘I grind slices of dry meat.’ (10) ttuprimeko ti-tupuri-me-ko 3-middle-clf:flat.hard-act ‘in the middle of something flat and hard (a bank or a back, for example)’

3

The Functions of Reduplication

Verb reduplication in Trinitario is a purely morphological process that does not influence the syntactic structure of the clause.9 It never affects the argument structure of the verb, whether it is transitive or intransitive, active or not. It is purely derivational, modifying the meaning of each verb root in a regular way. In this section, the functions of reduplication as defined by Gill are presented with some additional commentary. Reduplication conveys an attenuative or frequentative meaning in Trinitario (Gill 1957, 217). 9 Olza Zubiri, et al. (2002) mention that reduplication additionally occurs on nouns, adjectives (analyzed as verbs in my work), question words, and quantifiers in the Ignaciano dialect. Concerning the Trinitario dialect, this observation is not mentioned in Gill’s (1957) work. I found reduplication of these categories to be very marginal in my corpus.

when vowel deletion blurs reduplication in mojeño trinitario

383

3.1 Attenuation Reduplication may attenuate the meaning of the verb, with either a deintensifying or an approximative reading. In my corpus, this sense is most often found with stative predicates like the verb ixre ‘be hot’ in (11) and the adjective riko ‘rich’ in (12). It can, however, also be found on active verbs such as sopo ‘believe’ (13) and it(u)-ko ‘know/learn’ (14). (11) tixrerexi ti-ixVre~re-xi 3-be.hot~red-clf:bulk ‘It is warm (medium hot).’ (12) rikkokoʔo10 riko~koko-ʔo rich~red211-act ‘He is sort of rich.’ (13) psoppoxkonu pi-sopo~po-xi-ko-nu 2sg-believe~red-clf:bulk-act-1sg ‘You half-believe me.’ (14) mamuire 'mwettutupiko to βetʃxiri:wo ma-muire 'ma-itu~tutu-pi-ko to βi-etʃxiri:wo 3m-too 3m-know~red2-clf:long.flexible-act art.nh 1pl-language ‘He too knows a bit our language.’ 3.2 Repetition: Event-Internal Plurality The second meaning of reduplication in Mojeño is repetition. In my corpus, the repetitive meaning is found mostly on active verbs. (15) najus'titixiko na-jusVti~ti-xi-ko 3pl-cut~red-clf:bulk-act ‘They chopped it.’

10 11

The initial vowel of disyllabic roots is not deleted when it is word-initial. The double-copy pattern of reduplication shown in this example is discussed in section 4.5.

384

rose

(16) ene taeβiβik'riʔi to tektikwo. ene ta-eβi~βi-ko-riʔi to tektikwo and 3nh-swing~red-act-ass art.nh wind ‘The wind makes it (a scarecrow) move.’ Reduplication with a repetitive meaning does not entail argument plurality. Examples (15) and (16) show singular subjects and singular objects with reduplicated verbs. In fact, Trinitario reduplication denotes exclusively a repetitive action within the same event frame (event-internal plurality (Cusic 1981)). It does not express the repetition of the same event at different times, with empty intervals (event-external plurality). Therefore, reduplication more specifically expresses aspect rather than verbal plurality, also called pluractionality (Newman 1990, Corbett 2000). Pluractional markers express the plurality of participants, time and/or location. In fact, a distinct pluractional marker exists in Trinitario, distinct from reduplication: the -ri suffix and its allomorphs. It most often turns the action into an activity (à la Vendler 1967), as illustrated in the example below. The pluractional verb can either affect a plurality of objects at a specific time, or be extended in time with either a durative or a habitual reading. (17) ma pijaeni 'mwetko tiprux'riko ma pi-ija-ini 'ma-it-ko ti-iprux-ri-ko art.m 2sg-father-pas 3m-know-act 3-cure-plurac-act ‘Your late father knew how to cure (people).’ 3.3 Attenuation and Repetition Gill states that both meanings, attenuation and repetition, can be combined in a single example, as in (18) (Gill 1957, 217). My corpus does not illustrate this. However, the translation of the reduplication process is never self-evident. The determination of the core meaning of each occurrence of reduplication is not an easy task. (18) esu tkox'mamaxi. esu ti-ko-xuma~ma-xi pro.f 3-vz-sickness~red-clf:bulk ‘She is sickly (not real sick but is often rather sick).’ (Gill 1957, 217) The two meanings conveyed by reduplication, attenuation and repetition, seem a priori divergent. Attenuation expresses a lesser degree, while repetition expresses a greater number of realizations of the event. Nevertheless, they both express a dilution of the boundaries of the event expressed by the verb: ‘sickly’

when vowel deletion blurs reduplication in mojeño trinitario

385

is a less clearly defined state than ‘sick,’ and ‘chop’ is a less clearly bounded event than ‘cut’ (being less punctual in that case).12 Attenuation and repetition are typical meanings of reduplication (Rubino 2005, 19).

4

The Forms of Reduplication

This section investigates the forms of Trinitario verbal reduplication, i.e. its phonological structure and its morphological status. The item (generally a root) used as a model for reduplication is called the base of reduplication in this paper. The item resulting from reduplication is called the copy. The Trinitario reduplication process copies the last syllable of the verbal root. The copy immediately follows the root, as illustrated in (19) and (20). The copy is part of the verbal stem, as schematized in Figure 4. (19) tmopkukuʔi'ritʃʔo ti-mopVku~ku-ʔi-ritʃoʔo 3-be.dark~red-clf:atmosphere-still ‘It is still half dark.’ (20) tema'rereko13 ti-ema're~re-ko 3-be.lightning~red-act ‘There is a sequence of lightning.’

cau-root~red-clf/n-plurac-act/cau/apl figure 4

Mojeño verbal stem structure

The remainder of this section first discusses the position and status of the copy (4.1) and discusses its combination with the classifier -xi (4.2). In section 4.3, I 12

13

For some discussion about these apparently divergent though related meanings of reduplication, the reader could refer to Queixalós (2002), who discusses this under the term of “distensivité.” I cannot explain why this root is not affected by vowel deletion. An anonymous reviewer suggests that the unexpected failure of the /a/ to delete in (20) may be due to its being a low vowel, which would by virtue of its greater sonority be less likely to delete than high or mid vowels (see note 8). This still needs to be checked in a larger corpus.

386

rose

discuss the effect of vowel deletion on the basic pattern of reduplication. The issue of reduplication of monosyllabic roots is discussed in section 4.4. Finally, a second pattern of reduplication is presented (4.5). 4.1 The Copy as a Stem-Internal Suffix Since the root-final syllable and its copy are identical, there are logically two possible analyses: the copy either follows the base or is infixed before the final syllable. The first analysis is more economical since the copy would be considered a suffix immediately adjacent to the root rather than an infix for which a slot within the root would be added. This analysis is supported by the morphology of reduplicated stems. The copy shows a behavior typical of stem-internal suffixes (see 2.3). First, it causes the active suffix to be obligatory on all active stems, just as stem-internal suffixes do. This is the case with the verbs amo ‘swell up’ and soko ‘defecate,’ which do not take the active suffix in the absence of a stem-internal suffix. In (21), the presence of the active suffix -ko can only be attributed to the reduplication process. In (22), it can be attributed to the classifier -xi ‘bulk’ as well. See also example (5) where the active suffix is induced by a classifier on a non-reduplicated root. (21) βjammoʔojore βi-amo~mo-ʔo-jore 1pl-swell.up~red-act-fut ‘We are going to swell up.’ (22) tsokkox'kojre ti-soko~ko-xi-ko-jore 3-defecate~red-clf:bulk-act-fut ‘(S)he has diarrhea.’ Second, on regular roots, the copy is transparent for stress. The position of stress is calculated as if the copy were absent (23). Like other stem-internal suffixes, the copy can be stressed on irregularly stressed roots, like kémtone in (24). (23) tema'rereko14 ti-ema're~re-ko 3-be.lightning~red-act ‘There is a sequence of lightnings.’

14

I cannot explain why this root is not affected by vowel deletion.

when vowel deletion blurs reduplication in mojeño trinitario

387

(24) nkemton'nekpo nu-ko-emotone~ne-ko-po 1sg-vz-work~red-act-perf ‘I worked there and there.’ Third, the copy always conserves its vowel. Thus, in example (25), the first and third vowels are deleted. The fifth vowel, while expected to be deleted, is maintained because it is the vowel of the copy. We will return to this below. (25) tixrerexi ti-ixVre~re-xi 3-be.hot~red-clf:bulk ‘It is warm (medium hot).’ These three properties of the copy argue strongly that it is a stem-internal suffix. 4.2 Reduplication and -xi The reduplicated stem is very frequently followed by the classifier -xi ~ -x ‘bulk,’ inserted after the copy (and before the active suffix on active verbs). It is not obligatory, but most occurrences of reduplication show it, as in (26) and (27). (26) tkox'mamaxi ti-ko-xuma~ma-xi 3-vz-sickness~red-clf:bulk ‘(S)he is sickly.’ (27) najus'titixiko na-jusV 'ti~ti-xi-ko 3pl-cut~red-clf:bulk-act ‘They chopped it.’ -xi ~ -x is one of the classifiers described in 2.3. Its semantics are not easy to define. Gill offers the following definition: “-xi denotes usually bulky soft objects (grass, leaves, small branches, fresh meat, any cloth material, pillow; also mud, garbage; also clouds)” (Gill 1957, 84). Examples are given below. (28) nsu:xiko nu-suCV-xi-ko 1sg-fry-clf:bulk-act ‘I fry (fresh meat).’ (Gill 1957, 84)

388

rose

(29) nkoʔtʃaxi (Gill 1957, 84) nu-koʔVtʃa-xi 1sg-be.dirty-clf:bulk ‘I am dirty (my clothes).’ The use of -xi along with reduplication departs from Gill’s definition given above. In some examples, like the ones below, there is no bulky object it could refer to in an obvious way. (30) esu seno 'somuire cu:'nanaxi esu seno 'su-omuire ti-uri'na~na-xi pro3f lady 3f-also 3-be.beautiful~red-clf:bulk ‘This lady is also beautiful.’ (31) mwestataxiko to wije ma-esta~ta-xi-ko to wije 3m-whisp~red-clf:bulk-act art.nh ox ‘He often hits the ox.’ (Gill 1957, 218) For this reason, Gill considers -xi to be “part of the reduplication morpheme,” but still considers it an object classifier in some examples, specifically when it retains its full meaning. I consider it a morpheme separate from the copy. On the morphological level, it fills the classifier slot like other classifiers do (see (32) and (33)). Moreover, there are no examples of a reduplicated stem with -xi that combines with some other classifier. (32) tixreremo ti-ixre~re-mo 3-be.hot~red-clf:flat.flexible ‘It is medium hot (of something flat and flexible, like a pancake).’ (elicited) (33) ɲeβe'roroçeʔo ɲi-eβe'ro~ro-çe-ʔo 3m-lick~red-clf:bone-act ‘He licked the bones.’ Yet the meaning of the -xi classifier on reduplicated verbal roots is often bleached. Its use has become generalized in such a way that it resembles the fixed segmentism that occurs with reduplication in some languages. Alderete, et al. (1999) distinguish two types of fixed segmentism: phonological fixed seg-

when vowel deletion blurs reduplication in mojeño trinitario

389

mentism occurs when the presence of a default segment is phonologically motivated, and morphological fixed segmentism occurs when an affix is attached to the base concomitant with reduplication. The Trinitario -xi could be analyzed as a type of optional morphological fixed segmentism since it is not phonologically motivated. 4.3 Taking Vowel Deletion into Account A surface variant of the basic reduplication pattern occurs in Trinitario. This is not explained in Gill’s work. The process of vowel deletion may have the effect of deleting the final vowel of the root in the reduplicated form of the verb, as illustrated in (34) and (35). This occurs if the root-final vowel is in an odd-numbered syllable of the underlying structure of the word. Since the copy is suffixed to the root (consonant-final in those cases), reduplication surfaces in these cases as a sequence of two identical consonants. As a result, the identity between the copy and the base is not transparent in the phonological output. (34) viammoʔojre vi-amo~mo-ʔo-jore 1pl-swell~red-act-fut ‘We are going to swell.’ (35) tjuxpanniko ti-juxVpani~ni-ko 3-ask.favor~red-act ‘(S)he keeps asking for favors.’ This allomorphy does not constitute a different reduplication pattern. The alternation is predictable. This sequence of identical consonants cannot be analyzed as simple consonant reduplication. (36a) exemplifies the syllabic reduplication pattern on the root sopo ‘believe.’ /po/ is copied, and vowel deletion normally affects the first, third, and fifth syllables of the word. The rootfinal /o/ is deleted, being the third vowel of the word. As a result, the two /p/ consonants are adjacent. The sequence of identical consonants is created by the deletion of the vowel between the two successive consonants in the rootfinal syllable and the copy, these two being similar due to the reduplication process. (36b) illustrates the single-consonant reduplication analysis, applied to the same root. If /p/ only was doubled, the initial /i/, the root-final /o/ and the /o/ of -ko should be deleted, since vowel deletion affects the vowels of odd-numbered syllables within the word. The resulting phonological output is incorrect.

390

rose

(36a) psoppoxkonu pi-sopo~po-xi-ko-nu 2sg-believe~red-clf:bulk-act-1sg ‘You half-believe me.’ (36b)*psoppxiknu * pi-sopo⟨~p⟩-xi-ko-nu 2sg-believe~red-clf:bulk-act-1sg ‘You half-believe me.’ Vowel deletion can also lead to compensatory lengthening in reduplicated stems if the stranded consonant is not allowed in coda position. This is the case with /r/ in (37) and /w/ in (38). (37) matʃa:re'puekpo * matʃarrepuekpo ma-'tʃare~re-pue-ko-po 3m-pull.up~red-clf:ground-act-perf ‘He pulled them up.’ (38) tçi:'woxi * tçiw'woxi ti-çiwo~wo-xi 3-rain~red-clf:bulk ‘rainy season (it rains repeatedly)’ When a reduplicated stem undergoes both vowel deletion and compensatory lengthening, the base and the copy appear distinct at the surface level (compare /tʃa:/ and /re/ in (37) and /çi:/ and /wo/ in (38)). 4.4 Reduplication and Monosyllabic Roots In one example, the reduplicated syllable is arguably the last syllable of the stem, rather than the last syllable of the root. The parsing of this form is discussed a bit further below. (39) βnikkoxko βi-ni-ko~ko-xi-ko 1pl-eat-act~red-clf:bulk-act ‘We eat often.’ The root ni ‘eat’ is one of the few monosyllabic verbal roots. There are some external reasons to believe that the base niko is composed of the root ni ‘to eat,’

when vowel deletion blurs reduplication in mojeño trinitario

391

and the active suffix -ko (40). For example,-ru is a nominalizing suffix which regularly replaces the active suffix on active verbs. On the verb ‘to eat,’ -ko is replaced by -ru (41). Moreover, example (42) shows that a classifier can be inserted between ni and -ko. (40) nniko nu-ni-ko 1sg-eat-act ‘I eat.’ (41) to naniru to na-ni-ru art.nh 3pl-eat-pat.nz ‘their food’ (42) taniʧeko to juku ta-ni-ʧe-ko to juku 3nh-eat-clf:flat.bounded-act art.nh fire ‘The fire burnt the board.’ (elicited) In (39), it is thus the active suffix rather than the final syllable of the root that is reduplicated. This suffix is considered part of the reduplicative base. This can be explained via the minimality restrictions. Within the Theory of Prosodic Morphology, McCarthy and Prince (1995, 321–325) state that a “prosodic word” must contain at least one foot, i.e. two moras or syllables, according to whether the language is quantity-sensitive or not. Spring (1990, 140–163, cited in McCarthy and Prince 2001) studied reduplication in Axininca Campa, an Arawak language of Peru. She proposes that the base of reduplication is a prosodic word. Since all prosodic words must be bimoraic in this language, monomoraic bases must be augmented for reduplication to apply. Considering this, a possible analysis of example (39) is that there is a constraint in Trinitario on the size of the base: it must be bisyllabic or bimoraic. Reduplication cannot affect the monosyllabic root ni in a regular way because it is subminimal. The suffix following the root is included in the base in order to make it bisyllabic. Then, reduplication copies the final syllable of the base in a regular fashion, but this base-final syllable happens to be a suffix. The copy is immediately followed by the classifier -xi and an additional active suffix. This suggests that the original active suffix following the root was reanalyzed as part of the base in the reduplication process.

392

rose

An elicitation session conducted to collect additional examples of monosyllabic (and monomoraic) root reduplication did not produce any reduplicated form for the roots no ‘put’ and βe ‘take.’ However, the monosyllabic root tse ‘to be ready’ was found to be reduplicable without any sensitivity to minimality. The question of base minimality therefore remains open. (43) ntsetsekwo15 nu-tse~tse-ko-wo 1sg-be.ready~red-act-mid ‘I am almost ready.’ (elicited) Gill claims that in Trinitario, “on words of only one or two syllables AND on Initial Subjunctive words [a class of stative roots which mark the irrealis with a prefix rather than a suffix] the last syllable of the stem or word itself is repeated.” (Gill 1957, 218). He then provides five examples. In my view, only the example repeated in (39) and discussed above constitutes a possible counterexample to the fact that reduplication affects the last syllable of the root. I suggested that it could be explained via minimality constraints. The four other examples concern the trisyllabic derived root kó-tʃitʃa ‘to have children’ and the bisyllabic active roots ero ‘to drink,’ sopo ‘to believe,’ and u:na ‘to be good.’16 In all four of these, the last syllable of the root is normally reduplicated. 4.5 The Double-Copy Pattern In some examples, such as (44), (45) and (46), reduplication results in a surface form where apparently, the final consonant of the root is doubled and the root is followed by a copy of the final syllable. In my analysis, the root-final syllable is copied twice while deleting the root-final vowel (…C)VC∅~CVCV. This constitutes a second pattern of reduplication which could be labeled double-copy or triplication, and I have glossed it as ‘red2.’

15

16

There is no interest in analyzing this example as a case of full reduplication since full reduplication would then be restricted to monosyllabic verbs which show reduplication of the root. It is much more economical to explain this copy of the root as a monosyllabic reduplication of the final (and here, unique) syllable of the root. There is no pattern of full reduplication in Mojeño Trinitario. There are good arguments to synchronically consider u:na a bisyllabic root resulting from the lexicalization of uri ‘good’ and the human classifier -na.

when vowel deletion blurs reduplication in mojeño trinitario

393

(44) ʔtum'memexi ʔtume~meme-xi valiant~red2-clf:bulk ‘He is always ready.’17 (Gill 1957, 217) (45) mamuire 'mwettutupiko to βetʃxiri:wo ma-omuire ma-it(u)~tutu-pi-ko to βi-etʃxiri:wo 3m-also 3m-know~red2-clf:long.flexible-act art.nh 1pl-language ‘He too knows a bit of our language.’ (46) eɲi rikkokoʔo eɲi riko~koko-ʔo pro.m rich~red2-act ‘He is sort of rich.’ (elicited) This type of reduplication was first identified by Gill (1957). However, Gill does not consider it a second pattern of reduplication. His analysis refers to the basic reduplication pattern, i.e. a unique copy of the root-final syllable, with additional morphophonemic changes in the base (consonant doubling). However, his analysis predicts incorrect vowel deletion. (47a) illustrates the double-copy analysis and correctly predicts that the initial /i/, the rootfinal /u/ and the /o/ of -ko are deleted. The last vowel of the double-copy is unaffected by vowel deletion but is transparent. As shown in 2.2, this property is shared with other stem-internal suffixes, such as classifiers. (47b) illustrates the analysis of single-consonant reduplication. It shows that were the final consonant of the root itu ‘learn’ doubled, the initial /i/, the root-final /u/, and the /i/ of -pi would be deleted, being the first, third and fifth vowels of the underlying structure of the word. The resulting phonological output is incorrect. One must therefore reject the consonant doubling analysis.18

17 18

I elicited this translation for the reduplicated form given in Gill (1957). The root means ‘to be valiant, to be strong, to be motivated.’ Apart from the application of vowel deletion, there are other problems with this analysis. First, there is no obvious explanation for consonant doubling. As we have shown in 4.3, examples of apparent consonant doubling can be better explained by vowel deletion, a process pervasive in the language. Second, this analysis links consonant doubling with an additional copy of the root-final syllable. We have shown that the so-called consonant doubling can occur without an additional copy.

394

rose

(47a)esu sittutupik'ripo to esu su-itu~tutu-pi-ko-ripo to pro.3f 3f-learn~red2-clf:long.flexible-act-already art.nh βetʃxiri:wo βi-etʃxiri:wo. 1pl-language ‘She is learning our language bit by bit.’ (47b)*sitttupkoripo *su-itu⟨~t⟩~tu-pi-ko-ripo 3f-learn~red~red-clf:long.flexible-act-already ‘She is learning a language bit by bit.’ It remains an open question as to whether the double-copy is an allomorph of the basic reduplication pattern when it is applied to roots with final vowel deletion or whether it constitutes an independent pattern with its own meaning. In the pair of examples below, (49) differs from (48) both in the pattern of reduplication used and in the presence of an additional prefix: (48) uses the basic reduplication pattern, while (49) uses the double-copy pattern. (48) tkox'mamaxi ti-ko-xuma~ma-xi 3-vz-sickness~red-clf:bulk ‘He is always sick.’ (elicited) (49) wo takxum'mamaxi wo ti-a-ko-xuma~mama-xi neg 3-irr-vz-sickness~red2-clf:bulk ‘He is not always sick.’ (elicited) The comparison of these two examples supports the idea that the second reduplication pattern is related to vowel deletion. In (48), where the rootfinal vowel V4 is not deleted, basic reduplication is used. In (49), where the presence of an additional prefix triggers deletion of the root-final vowel V5, reduplication occurs with a double-copy. The relationship between doublecopy and root-final vowel deletion is also supported by the lack of a doublecopy pattern in the Ignaciano dialect. Ignaciano does not undergo vowel deletion. This double-copy pattern occurs on roots where the final vowel is in the third or fifth syllable of the underlying structure of the word, i.e. in the same

when vowel deletion blurs reduplication in mojeño trinitario

395

context where vowel deletion occurs on basic reduplication, as presented in 4.3. One possible explanation for this pattern is that the presence of a second copy is used to compensate for the loss of similarity between the base and the copy due to vowel deletion. The second copy recreates a sequence of two identical syllables. Surface level opacity would thus feed morphological doubling, and the use of the double-copy pattern would be due to the prosodic environment. Both the basic pattern with a unique copy of the last syllable and the doublecopy pattern can occur on some roots, with a slightly different meaning. This alternative between the two reduplication patterns is not productive.19 (50) βjammoʔo βi-amo~mo-ʔo 1pl-swell.up~red-act ‘We are swollen up (in a body-part).’ (elicited) (51) βjam'momoʔo βi-amo~momo-ʔo 1pl-swell.up~red2-act ‘We are swollen up (all over).’ (elicited) (52) sittupko su-itu~tu-pi-ko 3f-learn~red-clf:long.flexible-act ‘She is learning (a language) a little.’ (elicited) (53) esu sittutupik'ripo to esu su-itu~tutu-pi-ko-ripo to pro.3f 3f-learn~red2-clf:long.flexible-act-already art.nh βetʃxiri:wo. βi-etʃxiri:wo 1pl-language ‘She is learning our language bit by bit.’

19

Speakers accepted neither the double-copy on the roots kemtone and soko (illustrated in (24) and (22) with the basic reduplication pattern), nor the basic pattern of reduplication on the roots riko and ʔtume (illustrated in (46) and (44) with double-copy).

396

rose

(54) βkotʃittʃaxi20 βi-ko-tʃitʃa~tʃa-xi 1pl-vz-children~red-clf:bulk ‘We have many children.’ (Gill 1957, 217) (55) tkotʃittʃatʃaxi ti-ko-tʃitʃa~tʃatʃa-xi 3-vz-children~red2-clf:bulk ‘He has many children with various women.’ (elicited) In the examples above, the two phonological patterns (basic and double-copy) seem to convey different degrees in the reduplication meaning. They do not individually correspond with the two major semantic functions of Trinitario reduplication discussed in section 3 (attenuation and repetition).21 It rather seems that the double-copy pattern can be used to intensify the meaning of reduplication, whether attenuation as in (50) and (51) or repetition, as in (52) and (53). This suggests that the double-copy is not an allomorph of the basic reduplication pattern. A similar correlation between multiple copy and intensification of the meaning of reduplication is attested elsewhere. For example, in Tigre (Semitic, Erytrea), reduplication conveys attenuation. The number of copies goes from one to three. The greater the number of copies, the more attenuated the meaning of the verb is (S. Rose 2003, 113). It remains an open question as to whether the double-copy could be found independently of rootfinal vowel deletion.

5

Conclusion

This paper describes verbal reduplication in Trinitario. At the semantic level, partial reduplication of the verb conveys either an attenuative or a repetitive meaning. At the morphological level, the copy is a derivational suffix internal to the stem. At the phonological level, Trinitario reduplication involves copying the root-final syllable, with a possible double-copy of the same syllable. The base may be altered by vowel deletion and compensatory lengthening.

20 21

/tʃ/ is realized as /t/ before another /tʃ/. In Tupi-Guarani languages, each phonological template expresses a different meaning (Rose 2005, 2007).

when vowel deletion blurs reduplication in mojeño trinitario

397

With the help of a new Trinitario corpus and better knowledge of the phonological evolution of this dialect, a better understanding of the reduplication process was reached. This paper put forward two new analyses. First, Trinitario contains a phonologically-conditioned allomorph of the basic reduplication pattern, previously described by Gill (1957), that surfaces as consonant doubling. Second, there is a second reduplication pattern, where two copies of the root-final syllable occur and the root-final vowel is deleted. These patterns are explained as recent historical developments resulting from the pervasive vowel deletion process in the Trinitario dialect. Taking vowel deletion into account in the analysis here lets us explain the surface variants of various reduplication patterns in a manner that coheres better with the general prosodic system of the language. At the theoretical level, data on reduplication have nourished debates on the morphology-prosody interface. Among others, reduplication in the Arawak language Axininca Campa has triggered recent developments in prosodic morphology within Optimality Theory (McCarthy and Prince 1993; 2001). Without entering this theoretical topic, the Trinitario data demonstrate the great complexity of the morphology-prosody interface. First, the reduplication morpheme (the copy) is defined in prosodic terms (one or two syllables). Second, the base is defined morphologically (as the root) and prosodically (possibly as a “minimal word”). Finally, the prosodic process of vowel deletion applies only after reduplication. The form of the copy is determined by the underlying form of the base, before any prosodic operation applies. At the typological level, the fact that Trinitario displays partial reduplication (copying part of an element), but no full reduplication, is a clear counterexample to the hypothesis that “all languages that have partial reduplication also have total reduplication” (Moravcsik 1978, 328; Rubino 2005, 12). Another interesting aspect of Trinitario reduplication is that the regular processes of vowel deletion and compensatory lengthening result in a surface pattern where the copy and the base do not look alike. Differences between the base and the copy are usually attributed to phonological processes affecting the copy (Rubino 2005, 17). In Trinitario, it is the base that is altered by the vowel deletion and other prosodic rules, making the normally transparent phonological identity between base and copy opaque.

Acknowledgments This paper has benefited very much from commentary by Christian DiCanio, Gérard Philippson, the editors, and an anonymous reviewer.

398

rose

References Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John McCarthy, and Suzanne Urbanczyk. 1999. Reduplication with fixed segmentism. Linguistic Inquiry 30: 327–364. Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press. Crevels, Mily, and Pieter Muysken. 2009. Lenguas de Bolivia: presentación y antecedentes. In Lenguas de Bolivia, Tomo I: Ámbito andino, ed. Mily Crevels and Pieter Muysken, 13–26. La Paz: Plural editores. Cusic, David D. 1981. Verbal plurality and aspect. PhD diss., Stanford University. Fabricano Noé, Felicia, Justo Semo Guají, and Janneth Olivio. 2003. Guía del alfabeto mojeño trinitario. La Paz: Ministerio de Educación de Bolivia. Gill, Wayne. 1957. Trinitario grammar. Manuscript. San Lorenzo de Mojos: Misión Nuevas Tribus. Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Marbán, Pedro. 1701. Arte de la lengua Moxa, con su Vocabulario, y Cathecismo. Lima. McCarthy, John, and Alan Prince. 1995. Prosodic morphology. In The handbook of phonological theory, ed. John Goldsmith, 318–366. Cambridge: Blackwell. . 2001 [1993]. Prosodic morphology: Constraint interaction and satisfaction. Manuscript. . 1993. Generalized alignment. In Yearbook of Morphology: 79–153. Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of human language, ed. Joseph Greenberg, 297–334. Stanford: Stanford University Press. Newman, Paul. 1990. Nominal and verbal plurality in Chadic, Publications in African languages and linguistics 12. Dordrecht: Foris Olza Zubiri, Jesús, Conchita Nuni de Chapi, and Juan Tube. 2002. Gramática Moja Ignaciana. Caracas: Universidad Catolica Andres Bello. Ott, Willis, and Rebecca Ott. 1983. Diccionario ignaciano y castellano, con apuntes gramaticales. Cochabamba: Instituto Lingüístico de Verano. Queixalós, Francesc. 2002. Sur la distensivité. Mémoires de la Société de Linguistique de Paris XII, La pluralité: 55–71. Rose, Françoise. 2005. Reduplication in Tupi-Guarani languages: Going into opposite directions. In Hurch (ed.), 351–368. . 2007. Action répétitive et action répétée: Aspect et pluralité verbale dans la réduplication en émérillon. Faits de Langues 29, La réduplication: 125–143. . 2008. The word-prosodic system of Mojeño Trinitario and pervasive vowel deletion. Paper presented at A Estrutura de Línguas Amazônicas: Fonologia e Gramática II, November 24 in Recife, Brazil.

when vowel deletion blurs reduplication in mojeño trinitario

399

. 2011. Morphological and prosodic structure of the Trinitario verb. Paper presented at the Americanist meeting: Word-formation in South American languages, June 24, in Leipzig, Germany. . to appear. Mojeño trinitario. In Lenguas de Bolivia, Tomo III: Oriente, ed. Mily Crevels and Pieter Muysken. La Paz: Plural editores. Rose, Sharon. 2003. Triple take: Tigre and the case of internal reduplication. San Diego Linguistic Papers 1: 109–128. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Spring, Cari. 1990. Implications of Axininca Campa for prosodic morphology and reduplication. PhD diss., Tucson: University of Arizona. Vendler, Zeno. 1967. Linguistics and philosophy. Ithaca/NewYork: Cornell University Press.

Abbreviations act apl art cau clf cop d f fut irr m mid n nh

active applicative article causative classifier copula discourse marker feminine (singular) future irrealis masculine (singular) middle (incorporated) noun non-human

O pas pat.nz pl plurac pro red red2 rpt S sg tam vz

object past patient nominalizer plural pluractional pronoun reduplication double reduplication reportative subject singular Tense-Aspect-Mood verbalizer

chapter 15

Reduplication in Karitiana (Tupi) Luciana Storto

Reduplication and repetition processes in Karitiana (Tupi family, Arikém branch) are described in this article. Although verb reduplication is a very productive process in the language, there are verbs that do not reduplicate because they have suppleted forms for singular or plural events. In nouns, a repeated root is onomatopoeic and makes reference to a typical repetitive movement or sound characteristic of the living being denoted by the noun. In ideophonic phrases, repetition applies to activities and indicates a durative event. The quantifier -Vra that attaches to adjectival roots can be reduplicated to intensify the quantification.

1

Introduction

This article aims to describe and analyze reduplication processes in Karitiana (Arikém branch, Tupian family) affecting verbs and an affixal morpheme that functions as a quantifier of adjectives. Nouns that could arguably be described as reduplicated are rather shown to be onomatopoeic. A repetition of ideophones in narratives is also discussed but distinguished from reduplication. Müller, Storto and Coutinho-Silva (2006a) have shown examples in which Karitiana, a language that has no plural morphology whatsoever in noun phrases, uses complete root reduplication of verbs to mark plurality of events. SanchezMendes and Müller (2007) argue that verb roots are reduplicated in Karitiana to unambiguously indicate plural events—a linguistic phenomenon called pluractionality by Lasersohn (1995). I show that, although reduplication of verbs is a very productive process in Karitiana, there are verbs that do not reduplicate, but have a suppletive form instead, to indicate plural events. This does not necessarily falsify Sanchez-Mendes and Müller’s account since it appears that the suppletive plural forms in these cases are equivalent in meaning to the reduplicated forms, resulting from the same kind of semantic operation. In onomatopoeic nouns, an iconic phonological string is repeated and makes reference to a typical repetitive movement or sound characteristic of the animate being or object denoted by the noun. An affixal morpheme -Vra that typically attaches to adjectival roots to quantify them as “many” can also be reduplicated

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_016

402

storto

to intensify the quantification. In ideophonic phrases, repetition of an ideophone is used to indicate a durative event.

2

Verb Reduplication

2.1 Ambiguity between Singular and Plural Readings in Noun Phrases Before we discuss verb reduplication and demonstrate that it denotes plurality of the event represented by the reduplicated root, it is necessary to show that Karitiana does not mark the plural of nouns morphologically:1 (1)

pikom ∅-naka2-' y-t asyryty monkey 3-decl-eat-nfut banana ‘(The/a) monkey(s) ate (the/a) banana(s).’

As the translation shows, sentence (1) can mean that one or more monkeys— definite or indefinite—ate one or more bananas—definite or indefinite. This raises the following question: if a plural interpretation of the event is always available without verb reduplication, why should reduplication apply at all? Sanchez-Mendes and Müller (2007) give an explanation that will be discussed in section 2.3. Müller, Storto and Coutinho-Silva (2006a and 2006b) were the first to show that sentences in Karitiana are often ambiguous as a result of the fact that noun phrases headed by common nouns in the language do not have articles 1 All Karitiana examples are transcribed orthographically. Karitiana orthographic conventions represent five vowel qualities a, e, i, o, y (the latter represents a high central vowel), as well as phonemic vowel length (aa, ee, ii, oo, yy). Nasal vowels, both short and long, exist in the language, but the orthography only marks vowels as nasal when they are not contiguous to nasal consonants (labial m, alveolar n, palatal j̃ and velar ng). Only the latter requires a following vowel to be marked as nasal. The symbol j represents the nasal palatal in oral environments or the predictable consonantal use of i. Nasal consonants at other points of articulation are pre-oralized (bm, dn, gng) when preceded by oral vowels or post-oralized (mb, nd, ng) when followed by them. The pre- and post-oralized allophones of nasals that exist in the language are represented orthographically by mb, nd and ng. The syllabic pattern is (C)V(C) and stress falls on the last syllable of a root. Some suffixes have inherent stress, and when they occur in a word, the stress of the suffix is primary and that of the root is secondary (Storto 1999). 2 The allomorphs of the declarative mood are: na- after null agreement (third persons) and taafter overt agreement (first and second persons). The morpheme -ka is an augmentation of the declarative prefix that is present when the verb stem is stress-initial.

reduplication in karitiana (tupi)

403

or plural morphology. They explain this fact by saying that common nouns in Karitiana have a denotation that is neutral for number; that is, they can be interpreted as being either singular or plural depending on the context. Common nouns also differ from mass nouns in that only the latter require a measure unit to be counted. The authors also claim that Karitiana is a language without determiners of any kind (articles, demonstratives or quantifiers) in the noun phrase. Demonstrative constructions have the structure of clauses and are formed by a deictic element, the noun and a copula, as in (2). Universal quantification is expressed by a subordinate clause formed by the noun, a copula and a subordinator, as in (3): (2) dibm ∅-naka-tar3-i ony taso aka tomorrow 3-decl-go-fut deic man be ‘That man will go tomorrow.’ (3) taso ∅-na-sokõ'ĩ-t eremby aka-tyym4 man 3-decl-tie.up-nfut hammock be-sub ‘The man/men tied up all hammocks.’ Lexical items that can be translated as other types of quantifiers are adverbial in that they express the quantification of both nouns and verbs. In (4–5) the word si'ĩrimat is used in combination with a negator to mean either “nobody” or “never,” and in (6–7) the word kandat (‘much’) can quantify the noun phrase or the verb phrase, regardless of the position occupied by the adverb in the sentence:5 (4) i-semboko padni si'ĩrimat eremby 3-get.wet neg ever hammock ‘Hammocks never get wet’ or ‘No hammock(s) ever get(s) wet.’

3 The morphophonological process of lenition accounts for the change of a voiceless dental stop /t/ to a tap [r] in morpheme boundaries before a vowel-intial morpheme. It is a more general process that applies between a final stop or nasal followed by a vowel initial morpheme (Storto 1999), resulting in a voiced approximant when the point of articulation is labial, palatal or velar. 4 The morpheme tyym is used as a conjunction meaning ‘also’ in matrix clauses or as a subordinator sometimes translated as ‘when.’ Here, it occurs inside a universally quantified phrase, which is a clause formed by the noun phrase, a copula and the subordinator tyym. 5 Although the original paper from which these examples have been taken does not give two translations for each sentence, we have added them here for the sake of completion.

404

storto

(5) i-a-okooto padni si'ĩrimat y-'it 3-pass-bite neg ever 1s-son ‘My son was never bitten’ or ‘Nobody ever bit my son.’ (6) kandat ∅-naka-hor-i dibm taso a.lot 3-decl-leave-fut tomorrow man ‘Many men will leave tomorrow’ or ‘Men will be leaving a lot tomorrow.’ (7) ∅-pyry-kiit-dn6 taso pytim'adn kandat tyym 3-assert-exist-nfut man work a.lot also ‘There are many men that work as well’ or ‘There are men doing a lot of work as well.’ Sanchez-Mendes (2009) and Müller and Negrão (2010) have shown that adverbial quantifiers indeed have the same distribution that Storto (1999) has given for adverbs in the language. In a declarative SVO sentence, for instance, adverbial quantifiers, adverbs, and postpositional phrases can appear at the beginning or at the end of the sentence, as well as between the verb and the object. They do not form a constituent with the element over which they quantify, but they function as adverbial modifiers of NPs or VPs. Finally, when numerals combine with a common noun, they do not modify the noun directly as quantifiers and numerals are expected to do, but they first become the complement of a postposition and then combine with the noun, generating a construction that could be paraphrased, for instance, as “monkeys in five” in sentence (8): (8)

yj-py-t pikom ∅-naka-' y-t asyryty 1p-hand-posp monkey 3-decl-eat-nfut banana ‘Five monkeys ate (the/a) banana(s).’

In the next section, verb reduplication and suppletion are exemplified and described as pluralizing operations on verbs and auxiliaries, and in section 2.3 a discussion of the meaning of pluractionality in the language is presented.

6 In the assertative mood, the morpheme -n marks non-future tense. The allomorph -dn is predicted when the stem ends in an oral vowel.

reduplication in karitiana (tupi)

405

2.2

Ambiguity between Singular and Plural Event Readings in Verb Phrases 2.2.1 Verb Reduplication Complete reduplication of the verb root, exemplified in (10), (12), (14), (16) and (18) is extremely productive in Karitiana, as it can be used in most verb roots to indicate plurality of events (Sanchez-Mendes and Müller 2007): (9) Inácio ∅-na-mangat-∅7 Nádia ka'it Inácio 3-decl-lift-nfut Nádia today ‘Inácio lifted Nádia today.’ (10) Inácio ∅-na-mangat~mangat-∅ Nádia ka'it Inácio 3-decl-lift~lift-nfut Nádia today ‘Inácio lifted Nádia today (more than once).’ (11) pikom kyn ∅-naka-pon-∅ João monkey posp 3-decl-shoot-nfut João ‘João shot at a/the monkey(s).’ (12) pikom kyn ∅-na-pon~pon-∅ João monkey posp 3-decl-shoot~shoot-nfut João ‘João shot at a/the monkey(s) (more than once).’ (13) õwã ∅-naka-hit-∅ goojo-ty João child 3-decl-give-nfut canoe-obl João ‘The/A child(ren) gave a/the canoe(s) to João.’ (14) õwã ∅-na-hit~hit-∅ kinda'o-ty João child 3-decl-give~give-nfut fruit-obl João ‘(The/A )child(ren)) gave a/the fruit to João (more than once).’ (15) myhin-t 'ejepo ∅-na-aka-t i-'ot-∅8 one-posp stone 3-decl-be-nfut part-fall-abs.cop.agr ‘One stone fell.’

7 Declarative sentences have -t (after vowel-final stems) or the null morpheme -∅ (after consonant-final stems) as the allomorphs marking non-future tense. 8 This type of syntactic construction is a copular sentence with a nominalized clausal complement. The nominalizer is glossed participle. The suffix on the verb heading, the complement

406

storto

(16) sypom-p 'ejepo ∅-na-aka-t i-'or-ot-∅ two-posp stone 3-decl-be-nfut part-fall~fall-abs.cop.agr ‘Two stones fell (more than once, that is, two events of stone falling).’ (17) taso ∅-naka-kydn-∅ man 3-decl-wait-nfut ‘A/the man/men waited.’ (18) taso ∅-na-kydn~kydn-∅ man 3-decl-wait~wait-nfut ‘A/the man/men waited (more than once).’ Note that although common nouns are always ambiguous between a singular or plural interpretation, depending on the particular verb that is reduplicated, a certain reading may be forced. With reduplicated transitive verbs, the reading of a plural object is forced in verb phrases like ‘eat banana’ or ‘kill monkey.’ The same does not happen with a verb like ‘shoot at monkey,’ because many shooting events may target a single monkey. In other cases, the reading of a plural subject may be forced when certain verbs are reduplicated to imply that the subject is plural. The question that remains is whether this interpretation is imposed by the meaning of the verb or for pragmatic reasons. Sanchez-Mendes and Müller (2007) discuss the meaning of verb reduplication in Karitiana within a framework of formal semantics. They claim that the semantic operation that results from the application of reduplication pluralizes the event denoted by the verb phrase. The arguments given in favor of their hypothesis are: I: Reduplicated verb roots cannot be used with a singular interpretation, as in (19): (19) Inácio ∅-na-mangat~mangat-∅ Nadia ka'it Inácio 3-decl-lift~lift-nfut Nadia today ‘*Inacio lifted Nadia today (once).’ ‘Inacio lifted Nadia today (more than once).’

of the copula, glossed absolutive copular agreement, is present in copular sentences, as well as clefts (Storto 2008). This construction is widely used in Karitiana with intransitive verbs instead of a monoclausal declarative. In section 2.2.3 its structure is discussed in more detail.

reduplication in karitiana (tupi)

407

II: Reduplication is possible for any sentence denoting two or more events and not only for sentences denoting a significant number of events: (20) sypom-p ∅-na-pon~pon-∅ João sojxaty kyn two-posp 3-decl-shoot~shoot-nfut João wild.boar posp ‘João shot twice at a/the boar(s).’ III: Sentences with distributive readings of singular objects should not allow pluractional reduplication because in this kind of sentence one is distributing singular event predicates. The (reduplicated) adverbial quantifier tamyrỹ tamyrỹ can be roughly translated as ‘each,’ and is perfectly grammatical with a distributive reading in (21a), a sentence in which the event of “fruit giving” is interpreted as singular. Indeed, sentence (21b) is ungrammatical, as predicted, because the event of giving a fruit is plural, thus incompatible with the intended distributive reading: (21) a. ta-myrỹ ta-myrỹ ∅-naka-hit-∅ õwã myhin-t 3anaph-alone 3anaph-alone 3-decl-give-nfut child one-posp kinda'o fruit ‘Each child gave one fruit.’ b. *ta-myrỹ ta-myrỹ ∅-naka-hit~hit-∅ õwã 3anaph-alone 3anaph-alone 3-decl-give~give-nfut child myhin-t kinda'o one-posp fruit ‘Each child gave one fruit (more than once).’ In the examples seen so far, verb reduplication affected the root of the verb. Below, I list other verbs in which the reduplicated version of the verb is accompanied by affixal morphology, such as the causative morpheme (22e–22f) and stem formatives, to illustrate that the root is the only morpheme that is reduplicated:9 (22) a. taso ∅-naka-yt-∅ ohy man 3-decl-dig-nfut potato ‘A/the man/men dug a/the/some potato(es).’

9 I thank Suzi Lima for the data elicited in (22).

408

storto

b. taso ∅-naka-yt~yd-na-∅10 ohy man 3-decl-dig~dig-stmf-nfut potato ‘A/the man/men dug potatoes (more than once).’ c. taso ∅-na-ohok-a-t ohy man 3-decl-peel-stmf-nfut potato ‘A/the man/men peeled a/the/some potato(es).’ d. taso ∅-na-ohok~ohok-⟨o⟩na-t ohy man 3-decl-peel~peel-stmf-nfut potato ‘A/the man/men peeled potatoes (more than once).’ e. taso ∅-naka-m-hõron-∅ pykyp man 3-decl-caus-wash-nfut clothes ‘A/the man/men washed clothes.’ f. taso ∅-naka-m-hõron~hõron-∅ pykyp man 3-decl-caus-wash~wash-nfut clothes ‘A/the man/men washed clothes (more than once).’ g. Nadia ∅-na-aka-t i-kaj̃-∅ Nadia 3-decl-be-nfut part-dream-abs.cop.agr ‘Nadia dreamed.’ h. Nadia ∅-na-aka-t i-kaj̃~kaj̃-na-t Nadia 3-decl-be-nfut part-dream~dream-stmf-abs.cop.agr ‘Nadia dreamed (more than once).’ i. bola ∅-na-aka-t i-hop-∅ ball 3-decl-be-nfut part-blow.up-abs.cop.agr ‘A/the ball(s) blew up.’ j. bola ∅-na-aka-t i-hop~hob-na-t ball 3-decl-be-nfut part-blow.up~blow.up-stmf-abs.cop.agr ‘A/the ball(s) blew up (more than once).’

10

Root-final voiceless consonants followed by the stem formative -na become voiced predictably if they are labial and alveolar.

reduplication in karitiana (tupi)

409

k. taso ∅-na-aka-t i-kĩkĩ-t man 3-decl-be-nfut part-scream-abs.cop.agr ‘A/the man/men screamed.’ l. taso ∅-na-aka-t i-kĩkĩ~kĩkĩ-na-t man 3-decl-be-nfut part-scream~scream-stmf-abs.cop.agr ‘A/the man/men screamed (more than once).’ m.Nadia ∅-na-aka-t i-apii-t Nadia 3-decl-be-nfut part-argue-abs.cop.agr ‘Nadia argued.’ n. Nadia ∅-na-aka-t i-apii~apii-dna-t Nadia 3-decl-be-nfut part-argue~argue-stmf-abs.cop.agr ‘Nadia argued (more than once).’ 2.2.2 Verb Suppletion There are verb and auxiliary roots that do not reduplicate in Karitiana but have a suppletive form to indicate plural events as in (23–24): (23) yn ∅-na-oky-t pikom I 3-decl-kill-nfut monkey ‘I killed (a/the) monkey(s).’ (24) yn ∅-na-popi-t pikom I 3-decl-kill.pl-nfut monkey ‘I killed (the) monkeys.’ The suppletive root means that there were plural events of “monkey killing,” and that necessarily means that more than one monkey has been killed. To use sentence (24) to mean that a single monkey has been killed is ungrammatical, as represented in (25): (25) *yn ∅-na-popi-t myhin-t pikom I 3-decl-kill.pl-nfut one-posp monkey ‘I killed one monkey (more than once).’ Conversely, sentence (23) can mean that any number of “monkey killing” events took place (one or more), as examples (26–27) confirm:

410

storto

(26) yn ∅-na-oky-t sypom-p pikom I 3-decl-kill-nfut two-posp monkey ‘I killed two monkeys.’ (27) yn ∅-na-oky-t myhin-t pikom I 3-decl-kill-nfut one-posp monkey ‘I killed one monkey.’ We have seen that verb suppletion pluralizes the meaning of the verb phrase, and that sometimes implies that the object will necessarily be interpreted as plural. The subject of a transitive suppletive root, as expected, is not grammatically affected by suppletion and may be either singular (28) or plural (29): (28) myhin-t õwã ∅-na-popi-t pikom one-posp child 3-decl-kill.pl-nfut monkey ‘One child killed the monkeys.’ (29) sypom-p õwã ∅-na-popi-t pikom two-posp child 3-decl-kill.pl-nfut monkey ‘Two children killed the monkeys.’ The same pattern is true of the transitive verb ‘to catch.’ If there were many events of “fish catching,” more than one fish must have been caught, thus the ungrammaticality of (33): (30) myhin-t/sypom-p taso ∅-naka-ot-∅ myhin-t 'ip one-posp/two-posp man 3-decl-catch-nfut one-posp fish ‘One/two man/men caught one fish.’ (31) myhin-t/sypom-p taso ∅-naka-ot-∅ sypom-p 'ip one-posp/two-posp man 3-decl-catch-nfut two-posp fish ‘One/two man/men caught two fish.’ (32) myhin-t/sypom-p taso ∅-naka-piit-∅ 'ip one-posp/two-posp man 3-decl-catch.pl-nfut fish ‘One/two man/men caught fish (more than once).’ (33) *myhin-t/sypom-p taso ∅-naka-piit-∅ myhin-t 'ip one-posp/two-posp man 3-decl-catch.pl-nfut one-posp fish ‘One/two man/men caught one fish (more than once).’

reduplication in karitiana (tupi)

411

Intransitive verbs with suppletive roots mean, again, that a plural number of events took place: (34) ∅-na-yryt-∅ taso yj-akan pip 3-decl-arrive-nfut man 1p-village loc ‘A/the man/men arrived in our village.’ (35) ∅-na-ymbykyjt-∅ taso yj-akan pip 3-decl-arrive.pl-nfut man 1p-village loc ‘(The) men arrived in our village’ *‘A man arrived in our village.’ (36) ∅-naka-tat-∅ taso yj-akan pip 3-decl-go-nfut man 1p-village loc ‘A/the man/men went to our village.’ (37) ∅-naka-hot-∅ taso yj-akan pip 3-decl-go.pl-nfut man 1p-village loc ‘(The) men went to our village’ *‘A man went to our village.’ In the next two sub-sections we will see that not only lexical verbs have suppletive roots in Karitiana, but “functional verbs” such as the copula (cf. 2.2.3) and aspectual auxiliaries (cf. 2.2.4) may undergo suppletion as well. 2.2.3 Copula aka and kii We have seen examples of the copula aka before in this paper used in demonstrative constructions, universal quantification constructions, and in declarative copular sentences. At least two of these uses, in demonstrative constructions (38b) and copular sentences (39), allow for suppletion: (38) a. dibm ∅-naka-tar-i ony taso aka tomorrow 3-decl-go-fut deic man be ‘That man will go tomorrow.’ b. dibm ∅-naka-hor-i ari taso kii tomorrow 3-decl-go.pl-fut deic.pl man be.pl ‘Those men will go tomorrow.’

412

storto

(39) eje pip ∅-naka-kii-t gyryj-∅ soil posp 3-decl-be.pl-fut earthworm-abs.cop.agr ‘Earthworms are (stay, live) in the soil.’ Another interesting case of suppletion is found in copular sentences, when the verb that heads the complement of the copula is itself a copula. Before looking at those examples, however, we must discuss the structure of copular sentences. All copular sentences in Karitiana have clausal complements, headed by nouns, adjectives, or intransitive verbs (Storto 1999; 2008; 2010) as in (40– 42): (40) byyty ∅-na-aka-t kinda'o-t papaya 3-decl-be-nfut fruit-abs.cop.agr ‘Papaya is a fruit.’ (41) taso ∅-na-aka-t i-se'a-t man 3-decl-be-nfut part-beautiful-abs.cop.agr ‘The man is good/good-looking.’ (42) taso ∅-na-aka-t i-kat-∅ man 3-decl-be-nfut part-sleep-abs.cop.agr ‘The man is sleeping.’ The main evidence that the complement of a copula is clausal is the fact that its head is suffixed by the agreement marker -t/-∅. This suffix is present in copular constructions, in clefts, and wh-sentences whenever an absolutive argument— subject of an intransitive verb or object—is moved to sentence-initial position (Storto 2008; 2010). Another piece of evidence for the clausal status of copula complements is the participle prefix i- that nominalizes the clause when it is headed by an adjective or a verb but, crucially, not when it is headed by a noun. Turning now to the examples in which the complement of the copula is itself headed by a copula: (43) asori ∅-na-aka-t eem-∅ i-aka-t bat 3-decl-be-nfut black-abs.cop.agr part-be-abs.cop.agr ‘A bat is black.’

reduplication in karitiana (tupi)

413

(44) erery osiit ∅-na-aka-t pok-ora-t cotton.flower 3-decl-be-nfut white-qtfr-abs.cop.agr i-kii-t part-be.pl-abs.cop.agr ‘Cotton flowers are white.’ What seems to be happening here is that the adjectives eem and pok have been themselves extracted to the left periphery of the complement clause, triggering the absolutive copular agreement morphology. Independent of the correct structural analysis of these sentences, it is clear that the copula inside the complement clause suppletes for plural in (44). It remains to be determined whether the copula aka (as in example 3) can undergo suppletion with kii when it is used in universally quantified clauses. If it cannot, this is evidence that the copula and the subordinator tyym in these constructions have become lexicalized and now constitute a single lexical item. 2.2.4 Aspectual Auxiliaries Karitiana has imperfective auxiliaries that form a complex head with the verb in matrix or embedded sentences. Verb and imperfective auxiliaries are different phonological words, but syntactically they represent a complex head, forming a prosodic unit that can be described as a phonological phrase (Storto 1999). Examples (45–46) are assertative sentences in which the imperfective auxiliaries co-occur with the verb ‘to lift.’ In (46) the suppletive auxiliary agi contributes the meaning of plurality of imperfective events. (45) ∅-py-mangat tyka-dn taso Luciana 3-assert-lift impf-nfut man Luciana ‘Luciana is lifting a/the man/men.’ (46) ∅-py-mangat agi-dn taso Luciana 3-assert-lift impf.p-nfut man Luciana ‘Luciana is lifting a/the man/men (more than once).’ In embedded environments such as (47) and (48) the imperfective auxiliaries occur with a progressive morpheme to form one phonological word. Together, they form a syntactic unit with the verb that adds an imperfective progressive meaning to the event in the subordinate clause. In (48), the event is marked plural by the presence of the imperfective auxiliary agi:

414

storto

(47) [a-oty-p a-tat tyki-'oot] y-ta-so'oot-∅ yn an-ty 2s-bath-loc 2s-go impf-pgr 1s-decl-see-nfut I you.sg-obl ‘While you (sg.) were going to bathe, I met you (sg.).’ (48) [aj-oty-p aj-hot agi-'oot] y-ta-so'oot-∅ yn ajxa-ty 2p-bath-loc 2p-go.pl impf.p-pgr 1s-decl-see-nfut I you.pl-obl ‘While you (pl.) were going to bathe, I met you (pl.).’ 2.3 The Meaning of Pluractionality in Karitiana According to Sanchez-Mendes and Müller (2007), the semantic effect of reduplication is the pluralization of the event denoted by the verb phrase. They call the reduplicated element in Karitiana a pluractional affix, in the sense of Lasersohn (1995). Lasersohn claims that pluractional morphemes usually show up as verb affixes and mean that a multiplicity of events has occurred. The proposal of Sanchez-Mendes and Müller (2007) is that pluractional affixes in Karitiana perform a pluralization operation on verb denotations that excludes singular events. As a result, the singular or atomic interpretation of verb roots is no longer available after reduplication takes place. They claim that in a language like Karitiana, in which the interpretation of both nouns and verbs is cumulative in the sense of Kratzer (2005),11 pluractionality is similar to the pluralization of nouns in languages in which nouns are singular or atomic (i.e. count nouns). The same can be said about suppletive roots. The root that may have a singular meaning can be used in contexts where there is a singular or plural number of events. Suppletive roots, however, always denote a plural event, so they must be the result of a pluralizing operation, possibly the same described above for reduplicated verbs.

11

Cumulativity is a semantic property of predicates that can be tested in the following way: if it applies to two individuals in the denotation of the predicate, it also applies to their sum. All plural and mass nouns are cumulative, because if you add, for instance, apples and apples, you have a sum of individuals that can be described as apples, and if you add any amount of sugar with any other amount, you still are left with sugar. But if you add a singular count noun to another singular count noun the result is not a singular count noun, but a plural count noun. From this one concludes that singular count nouns are not cumulative. In Karitiana, all nouns are neutral for number, so they pass the cumulativity test: if you add two men ‘taso’ and two other men ‘taso,’ you are left with four ‘taso.’ We have seen that verbs in Karitiana can denote plural events in the same way as nouns, and therefore they are cumulative.

reduplication in karitiana (tupi)

415

Copulas and aspectual auxiliaries have suppletive forms to indicate plural events. As far as I understand their distribution, the analysis given by SanchezMendes and Müller (2007) for reduplicated roots can be applied to copulas and aspectual auxiliaries as well.

3

Adjectival Quantifier: Reduplication of Affixal Morphology

3.1 Evidence in Favor of a Class of Adjectives in Karitiana Before discussing the adjectival quantifier affix, it is necessary to give arguments for the existence of a class of adjectives in the language, distinct from the classes of nouns and verbs. Caleb Everett (2006) has a very useful presentation of the main evidence for a class of adjectives in Karitiana, reproduced below: The differentiation of adjectives from nouns and from verbs is apparent in four principal ways. First, their behavior in copular constructions distinguishes them from nouns. Second, they can occur without inflections when following nouns, so distinguishing them from verbs. Third, they can be derived from nouns via an adjectivizing morpheme -nã. Fourth, they can be suffixed with a -ra suffix when following a plural noun. (Everett 2006, 308) The behavior of adjectives in copular sentences has been exemplified and discussed in section 2.2.3 and the meaning and distribution of the -ra suffix will be addressed in section 3.2. The other two criteria used by Everett to differentiate adjectives from other categories are exemplified below (examples taken from Everett 2006, 309–311). In (49) and (50) adjectives are differentiated from verbs, and in (51) and (52) adjectives are shown to be derived from nouns by the addition of the adjectivizer suffix -na: (49) ombaky eem jaguar black/dirty ‘The jaguar is dirty/black.’ or ‘black/dirty jaguar’ (50) *ombaky pyt' y jaguar eat ‘The jaguar is eating.’

416

storto

(51) j̃onso 'ed-na12 woman child-advzr ‘the pregnant woman’ (52) him saa-dna animal smell-advzr ‘the smelly animal’ 3.2

Morphophonology, Syntax, and Semantics of the Adjectival Quantifier Interestingly enough, although nouns do not have plural morphology in Karitiana, adjectives may be quantified by the suffix -Vra, which can be freely translated as “many” in the sense of “a significant quantity of (something)”: (53) 'ewosiit som ∅-na-aka-t i-se'a-t flower red 3-decl-be-nfut part-beautiful-abs.cop.agr ‘A/The red flower(s) is/are beautiful.’ (54) 'ewosiit sõw-õrã ∅-na-aka-t i-se'a-t flower red-qtfr 3-decl-be-nfut part-beautiful-abs.cop.agr ‘Many (a lot of) red flowers are beautiful.’ If this suffix is added to a consonant-final adjective, the initial vowel of the suffix assimilates in quality to the last vowel of the root, and lenition of the last consonant of the root takes place. We must say that the vowel is part of the suffix and not epenthetic, because lenition only takes place between a final stop or nasal and a vowel-initial suffix in the language (Storto 1999). In cases in which the suffix is added to adjectival roots ending in a vowel, the initial vowel of the suffix is deleted: (55) 'ep saraka-ra ∅-na-aka-t i-se'a-t tree smooth-qtfr 3-decl-be-nfut part-beautiful-abs.cop.agr ‘Many (a lot of) smooth trees are nice.’

12

The orthography of Karitiana adopted in this paper does not mark nasal vowels with a tilde if they are adjacent to a nasal consonant (51), in contrast to Everett’s transcription -nã for the same adjectivizing morpheme. Instead, they mark the pre-oralization of the nasal when contiguous to an oral vowel, as in (52).

reduplication in karitiana (tupi)

417

When the suffix is reduplicated, as in (56), the initial vowel of the quantifier suffix is preserved and the last vowel of the reduplicant copy is deleted.13 The difference in meaning between sentence (54) and (56) is simply that reduplication of the affix creates an intensification of the plural quantification: (56) 'ewosiit sõw-õr~õrã ∅-na-aka-t i-se'a-t flower red-red~qtfr 3-decl-be-nfut part-beautiful-abs.cop.agr ‘Lots (and lots) of red flowers are beautiful.’ Table 1 is a list of adjectives in which suffixation with the adjectival quantifier -Vra and intensifying reduplication of that suffix occur: table 1

Adjectives, quantified adjectives, and reduplicated quantified adjectives

ky’ep ‘sharp’ josek ‘piercing’ syk ‘acidic’ osyk ‘full’ tek ‘short’ opap ‘distant’ penem ‘narrow’ horop ‘long, tall’ pendot ‘wide’ pyondyt ‘thick’ sembok ‘wet’ got ‘new, young’ terep ‘straight’ non ‘crooked, bent’ ’in’in ‘small’ bypiit ‘close’ eem ‘black, dirty’ hyk ‘old’ nam ‘rotten’

13

ky’ewera josekera sykyra osykyra tekera opawara penewẽrã horowora pendorora pyondyryra sembokora gorora terewera norõrã ’in’ĩrĩrã bypiirira ẽẽwẽrã hykyra nawãrã

ky’ewerera josekerera sykyryra osykyryra tekerera opawarara penewẽrẽrã horoworora pendororora pyondyryryra sembokorora gororora terewerera norõrõrã ’in’ĩrĩrĩrã bypiiririra ẽẽwẽrẽrã hykyryra nawãrãrã

We consider the longer form to be the base and the shorter form to be the reduplicant because the longer form preserves the phonological material of the base -Vra, whereas in the shorter form the final vowel of the base is deleted.

418

storto

Some speakers have different interpretations of the meaning of the affix -Vra and its reduplication when used with specific adjectives. To these speakers, a sentence pair like (54) and (56) with the adjective 'ewet ‘thin’ or okyp ‘hot’ could mean ‘a little thin/hot’ when suffixed by -Vra, and the plural of that when reduplicated, that is ‘many individuals who are a little thin/hot.’ This is similar to what has been reported in the literature about the same reduplication process being able to mean intensification or attenuation (Rubino 2005, 19; Gomes 2007, 393). In this case, however, the attenuative meaning is available with the suffixation of -Vra before it reduplicates, and reduplication adds the meaning of plural. More research must be done before we fully understand the morphophonology, syntax, and semantics of the morpheme -Vra. Verbs and adjectives in the language have properties in common that have yet to be explained because although they are separate categories, all adjectives can be inflected as verbs, and intransitive verbs behave like adjectives in causativization and copular sentences (Storto and Rocha, to appear). A detailed study of the syntactic and semantic properties of lexical items will, hopefully, solve some of the problems raised in this section.

4

Onomatopoeic Nouns

Many nouns in Karitiana seem to be formed via reduplication, but, in fact, most of them are simply onomatopoeic; that is, they are formed by an iconic phonological string, not morphemes, that represents or emulates a typical repetitive movement or sound characteristic of the living being or object denoted by the noun. table 2

Onomatopoeic nouns and intuitions on their origins by native speakers

siritsirit pyyppyyp tomtom-a ta’eret ta’eret ’ewewe ’oot’oot hoojhooj ketket

‘species of hummingbird’ (sound of wings) ‘species of owl’ (vocalization of bird) ‘guitar’ (sound of instrument) ‘species of bird’ (vocalization of bird) ‘species of bird’ (vocalization similar to humans) ‘species of egret’ (not sound) ‘species of frog’ (speakers have no intuition) ‘species of owl’ (speakers have no intuition)

reduplication in karitiana (tupi)

419

This particular type of repetition is not considered an instance of reduplication because the string of sounds that is repeated does not appear as a root elsewhere in the language. Although onomatopoeic nouns are often formed by the complete repetition of a string, they need not be, as in okokoj ‘species of bat,’ 'ewewe ‘species of bird’ and kyryryt ‘species of frog,’ and kydn kodna ‘butterfly.’ In nouns that involve the repetition of a string that is used as a root elsewhere in the language, it is necessary to investigate whether the meaning of the root is somehow present in the noun or not. If it is, we can consider that noun as formed via reduplication. If the meaning of the root does not have any relationship with the meaning of the repetitive string, then it could be a simple case of homophony and the noun may be considered onomatopoeic.

5

Repetition of Ideophones

This section describes the criteria for considering ideophones as a special word class in Karitiana and shows examples of ideophone repetition. The term repetition is used here to label a different process than reduplication. In short, the repeated ideophone roots are separated by a pause, which does not happen with reduplicated roots. 5.1 Evidence for the Existence of Ideophones in Karitiana Karitiana has a special class of words called ideophones, different from nouns, verbs, adjectives or adverbs (Landin 1983). The test I have been using to distinguish ideophones from verbs is their use in the imperative mood. If the root is an ideophone, it cannot be inflected with the person prefix (a- ‘2s,’ aj- ‘2p’) or with the imperative suffix (-a/-∅); whereas, verbs can. Ideophones may be used in the imperative as complements of the verb a ‘to do, to say’; in such cases, the verb is inflected and the ideophone is bare: (57) a-kar-a 2s-sleep-imp ‘Sleep!’ (58) *kat a-'a-∅ sleep 2s-do-imp ‘Sleep!’

420 (59) tẽ a-'a-∅ sleep 2s-do-imp ‘Sleep!’ (literally: ‘Do tẽ!’) (60) *a-tẽ-∅ 2s-sleep-imp ‘Sleep!’ (61) a-mbik-a 2s-sit-imp ‘Sit!’ (62) *bik a-'a-∅ sit 2s-do-imp ‘Sit!’ (63) dok a-'a-∅ sit 2s-do-imp ‘Sit!’ (literally: ‘Do dok!’) (64) *a-ndok-a 2s-sit-imp ‘Sit!’ (65) terek terek a-'a-∅ walk walk 2s-do-imp ‘Walk!’ (literally: ‘Do terek terek!’) (66) *a-terek tereg-a 2s-walk walk-imp ‘Walk!’ (67) a-tar-a 2s-go-imp ‘Go!’ (68) *tat a-'a-∅ go 2s-do-imp ‘Go!’

storto

reduplication in karitiana (tupi)

421

(69) i-okoor-a 3-bite-imp ‘Bite it/him!’ (70) *okoot a-'a-∅ bite 2s-do-imp ‘Bite it/him!’ (71) xak a-'a-∅ bite 2s-do-imp ‘Bite it/him!’ (literally: ‘Do xak!’) (72) *i-xag-a 3-bite-imp ‘Bite it/him!’ The examples given above include pairs of verbs and ideophones that have roughly the same meaning. Verbs are inflected for person and mood in imperative sentences. Storto (2008) analyzes such person prefixes as absolutive agreement: subject agreement for intransitive verbs and object agreement for transitive ones. Conversely, ideophones cannot be inflected but act as the complement of a verb “to do” in the imperative. 5.2 Examples of Repetition of Ideophones in Narratives One of the uses of ideophones in Karitiana is in narratives within what I call “ideophone phrases” as a resource, similar to a sound track or visual special effect in a movie, utilized to put the listener closer to the space and time of the narrative. It can be considered an instantiation of verbal art. In such uses, ideophones referring to events that are typically atelic, such as walking, talking, etc, are often repeated whereas events that are telic, or have an endpoint, such as sitting down or arriving, are not: (73) tarak tarak tarak otam dok walk walk walk arrive sit ‘(He) walked, walked, walked, arrived, sat.’ (74) sosyp ba, ∅-naka-'a-t saryt i-ohỹr-ym look open 3-decl-do-nfut ind.evid 3-top-loc ‘(He) looked and opened (it), (that is how) they say he did (it) on top of it.’

422

storto

(75) sosyp, iri-'a-j, tek tek jyk look cit-do-fut walk walk stop.standing ‘He looked, walked, walked and stopped.’ (76) ho ho ho jydn, ∅-na-tambyky oko ta'ã-t go go go take 3-decl-bring iter dir.ediv-nfut ‘(They) went, went, went, took it, and brought it again.’ Gabas Junior and van der Auwera (2004) identify ideophones in Karo, a Tupian language of the Ramarama branch. In Karo there is a process of repetition of ideophones that is very similar to the one just described in Karitiana. Crofts (1973; 1984) described ideophones in Munduruku (Munduruku branch, Tupi family), and, judging by the examples, they can also be subject to the kind of repetition seen in Karitiana.

6

Conclusion

The reduplication processes affecting verbs and adjectives in Karitiana are implemented via affixation and trigger similar meanings despite their different lexical categories: when both verbs and adjectives show reduplication (or when certain verbs and auxiliaries show suppletion), the interpretation is necessarily plural. In the case of the adjectival quantifier -Vra, reduplication can be translated as an intensification of the quantification. In nouns, what appears to the untrained observer to be reduplication is, in most cases, an instance of onomatopoeic repetition. In ideophones, reduplication is not present, but the repetition of an ideophone results in a durative reading for the event. One issue that must be addressed is whether the reduplication processes described in this chapter are instances of inflection or derivation. This is not a simple question because the phenomena encoded by reduplication in Karitiana, namely, pluractionality and intensification of plural quantification, are not very well understood cross-linguistically. In Karitiana, for instance, we have seen that using a non-reduplicated form of the verb gives both the reading of singular and plural event. Given the fact that reduplication processes are the result of semantic operations that are not obligatory in the language, it seems reasonable to classify them as derivational. However, we have seen that once reduplicated, a verb root has an exclusively plural reading, which can be useful as a way to avoid ambiguity. In this respect, speakers may start using it regularly when they want to convey the meaning of plurality and the process may

reduplication in karitiana (tupi)

423

end up regularized as inflectional. I have noticed that some speakers of Karitiana have started to use suppletive roots in such a way. However, since this is not yet true of all speakers, I must conclude that, at this moment in time, pluractionality is a derivational process in the language. The phenomenon of intensifying a quantifier through reduplication seems to be optional as well. It remains to be seen whether the facts reported in this paper about the semantics of verb and auxiliary reduplication and suppletion can be observed in all other Tupi languages. Lima (2007; 2008) claims that plurality of events is marked via reduplication in Juruna verbs in an exactly parallel fashion to what Sanchez-Mendes and Müller (2007) described for Karitiana. Lima (2007) includes an overview of the literature on Tupi languages, in which she reports what has been described about verb reduplication in six out of the ten families that comprise the Tupi linguistic family. Considering other Amazonian languages from two different genetic groups, I would like to comment on the works of Bruno (2003) and van der Voort (2003). Bruno discusses the phonological status of reduplication in Waimiri-Atroari (Karib) without mentioning its semantics, but one can see from the translation that reduplication means plurality of events (perhaps iterative) and duration (perhaps progressive aspect). For Kwaza (isolate), van der Voort describes many different meanings associated with reduplication, including repetition of an event, plurality of participants, intensification, attenuation, and progressive aspect. The most interesting and surprising of all is the reduplication of person affixes on verbs to create tense distinctions. The present paper describes reduplication of a suffixal quantifier in adjectives, meaning intensification of the quantification, which is yet another case of affixal reduplication reported to exist in Amazonia. If plurality of events and participants is a recurring meaning in reduplication processes in Amazonia, then it may be the case that, like in Karitiana, pluractionality is the process that underlies event and participant pluralization in other Amazonian languages as well. As we have seen, participant pluralization is not a necessary but a possible result of pluralizing the event.

Acknowledgements This research was developed during a sabbatical year at the University of Edinburgh, supported by FAPESP grant 2010/08571-9. I thank Caroline Heycock for hosting me during that period. A reduced version of this paper was published in Portuguese as Storto (2012).

424

storto

References Bruno, Ana Carla. 2003. Reduplicação em Waimiri-Atroari. Amérindia 28: 83–94. Crofts, Marjorie. 1973. Gramática Mundurukú. Brasília: SIL. . 1984. Ideófonos na narração Mundurukú. In Estudos sobre línguas Tupi no Brasil, ed. Robert A. Dooley, 207–218. [Série Linguística 11]. Brasília: SIL. Everett, Caleb. 2006. Gestual, perceptual and conceptual patterns in Karitiana. PhD diss., Rice University, Houston. Gabas Jr., Nilson, and Johan van der Auwera. 2004. Ideophones in Karo. In Language, culture and mind, ed. Michel Achard and Suzanne Kemmer, 397–413. Stanford: CSLI Publications. Gomes, Dioney. 2007. Reduplicação verbal em Munduruku. In Línguas e culturas Tupi, Vol. 1., ed. Aryon Dall’Igna Rodrigues and Ana Suelly Arruda Câmara Cabral, 391–396. Campinas: Editora Curt Nimuendajú / Brasília: LALI. Kratzer, Angelika. 2005. On the plurality of verbs. Manuscript. Semantics Archives. Landin, David. 1983. Dicionário e léxico Karitiana/Português. Brasília: SIL. Lasersohn, Peter. 1995. Plurality, conjunction and events. Dordrecht/Boston: Kluwer Academic Publishers. Lima, Suzi Oliveira de. 2007. Plurality and distributivity in Juruna: some considerations about verbal cumulativity. In University of Massachusetts Occasional Papers 35. Proceedings of the semantics of under-represented languages in the Americas 4, 117–128. Amherst: GLSA. . 2008. A estrutura argumental dos verbos na língua Juruna (Yudjá): Da formação dos verbos para a análise das estruturas sintáticas. MA thesis, Universidade de São Paulo. Müller, Ana, and Esmeralda Negrão. 2010. On distributivity in Karitiana. In University of Massachusetts Occasional Papers 41. Proceedings of semantics of under-represented languages in the Americas 5, 121–138. Amherst: GLSA. Müller, Ana, Luciana Storto, and Thiago Coutinho-Silva. 2006a. Número e a distinção contável-massivo em Karitiana. Revista da Associação Brasileira de Linguística 5/1–2: 185–213. . 2006b. Number and the count-mass distinction in Karitiana. In University of British Columbia Working Papers in Linguistics, vol. 19. Proceedings of WSCLA XI, ed. Atsushi Fujimori and Maria Amélia Reis Silva, 122–135. Vancouver: University of British Columbia. Rubino, Carl. 2005. Reduplication: form, function and distribution. In Studies on reduplication, ed. Bernhard Hurch with Veronica Mattes, 11–29. Berlin/New York: Mouton de Gruyter. Rodrigues, Carmen Lúcia. 1995. Étude morphosyntaxique de la langue Xipaya (Brésil). PhD diss., Université Paris VII.

425

reduplication in karitiana (tupi)

Sanchez-Mendes, Luciana. 2009. A quantificação adverbial em Karitiana. MA thesis, Universidade de São Paulo. Sanchez-Mendes, Luciana, and Ana Müller. 2007. The meaning of pluractionality in Karitiana. University of Massachusetts Occasional Papers 35. Proceedings of the semantics of under-represented languages in the Americas 4, 247–258. Amherst: GLSA. Storto, Luciana. 1999. Aspects of a Karitiana grammar, PhD diss., Massachusetts Institute of Technology. . 2008. Marcação de concordância absolutiva em algumas construções sintáticas em Karitiana. Ameríndia 32: 183–203. . 2010. Copular constructions in Karitiana: a case against case movement. In University of Massachusetts Occasional Papers 41. Proceedings of semantics of underrepresented languages in the Americas 5, 205–226. Amherst: GLSA. . 2012. Duplicação em Karitiana. In Nominais nus: Um olhar através das línguas, ed. Pires de Oliveira, Roberta and Meiry Meruchi Mezari, 213–233. Campinas: Mercado de Letras. Storto, Luciana, and Ivan Rocha. to appear. Strategies of valence change in Karitiana. In Valence increasing processes in Amazonia, ed. Stella Telles, Ana Carla Bruno and Francisco Queixalos. Bogota: Universidad Nacional de Colombia. Voort, Hein van der. 2003. Reduplication of person markers in Kwaza. Acta Linguistica Hafniensia 35: 65–94.

Abbreviations (abs.)cop.agr absolutive copular agreement advzr adjectivizer assert assertative mood cit citative mood decl declarative mood deic deictic dir.evid direct evidential fut future tense imp imperative mood impf imperfective aspect impf.p imperfective aspect (plural) ind.evid indirect evidential loc locative nfut non-future tense

obl part pl posp pgr qtfr red sg stmf sub 3

3anaph

oblique case participle plural postposition progressive aspect quantifier reduplicant singular stem formative subordinator third person prefix (singular or plural) or third person possessive clitic third person anaphoric prefix (singular

426

1s

1p

storto

or plural) or third per- 2s son possessive clitic first person singular prefix or first person singular possessive 2p clitic first person plural prefix or first person plural possessive clitic

second person singular prefix or second person singular possessive clitic second person plural prefix or second person plural possessive clitic

chapter 16

Is Reduplication an Areal Feature of the Guaporé-Mamoré Region? Hein van der Voort

Areal diffusion of productive reduplication seems to be a relatively rare phenomenon. The Guaporé-Mamoré region, in the southwestern Amazon, is host to a large number of language families and linguistic isolates. In spite of this genetic diversity, several grammatical traits are shared throughout parts of the region and suggest that it is a linguistic area. This article is intended as both a survey of productive reduplication in the Guaporé-Mamoré region and an examination of the question whether reduplication could represent an areal feature. As far as possible, the properties of reduplication are described individually for most of the region’s genetic linguistic units. Potential regional patterns of reduplication are discussed in the concluding section.

1

Introduction

In this article, reduplication is regarded as a productive morphosyntactic process that produces forms through non-recursive repetition of (parts of) an original form, with a different but systematically related meaning.1 Productive reduplication is a morphosyntactic process which is both similar to and different from other morphosyntactic processes. Partial reduplication is much like affixation in that it involves a morphological segment that changes the meaning (or the syntactic status) of the word to which it is applied, but it is different from affixation in that the form of the segment is variable and depends on the form of the word to which it is applied. Similarly, full reduplication is like compounding in that it causes a change of meaning that is not necessarily predictable on the basis of the meaning of the component parts of the resulting complex word form, but it is different from compounding in that the form of the modifier is not determined by lexical choice and depends entirely on the

1 This definition of reduplication corresponds to the one used in this book and derives from the one used in the Graz Reduplication Database Project (Hurch 2005ff.).

© koninklijke brill nv, leiden, 2014 | doi: 10.1163/9789004272415_017

428

van der voort

form of the head. Even though productive reduplication is in certain respects very different from other morphosyntactic processes, it has essential characteristics of a morphosyntactic process. An intriguing question is whether it can be the subject of diffusion through language contact, as attested for other morphosyntactic processes. In this article I will investigate the possibility of areal diffusion of reduplication in the Guaporé-Mamoré linguistic area, looking at examples from a number of genetically unrelated languages. Reduplication is almost a universal phenomenon. Most of the world’s languages have some kind of reduplication, and in many languages it is a productive grammatical process. Recently, questions of origin have been raised in reduplication studies. Most of the work that has resulted from these questions so far (Hurch, ed. 2009, Hurch and Mattes, eds. 2009) concerns productivity and historical development. However, the issue of diffusion through contact has hardly been touched upon in work on reduplication until recently (see Stolz 2008, Stolz, et al. 2011). The relative absence of reduplication in Indo-European languages should probably be attributed to genetic factors. Although some authors (Rubino 2005, 23) suggest that the prolific reduplication in creole languages based on IndoEuropean lexifier languages must be due to substratum effects, the picture is not that straightforward. Reduplication is almost universal in creole languages, regardless of whether the adstrate languages have reduplication or not. Bakker and Parkvall (2005, 522) suggest that only with regard to cross-linguistically uncommon types of reduplication could one claim a connection with adstrate languages.2 Note also that reduplication is almost completely absent in nonexpanded pidgins (Bakker 2003). In the introduction to her edited volume on reduplication in contact-induced languages Kouwenberg (2003, 3) states that “All of this suggests that reduplication emerges as a strategy of linguistic expansion between speakers of the same (contact) language, not in situations involving speakers of different languages.” In sum, pidgin and creole language specialists tend to attribute the emergence of reduplication to universal properties of spontaneous creole language formation. Looking at the distribution of reduplication across the world’s languages, Hurch and Mattes (2009, 302) note that “we do not find a specific concentration as to geographical regions or other linguistically defined areas.” In his 2 See e.g. Kouwenberg and LaCharité (2004, 321) for a likely case of substrate transfer of stative adjective reduplication from Gbe to the English-based Surinamese Creole. Yakpo (2012) presents evidence that certain tonal patterns in verbal reduplication shared by a number of genetically-related West African languages and certain African and Caribbean creoles are due to areal diffusion.

is reduplication an areal feature?

429

contribution to the symposium on reduplication in Amazonian languages at the 53rd International Congress of Americanists, Hurch (2009) doubted that there are linguistic areas in which reduplication is a characteristic feature; likewise, it is not discussed in literature on language contact in the Americas. It is true that reduplication rarely features in literature on areal linguistics, but it is not completely absent either. For example, reduplication is mentioned as an areal feature in the Caucasus region (Chirikba 2008) and possibly in Sri Lanka (Bakker 2006). Moreover, Stolz, et al. (2011) recently published a fascinating corpus-based study3 of full reduplication in Europe, which shows areal patterning with respect to the presence and absence of reduplication and with respect to certain functions. It is conspicuous that these areal patterns of reduplication do not correspond to the distribution of other features that characterize established linguistic areas in Europe. For this article, I have investigated whether reduplication might be an areal phenomenon in one of the linguistic areas of South America, the GuaporéMamoré region. This region forms a part of the southwestern Amazon that is defined by the basins of the Guaporé and Mamoré Rivers, the main tributaries of the upper Madeira River. This is the border region of Bolivia and Brazil, and it harbors an exceptional linguistic diversity. About the size of Germany, the Guaporé-Mamoré region is the original habitat of indigenous peoples who speak over 50 languages belonging to seven different linguistic (macro-)families, including ten languages that appear to be isolates. The region has been degraded by intense Western cultural and ecological pressure, and the indigenous languages and cultures are relatively under-documented and threatened by extinction. Even though the genetic and typological linguistic diversity in the region is high, descriptive linguistic studies, many of which have become available only recently, show a number of lexical and grammatical similarities between (genetically unrelated) languages. On that basis Crevels and van der Voort (2008) hypothesize that the region is a linguistic area. It is not quite clear yet what the limits and defining features of the GuaporéMamoré linguistic area are. Nevertheless, certain specific grammatical features that appear to have diffused in the region indicate several minor linguistic areas, which form a conglomerate that may define the Guaporé-Mamoré linguistic area. General features include minimal shared vocabulary, complex verbal morphology, evidentials, directionals, an inclusive/exclusive distinction, and a lack of classifiers. Sub-areas can be distinguished on the basis of features,

3 My lack of access to this study was compensated by Kim’s (2012) substantial book review.

430

van der voort

such as, perhaps surprisingly, classifier systems and forms, specific possessive constructions (cf. Crevels and van der Voort 2008), specific strategies for the expression of tense and modality (cf. van der Voort 2013), and certain lexical forms (cf. van der Voort 2005). Some features seem to extend beyond the boundaries of the Guaporé-Mamoré area and are characteristic of a larger linguistic area, such as certain valency-marking constructions and forms, and, again, certain lexical forms. Productive reduplication is not attested to similar degrees throughout the region but seems determined by genetic contingency. In some families and isolates it is prolific; in some it is limited. Most languages of the GuaporéMamoré area have productive reduplication to some extent, but there does not appear to be evidence for areal diffusion of patterns or functions of reduplication in the region. Many similarities that one encounters can either be found elsewhere and represent universal characteristics of reduplication, or they are characteristic of a language family and probably originate in genetic diffusion. Certain properties of reduplication attested in the Guaporé-Mamoré region are not found elsewhere, but they are limited to single languages and rather represent examples counter to areal diffusion. As an illustration, in Movima (isolate) reduplication has some typologically unusual functions, not found elsewhere in the region, and perhaps not even outside of the region. Another example that may disprove areal diffusion of reduplication is found in Kwaza (isolate), which shares recursive person marking with several unrelated languages but which also has reduplicative person marking that seems to be a cross-linguistic unicum (cf. van der Voort 2009). Rare properties of reduplication such as in Movima and Kwaza have likely emerged independently from the influence of other languages. The fact that they are hardly—or not at all— encountered elsewhere in the world shows us once again the importance of thorough descriptive research of every single language. In many sources the terms progressive, durative and continuative/continuous are used indiscriminately. In accordance with their definitions in standard typological literature on aspect (e.g. Bybee, at al. 1994, Comrie 1976, Dahl 1985, Timberlake 2006), I have distinguished them in this article: progressive for ongoing events (“is walking”), continuative for events that go on longer than expected (“keep on walking”) and durative for delimited events (“walk for awhile”). Note that the distinction between these aspects does not appear to be marked systematically by reduplication in the languages discussed in this article. Other aspectual notions distinguished here are iterative/repetitive for events that are repeated cyclically in identical manner, and frequentative for events that occur very frequently during a certain time. The term distributive refers to the scattering of events or items in time or space, respectively. The

is reduplication an areal feature?

431

terms intensive and attenuative refer to opposite aspectual notions. The term pluractionality (see e.g. Corbett 2000) refers to verbal number and can express lexical or grammatical aspectual notions such as repetitive, distributive, frequentative, and other types of event (or event-internal phase) plurality as well as participant plurality (which is different from argument agreement, which represents nominal number).

2

Reduplication in Guaporé-Mamoré Languages

The following subsections contain short descriptions of reduplication in the following languages and language families of the region: Arawak, Chapacura, Macro-Jê, Nambikwara and Tupi languages, and the isolates Aikanã, Cayuvava, Itonama, Kanoê, Kwaza, Mosetén and Yurakaré.4 For a number of GuaporéMamoré languages the reader is referred to other articles in the present volume, although some basic observations will be made. This concerns Trinitario (Arawak, Rose this volume), Cavineña (Pano-Tacana, Guillaume this volume), Karitiana (Tupi, Storto this volume), Yuki, Sirionó and Makurap (Tupi, Dietrich this volume) and Movima (isolate, Haude this volume). For certain languages of the region, including the Iranxe, Chiquitano and Pano languages, no information on reduplication was available to me. For Canichana hardly any descriptive information exists at all. The final section of this article contains a table that summarizes the properties of reduplication in the languages discussed here. For the general typological characteristics of individual languages, either grammatical or phonological, the reader is referred to the sources cited in this article. 2.1 Arawak Several Southern Arawak languages are spoken in the region. Reduplication in the highly endangered Baure language of Bolivia has been discussed by Danielsen (2007; p.c. 2011). It is not productive with nouns, but both full and partial reduplication are attested with the function of marking intensification of verbs, adjectives and certain adverbs. More unusual is partial reduplication of adverbialized pronouns, such as roti ‘he’:

4 Data on Aikanã, Arikapu, Djeoromitxi and Kwaza are from my own fieldwork. Information on the other languages is from published sources and personal comments. Unless otherwise indicated, an apostrophe (') in the linguistic forms represents a glottal stop.

432 (1)

van der voort

roti~ti-ko-e' he~red-abs-emph ‘he all alone’ (Danielsen 2007, 321)

Another very interesting possibility is that the reciprocal suffix -koko- in Baure has originated historically from reduplication of the verbal stem-final ‘absolute’ suffix -ko (Danielsen 2007, 244). The same pattern was also observed by Danielsen (p.c. 2011) in many other Arawak languages, so this development must have taken place at a proto-Arawak stage. Danielsen (2007, 133) also mentions productive reduplication of the augmentative suffix -ča and diminutive -či, with an intensifying function. Reduplication of bound morphemes is a very rare phenomenon cross-linguistically, and most examples of it actually represent reduplication of part of the derived stem of a word and cannot be distinguished from common partial reduplication. Danielsen (2007, 133) makes the interesting suggestion that this diminutive and augmentative reduplication could reflect influence from Spanish, where the diminutive is sometimes reduplicated. In Trinitario (see Rose this volume) partial reduplication marks verbal repetition or attenuation. Reduplication does not seem to be productive with other lexical categories. Trinitario is quite unusual because full reduplication is not unambiguously attested, which goes against Moravcsik’s (1978, 328) universal that full reduplication is a prerequisite for the existence of partial reduplication. 2.2 Chapacura The best described language of the Chapacura family is Wari.’ The grammar by Everett and Kern (1997) contains separate morphological and phonological sections on reduplication. Wari’ has full, partial and complex reduplication, and it appears to be quite productive. It occurs on verbs, verbal modifiers, inflectional clitics, demonstrative pronouns, and nouns. According to Everett and Kern (1997, 377), partial reduplication of verbs is one of the ways to create a “plural/intensive” form of the verb. Reduplication of transitive verbs concerns the CV of the stressed syllable: wac ‘cut (sg)’ > wa~wac ‘cut (pl).’ With intransitive verbs, an extra rV syllable is added to the reduplication: cat ‘break (sg)’ > cara~cat ‘break (pl)’; hwet ‘appear (sg)’ > hwere~hwet ‘appear (pl).’ Although there is no explanation of what is meant by “plural verbs,” many examples throughout the book suggest this concerns participant plurality:

is reduplication an areal feature?

433

(2) mao na go 3sg.rp/p ‘He went.’ (Everett and Kern 1997, 328) (3) ma~ma' nana red~go 3pl.rp/p ‘They went.’ (Everett and Kern 1997, 328) Some examples suggest event plurality: (4) cat na pana break 3sg.rp/p stick ‘The stick broke.’ (Everett and Kern 1997, 328) (5) cara~cat na pana red~break 3sg.rp/p stick ‘The sticks broke.’ (Everett and Kern 1997, 328) In view of the provided free translation in example (5) one would expect the plural inflectional particle nana instead of singular na.5 Also the notion “intensive” is not explained in the section on reduplication, but elsewhere (1997, 203) the “intensive” meaning of verbs is shown both in partial and full reduplication: cao' ‘eat’ > ca~ca~cao' ‘really eat’;6 'aca ‘cry’ > 'aca~'aca ‘really cry.’ Repetitive aspect is expressed iconically by full reduplication of the verb (1997, 316). Progressive aspect in the present tense can also be expressed by verb reduplication, although that is apparently not obligatory. In certain types of clauses, tense/mood clitics occur on the verb. Such clitics are usually omitted under reduplication of the verb. There is no meaning difference between a sentence involving such a clitic and its reduplicated counterpart. (6)

je 'i ma' cao' mao cao' mao caca 'oro wari' emph.3n n that.prox.hearer eat go eat go 3pl.m coll person ‘That is what the people eat.’ (Everett and Kern 1997, 53)

5 So perhaps example (5) refers to a plural event rather than a plural subject. 6 Sometimes, the CV syllable of transitive verbs is reduplicated twice. See also Key (1965, 93) who registered full triplication: kao'~kao'~kao' ärät ‘We ate and ate.’

434 (7)

van der voort

je 'i ma' ca cao' caca 'oro wari' emph.3n n that.prox.hearer infl.n.rp/p eat 3pl.m coll person ‘That is what the people eat.’ (Everett and Kern 1997, 53)

Reduplication is not only attested with verbs. Several third person plural inflectional clitics are based on reduplication of the third person singular forms (1997, 324, 415). As an example, (3) contains the third person plural particle nana, which originates from reduplication of third person singular na. Furthermore, the plural demonstrative pronoun ca~ram is formed through complex reduplication of the singular feminine demonstrative cam (1997, 299–300, 415). With regard to nominal bases, full reduplication can be used to derive names or descriptive terms from nouns, e.g., homa homa ‘fish species that is very fat’ (lit. ‘fat fat’); capija capija ‘talker’ (lit. ‘mouth mouth’) (1997, 379). In Oro Win, a closely related Chapacura language, it also seems that reduplication concerns mainly the first mora of verbs. Popky (1999) states that it indicates plural subject. Her thesis, which is based on several hours of field recordings made by Daniel Everett, includes some examples: tyt > ty~tyt ‘walk,’ mao > ma~ma' ‘go,’ tok > to~tok ‘drink’ (Popky 1999, 24). Popky notes that verb reduplication (both full and partial) may also indicate progressive aspect (1999, 31): (8) wirikam ma~ma' na 3f.emph red~come 3sg.rp/p ‘She’s coming.’ (Popky 1999, 31) Joshua Birchall (p.c. 2012) provided some examples from his own fieldwork that suggest that reduplication in Oro Win may furthermore indicate plural object: (9)

jyk kara~kat n-an pana push red~break 3sg.nfut-3n wood ‘He felled quite many trees.’ (Birchall p.c. 2012)

Perhaps certain Oro Win examples could be interpreted as pluractional, like some of the Wari’ examples. Less is known about reduplication in Moré, another Chapacura language. In the description by Angenot-de Lima (2002), full reduplication of the verb root has a repetitive effect. In some of the examples (represented here in a phonemicized orthography) reduplication appears to have an attenuative effect:

is reduplication an areal feature?

435

(10) kiw anãɲ pua bite 1.n.impf papaya ‘I’m eating a papaya.’ (Angenot-de Lima 2002, 340) (11) kiw kiw anãɲ pua bite bite 1.n.impf papaya ‘I’m nibbling a papaya.’ (Angenot-de Lima 2002, 340) 2.3 Macro-Jê Several Macro-Jê languages are spoken in the wider Guaporé-Mamoré region. Reduplication in Rikbaktsa of western Mato Grosso has been described in detail by Silva (2011). Reduplicative patterns observed with nouns are all inherent and mainly concern onomatopoeic animal names. Productive reduplication is attested with verbs, where it may indicate repetitive, progressive and habitual aspects and plural argument (subject or patient). There does not seem to be a functional division between full and partial reduplication. Partial reduplication concerns the first mora of a verb root, as in: bok ‘to fly’ > bo~bok ‘to fly (pl subject)’; parak ‘to run’ > pa~parak ‘to run (pl subject)’; boro ‘to eat’ > bo~boro ‘to bite’ (Silva 2011, 168). Full reduplication has similar functions, as in: dok ‘to arrive’ > dok~dok ‘to arrive (pl subject)’; kore ‘to pound’ > kore~kore ‘to pound (progressive)’; kodore ‘to see’ > kodore~kodore ‘to visit’ (Silva 2011, 169). The only function that does seem to be limited to one type of reduplication is imperative mood. Full reduplication of the verb root is obligatory with negative imperative expressions and optional with imperative: (12) piknu=tɨ hawk~hawk=bɨi=tɨ fish=acc eat~eat=neg.stat=imp.sg ‘Don’t eat fish!’ (Silva 2011, 173) (13) (kore~)kore=tɨ ktʃa pound~pound=imp.sg mod.emp ‘Pound (the fruits)!’ (Silva 2011, 173) In the Jabuti languages of Rondônia, reduplication of verb roots may symbolize repetitive events. In Arikapu no productive reduplication of nouns was attested. With verbs both partial and—less frequently—full reduplication are attested although they are not highly productive. There are many instances of reduplication where there does not seem to be much of a semantic difference between the reduplicated and non-reduplicated versions (which might be a reason not to regard this as reduplication at all). The verb roots it concerns

436

van der voort

usually refer to certain kinds of inherently or internally repetitive events such as sniffing, running, cutting, digging etc. Full reduplication was attested mainly with monosyllabic roots, e.g. hə̃ or hə̃~hə̃ ‘to split or cut firewood’; ho or ho~ho ‘to pierce’; to or to~to ‘to run after (e.g. a dog running after someone),’ including bimoraic roots, e.g. taj or taj~taj ‘play peteca (a type of hand-badminton) or ball’; mãj or mãj~mãj ‘to tie, roll up’; waj or waj~waj ‘to dig’; raj or raj~raj or rə~raj ‘to clean, sweep up.’ With disyllabic roots mostly partial reduplication was attested: nĩkrə̃ or nĩkrə̃~krə̃ ‘wipe (some)one’s ass’; kuhu or kuhu~hu ‘(to be) big’; rəjtxi or rəjtxi~txi ‘to be old’; ndukuho or ndukuho~ho ‘to vomit.’ In many of these instances, the reduplicated form represents the default; whereas, its non-reduplicated counterpart is rare. For example, the verb for the shaman’s practice of sniffing hallucinogenic paricá powder is ĩ but usually attested as ĩ~ĩ ‘to sniff.’ Reduplication of the verb root tɨ ‘to close (i.e. shutting off an opening, a hole)’ appears to indicate object plurality although that seems to be the case only when contrasted with the non-reduplicated form. There are, nevertheless, a number of reduplicative expressions that are semantically significant, e.g. with a repetitive meaning: ku ‘to bite, to ingest’ > ku~ku ‘to bite repeatedly, to sting (e.g. be stung by a swarm of wasps)’; tɨ ‘to crush (maize for flour)’ > tɨ~tɨ ‘to pound crushing (maize for flour)’; wa ‘to cut’ > wa~wa ‘to cut up, to cut (hair)’; paj ‘to take, catch, find’ > paj~paj ‘to search for’: (14) txitxi-hə̃ paj~paj maize-seed take~take ‘to be searching (for maize grains)’ Semantic effects of reduplication do not seem to represent a general pattern and probably depend on the contents of the particular verb root. Full reduplication of tənãj ‘to be crooked, bent, coiled’ has a repetitive meaning: tənãj~tənãj ‘to be repetitively crooked (e.g. a tree or an undulating line of body paint),’ while partial reduplication is intensifying: tənãj~nãj ‘to be entirely crooked.’ Reduplication of ro ‘to grow’ can be either transitivizing or nominalizing: ro~ro ‘to raise (an animal or child),’ ‘domesticated animal (lit. ‘raised one’).’ Verbalizing reduplication was attested with two nouns: ku ‘wood, tree, stick’ > ku~ku ‘to (be) fasten(ed) to wood or a tree’; (probably) nĩre ‘rope, yarn’ > nĩ~nĩre or nire~re ‘spin yarn by rolling on one’s thigh’: (15) nũrə̃ txiwə a-ni~nĩre or (16) ta-nire~re txiwə palm.sp braid 2-red~rope 3.cor-rope~red braid ‘You’re braiding yarn of palm fibre.’ ‘He is braiding yarn.’

is reduplication an areal feature?

437

There is very little information on reduplication in Djeoromitxi. Some examples suggest an intensifying function: tə ‘small’ > tə~tə ‘very small’ and perhaps todʒo~todʒo ‘to be crooked.’ Intensification might also explain the repeated second person pronoun adʒɛ ‘you (sg/pl)’: (17) adʒɛ~adʒɛ hõtõ-mã you~you play-fut ‘Are you (pl) going to play?’ Similar to Arikapu, there may not always be much of a semantic difference between the reduplicated and non-reduplicated forms: kuti or kuti~kuti ‘it is slippery.’ One example of verbalizing reduplication has been attested in Djeoromitxi: huwa ‘song’ > huwa~wa ‘to sing.’ The Chiquitano language of northeastern Bolivia and western Mato Grosso probably also belongs to the Macro-Jê family, but there is hardly any information on reduplication in the language. For a discussion of reduplication of a Macro-Jê language spoken outside of the region, see Salanova (this volume) on Mẽbengokre (Jê). 2.4 Nambikwara From the available sources on Nambikwara languages, productive reduplication seems to be a rather limited phenomenon. The language family has three main branches: Northern Nambikwara, Southern Nambikwara and Sabanê. There is almost no discussion of reduplication in Telles’ (2002) description of Latundê/Lakondê, a Northern Nambikwara language. Although she suggests that there is a morphological basis of reduplication in Latundê (2002, 31), the sparse cases attested in her book originate from sound symbolism (2002, 61; 170–171). In the section on ideophones she gives a handful of animal (mainly bird) names and points out that ideophones tend to have reduplicative structures and that they have a lexical aspectual value of ongoing or repetitive action. Ideophones may indeed express such aspectual values by symbolizing it through repetition of syllables or by imitating, e.g., repetitive bird calls through onomatopoeia, but that is not a productive morphological process. In Eberhard’s (2009) work on Mamaindê, another Northern Nambikwara language that is very closely related to Latundê/Lakondê, there is a full chapter (2009, 307–323)7 on reduplication. In the introduction to this chapter, Eber-

7 These are the page numbers of the online digital version. The page numbers in the printed version are 325–341.

438

van der voort

hard notes that most cases of ‘repetition of segmental material’ emphasize the meaning of a descriptive adjectival verb root, e.g., walo ‘rotten’ > walo~walo ‘very rotten’ and thã ‘soft’ > thã~thã ‘very soft.’ However, the rest of the chapter deals exclusively with the phonotactic aspects8 of reduplication and, lacking morphemic segmentation and minimal contrasts between reduplicated and non-reduplicated forms, it provides no grammatical or semantic analysis. From the few analyzed cases mentioned here it seems that reduplication in Mamaindê is similar to reduplication in the other Nambikwara languages. For Southern Nambikwara there is a minimal description of reduplication. Kroeker (2001, 71) states that reduplication occurs in attributive verb stems only when they lack an ‘instrumental prefix.’ He divides the verb stems into three classes: one which involves full reduplication, one which involves partial reduplication, and one without reduplication (even when the instrumental prefix is not attached). These different classes of attributive verbs are not shown to correspond to any semantic or pragmatic distinctions. Also, there is no discussion of possible semantic effects of reduplication. For Sabanê, Araujo (2004, 131) discusses only full verbal reduplication, suggesting in a footnote that there is no partial reduplication in Sabanê. He states that the distinction between predicative and attributive verbs (adjectives) can be observed in the semantic effect of reduplication (2004, 163–164). Reduplication of attributive verbal roots intensifies or emphasizes their meaning; whereas, reduplication of predicative verbal roots indicates repetitiveness. However, the few examples are not very convincing for such a strong statement. Surely the attributive verb stems are intensified: motu ‘round’ > motu~motu ‘perfectly round’; but is reduplication of siki ‘tickle’ > siki~siki ‘tickle too much/ tickle to death’ repetitive or intensifying? The distinction between intensity and repetition seems to depend crucially on the semantics of the particular verb root, e.g. hala ‘moan’ > hala~hala ‘sob’ (2004, 234), rather than on semantic subcategorization. Araujo (2004, 124) also mentions “reduplication” of a bound kinship morpheme -ko. The element is obsolescent but obligatory on kinship terms, which may suggest that -ko is in the process of lexicalizing together with the kinship term. This may necessitate the adding of another -ko in order to productively mark the terms as kinship terms. That would count against an analysis in terms of reduplication, which, by definition, is different from recursive application of a morpheme. The result may look like reduplication, but it is not a single productive operation.

8 Despite the fact that this chapter is located in the morphology section of the book.

is reduplication an areal feature?

439

2.5 Pano-Tacana The languages of the Tacana family are spoken exclusively in the Bolivian lowlands. Reduplication has been described for Cavineña by Guillaume (this volume and 2008). Both full and partial reduplication pervade the grammar of Cavineña in many different ways, and they can occur with various lexical categories with varying degrees of productivity. Some types of reduplication in Cavineña have well-known functions, such as repetition, distribution, and habituality. One of the cross-linguistically less commonly attested functions of reduplication that characterize Cavineña is the reduction of valence, which has antipassive effects and conveys reference to culturally codified activities. The Tacana language Ese Ejja (Vuillermet 2012) has a similar wealth of productive reduplication types and strategies, including valency reducing reduplication. Apparently unlike Cavineña, Ese Ejja allows reduplication of some members of its class of bound verbal aspectual suffixes, with an intensifying effect: (18) Pa-nei~nei-a=pa=pwa cry-very~very-rpas=rep=rpas ‘He cried a lot.’ (Vuillermet 2012, 495) (19) Xeki-kwaji~kwaji-kwe! go.get-fast~fast-imp ‘Go get (e.g. water) quickly!’ (Vuillermet 2012, 498) These suffixes, which convey adverbial-like notions of manner, may originate from independent lexemes or particles. The Pano languages are spoken in Bolivia, Brazil and Peru. Unfortunately, there are no descriptions of Pano languages spoken in the Guaporé-Mamoré region that include adequate examples of reduplication. 2.6 Tupi The Tupi language family most likely originated in the region that covers parts of Rondônia and western Mato Grosso since its internal diversity is mostly concentrated there. Reduplication in Karitiana, the last representative of the Arikém branch, is dealt with by Storto (this volume). In Mekens, also known as Sakurabiat, which is one of the five languages of the Tupari branch, productive reduplication mainly concerns verbs. Galucio’s (2001) description of Mekens contains several observations and a brief section (2001, 104–105) on reduplication. In principle any verb root can be reduplicated and can apparently express repetitive and progressive meaning, the latter as in

440

van der voort

set ‘go’ > set~set ‘She walked and walked’ (2001, 66). Depending on the context, the sense of ‘rapidly,’ for which Mekens does not have an adverb, can also be expressed by reduplication. Triplication seems to be common and expresses the same meanings as reduplication, as in ko ‘ingest’ > ko~ko~ko ‘eat a lot, eat rapidly’ (2001, 66). Note also the following examples: (20) sɨgɨ~sɨgɨ~sɨgɨ te pe=kɨmakãy raise~raise~raise foc obl=soil ‘He started to raise up soil dust.’ (Galucio 2001, 104) (21) poret ko~ko~ko te pe=o-iko then eat~eat~eat foc obl=1sg-food ‘Then I ate quickly.’ (Galucio 2001, 104) One example suggests that verb reduplication can also express plural events: (22) ɨa sɨk~sɨk pay lagoon stick~stick leave ‘It stuck them (the sticks) in the lagoon, and left (them) there.’9 (Galucio 2001, 105) Only very few instances of non-verbal reduplication were attested. The noun paak~paak ‘heron’ is derived through reduplication of the adjective paak ‘white,’ in clear reference to the bird’s white plumage (Galucio 2001, 28; 105). There are emphatic forms of the first and second person singular pronouns that derive from reduplication: õt ‘I’ > õr~õt ‘I (emphatic)’; ẽt ‘you’ > ẽr~ẽt ‘you (emphatic)’ (Galucio p.c. 2012). Partial reduplication has not been attested in Mekens. The Tupi-Guarani branch is the most expanded subfamily of Tupi and has many representatives in various South American countries. A survey of reduplication in Tupi-Guarani, contrasted with certain other Tupi languages, is presented in Dietrich (this volume) and includes Yuki and Sirionó (Bolivian Tupi-Guarani) as well as Makurap (Tupari). Several languages in this survey confirm earlier observations by Rose (2005) for a subset of Tupi-Guarani languages, in which different phonotactic reduplication patterns may correspond to pluractionality distinctions.

9 The example concerns two sticks, which were stuck into the bottom of the lagoon one after the other, to function as hammock poles (Galucio p.c. 2012).

is reduplication an areal feature?

441

2.7 Aikanã Aikanã is a fragmentarily described language isolate spoken by some 175 individuals on the Brazilian side of the Guaporé river. Reduplication seems to play a somewhat limited role in Aikanã and mainly concerns full reduplication of verb roots. In the survey by Kramer (1993),10 many examples of reduplication are cited as a productive means to express repetitive, progressive, and intensity. This is confirmed by my own field data. It appears that productive reduplication of verb roots often conveys a progressive meaning:11 (23) bi~bi-ke-ẽ write~write-3-dec ‘He is writing.’ (24) tẽtẽĩ kaw~kaw-ke-düka-ã-ẽ12 gnat bite~bite-3-cl:leg-1sg.o-dec ‘The gnat(s) is (are) biting my leg.’ (25) aityü~aityü-ne-ẽ change~change-cmpl-dec ‘It is changing bit by bit.’13 Repetitive, distributive and continuative, respectively, can also be expressed by reduplication: (26) bu~bu-he-dukari-ẽ epa~'epa-he-dukari-ẽ dance~dance-3–3pl-dec try~try-3–3pl-dec ‘They are practicing dancing.’

10

11 12 13

This is from a class report based on the fieldnotes of Harvey Carlson, a linguistics student at the University of California at Berkeley who spent some months among the Aikanã in 1984. The examples in the reduplication section include many instances of onomatopoeia and, since the examples are not segmented and analyzed, also many cases of pseudoreduplication, such as zõka-ka-naka-re-ka-ẽ (wash-1sg-cl:cloth-fut-1sg-dec) ‘I’m going to wash clothes’ (for the future tense construction with recursive person marking, see van der Voort 2009, 2013). As also observed by Vasconcelos (1996, 74). I have used a practical spelling of Aikanã in which ⟨'⟩ corresponds to IPA [ʔ], ⟨y⟩ to [j] and ⟨ü⟩ to [y ~ ∅]. The verb root aityü- means ‘to change, be different’ especially with regard to color.

442

van der voort

(27) hiba~hiba-'eye-ẽ urikai mama'ĩ-ye give~give-3pl.o-dec food chicha-obl ‘They are distributing food and chicha among the guests.’ (28) yoa~yoa-me-re-te-'ẽ cry~cry-2sg-fut-neg.imp-imp ‘You shouldn’t keep crying!’ Full verb root reduplication is also attested with a connotation of intensity: (29) ware~ware-'ẽ walk~walk-imp ‘Come on, walk!’ (30) hane~hane-pa-waru-a-ẽ burn~burn-cl:big-tongue-1sg.o-dec ‘It is burning my tongue.’ Color terms are expressed by verbal roots and the majority follow the pattern (C)VrVrV: hürürü- ‘green, blue,’ ururu- ‘brown, grey’; arara- ‘white’; parari- ‘yellow.’ Non-reduplicated versions of these verb roots are not known to exist, but it is conspicuous that these roots combine membership of such a semantically specific set with such similar syllable patterns. If their forms did not emerge out of a diachronic process of analogy, they could have originated from partial reduplication, involving two rV copies of the first syllable.14 2.8 Cayuvava In his sketch of the morphology of Cayuvava, a moribund language isolate of Bolivia, Key (1967) describes a number of mainly aspectual changes caused by different types of verb root reduplication (see also Crevels and Muysken 2012, 362–363). Unfortunately, the translations given by Key (1967, 25–27) are not contrasted and represent either only the meaning of the non-reduplicated form or only the meaning of the reduplicated form. Also, the few Cayuvava examples in an earlier survey article (Key 1965) are not helpful for resolving the puzzles. In the present subsection the most productive types are mentioned, presenting

14

Conspicuously similar to complex rV reduplication in Wari’. Kramer (1993) also noticed this pattern but was not able to find a semantic connection, because his survey includes many accidental instances.

is reduplication an areal feature?

443

only those examples for which the contrasting meanings can be inferred with reasonable confidence. One of the types of verb reduplication involves reduplication of the initial mora of the root and indicates a progressive state,15 as in: pe > pe~pe ‘state of waiting’; tïbï > tï~tïbï ‘process of awakening.’ Here we may assume that pe means ‘wait’ and tïbï ‘awaken.’ Another type involves reduplication of the second syllable of the root and expresses ongoing action: tïdï > tïdï~dï ‘carrying on shoulder’; põẽñẽ > põẽñẽ~ñẽ ‘(something) filling it.’ Probably tïdï means ‘carry on shoulder’ (Key 1965, 91: todo ‘carry in arms’ > totodo ‘carrying in arms’) and põẽñẽ ‘fill it.’ Yet another type of reduplication that expresses ongoing action involves full repetition of the verb root: ado > a'do~ado ‘wringing out in cloth’; rebe > re'be~rebe ‘blinking, opening eyes.’16 In addition to these relatively canonical types of reduplication, Key (1967, 19; 25) claims that inceptive aspect and causative are indicated by reduplication of the verb root’s last syllable, of which the initial consonant is represented by /r/.17 Regrettably, Key cites examples in this manner: /ße/18 plus /re/ = /ßere/ ‘run’ (1967, 25). One would assume that this implies ße ‘run’ > ße-re ‘start/cause to run.’ However, in his article (Key 1965, 91) ßere ‘run’ is contrasted with ßeßere ‘running.’ So, one wonders what the meaning of the base is. Similar examples are ro > ro-ro ‘carry’; hu > hu-ru ‘dig,’ ußæ̃ > ußæ̃ -ræ̃ ‘flavor.’ Only at the end of the reduplication section is a real contrastive example provided (1967, 27): je ‘move’ > je-re ‘cause to move.’ The analysis of this causative construction as (complex) reduplication, however, is not the most plausible. Considering that causative is an unusual function of reduplication, that Cayuvava does not appear to have another causative construction, and that Cayuvava is a language that displays vowel harmony phenomena, the rV syllable should probably just be analyzed as a dedicated causative morpheme. In case it, nevertheless, should be analyzed as complex reduplication, it would strengthen the hypothesis raised in the final section that valency-changing reduplication is an areal feature of the Guaporé-Mamoré region. Unfortunately, Cayuvava being practically extinct, it may not be possible to confirm the reduplication analysis.

15 16 17 18

“continuative state” in Key’s (1967, 25) terms. The apostrophe (’) in some of the Cayuvava examples indicates irregular or contrastive “strong stress” (Key 1967, 19–20). Which would be reminiscent of complex reduplication of intransitive verbs in Wari.’ ⟨ß⟩ in Key (1965) corresponds to ⟨b⟩ in Key (1967).

444

van der voort

According to Key, there is an ‘augmentative’ variant of his supposed causative reduplication that is expressed by reduplication of the verb’s final rV syllable. However, the examples are not contrastive and one can only speculate about the meaning of the non-reduplicated forms. For the example ße-re~re ‘flee’ we could speculate that the meaning of the root ße is ‘run.’ Apparently, some of the types of reduplication described above can be combined. One contrastive example is given: je ‘move’ > je-re ‘cause to move’ > je-re~je-re ‘moving right along’ > je-re~' je-re~re ‘cause to move along, to hurry one along.’ Key (1967) considers productive morphological processes expressing aspectual distinctions as inflectional. Therefore, the types of reduplication discussed above are considered by Key as inflectional operations. Less productive reduplications, such as those for which no synchronic non-reduplicated form was attested, are considered as derivational by Key (1967, 26). Productive nominal reduplication is not attested. The repetitive syllable structures encountered in Cayuvava, which are considered as derivational reduplication by Key (1967, 42), are sound symbolic and only attested with names of animals and certain natural phenomena. 2.9 Itonama There is no survey dedicated to the properties and uses of reduplication in the language isolate Itonama. However, from Crevels (2006; 2012) it becomes clear that partial reduplication has several important functions in this language. In Itonama, where number is mainly a verbal category, event number is expressed in various ways: by a distributive marker, by a pluractional marker, or by reduplication of part of the verb stem, as in example (31): (31) wa'ihna pini bo~bo'-te'-ka pa-nay-bo~bo'-cha md she red~dance-cnt-f.sg 3sg.f-sub-red~dance-cnt ‘And she danced and danced.’ (Crevels 2012, 283) Often, pluractional reduplication occurs in combination with the intensifier infix -ʔV- and andative infix -hV- (the vowels harmonize with the verb root). Note examples (32) and (33): (32) ohni ni-su~⟨h-u⟩suh-ne kay-chadï-ne-'o he foot-red~⟨and-ints⟩smell-neu face-find-neu-rep ‘He kept smelling his tracks and got him again.’ (Crevels 2006, 165)

is reduplication an areal feature?

445

(33) sosohte yumani ya⟨ka~⟩⟨'a⟩ka-ne'-ka wabï'ka all night sing⟨red~⟩⟨ints⟩-neu-f.sg woman ‘The woman sang every night.’ (Crevels 2006, 165; 2012, 271) This kind of pluractional reduplication is prefixed when applied to monosyllabic verbs (as in 32) and infixed before the last syllable with polysyllabic verbs (as in 33).19 It is often found with the specific meanings repetitive, progressive and continuative, depending probably on the lexical properties of the verb and on the pragmatic context. Example (34) shows infixation in a trisyllabic verb root: (34) mi-wahdi'-ke'-ka mi-kahana'-na'-ka may-yapï⟨la~⟩⟨'a⟩la'-cha'-k'a rel-sick-pl-f.sg rel-old-neu-f.sg sub-hurt⟨red~⟩⟨ints⟩-pl-ben pi-kas-no pi-bu-nu 3sg.f-side-nmloc 3sg.f-belly-nmloc ‘The sick old woman suffered from pain in her side, her belly all the time.’ (Crevels 2012, 271) In combination with reduplication, the infixed elements -ʔV- ‘ints’ and -hV‘and’ convey the meanings of repeated event and repeated motion, respectively. Because these infixed elements are possibly only attested in combination with reduplication (Crevels, p.c. 2013), it is tempting to analyze the entire combination as a case of complex reduplication,20 with an intensifying and progressive semantic effect. Another function of reduplication in Itonama is the expression of participant number. Note the following contrasting examples: (35) sih-k'i-ma-doh-ne upa'u 1pl.excl-inv-hand-bite-neu dog ‘The dog bit us on the hand.’ (Crevels 2006, 166; 2012, 271) (36) sih-k'i-ma-do~doh-ke upa'u 1pl.excl-inv-hand-red~bite-pl dog ‘The dogs bit us on the hand.’ (Crevels 2006, 166; 2012, 271) 19

20

One might wonder whether it is necessary to analyze cases of reduplication like (33) as infixing; however, the phonotactic distribution of Itonama negation and consecutive marking tends to follow a similar pattern. I.e., reduplication which involves additional phonological change (cf. Rubino 2005, 15; 18), similar to rV infixation in Wari’ (see section 2.2).

446

van der voort

Crevels (2006, 167; 2012, 272) explains that partial reduplication of the verb stem in (36) signals that more than one dog was involved in the event, whereas in (35) just one dog is involved. The pluractional suffix -ke indicates plurality of the event, and without the reduplication the free translation of sihk'imadohke upa'u would be ‘The dog bit us (several times) on the hand.’ In brief, in Itonama, partial reduplication denotes among other things event plurality and can represent one of the ways verbal number is expressed. Furthermore, partial reduplication may denote participant plurality and, therefore, represents one of the ways to express nominal number. Another function of partial reduplication of (attributive) verbs is intensification. Note example (37): (37) sewa-ne ihwana biluwa napa-bï⟨ya~⟩yah-na see-neu Juan snake length-be.big⟨red~⟩-neu ‘Juan saw a very long snake.’ (Crevels 2012, 252) Reduplication of nouns is reported as rare and is not attested unambiguously in the available sources. Full reduplication of locative and directional prefixes, which can function as nominal or verbal roots, is attested with an intensifying meaning and occurs often with reference to body parts: (38) uh-pa~pa-no 3-up~up-loc ‘the highest point (of the roof)’ (Crevels 2012, 276) (39) uh-pa-kos-cho~cho-no 3-up-side-inside~inside-loc ‘inside of the throat’ (Crevels, p.c. 2013) There is no evidence for polysyllabic full reduplication in Itonama. 2.10 Kanoê In the language isolate Kanoê described by Bacelar (2004) full verb root reduplication expresses frequentative aspect, progressive aspect, or intensity. The following examples illustrate these respective meanings.21

21

In Bacelar’s practical spelling ⟨j⟩ corresponds to IPA [j] and ⟨y⟩ to [ɨ ~ ə].

is reduplication an areal feature?

447

(40) aj oroe mana~mana õ-e-re I clay squeeze~squeeze 1-dec-aux ‘I’m kneading clay.’ (Bacelar 2004, 102) (41) kamitsi ivo mũ~mũ e-re yesterday yam plant~plant dec-aux ‘Yesterday I was planting yams.’ (Bacelar 2004, 221) (42) aj mo~mo-õ-kỹj e-re I sleep~sleep-1-cl:eye dec-aux ‘I slept a lot.’ (Bacelar 2004, 225) There is one example of an adverb reduplicated for emphasis: (43) oj tunemu vi~vi koro-e-to n-e-re he poison alone~alone commit.suicide-asp-tra 3-dec-aux ‘He poisoned himself.’ (Bacelar 2004, 212) There is no discussion of partial reduplication as a productive process. The only example in Bacelar’s (2004) description that shows a partial reduplicative pattern is onomatopoeic. 2.11 Kwaza Reduplication in the language isolate Kwaza was described in van der Voort (2003; 2009) and is briefly summarized here. In Kwaza, both full and partial reduplication occur with various functions, most of which are iconic and concern temporal and aspectual distinctions. Full repetition of verbal and adverbial roots often has repetitive, progressive, intensifying and attenuative meaning. Consider the following examples: (44) hãidi~hãidi-tse drip~drip-dec ‘It is dripping.’ (45) watxi~watxi ãw󰀟̃i-ra true~true see-imp ‘Aim right!’; ‘Look well!’

448

van der voort

(46) kuri~kuri-da-lɛ akɨ-da-tsɨ-tse stop~stop-1sg-prec take.food-1s-pot-dec ‘I’ll take food a little later (you go first).’ Productive reduplication of nominal roots is very rare. Reduplication of the interrogative pronoun dilɛ ‘who’ conveys the sense of ‘who else (besides you).’ Cross-linguistically, the reduplication of ‘who’ often has an indefinite meaning, such as ‘whoever’ (Moravcsik 1978, 319). Partial reduplication of verbal and adverbial roots may occur also with repetitive, frequentative, progressive, and intensifying meanings. Usually the first syllable is reduplicated, and occasionally the final syllable is reduplicated, but these different types are not associated with any semantic or functional division. Note the following contrasting examples: dote- ‘to shed’ > do~dote ‘to leak’; cari ‘to shoot’ > ca~cari ‘to shoot many’; esa- ‘to sting’ > e~'esa- ‘to sting frequently/by many (e.g. a swarm of wasps)’; durɨ- ‘to roll’ > durɨ~rɨ- ‘to be rolling by itself’; anũ- ‘to plant’ > a~'anũ- ‘to be planting’; txarwa ‘now, first’ > txa~txarwa ‘just now.’ Sometimes, full and partial reduplication were attested with the same verb root, resulting in different meanings: kahɛ- ‘to bite’ > intensive kahɛ~kahɛ- ‘to keep on biting (ferociously)’ > attenuative ka~kahɛ- ‘to bite playfully.’ Perhaps the analysis of partial reduplication in Kwaza should be based on the mora and not on the syllable. In the case of heavy syllables, apparently only the first mora is reduplicated. The following examples may have a continuative and/or distributive sense: hẽu- ‘to sniff’ > he~hẽu- ‘to keep sniffing/smelling (around)’; hou- ‘to take’ > ho~hou- ‘to gather, to collect.’ There are some instances of partial reduplication (on either side of the root, either syllable or mora) as a means to indicate participant number: hãrã- ‘to stop’ > hãrã~rã- ‘to stop (many)’; bui- ‘to leave, go outside’ > bu~bui- ‘to leave, go outside (many)’; tsje- ‘to grab’ > tsi~tsje- ‘to grab many.’ The reduplicative processes discussed above are all based on lexical or phonotactic boundaries and have the familiar semantic effects known from many other languages. One of the rarest phenomena in reduplication probably is morphologically-based reduplication of bound morphemes. Kwaza is the only example encountered so far of a language displaying this phenomenon in a consistent and indisputable manner. Thus, apart from normal phonotacticallybased reduplication, Kwaza also has reduplication of bound person markers to indicate remote past tense or habitual aspect.

is reduplication an areal feature?

449

(47) kukui-h󰀟̃-da-ki ill-nom-1sg-dec ‘I was ill.’ (48) kukui-h󰀟̃-da~daɨ-h󰀟̃-ki ill-nom-1sg~1sg-nom-dec ‘I was ill some time ago.’22 (49) aw󰀟̃i-ta-xa-h󰀟̃-ki see-1sg.o-2-nom-dec ‘You saw me.’ (50) aw󰀟̃i-ta~ta-xa-h󰀟̃-ki see-1sg.o~1sg.o-2-nom-dec ‘You saw me a long time ago.’ Bound morpheme reduplication in Kwaza is not a contingent phenomenon of phonotactically-based syllable reduplication, in which case the reduplicated part just happens to coincide with a bound morpheme. Reduplication of bound person markers with different syllable structures represents the evidence that it is truly determined by morpheme boundaries. The following example shows reduplication of a disyllabic person marker:23 (51) aure-lɛ-nã-axa~'axa-le-h󰀟̃-ki marry-reci-fut-1exc~1exc-frust-nom-dec ‘We were going to marry (but we didn’t).’ Kwaza reduplication was described more extensively in van der Voort (2003). Van der Voort (2009) investigated the possibility of a relationship with the quotative construction in Kwaza, which also involves the repeated occurrence of bound person markers:

22

23

Note that for some persons allomorphs occur, e.g. -da- > -daɨ- ‘1sg.’ Since these allomorphs are independently motivated, one could argue that they here reflect simple, non-complex reduplication. Third person singular expressions are ambiguous with regard to past or remote past because third person is unmarked.

450

van der voort

(52) kukuih󰀟̃-da-ki=da-ki ill-1sg-dec=1sg-dec ‘I said that I was ill.’ However, there is a fundamental difference between the remote past tense and the quotative constructions. The remote past tense is created by reduplication of a person marker and the meaning of the resulting construction is not predictable on the basis of its constituent parts. In contrast, the quotative involves both recursion of the person marker and of its meaning, resulting in a semantically fully transparent construction. The quotative construction involving recursion of person inflection is also attested in Aikanã (isolate), Wari’ (Chapacura) and possibly in Kanoê (isolate). It is one of the probable areal features found in the Guaporé-Mamoré region, and there may be a relationship with similar constructions in the Andean languages. Reduplication of person markers is attested exclusively in Kwaza and does not resemble anything in any other language of the region, although there are cases of bound morpheme reduplication in other languages that may not be accidental. Intensifying reduplication of the adjective quantifier suffix in Karitiana (Storto, this volume), of certain manner aspect suffixes in Ese Ejja (Vuillermet 2012, 495) and of the augmentative and diminutive suffixes in Baure (Danielsen 2007, 133), as well as the reduplicative character of the reciprocal suffix in many Arawak languages (Danielsen 2007, 244, p.c. 2011) are intriguing and merit further research. 2.12 Mosetén Together with Chimané, Mosetén is a member of the small isolate language family Mosetén. This subsection is based on Sakel’s (2004) description of Mosetén. Both full and partial (CV24) reduplication are attested with all parts of speech in Mosetén, and they have many of the functions that are known from other languages, such as plurality and emphasis. Sakel’s description devotes several subchapters (2004, 57–59; 79–84; 266–272) to the specific patterns of form and function of reduplication in Mosetén. The semantic outcome of reduplication depends on the part of speech among other things. Furthermore, 24

From the examples throughout Sakel (2004) it seems that the general tendency in partial reduplication is to copy the CV (the first mora) of the last syllable of the root, the derived stem, or the entire word. The question of which syllable is subjected to reduplication may depend on the particular stage of derivation at which reduplication is applied. Consequently, it may be the order of morphological operations that determines which syllable of the derived word is subjected to reduplication. This question merits further research.

is reduplication an areal feature?

451

full and partial reduplication appear to express different types of meaning. Full reduplication of nouns conveys a plural meaning; partial reduplication of nouns is associated with distributive plurality. With verbs, full reduplication expresses repetitive aspect; whereas, partial reduplication expresses progressive. Partial reduplication of adjectives and adverbs expresses a superlative meaning; while partial reduplication of certain adverbs has a comparative meaning. Full reduplication of adjectives tends to have an emphatic meaning. Finally, certain emphatic morphemes can be reduplicated to create further emphasis. Certain morphemes appear to contain fossilized reduplication, even though no corresponding simplex forms are attested. From the combination of their forms and meanings, it is likely that morphemes, such as25 -dyedyei ‘inceptive,’ -tyityi' / -sisi' ‘resultative,’ mimi' / mömö' ‘only, just,’ and yodyedye- ‘restrictive relative clause marker with plural antecedents,’ originate from reduplication (Sakel 2004, 59). Bound morphemes in Mosetén can also be subject to partial reduplication, although this is most likely determined by phonotactic structure (rather than by morphological structure, as with Kwaza person marking reduplication). In order to express distributed plurality, the CV sequence (the initial mora) of local case-like “relation markers,” such as comitative and adessive, is reduplicated. This results in partial reduplication of bimoraic morphemes, such as -tom ‘comitative’ > -to~tom ‘with several’ and in (contingently) full reduplication of monomoraic ones, such as -we ‘downriver’ > -we~we ‘down several rivers.’ The following example contains a partially reduplicated inessive morpheme: (53) öi-khan otel-in iits mintyi'-in i-kha~khan bae'-i-in dem.f-ine hotel-pl dem man-pl m-red~ine live-v.m.s-pl ‘In this hotel these men live in there (in various rooms).’ (Sakel 2004, 79) Sakel (2004, 78–79) explicitly distinguishes between recursion and reduplication of these relation markers.26 In combination with any of the other relation markers, the adessive marker -ya' creates the connotation ‘further,’ ‘less,’ or ‘better,’ and it can even be productively combined with another adessive marker: (54) a-dyaj i-ya'-ya' aj mi' yet-qy m-ade-ade yet 3m.sg ‘Is he already getting better?’ (Sakel 2004, 78) 25 26

In Mosetén ⟨j⟩ represents IPA [h], ⟨y⟩ is [j] and ⟨ae⟩ is [ə]. This distinction is not made consistently elsewhere in the book.

452

van der voort

In example (54), the adessive marker was applied recursively in an idiomatic expression where its second occurrence has a specific connotation in this particular construction. Reduplication of the adessive marker, however, involves omission of the final glottal stop of this morpheme, as is expected with partial reduplication in Mosetén, and it has the expected distributive connotation: (55) raej öi jäe'mä jinak-ya~ya all dem.f uhh river-red~ade ‘in all parts of/at the river’ (Sakel 2004, 79) In example (55) the CV of the final syllable of the morphologically complex location word was reduplicated, which happened to coincide with the initial consonant and vowel of the adessive morpheme. Thus, recursion of the adessive marker -ya' involves the whole morpheme, preserving the final glottal stop (- ya'-ya'). Under reduplication, however, the glottal stop is omitted from the complex word’s final syllable (- ya~ya). 2.13 Movima In the Bolivian isolate language Movima (see Haude this volume), reduplication does not have the usual iconic semantics like intensity, repetitive and plurality. Rather, it has several important grammatical functions that are, however, rarely attested cross-linguistically, such as certain voice-related functions with verbs and possession-related functions on nouns. Furthermore, Movima reduplication is not based on morphological or phonotactic units, but on the prosodic units, moras and feet. Full reduplication and a functional division with partial reduplication exist, but only as by-products of reduplication of different kinds of prosodic units. Reduplication is just one of the special features of this extraordinary language, which has been described in considerable detail by Haude (2006). 2.14 Yurakaré Yurakaré is a language isolate with approximately 2500 speakers in the Andean foothills region of central Bolivia. In his description of the language, van Gijn (2006) devotes a subsection to a concise survey of reduplication (2006, 53–58). In a later article (to appear) van Gijn expands on reduplication in relation to the issue of iconicity. Yurakaré has both full and partial reduplication, and their functions are determined by the specific type of reduplication at hand. Productive full reduplication usually concerns nouns and indicates the notion of ‘being distributed over a certain space’ or ‘provided with many x,’ e.g.: shonko ‘hole’ > shonko~shonko ‘full of holes’; sëmë ‘bald’ > sëmë~sëmë ‘naked’;

is reduplication an areal feature?

453

pujshi ‘fur’ > pujshi~pujshi ‘furry.’27 There are many adjectives with similar reduplicative forms that are apparently lexicalized, since the non-reduplicated forms are not attested (anymore). Several color terms are reduplicative, some of which are still transparent: yënnë ‘genipap’ > yënnëj~yënnë ‘purple’; tëbbë ‘blood,’ > tëbbët~tëbbë ‘red’; worew(o)reshi ‘black’; shüjshüshi ‘grey, black’; pëpëpë ‘brown.’ Some adjectives can be intensified, producing the same notion of ‘being distributed over a certain space,’ by infixation of the first mora of the final syllable of the root: waraj-ta ‘hollow’ > wa⟨~ra⟩raj-ta ‘completely hollow’; shuju-ta ‘warm’ > shu⟨~ju⟩ju-ta ‘warm all over,’ etc.28 This is not very productive, however. The most productive type of intensifying partial reduplication is prefixation of the first foot of the word. It is attested with adjectives, adverbs and verbs: binta ‘strong’ > bij~binta ‘really strong’; limeye ‘behind’ > lij~limeye ‘way behind’; mala ‘go.sg’ > maj~mala ‘walk really far.’29 Sometimes the first two syllables are copied, regardless of morphological structure. In example (56) reduplication affects only the prefixes of the verb: (56) tëtë-pshë milaj~mi-la-shojñe-jti-∅ meme what-entity red~2sg-afo-frighten-hab-3 mother ‘What really frightens you, mother?’ (van Gijn 2006, 55) Partial suffixing reduplication of verb roots can indicate repetitive30 and causative. In the case of a repetitive meaning, the first syllable is reduplicated, usually in combination with the verbal plural prefix i-: bache ‘send’ > i-bache~be ‘be sending repeatedly’; bobo ‘hit’ > i-bobo~bo ‘hit repeatedly’; bëj-ta ‘see’ > i-bëj-tu~bë ‘see repeatedly.’ The causative is expressed by suffixing reduplication of either the last syllable or (more usually) the first syllable of the verb root: shuyuj-ta ‘hide (intrans)’ > shuyuj~shu ‘hide (trans)’; shuñe-te ‘grow up’ > shuñe~ñe ‘raise’; werej-ta ‘untie (intrans)’ > werej~we ‘untie (trans).’31 Causative reduplication is restricted to roots that have an ideophonic origin (van Gijn to appear).

27 28 29 30 31

In Yurakaré ⟨j⟩ represents IPA [h] and ⟨y⟩ is [j]. The element -ta in Yurakaré corresponds to the middle voice marker -tA (different vowels may occur). The reduplicated syllable is followed by a ⟨j⟩ (i.e. [h]) or by voiceless lengthening of the vowel. Or ‘pluractional’ (van Gijn to appear). Van Gijn (2006, 57) describes the effect also as ‘distributive.’ Obviously, the middle voice marker -ta (or -te) is absent in causative constructions.

454

van der voort

Full reduplication of verb roots is attested only marginally and is probably lexicalized with a sense of plurality, either of events (pluractionality) or of arguments. The exclusion of the middle voice marker with some verbs is probably due to their ideophonic origin (van Gijn 2006, 58): nome-ta ‘move’ > nome~nome ‘move repeatedly’; wi-ta ‘arrive.sg’ > wi~wi ‘arrive.pl’; le-tu ‘sit.down.sg’ > le~le ‘sit.down.pl.’

3

Final Observations

Possibly because of its predominantly iconic nature, reduplication is almost a universal phenomenon. Also among the Guaporé-Mamoré languages it is widely attested. Nevertheless, one could imagine that reduplication as a grammatical construction with a specific function could, like certain other grammatical constructions, be borrowed. In Key’s (1965) early typological survey of productive reduplication, nine of the approximately 50 featured languages are spoken in the Guaporé-Mamoré region.32 A rather concise article, it represents a provisional but systematic inventory. However, it contains no evidence for areal diffusion in the Guaporé-Mamoré region. The overall picture that emerges from the descriptive sections in the present article is rather mixed. This survey still remains preliminary in view of the limited set of languages and incomplete data available. The properties of reduplication that are found throughout the entire sample are those that are universal, such as repetitive and intensifying meanings and the predominance of verbal reduplication. Some rather uncommon types of reduplication that are attested include possessive function and morphologically-based reduplication, but these are limited to single languages (Movima and Kwaza, respectively) and are not characteristic for the region. However, partial reduplication combined with a lack of full reduplication is often considered universally impossible, although one language, Trinitario (Arawak), appears to have this combination. In table 1, on the next page, I have listed the languages discussed here with regard to a number of possible properties of reduplication. Unfortunately, it was not possible to determine for every language the presence or absence of all properties listed.

32

It includes reduplication phenomena in Cayuvava (isolate), Iranxe (isolate), Chimané (Mosetén, isolate), Baure and Paresi (Arawak), Pacaas-Novas (i.e. Wari’, Chapacura), Nambikwara (Nambikwara), Cav(in)eña (Pano-Tacana), Sirionó (Tupi-Guarani, Tupi).

Full

y y y y y y y n y y y y y y y y y

Partial

y y y y y n y y y n y y n n y y y

Functional division

y y (y) (n) y n n y (n) y (n) n

suffix prefix bi, inf prefix bi prefix prefix bi suffix suffix, inf bi suffix bi

V, N V, N, Adj V, Adj, Adv, N V, (N), (DJE Pro) V V, (Adv) V, Adv, Adj, Pro V V, Adv, Pro V, Pro V, Adv, N, Pro, Clitic V, Vattr V, Adv V V, Adv, Adj, N V V, Vattr

Change of category y y (> Adj) (> V) ? (Adv?) (lex) ? n (?) y n n n n n ?

Valency effect various detransitive causative (causative) (causative) n (reciprocal) (n) n n n n n n n ? ?

Nominal plural n n (n) n n (n) (n) (n) n n n n n n y n ?

Plural arguments n n y (y) ? (?) (n) (n) y y y y (n) n n y ?

Distributive n y y n ? y (n) ? y (n) ? n (n) y y ? ?

Iterative repetitive n y y y ? y (lex) y y y y y y y y y y

Intensifying n y y y y y y y y y y y y y y y y

n n n n y y (n) ? y y y y y (n) y y ?

Progressive

Clarification of terms and abbreviations in table 1: ‘functional division’ = division between full and partial red; bi = bidirectional; inf = infixing; lex = lexical, lexicalized; n = no; Pro = pronoun; y = yes; Vattr = attributive verb root. Yes or no answers between parentheses indicate uncertainty, either due to an insufficient number of systematic examples or because of limited productivity.

Movima Cavineña (Pano-Tacana) Yurakaré Arikapu (Macro-Jê) Cayuvava Aikanã Baure (Arawak) Trinitario (Arawak) Kwaza Mekens (Tupi) Wari’ (Chapacura) Itonama Kanoê Karitiana (Tupi) Mosetén Rikbaktsa (Macro-Jê) Nambikwara

Directionality in partial red

Properties of reduplication across the Guaporé-Mamoré region

Limited to category

table 1

is reduplication an areal feature?

455

456

van der voort

In addition to the properties listed in table 1, there are other possible characteristics of reduplication that are shared by only two or three languages each. They are listed in Table 2: table 2

Properties of reduplication shared by few Guaporé-Mamoré languages

attenuative meaning color term reduplication complex reduplication limited productivity prosodically motivated triplication

Trinitario, Kwaza, Moré Yurakaré, Aikanã Wari’, Itonama?, Cayuvava? Arikapu, Nambikwara Movima, Chiquitano Mekens, Wari’

Those similar properties that could perhaps be explained by areal diffusion include complex reduplication and triplication. Other potential candidates for diffusion that may co-define the linguistic area or a sub-area are plural argument marking, pluractionality (perhaps originating from Tupi), and certain valency effects. The fact that one or perhaps two languages of the region can mark causative valence through reduplication is conspicuous, since that is rarely attested cross-linguistically. It is known from a handful of African and Austronesian languages, and the only other Amazonian language where a causative effect of reduplication is reported is Jarawara (Arawa, see Dixon and Vogel 1996). If one extends this phenomenon to include other valency effects, this subset of languages increases, adding at least Cavineña, its close relative Ese Ejja and Movima. Apart from being a universal tendency (cf. Hurch 2005 ff.), the clear predominance of verbal reduplication over nominal reduplication may be related to the rarity of a nominal plural (or collective) marking function in the region. One could speculate that borrowed reduplicative forms can function as a model and lead to the emergence of reduplication as a grammatical process.33 It is not surprising that (unproductive) reduplicated forms are borrowed. Evans (2009) shows that the Australian language Iwaidja has borrowed reduplicative patterns from other, distantly related and unrelated, languages. These patterns are not productive, unlike the original Iwaidja patterns, which

33

In that case, the language from which the reduplicated forms are borrowed does not necessarily have productive reduplication itself.

is reduplication an areal feature?

457

are productive. And there are indeed some reduplicative lexemes that have spread throughout the Brazilian part of the Guaporé-Mamoré area. One of them is the word for ‘maize’: atxitxi (Kwaza), atiti (Kanoê), txitxi (Jabuti languages), atsitsi (Mekens), atiti (other Tupari languages). One could imagine that the repeated syllable ti symbolizes the multiplicity or repetition of the relatively small size maize grains on a cob. The non-repeated Itonama form atxɨ may be related. Although maize must have been cultivated for millennia in the Guaporé-Mamoré region, it is not likely that this word stems from the time when it was introduced. Another characteristic word refers to ‘chicken,’ a domestic fowl that was originally introduced by Westerners: kurakura (Kwaza), kurakura (Kanoê), kalakala (Lakondê), korakora (Mekens) and kura'ka (Itonama). Yet another word that has spread is ‘owl’: pupure (Aikanã), pupu (Arikapu), popo (Djeoromitxi), popoba (Mekens) and perhaps also bububudɨ- ‘owl calling’ (Kwaza). Even though symbolic words such as these onomatopoeias are non-arbitrary, based on universal processes in language and also found elsewhere, they are strikingly similar across the region’s genetic-linguistic boundaries. One iconically reduplicative form that has apparently spread throughout a larger area is the word for ‘morning star’ or ‘Venus’: waruwaru (Kwaza), warɨwarɨ (Kanoê), warəwarə (Arikapu), wirəwirə (Djeoromitxi), parobaro (Mekens), waruwaru (other Tupari languages). This word is also attested in Aymara (Andean) as warawara. However, there is no evidence that these inherently reduplicated words have led to productive reduplication in any language of the region. The repetitive pattern found in numerals in several languages that have minimal counting systems, such as Aikanã, Kanoê, Kwaza, and the Jabuti languages, should not be regarded as reduplication. Rather, numerals like Arikapu heri heri ‘four’ (lit. ‘two two’) and Aikanã atuka atuka ‘four’ (lit. ‘two two’) are recursive additive constructions. Furthermore, as Green (1997) has shown, such constructions are also found elsewhere in the Amazon.34 There is considerable evidence for areal linguistic diffusion in the GuaporéMamoré region. Since productive reduplication is a morphosyntactic operation, one would expect that it could also occur as an areal feature. However, areal diffusion of reduplication is rarely attested in general and rarely discussed in the literature. Looking at the Guaporé-Mamoré region, shared characteristics of reduplication across different languages tend to be due to genetic or

34

Green’s overview of number terms in Brazilian indigenous languages does include some genuine cases of (intensifying) reduplication in minimal counting systems of two TupiGuaraní languages (1997, 184–185).

458

van der voort

universal factors and not to be identifiable as areal features. Reduplication as a means to modify valency, however, seems to be productive in three languages (Cavineña, Movima, Yurakaré),35 spoken in an area roughly following along the Mamoré river in the Bolivian Lowlands. It is possibly attested also in one language (Arikapu) on the Brazilian side of the Guaporé River but not productively. Although valency-changing reduplication is cross-linguistically quite rare, it is attested in a cluster of genetically-unrelated languages in the GuaporéMamoré region and may represent an areal trait.

Acknowledgements This article could not have been written without the input from my consultants Edileusa and Zezinho Kwazá, Raimunda and Mario Aikanã, Manoel Aikanã, Nazaré Arikapú, Armando Jabotí. I am also much indebted to the indigenous communities in the Tubarão-Latundê, Rio São Pedro, Rio Branco and Rio Guaporé reserves in Rondônia for their hospitality and help, and I want to thank the FUNAI and CNPq for the necessary research authorizations. Furthermore, I want to acknowledge the generous funding of the Netherlands Organization for Scientific Research (NWO, VIDI nr. nr. 276-70-005) and the Fundação de Amparo à Pesquisa do Estado do Pará (FAPESPA, PPDOC 004/2010). Finally, I am very grateful for the comments and corrections from the anonymous reviewer and my colleagues Joshua Birchall, Mily Crevels, Swintha Danielsen, Vilacy Galucio, Gale Goodwin Gómez and Rik van Gijn. All errors are mine.

References Angenot de Lima, Geralda. 2002. Description phonologique, grammaticale et lexicale du moré, langue amazonienne de Bolivie et du Brésil, 2 vols. PhD diss., Universiteit Leiden. Araujo, Gabriel Antunes de. 2004. A grammar of Sabanê, a Nambikwaran language. Utrecht: LOT Publications. http://www.lotpublications.nl/publish/articles/000919/ bookpart.pdf (accessed November 16, 2013). Bacelar, Laércio N. 2004. Gramática da língua Kanoê. PhD diss., Katholieke Universiteit Nijmegen. http://webdoc.ubn.kun.nl/mono/b/bacelar_l/gramdalik.pdf (accessed November 16, 2013).

35

Or perhaps four, if the Cayuvava causative is analyzed as reduplication.

is reduplication an areal feature?

459

Bakker, Peter. 2003. The absence of reduplication in pidgins. In Kouwenberg (ed.), 37–46. . 2006. The Sri Lanka Sprachbund: The newcomers Portuguese and Malay. In Linguistic areas: Convergence in historical and typological perspective, ed. Yaron Matras, April McMahon, Nigel Vincent, 135–159. Hampshire/New York: Palgrave Macmillan Bakker, Peter, and Mikael Parkvall. 2005. Reduplication in pidgins and creoles. In Hurch (ed.), 511–531. Bybee, Joan, Revere Perkins, and William Pagliuca. 1994. The evolution of grammar: Tense, aspect, and modality in the languages of the world. Chicago/London: The University of Chicago Press. Chirikba, Viacheslav A. 2008. The problem of the Caucasian Sprachbund. In Muysken (ed.), 25–93. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press. Crevels, Mily. 2006. Verbal number in Itonama. In What’s in a verb? Studies in the verbal morphology of the languages of the Americas, ed. Grażyna J. Rowicka and Eithne B. Carlin, 159–170. Utrecht: LOT. http://lotos.library.uu.nl/publish/articles/000159/ bookpart.pdf (accessed November 16, 2013). . 2012. Itonama. Lenguas de Bolivia, Tomo II: Amazonía, ed. Mily Crevels and Pieter Muysken, 233–294. La Paz: Plural editores. Crevels, Mily, and Pieter Muysken. 2012. Cayubaba. Lenguas de Bolivia, Tomo II: Amazonía, ed. Mily Crevels and Pieter Muysken, 341–375. La Paz: Plural editores. Crevels, Mily, and Hein van der Voort. 2008. The Guaporé-Mamoré region as a linguistic area. In Muysken (ed.), 151–179. Dahl, Östen. 1985. Tense and aspect systems. London/New York: Basil Blackwell. Danielsen, Swintha. 2007. Baure: An Arawak language of Bolivia. [Indigenous Languages of Latin America (ILLA) 6]. Leiden: CNWS Publications. Dietrich, Wolf. this volume. Forms and functions of reduplication in Tupian languages. Dixon, R.M.W., and Allan R. Vogel. 1996. Reduplication in Jarawara. Languages of the World 10: 24–31. Eberhard, David M. 2009. Mamaindê grammar: A Northern Nambikwara language and its cultural context, 2 vols. Utrecht: LOT Publications. http://www.lotpublications.nl/ publish/articles/003783/bookpart.pdf (accessed November 16, 2013). Evans, Nick. 2009. Doubled up all over again: borrowing, sound change and reduplication in Iwaidja. Morphology 19/2: 159–176. Everett, Daniel, and Barbara Kern. 1997. Wari’: The Pacaas Novos language of Western Brazil. London: Routledge. Galucio, Ana Vilacy. 2001. The morphosyntax of Mekens (Tupi). PhD diss., University of Chicago.

460

van der voort

Gijn, Rik van. 2006. A grammar of Yurakaré. PhD diss., Radboud Universiteit. http:// webdoc.ubn.ru.nl/mono/g/gijn_e_van/gramofyu.pdf (accessed November 16, 2013). . to appear. Reduplication in Yurakaré. In Word formation in South American languages, ed. Katja Hannss, Swintha Danielsen and Fernando Zúñiga. Amsterdam/Philadelphia: John Benjamins. Green, Diana. 1997. Diferenças entre termos numéricos em algumas línguas indígenas do Brasil. Boletim do Museu Paraense Emílio Goeldi (Antropologia) 13/2: 179–207. Guillaume, Antoine. 2008. A Grammar of Cavineña. Berlin/New York: Mouton de Gruyter. . this volume. The interaction of reduplication with word classes and transitivity in Cavineña. Haude, Katharina. 2006. A grammar of Movima. PhD diss., Radboud Universiteit. http:// webdoc.ubn.ru.nl/mono/h/haude_k/gramofmo.pdf (accessed November 16, 2013). . this volume. Reduplication in Movima: A prosodic morphology approach. Hurch, Bernhard. 2005ff. Graz Database on Reduplication. http://reduplication.uni-graz .at/redup/ (accessed February 24, 2013). . 2009. Areal typology and reduplication: The case of Amazonian languages. Talk presented at the 53rd International Congress of Americanists, July 19–24, in Mexico City, Universidad Iberoamericana. Hurch, Bernhard, with Veronika Mattes, ed. 2005. Studies on reduplication. [Empirical Approaches to Language Typology 28]. Berlin/New York: Mouton de Gruyter. Hurch, Bernhard, ed. 2009. Reduplication: Diachrony and productivity—a reprise. [Grazer Linguistische Studien 71]. Graz: Institut für Sprachwissenschaft der Universität Graz. Hurch, Bernhard, and Veronika Mattes. 2009. Typology of reduplication: The Graz database. In The use of databases in cross-linguistic research, ed. Martin Everaert, Simon Musgrave and Alexis Dimitriadis, 301–327. Berlin/New York: Mouton de Gruyter. Hurch, Bernhard, and Veronika Mattes, ed. 2009. Reduplication: Diachrony and productivity. [Morphology 19/2]. Berlin: Springer. Key, Harold. 1965. Some semantic functions of reduplication in various languages. Anthropological Linguistics 7/3: 88–102. . 1967. Morphology of Cayuvava. The Hague/Paris: Mouton. Kim, Yuni. 2012. Review of Total reduplication: The areal linguistics of a potential universal, by Thomas Stolz, Cornelia Stroh and Aina Urdze. Studies in Language 32/2: 440–448. Kouwenberg, Silvia. 2003. Introduction. In Kouwenberg (ed.), 1–6. Kouwenberg, Silvia, ed. 2003. Twice as meaningful: Reduplication in pidgins, creoles and other contact languages. [Westminster Creolistics Series 8]. London: Battlebridge. Kouwenberg, Silvia, and Darlene LaCharité. 2004. Echoes of Africa: reduplication in

is reduplication an areal feature?

461

Caribbean creole and Niger-Congo languages. Journal of Pidgin and Creole Languages 19/2: 285–331. Kramer, Marvin. 1993. Reduplication. Aikanã modules: A class report based on the fieldnotes of Harvey Carlson, ed. Leanne Hinton, section 19. Typescript. Berkeley: University of California. Kroeker, Menno H. 2001. A descriptive grammar of Nambikuara. International Journal of American Linguistics 67/1: 1–87. Moravcsik, Edith. 1978. Reduplicative constructions. In Universals of human language, Vol. 3, ed. Joseph H. Greenberg, 297–334. Stanford: Stanford University Press. Muysken, Pieter, ed. 2008. From linguistic areas to areal linguistics. Amsterdam/Philadelphia: John Benjamins. Popky, Donna. 1999. Oro Win: A descriptive and comparative look at an endangered language. MA thesis, University of Pittsburgh. Rose, Francoise. 2005. Reduplication in Tupi-Guarani languages: Going into opposite directions. In Hurch (ed.), 351–368. . this volume. When vowel deletion blurs reduplication in Mojeño Trinitario. Rubino, Carl. 2005. Reduplication: Form, function and distribution. In Hurch (ed.), 11–29. Sakel, Jeanette. 2004. A grammar of Mosetén. [Mouton Grammar Library 33]. Berlin/ New York: Mouton de Gruyter. Salanova, Andrés Pablo. this volume. Reduplication and verbal number in Mẽbengokre. Silva, Léia de Jesus. 2011. Morphosyntaxe du Rikbaktsa (Amazonie brésilienne). PhD diss., Université Denis Diderot (Paris 7). http://www.etnolinguistica.org/tese:silva -2011b (accessed November 16, 2013). Stolz, Thomas. 2008. Total reduplication vs. echo-word formation in language contact situations. In Language contact and contact languages, ed. Peter Siemund and Noemi Kintana, 317–350. Amsterdam/Philadelphia: John Benjamins. Stolz, Thomas, Cornelia Stroh and Aina Urdze. 2011. Total reduplication: The areal linguistics of a potential universal. [Studia Typologica 8]. Berlin: Akademie Verlag. Storto, Luciana. this volume. Reduplication in Karitiana (Tupi). Telles, Stella. 2002. Fonologia e gramática Latundê/Lakondê. PhD diss., Vrije Universiteit. Timberlake, Alan. 2006. Aspect, tense, mood. In Language typology and syntactic description, III: Grammatical categories and the lexicon, ed. Timothy Shopen, 280–333. Cambridge: Cambridge University Press. Vasconcelos, Ione P. 1996. Algumas considerações sobre a morfologia Aikanã. ABRALIN: Boletim da Associação Brasileira de Lingüística 19: 71–78. Voort, Hein van der. 2003. Reduplication of person markers in Kwaza. Acta Linguistica Hafniensia 35: 65–94. . 2005. Kwaza in a comparative perspective. International Journal of American Linguistics 71/4: 365–412.

462

van der voort

. 2009. Reduplication and repetition of person markers in Guaporé isolates. Morphology 19/2: 263–286. http://www.springerlink.com/content/ 1545301353748536/fulltext.pdf (accessed November 16, 2013). . 2013. Fala fictícia fossilizada: o tempo futuro em Aikanã. Boletim do Museu Paraense Emílio Goeldi (Ciências humanas) 8/2: 359–377. Vuillermet, Marine. 2012. A Grammar of Ese Ejja, a Takanan language of the Bolivian Amazon. PhD diss., Université Lumière Lyon 2. Yakpo, Kofi. 2012. Reiteration in Pichi. In The Morphosyntax of reiteration in creole and non-creole languages, ed. Aboh, Enoch O., Norval Smith and Anne Zribi-Hertz, 251–284. Amsterdam/Philadelphia: John Benjamins

Abbreviations acc ade afo asp aux ben cl cmpl coll cor dec dem emp emph excl f foc frust fut hab imp impf ine infl inv

accusative adessive affected object aspect auxiliary benefactive classifier completive collective coreferent declarative demonstrative pronoun empathy emphatic exclusive feminine focus frustrative future habitual imperative imperfective inessive inflectional particle inverse

nmloc m mod n, neu neg nom o obl pl pot prec prox qy reci red rel rep rp/p rpas s sg stat tra v

locative nominalization masculine modality neuter gender negative nominalizer object oblique plural potential preconditional proximate yes/no question marker reciprocal reduplication relative reported speech realis past/present remote past subject singular stative transitivizer verbal stem marker

Index of Names Abraham, W. 4 Acosta Alcaraz, F. 293n10 Adelaar, W.F.H. 40–43, 45, 46, 51, 51n7, 55, 61, 62, 62n19, 221 Aikhenvald, A. 79, 144, 329, 330 Albert, B. 161n1, 162n1, 181 Alderete, J. 3, 388 Alexander-Bakkerus, A. 40, 45–47, 53, 53n10, 55–57 Almeida, A. 297 Alves, F. de Castro 247 Alves, P. 301 Ammann, A. 1 Anchieta, P.J. de 278, 279 Angenot de Lima, G. 434, 435 Appah, C. 330n19 Araujo, G. Antunes de. 438 Araújo, L. 1 Augusta, F. 18, 20, 21, 23, 29–32 Bacelar, L.N. 446, 446n21, 447 Bakker, P. 428, 429 Bendor-Samuel, D. 298 Berlin, B. 100n7, 178n5 Bessa Freire, J.R. 117 Betts, L. 287n8 Beuchat, H 40, 64 Borges, M. 1, 138, 281 Braga, A. de Oliveira 301, 302n16 Broselow, E. 353 Brown, P. 103n9 Bruno, A.C. 1, 423 Bryant, D. 66 Burgess, E. 247 Bybee, J. 5, 430 Cabral, A.S. Arruda Câmara 118, 282, 282n6, 284n7 Callow, J. 247 Camp, E. 314, 332, 333, 336 Canese, N. Krivoshein de 293n10 Carvajal Carvajal, J. 41, 42

Caspar, F. 301 Cavalcante, M.P. 247, 268, 268n18, 269 Cerrón-Palomino, R. 41, 42 Chamorro, G. 293n10 Charachidzé, G. 330n20 Chirikba, V.A. 429 Clark, J. 45 Cole, P. 78 Comrie, B. 2, 275, 430 Corbett, G. 4, 384, 431 Coutinho-Silva, T. 401, 402 Creissels, D. 330, 330n20, 337n26, 341 Crevels, M. 11, 40, 375, 429, 430, 442, 444–446, 458 Croese, R. 18 Crofts, M. 422 Cruz, A. da 4, 7, 8, 118n4, 127n8, 130n10, 305, 330, 341 Curnow, T. 77, 110n14 Cusic, D. 128, 138, 266n17, 275, 384 Cusihuamán Gutiérrez, A. 50 D’Angelis, W. da Rocha 268 Dahl, Ö. 430 Danielsen, S. 3, 11, 431, 432, 450, 458 de Granda, G. 61 de la Mata, P. 45 Dedenbach-Salazar Sáenz, S. 51, 58, 60 DeLucca, M. 60, 60n15, 62n18 Díaz-Fernández, A. 6, 7, 17, 18, 20, 33, 40 Dickinson, C. 77, 79, 83n1 Dietrich, W. 9, 124n5, 185, 242, 276n2, 277n3, 431, 440 Dingemanse, M. 100 Dixon, R.M.W. 1, 99, 126n6, 329, 330, 456 Dooley, R. 291, 292 dos Santos, L 247 Dourado, L. 219, 247 Dressler, W. 148 Drude, S. 8, 11, 185, 187, 189n6, 194n16, 204, 210, 306 Dubois, F. 19

464 Eberhard, D.M. 437, 438 Emmerich, C. 219 Epps, P. 8, 11, 144, 146–150, 152, 154, 156, 157, 313n1 Evans, N. 456 Everett, C. 415, 416n12 Everett, D. 1, 3, 432–434 Fabricano Noé, F. 376n1 Facundes, S. 119 Fargetti, C.M. 1, 219, 276, 299, 299n13, 300n14 Felix, M.I. de Santana 117 Fernández Garay, A. 11, 18, 35 Fischer, O. 4 Floyd, S. 6, 7, 11, 78, 135 Gabas Jr., N. 234, 235n13, 238n16, 239, 240, 422 Galucio, A.V. 11, 439, 440, 440n9, 458 Giannecchini, D. 279, 294 Gil, D. 2, 155, 170 Gill, W. 10, 375, 376, 376n1, 378n4, 382, 382n9, 384, 387–389, 392, 393, 393n17, 396, 397 Girault, L. 53, 60 Gnerre, M. 40, 58, 60 Gomes, D. 1, 302, 418 Gomez-Imbert, E. 144 Goodwin Gómez, G. 1, 3, 4, 8, 139, 153, 157, 161, 162, 162n1, 170, 177, 244, 308, 313n1, 341, 371, 458 Green, D. 457, 457n34 Grimes, B. 17 Grinevald, C. 369 Guillaume, A. 4, 9, 139, 314, 315, 317, 321n9, 322n12, 330, 333, 334n23, 336n25, 431, 439 Guirardello, R. 9, 221, 235, 235n12 Haiman, J. 148 Halle, M. 191n11 Ham, P. 247 Hamp, E. 17 Hannß, K. 6, 7, 40, 44, 44n3, 45, 49, 50, 54, 55, 55n13, 59 Hardman, M. 40–42, 51, 51n9, 52, 58, 62, 62n18

index of names Hardy, H. 153 Harmelink, B. 18, 25, 27, 28, 30, 34 Harrison, S. 330 Hartt, C.F. 134n13 Haude, K. 2, 3, 10, 11, 343, 343n1, 345, 346, 354, 354n9, 354n10, 355, 359–361, 362n12, 362n13, 364–366, 366n17, 366n18, 367–370, 431, 452 Hernández Sallés, A. 18 Hoeller, A. 280 Hoggarth, L. 40, 48, 57 Hopper, P. 138 Huayhua Pari, F. 47, 48, 57 Hurch, B. 1, 155, 163, 427n1, 428, 429, 456 Huson, D. 66 Inkelas, S. 1, 89 Ito, J. 148 Jackson, J. 144 Jensen, C. 138, 190, 201, 279n4 Judy, J. 343 Judy, R. 343 Julião, M.R. Silva 284 Kager, R. 350 Kakumasu, J. 298 Kay, P. 100n7 Kemmer, S. 221 Kern, B. 3, 432–434 Key, H. 433n6, 442, 443, 443n15, 443n16, 443n18, 444, 454 Key, M.R. 17, 18 Kilian-Hatz, C. 80, 154, 163, 177, 234 Kim, Y. 429n3 Kita, S. 80 Koul, O.N. 20 Kouwenberg, S. 1, 61, 428, 428n2 Kramer, M. 441, 442n14 Kratzer, A. 414 Kroeker, M.H. 438 LaCharité, D. 61, 428n2 Landin, D. 419 Landman, F. 84

465

index of names Lasersohn, P. 401, 414 Lee, K. 117 Lehmann, W. 60 Liccardi, M. 314, 332, 333, 336 Lima, S. 11, 41, 299, 299n13, 407n9, 423, 434 Lindskook, J. 77 Loos, E. 17 Macaulay, M. 147 Macdonald, R. 50 Magalhães, M. 117n3, 287, 288n9 Malvestitti, M. 20 Marantz, A. 220n3, 267, 353, 371 Marbán, P. 375, 379 Matisoff, J. 258 Matras, Y. 44, 50 Mattes, V. 1, 163, 428 McCarthy, J. 348, 353, 391, 397 McKay, T. 84 Meira, S. 1, 117, 185, 221 Mello, A.A.S. 213 Mester, A. 148 Métraux, A. 60 Michaud, A. 1 Migliazza, E. 161 Milliken, W. 162n1 Mithun, M 126 Montler, T. 153 Montoya, A. Ruiz de 278, 279, 284 Moore, B. 11, 77, 89n4, 119 Moravcsik, E. 2, 5, 10, 146, 153, 241, 330, 330n19, 338, 397, 432, 448 Morgenstern, A. 1 Müller, A. 401, 402, 404–406, 414, 415, 423 Murphy, R. 235n12 Muysken, P. 6, 7, 40–44, 44n3, 45, 46, 50, 61, 62, 62n19, 71, 73, 375, 442 Nedjalkov, V 136, 137n15 Negrão, E. 404 Nelson, N. 154 Newman, P. 384 Newman, S. 63n20 Nuckolls, J. 242

Obando Ordóñez, P. 77 Oblitas Poblete, E. 49, 60, 62, 63, 63n21 Oliveira, C. da Cunha 247, 258, 259n13 Oliveira, R. Lopes Gomes de 117 Olza Zubiri, J. 375, 379, 382n9 Oré, L.J. de 43 Palmer, B. 330 Parkvall, M. 428 Pease, H. 286 Perri Ferreira, H. 11, 162n2 Picanço, G.L. 3, 302, 304 Pires, N. 119 Plaza Martínez, P. 42 Polinsky, M. 329, 330 Polo, J.T. 49, 58 Popjes, Jack 247 Popjes, Jo 247 Popky, D. 434 Porterie-Gutiérrez, L. 47 Pott, A.F. 1 Praça, W. Neiva 11, 297 Prince, A. 348, 391, 397 Quain, B. 235n12 Queixalós, F. 11, 138, 139, 385n12 Raimy, E. 191n11 Ramirez, H. 161, 166 Reiss, C. 191n11 Restivo, P. 278 Rivet, P. 40, 64 Rocha, I. 418 Rodrigues, A.D. 115n1, 117, 117n3, 138, 185, 190, 282, 282n6, 284n7, 285, 301, 301n15 Rose, F. 1, 3–5, 10, 11, 138, 190n10, 242–244, 273–276, 279, 281, 284, 294, 306, 341, 376, 396n21, 431, 432, 440 Rose, S. 396 Rouby, A. 51, 57 Rubino, C. 2, 3, 10, 19, 20, 48, 50, 58, 63n20, 146, 153, 153n7, 170, 221, 241, 314n2, 326, 364n15, 370, 385, 397, 418, 428, 445n20 Sakel, J.

3, 450, 450n24, 451, 452

466 Salanova, A.P. 9, 247–249, 258n11, 437 Salas, A. 25, 27, 28, 30, 34 Salmons, J.C. 147 Sanchez-Mendes, L. 401, 402, 404–406, 414, 415, 423 Sandvig, T. 20 Schermair, P. Fr. A. 295 Schiering, R. 148 Schuchard, B. 294, 294n12 Schultze-Berndt, E. 83n1 Schwade, M.C. de Deus Lima 117 Seki, L. 1, 138, 190, 201, 219, 234, 237, 239, 306 Sherzer, J. 155 Silva, A.C. Souza 296 Silva, L. de Jesus 139, 435 Silva, M. Amélia Reis 252n5, 253n5 Silva, R. Geraldine Pereira da 117 Simpson, M. 191n11 Smeets, I. 18, 25–27, 27n10, 28, 30, 31, 33–35 Soenjono, D. 50 Solano, E. de Jesus Bararuá 285, 286 Sorensen, A.P. 144 Soria Lens, L. 52 Spring, C. 391 Stark, L.R. 17, 43 Stolz, T. 1, 2, 4, 154, 161, 163, 170, 175n4, 180, 370n21, 428, 429 Storto, L. 10, 401, 402, 402n1, 403n3, 404, 406n8, 412, 413, 416, 418, 421, 423, 431, 439, 450 Stout, M. 248, 250n3

index of names Thompson, S.A. 138 Thomson, R. 248, 250, 250n3, 253n5 Timberlake, A. 430 Torero, A. 41, 43 Trapp, E. 252n5, 258 Turner, G. 46, 253n5 Urban, G. 268 Urdze, A. 1 Vallejos Yopán, R. 276, 289–291 van de Kerke, S. 40, 43, 46, 51, 55, 63, 73 van der Auwera, J. 234, 235n13, 238n16, 239, 240, 422 Van der Meer, T. 234 van der Voort, H. 1, 3, 5, 10, 11, 139, 157, 181, 244, 308, 341, 371, 423, 429, 430, 441n10, 447, 449 van Gijn, R. 1, 4, 11, 452, 453, 453n30, 454, 458 Vásquez de Ruiz, B. 77 Vellard, J. 49, 50, 54, 58 Vendler, Z 384 Villafañe, L. 296 Vittadello, A. 77 Voeltz, F.K. 80, 154, 234 Vogel, A.R. 1, 456 Vuillermet, M. 330, 439, 450 Vúletin, A. 18 Wiesemann, U. Yallop, C.

Tapuyo Pianchiche, V. 77 Tavo Mayo, V. 316, 331, 334, 336 Taylor, J. 118, 232n11

247

45

Zoll, C. 1, 89 Zúñiga, F. 6, 7, 17, 19, 30, 40, 73

Index of Languages Aché 274 Aikanã 13, 431, 431n4, 441–442, 450, 455–458 Akan 330n19 Alabama 153 Amondava 274 Anambé 13, 273, 275–278, 284–285, 306, 307 Andean 6, 7, 11, 17, 18, 39–41, 43, 45–47, 49, 49n5, 50, 55, 56, 60, 64–73, 450, 452, 457 Apiaká 274 Apinayé 247, 258 Araona 313, 336 Araucanian 18 Arawa 456 Arawak family 18 Arawak 10, 13, 144, 244, 375, 391, 397, 431–432, 450, 454, 454n32, 455 Arawakan 6, 43, 305 Arawaken 43 Araweté 13, 273, 275–278, 285–286, 306, 307 Arikapu 13, 431n4, 435, 437, 455–458 Asurini do Tocantins 13, 273, 275–278, 282–284, 306, 307 Australian 82n1, 456 Austronesian 1, 456 Avá-Canoeiro 13, 273, 275, 276, 278, 281–282, 306, 307 Avá-Nhandeva 273n1, 274 Avar 330n20 Awetí 9, 185, 186–187, 187, 189n6, 190–192, 192, 193n15, 194n16, 195, 198, 204–207, 210, 211n32, 213, 214, 306, 307 Aweti 13, 240n17, 306 Aymara 7, 13, 40–43, 43n3, 44, 45, 47, 48, 51, 52, 56, 57, 60, 60n14, 61, 62, 65, 67–72, 72n27, 457 Baniwa 115, 117, 144, 305 Barbacoa 13 Barbacoan 77, 79, 110, 110n14 Baré 115, 117, 130n10, 305 Baure 3, 11, 13, 431, 432, 450, 454n32, 455 Boumaa Fijian 330

Canichana 431 Carib 244 Caucasian 330n20 Cavineña 4, 9, 13, 313–316, 316n3, 317, 318–327, 329, 331, 333, 334, 337–339, 431, 439, 455, 456, 458 Cayuvava 13, 431, 442–444, 454n32, 455, 456, 458n35 Cha’palaa 6, 7, 13, 77–80, 80, 82–91, 92, 94, 96, 97, 99–101, 105–111 Chaco Guarani 294, 294n12 Chapacura 13, 431, 432–435, 450, 454n32, 455 Chedungun 17 Chesumun 17 Chezungun 17 Chimané 450, 454n32 Chiriguano 13, 273, 273n1, 275, 278–280, 280n5, 294–295, 306, 307 Cholón 7, 13, 40, 41, 45–47, 51–53, 55–57, 59, 66, 68, 68n26, 69–72 Cubeo 144 Cuzqueño Quechua 42–44, 46, 48, 65, 67, 68, 68n25, 69–72 Dâw 143n1 Djeoromitxi 13, 431n4, 437, 457 Emerillon 3, 242, 273, 273n1, 275, 279, 284, 306, 307 English 1, 110, 117n2, 134n14, 166, 176, 277n3, 428n2 Ese Ejja 13, 313, 330, 439, 450, 456 Gbe 428n2 Gê 244 German 195n18, 274 Greek, Ancient 275 Guajá 13, 273, 275, 276, 278, 287–289, 306, 307 Guajajara 13, 298, 306, 307 Guarani 13, 273, 273n1, 274–276, 278, 279, 279n4, 293–294, 294, 305

468 Guaraní 306, 307 Guarayo 273, 275, 276, 280, 306, 307 Hungarian 274, 276n2, 304n17 Hup 8, 13, 143, 143n1, 144–146, 146–148, 148–154, 155, 155n9, 156, 157, 313n1 Indo-European 275, 428 Iranxe 431, 454n32 Italian 274 Itonama 13, 431, 444–446, 455–457 Iwaidja 456 Japanese 148 Jaqaru 7, 13, 40–42, 47, 51, 52, 56–58, 65, 67–72 Jarawara 456 Jê 9, 13, 247, 267, 270, 437 Jivaro 13 Juruna 13, 274–278, 299–300, 306, 307, 423 Ka’apor 13, 274–277, 298, 306, 307 Kaingang 9, 247, 267, 268–270, 270, 270n19 Kaiwá 273n1, 274 Kallawaya 7, 13, 40, 41, 43, 43n3, 44, 44n3, 45, 47, 49–53, 55–60, 62, 63, 65, 68, 70–73 Kamaiurá 138, 234, 235, 235n12, 237, 239, 243, 244, 306, 307 Kamayurá 273, 275 Kanoê 13, 431, 446–447, 450, 455, 457 Karib 423 Karitiana 10, 13, 401, 402, 402n1, 403, 405, 406, 406n8, 409, 411–414, 414n11, 415–416, 416, 416n12, 418, 419–421, 421–423, 431, 439, 450, 455 Karo 240, 422 Kayabí 274 Kokama 13, 274–278, 289–291, 306, 307 Kokota 330 Kwa family 330n19 Kwaza 3, 5, 13, 423, 430, 431, 431n4, 447–450, 451, 454–457 Lakondê 437, 457 Latin 275

index of languages Latundê 13, 437, 458 Leko 7, 13, 40, 41, 44, 46, 51, 56, 57, 63, 66, 67, 69–72 lingua franca 116n1 Língua Geral 305 Macro-Jê 13, 431, 435–437, 455 Makurap 13, 274–278, 301–302, 306, 307, 431, 440 Malay 1 Mamaindê 13, 437, 438 Mapuche people 7, 17 Mapuchedungun 17 Mapuchezungun 17 Mapudungun 17, 19 Mapuzungun 7, 13, 17, 18, 18n1, 20–24, 24, 27n9, 30, 31n12, 32–35 Mawé 117, 185 Mayan family 17 Mbyá 13, 273, 273n1, 275, 276, 291–292, 306, 307 Me˜bengokre 9, 13, 247, 248–252, 252–257, 258–264, 265, 266–267, 268n18, 269, 270, 437 Mekens 13, 439, 440, 455–457 Mojeño Trinitario 10, 13, 376–382, 392n15 Mokilese 330 Moré 434, 456 Mosetén 3, 13, 431, 450–452, 454n32, 455 Movima 2, 3, 10, 13, 343–344, 344–345, 345–348, 348, 348n2, 348n3, 349, 349n5, 354, 356, 362, 362n13, 365, 369–371, 430, 431, 452, 454–456, 458 Mundurukú 3 Munduruku 13, 274–278, 302–304, 304n17, 306, 307, 422 Nadahup (Makú) 8, 143, 143n1 Nadëb 143n1 Nambikwara 13, 431, 437–438, 454n32, 455, 456 Nheengatu 7, 13, 115, 116–118, 118–121, 124, 126–128, 128–130, 130n10, 133, 135–139, 274, 277, 278, 305, 307, 330 Ninam 161

469

index of languages Northern-Arawak languages

Tapirapé 13, 273, 275–278, 297–298, 306, 307 Tariana 144 Oceanic languages 330 Tarma Quechua 40, 42, 51, 62 Oro Win 13, 434 Tembé 274 Pano-Tacana 13, 431, 439, 454n32, 455 Trinitario 5, 10, 375–379, 381, 382, 382n9, Pano-Tacanan family 18 384, 385, 389, 391, 392, 396, 397, 431, 432, Paraguayan Guarani 13, 273, 273n1, 274–276, 454–456 278, 279n4, 293–294, 294 Trumai 9, 13, 217, 218, 218n1, 219, 219n2, 220, Parakanã 13, 273, 275–279, 296–297, 306, 307 222n4, 223n6, 230, 232n11, 233–235, 235n12, Paresi 454n32 235n13, 237, 238n15, 239–244 Parintintin 13, 273, 275, 276, 278, 286–287, Tukanoan 144 306, 307 Tupari 13, 274, 276, 301, 306, 307, 439, 440, Portuguese 2, 115n1, 116, 117, 117n2, 119, 134n14, 457 136, 139, 218n1, 232n11, 234, 252, 305, 423 Tuparí 274, 275, 277, 278, 301 Proto-Panoan 17 Tupi-Guarani 9, 13, 115, 115n1, 126, 138, Proto-Tupi-Guarani 138 242–244, 273, 273n1, 274, 276–278, 281–305, Pukina 7, 13, 40, 41, 43, 43n3, 44, 45, 51, 54, 306, 330, 396n21, 440, 454n32 54n11, 55, 55n12, 56, 57, 65, 67, 68n26, 69–72 Tupí-Guaraní 185, 190n10, 199n22, 212 Tupi-Guaraní 273, 457n34 Quechua 6, 7, 13, 19, 40–43, 43n3, 44–46, 48, Tupi 9, 11, 13, 136, 138, 244, 273, 273n1, 50, 51, 53, 54n11, 55–58, 60, 60n14, 61, 62, 274–278, 281–305, 305, 306, 401, 422, 423, 65n22, 67, 68n25, 72, 73, 77, 79, 110, 111, 242 431, 439–440, 454n32, 455, 456 Tupí 185, 201, 204 Reyesano 313 Tupian 9, 190, 273–275, 306, 307, 401, 422 Rikbaktsa 13, 435, 455 Tupinambá 7, 13, 115–117, 130n10, 138, 273, 275, 278, 305, 307 Sabanê 13, 437, 438 Sakurabiat 439 Uru-Chipaya 13, 17, 44 Sanumá 161 Uru 7, 13, 40, 41, 44–51, 53–55, 55n13, 56–60, secret language 43, 50, 72 66, 68, 68n26, 69–72 Shuar 7, 13, 40, 41, 46, 51, 56–58, 60, 64, 66, Urubu-Kaapor 273 67, 69–72 Siriono 13, 273, 275, 276, 278, 295–296, 306, Waimiri-Atroari 423 307 Warekena 115, 117, 305 Sirionó 431, 440, 454n32 Wari’ 3, 13, 432, 434, 442n14, 445n20, 450, Southern Nambikwara 13, 437, 438 454n32, 455, 456 Spanish 19, 21, 22, 30, 34, 54n11, 79, 86, 110, Wayãpi 273, 273n1, 275, 306, 307 232n11, 279, 293n11, 294, 322, 347, 348n2, Williche 17 348n3, 351, 352n6, 362, 369, 370, 432 Suruí 274 Xetá 274 Tacana 13, 44n3, 313, 330, 439 Tacanan 9, 313, 330 Tagalog 1 Tapiete 273n1, 275

117

Yanam 161 Yanomae 3, 4, 8, 13, 153, 161, 162, 162n1, 163–165, 165, 166, 168, 170, 174, 175, 177, 178, 180, 181, 313n1

470 Yanomamɨ 161 Yanomami 8, 13, 161, 162, 170, 177, 181 Yuhup 143n1 Yuki 13, 273, 275–278, 295–296, 306, 307, 431, 440

index of languages Yurakaré Zo’e

274

4, 13, 431, 452–454, 455, 456, 458

Index of Subjects absolutive 189n5, 189n6, 235, 236, 268, 328, 406n8, 412, 413, 421 abstract phoneme 186n2 active verb 8, 83, 129n9, 185, 186, 191, 192–198, 198–200, 204, 205, 208, 380, 381, 383, 387, 391 active-stative system 126 additive 457 adessive 451, 452 adjectival quantification 10, 415–418, 422 adstrate 428 adverbial quantification 403, 404, 407 affix reduplication 9, 27, 40, 191, 198, 205, 207, 211, 332, 353, 364n15, 371, 415–418, 423, 427 Africa 116n1, 330n19, 428n2, 456 Amazon basin 6, 7, 247, 273n1, 305, 313 Amazonas 161 Amazonia 2, 6, 8, 46, 79, 115–117, 143, 143n1, 144, 157, 178, 313, 330, 423, 429, 456 andative 444 anticipated 275, 289–291 anticipatory reduplication 276, 297 antipassive 259, 260, 313, 316, 318, 327, 328–331, 332, 332n21, 333, 335, 337–339, 439 archiphoneme 9, 185, 186, 202, 203, 213 areal diffusion 427, 428, 428n2, 430, 454, 456, 457 associative quality 59, 62, 63, 63n20 atelic 115, 132, 364, 365n16, 370, 421 attenuation 4, 192, 375, 383, 384–385, 396, 418, 423, 432 attenuative 4, 205, 320, 382, 396, 418, 431, 434, 447, 448, 456 attributive 8, 77–79, 96–105, 105–107, 108, 109n13, 143, 144, 148, 152–154, 157, 301, 315, 317, 438, 446, 455 augmentative 4, 5, 99, 99n6, 432, 444, 450 automatic reduplication 3, 48, 326–327, 333, 335, 364n15 auxiliary 87n3, 316, 317, 319, 328, 331–333, 335–337, 409, 413, 423

bare nouns 163 bidirectional 455 bimoraic reduplication 344, 349, 350, 351, 356–359, 364, 370 bimoraic 145, 147, 289, 344, 349, 356–358, 364, 365, 369, 391, 436, 451 Bolivia 4, 9–11, 41, 43, 46, 295, 296, 313, 343, 348n3, 375, 429, 431, 437, 439, 440, 442, 452, 458 borrowing 86, 227, 235n12, 243, 258, 370 Brazil 7–9, 11, 116, 117, 117n2, 144, 157, 161, 162, 162n2, 181, 185, 217, 218, 218n1, 234, 247, 287, 293n10, 305, 429, 439, 441, 457, 457n34, 458 causative 27, 27n8, 29n11, 124, 169, 187, 205, 208, 327, 336, 337, 355, 407, 443, 444, 453, 453n31, 455, 456, 458n35 classifier 82, 83, 83n2, 84–87, 98, 100, 101, 103–105, 162, 162n2, 163, 164, 166, 173, 178n6, 179, 180, 255, 258, 259, 259n13, 260, 344, 364, 365, 370, 380, 381, 381n8, 385–388, 391, 392n16, 393, 429, 430 climax 277 clitic 8, 44, 55n13, 92, 99, 163–166, 169, 170, 180, 299, 315, 318, 432–434, 455 coda 118, 120, 145, 146, 146n3, 147, 147n5, 154, 186, 220, 221, 241, 242, 250, 250n2, 251, 252, 256, 257n9, 269, 276, 277, 279, 285, 287, 291, 297, 303, 344, 346, 377, 379, 390 collective 4, 57, 98, 164, 456 collectivity 20, 72 comitative 3, 206, 451 comparative 33, 66, 247, 267, 451 complete reduplication 19, 171, 193, 205n27, 405 completion 4, 20, 403n5 complex predicates 78, 82–87, 87n3, 90, 91, 317, 319, 333 complex reduplication 3, 49, 50, 304, 432, 434, 443, 443n17, 445, 456 composition juncture 195

472 compound 8, 19, 52, 98, 103n9, 117n3, 119, 144–146, 146n3, 147, 148, 152–154, 155, 156, 156n10, 157, 166, 173, 217, 228, 228n9, 231, 232, 344, 345, 364–365, 368, 370, 427 conditionality 20 consonant gemination 19 continuative 330n20, 430, 441, 443n15, 445, 448 contrastive 144, 145, 443, 443n16, 444 copula suppletion 413, 415 creation of new lexical item 63, 64, 66 creole 61n17, 428, 428n2 CV structure 119, 162, 185 derivation 5, 7, 64, 78, 79, 108, 165, 166, 180, 185, 188, 190, 208–210, 212, 214, 255, 258, 366, 369, 422, 450n24 detransitivization 339 diachronic 33, 68n26, 79, 80, 82, 118, 144, 155, 157, 337, 442 dialects 8, 10, 17, 161, 162n2, 252, 256, 280n5, 294, 305, 375, 376, 379, 382n9, 394, 397 diminutive 4, 61, 61n16, 62, 65, 68, 70, 99n6, 232n11, 250, 432, 450 direct 10, 27n9, 42, 48, 52, 238, 343, 344, 350, 352n7, 354–359, 369–371 directional 42, 50, 164, 167, 429, 446 distensitivité 4 distribution 56, 59, 61, 64–66, 69, 70, 72, 73, 80, 108, 128, 129n9, 133–135, 143, 185, 203, 249, 258, 306, 320, 334n23, 336, 339, 404, 415, 428, 429, 439, 445n19 distributive 1, 4, 9, 23, 25, 31, 34, 132, 134, 135, 139, 192, 273, 276, 291, 292, 294, 297, 304, 305, 340, 407, 430, 431, 441, 444, 448, 451, 452, 453n30 distributivity 20, 23, 28, 115, 136, 139 disyllabic 9, 19–21, 32, 89, 118, 119, 124, 163, 165, 169, 173–175, 187, 193, 200, 208, 223, 224n6, 225, 273–275, 279, 279n4, 280–288, 290–293, 296, 300, 304–306, 315, 338, 345, 347, 348, 348n2, 351, 354, 357, 358, 362, 367, 369, 377n2, 383n10, 436, 449 ditransitive 29, 30, 192 documentation 6, 44, 371

index of subjects domain of reduplication 186, 205, 213 double copy 375, 376, 383n11, 392–396, 396 duration 8, 25, 80, 151, 161, 163, 170, 172, 180, 302, 333, 423 durative 10, 28, 29, 34, 84, 153, 153n6, 154, 172, 239, 242, 243, 330, 370, 384, 401, 402, 422, 430 echo 175, 175n4, 176, 306 Ecuador 6, 7, 41, 46, 77, 79, 110, 111 emphatic 33, 48, 58, 232, 232n11, 251, 440, 451 endangered 6, 375, 431 epenthetic 18, 145, 195n18, 250, 252, 256, 315, 319n6, 335, 338, 416 ergative 162, 165n3, 170, 316, 318, 328, 332 event plurality 433, 446 event-external 274, 281, 285, 286, 306 event-internal 274 event-internal plurality 383–384 evidential 284, 429 fixed segmentism 3, 269, 304, 388, 389 formal feature 39, 67, 71, 72 fossilized 348n4, 451 frequentative 4, 242, 274–276, 282, 283, 285, 286, 295, 300, 301, 382, 430, 431, 446, 448 full reduplication 3, 5, 10, 19, 20, 39, 40, 47–51, 52–55, 57–59, 61n16, 65, 66, 77, 78, 81, 82, 87, 96, 108, 109, 165, 171, 175, 176, 185, 190, 217, 220, 222, 223, 223n6, 224, 226, 227, 230, 241, 242, 255, 318–324, 325–328, 328, 331, 331–333, 333–335, 337, 338, 349n5, 352n6, 392n15, 397, 427, 429, 432–436, 438, 441, 446, 451, 452, 454 functional feature 7, 39, 67–69, 71, 72 future 1, 56, 58, 441n10 genetic relationship 17, 67 glottal stop 42, 43, 80, 119, 147, 147n5, 278, 344–347, 348n4, 349, 431n4, 452 grammaticalization 109, 110, 126, 130n10 Graz Reduplication Project 2, 39, 154n8, 163, 313n1, 322

473

index of subjects habitual 4, 275–277, 306, 326, 330, 333–335, 340, 370, 384, 435, 448 heavy syllable 107, 221, 226, 278, 289, 344, 345, 347, 349, 350, 356, 377, 448 iambic 118, 145, 146n3, 343, 350–353, 362–363, 365, 366 iconic 4, 8, 10, 39, 40, 51n8, 56–60, 61–64, 69, 70, 72, 87, 94, 110, 148, 153, 154, 157, 163, 165, 166, 170–176, 180, 274, 276, 300, 365, 370, 370n21, 401, 418, 447, 452, 454 iconicity 4, 10, 56, 61n16, 110, 157, 181, 234, 243, 243n18, 274, 370, 452 ideophone 7–9, 60, 77–80, 80–82, 82, 83, 87, 88, 94, 96, 98, 100, 107–109, 109n13, 110, 148, 154, 163, 176, 177, 217, 218, 234–240, 240n17, 241–244, 263, 322, 323, 339, 340, 401, 402, 419–421, 421–422, 437 idiomatic 103, 452 imperative 108n12, 155n9, 235, 419, 421, 435 imperfective 56–59, 66, 70, 72, 188, 189, 195, 195n19, 197, 204, 302, 330, 413 inalienable 10, 49, 353, 354, 360–361, 361, 363, 367, 369 inceptive 443, 451 inchoative 4, 32, 34, 83, 192, 212, 324, 340 inclusive/exclusive 429 indefinite pronouns 7, 20, 47, 164 inessive 451 infixation 269, 353, 445, 445n20, 453 inflection 5, 21, 44, 78, 120–126, 185, 188, 212, 249, 255, 316, 332, 415, 422, 450 inherent 4, 8, 9, 18, 32–33, 34, 144, 146n4, 147, 150, 153, 163, 178, 181, 218, 233–234, 299, 303, 313n1, 322, 324, 366n18, 402n1, 435 instrumental 93, 165, 165n3, 438 intensification 4, 53, 56–59, 62, 64–66, 69, 73, 78, 111, 115, 116, 128, 129, 133–136, 138, 139, 161, 163, 291, 297, 302, 396, 417, 418, 422, 423, 431, 437, 446 intensifying 30, 34, 49, 54, 57, 59, 61–63, 172, 293, 301, 304, 383, 417, 423, 432, 436–439, 445–448, 450, 453–455, 457n34 intensive 41, 45, 53, 286, 294, 295, 305, 330n20, 431, 433, 448

interrogative 108n12, 448 intransitive 25–27, 29, 29n11, 115, 116, 126, 126n6, 127, 128, 130, 132–133, 133–135, 138, 164, 166, 168–170, 170, 180, 185, 186, 188, 189, 189n5, 189n7, 190–192, 199, 201, 202, 204, 205, 205n27, 206, 208, 210, 212, 258–260, 263, 267, 316–318, 318n5, 319, 324, 326, 328, 329, 331–335, 335n24, 336, 362, 364, 365n16, 366, 380, 382, 406n8, 411, 412, 418, 421, 432, 443n17 inverse 10, 343, 344, 350, 353, 354–359, 359, 360, 369–371 irregular 176, 381, 443n16 isolate 4, 6, 7, 9, 10, 13, 17, 45, 46, 50, 218, 244, 343, 423, 427, 429–431, 441, 442, 444, 446, 447, 450, 452, 454n32 iteration 9, 20, 25, 78, 80, 94, 108, 242, 243n18, 267 iterative/repetitive 151, 430, 455 kinship

33, 164, 318, 321, 438

language contact 136, 144, 428, 429 lengthening 347, 348, 351, 352, 356, 367, 367n19, 368n20, 375, 379, 390, 396, 397 lenition 187, 196, 403n3, 416 lexeme 19, 33–35, 49, 64, 163, 175, 177, 178, 250, 274, 279, 281, 439, 457 lexical base 10, 19, 52, 53n10, 55, 56, 64, 68, 70, 353 lexical classes 18, 83, 147, 317 lexicalization 126, 147, 148, 153n6, 156n10, 177–180, 199n22, 210, 212, 213, 231, 392n16 lexicalized 31, 34, 44, 47, 51, 60, 62n18, 64, 73, 97, 106, 143, 144, 147, 148, 152, 153n6, 154, 157, 168, 192, 201n25, 211–213, 231, 233, 301, 366, 368, 413, 453–455 lexifier 43, 428 light syllable 217, 220, 221, 223n6, 224n6, 226, 229, 233, 241, 278, 345–347, 349–351, 362, 365, 370, 371 linguistic area 6, 11, 79, 427–430, 456 linguistic exogamy 144 loanwords 19, 119, 130n10, 137

474 locative 48, 96, 164, 167, 169, 172, 284, 299, 446 ludic (creation, process, strategy) 276, 290, 299, 305 mass noun 133, 133n12, 403, 414n11 metrical system 118, 119 Middle Negro 8, 116, 133, 134–135, 136 middle voice 10, 343, 344, 350, 352n7, 353, 354, 359–360, 453n28, 453n31, 454 minimal word 124, 397 minimality 338, 347, 376, 377, 391, 392 mixed language 6, 44, 71 modality 21, 316, 430 monomoraic 241, 242, 279, 281, 289, 292, 296, 344, 349, 369, 391, 392, 451 monomoraic reduplication 344, 349–350, 352–353, 355–357, 370 monosyllabic 9, 19, 89, 119, 124, 138, 145, 147, 163, 166, 173, 187, 191, 195, 197, 199, 200, 211, 213, 220, 222, 224, 225, 241, 269, 270, 273–277, 278n4, 279, 281–291, 293, 295, 296, 300, 302, 304, 306, 315, 319n6, 334, 335, 338, 344, 350, 354–356, 358, 359, 366–368, 386, 390–392, 392n15, 436, 445 mood 41, 187, 188, 196, 300n14, 316, 344, 402n2, 404n6, 419, 421, 435 mora 3, 220, 220n3, 278, 281, 343–349, 367, 368, 391, 434, 435, 443, 448, 450n24, 451–453 morpheme boundaries 88, 148, 185, 188, 193n15, 202, 274, 284, 403n3, 449 morphologically-based 448, 454 morphology-prosody interface 10, 375, 376, 397 morphophonemic 27n8, 42, 43, 45, 46, 50, 276, 289, 306, 354n9, 393 morphophonology 254, 276, 416–418 multiplicity 128, 129, 138, 150, 151, 274, 414, 457 nasal harmony 186, 199n22, 203 nasalization 144, 144n2, 145, 175, 185, 199n22, 203, 250, 297 negation 209n30, 235, 284, 445n19

index of subjects Negro Içana 8, 116, 133–134, 135, 136 NeighborNet 7, 39, 66 nominal number 4, 431, 446 nominal plurality 56, 130 nominalization 4, 228, 230–232, 361 nominalizer 51, 127, 405n8 nominalizing 165–170, 170, 186, 355, 360, 391, 436 nucleus 131, 220, 290 number 1, 5, 9, 11, 20, 21, 25, 28, 29, 35, 40, 41, 43, 45, 46, 54n11, 62, 71–73, 78–80, 98, 104, 109, 126, 129–132, 136, 143, 154, 157, 162, 163, 191, 208, 211, 213, 218n1, 220, 220n3, 221, 229, 231, 241–243, 262, 267, 268–270, 270n19, 277, 277n3, 284, 294, 315, 317, 322, 324, 329, 344, 362n12, 370, 381n6, 384, 396, 403, 407, 409, 411, 414, 414n11, 427, 428, 428n2, 429, 431, 436, 437n7, 442, 444, 445, 448, 454, 455, 457n34 numeral 9, 49, 62n18, 93, 130n10, 220, 230, 231, 233, 241, 404, 457 numerals 230 object omission 8, 115, 116, 132, 138 onomatopoeia 8, 176, 177–180, 227, 233, 247, 254, 254n7, 255, 257, 265–266, 304, 315, 322, 339, 340, 437, 441n10, 457 onomatopoeic 4, 34, 60, 148, 163, 175, 180, 234, 239, 274, 305n17, 324, 325, 401, 418–419, 422, 435, 447 onset 119, 145, 220, 249, 249n1, 250, 287, 289, 290, 301, 305n17, 345, 377 opacity 97–99, 395 original 2, 18n1, 30, 139, 161, 165, 176, 180, 190–192, 195, 198, 205–208, 210, 211, 319, 337, 391, 403n5, 427, 429, 456 Pacific Coast 6, 79 paroxytonic stem 187 partial reduplication 3, 5, 7, 10, 19, 20, 32, 39, 40, 47, 49n5, 51–56, 58–60, 65, 65n23, 66–68, 70, 71, 77–79, 81, 82, 87, 96, 100, 108, 109, 109n13, 116, 120, 161, 163, 165, 166, 172, 174, 175, 191, 217, 220–223, 223n6, 224–226, 229, 230, 241, 242, 324, 337, 338, 375, 396,

index of subjects 397, 427, 431, 432, 435, 436, 438–440, 442, 444, 446–448, 450n24, 451–454 participant plurality 4, 9, 267, 431, 432, 446 persuasive speech 276 phonologically underspecified affix 343, 353 phonotactic 5, 8, 107, 143, 186–187, 438, 440, 445n19, 448, 451, 452 pidgin 428 pluractional 381, 384, 407, 414, 434, 444–446, 453n30 pluractionality 4, 7, 10, 115, 116, 128–130, 132, 136, 138, 139, 191, 257, 265–266, 266n17, 384, 401, 404, 414–415, 422, 423, 431, 440, 454, 456 plural/intensive 432 plurality 7, 9, 10, 20, 20n2, 29, 31, 56, 57, 59, 65, 66, 69, 70, 72, 84, 180, 192, 222n4, 247, 264, 266, 267, 273, 274, 276, 297, 304, 306, 343, 384, 401, 402, 405, 413, 422, 423, 436, 446, 450–452, 454 plurality, event-external 9, 273, 275, 277, 282, 285, 289, 291, 292, 294, 296, 297, 299–301, 303, 304, 306, 384 plurality, event-internal 4, 9, 273, 275, 276, 280, 285, 287–289, 291–294, 296, 297, 300–302, 302n16, 303, 304, 306, 384, 431 pluralizer 231, 236 polysyllabic 174, 334, 445, 446 polysynthetic 6, 316, 379 possession 10, 49, 304, 351, 353, 354, 360–361, 361, 362–363, 367–369, 452 possessive 10, 43, 48, 49, 55, 343, 344, 362, 362n13, 363, 363n14, 367, 369, 370, 430, 454 postposition 127n7, 129n9, 162, 190n9, 314, 315, 323, 324, 340, 404 potential 86, 427, 456 predicative 315–317, 320, 320n7, 321, 322, 324, 337, 338, 351, 354, 362–363, 438 prefixational 115, 116, 120, 138 productive 3, 4, 7, 11, 17, 20, 21, 23, 28, 33–35, 40, 44, 51, 52, 60, 71, 73, 77–79, 81, 82, 86, 91, 97, 99, 100, 103, 103n9, 104, 105, 108, 110, 110n14, 145, 151–153, 190, 191, 212, 223, 224, 230, 231, 233, 241, 257, 318, 320–323, 327–337, 338, 345, 350, 353, 362, 364–369,

475 369, 370, 395, 401, 405, 427, 428, 430–432, 435, 437–439, 441, 442, 444, 447, 448, 452–454, 456, 456n33, 457, 458 progressive 4, 25, 29, 30, 34, 56–58, 65, 66, 69, 71, 87n3, 115, 133, 139, 188, 189, 196, 204, 204n26, 276, 302, 413, 423, 430, 433–435, 439, 441, 443, 445–448, 451, 455 prosodic foot 115, 116, 118–124, 138 prosodic morphology 267, 343, 349, 353, 371, 391, 397 prosodic word 345–346, 354n9, 367, 378, 391 prosody 3, 106, 107, 254, 276, 345 pseudo-reduplication 441n10 punctual action 239, 242 quantifier 9, 133, 162–164, 222n4, 230, 231, 273, 314, 315, 323, 340, 382n9, 401, 403, 404, 407, 415–418, 422, 423, 450 question words 7, 20, 24, 315, 382n9 quotative 449, 450 reciprocal 4, 8, 115, 116, 124, 136–139, 187, 210, 210n32, 211, 316, 432, 450, 455 reciprocity 20, 115, 116, 136–138, 139, 211n32 recursion 450–452 recursive 2, 60, 430, 438, 441n10, 457 referentiality 138 reflexive 8, 29, 29n11, 50, 115, 116, 124, 136–139, 187, 210, 211, 211n32, 316 reflexivity 115, 137–139, 211n32 regressive reduplication 10, 344, 352 reiterative 254, 269, 333 relative clause 127, 332n21, 362n13, 451 repetition 1–3, 5, 7–10, 17, 19, 20, 28, 33, 35, 39, 51n8, 58, 77, 78, 80, 82, 108, 115, 132, 136, 143, 144, 146–151, 154, 155–156, 156n10, 157, 175, 177–180, 217, 221, 231, 233, 234, 242, 243, 247, 254, 265, 274, 275–276, 277, 279–282, 293–295, 314n2, 319, 320, 325–327, 330–333, 335, 336, 339, 340, 375, 383–384, 384–385, 396, 401, 402, 419–422, 423, 427, 432, 437–439, 443, 447, 457 repetitive 4, 10, 51, 150–152, 154, 242, 322, 383, 384, 396, 401, 418, 419, 431, 433–439, 441, 444, 445, 447, 448, 451–454, 457

476 resultative 23, 23n6, 208, 253n6, 451 retardation 276, 290, 305 rhythm 118, 276, 305, 322, 324, 370, 378 semantic weakening 281, 304 sesquisyllable 258, 259, 267 simple reduplication 3, 50, 338 sonorous 249, 322, 324 sound symbolism 87, 242, 243, 437 source 2, 7, 9, 18, 43–46, 49n4, 55, 55n13, 67, 77, 110n14, 118, 127, 177, 243, 282, 282n6, 317n4, 344, 348, 352, 430, 431, 431n4, 437, 446 stative verb 8, 115, 116, 127, 129, 129n9, 133–136, 138, 139, 143, 152, 162–164, 166–168, 168, 169, 171, 172, 180, 185, 186, 188, 190–192, 194n16, 199n23, 200, 204–205, 208, 209, 213 stem formatives 26, 27, 31, 407, 408n10 stock, linguistic 9, 244, 273, 273n1, 274, 275, 306 stress 19, 42–46, 80, 106, 107, 118, 119, 121, 145, 146n3, 156, 162, 187, 191–193, 198, 199n22, 219, 226n7, 234, 250, 254, 277–278, 344, 346–348, 354n9, 377, 378, 381, 381n7, 386, 402n1, 402n2, 443n16 subordination 51, 61, 188, 235, 237, 428 successive 91, 242, 274–276, 279, 283, 285, 287, 288, 296, 297, 306, 389 superlative 451 suppletive 2, 10, 401, 409–411, 413–415, 423 Surinamese Creole 428n2 symbolic 4, 10, 100, 154, 178, 234, 444, 457 synonyms 2, 102 syntactic effects 116, 313, 338–340 syntactic function 45, 313, 314 telic 132, 421 template 79, 96–107, 146, 148, 255, 256, 257n10, 258, 259, 270, 349n5, 396n21 templatic affix 349 temporal 81, 94, 320, 331, 339, 362n12, 447 tense 20, 41, 58, 162, 190, 235n14, 275, 277, 277n3, 294, 299, 331, 344, 405n7, 423, 430, 433, 441n10, 448, 450

index of subjects tense/mood 316, 433 thematic vowel 166, 169, 173, 175, 195, 195n18, 196–199, 199n21, 203, 204, 204n26, 205, 208 total reduplication 10, 222, 241, 397 transitive 4, 26, 27, 29n11, 115, 116, 126, 126n6, 127–130, 132, 138, 166, 170, 174, 180, 185, 186, 187n3, 189–192, 194, 205, 205n27, 206, 207n28, 208–212, 258–261, 262n15, 263, 264, 267, 301, 313, 316–318, 318n5, 319, 324, 325, 325n15, 326–332, 332n21, 333–336, 336n25, 337, 343, 354, 364–365, 380, 382, 406, 410, 421, 432, 433n6 transitivity 9, 20, 28, 126, 138, 259–261, 264, 267, 314, 316, 318, 328, 338 transitivized 25, 169 transitivizing 436 triplication 392, 433n6, 440, 456 trisyllabic 19, 21, 187, 224, 224n6, 225, 241, 269, 304, 320n8, 321n10, 327n16, 347, 367, 392, 445 truncation 247, 368, 369 typological linguistics 1, 429 unboundedness 4 universal 5, 155, 243, 403, 411, 428, 430, 432, 454, 456–458 valence 10, 190, 313, 316, 330, 338, 439, 456 valency 26, 211, 259, 319, 326, 331, 333, 430, 439, 443, 455, 456, 458 varieties 8, 18, 31n12, 35, 41, 48, 60, 60n14, 61, 65n22, 115–117, 129n9, 133, 136 Vaupés region 144 verbal number 4, 253, 257, 262, 270, 431, 446 verbal plurality 384 verbalization 21n3, 23n7, 86, 90, 164, 172–174, 176, 177, 299, 301, 324, 436, 437 voice 187, 210, 213, 250, 452 vowel deletion 52, 375–377, 377n2, 378, 378n4, 379, 381, 381n6, 381n8, 382, 385n13, 386, 386n14, 389–390, 393, 393n18, 394–397 vowel lengthening 19, 106, 107, 176, 379, 453n29 Watorikɨ

161, 181

477

index of subjects word class 10, 53, 57, 59, 82, 83, 97, 100, 103, 105, 107, 154, 165, 218, 221, 229, 230, 244, 313, 338–340, 419

word order 42, 92, 162, 218 word-internal reduplication 357

344, 350, 353,