Within Language, Beyond Theories (Volume I) : Studies in Theoretical Linguistics [1 ed.] 9781443879859, 9781443872041

Citation preview

Within Language, Beyond Theories (Volume I)

Within Language, Beyond Theories (Volume I) Studies in Theoretical Linguistics Edited by

Anna Bondaruk and Anna Prażmowska

Within Language, Beyond Theories (Volume I): Studies in Theoretical Linguistics Edited by Anna Bondaruk and Anna Prażmowska Reviewed by Dr. hab. Gréte Dalmi (Jan Kochanowski University) Prof. Dr. hab. Bogdan Szymanek (John Paul II Catholic University of Lublin) This book first published 2015 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2015 by Anna Bondaruk, Anna Prażmowska and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-7204-0 ISBN (13): 978-1-4438-7204-1

TABLE OF CONTENTS

Foreword .................................................................................................... ix Contributors .............................................................................................. xxi Part I: Studies in Syntax and Morphology Chapter One ................................................................................................. 3 The Nominal Left Periphery in Slavic: Evidence from Croatian Durdica Zeljka Caruso Chapter Two .............................................................................................. 27 The Categorial Status and Internal Structure of NPN Forms in English and Polish Wiktor Pskit Chapter Three ............................................................................................ 43 The DP Analysis of Polish Clausal Subjects Maágorzata Wietrzyk Chapter Four .............................................................................................. 59 How Relative are Cleft Clauses in English it-Clefts? Ewelina Mokrosz Chapter Five .............................................................................................. 79 A Minimalist Approach to Double Object Constructions in German: Their Syntactic Representation and Passivisation Possibilities Aleksandra Bartczak-MeszyĔska Chapter Six ................................................................................................ 98 Dual Interpretations of Object Neg-wh-Quantifiers (Neg-whQ) in Cantonese: A Feature-Based Approach Man-Ki Theodora Lee

vi

Table of Contents

Chapter Seven.......................................................................................... 116 Towards an Analysis of Clitic Doubling in Romanian Alina Tigău Chapter Eight ........................................................................................... 134 Old English Transitivity: A View from the Past Magdalena CharzyĔska-Wójcik Chapter Nine............................................................................................ 157 The Importance of Root Indices: Evidence from Arabic Nouns Abdullah S. Alghamdi Chapter Ten ............................................................................................. 172 Verbal Causative Structure for Polish Prefixed Transitives in the Root Based Approach Anna Malicka-Kleparska Chapter Eleven ........................................................................................ 188 The Linearization of Adjectives in Polish Noun Phrases: Selected Semantic and Pragmatic Factors BoĪena Cetnarowska Chapter Twelve ....................................................................................... 206 Onomasiological Strategies and Potential Forms: In Search of the Female Counterpart of Minister in Polish Maria Bloch-Trojnar Part II: Studies in Phonetics and Phonology Chapter Thirteen ...................................................................................... 223 Schwa: Phonological Coding versus Phonetic Manifestation Anna Bloch-Rozmej Chapter Fourteen ..................................................................................... 243 The Phonetics and Phonology of the Scottish Laterals Mateusz Urban and Sáawomir Zdziebko Chapter Fifteen ........................................................................................ 259 Welsh Epenthetic Vowels in the CV-Model of Government Phonology Tomasz Czerniak

Within Language, Beyond Theories (Volume I)

vii

Chapter Sixteen ....................................................................................... 274 On the Possibility of English and Slovak Consonant Clusters to Form Complex Segments Renáta Gregová Chapter Seventeen ................................................................................... 293 Within Phonology, Beyond GP: An Attempt to Determine the Structure Behind the West Saxon Digraphs Karolina Drabikowska Chapter Eighteen ..................................................................................... 312 Prosodic Actualization of a Moral Admonition in Small Folklore Texts Larysa Taranenko Chapter Nineteen ..................................................................................... 322 Phonetic Studies from the Perspective of an Energetics Approach Alla Kalyta Chapter Twenty ....................................................................................... 337 Communicative Entrainment and Linguistic Representations Štefan BeĖuš Part III: Studies in Cognitive Linguistics Chapter Twenty One................................................................................ 361 Not All Dads are Conceptualized the Same (Even if All are Created Equal) Adam Gáaz Chapter Twenty Two ............................................................................... 374 Motivation for Meaning Construction: Historical Semantics of English ‘Joy’ Vocabulary Konrad ĩyĞko and Angelina ĩyĞko Chapter Twenty Three ............................................................................. 386 Mental Spaces Beyond and Within: A Conceptual Blending Analysis of Monty Python’s Humour Joanna JabáoĔska-Hood Chapter Twenty Four ............................................................................... 397 Are Metonymic Part for Whole Relations a Mere Illusion? Wojciech Wachowski

viii

Table of Contents

Chapter Twenty Five ............................................................................... 407 Men’s Talk versus Women’s Talk in the Metaphorical Representation of Boil, Cook and Simmer in English: A BNC-Based Study Marietta Rusinek Index ........................................................................................................ 427

FOREWORD

The monograph Within Language, Beyond Theories presents a collection of insightful studies pertaining to the most perplexing problems in the areas of theoretical and applied linguistics. Contributors offer accounts of new evidence drawn from a number of the world’s languages and analyses that surpass the limits of contemporary frameworks in search of more explanatorily adequate solutions to linguistic dilemmas. We delve into the previously unexplored areas of linguistic reality, aiming to gain insight into the structure of the system and establish laws governing its inner organization. Importantly, linguists of different persuasions share the belief that our enhanced understanding of the grammar of language and its constituent modules will foster new advances in the novel application of the models proposed. Assisted by innovative ideas in corpus studies, translators and discourse researchers will be able to make invaluable contributions to the development of their fields. Volume One, entitled Studies in Theoretical Linguistics, comprises twenty five chapters, organized into three parts. Part I, called Studies in Syntax and Morphology, consists of twelve chapters devoted to current developments in syntactic and morphological theorizing. The leading framework adopted in the syntactic works is the Minimalist Program (henceforth, MP) of Chomsky (1995, 2000, 2001, 2008), in which the derivations are taken to be minimal and based on independently motivated extra-grammatical, cognitive constraints (cf. the third factor of Chomsky 2005). Another model that stems from the MP, which is used in Caruso’s work (cf. Chapter One) is the cartographic model (cf. Rizzi 1997; 2004, Cinque 1999; Belletti 2004, among others), in which the left periphery of the clause is split into several distinct projections to incorporate items such as topics, foci, force markers, finiteness markers, etc. By analogy with the CP, the split has recently been postulated within the structure of a DP to account for focalization, topicalization, informational prominence and quantification (cf. Ihsane and Puskás 2001; Aboh 2004; Länzlinger 2005, 2010; Giusti 2005, among others). The main syntactic area analysed in Part I relates to the DP in various languages, including English, Croatian, Polish, and Romanian (cf. the contributions by Caruso, Pskit, Wietrzyk, Tigău, and Cetnarowska), which is approached from distinct angles, including its internal structure, the order of elements inside a DP, the

x

Foreword

treatment of certain structures as DPs, and the sensitivity of clitic doubling to the type of DP involved. Information structure is central to Caruso’s and Mokrosz’s chapters, while Case valuation and ij-feature agreement figure prominently in Bartczak-MeszyĔska’s and Mokrosz’s analyses. The notion of transitivity is examined in two different languages from two different theoretical standpoints by CharzyĔska-Wójcik and Bartczak-MeszyĔska. The former focuses on the definition of transitivity in Old English, while the latter concentrates on the structure of the double object construction in English and German. Various interpretational possibilities are discussed in the chapters by Cetnarowska and Lee, with respect to the position of adjectives in a Polish DP in the former and negative quantifiers in Cantonese in the latter. There are three morphological chapters in Part I, which represent two very different traditions of morphological description. The chapter by Bloch-Trojnar adopts the Lexeme Morpheme Base Morphology of Beard (1995), in which morphology is taken to be a part of the lexicon. The contributions by Alghamdi and Malicka-Kleparska have their roots in the Chomskyan vision of grammar, with grammatical structure constituting the backbone of any linguistic analysis. Alghamdi’s work adopts the framework of Distributed Morphology (cf. Marantz 1984; Halle and Marantz 1993, 1994; Halle 1997; Harley and Noyer 1999, 2000; Arad 2005; Embick and Noyer 2007, inter alia). Malicka-Kleparska’s analysis utilises the root based approach and thus belongs to a new trend in structure dependent morphology, as outlined in Alexiadou and Anagnostopoulou (2004), Pylkkänen (2008), Doron and Labelle (2010), Embick (2009), Alexiadou and Doron (2012). The morphological chapters focus on data from Polish and Arabic. Part I of the volume is organized as follows. The first two chapters address the syntactic structure of the nominal domain of two Slavic languages – Croatian and Polish. Chapter One presents an analysis of the structure of the nominal left periphery within the split-DP approach. This approach stems from the split-CP analysis of clauses, where several functional projections, such as TopicP or FocusP, are located above the TP. Analogously, the split-DP analysis assumes the existence of a functional structure above the noun, which is generally associated with the category of determiners. It is argued that Croatian, although an articleless language, possesses a rich functional structure above noun phrases, which conveys information pertaining to the notions of (in)definiteness and specificity. Chapter Two examines the categorial status and the structure of NPN forms (e.g., day after day) in English and Polish. The categorial status of

Within Language, Beyond Theories (Volume I)

xi

these constructions has not yet been conclusively established in the literature and ranges from nominal (cf. Quirk et al. 1985), through quantificational (cf. Travis 2001, 2003), to prepositional (cf. Haïk 2013). Here, it is proposed that Polish NPNs belong to the nominal domain and, as in English, they seem to be derived through syntactic reduplication. Chapter Three also aims at establishing a categorial status but, this time, of Polish clausal subjects. The properties of Polish nominal and clausal subjects were juxtaposed and the subject status of CP subjects positively tested with reference to obligatory raising, agreement, coordination, control and case. It is proposed that Polish clausal subjects do have a nominal status and they do possess a DP layer within which the head of the DP is realized by to ‘it.’ Chapter Four addresses the syntax of clauses as well. More precisely, it offers an argument for the reanalysis of cleft clauses as relative clauses. Some significant similarities in agreement patterns between relative and cleft clauses are identified and employed as a starting point for an investigation into the possibility of analyzing the English it-clefts as either appositive or restrictive relative clauses. Chapter Five provides an insight into three types of Double Object Constructions in German, realized by different case patterns (dativeaccusative, accusative-dative, double accusative), and related structures employing a PP as one of the objects. The discussion of the structure and derivation of German DOCs is supplemented with an examination of certain processes that both objects undergo, such as Case and ij-feature valuation and passivisation. Chapter Six explores the syntax of object negative wh-quantifiers (Neg-whQ) from the perspective of a feature-based approach. The language under investigation is Cantonese, where object Neg-whQs are unique in that they can yield both a negative and an existential reading. In order to account for the overt raising of object Neg-whQs and their dual interpretation, two features are proposed, namely [Quant] (interpretable and strong), responsible for the movement phenomenon, and [uNeg], to which the distinction between the existential and negative reading is attributed. Chapter Seven offers an investigation into the clitic doubling phenomenon in Romanian, which is obligatory with definite pronouns, optional with proper names, definite descriptions and indefinites, and ungrammatical with bare quantifiers. It is observed that the presence of clitic doubling and the presence of PE marking in Romanian, although often correlated in the literature, are not always mutually dependent on each other. In the proposed analysis, the clitic is considered to be an

xii

Foreword

agreement marker and the double is argued to be an argument, rather than an adjunct. Chapter Eight offers a critical insight into the transitivity of Old English verbs viewed from the perspective of two traditional approaches to the concept of transitivity. One of them is referred to as the quantitative approach, where the number of arguments is taken into account, and the other one as the qualitative approach, concerned with the types of arguments. It is demonstrated that, although separately they are insufficient to successfully deal with all the intricacies of Old English passivisation facts, when applied jointly, the two approaches can cover a significant part of these facts (excluding, however, the issue of object case alternation). Chapter Nine examines the nature of roots within the framework of Distributed Morphology. It is shown that, at least in Standard Arabic, roots are devoid of compositional semantic features, which have been claimed to identify roots (Marantz 1995, 1998). Instead, it is proposed that a root index, i.e., the property of a root morpheme containing a primitive conceptual feature (Pfau 2009), is responsible for the proper identification of the root in the Vocabulary and the Encyclopedia components. Chapter Ten is concerned with the morpho-syntax of a group of prefixed causative verbs in Polish which lack corresponding synthetic anticausatives, despite their semantics and despite the availability of an applicable morphological pattern for anticausative formation in Polish. The proposed analysis, couched within the root-based approach, offers a bipartite structure where the prefixes of the aforementioned causatives function as heads of the active voice projection. The focus of Chapter Eleven is on attributive adjectives in Polish, in particular on classifying adjectives occupying not only post-nominal but also pre-nominal positions. The difference in interpretation between complex predicates, consisting of a noun followed by a classifying adjective (‘tight units’), and adjective + noun sequences is demonstrated and investigated with reference to some semantic and pragmatic factors. Chapter Twelve aims at establishing the female counterpart of the Polish noun minister ‘minister’ on morphological grounds. Several possible candidates are closely examined with reference to the etymology of the word minister, the patterns of paradigmatic derivation, markedness relationships, the economy principle, and the increasing tendency towards analytic structures in Polish. It is argued that pani minister ‘Mrs minister’ is the linguistically optimal form since the analytical form (as opposed to the synthetic variants) is semantically more precise and unambiguous, and,

Within Language, Beyond Theories (Volume I)

xiii

additionally, it is deemed to be more prestigious due to the use of the honorific pronoun. Part II of the volume, entitled Studies in Phonetics and Phonology, consists of eight chapters. The main research area in current phonological studies concerns the interface between phonology and phonetics. The issue of the phonology and phonetics interface arises only if these two disciplines are viewed as separate computational systems. Then, their interface can be taken to represent a spell-out operation converting the phonological output into units of phonetic representation (cf. Scheer 2014). If, on the other hand, phonology and phonetics are not treated as autonomous, but rather as closely integrated, then they must be seen as not interfacing with each other (Ohala 1990). The interface between phonology and phonetics is central to the chapters by Bloch-Rozmej, and Urban and Zdziebko. Some aspects of suprasegmental phonology are studied by Taranenko. Kalyta investigates speech energetics, i.e., the peculiarities of emotional utterances and their phonetic organization, relying, to some extent, on other disciplines, including pragmatics, stylistics, semantics, etc. BeĖuš, in turn, goes beyond the Articulatory Phonology of Browman and Goldstein (1986, 1989, inter alia) to examine entrainment, i.e., the tendency of interlocutors to synchronize, coordinate and speak in a more similar way over the course of mutual communicative interactions, manifested, among others, in the sphere of articulatory gestures. The predominant theoretical framework adopted in the phonological works is the Government Phonology of Kaye, Lowenstamm and Vergnaud (1985, 1990) (cf. also Harris 1994; Gussmann 2007; Cyran 2010; Bloch-Rozmej 2008). The model, which operates on just one level of representation and makes use of a limited number of primes, called elements, is adopted in the chapters by Bloch-Rozmej, Urban and Zdziebko, Czerniak, and Drabikowska. Another model which prominently figures in the work by Gregová is the framework proposed by Duanmu (2009). Duanmu (2009) suggests that all surface strings may be reduced to a fixed and invariant syllable template, C(onsonant) V(owel) X, and thus stands in sharp opposition to the CVCV model, advocated by Lowenstamm (1996), Scheer (2004), and Cyran (2010), inter alia. The languages that are analysed in the eight chapters devoted to phonology and phonetics include: English, Slovak, Welsh, Scottish English, Old English, and German. Part II is organized as follows. Chapter Thirteen provides a discussion of the nature of the mid central vowel schwa from the perspective of the Government Phonology framework. The distinct properties of representation and distribution of the English and the French schwa are demonstrated and explored. It is shown that, in English, the representation

xiv

Foreword

of the schwa vowel is that of a mono-elemental headless segment, whereas the French schwa is a phonetically realized ungoverned empty nucleus. Chapter Fourteen offers an empirical insight into the phonological constitution of Scottish laterals in terms of the Element Theory. On the basis of acoustic measurements, it is argued that the elements A and U cannot and do not constitute the representation of the Scottish lateral //. Instead, it is proposed that the non-vocalized laterals contain elements U and I (and, in the vicinity of /‫ܧ‬/, an additional element A), whereas vocalized laterals contain only the element U. Chapter Fifteen is concerned with melody spreading, in particular with vowel epenthesis and tonic lengthening in Welsh. The chapter offers a uniform representation of the two phenomena within the CV-model of Government Phonology, supporting the analysis with rich data from South Welsh. Chapter Sixteen examines English and Slovak word-initial consonant clusters and, applying the CVX syllable model, challenges the claim that, cross-linguistically, most consonant clusters can be reanalysed as complex segments (Duanmu 2009). It is shown that, although they can be phonemically considered as complex segments, phonetically, these clusters are much longer than single segments, which stands in direct opposition to the claim under investigation. Chapter Seventeen explores the structure of two Old English digraphs and from the perspective of the Government Phonology 2.0 framework (Pöchtrager 2006, among others), which borrows (and adapts for phonology) some concepts from minimalist syntax. It is demonstrated that GP 2.0 offers tools which can be efficiently employed in the analysis of certain sets of data that often prove problematic for the standard GP theory. Chapter Eighteen aims at establishing the correspondence between the pragmatic features of a fable containing moral admonition and its prosodic organization. An algorithmic analysis of several fable plots is conducted, resulting in the identification of two distinct patterns with distinct prosodic organization displaying differences in such prosodic features as volume, voice range, tempo, rhythm, melodic contour, and the frequency and duration of pauses. Chapter Nineteen offers a discussion of the energetic approach to phonetic research proposed therein, which presupposes that phonetic phenomena are the outcome of a non-deterministic speech generation process motivated by psycho-physiological energy redistributed over the speaker’s mental spheres of the conscious, the subconscious and the unconscious. The discussion is supplemented with a quantitative analysis

Within Language, Beyond Theories (Volume I)

xv

of different levels (high, mid and low) of the emotional-and-pragmatic potential of utterance actualization. Chapter Twenty investigates the nature of communicative entrainment, focusing on the search for a formalized link between linguistic (and also non-linguistic) entrainment and cognitive representations, or, in more general terms, a link between linguistic and social cognitive systems. In particular, the proposed analysis advocates modelling (acoustic and prosodic) entrainment analogously to the modelling of communicative intention. Part III, called Studies in Cognitive Linguistics, comprises five chapters. Cognitive Grammar represents a theory of grammar developed by Langacker (1991, 2000, 2009), which differs from the Generative Grammar of Chomsky in that it primarily focuses on the meaning of linguistic expressions, not on their form. Meaning in Cognitive Grammar is seen as flexible and dynamic, and as involving conceptualisation (Langacker 1991). In this framework, language is seen as a part of the cognitive faculty, not as a module autonomous with respect to other linguistic cognitive systems, as in the generative tradition (Croft and Cruse 2004). Some specific models stemming from Cognitive Grammar that have been adopted in the contributions by JabáoĔska-Hood and Rusinek correspond respectively to the Conceptual Integration Theory, devised by Fauconnier and Turner (2006), which is based on a mental spaces theory and the Conceptual Metaphor Theory, proposed by Lakoff and Johnson (1980, 2003) and Kövecses (2010). The major areas of research in Cognitive Grammar that have found their way into this volume include metonymy (cf. WiliĔski’s chapter), metaphor (see Rusinek’s chapter), conceptualisation or construction of meaning (cf. Gáaz’s, and ĩyĞko and ĩyĞko contributions), and mental spaces blending (see JabáoĔska-Hood’s work). The data in the final five chapters of the volume come primarily from English, including its earlier varieties, such as Old and Middle English. Part III is organized as follows. Chapter Twenty One examines the non-standard use of English articles in terms of Langacker’s model of Cognitive Grammar. In particular, the study focuses on the phrase the dad, employing in the analysis the juxtaposition of the concepts of actuality and virtuality. Chapter Twenty Two offers a discussion of the study of meaning construction, understood as the process of a word acquiring new senses, and its relation to the study of historical semantics. It is shown, on the basis of an analysis of the diachronic semantics of the English ‘joy’ vocabulary (bliss, cheer and delight), that the motivation underlying

xvi

Foreword

meaning construction is extra-linguistic in nature, being dependent on such factors as experience, mental associations and the influence of the dominant ideology in a particular society. Chapter Twenty Three provides an analysis of the notion of humour based on Monty Python’s work and couched within conceptual blending theory put forward by Fauconnier and Turner (2006). It is suggested that Conceptual Blending Theory, operating on mental spaces and their integration, i.e., blending, could prove to be a major stepping stone to creating a uniform theory of humour applicable to various kinds of comedy. Chapter Twenty Four is devoted to an investigation into one of the major metonymic relations, namely ‘the part for the whole’ relation. It is observed and illustrated that this relation, or rather its name, in particular the term ‘whole,’ is misleading in presupposing too wide a domain; hence, the term ‘metonymic expansion’ is proposed as a more precise one. The final chapter of this volume, Chapter Twenty Five, presents a corpus-based study of the gender differences in the metaphorical uses of ‘cooking’ terms, such as boil, cook and simmer, based on data from the British National Corpus and conducted within the Conceptual Metaphor Theory (Lakoff and Johnson 1980). It is demonstrated that men tend to use ‘cooking’ terms in the metaphorical sense more frequently than women. The gender distinctions made in employing different kinds of conceptual metaphors constructed on the basis of ‘cooking’ terms are also examined and discussed. We would like to thank Prof. Anna Malicka-Kleparska for the various forms of assistance she has provided during our work on this volume. We are grateful to Dr Sáawomir Zdziebko and Marietta Rusinek, for helping us to understand the recent developments in phonology and Cognitive Grammar. We are also deeply indebted to ElĪbieta Sielanko-Byford and Nigel Byford for having proofread the entire volume. Anna Bondaruk and Anna PraĪmowska

References Aboh, Enoch. 2004. “Topic and Focus within D.” Linguistics in the Netherlands 21:1–12. Alexiadou, Artemis, and Elena Anagnostopoulou. 2004. “Voice Morphology in the Causative-Inchoative Alternation: Evidence for a Non-Unified Structural Analysis of Unaccusatives.” In The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface,

Within Language, Beyond Theories (Volume I)

xvii

edited by Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert, 114–36. Oxford: OUP. Alexiadou, Artemis, and Edit Doron. 2012. “The Syntactic Construction of Two Non-Active Voices: Passive and Middle.” Journal of Linguistics 48:1–34. Arad, Maya. 2005. Roots and Patterns: Hebrew Morpho-syntax. Dordrecht: Springer. Beard, Robert. 1995. Lexeme Morpheme Base Morphology. Albany, NY: SUNY Press. Belletti, Adriana, ed. 2004. Structures and Beyond. The Cartography of Syntactic Structures. Oxford: OUP. Bloch-Rozmej, Anna. 2008. Melody in Government Phonology. Lublin: Wydawnictwo KUL. Browman, Catherine P., and Louis M. Goldstein. 1986. “Towards an Articulatory Phonology.” Phonology Yearbook 3:219–52. Browman, Catherine P., and Louis M. Goldstein. 1989. “Articulatory Gestures as Phonological Units.” Phonology 6:201–51. Cinque, Guglielmo. 1999. Adverbs and Functional Heads. Oxford: OUP. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. —. 2000. “Minimalist Inquiries: The Framework.” In Step by Step, edited by Roger Martin, David Michaels, and Juan Uriagereka, 89–155. Cambridge, MA: MIT Press. —. 2001. “Derivation by Phase.” In Ken Hale. A Life in Language, edited by Michael Kenstowicz, 1–52. Cambridge, MA: MIT Press. —. 2005. “Three Factors in Language Design.” Linguistic Inquiry 36:1– 22. —. 2008. “On Phases.” In Foundational Issues in Linguistic Theory. Essays in Honor of Jean-Roger Vergnaud, edited by Robert Freidin, Carolos P. Otero, and Maria Luisa Zubizarreta, 134–66. Cambridge, MA: MIT Press. Croft, William, and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge: CUP. Cyran, Eugeniusz. 2010. Complexity Scales and Licensing in Phonology. Berlin: Mouton de Gruyter. Doron, Edit, and Marie Labelle. 2010. “An Ergative Analysis of French Valency Alternations.” In Romance Linguistics 2010: Selected Papers from the 40th Linguistic Symposium on Romance Languages (LSRL), Seattle, Washington, March 2010, edited by Julia Herschensohn, 137– 54. Amsterdam: John Benjamins.

xviii

Foreword

Duanmu, San. 2009. Syllable Structure: The Limits of Variation. Oxford: OUP. Embick, David. 2009. “Roots, States, and Stative Passives.” Abstract for Root Workshop, University of Stuttgart, June 2009. Embick, David, and Rolf Noyer. 2007. “Distributed Morphology and the Syntax/Morphology Interface.” In The Oxford Handbook of Linguistic Interfaces, edited by Gillian Ramchand and Charles Reiss, 289–324. Oxford: OUP. Fauconnier, Gilles, and Mark Turner. 2006. “Mental Spaces. Conceptual Integration Networks.” In Cognitive Linguistics. Basic Readings, edited by Dirk Geeraerts, 303–71. Berlin: Mouton de Gruyter. Giusti, Giuliana. 2005. “At the Left Periphery of the Romanian Noun Phrase.” In On Space and Time in Language, edited by Martine Coene and Liliane Tasmowski, 23–49. Cluj: Clusium. Gussmann, Edmund. 2007. The Phonology of Polish. Oxford: OUP. Haïk, Isabelle. 2013. “Symmetric Structures.” Corela 11:1. Accessed September 08, 2013. http://corela.edel.univ-poitiers.fr/index.php?id=28 75. Halle, Morris. 1997. “Distributed Morphology: Impoverishment and Fission.” MIT Working Papers in Linguistics 30:425–49. Halle, Morris, and Alec Marantz. 1993. “Distributed Morphology and the Pieces of Inflection.” In The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, edited by Kenneth Locke Hale, Samuel Jay Keyser, and Sylvain Bromberger, 111–76. Cambridge, MA: MIT Press. Halle, Morris, and Alec Marantz. 1994. “Some Key Features of Distributed Morphology.” MIT Working Papers in Linguistics 21:275– 88. Harley, Heidi, and Rolf Noyer. 1999. “State-of-the-Article: Distributed Morphology.” Glot International 4(4):3–9. Harley, Heidi, and Rolf Noyer. 2000. “Formal Versus Encyclopedic Properties of Vocabulary: Evidence from Nominalisations.” In The Lexicon-Encyclopedia Interface, edited by Bert Peeters, 349–74. Amsterdam: Elsevier Press. Harris, John. 1994. English Sound Structure. Oxford: Blackwell. Ihsane, Tabea, and Genoveva Puskás. 2001. “Specific is not Definite.” GG@G (Generative Grammar in Geneva) 2:39–54. Kaye, Jonathan, Jean Lowenstamm, and Jean-Roger Vergnaud. 1985. “The Internal Structure of Phonological Elements: A Theory of Charm and Government.” Phonology Yearbook 2:305–28.

Within Language, Beyond Theories (Volume I)

xix

—. 1990. “Constituent Structure and Government in Phonology.” Phonology 7:193–231. Kövecses, Zoltan. 2010. Metaphor. A Practical Introduction. Oxford: OUP. Lakoff, George, and Mark Johnson. 1980/2003. Metaphors we Live by. Chicago: University of Chicago Press. Langacker, Ronald. 1991. Concept, Image and Symbol. The Cognitive Basis of Grammar. Berlin and New York: Mouton de Gruyter. —. 2000. Grammar and Conceptualization. Berlin: Mouton de Gruyter. —. 2009. Investigations in Cognitive Grammar. Berlin and New York: Mouton de Gruyter. Länzlinger, Christopher. 2005. “French Adjective Ordering: Perspectives on DP-internal Movement Types.” Lingua 115:645–89. —. 2010. “The CP/DP Parallelism Revisited.” Generative Grammar in Geneva GG@G 6:49–107. Lowenstamm, Jean. 1996. “CV as the only Syllable Type.” In Current Trends in Phonology. Models and Methods, edited by Jacques Durand and Bernard Laks, 419–41. Salford: ESRI. Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge, MA: MIT Press. —. 1995. “A Late Note on Late Insertion.” In Explorations in Generative Grammar: A Festschrift for Dong-WheeYang, edited by Young-Suk Kim, Byung-Choon Lee, Kyoung-Jae Lee, Hyun-Kwon Yang, and Jong-Yurl Yoon, 396–413. Seoul: Hanakuk. —. 1998. “‘Cat’ as a Phrasal Idiom: Consequences of Late Insertion in Distributed Morphology.” Ms., MIT, Cambridge, MA. Ohala, John J. 1990. “There is no Interface between Phonology and Phonetics: A Personal View.” Journal of Phonetics 18:153–71. Pfau, Roland. 2009. Grammar As Processor: A Distributed Morphology Account of Spontaneous Speech Errors. Amsterdam: John Benjamins. Pöchtrager, Markus A. 2006. “The Structure of Length.” PhD diss., University of Vienna. Pylkkänen, Liina. 2008. Introducing Arguments. Linguistic Inquiry Monograph 48. Cambridge, MA: The MIT Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In Elements of Grammar, edited by Liliane Haegeman, 281–337. Dordrecht: Kluwer.

xx

Foreword

—. 2004. The Structure of CP and IP. The Cartography of Syntactic Structures. Oxford: OUP. Scheer, Tobias. 2014. “Spell-out, Post-phonological.” In Crossing Phonetics-Phonology Lines, edited by Eugeniusz Cyran and Jolanta Szpyra-Kozáowska, 255–75. Newcastle-upon-Tyne: Cambridge Scholars Publishing. —. 2004. A Lateral Theory of Phonology, Vol. 1: What is CVCV and Why Should it Be? Berlin: Mouton de Gruyter. Travis, Lisa. 2001. “The syntax of reduplication.” Proceedings of NELS 31:455–69. —. 2003. “Reduplication feeding syntactic movement.” In Proceedings of the 2003 Annual Conference of the Canadian Linguistic Association, edited by Sophie Burelle and Stanca Somesfalean, 236–47. Université du Québec à Montréal, Département de linguistique et de didactique des langues.

CONTRIBUTORS

Abdullah S. Alghamdi Albaha University, Saudi Arabia [email protected] Aleksandra Bartczak-MeszyĔska University of Finance and Management in Biaáystok, Poland [email protected] Štefan BeĖuš Constantine the Philosopher University, Nitra, Slovakia [email protected] Anna Bloch-Rozmej John Paul II Catholic University of Lublin, Poland [email protected] Maria Bloch-Trojnar John Paul II Catholic University of Lublin, Poland [email protected] Durdica Zeljka Caruso University of Stuttgart, Germany [email protected] BoĪena Cetnarowska University of Silesia, Katowice, Poland [email protected] Magdalena CharzyĔska-Wójcik John Paul II Catholic University of Lublin, Poland [email protected] Tomasz Czerniak John Paul II Catholic University of Lublin, Poland [email protected]

xxii

Contributors

Karolina Drabikowska John Paul II Catholic University of Lublin, Poland [email protected] Adam Gáaz Maria Curie-Skáodowska University, Lublin, Poland [email protected] Renáta Gregová P. J. Šafárik University in Kosice, Slovakia [email protected] Joanna JabáoĔska-Hood Maria Curie-Skáodowska University, Lublin, Poland [email protected] Alla Kalyta National Technical University of Ukraine “Kyiv Polytechnic Institute” [email protected] Man-Ki Theodora Lee University of York, United Kingdom [email protected] Anna Malicka-Kleparska John Paul II Catholic University of Lublin, Poland [email protected] Ewelina Mokrosz John Paul II Catholic University of Lublin, Poland [email protected] Wiktor Pskit University of àódĨ, Poland / University of Economy in Bydgoszcz, Poland [email protected] Marietta Rusinek John Paul II Catholic University of Lublin, Poland [email protected]

Within Language, Beyond Theories (Volume I)

Larysa Taranenko National Technical University of Ukraine “Kyiv Polytechnic Institute” [email protected] Alina Tigău University of Bucharest, Romania [email protected] Mateusz Urban Jagiellonian University of Kraków, Poland [email protected] Wojciech Wachowski Kazimierz Wielki University in Bydgoszcz, Poland [email protected] Maágorzata Wietrzyk John Paul II Catholic University of Lublin, Poland [email protected] Sáawomir Zdziebko John Paul II Catholic University of Lublin, Poland [email protected] Angelina ĩyĞko Maria Curie-Skáodowska University, Lublin, Poland [email protected] Konrad ĩyĞko Maria Curie-Skáodowska University, Lublin, Poland [email protected]

xxiii

PART I: STUDIES IN SYNTAX AND MORPHOLOGY

CHAPTER ONE THE NOMINAL LEFT PERIPHERY IN SLAVIC: EVIDENCE FROM CROATIAN DURDICA ZELJKA CARUSO

1. Introduction It is generally assumed that in a communicative situation a speaker structures his utterances in such a way as to achieve an optimal exchange of information. Information Structure (IS) denotes the formal organization of linguistic expressions in relation to their discourse functions (cf. Halliday 1967; Chafe 1976; Lambrecht 1994). According to Aboh et al. (2010), information structure reflects the speaker’s hypotheses about the hearer’s state of mind (i.e., his assumptions, beliefs and knowledge) at the time of the utterance. (Aboh et al. 2010, 783)

Since its introduction in the late sixties, the term information structure has been widely used to refer to the partitioning of sentences into categories such as focus, topic or comment. However, no consensus has yet been reached as to which and how many categories of information structure can be distinguished and identified (cf. Büring 2005). Within the DP,1 the speaker’s hypotheses about the hearer’s familiarity with a particular referent are reflected in his choice of the determiner that

1

The abbreviations used in this chapter are the following: ACC – accusative, C(P) – Complementizer (Phrase), D(P) – Determiner (Phrase), DAT – dative, DEF – definite, Def(P) – Definiteness (Phrase), Dem(P) – Demonstrative (Phrase), F – feminine, FP – Functional Phrase, Foc(P) – Focus (Phrase), GEN – genitive, INDEF – indefinite, MASC – masculine, MED – medial, NP – Noun Phrase, NOM – nominative, Num(P) – Number (Phrase), PL – plural, POSS – possessive, PP –

Chapter One

4

marks the noun as either identifiable or non-identifiable for the addressee, as shown in the nominal expression a/the linguist from China below (Aboh et al. 2010, 783): (1)

a. John invited [a linguist from China] b. John invited [the linguist from China]

(indefinite; non-identifiable) (definite; identifiable)

Unlike English, which uses (in)definite articles to express this distinction, Turkish, for instance, marks the distinction between identifiable vs. nonidentifiable object noun phrases with the help of case marking (ibid.): (2)

a. Ahmet öküz- ü aldi. Ahmet ox-ACC bought ‘Ahmet bought the ox.’

(accusative case, identifiable)

b. Ahmet öküz aldi. Ahmet ox bought ‘Ahmet bought an ox.’

(no accusative marking, non-identifiable)

In a similar fashion, the difference between an identifiable and nonidentifiable object noun phrase in Croatian can be expressed via an accusative/genitive case marking (Pranjkoviü 2000, 345): (3)

a. dodati kruh (identifiable) to fetch bread-ACC.SG.MASC b. dodati kruha (non-identifiable) to fetch bread-GEN.SG.MASC

Adopting the view that the notions of non-familiarity (new information) vs. familiarity (known information) are associated with the categories of focus and topic respectively (cf. Rizzi 1997), the question arises as to whether the nominal domain contains these categories as well and how they are realized. According to Isac and Kirk (2008, 142), there are two types of evidence for the existence of topic and focus projections within the DP. The first type of evidence comes from the NP-internal morphology, e.g., specificity markers in Gungbe (Aboh 2004). The second type of evidence is based on DP-internal displacement phenomena used to express topic and focus, e.g., word order alterations often entail different Prepositional Phrase, SG – singular, TP – Tense Phrase, Top(P) – Topic Phrase, V(P) – Verb (Phrase), Voc(P) – Vocative Phrase.

The Nominal Left Periphery in Slavic: Evidence from Croatian

5

interpretations, such as focalization or emphasis on a particular nominal constituent. Since discourse-related properties like topic and focus are encoded in the clausal left periphery (Rizzi 1997), the most prominent discourserelated notions associated with noun phrases, namely (in)definiteness and specificity, are assumed to be realized within the nominal left periphery. Accordingly, the nominal left periphery is decomposed into various functional categories including projections related to focalization, topicalization, informational prominence and quantification (cf. Ihsane and Puskás 2001; Aboh 2004; Länzlinger 2005, 2010; Giusti 2005, among others). The various proposals regarding its structure have been summarized under the term ‘the Split DP-Hypothesis.’ As far as the nominal left periphery in Slavic languages is concerned, to my knowledge, no such subdivision has been undertaken so far.2 Noun phrases in Slavic are still considered to be either NPs (Boškoviü 2005, 2009, 2011; Zlatiü 1998) or DPs (Progovac 1998; Leko 1999; Pereltsvaig 2007). Both views are still a matter of controversy. In spite of the obvious lack of the overt discourse-related markers of (in)definiteness and specificity, such as (in)definite articles, and the NP-internal morphology involved in the expression of topic and focus, I argue for a split DP-analysis of nominal expressions in Croatian because the decomposition of the nominal left periphery, along with the overall nominal structure captured by the split DP-proposal, allows me to explain certain DP constructions and syntactic patterns within Croatian noun phrases (e.g., constructions including the numeral JEDAN ‘one,’ deictically used demonstrative pronouns and vocative constructions). The chapter is structured as follows. Section 2 introduces the categories of topic and focus, both on the clausal level and within the DP. Section 3 gives an overview of the different possibilities of (in)definiteness and specificity marking in Croatian and introduces nominal structures, whose syntactic analysis is provided in section 4. Section 5 offers a brief conclusion.

2

Progovac (1998) and Leko (1999) provide a DP-analysis of Serbo-Croatian nouns, including the projection DefP in their structure, but their definition of a split DP differs from the standard cross-linguistic approaches to the layered nominal left periphery.

Chapter One

6

2. The categories of Topic and Focus (Split CP vs. Split DP) The cartographic approach to syntactic structures (e.g., Rizzi 1997, 2004; Cinque 1999; Belleti 2004, among others), as Ihsane (2010) puts it, investigates the make-up of functional categories in the clause, and by extension in nominals. Essentially, it consists in identifying distinct positions in the structure dedicated to different interpretations. (Ihsane 2010, 8)

Following the idea that inflectional morphemes head their own functional projections (Chomsky 1986), clauses are viewed as being “articulated and formed of a succession of lexical and functional projections” (Ihsane 2010, 8). The cartographic approach has been successfully applied to the clausal inflectional domain (Pollock 1989) as well as to its left periphery (Rizzi 1997), and has recently been proposed for the nominal domain as well (cf. Ihsane and Puskás 2001; Aboh 2003, 2004; Länzlinger 2005; Giusti 2005). According to the cartographic approach to syntactic structures, the noun phrase displays a parallel structure to the one of the clause: both can be decomposed into three domains. The NP/VP-shells represent the thematic domain of a verb or a noun, that is, a domain where their external and internal arguments are merged (Larson 1988; Chomsky 1995; Grimshaw 1990). The inflectional domain is made up of functional projections that host modifiers of each lexical category, such as adverbs within the clause or adjectives within the noun phrase. Agreement, ij-features and case are also checked in this domain. Finally, the left periphery is associated with the notions of topic and focus in the clausal domain (Rizzi 1997) and, within the nominal domain, with features related to the D head, such as (in)definiteness, specificity or referentiality (Aboh 2004). The established parallelism3 between the two and the corresponding subdivision into three domains is illustrated in (4) below (Ihsane 2010, 17): (4)

a. [DP… [DP b. [CP… [CP left periphery

[FP adj1… [FP adj2… [FP adv1…[FP adv2… inflectional domain

[nP…[NP..]]]]]] [vP…[VP..]]]]]] NP/VP-shells

3 Länzlinger (2005) labels these three domains Vorfeld (left periphery), Mittelfeld (inflectional domain) and Nachfeld (thematic domain).

The Nominal Left Periphery in Slavic: Evidence from Croatian

7

Since my analysis focuses on the nominal left periphery, in what follows, I will briefly outline the Split CP-Hypothesis first before introducing the Split DP-Hypothesis.

2.1. Topic and Focus on the clausal level (Split CP-Hypothesis) The empirical motivation for advancing the idea that there must be more than one type of CP projection above TP comes from examples such as those in (5a), where the preposed constituent, here no other colleague, occurs after a complementizer like that (Radford 2004, 328): (5)

a. I am absolutely convinced [that no other colleague would he turn to]. b. I am absolutely convinced [ForceP [Force that [FocP no other colleague [Foc would [TP he would turn to no other colleague]]]]].

The embedded CP in (5a) contains a focused constituent and a complementizer, both of which, along with the moved finite verb would, occur in the CP-domain. In order for all these elements to be hosted under CP, Rizzi (1997) suggested that the CP should be split into a number of different projections: ForceP, TopicP, FocusP and FinitenessP. His analysis has been widely referred to as a Split CP-Hypothesis. Since complementizers specify whether a given clause is declarative, interrogative or exclamative in force, he suggested that complementizers are Force markers that head their own maximal projection, ForceP. The preposed focused constituents are contained within a separate Focus projection, FocP. In the above example, the focused constituent no other colleague, which originates as a complement of the preposition to, moves from its complement position within the PP into the specifier position of the FocP. The head of FocP is targeted by the auxiliary would, which moves there from T0. The rearrangement of the constituents results in the clausal structure given in (5b). A focused constituent typically represents a new piece of information within the discourse that the hearer is assumed not to be familiar with. In contrast to this type of constituent, there is another class of preposed expressions that represent information which has already been introduced into the discourse, hence old information, referred to as a topic. Let us consider the following example, adopted here from Radford (2004, 329): (6)

A: The demonstrators have been looting shops and setting fire to cars. B: That kind of behaviour, we cannot tolerate in a civilized society.

8

Chapter One

Being the complement of the verb tolerate, the nominal phrase that kind of behaviour does not surface in its base-generated position, a canonical complement position to the right of the verb tolerate, but rather undergoes movement (topicalization) to the specifier position in the left periphery of the clause. This type of movement, generally regarded to be an instance of A-bar movement, serves to mark the preposed constituent as the topic of the sentence (cf. Radford 2004). According to Rizzi (1997) and Haegeman (2000), topicalized constituents occupy the specifier position within a Topic Phrase, which appears below the ForceP and above FocP, bringing about the following partitioning of the CP-domain: [ForceP [Force [TopP [Top [FocP [Foc [TP…]]]]]]]. In addition to these three different types of projection, there is a fourth functional projection placed directly above TP, which Rizzi (1997) labels as FinP (Finiteness Phrase). The function of Fin0 is to mark the sentence as either finite or non-finite. In Italian, this position is occupied by the infinitival prepositional particle di ‘of,’ which introduces constructions involving infinitival control clauses (Radford 2004, 333): (7)

a. Gianni pensa, il tuo libro, di PRO conoscerlo bene. Gianni thinks, the your book, of PRO know-it well ‘Gianni thinks that your book, he knows well.’ b. Gianni pensa, [ForceP [Force Ø [TopP il tuo libro [Top Ø [FinP [Fin di [TP PRO conoscerlo bene]]]]]]].

As mentioned above, the decomposition of the clausal left periphery has led to a close inspection of the left periphery within the nominal domain. This, in turn, has generated various proposals concerning the decomposition of the nominal determination area. They are captured under the term ‘Split DP-Hypothesis’ and are briefly introduced in the next subsection.

2.2. Topic and Focus inside the DP (Split DP-Hypothesis) Following the view that the clausal left periphery encodes discourserelated properties, such as topic, focus or illocutionary force, which have been captured by Rizzi’s (1997) Split CP-Hypothesis, the question arises as to whether and to what extent the left periphery of the nominal domain encodes discourse-related notions as well. Among the most prominent discourse-related notions related to noun phrases are the notions of specificity and definiteness, both of which play an important role at the

The Nominal Left Periphery in Slavic: Evidence from Croatian

9

grammar-discourse interface. As already mentioned, the speaker’s hypotheses about the hearer’s familiarity with a particular referent are reflected in his choice of determiner (Aboh et al. 2010, 783). The appropriate examples are given below: (8)

a. John invited [a linguist from China] b. John invited [the linguist from China]

(indefinite; non-identifiable) (definite; identifiable)

(9)

There is this man who lives upstairs from me who is driving me mad because he jumps rope at 2 a.m. every night. (indefinite; specific)

(10)

a. A hünj hee tuswark. (+identifiable/–familiar) The dog (i.e., the speaker’s dog) has toothache ‘The dog has a toothache.’ b. Peetje hee jister an kü slachtet. Jo saai, det kü Peetje has yesterday a cow slaughtered. One says the cow wiar äi sünj. (+identifiable/+familiar) was not healthy. ‘Peetje has slaughtered a cow yesterday. One says that the cow was not healthy.’

The above examples illustrate that different languages use different means to express the distinction between identifiability and nonidentifiability: the situation in English is straightforward in that it makes use of indefinite and definite articles, as shown in (8a, b). The example given in (9), taken from Ionin (2006, 180), illustrates the use of the socalled ‘this-indefinite’ (referential indefinite this) in English, where the demonstrative this is not used deictically, but rather as a specificity marker. The referential indefinite this (thisREF) is an indefinite determiner that contributes to the specificity of indefinite classified DPs. Such DPs denote, as Ionin (ibid.) puts it, “a particular referent … about which further information may be given.” Fering, a variant of Frisian, employs two types of definite articles, a in (10a) and det in (10b), to distinguish between a familiar (D-linked) and a non-familiar (but identifiable) referent. The definite article a encodes the features [+identifiable/–familiar], whereas the definite article det/di is assigned the values [+identifiable/+familiar] (cf. Aboh et al. 2010, 783). The above examples show that the discourse-related notions of specificity and definiteness “have syntactic manifestations that need to be accounted for” (Aboh et al. 2010, 784). Moreover, these examples show that specificity and definiteness are features common to DPs across

Chapter One

10

languages. The only difference concerns their grammaticalization. Some languages predominantly employ articles to express these two interrelated notions, while others make use of alternative grammatical and lexical devices. In order to account for the subtle differences between nominal expressions with respect to their specificity and/or definiteness features, along with other IS-related notions such as topic or focus, for instance, a more fine-grained syntactic structure for the nominal left edge has been proposed (cf. Ihsane and Puskás 2001; Aboh 2004; Haegeman 2004; Länzlinger 2005, 2010; Giusti 2005, among others). According to Giusti (2005, 25), the fact that more than one functional element, or, in her words, “multiple occurrences of determiners,” can appear within the DP can be accounted for by assuming a split DP. In addition, the left periphery of the noun phrase “serves the function of complying with information structure” and is there in order to “host A-bar movement(s)” (Giusti 2005, 23). She assumes that unexpected, marked word orders within the NP in Romance can be explained if adjectival modifiers or genitival arguments move to left peripheral positions. The movements of these elements to the left periphery are triggered by the interpretive features [topic] and [focus]. Accordingly, she proposes that the determination area of a DP mirrors the clausal left periphery, consistently including the functional projections given in (11): (11)

DP > TopP* > FocP > TopP* > dP

Ihsane and Puskás (2001), on the other hand, place special emphasis on the discourse related notions of definiteness and specificity and their DPinternal syntactic realization. Like Giusti (2005) and many others, they also propose a structure of the nominal left periphery akin to Rizzi’s (1997) split CP: (12)

DP > TopP > FocP > DefP

However, in their view, the notions of definiteness and specificity are responsible for the DP-internal movement of various constituents within the noun phrase. Since definiteness and specificity clearly cannot be collapsed into one property, they project separately, that is, both definiteness and specificity are syntactically realized in a different functional projection. The feature [+specific] appears on Top0, whereas Def0 hosts the feature [+definite]. A closer look at their proposal is instructive. Definite Phrase (DefP). The lowest projection within the nominal left periphery, which corresponds to Rizzi’s (1997) FinP, is labelled Definite

The Nominal Left Periphery in Slavic: Evidence from Croatian

11

Phrase (DefP). As already argued by Giusti (2005), the head of DefP, Def0, is occupied by articles, which in some languages trigger different inflections on adjectival modifiers. Ihsane and Puskás (2001, 41) argue that “the choice of the article reflects certain properties of the nominal system.” This view is endorsed by two separate arguments. Firstly, determiners select the nominal domain. This means that certain types of nouns are selected by certain determiners, e.g., mass nouns in English can only be selected by a zero indefinite article (Ihsane and Puskás 2001, 41): (13)

John bought (*a) rice.

Secondly, in the same way that Fin0 anchors the event in time and determines the truth conditions of the proposition containing the predicate …, definiteness relates to nominals in the sense that it determines the presupposition of existence of the entity represented by the nominal. (ibid.)

The presupposition of existence is a property of definiteness. Definiteness is therefore considered to be an “existence-anchor.” The morpho-syntactic realization of definiteness within the nominal system differs from language to language. In languages like Swedish, definiteness markers can appear twice, see (14a), while in others, e.g., Romanian in (14b), the feature [+definite] sometimes does not have to be phonologically realized. Nevertheless, the nouns given in (14a, b) below are both definite (Ihsane and Puskás 2001, 42):4 (14)

a. det store huset the big house-the b. Merg la profesor(*ul). I’m going to professor-(*the)

The head of DefP, Def0, is characterized by the feature [+/–definite]. According to Ihsane and Puskás (2001, 42), Def0 hosts articles “as a possible realization of the feature [+/–definite].” Focus Phrase (FocP). Due to the fact that certain elements within the noun phrase can be emphasized (e.g., numerals and possessives), Ihsane and Puskás (2001) argue for a FocP, which is dominated by both the 4

The examples in (14), adopted from Ihsane and Puskás (2001, 42), originate from Granfeldt (1999) and Giusti (1997), respectively.

Chapter One

12

nominal TopP and the highest DP projection, the DP. The fact that FocP is lower than TopP is evidenced by the following Hungarian examples, where the emphasized lexical item, the numeral egy ‘a/one’ below, does not necessarily have a specific reading. This implies that the projection hosting [+focus] feature occurs below the projection where the [+specific] feature is checked (Ihsane and Puskás 2001, 48): (15)

a. egy könyv ‘a book/one book’

b. EGY könyv ‘(exactly) one book’

Topic Phrase (TopP). TopP hosts information that has been preestablished in the discourse. Nominal elements marked as [+specific] are entities that have been pre-established in the discourse. The projection TopP licenses some definite articles and demonstratives. Determiner Phrase (DP). The existence of DP is backed up by the following Hungarian data: (16)

A lánynak ez a könyve the girl-DAT this the book-POSS ‘this book of the girl’s’

The dative possessor5 a lánynak ‘the girl’ appears before the demonstrative ez ‘this,’ which occupies the head position of TopP. Ihsane and Puskás (2001, 50) assume that the dative possessor does not occur in the [Spec,TopP] because it is not associated with a specificity feature. Rather, it sits in the specifier position of the maximal projection placed over TopP, the [Spec,DP]. The idea that one additional projection dominates TopP is further supported by constructions in which the dative possessor can be extracted out of the DP to a higher position in the clause (ibid.): (17)

a. A ánynak vettem meg a könyvét. the girl-DAT bought-I part the book-POSS ‘I bought the girl’s book.’ b. * A lány vettem meg a könyvét. the girl-NOM bought-I part the book-POSS

5

Hungarian distinguishes between two different possessive constructions: a nominative and a dative one. For the details of Ihsane and Puskás’ (2001) analysis, the reader is referred to their paper.

The Nominal Left Periphery in Slavic: Evidence from Croatian

13

The fact that only the dative possessor in (17a) can be moved out of the DP, and not the nominative one in (17b), allows us to conclude that a position occupied by it must be an escape hatch position within the DP, which allows the possessor to move out.

3. Definiteness and specificity in Croatian One of the main arguments against the DP-analysis of nouns in Croatian, hence the non-availability of the nominal left periphery, relies on the fact that noun phrases appear without (in)definite articles (Zlatiü 1998; Trenkiü 2004; Boškoviü 2005, 2009, 2011). However, in spite of the lack of articles, Croatian noun phrases display a definite/indefinite distinction. This is particularly evident in contexts where the numeral jedan ‘one’ cooccurs with the demonstrative taj ‘that,’ where the former functions as the indefinite and the latter as the definite article (Pranjkoviü 2000, 347): (18)

Poslije polusatne ugodne šetnje došli su do jednoga trga after half-hour pleasant walk arrived are to one square i na tom se trgu zadržali do kasnih veþernjih sati. and at that are square stayed till late evening hours. ‘After having pleasantly walked for half an hour they arrived at a square where they stayed until late in the evening.’

In addition to the use of demonstrative pronouns and the numeral jedan ‘one,’ which are used to express the opposition definite/indefinite, Croatian employs other lexical items (e.g., possessive forms, indefinite pronouns, such as neki ‘some,’ nekakav ‘some,’ or indefinite quantifiers mnogo ‘many,’ malo ‘little/few,’ etc.) and morphosyntactic (in)definiteness markers (e.g., adjectival inflection, case or number) for the same purpose (cf. Siliü 2000; Pranjkoviü 2000). Moreover, other means can be used to express (in)definiteness, such as prosody (stress), word order both within a DP and on a clausal level, verbal aspect, and both restrictive and non-restrictive relative clauses (cf. Pranjkoviü 2000). To discuss all of them is beyond the scope of this chapter. Hence, I will discuss constructions which, in my view, involve movement to the TopP and FocP within the nominal left periphery. They include nominal expressions that contain demonstrative pronouns taj ‘this’/ovaj ‘this’ and onaj ‘that,’ the numeral JEDAN ‘one’ and jedan ‘one’ as an indefiniteness marker, along with vocative constructions. They are briefly introduced below.

14

Chapter One

3.1. Demonstrative pronouns As shown in (18), demonstrative determiners are sometimes used instead of definite articles. However, they are claimed to be optional elements within a noun phrase, because their omission does not lead to ungrammaticality (Zlatiü 1998, 3): (19)

(Ovaj) student voli Mariju. this student loves Mary ‘This/the student loves Mary.’

Nevertheless, in certain contexts, see (20), they appear to be obligatory. In the sentence given in (20), we have a discourse deictically used demonstrative pronoun onaj ‘that.’ Discourse deictically used demonstratives do not refer to any entity or a location, but rather “point to the meaning content of an immediately adjacent discourse segment” (Cleary-Kemp 2007, 335). Therefore, a demonstrative pronoun can either refer to the preceding or to the following discourse (anaphoric vs. cataphoric use). In (20), taken from Coelho (2006, 25), the former is the case. (20)

* (Onaj) razgovor sa sveüenikom, dok je još bio dijete, that conversation with priest, while is still been child, pretvorio se u sjeüanje. turned itself into memory ‘The conversation with the priest, when he was still a child, came to be a mere memory.’

The NP razgovor ‘conversation’ in (20) is accompanied by the distal demonstrative onaj ‘that,’ which establishes reference to a conversation that took place at some point in the past, here the childhood of the person being spoken of. The adverbial clause dok je još bio dijete ‘when he was still a child’ creates a particular referential context in this sentence, which makes the demonstrative determiner obligatory. The English translation equivalent shows that the noun conversation necessarily appears with the definite article the. Its omission would violate the grammaticality of the sentence.

The Nominal Left Periphery in Slavic: Evidence from Croatian

15

3.2. The numeral JEDAN vs. jedan as an indefiniteness marker Based on the assumption that the semantic interpretation results from the underlying syntactic structure, the following two expressions, which have different interpretations, must have different syntactic structures: (21)

a. JEDAN þovjek one man

b. jedan þovjek one-INDEF man

While in (21a) jedan ‘one’ refers to the number (one man as opposed to two men), in (21b), jedan ‘one’ is used to indicate the speaker’s assumption about the hearer’s familiarity with the denoted referent. Hence, in (21b), jedan ‘one’ is used to indicate the indefiniteness of the noun þovjek ‘man.’ The assumption that jedan ‘one’ indeed functions as an indefiniteness marker in Croatian is further endorsed by examples such as (22), where it appears together with nouns marked as plural: (22)

Na redu su još jedni savjeti za vašu kosu. following are still one-NOM.PL advices-NOM.PL for your hair ‘There still follows some advice concerning your hair.’ [http://videoteka.novatv.hr/multimedia/oblak-film-o-luki-ritzu.html]

At this point, the question arises as to how we can distinguish between the two. Siliü (2000) observed that the numeral jedan ‘one’ is emphasized, whereas jedan ‘one’ in its function as an indefinite article is not. This is confirmed by the following example, in which the use of jedan ‘one’ without emphasis, as in (21b), is ruled out: (23)

Ivan je kupio dvije knjige, a Marija je kupila JEDNU. John bought two books, and Mary bought one /*one-INDEF

I assume that jedan ‘one’ as a numeral appears in a position specified for the [+focus] feature, while jedan ‘one’ as an indefiniteness marker occupies a position where its [–def] feature is checked.

3.3. Vocative constructions As far as formal accounts of vocative phrases are concerned, vocatives are taken to either lack a D head (Longobardi 1994; Szabolcsi 1994) or are regarded as regular DPs (Moro 2003), which allow for definite articles and adjectival modification (cf. Hill 2007, 2079). Espinal (2011) postulates a separate functional projection called VocP above DP, whose head Voc0 is

16

Chapter One

endowed with a strong interpretable deictic [+DX] feature (cf. also Moro 2003; Stavrou 2009). My purpose, however, is not to give an extensive analysis of vocative constructions here, but simply to offer a possible explanation for the observed impossibility of demonstratives and simple nouns marked for vocative case to co-occur within the Croatian DP, see (25c). Consider the following examples: (24)

a. ova moja lijepa kuüa this my beautiful house b. * kuüa ova moja lijepa * house this my beautiful

In the neutral word order in (24a), both determiners and adjectives appear before the noun. Their postposition is ungrammatical, as evidenced by (24b). However, in vocative constructions, these lexical items display a different behaviour. Whereas adjectives or possessives6 may accompany the vocative noun in (25a, b), the determiner ta in (25c) cannot appear with it: (25)

a. lijepa djevojka / Djevojko lijepa! beautiful-F.SG.NOM girl-F.SG.NOM / girl-F.SG.VOC beautiful-F.SG.VOC ‘Beautiful girl / Beautiful girl!’ b. moja djevojka / Djevojko moja! my-F.SG.NOM girl-F.SG.NOM / girl-F.SG.VOC my-F.SG.VOC ‘My girl / My girl!’ c. ta djevojka / *Djevojko ta! / Djevojko! that-F.SG.NOM girl-F.SG.NOM / *girl-F.SG.VOC that / girl-F.SG.VOC ‘That girl / *Girl that! / Girl!’

Adopting the view that both vocatives and demonstratives are endowed with the features [+specific, +definite] (cf. Espinal 2011; Hill 2007; Ihsane and Puskás 2001) straightforwardly accounts for the ungrammaticality of (25c). If the elements equipped with the [+specific] feature occupy the head position of TopP, as proposed by Ihsane and Puskás (2001), both the 6

To my knowledge, the possible exceptions are restricted to the possessives moj ’my’ and naš ‘our,’ and the numeral jedan ‘one,’ e.g., we can say Dijete moje/naše! ‘child my/our’ or Kukavico jedna! ‘coward one,’ but not: *Dijete tvoje! ‘child your.’

The Nominal Left Periphery in Slavic: Evidence from Croatian

17

demonstrative ta ‘that’ and the noun djevojko ‘girl,’ which is assigned the vocative case, are competing for the same position. This is not the case in (25a, b). Since possessive pronouns are [+definite, non-specific], the cooccurrence of the vocative noun and the possessive pronoun is not banned, as shown in (25b). The appropriate syntactic analysis of this construction is given in 4.3.

4. The syntactic analysis Before we turn to the syntactic analysis of the constructions introduced above, let us have a look at (26), which mirrors the neutral word order of prenominal constituents within Croatian nominal expressions: (26)

sve te moje prve besmislene pogreške all-F.PL these-F.PL my-F.PL first-F.PL preposterous-F.PL mistakes-F.PL ‘all these first preposterous mistakes of mine’

The neutral word order of prenominal constituents in Croatian (26) roughly corresponds to the universal word order proposed by Greenberg (1966) and Hawkins (1983), the only difference being the co-appearance of possessives between demonstratives and numerals in Croatian: (27)

a. UNIVERSAL BASE ORDER Demonstrative > Numeral > Adjective > Noun b. Diese fünf großen Häuser (German) c. These five large houses (English) (Ihsane and Puskás 2001, 45)

Both Greenberg (1966) and Hawkins (1983) assume that all of the abovementioned prenominal elements are generated in the inflectional domain of the noun. Adopting this view and following Giusti (1994), Ihsane and Puskás (2001, 45) take the position that demonstratives are generated in the specifier position of the highest functional projection of the inflectional system, directly below the DP area, to which they move in order to check their [+specific] feature:

Chapter One

18 (28)

DP Spec,DP

D‫މ‬ D

FP Dem

FP

Num

FP Adj

NP

Can DemP be regarded as the highest functional projection within the inflectional domain cross-linguistically? The following examples seem to contradict this assumption, at least in Croatian: (29)

a. [Svi ti dokazani sluþajevi] detaljno all-NOM.PL these-NOM.PL proven-NOM.PL cases-NOM.PL in detail su bili dokumentirani tijekom posljednjih godina. are been documented during last years ‘All the proven cases have been documented in detail during the last years.’ b. [Svih tih all-GEN.PL these-GEN.PL sluþajeva] detaljno cases-GEN.PL in detail

nekoliko tisuüa dokazanih several thousand-GEN.PL proven-GEN.PL je bilo dokumentirano… is been documented

c. [Mojih prvih pet reþenica] odnosile my-GEN.PL first-GEN.PL five sentences-GEN.PL referred su se na... are self to…

The example given in (29a) mirrors the neutral word order within a DP, where the demonstrative appears in front of all the other constituents, being preceded only by the universal quantifier svi ‘all.’ In addition, all the prenominal elements agree with the noun in number, gender and case. However, in spite of the preserved word order in (29b), both the universal quantifier svi ‘all’ and the demonstrative determiner ti ‘that-MED.PL’ are marked for genitive, although nekoliko ‘several’ quantifies tisuüe dokazanih sluþajeva ‘thousands of proven cases.’ These examples show that, since all constituents preceding quantifiers and numerals are marked for genitive, QP must be the highest functional projection within the inflectional domain, followed by DemP, NumP and others. After having

The Nominal Left Periphery in Slavic: Evidence from Croatian

19

been assigned case in their base-generated position within the inflectional domain, some prenominal items subsequently move out of it to the left periphery, where they check their specificity, focus and (in)definiteness features. Following this idea, let us have a look at the syntactic structure of the constructions introduced in sections 3.1–3.3.

4.1. DPs containing demonstrative pronouns Before turning to the analysis of DPs that contain demonstrative pronouns, let us have a brief look at (30) below: (30)

Profesorov-a procjena Vaših sposobnosti professor-POSS.NOM assessment-NOM your abilities-GEN

In the argument-supporting nominalization (ASN) in (30), one of the nominal arguments is realized as a possessive adjective (profesorova ‘professor’s’). Being the argument of the noun procjena ‘assessment,’ the possessive element is c-selected and theta-marked by the noun. This further implies that it must be generated in a position where it receives its theta-role, that is, within the nP-shell. Due to the fact that it also contributes to the definiteness status of the whole nominal expression, it moves to the nominal left periphery, where these features are checked: the possessive suffixes -ov, -ev, -in, etc., are generated in the head position of PossP and move to Def0 (Caruso 2012). Now consider the following examples: (31)

a. * moja ova ruža my this rose * ‘my this rose’

b. * Petrovi oni prijatelji Peter-POSS those friends * ‘Peter’s those friends’ (Zlatiü 1997, 36ff.)

As we can see in (31), the ordering Poss-Det is ungrammatical, which means that demonstrative pronouns occupy a position above PossP. If possessive elements move to DefP, as briefly outlined above, then demonstrative pronouns need to move farther up to TopP (to check their [+spec] feature). Hence, the syntactic analysis of the nominal expression ona tvoja kuüa ‘that house of yours’ in (32) looks as follows:

Chapter One

20 (32)

ona tvoja kuüa that your house ‘that house of yours’

TopP Top‫މ‬ ona

DefP

Spec,DefP Def [+def] tvoja

Def‫މ‬ FPNPagr ….… nP¨ kuüa

4.2. The numeral JEDAN vs. jedan as an indefiniteness marker As mentioned previously, the unstressed version of the numeral jedan ‘one’ is used to indicate the indefiniteness of the nominal referent and functions as a real indefinite article (cf. Siliü 2000). Being an indefiniteness marker, jedan ‘one’ is base-generated in Def0, a position where both definite and indefinite articles in DP-languages occur (33a). In contrast, the numeral JEDAN ‘one,’ which necessarily needs to be emphasized, appears in a position specified for a [+focus] feature, that is, in FocP (see (33b)):

The Nominal Left Periphery in Slavic: Evidence from Croatian

21

DefP

(33) a.

Spec,DefP

Def‫މ‬

Def [–def] jedan

FPNPagr …… nP¨ þovjek

b.

FocP Foc‫މ‬

JEDAN

DefP

Spec,DefP Def [+def]

Def‫މ‬ FPNPagr …… nP¨ þovjek

4.3. Vocative constructions Recall the example given in (25b), which is repeated here for the sake of convenience:

Chapter One

22 (34)

moja djevojka / Djevojko moja! my-F.SG.NOM girl-F.SG.NOM / girl-F.SG.VOC my-F.SG.VOC ‘My girl / My girl!’

As already mentioned, since possessive pronouns are [+definite, nonspecific], the co-occurrence of the vocative noun and the possessive pronoun is not banned. As we have seen, possessive pronouns move to DefP. Since the vocative noun djevojko ‘girl’ precedes it, the noun moves from its base-generated position to Top0, leaving the possessive pronoun moja ‘my’ behind in Def0. Hence, the proposed syntactic structure looks as follows: (35)

TopP Top‫މ‬ djevojkoi

DefP

Spec,DefP Def [+def] moja

Def‫މ‬ FPNPagr …… nP¨ ti

5. Concluding remarks In this chapter, I argue for a split DP-analysis of nominal expressions in Croatian, despite the fact that Croatian lacks (in)definite articles, which are considered to be the most-prominent overt discourse-related markers of (in)definiteness and specificity.

The Nominal Left Periphery in Slavic: Evidence from Croatian

23

Although no NP-internal morphology is involved in the expression of topic and focus (in the sense of Aboh 2004), certain DP constructions that involve emphasized elements (e.g., the numeral JEDAN ‘one’ vs. the indefiniteness marker jedan ‘one’), or syntactic patterns that deviate from the regular DP-internal word order (e.g., vocative constructions), can be best explained if we adopt a layered nominal left periphery that consists of several functional projections, among them FocP and TopP.

References Aboh, Enoch. 2003. The Morphosyntax of Complement-Head Sequences. Oxford: OUP. —. 2004. “Topic and Focus within D.” Linguistics in the Netherlands 21:1–12. Aboh, Enoch, Norbert Corver, Marina Dyakonova, and Marjo van Koppen. 2010. “DP-Internal Information Structure: Some Introductory Remarks.” Lingua 120:782–801. Belletti, Adriana, ed. 2004. Structures and Beyond. The Cartography of Syntactic Structures. Oxford: OUP. Boškoviü, Željko. 2005. “On the Locality of Left Branch Extraction and the Structure of NP.” Studia Linguistica 59:1–45. —. 2009. “Good Monsters are Back, Now We Need to Rename Them.” Paper presented at the Conference “Atoms and Laws of the Noun Phrase” at the University of Utrecht, July 2. —. 2011. “More on the DP/NP Analysis of Languages with and without Articles.” Paper presented at the Conference on Languages with and without articles, Paris (CNRS/Paris 8), March 3–4. Büring, Daniel. 2005. “Semantics, Intonation and Information Structure.” Available at: http://semanticsarchive.net/Archive/GQ0YjgxM/buring.information.str ucture.v2005.pdf. Caruso, Durdica Zeljka. 2012. “The Syntax of Nominal Expressions in Articleless Languages: A Split DP-Analysis of Croatian Nouns.” PhD diss., University of Stuttgart. Chafe, Wallace. 1976. “Giveness, Contrastiveness, Definiteness, Subjects, Topics and Point of View.” In Subject and Topic, edited by Charles N. Li, 25–56. New York: Academic Press. Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press. —. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cinque, Guglielmo. 1999. Adverbs and Functional Heads. Oxford: OUP.

24

Chapter One

Cleary-Kemp, Jessica. 2007. “Universal Uses of Demonstratives: Evidence from Four Malayo-Polynesian Languages.” Oceanic Linguistics 46:325–47. Coelho, Paulo. 2006. Peta Gora. Zagreb, VBZ. Tiskarna Ljubljana d.d., Ljubljana. Espinal, M. Teresa. 2011. “On the Structure of Vocatives.” Available at: http://filcat.uab.cat/clt/publicacions/reports/pdf/GGT-11-04.pdf Giusti, Giuliana. 1994. “Enclitic Articles and Double Definiteness: A Comparative Analysis of Nominal Structure in Romance and Germanic.” The Linguistic Review 11:231–55. —. 1997. “The Categorial Status of Determiners.” In The New Comparative Syntax, edited by Liliane Haegeman, 95–123. Harlow: Addison Wesley Longman. —. 2005. “At the Left Periphery of the Romanian Noun Phrase.” In On Space and Time in Language, edited by Martine Coene and Liliane Tasmowski, 23–49. Cluj: Clusium. Grandfeldt, Jonas. 1999. Sur la Structure du DP dans les Langues Scandinaves. Presentation in the DES seminar, Geneva. Greenberg, Joseph H. 1966. “Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements.” In Universals of Language (2nd ed.), edited by Joseph H. Greenberg, 73–113. Cambridge, MA: MIT Press. Grimshaw, Jane. 1990. Argument Structure. Cambridge, MA: MIT Press. Haegeman, Liliane. 2000. “Inversion, Non-Adjacent Inversion and Adjuncts in CP.” Transactions of the Philological Society 98:121–60. —. 2004. “DP-Periphery and Clausal Periphery: Possessor Doubling in West Flemish.” In Peripheries, edited by David Adger, Cécile de Cat, and George Tsoulas, 211–40. Dordrecht: Kluwer. Halliday, Michael Alexander Kirkwood. 1967. “Notes on Transitivity and Theme in English, Part II.” Journal of Linguistics 3:199–244. Hawkins, John A. 1983. Word Order Universals. New York: Academic Press. Hill, Virginia. 2007. “Vocatives and the Pragmatics-Syntax Interface.” Lingua 117:2077–105. Ihsane, Tabea, and Genoveva Puskás. 2001. “Specific is not Definite.” GG@G (Generative Grammar in Geneva) 2:39–54. —. 2010. The Layered DP: Form and Meaning of French Indefinites. Amsterdam and Philadelphia: John Benjamins. Ionin, Tabea. 2006. “‘This’ is Definitely Specific: Specificity and Definiteness in Article Systems.” Natural Language Semantics 14:175–234.

The Nominal Left Periphery in Slavic: Evidence from Croatian

25

Isac, Daniela, and Allison Kirk. 2008. “The Split DP Hypothesis: Evidence from Ancient Greek.” Rivista di Grammatica Generativa 33:137–55. Available at: http://lear.unive.it/bitstream/10278/ 2506/1/Isac-Kirk.pdf. Länzlinger, Christopher. 2005. “French Adjective Ordering: Perspectives on DP-internal Movement Types.” Lingua 115:645–89. —. 2010. “The CP/DP Parallelism Revisited.” Generative Grammar in Geneva GG@G 6:49–107. Lambrecht, Knud. 1994. Information Structure and Sentence Form. Topic, Focus and the Mental Representations of Discourse Referents. Cambridge: CUP. Larson, Richard K. 1988. “On the Double Object Construction.” Linguistic Inquiry 19:335–92. Leko, Nedzad. 1999. “Functional Categories and the Structure of the DP in Bosnian.” In Topics in South Slavic Syntax and Semantics, edited by Mila Dimitrova-Vulchanova and Lars Hellan, 229–52. Amsterdam: John Benjamins. Longobardi, Giuseppe. 1994. “Reference and Proper Names: A Theory of N-Movement in Syntax and Logical Form.” Linguistic Inquiry 25(4):609–65. Moro, Andrea. 2003. “Notes on Vocative Case: A Case Study in Clause Structure.” In Romance Languages and Linguistic Theory 2001. Selected Papers from Going Romance, edited by Josep Quer, Jan Schroten, Mauro Scorretti, Petra Sleeman, and Els Verheugd, 247–61. Amsterdam: John Benjamins. Pereltsvaig, Asya. 2007. “On the Universality of DP: A View from Russian.” Studia Linguistica 61(1):59–94. Pollock, Jean-Yves. 1989. “Verb Movement, Universal Grammar and the Structure of IP.” Linguistic Inquiry 20:365–424. Pranjkoviü, Ivo. 2000. “Izražavanje Neodreÿenosti/Odreÿenosti Imenica u Hrvatskome Jeziku.” In Rijeþki Filološki Dani, Zbornik Radova 3:343– 50. Progovac, Ljiljana. 1998. “Determiner Phrase in a Language without Determiners.” Journal of Linguistics 34:165–79. Radford, Andrew. 2004. Minimalist Syntax: Exploring the Structure of English. Cambridge: CUP. Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In Elements of Grammar, edited by Liliane Haegeman, 281–337. Dordrecht: Kluwer. —. 2004. The Structure of CP and IP. The Cartography of Syntactic Structures. Oxford: OUP.

26

Chapter One

Siliü, Josip. 2000. “Kategorija Neodreÿenosti/Odreÿenosti i Naþini Njezina Izražavanja.” Rijeþki Filološki Dani, Zbornik Radova 3:401– 06. Stavrou, Melita. 2009. "Vocative!” Ms. Aristotle University of Thessaloniki. Szabolsci, Anna. 1994. “The Noun Phrase.” Syntax and Semantics 27:179–274. Trenkiü, Danijela. 2004. “Definiteness in Serbian/Croatian/Bosnian and Some Implications for the General Structure of the Nominal Phrase.” Lingua 114:1401–27. Zlatiü, Larisa. 1997. “The Structure of the Serbian Noun Phrase.” PhD diss., University of Texas, Austin. —. 1998. "Slavic Noun Phrases are NPs not DPs.” Workshop on Comparative Slavic Morphosyntax, Bloomington, Indiana.

CHAPTER TWO THE CATEGORIAL STATUS AND INTERNAL STRUCTURE OF NPN FORMS IN ENGLISH AND POLISH WIKTOR PSKIT

1. Introduction This chapter is concerned with NPN forms in English (e.g., day after day) and Polish (e.g., dzieĔ po dniu). This type of structure has recently attracted the attention of a number of linguists representing various theoretical frameworks (Travis 2001, 2003; Jackendoff 2008; Haïk 2013).1 The first part of the study deals with an account of the relevant expressions in English descriptive grammars. This is followed by a detailed presentation of the approach to NPN structures developed in Travis (2001, 2003). Another section is devoted to an analysis of Polish NPNs. Given that the English data does not exclude the possibility that NPNs are NPs, and that the Polish data points to the nominal character of such structures, the relevant forms can be categorized as NPs or, following Travis (2001, 2003), QPs, where, broadly speaking, QP is a functional projection within the nominal domain.

2. NPN forms in English An account of English NPN forms is provided in large descriptive grammars such as Quirk et al. (1985) and Huddleston and Pullum (2002). It is worth looking at both analyses since, while sharing some insights, 1

Pskit (2012) is an attempt at a comparison of the different approaches to NPN forms in English in the context of the distinction between words and constructions.

Chapter Two

28

they also differ in a number of details as far as the structure of NPNs is concerned. The discussion of the two descriptive approaches to English NPNs is followed by a presentation of more recent proposals in Travis (2001, 2003).

2.1. NPN forms in Quirk et al. (1985) In Quirk et al. (1985), the relevant forms are discussed in a chapter on nouns and determiners, in a subchapter devoted to the use of articles. NPNs exemplify a type of structure where the article, normally expected with countable singular nouns, is omitted. They are labelled parallel structures “where two nouns are placed together in a parallel structure” (Quirk et al. 1985, 280) as in (1): (1)

a. arm in arm b. mile upon mile c. back to back d. side by side

There is some affinity between the structures in (1) and expressions where “one noun is balanced against another noun of contrasting meaning” (Quirk et al. 1985, 280): (2)

a. from father to son b. from (the) right to (the) left c. from (the) beginning to (the) end

Interestingly, as the data in (2) demonstrate, some of those “contrasting” NPNs can “grow into” PNPNs (cf. (2a, b, c)) and the presence of the determiner is optional in some of them (cf. (2b, c)). As suggested by Quirk et al. (1985, 280), the idiosyncratic nature of such structures may stem from the loss of “independent nominal status” by the relevant nouns. This is observable in the impossibility of adjectival premodification or pluralization of the nouns: (3)

a. * They talked old man to young man. b. * They stood toes to toes.

Such behaviour makes NPNs idioms with “‘frozen’ article use” (Quirk et al. 1985, 280). On the other hand, an argument against the analysis of NPNs as idiosyncratic adverbial idioms is the existence of a whole range

The Categorial Status and Internal Structure of NPN Forms

29

of mostly fixed expressions containing countable singular nouns with a zero article, including prepositional phrases (e.g., at home, on foot, in turn) as well as complex prepositions (e.g., on top of, by way of) and idioms headed by verbs (e.g., take advantage of, set fire to) (Quirk et al. 1985, 280–81). Thus, while Quirk et al. (1985) point out a number of properties that make the relevant forms non-prototypical NPs, the NPNs still belong to the nominal domain.

2.2. NPN forms in Huddleston and Pullum (2002) A different perspective on NPN forms can be found in a more recent descriptive grammar of English. Huddleston and Pullum (2002, 409–10) note the presence of bare NPs (i.e., NPs without a determiner) that are regarded as fixed expressions or frames. These include NPNs with two identical nouns (e.g., day after day) and PNPNs with contrasting (or matched) nouns (e.g., from father to son). However, unlike Quirk et al. (1985), Huddleston and Pullum (2002) discuss the categorial status of NPNs in a chapter concerned with prepositions and PPs, which indicates an important change in the treatment of these structures. Huddleston and Pullum (2002, 632–34) distinguish between NPN forms that have the status of an NP and those that are PPs. English NPNs with the preposition after are NPs since they have a distribution typical of nominals. For instance, they can be subjects, as in (4a), objects, as in (4b) and adjuncts, as in (4c) (Huddleston and Pullum 2002, 634): (4)

a. Day after day was wasted. b. They wasted day after day. c. She worked on it day after day.

It is argued that the distribution of such expressions is determined by the first noun, which functions as a head, with the following PP as its complement. Another fact that corroborates the nominal analysis of these NPNs is the possibility of postmodification by another PP of the rightmost noun (Huddleston and Pullum 2002, 634): (5)

They climbed [flight after flight of stairs].

The category, internal structure and behaviour of day after day can be contrasted with those of spoonful by spoonful or day by day. The distribution of the latter appears to depart from the prototypical

Chapter Two

30

characteristics of NPs because they only occur as adjuncts and cannot function as subjects or objects, as manifested below (Huddleston and Pullum 2002, 632): (6)

a. * I used spoonful by spoonful. b. I drank my milk spoonful by spoonful. c. * One by one exited. d. They exited one by one.

This prompts Huddleston and Pullum (2002) to analyse these NPNs as PPs with an unusual internal structure. There is a prepositional head with two NP complements: the right-hand noun is the internal complement, whereas the left-hand noun is the external complement. Thus, there is a parallel between the internal structure of such PPs and the structure of clauses or NPs, all of them potentially involving internal and external arguments. The noun following the preposition is considered the internal complement because the head and the complement form a ‘lower-level’ or ‘inner’ PP. This yields the following structure (Huddleston and Pullum 2002, 632): (7)

PP Comp: NP

Head: PP Head: Prep

spoonful

by

Comp: NP spoonful

As Huddleston and Pullum (2002, 633) observe, the treatment of a subclass of NPNs as PPs can be extended to lexicalised expressions, such as arm in arm or face down, which function as adjuncts or predicatives. Examples like face down seem to be even more unusual – the inner PP consists of the prepositional head only, the preposition being intransitive. Summing up the two descriptive accounts of the expressions in question, it is important to note that both grammars recognise certain idiosyncratic properties of NPNs. However, while Quirk et al. (1985) treat them in a uniform way as nominals, Huddleston and Pullum (2002) make an important distinction between two subtypes of the relevant forms

The Categorial Status and Internal Structure of NPN Forms

31

characterized by two different types of internal composition. Importantly, the distinction advocated by Huddleston and Pullum (2002) is motivated by distributional considerations.

2.3. Travis (2001, 2003) Travis (2001, 2003) proposes the mechanism of syntactic reduplication for the analysis of NPN forms in English. Although syntactic reduplication seems to be much less common than phonological reduplication, Travis argues that reduplication also occurs in syntax and, when it does, it is subject to syntactic constraints. In fact, all cases of reduplication involve phonological copying that is triggered by syntactic environments. This also refers to instances that seem to be morpho-phonological reduplication, where reduplication is triggered by the syntactic configuration created via head movement (Travis 2003). Travis (2001) reconsiders data presented in Pi (1995) in (8) below. (8)

a. The butterfly fluttered from flower to flower. b. Jon washed plate after plate for hours after the party. c. The careful artist completed the mosaic tile by tile. d. Eric can drink mug upon mug of coffee in a single hour. e. In fairy tale after fairy tale, good triumphs over evil.

Pi (1995) argues that certain prepositions in English like after, upon, and by appear in constituents where the material that appears before the preposition is a copy of (part of) the material in the complement of the preposition. According to Pi, the material preceding the preposition is a linearization of what is generated in the complement position. In a sense, in (8), the material is repeated to encode plurality and the repeated items function like bare plurals, which explains why they occur without determiners. Interestingly, the lack of repetition yields ungrammatical structures, indicating that the repetition is not optional (Travis 2003): (9)

a. * The butterfly fluttered to flower. b. * Jon washed plate.

The mechanism responsible for the construction of NPN forms, such as those in (8), is a form of syntactic reduplication, and these structures and phonological reduplication structures should be seen as two examples of the same process. Evidence for this idea comes from a number of

32

Chapter Two

similarities between phonological reduplication and syntactic reduplication. First of all, in both subtypes of reduplication, differently sized domains can be affected by reduplication. Phonological reduplication exhibits variation in the size of prosodic domains that can be copied, ranging from a core syllable to a minimal word. The same holds for syntactic reduplication, where reduplicated items are of varying size/complexity (data from Pi 1995): (10)

a. cup after cup of coffee b. cup of coffee after cup of coffee c. cup after steaming cup of coffee d. steaming cup after steaming cup of coffee e. steaming cup of coffee after steaming cup of coffee

Further, in both types of reduplication, information must be copied from the rightward host and nothing can be added if it is not included in the host. The following shows that the addition of information beyond what is in the host to the copy results in an ungrammatical expression (Pi 1995): (11)

* steaming cup after cup of coffee

In addition to this, in both types, the reduplication mechanism is sensitive to constituency. In Pi (1995), the structure for steaming cup of coffee is represented by (12a), which renders (12b) ungrammatical because it involves copying a non-constituent. (12)

a. [[steaming cup] of coffee] b. * cup of coffee after steaming cup of coffee

Phonological reduplication and syntactic reduplication also share a number of properties that have to do with function. Phonological reduplication is used in a similar manner across languages. According to Moravcsik (1978), this type of reduplication relates to repeated events. Likewise, syntactic reduplication can designate plural nouns or plural events. An iteration of events is illustrated in (13a, b) (data from Pi 1995, after Travis 2001): (13)

a. Student after student visited the professor on Monday. b. Gertrude watched program after program all afternoon.

The Categorial Status and Internal Structure of NPN Forms

33

In Travis (2001), reduplication is triggered by the need to check a formal feature. Head movement takes place before the copying involved in reduplication occurs, which entails the occurrence of head movement prior to phonological reduplication. By contrast, in syntactic reduplication, feature checking does not force movement to the position of the specifier. In this case, a copy created by the head occupies the specifier position and checks the relevant feature. In other words, syntactic reduplication requires that copying precede checking. Yet, there are differences between the two subtypes of reduplication as well. First of all, phonological and syntactic reduplication differ with respect to the type of domain that is subject to copying, defined in prosodic terms for the former and in syntactic terms for the latter. Travis (2001) attributes this sort of difference to the difference in syntactic configurations. Then in phonological reduplication, since reduplication follows head movement, a subdomain of X0, which is the sister of the reduplicative head, is assumed to be sub-syntactic, and hence invisible in syntax. On the other hand, in the case of syntactic reduplication, as reduplication precedes head movement, giving rise to XP as the sister of the reduplicative head, a subdomain of XP is determined syntactically. Another source of difference concerns the size of the reduplicated item, which is fixed for phonological reduplication, by virtue of being reduced to affixal forms with prosodically defined inherent shapes, and variable for syntactic reduplication, because specifiers – hosts for copied material – are not required to have an inherent shape defined in prosodic terms. Furthermore, whereas phonological reduplication involves left-to-right mapping, syntactic reduplication is based on bottom-up derivation. In the case of syntactic reduplication, the head of the complement is included in the copy irrespective of whether it is the left-most or right-most element in the structure. At the same time, syntax permits copying merely a head, but phonology does not allow just a vowel to be copied. Travis (2001) suggests that this difference stems from different default forms in each domain, with a head being the minimal possible default form in syntax, but with CV being the minimal default in phonology. Travis (2001) also points out that, contrary to phonological reduplication, which allows only a single copy, syntactic reduplication enables more than one copy to appear in the relevant slot, as shown in (14a). This is possibly due to a general tolerance for emphatic iteration in syntactic structures (14b). (14)

a. He drank cup after cup after cup of coffee. b. He drank many many many cups of coffee.

34

Chapter Two

Interestingly, Pi (1995) observes that the preposition upon is exceptional in that it occurs with plural nouns in NPN forms: (15)

Bags upon bags of marshmallows were stolen this week.

On the other hand, this is not possible with after, a preposition which appears to behave in a way that is very similar to upon: (16)

a. Student after student visited the professor on Monday. b. * Students after students visited the professor on Monday.

According to Travis (2001), this difference in behaviour between the two prepositions is based on the variable nature of the reduplicative morpheme. The morpheme can occupy various head positions: if realized by upon it occurs in a head above Number, thus allowing plural nouns, and if realized as after, it appears in the Number head, thus excluding plural nouns. Although it is always a head, the reduplicative morpheme does not have to constitute the final site for the copy. Then, if the morpheme has a head (a minimal projection) as its sister, the copied material may appear in the very position of the morpheme. However, if the morpheme has a phrase (maximal projection) as its sister, then the copied material, probably owing to its structural complexity, has to appear in the specifier of the projection headed by the reduplicative morpheme. From this point of view, Travis (2001) takes after and upon to be reduplicative morphemes since they give rise to a copy of a subdomain of their sisters, but the two prepositions differ from other reduplicative morphemes because they do not constitute the copy themselves. Among major issues that require explanation is the question of the constraint on the domain that is copied. As noticed in (12) above, it seems that syntactic reduplication is sensitive to constituency. However, this generalization is slightly undermined by the data in (17) below, where one can observe that it is easier to copy a domain larger in size than a head noun if the noun functions as a measure phrase (Travis 2001): (17)

a. I met with student after student. b. * I met with student of linguistics after student of linguistics. c. I ate bunch after bunch of bananas. d. I ate bunch of bananas after bunch of bananas.

Although cup of coffee can be copied, as (10b) above demonstrates, the copying of student of linguistics in (17b) is not licensed.

The Categorial Status and Internal Structure of NPN Forms

35

Travis (2001, 2003) claims that all reduplication is actually a syntactic phenomenon. One of the reasons for this conclusion is the presence of basic syntactic relations, viz. specifier-head, head-head and modifier-head as well as the fact that the copy occurs in an independently available syntactic position. This is interpreted as follows: although reduplication is syntactically driven, it may need to resort to phonological terminology whenever the syntactic environment of reduplication involves subsyntactic material. Certain aspects of syntactic reduplication and contrastive reduplication seem to be of significance in a broader theoretical context. For instance, grammatical structures such as (10b) contrasted with ungrammatical ones such as (17b) suggest that syntactic reduplication does not allow reduplication of all the complements of heads, which has consequences for the internal structure of the relevant reduplicated expressions (Travis 2001). As emphasized by Travis (2001), the type of copying observed in syntactic reduplication is essentially different from a similar mechanism found in movement. One difference has to do with the distinction between feature checking by means of head movement and by means of occurrence in the specifier position. Head movement precedes copying, whereas setting up the specifier-head relationship follows copying. Also, copying for reduplication appears to be different from copying for movement. The former permits copying subdomains of the complement XP which independently cannot undergo movement. Travis (2003, 240) proposes the following representation for syntactic (iterative) reduplication: (18)

Iterative (syntactic) reduplication QP Spec

Q‫މ‬

COPY Q

XP X

ZP Y

X

36

Chapter Two

The representation in (18) above involves a syntactic configuration with a head Q (quantity) merged into the structure. The Q head is realized as a reduplicative morpheme and it creates a copy of its sister or the subpart of its sister. If the sister is a maximal projection (XP), as in (18), the copy is realized in the specifier position. Furthermore, in syntactic (iterative) reduplication in (18), no head movement takes place. Instead, the copying of a subpart of the sister of Q means the copying of a subpart of XP, not of X0. The subpart of XP that is copied is determined in syntactic terms. Summing up, Travis (2001, 2003) proposes a single mechanism for the derivation of all NPN forms, unlike Huddleston and Pullum (2002), who distinguish NPNs that are NPs and those that are PPs, or Haïk (2013), who divides NPN forms into lexical small clauses and lexical coordinate nouns.

3. NPN forms in Polish It appears that NPN structures have not been extensively investigated in Polish as yet. The only studies that the present author is aware of include DaszczyĔska (1997), Dobaczewski (2009) and Rosalska (2011). One of the potential reasons why Polish NPN forms have not attracted attention comparable to that devoted to their English counterparts is that, at least superficially, they do not differ from other phrases. In particular, since Polish is an articleless language, it is not surprising to find a PP with a preposition taking a bare noun (N2) as its complement or a bare noun (N1) taking the thus formed PP as its postmodifier. In other words, such bare nouns would be expected in any NP with a prepositional postmodifier. Thus, the structure of NPN forms does not depart from the standard structure of NPs with embedded PPs. It seems that Polish NPN forms basically involve three prepositions, i.e., po ‘after,’ za ‘behind’ and w ‘in.’ Moreover, expressions with po and za function as subjects (19a), direct objects (19b-c), indirect objects (19d) and adjuncts (19e), whereas those with w occur only as adjuncts (19f):2

2

Abbreviations used in the glosses: 1 – first person, 3 – third person, ACC – accusative, DAT – dative, FEM – feminine, INSTR – instrumental, MASC – masculine, NEUT – neuter, NOM – nominative, PAST – past tense, S – singular.

The Categorial Status and Internal Structure of NPN Forms (19)

37

a. Mijaá dzieĔ za dniem. pass-3S.NEUT.PAST day-NOM after day-INSTR ‘Day after day passed.’ b. Pisaáa ksiąĪkĊ za ksiąĪką. write-3S.FEM.PAST book-ACC after book-INSTR ‘She wrote book after book.’ c. Czytaáem wiersz po wierszu. read-1S.MASC.PAST poem-ACC after poem-INSTR ‘He read poem after poem.’ d. Mówiá studentowi za studentem… tell-3S.MASC.PAST student-DAT after student-INSTR ‘He told student after student…’ e. DzieĔ po dniu zbliĪaliĞmy siĊ do celu. day-NOM after day-INSTR approach-PAST to goal ‘Day after day we were approaching our goal.’ f. Szli áeb w áeb go-3P.PAST head-NOM in head-ACC ‘They went/ran neck and neck.’

Importantly, the case-form of N1 is indicative of the grammatical function of the whole NPN: nominative – subject or adjunct, accusative – direct object, dative – indirect object. This is clearly an argument for categorizing NPN as nominals. Rosalska (2011) focuses on the selected Polish NPNs involving the noun krok ‘step,’ namely krok po kroku lit. ‘step after step,’ krok za krokiem lit. ‘step behind step,’ and krok w krok lit. ‘step in step,’ in the context of the distinction between lexical units and syntactic units. Rosalska (2011) argues that the relevant structures involve the meaning of a succession or sequence of entities or events, emphasizing the gradual character of an event or process. She observes that while Polish NPNs with the prepositions po ‘after’ and za ‘behind’ allow a whole range of nouns, the selection of nouns is restricted in the case of the structure with the preposition w ‘in’ (Rosalska 2011, 157):

38 (20)

Chapter Two a. krok w krok lit. ‘step in step’ (iĞü za kimĞ krok w krok = follow someone closely) b. oko w oko lit. ‘eyeball in eyeball’ (= eyeball to eyeball) c. kropla w kroplĊ lit. ‘drop in drop’ (= exactly the same) d. ramiĊ w ramiĊ lit. ‘shoulder in shoulder’ (= shoulder to shoulder) e. jota w jotĊ lit. ‘iota in iota’ (= exactly the same) f. kubek w kubek lit. ‘mug in mug’ (= exactly the same) g. twarzą w twarz lit. ‘face in face’ (= face to face)

Apart from this, in Polish NPNs with w ‘in,’ N1 is in the nominative and N2 is in the accusative, with the exception of twarzą w twarz ‘face to face,’ where N1 is in the instrumental and N2 is in the accusative. While English NPNs sometimes allow premodification or postmodification of N1 and/or N2 (cf. (10) above), it seems that in Polish NPNs premodification of N1 or N2, at best, results in semantic oddness, if not deviance (Rosalska 2011, 159–60): (21)

a. Czytaáam ksiąĪkĊ za ksiąĪką. ‘I was reading book after book.’ b. ? Czytaáam ciekawą ksiąĪkĊ za ciekawą ksiąĪką. ‘I was reading interesting book after interesting book.’ c. Krok za krokiem posuwaáam siĊ naprzód. ‘Step by step, I was moving forward.’ d. ? Maáy krok za maáym krokiem posuwaáam siĊ naprzód. ‘Small step by small step I was moving forward.’

Also, compared to the English data in (10) above, where either N2 or both N1 and N2 can be premodified and/or postmodified, in Polish the omission of premodifiers and postmodifiers on N1 (22b) yields structures which are probably even less acceptable than those with premodification and postmodification on both nouns (cf. (21) above): (22)

a. ? Czytaá interesującą ksiąĪkĊ za read-3S.MASC.PAST interesting-ACC book-ACC after interesującą ksiąĪką. interesting-INSTR book-INSTR ‘He read interesting book after interesting book.’

The Categorial Status and Internal Structure of NPN Forms

39

b. ?? Czytaá ksiąĪkĊ za interesującą read-3S.MASC.PAST book-ACC after interesting-INSTR ksiąĪką. book-INSTR ‘He read book after interesting book.’

Instead, it appears that in Polish premodifiers on N2 can be omitted if they occur on N1: (23)

Czytaá interesującą ksiąĪkĊ za ksiąĪką. read-3S.MASC.PAST interesting-ACC book-ACC after book-INSTR

By contrast, the presence of a premodifier on N1 and its deletion on N2 is excluded in English NPNs (cf. (11) above).

4. Prospects for a Minimalist account of NPNs in English and Polish The account of the structure and derivation of NPN forms in terms of a reduplication mechanism seems to be an interesting unified approach to the relevant expressions, irrespective of a number of issues that are still unresolved, as far as the structure of the nominal domain is concerned. For instance, the mechanism of reduplication is a workable solution independently of the universal or non-universal nature of D heads, i.e., independently of the fact of whether Polish nominals project a DP-layer (e.g., Rutkowski 2007; Pereltsvaig 2007) or not (e.g., Willim 1998; Boškoviü 2005), as is the case with English nominal expressions (e.g., Abney 1987; Boškoviü 2005). The differences between English and Polish NPNs concerning the distribution and omission of premodifiers and postmodifiers on N1 and N2 could be explained on the basis of the nature of the relevant features on the reduplicative morpheme, i.e., a feature would decide whether premodifiers and postmodifiers are deletable in appropriate circumstances in a given language. It is commonly assumed in the Minimalist literature (e.g., Radford 2000; Adger 2003; Boeckx 2009) that the extended nominal projection includes the lexical projection NP and a number of functional projections, invariably with layers such as nP and DP. The area between nP and DP can optionally involve further functional projections responsible for the characteristics of the nominal domain with respect to grammatical categories such as Number, Gender, etc.

40

Chapter Two

Following Radford’s (2000) nP-shell approach, nominal expressions include an inner NP core headed by a lexical noun and an outer nP shell headed by a light noun. Moreover, Radford (2000) argues that, in English, the light n head is always strong and triggers movement of the noun from the lexical N head to the head n position of nP. Thus, following the relevant movement, the original structure in (24a) becomes (24b): (24)

a. [DP D [nP [NP student ]]] b. [DP D [nP student [NP … ]]]

Adopting Travis’s (2001, 2003) reduplication mechanism, the reduplication of the DP must be triggered by an appropriate Q head (a reduplicative head/morpheme). An important question that remains is whether in NPNs (e.g., student after student) the reduplicative head is hosted by the P head or it merges with the PP after student, making the whole structure a QP (reduplicative phrase). In the latter variant, the reduplicated DP occupies the SpecQP position. Indeed, Travis (2001) suggests that the reduplicative morpheme has a variable nature. Combined with insights from Radford (2000), this line of thinking would produce the following structure: (25)

[QP [DP D [nP student [NP …] Q ] [PP after [DP D [nP student [NP …]]]]]]

The Q label for the relevant functional category is motivated by the quantifying nature of the reduplicative morpheme: the process of reduplication has to do with the plurality of the entities or events. This is corroborated by Radford’s (2009, 132) claim that, while bare definite nominals (e.g., John) are DPs headed by a null D (which is definite in interpretation), bare indefinite nominals (e.g., toast, eggs) are QPs headed by a null Q (which is generic or partitive in interpretation). What also remains problematic is the content of the D head – under the analysis in (25) this would have to be a null D, otherwise it might be assumed that perhaps the preposition takes an nP rather than a DP as its complement.

5. Conclusion Polish data show that NPN forms are nominal expressions. Unless the NPN functions as an adjunct, the form of N1 is determined by the head selecting the whole NPN as its complement. On the other hand, the preposition imposes an appropriate case-form on N2. The prepositional phrase formed by the preposition and N2 serves as a complement

The Categorial Status and Internal Structure of NPN Forms

41

(postmodifier) for N1. Another issue is the way such structures are derived, though it seems that the operation of syntactic (iterative) reduplication proposed in Travis (2001, 2003) might be a solution for both English and Polish NPN structures. While the English data seem to blur the picture because N1 and N2 are not distinguishable in terms of case inflection, Polish NPNs enable a different perspective on NPN structures in general. Obviously, the present chapter is merely a preliminary study and proposals in section 4 above are merely tentative. There are a number of issues that need to be taken into consideration in the further exploration of NPN forms in English, Polish, and possibly in other languages. The reduplication-based approach proposed in Travis (2001, 2003) advocated above needs to be modified in the context of the current advances in the Minimalist Program, in particular phase-based derivations. As for the categorial status and internal structure of NPN forms, given that NPNs are QPs, what matters is the presence of various functional layers in the extended nominal projection, as well as the question of whether the relevant functional categories are overt or covert, and the status of nominal expressions such as DPs or NPs.

References Abney, Steven Paul. 1987. “The English Noun Phrase in its Sentential Aspect.” PhD diss., MIT. Adger, David. 2003. Core Syntax: A Minimalist Approach. Oxford: OUP. Boeckx, Cedric. 2009. Bare Syntax. Oxford and New York: OUP. Boškoviü, Željko. 2005. “On the Locality of Left Branch Extraction and the Structure of NP.” Studia Linguistica 59:1–45. DaszczyĔska, Izabela. 1997. “Reduplikacja jako Jeden ze Sposobów Pierwotnej Derywacji Frazeologicznej.” In Wspóáczesne Tendencje Rozwoju JĊzyków SáowiaĔskich, Vol. 2, edited by Michaá Blicharski and Henryk FontaĔski, 54–62. Katowice: Wydawnictwo Uniwersytetu ĝląskiego. Dobaczewski, Adam. 2009. “Operacje Iterujące w JĊzyku Polskim (Wprowadzenie do Opisu).” Poradnik JĊzykowy 9:26–36. Haïk, Isabelle. 2013. “Symmetric Structures.” Corela 11:1. Accessed September 08, 2013. http://corela.edel.univ-poitiers.fr/index.php?id=28 75. Huddleston, Rodney, and Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: CUP.

42

Chapter Two

Jackendoff, Ray. 2008. “Construction after Construction and its Theoretical Challenges.” Language 84:8–28. Moravcsik, Edith. 1978. “Reduplicative Constructions.” In Universals of Human Language, edited by Joseph H. Greenberg, 297–334. Stanford, CA: Stanford University Press. Pereltsvaig, Asya. 2007. “On the Universality of DP: A View from Russian.” Studia Linguistica 61:59–94. Pi Chia-Yi, Tony. 1995. “The Structure of English Iteratives.” Canadian Linguistics Association Proceedings (Toronto Working Papers in Linguistics), 434–45. Pskit, Wiktor. 2012. “The English NPN Forms: Words or Constructions?” In Exploring Language through Contrast, edited by Waldemar Skrzypczak, Tomasz Fojt, and Slawomir Wacewicz, 126–44. Cambridge: Cambridge Scholars Publishing. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language, London: Longman. Radford, Andrew. 2000. “NP Shells.” Essex Research Reports in Linguistics 33:2–20. —. 2009. Analysing English Sentences. A Minimalist Approach. Cambridge: CUP. Rosalska, Paulina. 2011. “Krok po Kroku, Krok za Krokiem, Krok w Krok – Konstrukcje Skáadniowe czy Jednostki JĊzyka?” Linguistica Copernicana 2(6):149–62. Rutkowski, Paweá. 2007. “Hipoteza Frazy Przedimkowej jako NarzĊdzie Opisu Skáadniowego Polskich Grup Imiennych.” PhD diss., Uniwersytet Warszawski. Travis, Lisa. 2001. “The Syntax of Reduplication.” Proceedings of NELS 31:455–69. Travis, Lisa. 2003. “Reduplication Feeding Syntactic Movement.” In Proceedings of the 2003 Annual Conference of the Canadian Linguistic Association, edited by Sophie Burelle and Stanca Somesfalean, 236–47. Université du Québec à Montréal, Département de linguistique et de didactique des langues. Willim, Ewa. 1998. “On the DP-Hypothesis in Polish, an Articleless Language.” In Projections and Mappings. Studies in Syntax. PASE Studies and Monographs 5, edited by Piotr Stalmaszczyk, 137–58. Lublin: Folium.

CHAPTER THREE THE DP ANALYSIS OF POLISH CLAUSAL SUBJECTS MAàGORZATA WIETRZYK

1. Introduction The aim of this chapter is to derive a uniform theory of the categorial status of Polish subjects through a comparison of the properties of nominal and clausal subjects. The chapter consists of six sections. Section 2 enumerates the possible categories and structures of the elements appearing in the subject position. Section 3 outlines the structure of sentences with clausal subjects. In section 4, clausal subjects are tested against the subjecthood properties typical of nominal subjects. Section 5 provides arguments for the presence of a DP layer in Polish. Finally, section 6 offers some conclusions.

2. Elements appearing in the subject position – categories and structures Because of the fact that a subject represents an element of the reality being discussed, its position is most often occupied by a noun, pronoun, or another nominal element, one of whose defining characteristics is the very property of denoting objects. Consider the examples in (1).1

1 For the sake of clarity, subjects appearing in the presented examples will be italicised throughout the chapter.

44 (1)

Chapter Three a. Drzewa uginaáy siĊ pod ciĊĪarem Ğniegu. snow trees were-bending REFL2 under load ‘Trees were bending under the load of snow.’ b. Oni wiedzieli co robiü. what do-INF they-VIR knew ‘They knew what to do.’ c. WieĪa Eiffla zostaáa zbudowana w 1889. Tower Eiffel’s was built in 1889 ‘The Eiffel Tower was built in 1889.’

However, a subject can also take the form of an infinitive, as in sentence (2). (2)

Wystarczyáo spojrzeü tylko raz. was-enough-IMP look-INF only once ‘It was enough to look only once.’

As a test for the subject function of the infinitive, we can try, as suggested by Jodáowski (1976, 66), to turn it into a noun, e.g., spojrzeü ‘to look’ > spojrzenie ‘a look,’ czytaü ‘to read’ > czytanie ‘reading.’ Among other elements functioning as subjects, we can enumerate, after Klemensiewicz (1969), Jodáowski (1976), and StrutyĔski (2009), numerals and adjectives. Consider sentences (3) and (4), respectively. (3)

Byáa juĪ dziesiąta. was already tenth-3SG.FEM ‘It was already ten.’

(4)

Niski wyciągnąá pistolet. short pulled-out gun ‘The short man pulled out a gun.’

These two types of subjects, however, could be analysed, after Kallas (1991), as examples of ellipsis with the words godzina ‘hour’ elided in 2

The list of abbreviations: ACC – accusative; DAT – dative; F/FEM – feminine; FUT – future; GEN – genitive; IMP – impersonal; INF – infinitive; INSTR – instrumental; iT – interpretable tense feature; M/MASC – masculine; N/NEUT – neuter; NOM – nominative; NON-VIR – non-virile; PL/PLUR – plural; REFL – reflexive; SG/SING – singular; uT – uninterpretable tense feature; V/VIR – virile; 1 – 1st person; 2 – 2nd person; 3 – 3rd person.

The DP Analysis of Polish Clausal Subjects

45

sentence (3) (dziesiąta instead of dziesiąta godzina) and mĊĪczyzna ‘man’ in sentence (4) (niski instead of niski mĊĪczyzna). Ellipsis is possible because the utterance remains unambiguous and the meaning of the elided word can be easily recovered from the context. As far as the grammatical structure of a subject is concerned, the following types can be enumerated (cf. Jodáowski 1976, 69): a. a single word (5) Ubodzy proszą o pomoc. poor-NOM.3PL.MASC ask-3PL for help-ACC ‘The poor ask for help.’ b. a sequence of coordinated words or phrases (i.e., Polish podmiot szeregowy) (6) Ania i Kasia poszáy wczoraj Ann-NOM and Kate-NOM went-3PL.NON-VIR yesterday do szkoáy. to school-GEN ‘Ann and Kate went to school yesterday.’ c. an inseparable cluster of a noun and its determiner (7) Jego córka zdobyáa nagrodĊ. his daughter won prize-ACC ‘His daughter won the prize.’ d. a prepositional phrase (8) MiĊdzy szóstą a siódmą pasowaáo wszystkim. between sixth-INSTR and seventh-INSTR suited-IMP everyone-DAT ‘Between six and seven suited everyone.’ (Nagórko 2012, 286)

The examples presented above show that, no matter what its structure or surface form, the subject position in Polish is most often occupied by a nominal expression. The only structure that may seem problematic for the uniformity of the theory (postponing the analysis of clausal subjects until sections 3 and 4) is exemplified by the sentence in (8), where the subject position appears to be occupied by a prepositional phrase. It may be posited, however, that, similarly to examples (3)–(5), some part of the subject has been elided here. It may be assumed that in (8) the element that has undergone ellipsis is the infinitive spotkaü siĊ ‘to meet’ (leaving miĊdzy szóstą a siódmą instead of spotkaü siĊ miĊdzy szóstą a siódmą). The subject status of infinitives has already been demonstrated at the beginning of this section.

46

Chapter Three

3. Clausal subjects – outline of the structure In Sáownik terminologii jĊzykoznawczej, Goáąb, Heinz, and PolaĔski (1968, 426–27) define a subject clause as a subordinate clause functioning as an unexpressed or generally indicated subject of the superordinate clause. Consider examples (9) and (10). (9)

Byáo oczywiste (to), Īe Jan wróci. was obvious it that John-NOM will-come-back-3SG ‘It was obvious that John would come back.’

(10)

To, Īe Jan wróci, byáo oczywiste. it that John-NOM will-come-back-3SG was obvious ‘That John would come back was obvious.’

In both sentences, to, Īe Jan wróci ‘that John would come back’ functions as the subject of the superordinate clause. In (9), the subject clause is said to be extraposed, while in (10) the clause fills the subject slot.3 Clausal subjects in Polish may be introduced by three types of hypotactic, i.e., subordinate conjunctions: declarative Īe ‘that,’ irrealis Īeby ‘so that,’ and interrogative czy ‘whether/if.’ Each of the three groups can be further divided into subgroups, according to the type of the predicate appearing in the matrix clause (e.g., a copula verb, a psych verb, a relational verb, a causative verb, a raising verb, or a passive structure). In this chapter, however, we will focus only on the examples introduced by Īe ‘that,’ which are the most numerous type of clausal subjects.

3

Similarly, Bondaruk (2000, 61) defines a subject clause as “a sentence which either itself appears in the subject position or is extraposed from the subject position and linked with the word it [or its Polish counterpart to].” (i) To, Īe oblaá egzamin rozczarowaáo nas. it that he-failed exam disappointed us ‘That he failed the exam made us disappointed.’ (ii) Rozczarowaáo nas to, Īe oblaá egzamin. disappointed us it that he-failed exam ‘It made us disappointed that he failed the exam.’ (Bondaruk 2000, 61)

The DP Analysis of Polish Clausal Subjects

47

4. CP subjects checked against the subjecthood properties of NP/DP subjects The majority of subjects within the most common category (NP/DP subjects) display the set of subjecthood properties originally enumerated for English data by Falk (2006, 1–2). Let us examine whether Polish subject clauses can be associated with a full range of prototypical subject properties, and, thus, whether they can be treated as NPs/DPs themselves. Obligatory raising. NPs/DPs undergo raising precisely because they are NPs/DPs. In other words, NPs/DPs raise because, as opposed to other categories, they need Case, or because they have the capacity, not possessed by other categories, to check a formal feature in T. (11)

Jani zdaje siĊ [ ti byü dobrym nauczycielem]. teacher John-NOM seems REFL be-INF good ‘John seems to be a good teacher.’

Clausal subjects, like NP/DP subjects, undergo obligatory raising, as proved by the example in (12). Moreover, what is proved by the ungrammaticality of the example in (13), is that not only is this subject able to be raised, but it has to be. (12)

[CP To, Īe jest niewinny]i, zdaje siĊ [ti byü it that is-3SG innocent-NOM.MASC appears REFL be-INF prawdą]. truth-INSTR ‘That he is innocent appears to be true.’

(13)

jest niewinny] byü * Zdaje siĊ [[CP to, Īe appears REFL it that is-3SG innocent-NOM.MASC be-INF prawdą]. truth-INSTR * ‘It appears that he is innocent to be true.’

Agreement. According to Jodáowski (1976), the form of the predicative verb must agree in person, number, and gender with the form of the subject. Grzegorczykowa (1996), as well as PolaĔski and Nowak (2011), paraphrase this requirement and claim that the relation between the subject and the predicate is characterized by mutual accommodation. It means that the nominal phrase bearing the nominative case is accommodated by the verb (e.g., Janek przybiegá ‘John.NOM [came running].3SG.MASC’; przybiec ‘to come running’ is a verb requiring the

48

Chapter Three

nominative), while the noun determines the three features of the verb mentioned by Jodáowski (1976). In the case of several coordinated subjects appearing within one clause, the form of the predicate is plural when the subjects are joined by the conjunction i ‘and,’ as shown in (6), repeated below as (14). (14)

Ania i Kasia poszáy wczoraj Ann-NOM and Kate-NOM went-3PL.NON-VIR yesterday do szkoáy. to school-GEN ‘Ann and Kate went to school yesterday.’

To check whether the same property holds true for Polish clausal subjects, let us consider examples (15a) and (15b). (15)

a. [CP To, Īe káamie], jest oczywiste. it that lies-3SG is-3SG obvious ‘That he is lying is obvious.’ jej nigdy wiĊcej]. b. Okazaáo siĊ, [CP Īe nie zobaczą turned-out-3SG REFL that not will-see-3PL her never more ‘It turned out that they would never see her again.’

In (15a), to, Īe káamie ‘that he is lying’ triggers third person singular agreement on the tensed verb jest ‘is.’ Similarly, in (15b), where the clause Īe nie zobaczą jej nigdy wiĊcej ‘that they would never see her again’ has been extraposed from the subject position, the verb takes the form of the third person singular neuter okazaáo siĊ ‘it turned out.’ However, in order to check whether the CP really determines the morphology of the verb, sentences with coordinated CPs, such as the ones in (16), should be analysed. (16)

a. [CP[CP To, Īe straciá pracĊ] i [CP to, Īe it that lost-3SG job-ACC and it that wszystko przemilczaá]] naprawdĊ nas zdziwiáo/ everything-ACC left-unsaid-3SG really us surprised-3SG.NEUT/ *zadziwiáy/ *zadziwili. surprised-3PL.NON-VIR./ surprised-3PL.VIR ‘That he lost his job and that he left everything unsaid really surprised us.’

The DP Analysis of Polish Clausal Subjects

49

káamie] b. Denerwuje/ *denerwują mnie, [CP [CP Īe ciągle me-ACC that all-the-time lies-3SG annoys-3SG/ annoy-PL i [CP Īe nigdy nie przyznaje siĊ do báĊdu]]. and that never not admits-3SG REFL to mistake-GEN ‘It annoys me that he lies all the time and that he never admits to his mistakes.’

Once again, in both sentences, the verbs display the third person singular morphology. In (16a), where the to ‘it’ element is present twice, as well as in (16b), where this element is left out and the CPs are extraposed from the subject position, the forms of the verbs zdziwiáo ‘surprised,’ and denerwuje ‘annoys’ are not influenced by the coordinated CPs and do not change to the plural. The fact that plural agreement is not possible in Polish sentences with conjoined clausal subjects may be used as an argument against treating CPs as nominal subjects. The situation changes, however, when the predicate is replaced with a copula verb. (17)

[CP [CP To, Īe wygra] i [CP to, Īe przegra]] it that will-win-3SG and it that will-lose-3SG są dwoma scenariuszami wartymi rozwaĪenia. are two scenarios worth consideration ‘That he’ll win and that he’ll lose are the two scenarios worth taking into consideration.’

Moreover, the grammaticality of (18b) shows that conjoined CPs in the subject position can license the presence of the quantificational adverb równie ‘equally’ in Polish, like a plural DP (cf. (18a)). (18)

a. Obie propozycje wydają siĊ równie interesujące. both proposals seem-PL REFL equally interesting ‘Both proposals seem equally interesting.’ jest niewinny] i [CP to, Īe b. [CP [CP To, Īe it that is-3SG innocent-NOM.MASC and it that zabiá prezydenta]] wydaje siĊ równie moĪliwe. killed-3SG president-ACC seems-3SG REFL equally possible ‘That he is innocent and that he killed the president seem equally possible.’

Coordination. Another feature common to NP/DP and CP subjects is the fact that not only the prototypical subjects but also the latter type can function as arguments of coordinated clauses. Consider (19) and (20).

50

Chapter Three

(19)

Maria wyszáa i jeszcze nie not Mary-NOM left-3SG.FEM and yet ‘Mary left and hasn’t come back yet.’

wróciáa. came-back-3SG.FEM

(20)

To, Īe przyszedáeĞ ucieszyáo mnie i wywoáaáo it that came-2SG made-happy-3SG me and brought-3SG burzĊ oklasków. storm-ACC applause-GEN ‘That you came made me happy and brought a storm of applause.’

ĝwidziĔski (1992, 3) additionally points out that it is possible to coordinate a sentential subject with a nominal subject, as presented in (21), which, assuming that only elements of the same category can be coordinated, proves the nominal status of clausal subjects. (21)

Jana dziwi, Īe Maria wybraáa Piotra, John-ACC surprises that Mary chose-3SG.FEM Peter-ACC i jej zupeány brak gustu. and her complete lack taste-GEN ‘It surprises John that Mary has chosen Peter and her complete lack of taste.’ (ĝwidziĔski 1992, 3)

An argument that seems to support this claim follows from the selectional properties of nominal copula clauses. In this type of structure, presented in (22a), the subject and the predicate are required to have the same categorial status (cf. Citko 2008). Thus, the sentential subject present in example (22b) should also be analysed as a DP. (22)

a. [DP Janek] to [DP mój przyjaciel]. Janek is my friend-NOM ‘Janek is my friend.’ b. [DP To, Īe nie zdaáeĞ egzaminu] to [DP skandal]. scandal-NOM it that not passed-2SG exam-GEN is ‘That you haven’t passed the exam is a scandal.’

Control. Moreover, clausal subjects, like NP/DP subjects, control the PRO argument of a subordinate clause (its subject), as in (23). (23)

Mariai wyszáa nie PROi mówiąc dokąd idzie. Mary-NOM left-3SG.FEM not PRO saying where go-3SG ‘Mary left without saying where she was going.’

The DP Analysis of Polish Clausal Subjects (24)

51

dotaráo do nas nagle [To, Īe siĊ rozstali]i it that REFL broke-up-3PL dawned-3SG to us suddenly PROi wywoáując wielkie zamieszanie. PRO causing [great commotion-ACC] ‘That they broke up dawned on us suddenly causing a great commotion.’

Case. Nominal subjects in Polish sentences typically bear the nominative case and they are referred to as grammatical subjects (cf., e.g., Jodáowski 1976; StrutyĔski 2009; PolaĔski and Nowak 2011).4 (25)

Autor odczytaá pierwszy rozdziaá ksiąĪki. author-NOM read-out-3SG.MASC [first chapter-ACC] book-GEN ‘The author read out the first chapter from the book.’

If clausal subjects are to be treated as NPs/DPs, they should also require Case to satisfy the Case Filter. This is claimed by Boškoviü (1995), who analyses English and suggests that the complementizer that heading the subject clause has the nominative case, which should agree in Case with T. The responsibility for case assignment to clausal subjects in English has also been attributed to the element that by Pesetsky and Torrego (2001, 2004, 2007) (henceforth P&T). Can their theory, however, be applied to Polish data? In their analysis of clausal subjects in English, P&T (2001, 2004, 2007) explain that the obligatoriness of the element that introducing subject clauses is the result of an iT feature on that which is capable of deleting uT on the matrix clause C. As in the Polish examples, it is the element to ‘it’ which is obligatory in subject clauses appearing in the canonical subject position (Īe, assumed to be a counterpart of that, is always present in embedded clauses), it might be analysed as an item with a T feature. Moreover, following P&T (2001) who claim that the Nominative is uT on D, it could be assumed that the T feature on to ‘it’ is also uninterpretable, as this element shows obvious differences in its form depending on case, i.e., to ‘it’ Nominative, Accusative, tego ‘it’ Genitive, temu ‘it’ Dative, tym ‘it’ Instrumental, Locative. That to ‘it’ is the exponent of the checked Case is also claimed by Miechowicz-Mathiasen 4

However, although far less numerous, there also exist structures with logical subjects expressed by the genitive, accusative, and dative or, less frequently, by the instrumental. Although Polish subjects may bear various case forms, only nominative is considered typical of nominal subjects. The main functions of other cases that may be assigned to Polish subjects are concisely described, e.g., by Milewski (2009, 85) and Mecner (2007, 217–22).

52

Chapter Three

(2013) who shows that it bears nominative in the subject position (cf. (26a)), accusative in the object position (cf. (26b)), and genitive under negation (cf. (26c)). (26)

a. [To, [CP Īe siĊ spóĨniáa]] zaskoczyáo go. [this-NOM that REFL late-3SG.F.PAST] surprised-3SG.N.PAST him ‘That she was late surprised him.’ spóĨni]]. b. Wiedziaá [(to), [CP Īe ona siĊ knew-3SG.M.PAST (this-ACC) that she REFL be-late-3SG.FUT ‘He knew that she was going to be late.’ spóĨni]]. c. Nie wiedziaá [(tego), [CP Īe ona siĊ not knew-3SG.M.PAST (this-GEN) that she REFL be-late-3SG.FUT ‘He didn’t know that she was going to be late.’ (Miechowicz-Mathiasen 2013, 197)

The examples presented above seem to prove the subjecthood properties of clausal subjects. They undergo obligatory raising, can function as arguments of coordinated clauses, control the PRO argument of the subordinate clause, and can be coordinated with prototypical subjects. If subject clauses have so many properties characteristic of prototypical subjects, then the question of their categorial status should be asked.

5. Arguments for a DP layer in Polish Davies and Dubinsky (2001) claim that all subjects must be DPs (sometimes covertly so). That is why, CPs, which are non-NP subjects, are thought to have a DP-shell structure, where the head of the DP carries the usual set of ϕ-features. This head, which is thought to be null in English, could be analysed as to ‘it’ for Polish clausal subjects, following Wietrzyk (2010) and Miechowicz-Mathiasen (2012), as shown in (27), where the element to ‘it’ carrying the usual set of ϕ-features is in the head position of a DP built up on top of a CP.

The DP Analysis of Polish Clausal Subjects (27)

53

DP D‫މ‬ D to

CP C‫މ‬ C

TP

To adopt the above-mentioned structure, however, we need to assume that Polish, being an articleless language, in fact employs a DP layer. And, indeed, there exist arguments in favour of this hypothesis. LindeUsiekniewicz and Rutkowski (2007) observe that NP cannot be the highest nominal projection in Polish as coordinated singular NPs may be premodified by a singular adjective, as in (28a), which suggests that the position occupied by the modifier cannot be NP-internal. (28)

a. Pijam gorzką herbatĊ i kawĊ. and coffee I-drink unsweetened-SG tea ‘I drink unsweetened tea and coffee.’ (Linde-Usiekniewicz and Rutkowski 2007, 112) b. * gorzkie [ConjP herbata i kawa] unsweetened-PL tea and coffee (Linde-Usiekniewicz and Rutkowski 2007, 110)

Example (28b) is clearly ungrammatical, which proves that coordinated phrases cannot be treated as a single plural constituent with respect to adjectival modification. On the other hand, as shown in (28a), both conjuncts in the coordinated NP are modified by the adjective, which means that the AP must occupy a position above the conjunction. Another piece of evidence comes from noun/pronoun asymmetries exemplified in (29). (29)

a. [sam Chomsky] czytaá mój artykuá alone Chomsky read my article ‘even Chomsky read my article’

Chapter Three

54

b. [on sam] czytaá mój artykuá he alone read my article ‘even he read my article’ c. *[sam on] alone he

czytaá mój artykuá read my article (Rutkowski 2002, 161)

When adjectives such as sam ‘alone’ appear with personal pronouns, they may only occupy the position to the right of the pronominal head. On the other hand, example (29a) shows a pattern, where the adjective precedes the noun, which is a general, unmarked order of the Polish language. An explanation of the above-mentioned asymmetries is provided by Abney’s (1987) DP hypothesis if we assume that personal pronouns occupy the D position, while nouns reside in N. Rutkowski (2002), following Cardinaletti (1993) and Progovac (1998), claims that personal pronouns are base-generated in N and raise to D (crossing the adjective in its fixed Spec,NP position). The process is shown in (30). DP

(30)

Spec

D‫މ‬ D

oni

NP Spec

N‫މ‬

AP

N

sam sam

Chomsky ti (Rutkowski 2002, 163)

The suggestion that to ‘it’ residing in D selects a CP complement and projects into DP is proved, as noticed by Miechowicz-Mathiasen (2012), by adjectival Left Branch Extractions (LBE) and adjunct extractions,

The DP Analysis of Polish Clausal Subjects

55

which are blocked in languages with articles but possible in articleless ones. Consider (31) and (32). (31)

a. Przyjechali (oni) wielcy jĊzykoznawcy. arrived-3.PL.VIR [(they) great linguists-NOM] ‘The great linguists have arrived.’ b. Wielcy przyjechali (*oni) jĊzykoznawcy. great-NOM arrived-3PL.VIR [(*they) linguists-NOM] ‘* The great have arrived linguists.’ c. Witali nas (oni) wielcy jĊzykoznawcy welcomed-3PL.VIR us [(they) great linguists-NOM] z Cambridge. from Cambridge ‘We were welcomed by the great linguists from Cambridge.’ d. Z jakiego miasta witali nas (*oni) wielcy from [which city-GEN] welcomed-3PL.VIR us [(they) great jĊzykoznawcy? linguists-NOM] ‘* [From which city] did the great linguists ___ welcome us?’ (Miechowicz-Mathiasen 2012, 78)

(32)

a. Podejrzewaáem (to), Īe Piotr okradnie MariĊ. suspected-1SG.M (this) that Peter-NOM rob-3SG.FUT Mary-ACC ‘I suspected that Peter would rob Mary.’ b. Ktoi podejrzewaáeĞ, Īe ti okradnie MariĊ? rob-3SG.FUT Mary-ACC who-NOM suspected-2SG.M that ‘Who did you suspect would rob Mary?’ c. * Ktoi podejrzewaáeĞ, to Īe ti okradnie MariĊ? who-NOM suspected-2SG.M this that rob-3SG.FUT Mary-ACC ‘Who did you suspect would rob Mary?’ (Miechowicz-Mathiasen 2012, 79)

Examples in (31b) and (31d) show that both adjectival LBE and adjunct extraction are blocked if a nominal, out of which we want to extract, is introduced by a personal pronoun (an instantiation of D). Similarly, extraction is not possible in (32c), which proves that to ‘it’ projects into a DP built on top of CP. If, however, the introductory element is absent from the clausal argument, the sentence is perfectly grammatical, as in (32b).

56

Chapter Three

6. Conclusions An analysis of the possible structures functioning as subjects in the Polish language shows that the majority of them are represented by NPs/DPs. Even potentially problematic phrases, such as niski ‘short,’ dziesiąta ‘tenth,’ miĊdzy szóstą a siódmą ‘between sixth and seventh’ that we may be inclined to treat as AP, NumP, and PP, respectively, can be proved to have a nominal status. Thus, the need arises to examine whether it is possible for clausal subjects to become part of the uniform theory. Although the analyses of agreement phenomena and Case do not prove unambiguously the nominal status of clausal subjects, the ones that support their NP/DP character, i.e., the arguments from raising structures, coordination and control facts, outweigh the possible drawbacks of such a proposal. Finally, it has been shown that there are reasons, based on the analysis of adjectival Left Branch Extraction, adjunct extraction, and noun/pronoun asymmetries, to posit a DP layer in Polish and to claim that Polish clausal subjects are in fact DPs with the head of the phrase filled with the element to ‘it,’ carrying the usual set of ϕ-features.

References Abney, Steven Paul. 1987. “The English Noun Phrase in its Sentential Aspect.” PhD diss., MIT. Bondaruk, Anna. 2000. “Subject Clauses in English and Polish.” In PASE Papers in Language Studies. Proceedings of the 8th Annual Conference of the Polish Association for the Study of English, edited by BoĪena Rozwadowska, 61–71. Wrocáaw: Aksel. Boškoviü, Željko. 1995. “Case Properties of Clauses and the Greed Principle.” Studia Linguistica 49:32–53. Cardinaletti, Anna. 1993. “On the Internal Structure of Pronominal DPs.” University of Venice Working Papers in Linguistics 3.2:1–20. Citko, Barbara. 2008. “Small Clauses Reconsidered: Not So Small and Not All Alike.” Lingua 118:261–95. Davies, William D., and Stanley Dubinsky. 2001. “Functional Architecture and the Distribution of Subject Properties.” In Objects and Other Subjects, edited by William D. Davies and Stanley Dubinsky, 247–79. Dordrecht: Kluwer. Falk, Yehuda. 2006. Subjects and Universal Grammar: An Explanatory Theory. Cambridge: CUP.

The DP Analysis of Polish Clausal Subjects

57

Goáąb, Zbigniew, Adam Heinz, and Kazimierz PolaĔski. 1968. Sáownik Terminologii JĊzykoznawczej. Warszawa: PWN. Grzegorczykowa, Renata. 1996. Wykáady z Polskiej Skáadni. Warszawa: PWN. Jodáowski, Stanisáaw. 1976. Podstawy Polskiej Skáadni. Warszawa: PWN. Kallas, Krystyna. 1991. “JednorodnoĞü Polskich Konstrukcji WspóárzĊdnych.” In Words are Physicians for an Ailing Mind, edited by Maciej Grochowski and Daniel Weiss, 257–64. München: Verlag Otto Sagner. Klemensiewicz, Zenon. 1969. Zarys Skáadni Polskiej. Warszawa: PWN. Linde-Usiekniewicz, Jadwiga, and Paweá Rutkowski. 2007. “NP Coordination as a New Argument in the Debate on the DP Analysis of Polish.” LSO Working Papers in Linguistics 6:103–17. Madison: University of Wisconsin-Madison. Mecner, Paweá. 2007. Elementy Gramatyki Umysáu. Kraków: Universitas. Miechowicz-Mathiasen, Katarzyna. 2012. “Case, Tense and Finite Clausal Arguments in Polish.” Studies in Polish Linguistics 7:63–81. —. 2013. “Case, Tense and the that-Omission Asymmetry in Polish Clausal Arguments.” In Formal Approaches to Slavic Linguistics: The Third Indiana Meeting 2012, edited by Steven Franks, Markus Dickinson, George Fowler, Melissa Witcombe, and Ksenia Zanon, 190–203. Michigan: Michigan Slavic Publications. Milewski, Tadeusz. 2009. JĊzykoznawstwo. Warszawa: PWN. Nagórko, Alicja. 2007. Zarys Gramatyki Polskiej. Warszawa: PWN. —. 2012. PodrĊczna Gramatyka JĊzyka Polskiego. Warszawa: PWN. Pesetsky, David, and Esther Torrego. 2001. “T-to-C Movement: Causes and Consequences.” In Ken Hale: A Life in Language, edited by Michael Kenstowicz, 355–426. Cambridge, MA: MIT Press. —. 2004. “Tense, Case, and the Nature of Syntactic Categories.” In The Syntax of Time, edited by Jacquelin Guéron and Jacqueline Lacarme, 495–537. Cambridge, MA: MIT Press. —. 2007. “The Syntax of Valuation and the Interpretability of Features.” In Phrasal and Clausal Architecture: Syntactic Derivation and Interpretation. In Honor of Joseph Emonds, edited by Simin Karimi, Vida Samiian, and Wendy K. Wilkins, 262–94. Amsterdam: John Benjamins. PolaĔski, Edward, and Tomasz Nowak. 2011. Najnowszy PodrĊcznik Gramatyki JĊzyka Polskiego. Kraków: Petrus. Progovac, Ljiljana. 1998. “Determiner Phrase in a Language without Determiners.” Journal of Linguistics 34:165–79.

58

Chapter Three

Rutkowski, Paweá. 2002. “Noun/Pronoun Asymmetries: Evidence in Support of the DP Hypothesis in Polish.” Jezikoslovlje 3.1-2:159–70. StrutyĔski, Janusz. 2009. Gramatyka Polska. Kraków: Wydawnictwo Tomasz StrutyĔski. ĝwidziĔski, Marek. 1992. “Realizacje Zdaniowe Podmiotu-Mianownika, czyli o Strukturalnych Ograniczeniach Selekcyjnych.” In Opisaü Sáowa. Materiaáy Ogólnopolskiej Sesji Naukowej w RocznicĊ ĝmierci Profesor Danuty Buttler “Teoretyczne i Metodologiczne Zagadnienia Leksykologii,” edited by Andrzej Markowski, 188–201.Warszawa: Dom Wydawniczy Elipsa. Wietrzyk, Maágorzata. 2010. “Clausal Subjects in English and Polish in the Theory of David Pesetsky and Esther Torrego (2001, 2004).” In Young Linguists in Dialogue. The 3rd Conference, edited by Ewelina Mokrosz, Ewa Pająk, and Sáawomir Zdziebko, 153–64. Lublin: Wydawnictwo KUL.

CHAPTER FOUR HOW RELATIVE ARE CLEFT CLAUSES IN ENGLISH IT-CLEFTS? EWELINA MOKROSZ

1. Introduction The aim of this chapter is to present another argument in favour of the analysis of cleft clauses as relative clauses. The findings will shed new light on the syntactic analysis of cleft clauses with clefted personal pronouns. The two types of relative clauses (RCs) to which cleft clauses (CCs) are most legitimately compared are restrictive relative clauses (RRCs) and appositive relative clauses (ARCs). All three types are presented below. (1)

(It is the lecturers) who failed the test on didactics.

CC

(2)

a. (Jill spoke to) the lecturers that failed the test on didactics. RRC b. (Jill spoke to) the lecturers, who failed the test on didactics. ARC (de Vries 2002, 16)

The chapter focuses only on examples with head nominal expressions corresponding to the subject position in a subordinate clause, as in (1) and (2) above. In particular, we will examine clauses with head personal pronouns. Section 2 unveils a variety of agreement patterns found in both CCs and RCs. In section 3, attention is paid to a syntactic analysis of it-clefts with clefted personal pronouns. Here, the aim is twofold. Section 3.1 is an attempt to solve the relative puzzle of cleft clauses by defining their exact status. In section 3.2, we propose a solution that demonstrates how the agreement patterns in cleft clauses can be reconciled with their relative nature.

Chapter Four

60

2. Agreement patterns in cleft and relative clauses This part of the chapter deals with agreement patterns observed in CCs and RCs. Agreement patterns represent another point of reference which can be used to either validate or refute the comparison between CCs and RCs. Firstly, some observations will be made with regard to agreement in cleft clauses. Secondly, it will be shown whether the patterns recognized in cleft clauses comply with those discovered in relative clauses.1 The agreement in question refers to a relation between items in the following two pairs: one pair consists of a personal pronoun in the main clause and a verb in the subordinate clause, while the other pair encompasses a personal pronoun in the main clause and a reflexive pronoun in the subordinate clause.

2.1. Agreement patterns in it-clefts As reported by Akmajian (1970), Sornicola (1988) and Mokrosz (2014), it-clefts exhibit a variety of agreement patterns. Examples in (3)–(5) present patterns from different dialects of English found in Akmajian (1970).2 (3)

Dialect I It is me who is responsible.

(4)

Dialect II a. It is me who(m) John is after. b. It is I who is sick.

(5)

Dialect III a. It is I who am/*is responsible. b. It is me who *am/is responsible.

Akmajian (1970, 150)

Akmajian (1970, 152)

Akmajian (1970, 153)

1 This study of relative clauses was based on a corpus research (The Corpus of Global Web-Based English created by Mark Davies 2013). If not indicated otherwise, all English examples presented in this chapter were taken from the corpus. 2 Akmajian (1970) does not name the dialects in question.

How Relative are Cleft Clauses in English it-Clefts?

61

In Dialect I, there is a partial agreement, i.e., solely in number, between the italicized personal pronoun and the verb.3 The same type of agreement is present in Dialects II and III. In the former, as opposed to the latter, the agreement is indifferent to the case on the pronoun. Full agreement, i.e., in person and number, is less common and appears only in Dialect III. Example (6) below incorporates another pattern which, in a questionnaire conducted by Mokrosz (2014), appears to be preferred with verbs such as like, as in (6a), rather than with a copula verb, as in (6b). The judgements are provided next to each sentence with a given verb. The corpus-based research confirmed the scarcity of patterns such as me who am. It was possible, however, to find examples with a full agreement if the person feature of the pronoun changes, as demonstrated in (6c). In the case of the 2nd person pronoun, however, it is impossible to decode the case. We consider the pattern me who am as highly marked. (6)

me who like/*?am a. It is me who like flowers. 3 – ungrammatical, 1 – acceptable, 1 – grammatical b. *? It is me who am young. 4 – ungrammatical, 1 – acceptable c. No it is you who are inconceivable DERRICK…

In (7) below, there are agreement patterns involving a reflexive pronoun. (7)

a. It’s me who has to protect myself/himself. (Akmajian 1970, 156) b. It is I who like myself/*himself. (Mokrosz 2014, 66)

On the basis of the examples in (7), it can be concluded that reflexive pronouns show either a 3rd person or the person of the clefted pronoun when the verb is marked for the 3rd person (see Akmajian 1970; Sornicola 1988). When, however, the verb shows a non-3rd person as in (7b), the reflexive obligatorily agrees in person with the clefted pronoun.

3

Throughout the chapter, the term ‘partial agreement’ refers to agreement in number, while ‘full agreement’ refers to agreement in both person and number.

62

Chapter Four

2.2. Agreement patterns in relative clauses In English relative clauses, a head noun agrees with an embedded verb as in (8). (8)

John is in love with a girl who studies/*study English.

RRC

Akmajian (1970) is one of the few who draw attention to the difference in agreement among relative clauses, in particular, a group of appositive relative clauses. The relevant examples are presented below. (9)

a. I, who am/*is tall, was forced to squeeze into that VW. b. He had the nerve to say that to me, who has/*have made him what he is today. (Akmajian 1970, 154)

According to Akmajian (1970), the nominative case on the pronoun induces a person agreement between the pronoun and the verb. The accusative case on the pronoun, however, invariably co-occurs with a 3rd person feature on the verb. Corpus-based research has shown that relative clauses, like cleft clauses, do show variation when it comes to agreement. In fact, variation in agreement patterns is much greater than that observed by Akmajian (1970). In the corpus-based research we were looking for relative clauses with the same agreement pairs reported in cleft clauses presented in (3)– (7) above. There are well-known features distinguishing restrictive from appositive relative clauses and yet, in the minority of cases, the corpus shows that they appear to be disregarded. For example, there are cases in which commas precede the restrictive complementizer that or cases in which personal pronouns seem to be modified by a RRC. Personal pronouns have unique referents, which is why no restrictive identification is necessary. In this chapter, we would like to assume that the relative clauses modifying personal pronouns are appositive in nature, with the exception of relatives containing personal pronouns modified by a determiner, as presented in (10) below. The interpretation of such examples can be argued to be devoid of the unique referent usually created by a personal pronoun. (10)

a. Someone acknowledged that inner me that is always smiling. b. … who is the me who is actually making that decision?

How Relative are Cleft Clauses in English it-Clefts?

63

In examples like the ones in (10), the verb always shows a 3rd person feature. Table 4-1 summarizes the results of the corpus-based research. Below the table there are example sentences embodying each pattern. Table 4-1. Agreement patterns in relative clauses4 5 a. me that am b. me who am c. me that is d. me who is e. I that am f. I who am g. I that is h. I who is

ARCs ?V V – V ?V ?V – ?V

(11)

me that am He has healed me, poor me that am unworthy, but Glory is to the Father, Son, and the Holy Ghost.

(12)

me who am Glaucus knew how kind he has always been to me who am poor and blind.

(13)

me who is I agree that President Obama did a terrific job as president of the United States that people like me who is not an American began to pay respect to the American people having such a brilliant president.

(14)

I that am If I find myself amongst the idolaters of America, shall I that am a Christian, though not in orders, think it a sin to preach Jesus Christ, till I have received orders from Rome?

(15)

I who am It is unthinkable that even I who am an engineer and well read did not know this fact.

4

V stands for present/acceptable, while a hyphen indicates either absence or inapplicability. A question mark put next to V signifies scarcity of occurrences. 5 It has to be underlined that the cleft clauses examined by Akmajian (1970), Sornicola (1988) and Mokrosz (2014) contained not only copular verbs.

64 (16)

Chapter Four I who is The parents of these children, some of whom were in their 30s and 40s are much more able to relate to their children’s lived experience than I, who is much closer in age, will ever be able to.

Pattern (a) from Table 4-1, exemplified in (11), is very rare among ARCs. One of the examples found in the corpus comes from a religious text while another one comes from a work that is a translation of a Latin novel that dates back to the 170s. A similar pattern with a relative pronoun, as in (12), appears to be much more popular. Pattern (c) can be found in relative clauses in which, in most of the hits, the pronoun is modified by the definite article the, which, in turn, can be argued to have an effect on the person of the embedded verb. No examples with an appositive interpretation were found. The insertion of the relative pronoun, as in (13), pattern (d), once again resulted in a greater number of hits. The next set of examples contains pronouns in the nominative case. Generally, these were much less popular than the ones with the accusative case. The pattern with a relative complementizer and a full agreement on the verb, as in (14), is very rare and, as was the case with the pattern in (11), it appears in religious texts. The frequency slightly increases when a relative complementizer is replaced with a relative pronoun, as in (15). In patterns (g) and (h), the verb shows a partial agreement. The corpus produced no hits containing pattern (g), i.e., the pattern with a relative complementizer. The change of the relative complementizer, as in (h) exemplified in (16), similarly to the cases mentioned above, contributed to an increase in the number of the hits found. Let us now examine reflexive pronouns in relatives. Table 4-2 presents different patterns and the information about their presence/absence in the corpus. The table is followed by examples illustrating existing patterns. Table 4-2. Agreement patterns with reflexive pronouns a. I who consider myself b. I who calls myself' c. I who (non 3rd person verb) himself d. I who (3rd person verb) himself e. me who calls myself f. me who prides himself g. me who (non 3rd person verb) himself/myself

ARCs V V – ? V ? –

How Relative are Cleft Clauses in English it-Clefts?

65

(17)

I who consider myself a brave writer, one unafraid to push boundaries…

(18)

I who calls myself ‘Silver’ now.

(19)

O you who wraps himself in clothing.

(20)

Here’s an article by me who calls myself ‘aam aadmi’…

(21)

… a massive source of shame for someone like me who prides himself on being able to count on one hand the number of times…

Between a verb and a reflexive pronoun in what has been assumed to be appositive relative clauses, we observe either a full agreement, as in (a, d, f) in Table 4-2 or a partial agreement (see b, e in Table 4-2, exemplified in (18) and (20), respectively).6 The corpus produced no hits in the case of patterns exemplified in (c) and (g). In an ungrammatical example, i.e., (c), the verb shows a non-3rd person and the reflexive pronoun a 3rd person with the personal pronoun in the main clause bearing the nominative case. When the personal pronoun shows the accusative case, as in (g) in Table 4-2, both the 1st and 3rd person on the reflexive pronoun turn out to be infelicitous.7 As for the relative complementizer, we found it only in examples with pattern (d). We have put question marks next to two patterns. Pattern (d) could be found with a 2nd person pronoun, as in (19), but not a 1st person pronoun. In sentence (21) with pattern (f), the pronoun me with the preposition like may be argued to remain in an adjunction relation to the pronoun someone. Thus, the 3rd person on the verb and the reflexive may in fact be related to someone preceding like.

6 As a reviewer rightly points out, reflexive pronouns are required to agree with their antecedents and not with verbs. Here, attention is drawn to the relation between verbs and reflexives because of the dependency that can be observed between them with the person feature on an antecedent remaining constant. This, in turn, is possible only in examples under examination in which a verb exhibits variation with regard to the value of the person feature. Later in the chapter, we adopt an analysis in which a set of ij-features is passed from an antecedent via the functional head v to a reflexive pronoun. This transmission is another argument in favour of a relation holding between a verb and a reflexive. 7 We would also like to present an example that is worth considering in some future research. Specifically, the reflexive bears the person of a nominal, here the pronoun I, which is not the c-commanding nominal closest to it. (i) I'm probably not the only person who finds myself envying him this, just a little bit.

66

Chapter Four

In this section, we have tried to demonstrate that relative clauses exhibit agreement patterns similar, if not identical, to cleft clauses, which can serve as another argument in favour of a common syntactic analysis for both types of constructions. The study produced more examples with the accusative case than with the nominative one on the personal pronoun. The rare occurrences of the relative complementizer that can be explained by reference to the nature of the heads of the studied clauses. Specifically, personal pronouns are not modified by restrictive relative clauses which allow the use of a relative complementizer.

3. An account of it-clefts Section 3 is devoted to a syntactic analysis of it-clefts that would be able to account for the agreement patterns outlined in section 2. The agreement patterns in question provide a sound basis for the claim that it-clefts and relative clauses share a common, derivative denominator. We will start with an outline of different analyses of relative clauses. Subsequently, we will propose an analysis of it-clefts that will be able to account for the agreement patterns found in it-clefts.

3.1. Relative analysis of cleft clauses Since cleft clauses are commonly analyzed as restrictive relative clauses (see, e.g., Schachter 1973; Hedberg 1990; Reeve 2011), the section starts with an outline of two analyses usually assigned to RRCs. One of the most distinctive features of appositive clauses is the fact that they can modify personal pronouns. This will be a springboard for the analysis of cleft clauses as ARCs. 3.1.1. Restrictive analysis of cleft clauses Restrictive relative clauses have received various syntactic analyses. For example, the raising analysis assumes the movement of a nominal phrase from the relative clause to the matrix clause (Brame 1968; Schachter 1973; Vergnaud 1985; Kayne 1994; Bianchi 1999; Safir 1999; de Vries 2002; Hornstein 2001; Bhatt 2002, among others). Under the matching analysis (Sauerland 1998; Citko 2001), there are two representatives of the same nominal head. One is nested in the relative clause and the other in the matrix clause. The former is deleted under identity and the latter is

How Relative are Cleft Clauses in English it-Clefts?

67

spelled-out. There are also scholars, e.g., Bhatt (2002), Hulsey and Sauerland (2006), who argue for the co-existence of both derivations. Exemplary structures illustrating raising and matching derivations are presented in (22) below. (22)

the book that John read a. Raising analysis DP

D

the

CP

NP

C‫މ‬

booki

that John read ti

b. Matching analysis DP

D

the

NP

NP

booki

CP

NP

booki

C‫މ‬

that John read ti (Hulsey and Sauerland 2006, 112)

Chapter Four

68

The arguments in favour of the raising analysis of it-clefts include binding, idiom chunks, quantifier scope and weak cross-over effects (see, e.g., Reeve 2011). All of them involve reconstruction. Specifically, the phrase in the clefted, i.e., immediately post-copular, position is interpreted in the position in which it originates, namely inside the cleft clause. The difference between the position at which an item is interpreted and its surface position is indicative of movement/raising. It is impossible to check how the it-clefts under investigation behave with regard to each of the abovementioned arguments as they are limited with regard to the type of clefted element, namely they feature only personal pronouns. Let us consider just one of the arguments in favour of the raising analysis, namely binding.8 In accordance with the binding Condition A, a reflexive has to be ccommanded by its binder within a local domain. By looking at (23), it can be concluded that me originates within the embedded clause, which, in turn, ensures a proper binding relation between the personal pronoun and the reflexive. (23)

It is me who likes myself.

Another argument in favour of the raising analysis of cleft clauses with clefted personal pronouns relates to the binding Condition B under which pronouns have to be free, i.e., they cannot be bound in a local domain. In (24), the proper name Mary and the pronoun she are in different clauses, thus, co-reference is possible. The same kind of co-reference appears to be infelicitous in the cleft in (24b). The lack of co-reference between the name and the pronoun in (24b) can be linked to the fact that the two items originate within the same clause.

8

Idioms with personal pronouns embedded in it-clefts constitute an area of research that deserves a separate investigation. As can be seen in (i) below, the first person reflexive pronoun does not appear to be grammatical. The contrastive interpretation carried by it-clefts entails the third person on the reflexive, which is valid not only with regard to a 1st person personal pronoun but also a 2nd person personal pronoun. In the case of clefts which do not involve idioms, as in (ii), 1st as well as 3rd person reflexives are possible. (i) It is me who is beside *myself/himself (and not you (who is beside *myself/himself)). (ii) It is me who likes himself/myself.

How Relative are Cleft Clauses in English it-Clefts? (24)

69

a. Maryi thinks shei is clever. b. * It is Maryi that shei likes.

The pieces of evidence against the raising analysis, which, in turn, support the matching analysis, include anti-connectivity effects such as the one presented in (25) and (26). (25)

Billi asked Sue to wash *himselfi.

(26)

It was himselfi that Billi asked Sue to wash. (Pinkham and Hankamer 1975, 431)

According to Pinkham and Hankamer (1975), the licit co-reference between the pronoun himself and the proper name Bill in (26), but not in (25), argues against the movement of the clefted element from within the embedded clause. It is possible, however, that himself in (26) functions as an emphatic pronoun and originates within the same phrase as Bill and not wash. Taking into consideration the evidence connected with binding, it appears that the raising analysis fares better than the matching analysis. 3.1.2. Appositive analysis of cleft clauses In the case of clefted personal pronouns, restrictive relative clauses should be excluded on a priori grounds. Personal pronouns as well as proper names have unique referents, which is why they can be modified only by appositive relative clauses. To that end, the restrictive analysis of cleft clauses with clefted personal pronouns should be abandoned as well. In order to verify that claim, some pieces of evidence in favour of the appositive status of cleft clauses will be confronted with the data on itclefts. Firstly, we will look at the semantics of appositives. According to Chierchia and McConnell-Ginet (1990), appositive relative clauses entail backgrounded assertions and not presuppositions because their truth is not taken for granted. Thus, the sentence with an appositive relative clause in (28a), which carries backgrounded information, can follow sentence (27), but the pseudo-cleft in (28b), which establishes that Jill lost something, is an infelicitous continuation. (27)

Let me tell you about Jill Jensen, a woman I met while flying from Ithaca to New York last week.

70 (28)

Chapter Four a. Jill, who lost something on the flight from Ithaca to New York, likes to travel by train. b. What Jill lost on the flight from Ithaca to New York was her new flute. (Chierchia and McConnell-Ginet 1990, 282)

It-clefts, however, carry the presupposition of existence together with an exhaustive interpretation (see, e.g., Chierchia and McConnell-Ginet 1990; Rooth 1999). Thus, sentence (29a) carries the presupposition articulated in (29b). (29)

a. It was me who opened the door to the porch. b. Someone opened the door to the porch. (Chierchia and McConnell-Ginet 1990, 314)

Another semantic difference between ARCs and CCs concerns the fact that relative pronouns in ARCs are not semantically vacuous. In other words, they are of type , while the appositive clause is of type , i.e., it is a sentence (see, e.g., Sells 1985; Del Gobbo 2002). As for cleft clauses, Hedberg (2000, 917) maintains that they are of type . No less important is the fact that appositive relative clauses do not cooccur with the relative complementizer, which, however, is used interchangeably with relative pronouns in it-clefts. From a syntactic point of view, ARCs and CCs appear to be more alike as both cannot stack in contrast to restrictive relative clauses. (30)

a. the man who came to dinner who hated lox. b. * the man, who came to dinner, who hated lox.

RRC ARC (de Vries 2002, 197)

If cleft clauses co-occurring with clefted pronouns were restrictive relative clauses, it should be possible to find examples like the one in (31) grammatical. (31)

* It is me who likes flowers who did it.

In the it-cleft in (31), only the first clause restricts, as opposed to what can be observed in a sentence with a noun modified by two restrictives, as in (30a). Since personal pronouns cannot be modified by restrictives, the two alleged RRCs in (31) yield an illicit result. Optionally, the first restrictive in (31) could be replaced by an appositive, as in (32). (32)

It is me, who likes flowers, who did it.

How Relative are Cleft Clauses in English it-Clefts?

71

Since appositives do not stack and cannot precede restrictives (see Jackendoff 1977, 171), it can be concluded that cleft clauses with clefted personal pronouns are neither appositive nor restrictive. There are more features shared only by CCs and ARCs. Firstly, both are optional in contrast to restrictives. As reported by Emonds (1979), citing Milner (1973) and Ogle (1974), there is a certain group of sentential adverbs that can appear inside appositive relative clauses but not restrictive relative clauses, as presented in (33). (33)

a. The boys, who have frankly lost their case, should give up. b. * The boys that have frankly lost their case should give up. (Emonds 1979, 239)

Adverbs such as frankly or apparently can be found only in main clauses. This observation induces Emonds (1979) to assign the status of main clauses to appositive clauses. The corpus produces hits with a sentential adverb such as apparently inside a cleft clause, as in (34).9 (34)

It was Rafa who was apparently going to knock us…

The sentence in (34) may, again, point to an appositive nature of cleft clauses with clefted pronouns. As an interim summary, it can be concluded that cleft clauses with clefted personal pronouns resemble, to a certain extent, both restrictive and appositive relative clauses and, thus, cannot be unanimously classified as one of these two types.

3.2. Non-relative it-clefts It appears that any relative analysis, either strictly restrictive or strictly appositive, turns out to be problematic and cannot account for the behaviour of it-clefts with clefted personal pronouns. We would like to propose an analysis in which the clefted element originates inside the cleft clause and undergoes A-bar movement. The place of origin of the clefted pronoun is motivated by semantics and the richness of agreement patterns.

9

It has to be admitted that no hits contained clefted personal pronouns.

Chapter Four

72

Similarly to it-clefts, sentences with a leftward moved nominal phrase, as in (35b), entail presupposition, thus, as opposed to appositives, they cannot follow the sentence in (35a). (35)

a. Let me tell you something about Jill Jensen, a woman I met while flying from Ithaca to New York last week. b. JILL lost something on the flight from Ithaca to New York.

Since raising analysis applies to restrictive relative clauses, it can be argued that RRCs also entail presupposition. This observation is made, for example, by Del Gobbo (2002). The sentence in (36) cannot follow the utterance in (35a). (36)

The woman that lost her luggage on the flight from Ithaca to New York was pretty upset. (Del Gobbo 2002, 180)

One could conclude that, semantically, cleft clauses are like restrictive relative clauses. Yet, a set of sentences originally provided by Reeve (2011) in favour of a matching analysis may testify against such a claim. Specifically, the pronoun me in (37) can refer to I only when it appears in a different clause, as in (37b), and does not undergo reconstruction. (37)

a. I dislike myself/*me. b. It’s myself/me that I dislike. (Reeve 2011, 165)

Reeve (2011) also draws attention to an example such as (38) with a contrastively focused pronoun. (38)

ME, I dislike, not JOHN. (Reeve 2011, 165)

In sentences such as (38), it is generally agreed that movement takes place in contrast to what can be argued for in (37b) above. On the basis of examples like (38), it can be concluded that a similar movement is part of the derivation of it-clefts with clefted personal pronouns such as (37b). Another argument in favour of movement in the it-clefts under investigation is the variety of agreement patterns they exhibit. The analysis proposed here tries to explain the way the value of the person feature on the clefted pronoun in the matrix clause appears on the reflexive pronoun or the verb in the embedded clause. We would like to assume that both the personal pronoun with valued person and number features and the relative

How Relative are Cleft Clauses in English it-Clefts?

73

pronoun with a 3rd person feature and an unvalued number originate within the same DP (DP2 in (39)). As a result of their merge, the head D bears two values for the person feature and one for a number feature. Here, after Kratzer (2009), we would like to argue that a 3rd person can be considered a non-person, which allows for the co-existence of apparently different values of a person feature. Kratzer (2009) also maintains that bound pronouns, such as reflexives, which are born with unvalued ij-features, receive their values via functional heads such as v. Thus, we have to assume that, for pronouns to be clefted, they must originate inside the cleft clause where they can pass their ij-features via functional v to the reflexive. The variety of agreement patterns is linked to the idea of the markedness of features. According to Kratzer (2009), a person feature is marked on verbs while a non-person, i.e., a 3rd person, is marked on nominal expressions. Different degrees of markedness can be the cause of the variation in agreement patterns in CCs.

Chapter Four

74 (39) …

…

CP

DP2j

DP1

D

me

who

C‫މ‬

C

TP

Spec,TP

tj

T‫މ‬

T

vP

Spec,vP

tj

v‫މ‬

v

likesi

VP

V

DP

ti

himself/ myself

Importantly, the analysis proposed above can account for the fact that a cleft clause has a different interpretation from an appositive relative clause modifying the pronoun, as in (32). In other words, a cleft clause and an

How Relative are Cleft Clauses in English it-Clefts?

75

appositive relative clause do not form a relation accompanying stacking. In order to present such a relation, we would follow the proposal made by Griffiths and de Vries (2013, 337), who assume that the relation between an appositive relative clause and a given head noun is mediated by a functional head Par which heads the Parenthetic Projection (ParP). ParP appears in the place of the nominal phrase which is inserted in the Specifier of ParP. The appositive relative clause in (32) would be positioned within DP2 and the relevant structure would look as follows: (40)

…

…

CP

DP2

ParP

DP1

me

D

Par‫މ‬

Par

C‫މ‬

who

C

TP

…

ARC

who likes flowers

4. Conclusion The chapter aspired to be another work in a series confirming the status of cleft clauses as relative clauses. Despite the fact that both types of constructions show great similarity when it comes to agreement patterns,

76

Chapter Four

an attempt to assign an analysis of one type of relative clause to CCs turned out to be infelicitous. Even though personal pronouns are traditionally modified by appositive relative clauses, the semantics of appositives did not match the semantics of it-clefts. The interpretive discrepancy between cleft clauses and appositive relative clauses forced us to abandon an appositive analysis of it-clefts. Eventually, the binding facts, as well as the possibility of different person features on a reflexive pronoun and a verb inside the embedded clause, induced us to adopt an analysis with the movement of the clefted item from within the embedded clause.

References Akmajian, Adrian. 1970. “On Deriving Cleft Sentences from Pseudo-Cleft Sentences.” Linguistic Inquiry 1:149–68. Bhatt, Rajesh. 2002. “The Raising Analysis of Relative Clauses: Evidence from Adjectival Modifiers.” Natural Language Semantics 10:43–90. Bianchi, Valentina. 1999. Consequences of Antisymmetry: Headed Relative Clauses. Berlin: Mouton de Gruyter. Brame, Michael. 1968. “A New Analysis of the Relative Clause: Evidence for an Interpretive Theory.” Ms. MIT. Chierchia, Gennaro, and Sally McConnell-Ginet. 1990. Meaning and Grammar: An Introduction to Semantics. Cambridge, MA: MIT Press. Citko, Barbara. 2001. “Deletion Under Identity in Relative Clauses.” In Proceedings of North East Linguistics Society 31, edited by Minjoo Kim and Uri Strauss, 131–45. Amherst: GLSA of the University of Massachusetts. Davies, Mark. 2013. “Corpus of Global Web-Based English: 1.9 Billion Words from Speakers in 20 Countries.” Available at http://corpus2.byu.edu/glowbe/. Del Gobbo, Francesca. 2002. “Appositives and Chinese Relative Clauses.” In CLS 38: The Main Session. Papers from the 38th Meeting of the Chicago Linguistic Society, edited by Mary Andronis, Erin Debenport, Anne Pycha, and Keiko Yoshimura, 1:175–90. Chicago: Chicago Linguistic Society. Emonds, Joseph. 1979. “Appositive Relatives Have no Properties.” Linguistics Inquiry 10(2):211–43. Griffiths, James, and Mark de Vries. 2013. “The Syntactic Integration of Appositives: Evidence from Fragments and Ellipsis.” Linguistic Inquiry 44(2):332–44.

How Relative are Cleft Clauses in English it-Clefts?

77

Hedberg, Nancy. 1990. “Discourse Pragmatics and Cleft Sentences in English.” PhD diss., University of Minnesota. —. 2000. “The Referential Status of Clefts.” Language 76:891–920. Hornstein, Norbert. 2001. Move! A Minimalist Theory of Construal. Oxford: Wiley-Blackwell. Hulsey, Sarah, and Uli Sauerland. 2006. “Sorting out Relative Clauses.” Natural Language Semantics 14:111–37. Jackendoff, Ray S. 1977. X' Syntax. Cambridge, MA: MIT Press. Kayne, Richard S. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Kratzer, Angelika. 2009. “Making a Pronoun. Fake Indexicals as Windows into the Properties of Pronouns.” Linguistic Inquiry 40(2):187–237. Milner, Jean-Claude. 1973. Arguments Linguistiques. Paris: Mame. Mokrosz, Ewelina. 2014. “Agreement Patterns in It-Clefts: A Minimalist Account.” In Proceedings of the First Postgraduate and Academic Researchers in Linguistics at York, edited by Theodora Lee, 1:63–73. York: The University of York. Ogle, Richard Arthur. 1974. “Natural Order and Dislocated Syntax. An Essay in the History of Linguistic Ideas.” PhD diss., University of California, Los Angeles. Pinkham, Jessie, and Jorge Hankamer. 1975. “Deep and Shallow Clefts.” In Papers from the Eleventh Regional Meeting of the Chicago Linguistic Society, edited by Robin E. Grossman, L. James San, and Timothy J. Vance, 429–50. Chicago: Chicago Linguistics Society. Reeve, Matthew. 2011. “The Syntactic Structure of English Clefts.” Lingua 121:142–71. Rooth, Mats E. 1999. “Association with Focus or Association with Presupposition?” In Focus: Linguistic, Cognitive, and Computational Perspectives, edited by Peter Bosch and Rob van der Sandt, 232–44. Cambridge: CUP. Safir, Ken. 1999. “Vehicle Change and Reconstruction in A-bar Chains.” Linguistic Inquiry 30:587–620. Sauerland, Uli. 1998. “The Meaning of Chains.” PhD diss., MIT. Schachter, Paul. 1973. “Focus and Relativization.” Language 49:19–46. Sells, Peter. 1985. “Restrictive and Non-Restrictive Modification.” CSLI Technical Report CSLI-85-28. Stanford, California. Sornicola, Rosanna. 1988. “It-Clefts and Wh-Clefts: Two Awkward Sentence Types.” Linguistics 24:343–79. Vergnaud, Jean-Roger. 1985. Dépendances et Niveaux de Répresentation en Syntaxe. Amsterdam: John Benjamins.

78

Chapter Four

Vries, Mark de. 2002. “The Syntax of Relativization.” PhD diss., University of Amsterdam.

CHAPTER FIVE A MINIMALIST APPROACH TO DOUBLE OBJECT CONSTRUCTIONS IN GERMAN: THEIR SYNTACTIC REPRESENTATION AND PASSIVISATION POSSIBILITIES ALEKSANDRA BARTCZAK-MESZYēSKA

1. Introduction The aim of this chapter is to discuss two issues: the syntactic structure of the Double Object Construction (DOC) in German and the passivisation processes that the objects in German DOCs undergo. This analysis of DOCs in German focuses on three types of construction with the following object patterns: dative-accusative, accusative-dative and double accusative, as well as structures where one of the objects is realised by a Prepositional Phrase (PP). Our analysis of the syntactic structure of German DOCs is not limited to a mere discussion of the projections constituting DOCs but it also involves the syntactic operations which bring about feature valuation. In our account, we utilise McFaddden’s (2004) and Citko’s (2011) suggestions. This discussion is carried out within the minimalist framework of Chomsky (2007, 2008). This chapter consists of three sections. Section 2 is a presentation of the relevant data. Section 3 discusses the underlying syntactic structure of each type of DOC. The syntactic operations affecting the syntactic objects constituting each DOC, such as Case and ij-feature valuation, are also analysed in detail in this section. The aim of section 4 is to characterise the passivisation patterns that German DOCs exhibit. The chapter ends with our conclusions.

80

Chapter Five

2. German DOC – the data Since German has a relatively rich system of morphological marking (especially compared to English), it is not surprising that there are several types of acceptable DOCs with various object case patterns. The discussion here is limited to the three most common types whose underlying structure and feature valuation operations seem to be quite problematic to account for.

2.1. Dative-accusative DOC (Dat-Acc DOC) The dative-accusative DOC seems to be the default ditransitive structure in German. One of its most important properties is its strong tendency to allow only animate nominals as datives. The verb classes surfacing in this construction are highly productive, as newly coined verbs (e.g., bebauen ‘cover sth. with buildings,’ bebaumen ‘cover sth. with trees’) or borrowed lexical items (such as mailen ‘to mail,’ faxen ‘to fax’) are most likely to require a double object complementation of this type, as illustrated by the following examples. (1)

a. Wir werden Ihnen den Hof bebaumen. we will you-DAT the courtyard-ACC betree ‘We will plant the courtyard with trees for you.’ b. Er hat uns die Angaben gefaxt. he has us-DAT the data-ACC faxed ‘He faxed us the data.’ (modified from McFadden 2004, 110)

The accusative-genitive object sequence (cf. (2a)), which is nowadays perceived as archaic, seems to have been replaced by the dative-accusative word order, as in (2b).1

1

There is a general tendency in German (especially spoken) to replace the genitive with the dative after prepositions traditionally assigning the genitive to their complements (Helbig and Buscha 2001), e.g., trotz des Regens-GEN – trotz dem Regen-DAT ‘despite the rain,’ während des Frühstücks-GEN – während dem Frühstück-DAT ‘during breakfast.’ A reverse process has taken place in Greek where the dative has merged with the genitive (e.g., Anagnostopoulou 2003).

A Minimalist Approach to Double Object Constructions in German (2)

a. jemanden seines Wunsches gewähren wish-GEN grant someone-ACC his ‘grant someone his wish’

81

(archaic)

b. jemandem seine Bitte gewähren (modern) someone-DAT his request-ACC grant ‘grant someone his request’ (modified from McFadden 2004, 111)

The objects in this construction can be freely realized by pronouns, either one at a time or together. (3)

a. Hans gab dem Lehrer einen Brief. John gave the teacher-DAT a letter-ACC ‘John gave the teacher a letter.’ b. Hans gab ihm einen Brief. John gave him-DAT a letter-ACC ‘John gave him a letter.’ c. Hans gab es dem Lehrer. John gave it-ACC the teacher-DAT ‘John gave it to the teacher.’ d. Hans gab es ihm. John gave it-ACC him-DAT ‘John gave it to him.’

The Indirect Object (IO) can also be realized by a Prepositional Phrase (PP) with the prepositions für ‘for’ or zu ‘to.’ (4)

Hans gab einen Brief für den Lehrer/ zu dem John gave a letter-ACC for the teacher-ACC/ to the Lehrer. teacher-DAT ‘John gave a letter for the teacher/ to the teacher.’

2.2. Accusative-dative DOC (Acc-Dat DOC) This is the basic sequence for a group of verbs which are significantly less numerous than Dat-Acc verbs, e.g., aussetzen ‘expose,’ unterziehen ‘subject to,’ etc.

82 (5)

Chapter Five a. Die Mutter hat ihre Kinder der Kälte ausgesetzt. the mother has her children-ACC the cold-DAT exposed ‘The mother exposed her children to the cold.’ b. Der Arzt hat den Patienten einer Operation unterzogen. surgery-DAT subjected to the doctor has the patient-ACC a ‘The doctor subjected his patient to a surgery.’

In contrast to the previously discussed construction, the Acc-Dat DOC only allows one of its objects to be realised by a pronoun – instances of this type of DOC with two pronouns are ungrammatical. (6)

a. Der Arzt hat den Patienten einer Operation unterzogen. the doctor has the patient-ACC a surgery-DAT subjected ‘The doctor subjected the patient to a surgery.’ b. Der Arzt hat ihn einer Operation unterzogen. the doctor has him-ACC a surgery-DAT subjected ‘The doctor subjected him to a surgery.’ c. Der Arzt hat ihr den Patienten unterzogen. the doctor has her-DAT the patient-ACC subjected ‘The doctor subjected a patient to her.’ d. * Der Arzt hat ihn ihr unterzogen. the doctor has him-ACC her-DAT subjected ‘The doctor subjected him to her.’

Additionally, none of the objects can be substituted by a PP, as in (7). (7)

Der Arzt hat den Patienten *zu einer/ *für eine Operation for a surgery-DAT the doctor has the patient-ACC to a/ unterzogen. subjected ‘The doctor subjected the patient to a surgery.’

2.3. Double accusative DOC (Acc-Acc DOC) There is yet another type of double object complementation that needs to be discussed here, namely the structure where both objects are assigned the accusative case. Verbs that require double accusative objects include, among others, lehren ‘to teach,’ abfragen ‘to question,’ kosten ‘to cost,’ schimpfen ‘to swear,’ etc.

A Minimalist Approach to Double Object Constructions in German (8)

83

a. Der Junge hat den Studenten den Trick gelehrt. the boy-NOM has the student-ACC the trick-ACC taught ‘The boy taught the student the trick.’ b. Die Tasche kostete mich einen Euro. me-ACC a euro-ACC the bag-NOM cost ‘The bag cost me one euro.’

The Accusative IO cannot be realized by a PP, which makes Acc-Acc structures similar to Acc-Dat DOCs and different from Dat-Acc ones which permit IO PP substitution. (9)

* Der Junge hat den Trick für den Studenten/ zu dem the boy-NOM has the trick-ACC for the student-ACC/ to the Studenten gelehrt. student-DAT taught ‘* The boy taught the trick for the student/ to the student.’

As we can observe, the ability to substitute one object (IO) with a PP is what makes the Dat-Acc type of DOC in German distinctive. Such a PP is usually introduced by either the Preposition zu ‘to,’ assigning the dative to its object, or für ‘for,’ assigning the Accusative to the Determiner Phrase (DP) in its domain. However, a ditransitive verb complementation with a nominal object and the other object realized as a PP headed by another P is also well attested in German. This is the issue we would like to briefly discuss in the next section.

2.4.

Ditransitive complementation (DC) with the Prepositional Phrase (PP)

Some of the PPs in German DOCs are interchangeable with objects realized by DPs or pronouns, as illustrated in (10) below, but there are also instances of ditransitive complementation with a PP where the PP cannot be omitted or changed because this results in ungrammaticality, as in (11b). (10)

a. Peter schrieb dem Jungen den Brief. Peter wrote the boy-DAT the letter-ACC ‘Peter wrote the boy the letter.’

84

Chapter Five b. Peter schrieb den Brief an den Jungen/ für den Jungen. Peter wrote the letter-ACC to the boy-ACC/ for the boy-ACC ‘Peter wrote a letter to the boy/for the boy.’

(11)

seine Ferien]. a. Er schreibt seinem Vater [PP über he writes his father-DAT about his holidays-ACC ‘He writes his father about his holidays.’ b. * Er schreibt seinem Vater seine Ferien-ACC he writes his father-DAT his holidays-ACC ‘He writes his father about his holidays.’

Having briefly introduced the four most common types of ditransitive complementation, let us now turn to a detailed discussion of the syntactic characteristics of German DOCs.

3. The syntactic structure of DOCs in German Before discussing the actual syntactic representations of DOCs in German, we have to consider one issue, namely, whether the Dat-Acc and Acc-Dat DOCs constitute separate phenomena or are just different surface realizations of one construction. Whereas the derivational proposals (e.g., Grewendorf 1988, 2002; Müller 1995, 1997; Müller and Sternefeld 1994; Sternefeld and Featherston 2002, among many others) mostly concern variations in the object sequence of the Dat-Acc DOC type (the theta roles of the dative and the accusative objects remain identical in all the patterns, the dative mainly being Recipient or Benefactive, and the accusative being Theme), McFadden (2004) provides an analysis of Dat-Acc and Acc-Dat DOCs (the theta role of the accusative in the Acc-Dat DOC is usually Patient) and, taking into consideration processes such as the constituency test based on Topicalisation and the acceptable positions of the negation nicht, concludes that these two types of DOC require, in fact, two different underlying syntactic representations. In our proposal, we modify McFadden’s (2004) representation of the Acc-Dat DOC since his theory does not seem to consider the object fronting phenomena.2 2

Surely, the syntactic representations of the Dat-Acc and Acc-Dat DOCs in German need to reflect their different syntactic properties. However, German dative and accusative objects can freely move to the sentence initial position in both Dat-Acc and Acc-Dat DOCs, which entails that the syntactic representations of both types of DOCs must exhibit certain structural similarities to allow object

A Minimalist Approach to Double Object Constructions in German

85

3.1. Dat-Acc DOC Following numerous approaches to this construction (see, e.g., Georgala 2011 for a discussion of the Applicative in German, among others), we adopt the most popular representation of the Dat-Acc DOC with the Applicative Phrase (Appl): a High Applicative where the dative is Benefactive (illustrated in (12) below) and a Low one with the dative as Recipient, as in (13). (12)

How can the structure in (12) be obtained? We assume that the computation and feature valuation happens in the following manner: in a vP phase, Verb (V) merges with a Direct Object (DO) and together they build the Verb Phrase (VP). This VP merges with the Applicative. The IO merges in the Specifier (Spec) of the Applicative and has its inherent dative case valued at first Merge in this position. The Applicative merges with the phase head, v, with the External Argument (EA) in the Spec,vP position. The phase head v is equipped with ij-features, as in Chomsky’s (2008) model, and it hands these features down onto V in the course of feature inheritance. Here, we would like to follow the proposal made by Gallego (2010), according to which all projections contained within one phase share the same features. Under this scenario, the phase head v shares its ij-features with all the heads within the vP phase, namely, the Applicative and V. As a result of feature sharing, both the Appl head and V are equipped with ij-features and, therefore, either can enter Agree with the DO and value its Case as accusative (the choice of the Probe is movement. This observation forces certain modifications upon McFadden’s (2004) proposal (for more details concerning the relation between object displacement phenomena and syntactic representations of German DOCs, see BartczakMeszyĔska 2013).

86

Chapter Five

determined by Locality constraints – the feature-valuation relation happens between the closest available Probe and Goal). Therefore, in accordance with Locality, it is the Applicative which values the case of the DO. Let us now consider the structure of the Dat-Acc DOC with a Low Applicative, represented in (13) below. (13)

In order to derive a structure with a Low Applicative, the DO merges directly with an Applicative which also merges with the IO in its Specifier position. Whereas the valuation of the inherent dative on the IO in the first Merge position seems quite straightforward, the valuation of the structural accusative on the DO is more complicated. The structural accusative on the DO is valued in a similar way to the one proposed for the High Applicative above. We assume that the process of feature-sharing occurs – the phase head shares its ij-features with all the other heads in the vP phase, namely V and Appl. Consequently, since the Applicative is the closest to the DO, it undergoes Agree with the DO and values its case as accusative, just like in the structure with a High Applicative in (12). The structural case valuation does not affect the inherently marked IO (cf. Jeong 2007; this is where our analysis departs from the one proposed by Citko 2011),3 the inherent Dative does not count as an intervener for syntactic operations concerning structural Case valuation.4 3

Citko (2011) suggests that the dative, especially quirky, consists of both inherent and structural properties, therefore, it does intervene in structural relations. 4 The fact that the dative IO is not an intervener here is additionally supported by the observation that raising across the dative is possible in German (as illustrated in (i) below). ti langeweilt zu sein. (i) Helmuti scheint mir Helmut seems to me-DAT bored to be ‘It seems to me that Helmut is bored.’

A Minimalist Approach to Double Object Constructions in German

87

3.2. Acc-Dat DOC The derivation of the Acc-Dat DOC happens in a manner significantly different from the one for the Dat-Acc DOC, which is another argument for regarding these two structures as independent phenomena. It is represented in (14) below. (14)

The first step here is the incorporation of the preposition to the verb, the two of which build a complex head V together. This element merges with the dative object. The inherent dative case on the IO is valued at first Merge. The accusative DP enters the derivation and, together with the previously merged items, it constitutes a VP. This DP has its accusative valued by the phase head v directly, when these two undergo Agree. Finally, the EA merges in the Spec,vP position.

3.3. Double accusative DOC This type of ditransitive complementation requires a separate discussion, since it is the only type of German DOC which builds the symmetric passive (as discussed in section 4.3). This fact leads us to the conclusion that the case of both objects is structural.5 Hence, we need a different mechanism to value cases on both objects, and this is why we apply the approach proposed by Citko (2011) for (mainly) Polish to the German double accusative. She claims that constructions where two cases are structural require two active Probes and this proposal introduces changes in the properties of Applicative phrases (they are usually associated with 5

We need to remark here that instances of the inherent accusative do occur in German, in infrequent structures like mich friert ‘I am cold’ where the reflexive mich ‘myself’ occurs in the accusative.

88

Chapter Five

the inherent dative case) since there can be only two sources of structural case, as in (15) below: one is the phase head v, the other has to be the Applicative. The Goals of these Probes are then different from the ones in the traditional approach, too. Due to locality constraints, the Applicative values the case of the DO and the phase head v values the case of the IO. The opposite situation is excluded: the Applicative would act as an intervener between v and the DO (also the IO would remain caseless). How does the derivation proceed then? The derivation of the double accusative DOC begins with the merger of the verb with the DO, as in the Dat-Acc DOC.6 The VP merges with the Applicative, the IO is merged in the Specifier of the Applicative and the phase head vP merges with the Applicative. The Applicative values the structural case on the DO and the phase head v values the structural case on the IO in the Spec,ApplP position. This derivation is presented in (15) below. (15)

This proposal for the Case valuation entails an even more fine-grained distinction among the Applicative phrases: some can value structural Case, becoming active Probes, some cannot and, hence, remain syntactically inactive. We need to emphasize here that the Applicative in the double accusative DOC is able to value the accusative from the start – it does not inherit this ability from the phase head, the process of feature sharing does not take place. Therefore, the Applicative in the double accusative DOC differs in this respect from the Applicative found in the Dat-Acc DOC, which inherits the ij-features of v and, hence, can value the accusative only after feature-sharing occurs (cf. (13) and (14)).

6

The fact that double accusative DOCs do not involve a literary transfer of possession supports the claim that the double Acc DOC instantiates a High Applicative.

A Minimalist Approach to Double Object Constructions in German

89

The variation in the properties of particular Applicative structures could then be employed to explain the interchangeability between the use of the dative and the accusative case on the IO which some German native speakers allow: Er lehrte den StudentenACC den TrickACC (as in (8a)) vs. Er lehrte dem StudentenDAT den TrickACC. We need not suggest two separate structures to account for this phenomenon, the fact that the speaker uses a different type of Applicative in these structures would suffice to explain the case change.

3.4. DC with PP In such structures, case valuation seems to be quite straightforward, the phase head v values the case on the DO in the course of Agree, whereas the Preposition values the Case on the DP within its domain, as represented in (16) below. Still, we need to discuss this process in greater detail. (16)

The derivation of the DC with a PP begins with the merger of a P with a DP. We follow the proposal of Pesetsky and Torrego (2007)7 and assume that a P possesses the ability to value Case. Hence, P values the Case of its object and together they form a PP. The PP merges with the verb and the DO is merged in the Spec,VP position. VP merges with the phase head v, which, by means of its ij-features, acts as a Probe and looks for an active Goal within its domain. The DO is active by virtue of having an unvalued 7

Pesetsky and Torrego (2007) regard P as a kind of T, able to value a structural case on the DP within its domain.

90

Chapter Five

uninterpretable Case feature. The Agree between v and the DO results in the valuation of the ij-features of v and the valuation of the Case feature of the DO as accusative. The EA merges in the Spec,vP position.

4. The computation of passive sentences with the DOC The aim of this section is to provide a step-by-step description of the derivation of passive sentences with the previously discussed types of DOC in German. As canonically assumed, passive sentences involve the change of case into the nominative (in the Nom-Acc language family) of an object that becomes a subject. In German, only accusative objects can become nominative subjects and value the EPP feature on T (that is, undergo A-movement).8 One issue remains to be discussed: the way this movement proceeds. If we adopt Chomsky’s (2007, 2008) approach to phases, a passive vP is a weak phase, unable to project an EA, therefore the object cannot move via the Spec,vP. Since a vP does not constitute a spell-out domain, the object, whose structural Case gets absorbed by the passive morphology, is not at risk of being trapped within a phase with its features unvalued. It can move directly to the TP projection to satisfy the EPP feature of T.9 On the other hand, if Legate’s (2003) or Richards’ (2004) approach to phases is preferred, where they assume that passive vP are phases, then the movement of the object would proceed via the Specifier of the passive vP. Let us discuss the derivations of the relevant passive sentences in detail.

4.1. Dat-Acc DOC Since this is the most common type of DOC in German, it is not surprising that this is the most often occurring kind of passive. Let us begin the discussion of this type of DOC with a presentation of the passivization of sentences with a High Applicative.

8 Again, as in the discussion of active sentences, we do not regard objects with inherent case as interveners for the relations concerning structural cases, following Jeong (2007). 9 This would be its final position in the derivation, as we assume, following, e.g., Wurmbrand (2006), that subject initial passive sentences are TPs.

A Minimalist Approach to Double Object Constructions in German (17)

91

Der Mantel wurde der Frau gehalten. the woman-DAT held the coat-NOM was ‘The coat was held for the woman.’

(18)

The derivation above proceeds in the following way: the DO merges with the verb and they constitute a VP, which merges with the Applicative Phrase, with the IO in the Spec,ApplP position. The IO has its inherent dative valued at first Merge. The ApplP merges with vP, but v cannot share its features with the Applicative since v’s probing abilities have been absorbed by the passive morphology. Consequently, the Applicative cannot value the Case of the DO. The DO seeks another Probe to have its Case feature valued – it enters into Agree with T, which values its Case as the nominative. The DO moves to Spec,TP to satisfy the EPP feature of T. Although the DO crosses the IO on its way to Spec,TP, no intervention effects occur. We assume, following Jeong (2007), that the IO with an inherent case does not intervene in relations valuing structural Case.10 Let us now focus on the following sentence with a Low Applicative:

10

Citko’s (2011) approach is justified in the languages she analyses, e.g., Icelandic with its quirky datives, but it seems to be unjustified in German where the dative never changes, except for one structure, the so called kriegen passive – in the kriegen passive the dative IO changes case and the accusative DO does not. However, since this is the only occurrence of this phenomenon, we have concluded that the case change must be induced by the properties of the kriegen structure and does not result from the nature of the Case itself. Hence, it does not seem appropriate from the minimalist perspective to propose the constant presence of a feature which is activated in only one particular configuration and does not seem to matter in all the others.

92 (19)

Chapter Five Der Ball wurde der Frau gegeben. the woman-DAT given the ball-NOM was ‘The ball was given to the woman.’

(20)

The derivation takes place in the following way: since the passive v has no ability to value Case (its ij-features have been absorbed by the passive morphology), the DO cannot be assigned Case in the vP and seeks another Probe outside vP to value its Case. Hence, the DO der Ball enters Agree with T, which values its case as the nominative and is attracted by the EPP feature of T to Spec,TP. Again, we assume here, following Jeong (2007), that inherent case does not intervene for structural Case valuation The DO moves to Spec,TP, either directly, if we adopt Chomsky’s (2007, 2008) proposal, which is what we have chosen here, or via the Specifier of vP, if Legate’s (2003) version is preferred.11 The verb movement occurs simultaneously. The auxiliary verb moves out of its base position v to T.

4.2. Acc-Dat DOC The accusative case of the DO in the active changes into the nominative in the passive, as in (21). The derivation of (21) is illustrated in (22).

11

We do not assume one proposal’s superiority over the other but adopt Chomsky’s version for the ease of exposition.

A Minimalist Approach to Double Object Constructions in German (21)

93

Der Junge wurde der Kälte ausgesetzt. the cold-DAT subjected the boy-NOM was ‘The boy was subjected to the cold.’

(22)

As in the active, the verb combines with the Preposition and together they merge with the IO. Since the passive morphology absorbs the case valuation property of v (ij-features), the DO undergoes Agree with T and moves to the Spec,TP to satisfy the EPP feature of T. The DO terminates its movement there.

4.3. Double Acc DOC This type of DOC allows the symmetric passive in German, which means that both the DO and the IO can become the subjects of passive sentences. The examples shown in (23) below, with either object passivised, are equally grammatical. However, since both objects have a structural Case, we expect intervention effects to occur – the object closer to the Probe should prevent relations between the Probe and the object more embedded in the structure. This observation makes the derivation of these sentences more complicated and requires a meticulous explanation. Let us begin the analysis of the symmetric passive in German with the examples in (23) below. (23)

a. Der Trick wurde den Studenten gelehrt. the student-ACC taught the trick-NOM was ‘The trick was taught to the student.’

(passivised DO)

94

Chapter Five b. Der Student wurde den Trick gelehrt. the trick-ACC taught the student-NOM was ‘The student was taught the trick.’

(passivised IO)

(24)

(25)

Here, since both v and the Applicative value the structural accusative, the passive morphology can affect the properties of either Probe. The situation is quite straightforward in (25) where no element intervenes between the Probe and the Goal – v loses its ability to value Case, therefore the IO enters Agree with T in-situ and moves to the Spec,TP, satisfying the EPP feature of T. The final step of the derivation is the movement of the finite verb form to T. The situation is more complicated in (24) where the passive morphology absorbs the ability of the Applicative to value the accusative Case. The Case on the DO is valued as

A Minimalist Approach to Double Object Constructions in German

95

the nominative by T. Since the Applicative is an independent Probe, we assume (following Citko 2011) that it also has the ability to project an additional Specifier position with the EPP feature, and this is exactly what happens here: when the properties of the Applicative are absorbed by the passive morphology, it projects another Specifer position. The DO moves to this position to satisfy the EPP feature of Appl. Now it is closer to the Probe than the IO and, when T seeks a nominal expression to satisfy its EPP feature, the DO moves to Spec,TP. This account is advantageous since it does not require locality violations – if the DO remained in its initial position, the IO would be closer to T and, although its accusative Case could be valued by v, it would be unable to move to T to satisfy its EPP and it would also block Agree between T and the DO.

4.4. DC with PP This structure is clearly very different from the others. Although the passive morphology absorbs the Case valuation properties of v, only the DO needs Case valuation by the Probe and it is the only expression to satisfy the EPP feature of T, the PP object remains in its original position. The representation in (27) illustrates the sequence of operations in this type of passivization. (26)

Der Brief wurde an den Sohn geschrieben. to the son-ACC written the letter-NOM was ‘The letter was written to the son.’

(27)

As we can observe in the example in (27), only the DP object can change into the subject of the passive sentence (undergoes Agree with T,

96

Chapter Five

which values the nominative case on the DO) and satisfy the EPP on T. As in the previously discussed types of derivations, the movement of the object is accompanied by a movement of the finite verb form.

5. Conclusion The aim of the chapter has been to discuss the syntactic structure of the most common types of DOC in German and to analyse the passivisation possibilities of these types of DOC. The structures presented in this chapter seem to be useful, not only when active sentences but also when passive sentences containing DOCs are subject to scrutiny. The syntactic properties of the Dat-Acc DOC are based on the distinction between the High and Low Applicative, as proposed for German by Georgala (2011), whereas the representation of the Acc-Dat DOC suggested here is an attempt to simultaneously account for the differences (as in McFadden 2004) between the Acc-Dat and the Dat-Acc DOC, as well as the similarities which these two types of DOC exhibit (object fronting). Our proposal for the structure of the double accusative DOC, a modified version of Citko’s (2011) account of the Polish DOC, seems to be a suitable one to account for the properties of both active and passive sentences.

References Anagnostopoulou, Elena. 2003. The Syntax of Ditransitives. Evidence from Clitics. Studies in Generative Grammar 54. Berlin and New York: Mouton de Gruyter. Bartczak-MeszyĔska, Aleksandra. 2013. “The Double Object Construction in English and German. Structural Position, Case Valuation and Movement of Objects.” PhD diss., John Paul II Catholic University of Lublin. Chomsky, Noam. 2007. “Approaching UG from Below.” In Interfaces + Recursion = Language?: Chomsky’s Minimalism and the View from Syntax-Semantics, edited by Uli Sauerland and Hans-Martin Gartner, 1–29. Berlin and New York: Mouton de Gruyter. —. 2008. “On Phases.” In Foundational Issues in Linguistic Theory. Essays in Honour of Jean-Roger Vergnaud, edited by Robert Freidin, Carlos P. Otero, and Maria Luisa Zubizaretta, 134–66. Cambridge, MA: MIT Press.

A Minimalist Approach to Double Object Constructions in German

97

Citko, Barbara. 2011. Symmetry in Syntax. Merge, Move and Labels. Cambridge: CUP. Gallego, Angel. 2010. Phase Theory. Amsterdam: John Benjamins. Georgala, Effi. 2011. “Why German is not an Exception to the Universal Base Order of Double Object Constructions.” In Proceedings of the 28th West Coast Conference on Formal Linguistics (WCCFL 28), edited by Mary Byram Washburn, Katherine McKinney-Bock, Erika Varis, Ann Sawyer, and Barbara Tomaszewicz, 96–105. Sommerville: Cascadilla Proceedings Project. Grewendorf, Günther. 1988. Aspekte der Deutschen Syntax. Tübingen: Narr. —. 2002. Minimalistische Syntax. Tübingen: A. Francke Verlag. Helbig, Gerhard, and Joachim Buscha. 2001. Deutsche Grammatik. Berlin, München: Langenscheidt KG. Jeong, Youngmi. 2007. “The Landscape of Applicatives.” PhD diss., University of Maryland. Legate, Julie. 2003. “Some Interface Properties of the Phase.” Linguistic Inquiry 34:506–16. McFadden, Thomas. 2004. “The Position of Morphological Case in the Derivation. A Study on the Syntax–Morphology Interface.” PhD diss., University of Pennsylvania. Müller, Gereon. 1995. A-bar Syntax. Berlin: Mouton de Gruyter. —. 1997. “Parallel Movement.” Ms., University of Stuttgart. Müller, Gereon, and Wolfgang Sternefeld. 1994. “Scrambling as A-bar Movement.” In Studies on Scrambling. Movement and Non-Movement Approaches to Free Word Order Phenomena, edited by Norbert Corver and Henk van Riemsdijk, 331–85. Berlin and New York: Mouton de Gruyter. Pesetsky, David, and Ester Torrego. 2007. “The Syntax of Valuation and the Interpretability of Features.” In The Syntax of Valuation and the Interpretability of Features, edited by Simin Karimi, Vida Samiian, and Wendy K. Wilkins, 262–94. Amsterdam: John Benjamins. Richards, Marc. 2004. “Object Shift and Scrambling in North and West Germanic: A Case Study in Symmetrical Syntax.” PhD diss., University of Cambridge. Sternefeld, Wolfgang, and Sam Featherston. 2002. “The German Reciprocal ‘einander’ in Double Object Constructions.” In Arbeiten zur Reflexivierung, edited by Lutz Gunkel, Gereon Müller, and Gisela Zifonun, 239–66. Tübingen: Niemeyer. Wurmbrand, Susane. 2006. “Licensing Case.” Journal of Germanic Linguistics 18.3:175–236.

CHAPTER SIX DUAL INTERPRETATIONS OF OBJECT NEGWH-QUANTIFIERS (NEG-WHQ) IN CANTONESE: A FEATURE-BASED APPROACH MAN KI THEODORA LEE

1. Introduction In Cantonese, the combination of a negative morpheme and a wh-phrase can be used as a strong indefinite. This chapter is concerned with the features involved in these negative wh-quantifiers (neg-whQs). Much like wh-words in Mandarin Chinese, wh-words in Cantonese also give rise to existential interpretations, in the same way as indefinites (some- or any- or every-) in non-interrogative contexts besides interrogative interpretations.1 The literature (Cheng 1991, 1994; Huang 1982; Li 1992; Lin 1998) reports that Chinese wh-words need to be interpreted as indefinites under a licensing relationship with licensers. In order to obtain the existential interpretations, wh-words are required to appear in negative polarity contexts,2 non-factive contexts and contexts of tentativeness, inference (Cheng 1994) and the dou-quantification (Cheng 1995) context. In addition, some short forms of wh-expressions (e.g., bin) are used as 1

With dou-quantification, the wh-phrase matje ‘what’ is licensed as a universal quantifier in Cantonese. (i) Ngo matje dou zongji sik gaa. I what also like eat SP ‘I like to eat everything.’ 2 Like wh-phrases in Mandarin Chinese, wh-phrases in Cantonese are also licensed as Negative Polarity Items (NPI) in a negated context. (i) Ngo mou zongji bingo. I not like who ‘I don’t like anybody.’

Dual Interpretations of Object Neg-wh-Quantifiers

99

negative propositions rather than interrogatives. Cheng (1995) reports a few languages (e.g., Cantonese, Spanish, Korean, English, German, Japanese and Hebrew) in which wh-expressions are used as negative propositions rather than interrogatives, like since when in English, eti/ettehkhey ‘where/how’ in Korean, de dónde ‘of where’ in Spanish and bindou ‘where’ in Cantonese. Even though Mandarin Chinese and Cantonese share a lot of syntactic similarities, an object neg-whQ is an exception that does not appear in Mandarin Chinese. As a negative quantifier, an object neg-whQ undergoes overt raising and gives rise to both negative and existential presupposition interpretations. This chapter aims to report the overt raising phenomenon which results in a unique SOV word order in Cantonese and also dual interpretations, in a featurebased approach. In section 2, I propose an interpretable and strong [Quant] feature and a [uNeg] feature to account for the possible dual readings of object neg-whQs in Cantonese. I attempt to explain how the overt raising phenomenon is accounted for and each available interpretation is licensed in a feature-based approach. Section 3 supports my proposal by presenting data where Cantonese wh-expressions are used as negative presuppositions and briefly discusses Cantonese wh-words in a feature-based approach. Section 4 provides additional data where wh-words give rise to special existential interpretations in negative sentences. Section 5 summaries the proposed account.

2. The proposal This study proposes an interpretable and strong [Quant] feature and an uninterpretable feature [uNeg], which would account for the possible dual readings of object negative wh-quantifiers (neg-whQs) in Cantonese. Unlike other ordinary negative quantifiers (e.g., mouje ‘nothing,’ moujan ‘nobody’) that give rise to only a non-existential reading, neg-whQs give rise to either negative or existential readings, depending on the different licensers involved in the contexts. Morphologically, Cantonese neg-whQs consist of a negative morpheme mou with any wh-phrases, for example, mou-bingo (no-who), mou-matje (no-what) and mou-bindou (no-where). (1)

Neg-whQ e.g., mou-bingo, mou-bindou, mou-matje [Quant: strong] [uNeg]

Chapter Six

100

Empirical studies (Kratzer 1995; Potts 2000; Penka and von Stechow 2001) suggest that negative phrases are decomposed into negation and an existential/indefinite element. Neg-whQs are most likely equivalent to English negative quantifiers such as nobody, nothing and nowhere in semantic terms of the negative interpretation. Given that the canonical word order in Cantonese is SVO, objects follow the verb, as in (2), and so does a numeric NP, as in (3). On a par with ordinary negative quantifiers (negQs), neg-whQs are also a kind of strong quantifiers and observe a unique SOV word order, apart from their equivalent negative reading. A construction with an object negQ mou-je ‘nothing’ renders grammaticality in the SOV word order, as in (4). (2)

Ngo zungji fa. I like flower ‘I like flowers.’

(3)

Ngo sik-zo saam go pingguo. I eat-PFV3 three CL4 apple ‘I ate three apples.’

(4)

Ngo mou-je sik-guo. I no-thing eat-ASP5 ‘I ate nothing.’

An object neg-whQ construction also displays such a unique SOV word order. In addition, it not only gives rise to a non-existential or sentential negation interpretation, but also to an existential presupposition interpretation. In certain contexts, such as (5) below, there can be ambiguity due to the difference between a non-existential interpretation, as in (5a), and an existential presupposition interpretation, as in (5b). This neg-whQ is unique in Cantonese, and not found in Mandarin Chinese, despite the syntactic similarities between the two languages. The same combination of a negative morpheme and a wh-phrase in Mandarin Chinese neither appears as an object in canonical SVO nor in the unique SOV structure in (6). On the contrary, such a combination, appearing as the subject in a sentence, leads to grammaticality in Mandarin Chinese, as in (7). We, thus, focus on object neg-whQs in Cantonese.

3

PFV, perfective aspect. CL, classifier. 5 ASP, aspect marker. 4

Dual Interpretations of Object Neg-wh-Quantifiers (5)

Ngo mou-bindou soeng heoi. I no-where want go a. ‘I want to go to nowhere.’ (Lit. ‘I don’t want to go to anywhere.’) b. ‘I want to go to a few places.’

(6)

* Wo meiyou-shei xihuan. I no-who like * Wo xihuan meiyou-shei. I like no-who ‘I like nobody.’

(7)

Meiyou-shei xihuan ni. no-who like you ‘Nobody likes you.’

101

On the one hand, the existential reading is facilitated with the help of sentence final particles (SP) or a rising tone in contexts where exaggeration or emphasis is made. The literature on Cantonese sentence particles (Law 2002; Tang 1998) suggests that zaa, for example, is an indication of a restrictive focus “only.” (8)

Ngo mou-bindou soeng heoi zaa. I no-where want go SP ‘I want to go to only a few places.’

Similarly, negative quantifiers can imply an existential reading in English double negated contexts when stress applies. (9)

I don’t love NOBODY. (In other words, ‘I love SOMEBODY.’)

On the other hand, a negative reading is embedded in neg-whQs in cases where an existential reading is maintained in double negated contexts. (10)

Ngo mou-matje m zungji. I no-what not like ‘I don’t like nothing.’ (In other words, ‘I like something.’)

Neg-whQs, as indefinites, require an external licenser to allow for the full interpretations in the same way as wh-words do in Cantonese (or Mandarin Chinese). Lewis (1975) and Heim (1982) suggest that there is no inherent quantificational force within indefinites and, therefore, indefinites require a trigger to be licensed. For instance, Mandarin Chinese wh-words remain in-situ and require an external force to be licensed as interrogative.

Chapter Six

102

Aoun and Li (1993) suggest that question particles in Chinese are an overt Qu-marker belonging to an X0 category and they undergo optional raising, depending on the presence of a Qu-operator in LF. When a question particle is overt, it appears in the Spec of CP and creates such a licensing effect for the wh-word in-situ. When it is absent, however, the wh-word insitu can also be licensed as interrogative by raising the question operator in LF. Unlike wh-phrases in English, the morphological realization of the [+/–Q] feature of Cantonese wh-words does not rely on the wh-word itself but depends on elements in other parts of the sentence. In other words, an unvalued (uninterpretable) [Q] feature is involved in wh-words in Mandarin. Empirical studies (Aoun and Li 1993; Cheng 1991, 1994; Huang 1982; Lin 1998, 2004; Tsai 1994a, 1994b) suggest that wh-words in Mandarin can be interpreted as indefinites and, since wh-words in Cantonese function in a similar syntactic way, wh-words in Cantonese are proposed to have an unvalued [Quant] feature as well. The difference between the mechanisms of negative quantifiers in English and neg-whQs in Cantonese is parallel to those of wh-phrases in these two languages. While English negative quantifiers are internally negative with a [+Neg] feature, neg-whQs as indefinites require an external valuation for the [uNeg] feature inherited from the lexical ambiguity of mou in Cantonese. Table 6-1. Comparison of Cantonese neg-whQs, Cantonese ordinary negative quantifiers and English negative quantifiers in the object position

Quantifiers Syntactic features Word Order Interpretation

Cantonese Neg-quantifiers Neg-whQs e.g., moujie e.g., mou-matjie (‘nothing’) (‘no-who’) [Quant: strong] [Quant: strong] [+Neg] [uNeg] SOV SOV Sentential Sentential negation/ negation existential presupposition

English Neg-quantifiers e.g., nobody, no one [Quant: weak] [+Neg] SVO Sentential negation

My proposal here follows the structure of NegQP (Lee 2011) in that a neg-whQ appears as one constituent in terms of possible movements or extractions. Modifications are made to the NegQP in order to account for dual interpretations. I hereby propose a (Neg)QuantP which has an internal unpronounced quantifier operator Ø that carries a [Quant:Strong] feature as its head, and a negator mou in the Spec position that specifies the phrase with a [uNeg] feature and takes a wh-phrase as its complement (any DP

Dual Interpretations of Object Neg-wh-Quantifiers

103

for other non-existential quantifiers). Such a QuantP accounts for the internal structure of neg-whQs and captures the features involved, triggering the overt quantifier raising and the dual interpretations. The [Quant:Strong] feature in the invisible quantifier operator Ø marks the neg-whQ as a strong quantifier and forces it to undergo overt movement. Overt raising of neg-whQs is driven by the [uQuant] in Spec,vP under Agree. The underspecified [uNeg] feature triggers vague variation between negative and existential presuppositions. Hence, neg-whQs, as strong quantifiers and colloquial terms, have both non-existential and existential presupposition interpretations in Cantonese. This is shown in the following structure. (11)

(Neg)QuantP[Quant:Strong, uNeg] Mou[uNeg]

Quant‫މ‬

Quant Ø[Quant:Strong]

whP bingo[uQuant:_]

The QuantP with mou serves as a classifier. The internal quantifier operator has a quantifying force, after Merge ({mou{Ø, bingo}}), which accounts for the structure of neg-whQs and endows such strong quantifiers with properties of ambiguity. This ambiguity is possibly an outcome of combining the negative morpheme mou with wh-words in Cantonese, given that there is some sort of [+Neg] feature within wh-words. The short form of wh-words in Cantonese can imply negative presuppositions. This will be discussed in the next section, using data from Cantonese. The NegP is projected, with a covert negative operator Op炌 as its head, after Merge of an object neg-whQ in the derivation. I take Beghelli and Stowell’s (1997) location in the functional structure of the clause of the five QP-types into account, where neg-whQs “checks [+Neg] in Spec of NegP, under agreement with the Neg-operator in Neg0.” This covert negative operator carries an [iNeg] feature and creates a licensing relationship with a neg-whQ in the base-generated object position of an SVO structure. The full interpretation of the neg-whQ construction is accounted for under the following assumptions for the two scenarios:

Chapter Six

104 (12)

The two scenarios for the pre-Spell-out mechanisms of neg-whQs: a. Obligatory quantifier raising (QR) takes place before Spell-Out and decomposition for internally complex QPs must follow. b. Decomposition for internally complex QPs takes place before Spell Out and overt movement must follow.

This chapter focuses mainly on the first scenario to capture the dual interpretations of object neg-whQs in an SOV structure. Neg-whQs, as strong quantifiers, undergo obligatory quantifier raising (QR) as one constituent. I follow Diesing (1992) and É. Kiss (1995) in that all “strong quantifiers” in Cantonese, such as ordinary negative (non-existential) quantifiers like moujan ‘nobody,’ universal quantifiers like sojaujan and muigojan ‘everyone’ with dou-quantification (Cheng 1995), existential quantifiers like jaujan ‘someone’ and neg-whQs like mou-bingo ‘nobody’ are required to undergo QR. In contrast, only quantifiers with douquantification in Mandarin Chinese appear in the SOV structure. Negative phrases like ekkert (Rögnvaldsson 1987) in Icelandic undergo overt quantifier raising. The principle of Full Interpretation (FI) … requires that every element of PF and LF, taken to be the interface of syntax (in the broad sense) with systems of language use, must receive an appropriate interpretation-licensed in the sense indicated. (Chomsky 1986, 98).

In order to receive the full quantificational interpretations, either nonexistential or existential, neg-whQs have to undergo overt QR to the preverbal position to be licensed as strong quantifiers, obtaining both possible presuppositions. In Chomsky’s Minimalist Program (MP) (Chomsky 1995a), Agree and EPP features yield obligatory movements of neg-whQs into the specifier position of a structurally higher position. Attract F and Minimal Link Condition (Chomsky 1995b) are defined as follows, which constrains any possible syntactic movement involved for a probe in offering a position for any movement of a valid goal. (13)

Attract F K attracts F if F is the closest feature that can enter into a checking relation with a sublabel of K.

(14)

Minimal Link Condition K attracts Į only if there is no ȕ, ȕ closer to K than Į, such that K attracts ȕ.

Dual Interpretations of Object Neg-wh-Quantifiers

105

Based on the first scenario in (13a), overt quantifier raising is triggered by the uninterpretable features, as described below with the representation in (15). TP

(15)

Subj Mary

T‫މ‬

T

NegP Neg‫މ‬

AGREE

Op炌[iNeg]

vP

mou-bindoui [uNeg,Quant:Strong]

v‫މ‬

v[uQuant, EPP] want

VP V go

QuantP ti

QUANTIFIER RAISING

The object neg-whQ mou-bindou undergoes overt QR to satisfy unvalued features such as [uQuant] and [uNeg]. The EPP feature on v, as Last Resort, triggers movement of the neg-whQ into Spec,vP and possible further successive movements. The v probes its c-command domain and attracts mou-bindou for feature matching and valuation (and deletion). Mou-bindou first lands in Spec,vP, where [uQuant] is valued and deleted. In addition, the EPP feature on v is also checked. Such constructions, with the object neg-whQ carrying an [uNeg], require the projection of NegP in the derivation and allow a sentential negation. The covert negative operator Op炌, as an active probe in the Neg0, carries an [iNeg] feature

106

Chapter Six

and, therefore, Agrees with mou-bindou, and checks and deletes the [uNeg] feature on mou-bindou. All features are checked for the probes and they become inactive and drive no further syntactic operation. Hence, object neg-whQs move to Spec,vP and nowhere else obligatorily in overt syntax in Cantonese, resulting in the unique SOV structure. Such overt movement licenses the negative reading of neg-whQs, and, therefore, gives rise to the sentential negation interpretation. The recent work on the syntax of LF suggests that QR is required for antecedent-contained deletion (ACD) (Kennedy 1997; May 1985) construction as a condition for grammaticality. Kennedy (1997) suggests that the principles forcing LF movement of any lexical material should be the same as those forcing overt (PF) movement. The data with ACD in Cantonese in example (16) supports the claim of the existence of the overt QR in Cantonese. VP-deletion in Cantonese is marked by modal verbs like jau ‘have,’ hoji ‘can’ and in infinitival constructions after hui ‘to.’ This patterns with ACD constructions involving a relative clause in Mandarin Chinese, where modal verbs such as neng ‘can’ or gan ‘dare to’ (Soh 2005) are required. The neg-whQ mou-matje in the following example survives in an ACD construction. Although Cantonese is subject to scrambling, such raising of neg-whQs is not simply an object shift (Chomsky 2001) because the unique SOV structure is restricted to object neg-whQs. (16)

Ngo mou-matjei maai-zo ti (ji) nei dou jau ge. I no-what buy-PVF (that) you also have GE a. ‘I bought nothing that you did.’ b. ‘I bought only a few things that you did.’

Under the assumption in (12), the decomposition for a neg-whQ follows after the obligatory and overt QR because it is internally complex in morphology. The existential reading is triggered where the split reading is made available due to the decomposition of the neg-whQ into negation and indefinites. This is in accord with Kayne (1998), who proposes that Norwegian neg-phrases are a combination of a negation ‘not’ plus an existential/indefinite. (17)

The NEG-criterion: a. A NEG-operator must be in Spec-head agreement relation with an X0 [NEG]. b. An X0 [NEG] must be in Spec-head agreement relation with a NEG operator.

Dual Interpretations of Object Neg-wh-Quantifiers

107

I follow Haegeman and Zanuttini’s (1991) Neg-criterion, which forces the negative morpheme mou to move to the specifier of NegP after decomposition in order to be in a Spec-head agreement relation with the Op炌 at X0 [NEG]. Neg-raising is forced by the lexical ambiguity of mou, being either interpretable or uninterpretable, and mou cannot stay under Op炌. This neg-raising takes place after the obligatory overt movement to the preverbal position and still preserves the syntactic structure of an object neg-whQ construction prior to Spell-out. The existential reading is made available under the double negated context as an outcome. TP

(18)

Sub Mary

T‫މ‬ T

NegP mouj [iNeg]

Neg‫މ‬ vP

Op炌[iNeg] QuantP[uNeg,Quant:Strong]i tj

bindou

v‫މ‬

v[uQuant, EPP] VP want V go

QuantP ti

QUANTIFIER RAISING

108

Chapter Six

(19)

Semantic representation of the existential reading: 炌…炌‫(׌‬x) [place (x) I want to go to x] ‘It is not the case, such that there is not a place x that I want to go.’ (Lit. ‘There is at least somewhere I want to go.’)

Given that constructions with object neg-whQs have dual interpretations, the choice of interpretation in these constructions is always context dependent in Cantonese (given that neg-whQs are typical colloquial terms). On the one hand, the “only a few” reading is facilitated when there is an overt sentence particle (SP) such as zaa (“only”), ze/zek (“emphatic”), zimaa (“only”) or a rhetorical rising tone. According to Tong and James (1994), some sentence particles “express moods and achieve certain rhetorical functions” (Tong and James 1994, 17). On the other hand, the definite negative reading is forced in affirmed contexts, for example, with an overt sentence particle (e.g., ge “assertion or emphasis”) and a lowering tone. Taking SP laa as an example, with a lowering tone, it has the function of “giving affirmation” (Tong and James 1994, 17). (20)

(Houcoi) ngo mou-matje jiu zou zaa. (luckily) I no-what need do SP a. * ‘(Luckily), I have to do nothing.’ b. ‘(Luckily), I have to do only a few things.’

The sentence particle zaa (“only”) above indicates a restrictive focus “only” (Law 2002; Tang 1998). With the presence of an SP zaa “only,” the existential interpretation of the object neg-whQ mou-matjie is forced in (20b), and the non-existential interpretation in (20a) is no longer available. In accordance with the literature (A. Law 2002; S. Law 1990), zaa occurs in CP and it is high enough to license the existential interpretation of moumatje in (20). It is argued that the sentence particle zaa is quantificational in nature and occupies SFP2 head, where a topic is located, which is higher than the Neg head in the clausal structure. It follows that it blocks any LF (covert) movements of object neg-whQs to take wider scope, as Relativised Minimality (Rizzi 1990) excludes further movements across C0.6

6

However, sentence particles are beyond the scope of this chapter and are not discussed in detail.

Dual Interpretations of Object Neg-wh-Quantifiers (21)

Ngo mou-matje maai-zo ze/zek. I no-what buy-PFV SP a. ? ‘I bought nothing.’ b. ‘I bought only a few things.’

(22)

Ngo mou-matje maai-zo (gaa)wo. I no-what buy-PFV SP a. ‘I bought nothing.’ b. ? ‘I bought only a few things.’

(23)

Keoi mou-bingo zungji ze/zek. he no-who like SP a. ? ‘He likes nobody.’ b. ‘He likes only a few people.’

(24)

Keoi mou-bingo zungji (gaa)wo. he no-who like SP a. ‘He likes nobody.’ b. ? ‘He likes only a few people.’

109

An existential reading seems to be preferable to a non-existential one in neg-whQ constructions in (21) and (23), which end with ze/zek, and a nonexistential reading is preferred to an existential one in (22) and (24), which end with (gaa)wo. This is expected under our assumption since (gaa)wo (“reminder”) tends to have a lowering tone whereas ze/zek (“emphatic”) tends to have a rising tone. To summarize, the neg-whQ inherits both the [uNeg] and [Quant:Strong] features from the internal negative morpheme mou and unpronounced quantifier operator Ø. As a strong quantifier, it is forced to undergo overt quantifier raising. In order to obtain a full interpretation, it is attracted to Spec,vP, then to Spec,NegP, checking and deleting uninterpretable features under Agree. Therefore, it gives rise to the sentential negation interpretation. The additional existential presupposition is triggered by the following decomposition mechanism that takes place. The choice between the dual interpretations is context-dependent. Rhetorical contexts and contexts with sentence final particles with a rising tone such as zaa (“only”), ze/zek (“emphatic”) and zimaa (“only”) indicate presuppositions of existence. They tend to give rise to existential presupposition interpretations, whereas sentence particles with a lowering tone like (gaa)wo tend to infer negative readings.

Chapter Six

110

3. Cantonese wh-words in a feature-based account In this section, I support my proposal by presenting data where Cantonese wh-expressions are used as negative presuppositions, and briefly discuss Cantonese wh-words in a feature-based approach. Cheung (2006, 2009) suggested that Negative Wh-words (NWHs) in Cantonese could also be used to give a negative reading and be paraphrased closely as “no way… .” NWHs are restricted mainly to the short forms of Cantonese whwords, such as bin(dou) ‘where,’ dim ‘how’ and me ‘what’ (but not the long forms of ‘how’ dimjoeng and ‘what’ matjie). The negative interpretation of these words can only be maintained in a pre-modal position. The distribution of NWHs is as follows ((25) and (26) are based on Cheung 2006): (25)

Keoi bin jau luksap seoi aa?! he where have sixty year-old SP a. ‘No way is he 60 years old.’ b. * ‘Where will he be 60 years old?’

(26)

Keoi dim wui maai go bun syu aa?! he how will buy the CL book SP a. ‘No way will he buy the book.’ b. * ‘How will he buy the book?’

On a par with shenme ‘what’ in Mandarin Chinese, matje ‘what’ can also be used to give a strong negative implication when it appears after auxiliaries. The following illustrate the use of shenme in Mandarin Chinese (27) (Hsieh 2001) and its counterpart matje in Cantonese (28), which give rise to a strong negative implication. (27)

Zhe you shenme hao? this have what good ‘What good is this?’

(28)

Gam jau matje hou? this have what good ‘What good is this?’

Wh-elements occurring either in pre- or post-auxiliary positions, as shown by (27) and (28), have negative interpretations but not interrogative interpretations. I have further analyzed Cheung’s NWHs in Cantonese as a variation from other wh-words in their inherent features. On the one hand, NWHs appear to bear an internal (uninterpretable) [uNeg] to be licensed in

Dual Interpretations of Object Neg-wh-Quantifiers

111

the pre-auxiliary position and a [–Q] feature, and, therefore, a negative rather than an interrogative reading is easily obtained. On the other hand, ordinary wh-words bear the [+/–Q] feature. Wh-arguments such as matje ‘what,’ bingo ‘who’ and bindou ‘where’ even bear a [uQuant] feature, which allows them to be interpreted as either interrogative or indefinite depending on the contexts they appear in. (29) NWHs e.g., bin(dou), dim, me, mat(je) [uNEG] [–Q]

WH e.g., matje, bingo, bindou, dimgaai, dimjoeng [+/–Q]

Since NWHs are normally short forms of ordinary wh-words, perhaps such [uNeg] elements are embedded in every wh-word and the negative reading is retrieved in different ways, for example, where NWHs appear in the preor post-auxiliary position. In some cases, even the long form of wh-words give rise to negative meanings, in particular contexts like (30). (30)

Bingo wui zoek ji gin saam gaa, gam watdak. who will wear this piece clothes SP so ugly ‘Who will ever wear this top? This is so ugly.’ (Lit. ‘Nobody will wear this top, so ugly (it is).’)

The reason why bingo in (30) gets a negative interpretation is beyond the scope of this study. However, the data suggest some sort of [uNeg] feature embedded within wh-words and the negative reading is waiting to be triggered in some way. Therefore, dual interpretations can somehow be hinted at from the complexity of a neg-whQ. Ambiguity arises from the interaction of the lexical ambiguity of the negative morpheme mou itself and some sort of [uNeg] features within wh-words in Cantonese. This study needs to be further extended to provide a unified account for negwhQs and wh-words in Cantonese.

4. Existential interpretations of Cantonese wh-words This section examines Yip and Matthews’ (2000) data in Cantonese where wh-words give rise to special existential interpretations in negative sentences. The data supports our proposal that an existential interpretation is possible in negative (neg-whQ) constructions.

112 (31)

Chapter Six Mou bingo wui gam chun ge. no who will so stupid SP ‘Hardly anyone would be so stupid.’

The interaction of a negative word and a wh-word gives rise to the meaning “hardly at all” rather than “not at all,” according to Yip and Matthews (2000). Mou bingo in (31), as a subject, is also a neg-whQ, however, in my proposal, a subject neg-whQ has no optional existential interpretation and only gives rise to a sentential negation interpretation. The same “hardly” interpretation remains even if mou bingo is replaced by the ordinary negative quantifier mou-jan ‘nobody’ in (31). Therefore, it is questionable whether the ambiguous mou and the sentence particle ge have an effect on giving rise to existential interpretations here, rather than mou and wh-word combined as one constituent in such a case. Neg-whQs in the subject position are outside the scope of this study. To better explain the “hardly at all” interpretation in these kinds of constructions, I focus on constructions with an object neg-whQ where the interpretation changes if it is replaced by an ordinary negative quantifier. In the following examples cited from Yip and Matthews’ (2000) data, the so-called “hardly at all” interpretation is only available with object neg-whQs or a negation and wh-word as a combination in the preverbal position. (32)

Nei gamjat mou matje zou. you today no what do ‘You don’t have anything much to do today.’

(33)

Ngodei mou bindou heoi. we no where go ‘We don’t have anywhere much to go.’

(34)

Ngo mou dim(jeong) lam-guo. I no how think-PFV ‘I hardly gave it any thought.’

(35)

Nei gamjat mou matje zou, zinghaai jiu daa fon sun zek. you today no what do just need type CL letter SP ‘You don’t have anything much to do today, just type up a letter.’

The above examples support the claim that an existential interpretation is available even in the absence of sentence particles. However, in replacing all neg-whQs by ordinary negative quantifiers, the “hardly at all” interpretations immediately disappear. These support the proposal made in this chapter that there are dual interpretations in constructions with object neg-whQs.

Dual Interpretations of Object Neg-wh-Quantifiers

113

5. Conclusion To conclude, this study has looked at neg-whQs in Cantonese in a featurebased approach. The possible dual readings of an object neg-whQ, which is morphologically composed of a negative morpheme mou and a whphrase, are made clear by the proposed interpretable [Quant:Strong] feature and the [uNeg] feature. Both non-existential (negative) and existential interpretations are available in neg-whQs, which makes negwhQs distinct from any other ordinary negative quantifiers. The alternation of the two possible readings is context-dependent in the absence or presence of sentence particles (e.g., “only” particle zaa) implying existential presuppositions, or the change of tone of sentences. The study has also attempted to unify the features of wh-phrases in Cantonese in which [uNeg] could be involved, by suggesting NWHs are a subset of WH. The additional existential reading is supported by data with a “hardly at all” reading in Cantonese. The scope of this study is limited to object neg-whQs and further investigation is needed for a unified account for wh-phrases in Cantonese.

References Aoun, Joseph, and Audrey Li. 1993. Syntax of Scope. Cambridge, MA: MIT Press. Beghelli, Fillippo, and Tim Stowell. 1997. “Distributivity and Negation: The Syntax of each and every.” In Ways of Scope Taking, edited by Anna Szabolcsi, 71–197. Dordrecht: Kluwer. Cheng, Lai-Shen Lisa. 1991. “On the Typology of Wh-Questions.” PhD diss., MIT. —. 1994. “Wh-Words as Polarity Items.” Chinese Languages and Linguistics 2:615–40. —. 1995. “On Dou-Quantification.” Journal of East Asian Linguistics 4:197–234. Cheung, Yam-Leung Lawrence. 2006. “Negative Wh-Words in Cantonese.” In Proceedings of the 18th North America Conference on Chinese Linguistics (NACCL-18), edited by Janet Xing, 90-102. Los Angeles: GSIL Publications, University of Southern California. —. 2009. “Negative Wh-Construction and its Semantic Properties.” Journal of East Asian Linguistics 18:297–321. doi:10.1007/s10831009-9051-2. Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press.

114

Chapter Six

—. 1995a. “Bare Phrase Structure.” In Government and Binding Theory and the Minimalist Program, edited by Gert Webelhuth, 383–440. Oxford: Blackwell. —. 1995b. The Minimalist Program. Cambridge, MA: MIT Press. —. 2001. “Derivation by Phrase.” In Ken Hale. A Life in Language, edited by Michael Kenstowicz, 1–52. Cambridge, MA: MIT Press. Diesing, Molly. 1992. Indefinites. Cambridge, MA: MIT Press. É. Kiss, Katalin. 1995. “NP Movement, Operator Movement, and Scrambling in Hungarian.” In Discourse Configurational Languages, edited by Katalin É. Kiss, 207–43. Oxford: OUP. Haegeman, Liliane, and Raffaella Zanuttini. 1991. “Negative Heads and the Neg-Criterion.” The Linguistic Review 8:233–51. Heim, Irene Roswitha. 1982. “The Semantics of Definite and Indefinite Noun Phrases.” PhD diss., University of Massachusetts, Amherst, MA. Hsieh, Miao-Ling. 2001. “Form and Meaning: Negation and Question in Chinese.” PhD diss., University of Southern California. Huang, Cheng-Teh James. 1982. “Logical Relations in Chinese and the Theory of Grammar.” PhD diss., MIT. Kayne, Richard. 1998. “Overt vs. Covert Movement.” Syntax 1:2:128–91. Kennedy, Christopher. 1997. “Antecedent-Contained Deletion and the Syntax of Quantification.” Linguistic Inquiry 28(4):662–88. Kratzer, Angelika. 1995. “Stage-Level and Individual-Level Predicates.” In The Generic Book, edited by Gregory N. Carlson and Francis Jeffry Pelletier, 125-75. Chicago: University of Chicago Press. Law, Sam-Po. 1990. “The Syntax and Phonology of Cantonese SentenceFinal Particles.” PhD diss., Boston University. Law, Ann. 2002. “Cantonese Sentence-Final Particles and the CP Domain.” UCL Working Papers in Linguistics 14:375–98. Lee, Man-Ki Theodora. 2011. “Overt Quantifier Raising of Negative-WhQuantifiers in Cantonese.” In Proceedings of the Sixth Cambridge Postgraduate Conference in Language Research, edited by Chris Cummins, Chi-Hé Elder, Thomas Godard, Morgan Macleod, Elaine Schmidt, and George Walkden, 92–107. Cambridge: Cambridge Institute of Language Research. Lewis, David. 1975. “Adverbs of Quantification.” In Formal Semantics of Natural Language, edited by Edward L. Keenan, 3–15. Cambridge: CUP. Li, Yen-Hui Audrey. 1992. “Indefinite Wh in Mandarin Chinese.” Journal of East Asian Linguistics 1:125–55. Lin, Jo-Wang. 1998. “On Existential Polarity Wh-Phrases in Chinese.” Journal of East Asian Linguistics 7:219–55.

Dual Interpretations of Object Neg-wh-Quantifiers

115

Lin, Jo-Wang. 2004. “Choice Functions and Scope of Existential Polarity WH-Phrases.” Linguistics and Philosophy 27:451–91. May, Robert. 1985. Logical Form: Its Structure and Derivation. Cambridge, MA: MIT Press. Penka, Doris, and Arnim von Stechow. 2001. “Negative Indefinita unter Modalverben.” In Modalität und Modalverben im Deutschen, edited by Reimar Müller and Marga Reis, 263–86. Hamburg: Buske Verlag. Potts, Christopher. 2000. “When Even No’s Neg is Splitsville.” Jorge Hankamer’s Web Fest, edited by Sandy Chung, Jim McCloskey, and Nathan Sanders. Available at http://ling.ucsc.edu/Jorge/potts.html. Progovac, Ljiljana. 1988. “A Binding Approach to Polarity Sensitivity.” PhD diss., UCLA. Rizzi, Luigi. 1990. Relativized Minimality. Cambridge, MA: MIT Press. Rögnvaldsson, Eiríkur. 1987. “OV Word Order in Icelandic.” In Proceedings of the Seventh Biennial Conference of Teachers of Scandinavian Studies in Great Britain and Northern Ireland, edited by Robin M.A. Allan and Michael P. Barnes, 33–49. London: London University College. Soh, Hooi-Ling. 2005. “Wh-in-Situ in Mandarin Chinese.” Linguistic Inquiry 36:143-55. Tang, Sze-Wing. 1998. “Parametrization of Features in Syntax.” PhD diss., University of California, Irvine. Tong, S.-T. Keith, and Gregory James. 1994. Colloquial Cantonese – A complete Language Course. London and New York: Routledge. Tsai Wei-Tien, Dylan. 1994a. “On Nominal Islands and LF Extractions in Chinese.” Natural Language and Linguistic Theory 12:121–75. —. 1994b. “On Economizing the Theory of A-bar Dependencies.” Unpublished PhD diss., MIT. Yip, Virginia, and Stephen Matthews. 2000. Intermediate Cantonese: A Grammar and Workbook. London: Routledge.

CHAPTER SEVEN TOWARDS AN ANALYSIS OF CLITIC DOUBLING IN ROMANIAN1 ALINA TIGĂU

1. Introduction2 Clitic doubling (CD) is a phenomenon by means of which clitic pronouns appear in verb phrases together with the full noun phrases that they refer to (as opposed to the cases where such pronouns and full noun phrases are in complementary distribution, e.g., French). Clitic doubling is found in many languages, including Spanish, Romanian, etc. In each case, this phenomenon seems to go by different rules. Romanian seems to correlate the use of the case marker PE with the possibility of clitic doubling the overtly marked constituents in the accusative to such an extent that various linguists have argued that the accusative feature on the verb is checked by means of the clitic pronoun: in those cases where the clitic co-occurs with a lexical DP, the preposition PE would be required in order to check the accusative feature on the direct object DP. However, such an account would imply that PE marking and CD are part of a more complex phenomenon, a fact which is not accurate: historical data show that the two are independent phenomena which have developed at different stages in the language. Thus, we would rather view 1

Work on this study was financed by research project PN-II-ID-PCE-2011-30959. 2 Abbreviations: ACC – accusative; CD – Clitic Doubling; CLLD – Clitic Left Dislocations; cl. – clitic; DAT – dative; DP – determiner phrase; FEM – feminine; PL – plural; P(R)E – accusative case marker in Romanian; Refl. – reflexive; SĂ – particle introducing the ‘Conjunctiv’ (roughly the ‘Subjunctive’) mood in Romanian; SG – singular; M – masculine.

Towards an Analysis of Clitic Doubling in Romanian

117

the two phenomena as independent but having similar interpretational effects. In a diachronic study carried out on a number of old Romanian texts (Tigău 2010), we discovered that the use of PE seems to be a far more remote phenomenon, dating back as far as the 16th century, than clitic doubling which developed at a later stage, after the advent of Clitic Left Dislocation (CLLD). Stan (2009) also points to the existence of very few doubling structures (the direct object is mainly doubled by personal and reflexive pronouns), whereas GheĠie (1974) proves that such a phenomenon is totally absent from most of the texts dating back to the 16th century. Furthermore, Asan (1961) maintains that the doubling of DPs is a rare phenomenon in 16th century texts. In line with these data, Coteanu (1963), Niculescu (1965) and Rosetti (1978) point out that the syntactic constraints concerning the doubling of the direct object were much more stable when it came to CLLD because, in this latter case, the role of the clitic would have been to resume a lexical object which had been dislocated from its base-generated position. In the 16th century texts3 that we studied, we found no instances of CD. However, as Stan (2009) points out, the phenomenon had already appeared by that time in some CD constructions where personal pronouns and reflexives seemed to be the only elements prone to CD. Proper names and definite descriptions, which could undergo PE marking, were not clitic doubled: (1)

ne-au rămas [“ne-au învins”] pre noi de bani us have left [us have defeated] PRE us of money

(2)

să ne piiarză úi pre noi SĂ us kill and PRE us ‘to kill us too’

3 We analyzed Catehismul lui Coresi (1559) (Coresi’s Cathehism), Pravila lui Coresi (1570) (Coresi’s Code of Laws) as well as various prefaces and epilogues to texts dating from the 16th century: Coresi: Tetraevanghel (1561) (The Four gospels), Coresi: Tîlcul evangheliilor (1564) (Explaining the Gospels), Coresi: Molitvenic (1564) (The Prayer Book), Coresi: Psăltire Romînească (1570) (The Romanian Psalm Book), Coresi: Psăltire Slavo-Romînă (1570) (The SlavicRomanian Psalm Book), Coresi: Evanghelie cu învăĠătură (Gospel with Advice), Palia de la Orăútie (1582) (The Old Testament from Orăútie). To these texts we have added a number of documents, testaments, official and private letters.

Chapter Seven

118 (3)

úi-L prinsără pre El and-him-cl. caught PRE Him ‘and they caught Him’

(4)

să nu să aibâ pre sine SĂ not SĂ have PRE self ‘so as not to have oneself’ (Stan 2009, 19)

The phenomenon increases in frequency, reaching a climax in the 19th century, when most of the strong forms of personal pronouns are accompanied by their clitic counterparts: (5)

17th c.:

úi lăsă pre ea frigul and left PE she fever

17th c.:

ùi lăsă pre ea frigurile and left PE she fever-PL

19th c.:

ùi o au lăsat pe ea frigurile. and she-cl. have-they left PE she fever-PL ‘And the fever left her’ (von Heusinger and Onea 2008)

Thus, pronouns are obligatorily case marked by PE and systematically accompanied by their clitic counterparts in 19th century Romanian, as in (5) above, where the strong form of the personal pronoun pe ea is accompanied by the 3rd person, feminine clitic o. This increase in CD instances steadily decreases towards the 20th century, when the co-occurrence of both the strong form of a pronoun and of its clitic counterparts is dispreferred; the clitic pronoun seems to suffice. (6) is a 20th century rendition of (5) above. (6)

ùi au lăsat-o frigurile. and have-they left-her-cl. fever-the ‘And the fever left her.’

Thus, we may conclude that CD developed at a different point in time and independently from PE marking. Furthermore, CD does affect the same type of DPs as PE marking. In the following section, we will analyze a corpus of present day Romanian CD structures with the purpose of identifying those types of DPs which undergo CD. We will be equally interested in overlapping the sets of DPs which undergo PE marking with those DPs which accept CD.

Towards an Analysis of Clitic Doubling in Romanian

119

2. CD in present day Romanian In this section, we will consider the case of CD in present day Romanian. It seems that the cases where CD appears form a subset of those cases where the direct object is overtly case marked by means of PE. Before embarking upon a survey of CD cases, we need to draw attention to the fact that we will split all the possible situations into: cases where CD is impossible, cases where it is obligatory and, finally, cases where CD is optional.

2.1. Ungrammatical CD Bare quantifiers may be PE marked when they are semantically [+Person]. On the other hand, bare quantifiers cannot be clitic doubled due to the fact that there is a breach in agreement between the clitic, which is ij-complete, and the bare quantifier, which is not. Even when bare quantifiers are specified for [+Person], they still do not match the clitic because they have the default singular number and they are unmarked for gender: (7)

La facultate nu am întâlnit (*-o) pe at faculty not have-I met her-cl. PE nimeni cu care să discut. nobody with who SĂ talk ‘I didn’t meet anyone at the faculty to talk about my paper with.’

Example (7) above becomes ungrammatical if we choose to double the direct object pe nimeni. However, when the bare quantifier nimeni ‘nobody’ is replaced by nici un/nici o ‘none,’ CD is possible as this DP is marked for gender: (8)

Mihai n-a vrut s-(o) ajute pe Mihai not-has wanted SĂ-her-cl. help PE nici una dintre surorile sale la teme. none of sisters his at homework ‘Mihai didn’t want to help any of his sisters with their homework.’

The explanation above seems to hold as bare quantifiers may combine with dative clitics which are not marked for gender (the exact feature which accounts for the breach in agreement between direct object bare quantifiers and accusative clitics). Consider the example below, where the

120

Chapter Seven

dative bare quantifier nimănui ‘nobody’ is clitic doubled by means of the appropriate clitic in the dative: (9)

Nu i-a venit nimănui să creadă not him-cl.-has come nobody SĂ believe că Paul supravieĠuise accidentului, that Paul had survived accident.the nici măcar soĠiei lui care mai nutria not even wife his who still hoped speranĠa să-l vadă din nou acasă. SĂ-him-cl. see again home ‘Nobody could believe that Paul had survived the accident, not even his wife, who still hoped to see him at home.’

Thus, the bare quantifier nimănui ‘nobody’ may be doubled by a dative clitic because this pronoun is not marked for gender, i.e., it does not require that the DP it doubles should be ij-complete as accusative clitics do.

2.2. Obligatory CD It would be very interesting to see exactly why CD is obligatory with definite pronouns. At face value, if we were to compare bare quantifiers (which disallow CD altogether) and pronouns, we would see that pronouns are ij-complete and, therefore, can be a perfect match for the clitic which is also ij-complete. A fact which seems to confirm this hypothesis is that, in the 17th century texts we studied,4 the only (very few) cases which could be clitic doubled were those cases where a strong pronoun or a reflexive was employed. Notice also that definite pronouns are also obligatorily marked

4

We have studied the following works: Istoria ğării Româneúti de la octombrie 1688 până la martie 1718 (The History of ğara Românească from October 1688 until March 1718), Istoriile domnilor ğării Rumâneúti. Domnia lui Costandin – vodă Brâncoveanu (Radu Popescu) (The Lives of the Rulers of ğara Românească. The reign of Costandin Brâncoveanu (Radu Popescu)), Istoria Ġării rumâneúti de când au descălecat pravoslavnicii creútîni (LetopiseĠul Cantacuzîno) (The History of ğara Românească since the Advent of the Christian Orthodox Believers) (The Cantacuzino Chronicle), LetopiseĠul ğării Moldovei (Ion Neculce) (The Chronicle of Moldavia by Ion Neculce).

Towards an Analysis of Clitic Doubling in Romanian

121

by PE. Therefore, this domain is one where the two mechanisms overlap. Consider some cases of clitic doubled definite pronouns: (10)

Personal pronouns: Au ajutat-o pe ea părinĠii, dar ce folos have-they helped-her-cl. PE her parents.the but what use dacă n-o duce mintea. if not-her.cl. help mind ‘Her parents helped her but it was no use as she’s not bright.’

(11)

Pronouns of politeness: Nu cumva crezi că eu te-am fluierat pe not somehow believe.you that I you-have-I whistled PE dumneata? you ‘Do you believe that I have whistled at you?’

(12)

Reflexive pronouns: Matei se va distruge pe sine până la urmă dacă Matei Refl. will destroy PE self eventually if mai bea fără măsură. more drink without measure ‘Matei will sicken himself eventually if he keeps drinking so much.’

(13)

Possessive pronouns: Nu i-am mai văzut pe ai mei de un car de ani. not them-cl.-have-I more seen PE mine for ages ‘I haven’t seen my parents for ages.’

(14)

Demonstrative pronouns: Deúi avem o mulĠime de caiete, although have-we a multitude of notebooks, Ioana nu-l vrea decât pe acela. Ioana not-him-cl. want only PE that one. ‘Although we have plenty of notebooks, Ioana only wants that one.’

Thus, both PE marking and CD are obligatory with definite pronouns. Notice also that the two mechanisms may equally mark pronouns whose referents are either animate or inanimate. Consider example (14) above, where the demonstrative pe acela pointing to an inanimate referent, i.e., a notebook, undergoes obligatory PE marking and CD.

Chapter Seven

122

2.3. Optional CD Proper names. Unlike PE marking, which is obligatory with proper names pointing to animate referents, CD is merely optional with this type of DP. Consider the examples below which prove this point. Notice that the proper name Ionel may not be clitic doubled if it is not PE marked first: (15)

Maria (îl) ajută mereu *(pe) Ionel la teme. Maria (him-cl.) help always PE Ionel at homework ‘Maria always helps Ionel with his homework.’

Definite descriptions. Definite descriptions are also optionally CD-ed, just like proper names. Unlike proper names, however, these DPs are also optionally PE marked: (16)

a. Maria îi duce pe copii la grădiniĠă în Maria them-cl. takes PE children at kindergarten in fiecare zi. every day ‘Maria takes the children to the kindergarten every day.’ b. Maria duce copiii la grădiniĠă în fiecare zi. Maria takes children-the at kindergarten in every day c. * Maria Maria fiecare every

îi duce them-cl. takes zi. day

copii la grădiniĠă în children at kindergarten in

In (16a), the definite description pe copii is doubled by the clitic îi. The sentence remains grammatical in the absence of PE marking and CD, provided that the DP in question acquires a definite article (as in (16b)). Example (16c) points to the fact that a definite description may not be clitic doubled if it is not PE marked. Definite descriptions which refer to [–human, +animate] entities may be clitic doubled, provided that they are PE marked: (17)

Ce motiv a avut copilul când l-a lovit pe what reason has had child-the when him-cl.-has struck PE căĠeluú? puppy ‘Why did the child strike the puppy?’

Towards an Analysis of Clitic Doubling in Romanian

123

Definite descriptions which refer to inanimate entities may not be marked by CD since they are not PE marked. Indefinite descriptions. Both PE marking and CD are optional with indefinites, and CD cannot occur in the absence of PE (when these two mechanisms do occur). (18)

Atunci (l)-au arestat poliĠiútii *(pe) un then (him-cl.)-have arrested policemen-the PE a trecător nevinovat ca să nu zică lumea passer-by innocent so as not to say the people că nu-úi fac treaba that not-Refl.-they do their job. ‘Then, the policemen arrested an innocent passer-by so as people could not say that they were not doing their job.’

Notice also that CD is restricted to [+animate] indefinites. The [íhuman, +animate] indefinite DP o căĠeluúă ‘a puppy,’ may be both PE marked and clitic doubled: (19)

Am văzut-o pe o căĠeluúă lingându-úi lăbuĠa have-I seen-her-cl. PE a puppy licking-Refl. paw úi pe alta dând din coadă. and PE another wagging from tail ‘I saw one of the puppies licking its paw and another one wagging its tail.’

On the other hand, [–animate] DPs cannot be clitic doubled or PE marked. (20)

Prisecariu (*l-) a vândut (*pe) un teren Prisecariu it-cl.-has sold PE a piece of land către Praktiker. to Praktiker ‘Prisecariu has sold a piece of land to Praktiker.’

Thus, CD may only occur in some of those cases where PE marking appears.

2.4. Conclusions In this section, we have considered the use of CD in present day Romanian, and we have seen that this mechanism applies to a subset of those cases which may be affected by PE marking. Thus, bare quantifiers

124

Chapter Seven

may be PE marked, but they are never clitic doubled on account of their ij-incompleteness; definite pronouns are obligatorily PE marked and CDed, irrespective of whether their referent is animate or not. Moreover, proper names, definite descriptions and indefinites are optionally PE marked and CD-ed. Furthermore, CD does not appear in the absence of PE marking. The reverse does not hold. Finally, CD is sensitive to animacy in that DPs whose referents are inanimate may not be clitic doubled.5

3. Towards an account of CD in Romanian The analysis with respect to CD that we will be adopting goes along the lines of Cornilescu (2006), Cornilescu and Dobrovie-Sorin (2006), Roberts (2010) and Tigău (2008). According to these accounts, the clitic in the CD structure forms a chain with a DP in an argument position. As for the clitic pronoun, Cornilescu (2006) argues that it has lost (some of) its argumental status, gradually turning into an agreement morpheme on the verb. The nature of the clitic is thus correlated with the type of position (an argument one) occupied by the DP double which functions as a true syntactic argument.

3.1. Clitics as object agreement morphemes Franco (2000) claims that, to different degrees, Romance pronominal clitics are in the process of acquiring the status of object verbal inflection (i.e., agreement markers). In his view, object clitics should be analyzed as object agreement morphemes on the verb, on a par with subject-verb agreement, and not as pronominal arguments that are phonologically dependent. In the case of CD, the relationship between the clitic and the element they double is the same as a verb-argument agreement relationship. As we will show,6 object clitics in Romanian pattern very much like their counterparts in Spanish so that we have reasons to believe them to be very advanced in the process of acquiring the status of verbal inflection. 5

Definite pronouns, however, are obligatorily case marked and CD-ed, irrespective of whether their referents are animate or not. 6 This section is part of Tigău (2008, 2010).

Towards an Analysis of Clitic Doubling in Romanian

125

Aware of their mixed syntactic and morphological properties (which poses problems with classification), Franco (2000) assumes that clitics, in general, are intermediate elements between bound words and inflectional affixes, that is, between pronouns and inflectional affixes. Furthermore, due to the fact that not all pronominal clitics in the languages of the world have the same behavior or distribution, he places them along a spectrum such as the one given in (21) – their position on this spectrum depends on how much the properties of each type of clitic resemble those of a pronoun or those of a verbal inflectional affix: (21)

Inflectional Affix [-Z-Y-X-W-] Pronouns Pronominal clitics

In this section, we would like to prove that Romanian pronominal clitics are at an advanced stage in the process of becoming inflectional affixes in line with Franco (2000). The Same Specific Host [Aux/V]. Pronominal clitics in contemporary French, Italian, and Spanish (unlike their medieval counterparts which showed much freedom in affixation, a fact that placed them closer to a word like status rather than to an inflectional one) are restricted to preverbal or post-verbal positions depending on the verb’s feature [Į finite]. Consider the following examples for Romanian: (22)

a. Maria îl cumpără. Maria it-3.SG.M buys ‘Maria buys it.’ b. Maria l-a cumpărat. Maria it-3.SG.M-has bought ‘Maria bought it.’ c. Maria nu l-a cumpărat. Maria not it-3.SG.M-has bought. ‘Maria didn’t buy it.’ d. Maria nu-l cumpără. Maria not it-3.SG.M buy ‘Maria doesn’t buy it.’ e. Maria nu-l mai cumpără. Maria not-it-3.SG.M more buy ‘Maria doesn’t buy it anymore.’

126

Chapter Seven f. Maria vrea să-l cumpere. Maria want sa-it-3.SG.M buy ‘Maria wants to buy it.’ g. Daca-l vede aici se va supăra. if-it-3.SG.M sees here Refl. will get mad ‘If she sees it here she will get mad.’ h. De-l prind că vine, vă alung pe-amândoi. if-it-3.SG.M catch-I that comes-he, you-cl. chase-I PE-both ‘If I catch him coming here, I’ll chase both of you away.’ i. Maria-mi spune tot. Maria-cl-1.SG tells everything ‘Maria tells me everything.’ j. Nemaivăzându-l, am crezut că plecase. not.yet.seeing-cl-3.SG.M have-I thought that left-he ‘Not seeing him, I thought that he had left.’ k. E Ion, pe care-l útii deja. is Ion PE whom-cl-3.SG.M know-you already ‘This is Ion whom you already know.’

Examples (22a, b, c) seem to obey the generalization holding for modern French, Italian, and Spanish, i.e., the pronominal clitic exhibits a preference for a verbal host (either the verb or the auxiliary). The same can be said about example (22j), where the pronominal clitic attaches to the non-finite verb. Notice also that because the form of the verb is non-finite, the clitic appears post verbally in this case. Furthermore, the negation and the adverb mai ‘more’ behave like clitics as well and attach to the verb just like them. The entire complex construct moves to the final landing site of the verb which appears to be MoodP for Romanian. The rest of the examples seem to function as counterexamples to our assumption. Thus, in (22d), the clitic attaches to the negation nu and not to the verb. To make matters worse, example (22e) presents us with a variant wherein the clitic is clearly severed from the verb by the adverb mai. Examples (22f, g, h) are cases where the clitic seems to attach to a complementizer: the ‘conjunctiv’ mood marker să, and the connectors dacă and de. In (22i), the clitic mi attaches to the DP Maria. Lastly, in example (22k), the clitic attaches to the relative pronoun. As we can see, the pronominal clitic can accept a variety of hosts of different nature. However, if we consider examples (22d, e) again, we will notice that the pronominal clitic interacts with elements of the same type,

Towards an Analysis of Clitic Doubling in Romanian

127

i.e., both nu and mai are actually clitics, as we can clearly see from example (22j); this amounts to saying that the three clitics cluster together on the verb. As for examples (22f, g, h, i, k), we would suggest that these are not actually cases of syntactic cliticization but of prosodic cliticization along the lines of Dobrovie-Sorin (1994), i.e., Romanian clitics undergo prosodic cliticization to a word that has nothing to do with the selection of the clitic. Strict Adjacency to [Aux/V]. This refers to the absence of interpolation phenomena which involve the occurrence of intervening elements between the clitic and the host which is either the verb or the auxiliary. Consider some Romance languages where nothing can intervene between the clitic and the verb; object clitics require strict adjacency to their host. (23)

a. (F) * Martine le souvent b. (I) * Martina lo non c. (S) * Martina lo no Martina it often ‘Martina often reads it.’

lit. legge. lee reads.

Romanian behaves differently in that it allows the adverbial clitic mai to occur between the object clitic and the verb. (24)

Maria nu-l mai cumpără. Maria not-him-cl. more buy. ‘Maria does not buy it any longer.’

However, in line with Cornilescu (2006), we will not consider this as a violation of the strict adjacency condition since the intervening element is of the same nature as the clitic. Syntactic Unit with the Host. This property refers to whether or not the clitic is left stranded after the host moves in syntax. Indeed, in Romanian, verbs and clitics form a syntactic unit – this can be seen in instances in which the verb undergoes syntactic operations such as verbsubject inversion in questions: the clitic and the verb remain together as a unit. (25)

a. Maria a citit cartea cu uúurinĠă. Maria has read book-the easily. ‘Maria easily read the book.’

128

Chapter Seven b. Maria a citit-o cu uúurinĠă. Maria has read-it-cl.-3.SG.FEM easily ‘Maria easily read it.’

Fixed Order. On a closer look at the morphology of cliticization, we can see that Romance languages (including Romanian) are subject to ordering constraints in relation to other inflectional affixes. The order of pronominal clitics in the cluster is fixed. Romanian only has accusative and dative object clitics which occur in the order dativeaccusative, as shown in (26a). The opposite order would be ungrammatical, as can be seen in (26b): (26)

a. Alexandru mi-l trimite astăzi Alexandru cl.-DAT-cl.ACC sends today ‘Alexandru sends it to me today.’ b. * Alexandru îl-mi trimite astazi

As pointed out by Anderson (1992), there is a clear similarity between clitics and affixes, which also exhibit rigid ordering. Independent words are, on the other hand, usually allowed a certain degree of free ordering. The properties we presented above rank Romanian quite highly along the agreement continuum argued for by Franco (2000), and pronominal clitics in Romanian are agreement markers on the verb. This entitles us to maintain that Romanian pronominal clitics are at a very advanced stage in the process of acquiring the status of verbal inflection.

3.2. The status of the double The double has been perceived as either an adjunct (Kayne 1989) or an argument (Franco 2000; Cornilescu 2006; Cornilescu and Dobrovie-Sorin 2008; Tigău 2008, etc.). If one considers the DP double an adjunct, then one should account for the fact that (under this view) the double has inert ij-features and cannot participate in the case-agreement system. On the other hand, there are several tests proving that the double is an argument and that it has active ij-features, as well as case features, and that it occupies an A position. Firstly, Binding Theory supports the A status of the doubled DP since this DP can serve as the antecedent of an anaphor, and anaphors must be bound from A-positions, according to Principle C of this theory.

Towards an Analysis of Clitic Doubling in Romanian (27)

129

supăraĠi pe ei înúiúii Îi útiam pe copiii them-cl. knew-I PE children upset PE them themselves pentru că pierduseră meciul. because had lost-they game-the ‘I knew that the children were upset with themselves because they had lost the game.’

In this respect, the clitic doubled DP behaves like the undoubled (argument) DP, which can also serve as an antecedent for an anaphor. As pointed out by Cornilescu (2006, 34), “it is likely that the (prepositional) Accusative objects are arguments.” (28)

La concursul de matematică am văzut copii mulĠumiĠi at contest of Mathematics have-I seen children satisfied de ei înúiúi, dar úi copii supăraĠi de with them themselves but also children upset of performanĠa lor. performance their ‘At the contest in Mathematics I saw children that were pleased with themselves and children that were upset about what they had done.’

Cornilescu (2006) contributes a second argument in favour of the argumental status of the clitic-doubled element: the case of transitive verbs which require that a strong reflexive pronoun co-occur with the reflexive clitic in order for the verb to acquire a reflexive meaning. The fact that the reflexive pronoun is obligatorily required with such verbs points to the fact that this DP is an argument and not an adjunct: (29)

Bogdan se deplânge pe sine prea mult ca să Bogdan Refl. pities PE self too much to SĂ mai bage în seamă úi pe altcineva. more take notice and PE somebody else. ‘Bogdan pities himself too much to pay attention to anybody else.’

A third argument in favour of the argument status of the DP double comes from the field of small clauses. This is because the DP double may function as the subject of a small clause, and these subject DPs are known to occupy argument positions: (30)

O consider [SM pe Maria fată deúteaptă]. her-cl. consider-I PE Mary girl smart. ‘I consider Mary a smart girl.’

130

Chapter Seven

Thus, after taking all these arguments into consideration, one may assert that the double is an argument in Romanian.

3.3. An important property of CD Sentences where the direct object DP is overtly case marked by means of PE differ syntactically from those structures where, apart from PE marking, CD is present. The two types of sentences differ with respect to the phenomenon of binding in that the latter type gives rise to ‘inverse binding,’ i.e., the object DP may bind into the subject DP: (31)

Pe un profesor buni orice elev de-al luii PE a teacher good any pupil of his ‘Any pupil will admire a good teacher.’

îl admiră. him-cl admires.

In example (31) above, the clitic doubled and PE marked indefinite pe un profesor correctly binds the subject DP which it appears to c-command. The same relation of co-reference between the two DPs should be maintained even when the word order changes from OSV to SOV (which can be found in example (32) below): (32)

Orice elev de-al luii îl admiră pe un profesor buni. any pupil of his him-cl. admires PE a teacher good. ‘Any pupil admires a good teacher.’

Thus, a clitic doubled and PE marked object DP may bind into the subject DP, irrespective of the word order. On the other hand, the unmarked counterpart may only bind into the subject if it is in a c-commanding position with respect to this DP. Thus, of the two sentences presented under (33), only variant (33a) is correct because the indefinite DP un profesor duly c-commands the subject DP. Variant (33b) is ungrammatical if co-indexation between the two DPs is maintained because the pronoun lui does not have an appropriate antecedent. (33)

a. Un profesor buni orice elev de-al luii admiră. a teacher good any pupil of his admires. ‘Any pupil will admire a good teacher.’ b. * Orice elev de-al luii admiră un profesor buni. any pupil of his admires a teacher good.

Towards an Analysis of Clitic Doubling in Romanian

131

Thus, with clitic doubled DPs, the binding relationship between the subject and the object holds even if the object has moved past the subject, a fact which should have led to the reversal of c-command relations. On the other hand, binding between the two constituents no longer holds when the order between them is reversed and when no clitic is present, a fact which points to the syntactic difference between the two structures. These results count as evidence that, at some point in the derivation, the clitic doubled direct object c-commands the subject. Thus, at some point, the doubled direct object leaves the vP and moves into a c-commanding position for the subject (which at this point is to be found in-situ, i.e., Spec,vP), which is situated above Spec,vP and below TP. Since binding is involved, such a position can only be an A position as also proved above. Direct objects which are not clitic doubled but which are overtly case marked do not leave the VP.

3.4. Analyzing CD in Romanian Following the lines of analysis of Franco (2000), Cornilescu (2006) and Cornilescu and Dobrovie-Sorin (2008), we have seen that the clitic from CD structures is on the way to becoming an agreement marker and that the DP double is an argument and not an adjunct. In this section, we adopt Uriagereka’s (1995) Big DP analysis of CD as it is defended for Romanian by Cornilescu (2006). According to the Romanian linguist, the clitic is a strong determiner in Romanian which projects a big DP structure as in the one below: (34)

DP

D

PP P pe

DP [ij] (Cornilescu 2006, 36)

This analysis of CD in terms of a Big DP has the advantage of capturing the fact that the clitic and the DP double represent the same ij-argument at merge. Furthermore, it accounts for the selectional

132

Chapter Seven

properties of the clitic, which is ij-complete and which requires that its complement be ij-complete as well. Thus, as already pointed out above, bare quantifiers such as nimeni ‘nobody’ cannot be clitic doubled as they are unmarked for gender.

4. Conclusions We started this chapter with a diachronic account of CD in Romanian and then went on to analyze this structure in present day Romanian. Following the steps of Franco (2000), we provided arguments in favour of the hypothesis that clitics are agreement markers. The DP double was shown to be an argument and a complement of the clitic which acts as a clitic in the Big DP, put forth by Uriagereka (1995).

References Anderson, Stephen R. 1992. A-Morphous Morphology. Cambridge: CUP. Asan, FinuĠa. 1961. PropoziĠia Apozitivă úi ApoziĠia. Limba română 10, 4:312–16. Cornilescu, Alexandra. 2006. “On Clitic Doubling and Parasitic Gaps in Romanian.” Revue Roumaine de Linguistique 1:23–42. Cornilescu, Alexandra, and Carmen Dobrovie-Sorin. 2008. “Clitic Doubling, Complex Heads and Interarboreal Operations.” In Clitic Doubling in the Balkan Languages, edited by Dalina Kallulli and Liliane Tasmowski, 289–319. Amsterdam: John Benjamins. Coteanu, Ion. 1963. Elemente de Lingvistică Structurală. Bucureúti: Editura ùtiinĠifică. Dobrovie-Sorin, Carmen.1994. The Syntax of Romanian. Berlin: Mouton de Gruyter. Franco, Jon. 2000. “Agreement as a Continuum.” In Clitic Phenomena in European Languages, edited by Frits Beukema and Marcel den Dikken, 146–89. Amsterdam: John Benjamins. GheĠie, Ion. 1974. Începuturile Scrisului în Limba Română. ContribuĠii Filologice úi Lingvistice. Bucureúti: Editura Academiei Republicii Socialiste România. von Heusinger, Klaus, and Edgar Onea. 2008. “Triggering and Blocking Effects in the Diachronic Development of DOM in Romanian.” Probus 20:67–110.

Towards an Analysis of Clitic Doubling in Romanian

133

Kayne, Richard S. 1989. “Null Subjects and Clitic Climbing.” In The Null Subject Parameter, edited by Osvaldo Jaeggli and Kenneth J. Safir, 239–63. Dordrecht: Kluwer. Niculescu, Alexandru. 1965. Obiectul Direct Prepozitional in Limbile Romanice. Individualitatea Limbii Romane intre Limbile Romanice. Bucuresti: Editura Stiintifica. Roberts, Ian. 2010. Agreement and Head Movement. Clitics, Incorporation, and Defective Goals. Cambridge, MA: MIT Press. Rosetti Alexandru. 1978. Istoria Limbii Romane. Bucuresti: Editura Stiintifica. Stan, Cornelia. 2009. “Complementul Direct.” Unpublished Ms. Tigău, Alina. 2008. “Towards an Account of Romanian Pronominal Clitics. An Overview of Clitic Doubling.” Analele UniversităĠii Bucureúti, 41–58. —. 2010. Syntax and Interpretation of the Direct Object in Romance and Germanic Languages with an Emphasis on Romanian, German, Dutch and English. Bucureúti: Editura UniversităĠii Bucureúti. Uriagereka, Juan. 1995. “Aspects of the Syntax of Clitic Placement in Western Romance.” Linguistic Inquiry 26(1):79–123.

CHAPTER EIGHT OLD ENGLISH TRANSITIVITY: A VIEW FROM THE PAST MAGDALENA CHARZYēSKA-WÓJCIK

1. Introduction There are two major variants of traditionally understood transitivity: broad and narrow.1 The former makes reference to the number of arguments, hence it will be referred to as quantitative, the latter is of a qualitative type, i.e., it is related to the presence of specific argument types. An instance of a quantitative approach to transitivity is exemplified by Bosworth and Toller’s classification, which will be discussed in section 2. A qualitative approach is represented by Visser’s (1963-73) typology of Old English verbs, which will be presented in section 3. As the two accounts are not only based on different defining criteria but also result in different classifications of particular verbs, the two approaches will be evaluated in section 4 with the help of independent Old English data. The concluding remarks will be presented in section 5.

2. Quantitative criterion The oldest study of OE transitivity is Bosworth and Toller’s (1898) AngloSaxon Dictionary and Toller’s (1921) Supplement2 to the dictionary. These 1

After Beedham (2010, 23 fn. 5), [i]n languages with a case system there is a narrow definition of the transitive verb which covers objects in the accusative case only, and a broad definition which incorporates objects in oblique cases, e.g., dative, genitive. 2 It should perhaps be explained that the Supplement does not offer systematic additions with respect to the main volume. Instead, it either corrects and/or

Old English Transitivity: A View from the Past

135

works do not constitute an account of OE transitivity per se but they offer detailed classifications of (most) OE verbs and provide them with illustrative examples. The picture that emerges from the analysis of the verbal entries available in these two sources leaves no doubt as to what property distinguishes transitive and intransitive verbs in OE. These two works are, therefore, a legitimate starting point of this discussion. The first interesting observation with regard to transitivity is that the very terminology used by the two dictionaries changes and the fact that it changes more or less half-way through the 1898 volume. In particular, while the first part of the 1898 volume resorts to the contrast between verbum activum (v.a.) and verbum neutrum (v.n.), the latter part of the volume and the 1921 supplement uniformly use the terms transitive and intransitive verb. It has to be emphasised that the change in terminology does not represent a change of focus or of the defining properties of transitivity in any way, as evidenced by the fact that the same verb is marked as verbum activum in the 1898 volume and as a transitive verb in the Supplement. The same applies to the terms verbum neutrum vs. intransitive verb. Similarly, the marking of verbs which appear both in transitive and intransitive structures in the Supplement corresponds to the verbum activum vs. verbum neutrum distinction in the first part of the dictionary. The terms verbum neutrum and verbum activum are slightly problematic and therefore require an explanation. Both are transplanted to the linguistic description of (and in) English from Latin grammars, where their significance is morphosyntactic: a verb is classified as active if its form contrasts with the passive verbal inflection. In effect, this is equal to the verb’s ability to appear in the passive voice – itself a clear indication that the verb can take an object. However, if a verb does not show the contrast, i.e., it never appears in the passive voice because it has no object, it is a neuter verb. The term active verb is first recorded in English in the early 16th century but it soon started to give way to a term introduced around the end of the same century, i.e., transitive verb; the term active is most frequently (though not exclusively) used as an adjective classifying voice. The term neuter verb is first recorded in English by the side of the term active verb in the early 16th century, roughly speaking to denote verbs not capable of appearing in the passive, and it was used in this sense until the 19th enriches the information given in the main volume or introduces entries which are not covered in the 1898 volume at all. Therefore, an analysis of any verb needs to encompass the information provided by both volumes.

136

Chapter Eight

century, when the term intransitive verb replaced it completely. Interestingly, the term intransitive verb started to be used in the early 17th century with reference to the same type of verbs, though here the focus was more on the syntactic aspect (‘verbs not taking a direct object’). It has to be pointed out, however, that when the term neuter verb was applied in grammatical descriptions of English, the criteria qualifying a verb as neuter differed widely. Samuel Johnson’s famous Dictionary of the English Language, first published in 1755, regularly employs the terms active and neuter verb. In the Preface, Johnson provides the following (very) brief description: English verbs are active, as I love; or neuter, as I languish. The neuters are formed like the actives. Most verbs signifying action may likewise signify condition or habit, and become neuters, as I love, I am in love; I strike, I am now striking.

In effect, according to Johnson (and five other authors), a verb formally qualifies as neuter if it does not have a past participle. There are, of course, problems inherent in such an approach (cf. Michael 1970, 382), which indicate that, while it works for a description of Latin verbs (except for deponentia), it cannot be adapted to a description of English verbs. Another renowned lexicographer and grammarian, the principal editor of the Oxford English Dictionary, Lindley Murray, uses the same terms, for which he provides the following definitions in his 1808 grammar:3 A Verb Active expresses an action, and necessarily implies an agent and an object acted upon, as: to love; “I love Penelope.” … A Verb Neuter expresses neither action nor passion, but being, or a state of being: as, “I am, I sleep, I sit.” The verb active is also called transitive, because the action passes over to the object, or has an effect on some other thing, as: “The tutor instructs his pupils;” “I esteem the man.” 3

It has to be clarified, however, that Murray (1808 Vol. I) distinguishes not two but three types of verbs: active, passive and neuter (in agreement with Latin grammars), which seems to invalidate the definitions provided for active and neuter verbs. The definition of passive verbs quoted after Murray (1808 Vol. I, 99) below, however, makes it clear that they do not form a separate group of verbs as such, but they are structures with verbs which do not belong to the neuter division. A Verb Passive expresses a passion, or a suffering, or the receiving of an action; and necessarily implies an object acted upon, and an agent by which the object is acted upon: as, to be loved: “Penelope is loved by me.” (Murray 1808 Vol. I, 99)

Old English Transitivity: A View from the Past

137

Verbs neuter may properly be denominated intransitives, because the effect is confined within the subject and does not pass over to any object: as, “I sit, he lives, they sleep.” (Murray 1808 Vol. I, 99-100)

As is clear, Murray’s definition equates active verbs (verba activa) with transitives and neuter verbs (verba neutra) with intransitives. Note that the defining properties are both semantic (whether a verb ‘expresses an action’ vs. ‘being, or a state of being’) and syntactic (the presence vs. absence of the object the action passes over to).4 Murray (1808 Vol. I, 99) also reports that some writers distinguish the following kinds of verbs: 1st. Active-transitive, or those which denote an action that passes from agent to some object; as Cæsar conquered Pompey. 2d. Active-intransitive, or those which express that kind of action, which has no effect upon any thing beyond the agent himself: as, Cæsar walked. 3d. Passive, or those which express, not action, but passion, whether pleasing or painful; as, Portia was loved; Pompey was conquered.

As if this is not confusing enough, other authors propose still different definitions of neuter verbs, ranging from classifications which could be summarised as expressing an “ambiguous relation to the passive” (for example, neuter verbs are neither active nor passive, if being active necessarily implies a contrast with active), verbs which “can have a nominative after them” or “express being” (Michael 1970, 388).5 In conclusion, the terms verbum activum and verbum neutrum, though perhaps not clearly defined, are clearly synonymous with the terms transitive verb and intransitive verb, at least in the way they are used by Bosworth and Toller. The complete interchangeability of the two sets of terms can be fully appreciated by comparing the entries of the same verb in the two parts of the dictionary, such as bregdan ‘to move to and from, turn into,’ efestan ‘to hasten, be quick,’ gælan ‘to hinder, hesitate,’ to name but a few. The verbs exhibit two types of uses, which are referred to 4 In a further part of the discussion (Vol. I, 100), however, it appears that the two criteria are not consistently applied: Some of the verbs that are usually ranked among neuters, make a near approach to the nature of a verb active; but they may be distinguished from it by their being intransitive: as, to run, to walk, to fly, &c. The rest are more obviously neuter and more clearly expressive of a middle state between action and passion: as, to stand, to lie, to sleep, &c. 5 As noted by Michael (1970), grammarians who used the term neuter verb can be divided on the basis of the defining criteria into as many as six different groups.

138

Chapter Eight

by a different set of terms, depending on the volume. An examination of the dictionary entries of these and other verbs of the same type makes it clear that the difference is purely terminological. In effect, while not bearing directly on the issue, the terminological shift and the problems connected with defining the concept point to two important observations. First of all, there seems little justification for the application of the term neuter verbs to English – even less so than there is in Latin. Secondly, on a more general level, it forces upon us the often carefully avoided question of how much of the grammatical description in Western linguistics is influenced, if not in fact distorted, by the adaptation of the Latin terms to descriptions of languages which exhibit altogether different grammatical contrasts.6 Leaving these issues aside, let us return to the two types of verbs recorded in the Anglo-Saxon dictionaries. In line with the terminological switch observed above, I will call them transitive and intransitive. An examination of the dictionary entries leaves no doubt as to the defining characteristics of the two verb types: verbs which are accompanied by a nominal object in any of the available verbal cases, i.e., accusative, dative or genitive, are classified as transitive (cf. (1) below), while a verb not accompanied by an object is classified as intransitive (cf. (2)). (1)

6

transitive verbs a. ACC Þonne ÿu hine cwelst when you him-ACC kill ‘When you kill him.’ Exod; B8.1.4.2 (CCOE)

One seventeenth-century grammar of English presents the English adjective in a chart with six cases, two numbers and three genders. As a result, it shows 36 cells with the same form of the adjective, which had become an uninflected category centuries before the grammar was written. That this was not an attempt at a diachronic description of the English adjective but a side-effect of imposing Latin categories on English is best indicated by the fact that English has never had as many as six different cases. The detrimental impact of grammatical description being so strongly Latin-based is also reported by Næss (2007), albeit she addresses a problem of the opposite nature: [t]raditional grammatical terminology and analysis has been, and is still being, continually challenged by the discovery of grammatical phenomena which do not seem to fit neatly into the categories inherited from Greek and Latin grammar. (Næss 2007, 2)

Old English Transitivity: A View from the Past

139

b. DAT Manige tiliaÿ Gode to cwemanne many strive God-DAT to please ‘Many people strive to please God.’ Or4; B9.2.5 (CCOE) c. GEN þæs his lufu bædeÿ this-GEN his love demands ‘His love demands that.’ Max I; A3.13 (CCOE) (2)

an intransitive verb and his blæda ne fealwiaÿ and its flowers not wither ‘And its flowers do not wither.’ PPs (prose); B8.2.1 (CCOE)

In effect, the defining parameter of transitivity within Bosworth and Toller’s classification embodies the most traditional interpretation of the notion: a purely quantitative requirement that a verb be accompanied by an NP object in an unspecified case. Therefore, according to Bosworth and Toller, a verb is transitive if it occurs with a nominal object, regardless of its case and a verb which appears without it is intransitive.7 An invaluable asset of this type of approach, or perhaps its raison d’être, is a peculiarity of Old English complementation, which shows a strong propensity for object case alternations. In other words, one and the same verb tends to take objects in different cases without exhibiting any change of meaning.8 Instances of such alternations can be found with all verb types and though 7

It has to be admitted that verbs accompanied by a prepositional phrase are not classified into any type. This omission is to be expected in such a binary formal approach to transitivity. Interestingly, Doroszewski’s (1963) definition of transitivity, which relies on exactly the same purely formal criteria introduces a subtype in order to accommodate verbs accompanied by prepositional phrases. In effect, according to Doroszewski, a verb is transitive if it is accompanied by a nominal object, intransitive if no object accompanies it, and indirectly transitive if it governs a prepositonal object. 8 The limitations of space do not allow me to expand further on that topic but as I show in CharzyĔska-Wójcik (2014), this means that the basic sense of the verb stays the same but the overall meaning of the clause is, according to some researchers (Plank 1982; Goh 2000; Toyota 2009), influenced by the choice of cases.

140

Chapter Eight

they may be more common with some verbs than with others, they are encountered across the whole system in all configurations, i.e., ACC/DAT, ACC/GEN, GEN/DAT, and ACC/DAT/GEN. Some examples of the alternations are shown below. (3)

ACC/DAT: blissian ‘to gladden’ a. ACC Ða se halga ongann hæleþ blissigean then the saint started man-ACC to-gladden ‘Then the saint started to gladden the man.’ And; A2.1 (CCOE) b. DAT swa þu, engla god, eallum blissast as you angels’ God all-DAT gladden ‘As You, the God of angels, gladden everybody.’ LPr II; A20 (CCOE)

(4)

ACC/GEN: bidan ‘to expect’ a. ACC hire sweostor… baad þone ecan sige her sister expected the eternal victory-ACC ‘He sister expected eternal victory.’ Bede 4; B9.6.6 (CCOE) b. GEN Nu hie softe þæs bidon now they quietly that-GEN expected ‘Now they quietly expected this.’ ChristA,B,C; A3.1 (CCOE)

(5)

ACC/DAT/GEN: wiðsacan ‘to reject’ Se fæder wiðsoc his bearne, the father rejected his child-DAT and ðæt bearn wiðsoc ðone fæder, and that child rejected the father-ACC and æt nextan ælc freond wiðsoc oðres and next each friend rejected other-GEN ‘The father rejected the child and that child rejected the father and in the end friends rejected each other.’ LS 34; B3.3.34 (CCOE)

The verbs in all the clauses quoted in (3)–(5) above are seen as transitive in this approach. Note that the non-discriminating nature of this classification, which does not rely on the type of case exhibited by a

Old English Transitivity: A View from the Past

141

nominal object, is most fully appreciated with reference to the verbs exhibiting object case alternations. This is summarised in Table 8-1 below. Table 8-1. Verb types in OE according to Bosworth and Toller Intransitive verbs V+0

Transitive verbs V + NP-ACC V + NP-DAT V + NP-GEN V + NP-ACC/DAT V + NP-ACC/GEN V + NP-GEN/DAT V + NP-ACC/DAT/GEN

The pervasive variability of transitive complementation patterns contrasts sharply with the single intransitive type and naturally provokes the following set of questions: (6)

i.

Do OE intransitive (monovalent) verbs really represent a homogenous pattern? ii. What is the mutual relationship between transitive and intransitive structures with the same verb, as in the case of bregdan, efestan, gælan discussed above? iii. What are the differences between the many transitive patterns shown in Table 8-1?

As will soon be clear, points (6i) and (6ii) can receive a common answer, while point (6iii) requires a separate treatment. It has already been mentioned that some OE verbs can be used both transitively and intransitively with the effect that they are classified as transitive in some uses and as intransitive in others. These valency alternations qualify as alternations of the type discussed in Levin (1993). Therefore, it seems that the type is not homogenous and requires the introduction of finer internal distinctions. This issue falls beyond the scope of the present chapter but an investigation into this type of alternation is offered in van Gelderen (2011). In addition, some transitive verbs can appear without an object which they normally require if the object is sufficiently clear from the context. Such cases need to be set apart from

Chapter Eight

142

the former type, as they represent elliptical structures (also referred to as object drop).9 It can, therefore, be concluded that OE monovalent structures do not represent a homogenous type. Instead, they fall into (at least) as many as three different subtypes: (7)

i. intransitive verbs which never appear with an object; ii. verbs which show both transitive and intransitive uses, representing classifiable alternations of the type discussed in Levin (1993) and van Gelderen (2011); iii. transitive verbs used in elliptical structures.

In effect, both transitive and intransitive clauses represent various subtypes and any account of OE transitivity should be able to properly express the mutual relationships between them. Bosworth and Toller’s modest theoretical apparatus naturally lacks the necessary formal means, not being a classification per se but it has to be admitted that it has a way of setting apart transitive uses of bivalent verbs from elliptical structures. In particular, some uses of verbs otherwise classified as transitive are marked as absolute. These are illustrated in (8)–(10) below, which contain quotations listed in Bosworth and Toller under the relevant classifications. The actual text, however, is given in the form in which it appears in the Complete Corpus of Old English to ensure consistency in the format of the data quoted. (8)

a. ACC Ic dealf ðisne pytt. I dug this hole-ACC ‘I dug this hole.’

Gen. 21, 30. (B&T)10

b. absolute Þa dulfon hi in þere ylcan stowe. when dug they in the same place ‘When they dug in the same place.’ Shrn. 113, 13. (BTs) 9

A more insightful discussion of the type is offered in Visser (1963-73) and CharzyĔska-Wójcik (2013). 10 The abbreviations B&T and BTs placed next to the primary sources identify the part of the dictionary which a given quote comes from, with the former indicating the main volume of the dictionary and the latter the supplement volume.

Old English Transitivity: A View from the Past (9)

143

a. DAT Sum sceal on heape hæleþum cweman. one shall in company men-DAT delight ‘One shall in company give pleasure to men.’ Exon. 88a; Th. 331, 33; Vy. 77. (B&T) b. absolute Ic cweme I delight ‘I am happy.’ Wrt. Voc. ii. 12, 44. (BTs)

(10)

a. GEN His frynd þæs micclum wundrodon and blissodon. wondered-at and rejoiced his friends that-GEN much ‘His friends very much wondered at this and rejoiced it.’ Hml. Th. ii. 26, 11. (BTs) b. absolute He… sæt mid þam gebeorum blissigende samod. he sat with the guests rejoicing together ‘He sat with the guests rejoicing.’ Hml. S. 26, 327. (BTs)

While the very term absolute is, again, one with a long and confused history in the description of Latin, let us only remark that it is used in Bosworth and Toller to signify an object-less structure with a transitive verb, i.e., to mark elliptical uses (which does not fully correspond to Priscian’s definition of the term). In effect, Bosworth and Toller’s classificatory system is capable of differentiating (even if only terminologically) between transitive verbs that are used transitively and those that are used absolutely and can formally mark verbs of dual membership. It does not, however, articulate the differences between the subtypes of transitive verbs, implying thereby that they represent a homogenous group. This brings us to question (6iii) posed above, which will, therefore, have to remain unanswered for the time being. I will return to it in section 3, where I will show that treating these verbs as a homogenous group is an oversimplification with implications producing wrong empirical results for OE. Having looked at the advantages and disadvantages of Bosworth and Toller’s treatment of transitivity with respect to the Old English data, I would like to evaluate the theoretical status of transitivity defined solely with respect to the quantitative parameter. Despite important advances in the study of transitivity viewed as a notion of cross-linguistic validity, one

144

Chapter Eight

can still meet in the current literature definitions of transitivity which rely on the quantitative criterion. This reflects the assertion that all human languages classify actions into two basic types: those involving one obligatory participant, which are described by intransitive sentences, and those involving two obligatory participants, which are dealt with by transitive sentences. (Dixon 1979, 102)

The status of this criterion is discussed in LaPolla, Kratochvíl, and Coupe (2011). In their evaluation of syntactic (as opposed to semantic) definitions of transitivity, where a quantitative interpretation of the notion certainly belongs, they note what follows. The traditional syntactic definition of transitivity says that a language has one or more constructions where two arguments are given special status in the clause as core (obligatory) arguments, as opposed to only one argument being given that status. This is straightforward, but defining transitivity in this way doesn’t help us understand very much about the language given the circularity of identifying a clause as transitive because it has two arguments, and saying that it has two core arguments because it is a transitive clause. The traditional view also does not recognise the diversity of morphosyntactic phenomena that show that clauses with two core arguments are not all alike … . (LaPolla, Kratochvíl, and Coupe 2011, 471)

In conclusion, the major empirical assets of Bosworth and Toller’s quantitative approach to OE transitivity consist in: (i) accommodating the variability of the case-marking of verbal objects; (ii) distinguishing between intransitive (absolute) uses of transitive verbs and: a. genuinely intransitive verbs, b. verbs of dual membership, referred to as ambitransitive or labile. As for its drawbacks, as already indicated, these are both of a theoretical and an empirical nature. First of all, defining transitivity solely on the basis of the number of core arguments is, as pointed out above, circular. Secondly, Bosworth and Toller’s bi-partite division into transitive and intransitive verbs fails to differentiate between various subtypes of OE transitive verbs – a distinction which will be brought to light in section 3.

3. Qualitative criterion Another type of traditional approach to transitivity, though with a slightly different focus, takes the presence of a direct object as the defining

Old English Transitivity: A View from the Past

145

property of transitive structures. It is represented by Visser (1963-73), whose An Historical Syntax of the English Language constitutes the only attempt at a full, independent classification of OE verbs. Another seminal work devoted to Old English syntax, Mitchell (1985), relies heavily on Visser’s classification, introducing only minor modifications. Similarly, Ogura’s (1996) classification of OE verbs is based on Visser’s typology. Moreover, Ogura focuses more on semantic groupings than on formal aspects. No other authors have explicitly dealt with OE transitivity in any holistic way on such a grand scale or attempted a classification of OE verbs into the relevant types, though there have been a few isolated attempts at interpreting case variation in OE from the perspective of transitivity (cf., for example, Plank 1982; Goh 2000; Toyota 2009; van Gelderen 2011). In spite of these, Visser’s view on OE transitivity remains the definitive word on the matter. As signalled by the title of the section, Visser’s approach to transitivity relies on the presence of an NP object but, in contrast to Bosworth and Toller’s classification, it differentiates between NP object types, to the effect that it is only the presence of a direct object that makes a clause transitive. Visser’s definition of the term direct object is as follows: The traditional term ‘direct object,’ in which the meaning of the adjective ‘direct’ is obscure, has been kept here for want of a better. In Old English the direct object is easily recognizable since it is the (pro)noun in the accusative complementing the subject + verb group. … Grammarians have often tried to give a definition based on the kind of notional or logical relation between this complement and its verb, but so far these attempts have been futile, the relations being too multifarious and too heterogeneous to be comprehended in one single term. Only in a few special cases is the direct object “the person or thing to which something is done” (Onions); “the receiver of the action” (Kittredge); “the thing directly affected by the action” (Grattan); “the goal, the object, actually hit or affected by the activity, or thing representing the goal, the real object of the activity” (Curme); “the sufferer(s) of the action” (Stokoe). (Visser 1963-73, §418)

In consequence, a verb accompanied by an indirect object is classified by Visser as intransitive. The crucial aspect of transitivity, then, is what qualifies as a direct object? As shown above, Visser draws a dividing line between objects in the accusative case on the one hand and dative or genitive objects on the other: only the former count as direct objects.11 So 11 In accordance with Visser’s definition of transitivity, Brody (1989), when discussing OE impersonal constructions, says that lician ‘to like’ could not appear

Chapter Eight

146

only verbs with accusative objects are classified as transitive in Visser’s typology, while verbs whose objects are dative or genitive are intransitive, just as verbs which do not take an object of any kind. However, Visser remains silent on the matter of case alternations exhibited by OE verbs and does not classify verbs which show variable object cases. In consequence, the aspect of structure which is crucial in defining transitivity remains underspecified and, while it is clear how to classify verbs with only accusative objects on the one hand, and those with exclusively nonaccusative objects on the other, it is not altogether obvious how to treat objects which exhibit alternative accusative vs. non-accusative case marking. This is summarised in Table 8-2 below. Table 8-2. Visser’s verb types Intransitive verbs

?

V+0

V + NP-ACC/DAT

V+ NP-DAT

V + NP-ACC/GEN

V + NP-GEN

V + NP-ACC/DAT/GEN

Transitive verbs V + NP-ACC

V + NP-GEN/DAT

In effect, it might seem that Visser’s classification does not focus on a verb with the full array of its complementation frames but on individual structures. This impression, however, vanishes when one is confronted with Visser’s classification of “syntactical units in Old English that consist of subject + verb without further complement” (§129), where he distinguishes as many as four different types of verbs, depending on whether these verbs can take an object and, if so, of what type: direct or indirect.12 What has to be added, however, in defence of this classification is that it is only seemingly tailor-made for Old English: in fact, it is more of a diachronic account of the changes to the system. Since later changes to the English language eliminated case distinctions, the variability of case in transitive constructions, i.e., “it could not appear with a nominative experiencer argument and a theme in the accusative case” (Brody 1989, 262). However, in contrast to Visser’s formulation, Molencki (1991) talks about the direct object in OE in the dative or accusative case, clearly indicating the equal status of the two cases. The presence of these contrasting interpretations of what is the direct object in OE shows that the terms direct object and, in consequence, transitivity are understood differently even with reference to the same language. 12 Cf. CharzyĔska-Wójcik (2011, 2013) for detailed comments on Visser’s groupings presented in his §129.

Old English Transitivity: A View from the Past

147

marking is attested only in OE and eME. As a result, the direct vs. indirect object alternations, which are crucial in defining transitivity in OE, do not constitute a valid aspect of structure when later states of English are examined. Since Visser’s focus is diachronic, the OE data are not viewed from a purely synchronic perspective. Instead, their interpretation is strongly influenced by the later stages of the system, where case alternations do not play a part in defining transitivity. In conclusion, Visser defines transitivity with respect to the notion of the direct object, without taking a stand on the matter of case alternations, inherent in the system. In effect, his classification works only for verbs with non-alternating object cases or for individual structures. This kind of underspecification of a crucial notion is a recurring theme in qualitative definitions of transitivity. It is reported by Lapolla, Kratochvíl, and Coupe (2011, 470), who remark that, while standard definitions of transitivity involve the notion of the direct object, “[n]othing is said in these definitions about what a direct object is and how to identify it.” A subtype of this kind of qualitative (rather than quantitative) interpretation of transitivity, though in a different theoretical guise, is represented by approaches resorting to S A P13 in defining transitivity. S denotes the subject of an intransitive clause, while A and P are relations within a monotransitive one. As noted by Haspelmath (2011, 535), these terms “are often taken for granted,” but “there are substantial differences in the literature in the way these terms are understood” (Haspelmath 2011, 538). This is (at least partly) due to the fact that there are no universal criteria for identifying transitive clauses. Precise criteria defining (in)transitivity can often be found in individual languages, but transitivityrelated phenomena are so diverse that these criteria “are not generalizable across languages” (Haspelmath 2011, 542). What transpires from the above is this: if transitivity is defined by the presence of a particular category in a particular language, then classifying 13

The full set of terms is: S, A, P (or O), T, R (or G) but only S, A and P are relevant for our discussion here, while T and R (alongside A) represent relations within ditransitive clauses. The terms appeared in the linguistic literature in the 1970s as tools of comparative linguistics. With time, they were adopted by descriptive linguistics, and the use of the same categories in descriptive and comparative studies has resulted in the categories taking on different meanings. As noted by Haspelmath (2011, 536), there are now three major types of approaches which apply the term: “the Dixonian approach (Dixon 1994, 2000), the Comrian approach (Comrie 1989; Lazard 2002) and the Bickelian approach (Bickel and Nichols 2009; Bickel 2011).”

148

Chapter Eight

clauses in this language as transitive on the basis of the presence of this category is, in effect, perfectly circular. Another very unwelcome reflection following from the same set of observations is whether by applying different criteria to different phenomena approached from different perspectives linguists have not, in effect, defined a different category? Therefore, it seems that, if transitivity is to be a valid notion in OE, we need to see what it entails.

4. Passivisation as an indicator of transitivity The availability of passivisation, though very imperfect as a universal diagnostic, is normally associated with transitivity.14 Under the most traditional understanding of transitivity, it is the ability to appear in the passive voice (verbum passivum) that qualified a verb as verbum activum. As noted above, the term verbum activum was replaced with transitive verb, but its defining property did not change. If the contrast was not available for a given verb (i.e., the verb was unable to appear in the passive voice), it was classified as verbum neutrum – again, a term which gave way to a newer denotation, i.e., intransitive verb. This basic distinction is still recognised in the modern literature on transitivity, as shown below. Passivization makes it in many (but not all) cases possible to separate transitive clauses from less transitive ones, since ... only clauses conceived of as somehow transitive are to be passivized in many languages. The acceptability of passivization correlates to some extent with transitivity: the more transitive a clause is, the more readily it can be passivised. (Kittilä 2002, 23)

Nowadays, it is mainly the semantic understanding of transitivity that is directly correlated with passivation (see, for example, Toyota 2009, 11 and de Mattia-Viviès 2009, 105). However, transitivity defined semantically correlates with the presence of a passivisable object and, in this sense, it can serve as a diagnostic for our OE data: since our problem is precisely the lack of a reliable definition of transitivity, we can try using the passivisation test to see how it divides OE verbs into passivisable and 14

This is of course an oversimplification, especially in view of the fact that, as noted by Siewierska (1984), there is not a single property of passive clauses which is universally attested.

Old English Transitivity: A View from the Past

149

non-passivisable ones and whether this division correlates with the transitive vs. intransitive distinction. Note that my reasoning here is not circular: I am not defining transitivity with respect to passivisation. My intention is to see how passivisation relates to the two approaches to transitivity presented in sections 1 and 2. Let me begin with the predictions. If transitive verbs are understood in the sense of Bosworth and Toller (a broad, quantitative definition, by which a verb is transitive if accompanied by an NP object regardless of its case), we expect all verbs which can take an NP object to be passivisable. If, on the other hand, transitivity is viewed in Visser’s terms (a narrow qualitative definition, where only verbs with direct objects are transitive), only verbs which take accusative objects are expected to undergo passivation in OE. An examination of the passivisation of OE verbs, as shown in Mitchell (1985), CharzyĔska-Wójcik (2002) and Quinn (2005), indicates that, while the major division between verbs which can passivise and those that cannot runs along the lines of Bosworth and Toller’s classification, a secondary division within passivisation reflects Visser’s understanding of the notion. In particular, the criterion which qualifies an OE verb for passivisation is its ability to take an NP object, regardless of the actual case of the object. In other words, verbs classified by Bosworth and Toller as transitive (or active) can passivise. However, verbs taking an accusative object (i.e., transitive verbs under Visser’s classification) produce personal passives with a correspondence between the active accusative object and the nominative subject in the passive, as shown in (11) below. In contrast, verbs which take non-accusative objects (in the dative or genitive case) produce impersonal passives, in which the dative or genitive argument which shows up in the active voice remains unchanged under passivisation and the resulting clause has no nominative subject, hence the 3SG form of the verb, as shown in (12) and (13).15

15 It needs to be borne in mind that some verbs which were never accompanied by accusative objects in the earlier part of the period started to appear with accusative NPs in late OE. In consequence, the type of passive these verbs produced changed from the impersonal to the personal type. If the diachronic change of the complementation pattern were ignored, one might get a mistaken impression that verbs which assigned the accusative to the object could appear in impersonal passives.

150 (11)

Chapter Eight personal passivisation a. active clause with ACC And he eac ofsloh æÿele cyningas, and he also killed noble kings-ACC ‘And he also killed noble kings.’ PPs; A5 (CCOE) b. personal passive Næs seo eadige maria na ofslegen not-was the blessed Mary-NOM not-at-all killed ne gemartyrod lichomlice: Ac gastlice; nor martyred physically but spiritually ‘The blessed Mary was not killed or martyred physically but spiritually.’ ÆCHom I, 9; B1.1.10 (CCOE)

(12)

impersonal passivisation with a dative NP a. active clause with DAT þonne god demeÿ manna gehwylcum be ærran gewyrhtan. by earlier merits then God will-judge of-men each-DAT ‘Then God will judge each man by his earlier merits.’ HomU 32 (Nap 40); B3.4.32.3 (CCOE) b. impersonal passive Siÿÿan hi bioÿ læded to þam heofone and to þam soon they are led to the heaven and to the heahsetle þære halgan Ĉrinnisse, and him biÿ þær demed. throne of-the holy Trinity and them-DAT is there judged ‘They are soon led to the heaven and to the throne of the holy Trinity and they are judged there.’ HomU 12.2 (Willard); B3.4.12.2 (CCOE)

(13)

impersonal passivisation with a genitive (and dative) NP a. active clause with GEN (and DAT) For ÿæm þu him sealdest his modes willan, because you him granted his spirit’s wish and þæs þe he mid his weolorum wilnade, and that which he with his lips asked-for þæs þu him ne forwyrndest. that-GEN you him-DAT not refused ‘Because, you granted him the wish of his spirit and you did not refuse to him what he asked for with his lips.’ PPs (prose); B8.2.1 (CCOE)

Old English Transitivity: A View from the Past

151

b. impersonal passive and him wæs swa forwyrned ÿæs inganges syÿÿan. the entrance-GEN soon and him-DAT was so refused ‘And he was soon refused entry.’ ÆHex; B1.5.13 (CCOE)

Note that it is also necessary to see how the verbs not classified by Visser behave under passivisation. As shown in Mitchell (1985), verbs which optionally take an accusative object tend to produce personal passives: (14)

a. active with ACC þu hæfst afandod min mod, you have tested my spirit-ACC ‘You have put my spirit to trial.’ PPs (prose); B8.2.1 (CCOE) b. active with GEN Þa ÿa crist axode philippum and he his afandode when Christ asked Philip and he him-GEN tested ‘When Christ tested Philip and put him to trial.’ ÆCHom I, 12; B1.1.13 (CCOE) c. personal passive swa swa gold biÿ on fyre afandod; just as gold-NOM is in fire tested ‘Just as gold is put to trial in fire.’ ÆCHom I (Pref); B1.1.1 (CCOE)

In effect, the difference between personal vs. impersonal passivisation derives from a difference between object types, i.e., it is based on Visser’s contribution to the notion of OE transitivity. It transpires that objects which can (at least potentially) be assigned the accusative case contrast with objects incapable of bearing the accusative. Regardless of whether the two different object types are formally referred to as direct and indirect, the fact remains that there is a distinction between the two object types which is clearly visible under passivisation: accusative objects form personal passives, while non-accusative objects are correlated with impersonal ones. In conclusion, OE passivisation shows that each of the two approaches single-handedly produces the wrong empirical results but a combination of the two interpretations of transitivity accurately accounts for the specificity of OE passivisation.

152

Chapter Eight

5. Concluding remarks In this chapter, I have presented two traditional accounts of OE transitivity. They represent a broad and narrow understanding of the notion and, as I have shown, each of them has its strengths and weaknesses. The former, i.e., the broad quantitative interpretation of Bosworth and Toller, relies on the number of nominal arguments of the verb in defining transitivity. It has been shown to accommodate the pervasive variability of OE object case alternations by classifying a verb as transitive if it is accompanied by an NP object regardless of its case. In contrast, the narrow, i.e., qualitative approach to transitivity of Visser relies on the presence of the direct object in defining transitivity and by direct object Visser means only the NP object formally marked with the accusative case. The variability of object cases is not tackled at all in this approach. The two approaches have been evaluated on theoretical and empirical grounds. Both are viewed in the current linguistic literature as insufficient and circular – a conclusion only to be expected in view of the more recent advances in linguistic theory and description. As for their empirical power, each of them separately is shown to make wrong predictions with respect to the peculiarities of OE passivisation. However, the OE facts can be readily explained by appealing to the defining parameters of transitivity of both of the two classifications jointly. In particular, the quantitative classification separates verbs which can be passivised (transitive) from those that cannot (intransitive); the results of the qualitative classification coincide with the division of verbs between those which form personal passives (transitive verbs with a direct object in the accusative case) and those which form impersonal passives (intransitive verbs with an indirect object in the dative or genitive case). An important point where the two views can be shown to converge, although not related to passivisation, is strictly connected with another syntactic operation which affects the number of arguments in a clause, namely object dropping. The treatment of elliptical structures is the same in both approaches. Bosworth and Toller on the one hand, and Visser on the other recognise the tendency of OE verbs to drop the NP object if it is sufficiently implied in the context. The relevant structures are referred to in both sources as absolute. In addition to that, both classifications recognise the existence of verbs which show transitive and intransitive uses. However, Bosworth and Toller classify the different structures as representing the same verb (transitive and intransitive within one dictionary entry), while Visser

Old English Transitivity: A View from the Past

153

(§129) treats them as “etymologically related homonym[s],” i.e., as different verbs, thereby denying a syntactic relationship between the two structures.16 Note, however, that neither Bosworth and Toller, nor Visser approach the question of what is encoded in the differences exhibited by object cases, i.e., the question raised in (6iii) remains unanswered. Concluding, some of the phenomena related to transitivity are properly accounted for in either of the two theories individually or in both of them jointly, but the problem of object case alternation is not even addressed by either of them. Importantly, if transitivity is related to cases, and on the basis of passivisation we have seen that it is, this dimension cannot be ignored in a classification of OE verbs. It would, therefore, be interesting to see how the current linguistic models with their refined theoretical apparatus fare with respect to OE transitivity. This issue, however, falls beyond the scope of the present chapter but is exhaustively dealt with in CharzyĔska-Wójcik (2014).

References Beedham, Christopher. 2010. “The Equivocation of Form and Notation in Generative Grammar.” In Chomskyan (R)evolutions, edited by Douglas A. Kibbee, 19–42. Amsterdam and Philadelphia: John Benjamins. Bickel, Balthasar. 2011. “Grammatical Relations Typology.” In The Oxford Handbook of Linguistic Typology, edited by Jae Jung Song, 399–444. Oxford: OUP. Bickel, Balthasar, and Johanna Nichols. 2009. “Case Marking and Alignment.” In The Oxford Handbook of Case, edited by Andrej Malchukov and Andrew Spencer, 304–21. Oxford: OUP. Bosworth, Joseph, and Thomas Northcote Toller. 1898. An Anglo-Saxon Dictionary Based on the Manuscript Collections of the Late Joseph Bosworth, Edited and Enlarged by T. Northcote Toller. London: Clarendon Press. Brody, Michael. 1989. “Old English Impersonals and the Theory of Grammar.” UCL Working Papers in Linguistics 1:262–94.

16

The interpretation of this type of alternation in the current linguistic literature agrees with Bosworth and Toller’s treatment (cf. van Gelderen 2011 and CharzyĔska-Wójcik 2011, 2013).

154

Chapter Eight

CCOE = Cameron, Angus, and Roberta Frank, eds. Complete Corpus of Old English: The Toronto Dictionary of Old English Corpus. University of Toronto Centre for Medieval Studies. The Oxford Text Archive. http://ota.ahds.ac.uk/ CharzyĔska-Wójcik, Magdalena. 2002. “The Impersonal Passive in Old and Middle English.” Beyond Philology 2:35–51. —. 2011. “Reflections on Structural Variation in Old English Verbs.” In New Perspectives in Language, Discourse and Translation Studies, edited by Mirosáaw Pawlak and Jakub Bielak, 137–51. Heidelberg, Dordrecht, London and New York: Springer. —. 2013. “The Salient Silence – Against a Unique Membership-Based Typology of Old English Verbs.” In PASE Papers in Linguistics, edited by Marcin Kleban and Ewa Willim, 37–53. Kraków: Wydawnictwo Uniwersytetu JagielloĔskiego. —. 2014. “Transitivity à la Old English.” In Language Change: Faces and Facets, edited by Magdalena CharzyĔska-Wójcik, Jerzy Wójcik, and Anna Bloch-Rozmej, 43–75. Lublin: Wydawnictwo KUL. Comrie, Bernard. 1989. Language Universals and Linguistic Typology: Syntax and Morphology. 2nd edition. Chicago: University of Chicago Press. Curme, George Oliver. 1931. A Grammar of the English Language. In Three Volumes. Volume III: Syntax. Boston and New York: D.C. Heath and Company. Dixon, Robert Malcolm Ward. 1979. “Ergativity.” Language 55:59–138. —. 1994. Ergativity. Cambridge: CUP. —. 2000. “A Typology of Causatives: Form, Syntax and Meaning.” In Changing Valency: Case Studies in Transitivity, edited by Robert Malcolm Ward Dixon and Alexandra Yurievna Aikhenvald, 30–83. Cambridge: CUP. Doroszewski, Witold. 1963. Podstawy Gramatyki Polskiej. CzĊĞü 1. Wydanie 2. Warszawa: PWN. Gelderen, Elly van. 2011. “Valency Changes in the History of English.” Journal of Historical Linguistics 1(1):106–43. Goh, Gwang-Yoon. 2000. “Alternative Case Markings in Old English Texts.” English Studies 3:185–98. Grattan, John Henry Grafton, and Percival Gurrey. 1925. Our Living Language: A New Guide to English Grammar. London and Edinburgh: T. Nelson and Sons, Ltd. Haspelmath, Martin. 2011. “On S, A, P, T, and R as Comparative Concepts for Alignment Typology.” Linguistic Typology 15(3):535– 67.

Old English Transitivity: A View from the Past

155

Johnson, Samuel. Preface to A Dictionary of the English Language: A Digital Edition of the 1755 Classic by Samuel Johnson, edited by Brandi Besalke. Last modified December 16, 2013. Accessed February 20, 2014. http://johnsonsdictionaryonline.com/?page_id=8. Kittilä, Seppo. 2002. Transitivity: Towards a Comprehensive Typology. Turku: University of Turku Press. Kittredge, George Lyman, and Frank Edgar Farley. 1913. An Advanced English Grammar: With Exercises. Boston, New York, Chicago and London: Ginn and Company. Lapolla, Randy John, František Kratochvíl, and Alexander Robertson Coupe. 2011. “On Transitivity.” Studies in Transitivity: Insights from Language Documentation, special issue of Studies in Language 35(3):469–92. Lazard, Gilbert. 2002. “Transitivity Revisited as an Example of a More Strict Approach in Typological Research.” Folia Linguistica 36(3-4):141–90. Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation. Chicago and London: The University of Chicago Press. Mattia-Viviès, Monique de. 2009. “The Passive and the Notion of Transitivity.” Review of European Studies 1(2):94–109. Michael, Ian. 1970. English Grammatical Categories and the Tradition to 1800. Cambridge: CUP. Mitchell, Bruce. 1985. Old English Syntax. Oxford: Clarendon Press. Molencki, Rafaá. 1991. Complementation in Old English. Katowice: Uniwersytet ĝląski. Murray, Lindley. 1808. An English Grammar: Comprehending the Principles and Rules of the Language, Illustrated by Appropriate Exercises and a Key to the Exercises. In Two Volumes. York: Thomas Wilson and Son. Næss, Åshild. 2007. Prototypical Transitivity. Amsterdam and Philadelphia: John Benjamins. Ogura, Michiko. 1996. Verbs in Medieval English. Differences in Verb Choice in Verse and Prose. Berlin: Mouton de Gruyter. Onions, Charles Talbut. [1905] 1927. An Advanced English Syntax Based on the Principles and Requirements of the Grammatical Society. 4th edition. London: Swan Sonnenschein and Co. Plank, Frans. 1982. “Coming into Being among the Anglo-Saxons.” Folia Linguistica 16:73–118. Quinn, Heidi. 2005. The Distribution of Pronoun Case Forms in English. Amsterdam and Philadelphia: John Benjamins.

156

Chapter Eight

Siewierska, Anna. 1984. The Passive: A Comparative Linguistic Analysis. London: Croom Helm. Stokoe, Henry Robert. 1937. The Understanding of Syntax. London: W. Heinemann. Toller, Thomas Northcote. 1921. An Anglo-Saxon Dictionary Based on the Manuscript Collections of the Late Joseph Bosworth. Supplement. London: Clarendon Press. Toyota, Junichi. 2009. “Passive as an Indicator of Alignment Change in Terms of Transitivity.” Groninger Arbeiten zur Germanistischen Linguistik 49:41–52. Visser, Frederick Theodore. 1963-73. An Historical Syntax of the English Language. Leiden: Brill.

CHAPTER NINE THE IMPORTANCE OF ROOT INDICES: EVIDENCE FROM ARABIC NOUNS∗ ABDULLAH ALGHAMDI⊥

1. Introduction The nature of roots has been one of the most controversial issues in the literature of Distributed Morphology (DM) (Halle and Marantz 1993, 1994). In DM, roots are devoid of any phonological or semantic representations in the computational system. They receive their phonological forms, known as Vocabulary Items (VIs), and semantic meanings later in the derivation process. However, the way in which originally indistinct roots receive distinct VIs and meanings was not made clear in the early works on the theory. This has stimulated further investigations into the nature of roots, the ultimate aim of which is an explanation of how the choice of the appropriate root VIs and meanings is determined. The present study contributes to this issue by providing evidence from Standard Arabic (SA) nouns supporting the view that roots are better identified by indices (Pfau 2009; Harley 2011) than by compositional semantic features (CSFs) such as [±Animate] and [±Count] (Marantz 1995, 1998). It argues against Marantz’s view by showing that CSFs cannot be the property of roots, let alone identify them. Consequently, this emphasises the importance of indices as guiding and identifying elements. ∗

I would like to thank my supervisors Dr. Hugues Peters and Dr. Mengistu Amberber for their guidance and helpful feedback, and Dr. Abdel El Hankari for his precious comments. ⊥ PhD candidate at the Faculty of Arts and Social Sciences, University of New South Wales, Australia; Lecturer at the Faculty of Arts and Humanities, Albaha University, Saudi Arabia.

158

Chapter Nine

In addition, this study will show that the role that indices play in the derivation process is not limited to the choice of root VIs and meanings. Indices can also motivate some language-specific morphological modifications. This study will provide evidence supporting this claim by analysing the structural differences between “affixal” and “templatic” plural nouns in SA.

2. Theoretical background This section provides an overview of the derivation process introduced by DM and the theoretical proposals that are relevant to the discussions in this chapter. Section 2.1 outlines the architecture of the grammar and the major principles of DM. Section 2.2 shows how stem formation is acknowledged in the theory. Finally, section 2.3 sketches out two different proposals on how the appropriate representations of roots are determined, especially in the Vocabulary. One proposal states that the choice is conditioned by the CSFs of the roots (Marantz 1995, 1998) and the other claims it is conditioned by the indices (Pfau 2009; Harley 2011).

2.1. The architecture of DM DM is a theory that makes a sharp distinction between morphosyntactic features, phonological representations and semantic realisations. According to this view, a plural noun such as cows consists of three layers: the morphosyntactic layer ([ROOT [+Plural]]), the phonological layer (/kaw + z/) and the semantic layer, which is simply the meaning that the noun can convey. In DM, handling these layers is a responsibility that is distributed across five components (Halle and Marantz 1993, 1994; Marantz 1997; Halle 1997; Harley and Noyer 1999, 2000; Embick and Noyer 2007): a. Narrow Lexicon is the component in which morphosyntactic features are stored as morphemes. In this component, morphemes are taken to be devoid of any phonological or semantic information. Early works on DM distinguish between functional morphemes (e.g., [Plural] and [–Definite]) and root morphemes (traditionally represented as [¥ROOT]). This distinction is based on how the morphemes of each type are filled with their phonological representations, as will be explained in (d) below.

The Importance of Root Indices: Evidence from Arabic Nouns

159

b. Syntax is the universal engine that builds hierarchical syntactic structures by merging morphemes supplied by the Narrow Lexicon. c. Morphology is the component in which the functional morphemes of syntactic structures can be modified according to the rules of the language. Morphosyntactic features of a morpheme can be deleted (Impoverishment), split into more than one morpheme (Fission), or merged with another morpheme (Fusion). d. Vocabulary is the storehouse of the phonological representations available in the language (VIs). In this component of DM, each functional morpheme, whether it is impoverished, fissioned, fused or unchanged, is expressed by one VI, in what is known as Vocabulary Insertion. When two or more VIs are fully or partially specified for the terminal functional morpheme, they start a competition for insertion. The winning VI is then determined by the Subset Principle: The phonological exponent of a Vocabulary item is inserted into a morpheme in the terminal string if the item matches all or a subset of the grammatical features specified in the terminal morpheme. Insertion does not take place if the Vocabulary item contains features not present in the morpheme. Where several Vocabulary items meet the conditions for insertion, the item matching the greatest number of features specified in the terminal morpheme must be chosen. (Halle 1997, 427)

As to the insertion of root VIs, it has been traditionally believed that they are not in competition for insertion. The choice of the appropriate root VI is a language-specific matter that might be conditioned by the CSFs of the root morpheme, the entire syntactic structure and the speaker’s communicative intents (Marantz 1995, 1998; Harley and Noyer 1999). This issue will be addressed in greater detail later in this chapter. e. Encyclopedia is composed of semantic entries that give the syntactic structures their meanings. The meanings can be canonical and assigned according to the root itself, e.g., cats or dogs, or noncanonical, that is, assigned according the position of the root in the sentence, e.g., it is raining cats and dogs.

2.2. Stem formation in DM From a DM perspective, words are the outcomes of merging category-less roots with category-assigning heads. For example, cat is a complex noun composed of a category-less root morpheme merged with a category-

Chapter Nine

160

assigning [n] morpheme, of which the latter happens to be a zero-exponent morpheme that surfaces as Ø, as shown below (Marantz 1997, 1998, 2001; Arad 2003, 2005; Harley 2005; Embick and Noyer 2007; Embick and Marantz 2008, among others): (1)

NP

[¥ROOT] /kæt/

[n] Ø

[n] morphemes are not always a zero-exponent in English. There are cases where an [n] morpheme surfaces with an overt phonological representation. For example, in a noun such as teacher, a category-less root morpheme is merged with a category-assigning [n] morpheme that surfaces as /ԥr/, as shown below: (2)

NP

[¥ROOT] /ti‫ޝ‬t‫ݕ‬/

[n] /ԥr/

This mechanism of stem formation provides the current analysis with an important tool for investigating the location of CSFs in SA nouns, as will be addressed in section 3.

2.3. The determination of root representations As mentioned earlier, roots are taken to be indistinct morphemes in the Syntax that receive different representations in the Vocabulary. This raises a significant question about how each representation is determined in the Vocabulary. There are two major claims about how root VIs are chosen for insertion. Firstly, according to Marantz, CSFs are present in the Syntax, and they can condition the insertion of VIs:

The Importance of Root Indices: Evidence from Arabic Nouns

161

The syntactic and compositional semantic features that condition the insertion of Vocabulary items at PF are taken to be involved in the grammar qua computational system … . (Marantz 1995, 403)

Marantz did not specifically mention whether CSFs are present in the Syntax as the property of root morphemes or other morphemes. It is not clear whether they condition the insertion of root VIs or something else. However, Marantz (1998, 8) clearly stated that CSFs are the property of root morphemes; consequently, they condition the insertion of root VIs in the Vocabulary: As far as the licensing structure of grammar is concerned, all we need to know about cat is that a particular bunch of phonology can be associated with a syntactic node with the features, singular count noun (perhaps also, animate). (Marantz 1998, 8)

Marantz’s view has been criticised due to the insufficient ability of CSFs to precisely identify roots. According to Pfau (2009) and Harley (2011), a root morpheme with CSFs such as [+Count] and [+Animate] gives equal eligibility to /kæt/ and /d‫ܥ‬g/, and many other root VIs, to be inserted into it in the Vocabulary. In this case, the CSFs that belong to the root morpheme do not have a decisive role to play in the choice of the appropriate root VI. Pfau (2009) investigated speech errors and showed that they are semantically motivated. That is, when errors occur, they involve substituting the desired words with conceptually related ones, e.g., door vs. window or blackboard vs. eraser. This gives an indication that the root morpheme of the desired word has some sort of information that allows the root VI which corresponds to a conceptually similar word to be mistakenly inserted instead. Pfau (2009) argued that every root morpheme is equipped with an index that contains a primitive conceptual feature, giving the root a unique identity. Thus, in order to show how roots are individuated in the Narrow Lexicon, Pfau (2009) used notations such as [¥(CAT)] and [¥(DOG)] instead of the unified [¥ROOT] notation that is traditionally used in DM. He emphasised that root indices do not contribute their contents to the Syntax, which only deals with morphosyntactic features. According to Harley (2011), this view motivates the assumption that root VIs are in a competition that is similar to the competition of functional VIs. Given how a word is substituted by a conceptually related one when a speech error occurs, it can be assumed that the competition is limited to root VIs that represent conceptually related roots. In the Encyclopedia, the semantic meaning of a root is given according to its index and the surrounding syntactic environment. In other words, the choice of the

Chapter Nine

162

phonological and semantic representations of a root morpheme is now conditioned by its index. Although this chapter agrees that CSFs are unable to condition the insertion of root VIs, it argues that this is attributable to CSFs being out of the root domain in the first place, that is, CSFs cannot condition the insertion of root VIs because they are not the property of root morphemes, not because there are too many root VIs that are equally specified for each combination of CSFs.

3. CSFs in Standard Arabic nouns This section investigates the location of CSFs in SA nouns in light of the stem formation mechanism provided by DM. SA is a non-concatenative language (McCarthy 1981; Ryding 2005; Watson 2006). In such a morphological system, stems are formed by combining consonantal roots (radicals),1 mostly tri-consonantal, with patterns of vowels, known as binyanim2 (singular = binyan), as shown below: (3)

a. /ktb/ + /-a‫ޝ‬-i-/ = /ka‫ޝ‬tib-/ b. /ktb/ + /-i-a‫ޝ‬-/ = /kita‫ޝ‬b-/

(writer) (book)

(4)

a. /ktb/ + /- -u-/ = /-ktub-/ b. /ktb/ + /-a-a-/ = /katab-/

(write) (wrote)

The above examples are stems derived from combining the radicals /ktb/ with different binyanim. Stems can be nominal, as in (3), and verbal, as in (4). Nominal stems can refer to humans, as shown in (3a), or to nonhumans, as shown in (3b). What determines whether a stem is nominal or not, and whether it has a human or non-human denotation, is the binyan, 1

Radicals alone have a vague meaning that becomes more specific when they interlock with the appropriate binyan. For example, /ktb/ conveys a general meaning of “writing” or “writing-related,” from which more specific meanings such as “book,” “write” and “writer” can emerge. 2 The term is Hebrew, and it has been translated into English as ‘conjugation,’ ‘formative,’ ‘verbal pattern,’ ‘verbal stem’ and ‘modification,’ (Aronoff 1994, 123). “Bunyan” (structure) is a slightly different Arabic term, which can also be used. However, I will maintain the Hebrew term, as it has been widely used to refer to this type of structure in Semitic languages. See, e.g., Yip (1988), Noyer (1997), Arad (2003, 2005), Ravid and Schiff (2006), Harley (2011), Fassi-Fehri (2012).

The Importance of Root Indices: Evidence from Arabic Nouns

163

rather than the radicals, of the stem. The radicals /ktb/ in these different structures are the same and are always in the same sequence, and they are not exclusively associated with nouns or verbs. Following Arad’s (2003, 2005) analysis of Hebrew, this study assumes that these radicals are the VIs inserted into the root morphemes; consequently, these morphemes cannot be specified for any grammatical category on their own, i.e., they are category-less. As for the binyanim, this study assumes that they are the VIs inserted into the category-assigning [n] morphemes, as shown in (5): (5)

NP

[¥(ktb)] /ktb/

NP

[n] /-a‫ޝ‬-i-/

[¥(ktb)] /ktb/

[n] /-i-a‫ޝ‬-/

Within a purely phonological domain, inserted VIs become subject to the phonological readjustment rules of the language (Halle 1990, 1995) by which the phonological constraints and the overall prosody of Arabic (McCarthy 1981) are guaranteed. This includes interlocking the chosen radicals with the chosen binyanim in order for stems to surface. As far as SA nominal stems are concerned, CSFs such as [±Human] cannot be the property of root morphemes. In addition to the fact that root morphemes are category-less, they are CSF-free since the same root can be part of nominal structures that, ultimately, have human and non-human denotations. In contrast with Marantz’s (1995, 1998) view of the location of CSFs, this study claims that CSFs are the property of the categoryassigning [n] morphemes rather than of the roots. When an [n] morpheme is inherently specified for [+Human], the whole structure surfaces with a human denotation, as shown in (3a). When an [n] morpheme is inherently specified for [–Human], the whole structure will have a non-human denotation, as shown in (3b). It is important to point out that there are different human and nonhuman binyanim in SA. Consider the following table:

Chapter Nine

164

Table 9-1. Human and Non-Human Binyanim in SA Nouns [±Human]

Binyan /-a‫ޝ‬-i-/

[+Human]

/-a-i‫ޝ‬-/

/-a--a‫ޝ‬-/3

/-i-a‫ޝ‬-/

[–Human]

/-a-a-/

/-a-a‫ޝ‬-/

Radicals /ktb/ /ƫrs/ /‫ݧ‬ml/ /t′bb/ /qrb/ /ƫbb/ /t′bx/ /ƫml/ /t′bl/ /ktb/ /qtl/ /kfƫ/ /jbl/ /qlm/ /t′lb/ /œrb/ /jml/ /‫ݧ‬ðb/

Nominal Stem ka‫ޝ‬tib- (writer) ƫa‫ޝ‬ris- (guard) ‫ݧ‬a‫ޝ‬mil- (worker) t′abi‫ޝ‬b- (doctor) qari‫ޝ‬b- (relative) ƫabi‫ޝ‬b- (lover) t′abba‫ޝ‬x- (cook/chef) ƫamma‫ޝ‬l- (carrier) t′abba‫ޝ‬l- (drummer) kita‫ޝ‬b- (book) qita‫ޝ‬l- (fight) kifa‫ޝ‬ƫ- (struggle) jabal- (mountain) qalam- (pen) t′alab- (request) œara‫ޝ‬b- (drink) jama‫ޝ‬l- (beauty) ‫ݧ‬aða‫ޝ‬b- (torture)

In the table, it can be seen that [n] can take different binyanim in the Vocabulary, and each binyan has either a human or non-human denotation. Following Kihm (2005), who claims that [n] is a functional set that consists of different members, e.g., [nI], [nII], [nIII] and so forth, this chapter argues that each member of [n] in SA has a different binyan in the Vocabulary. That is, there are different members of [n], and each member is inherently specified for the CSF [+Human] or [–Human]. In the Vocabulary, there are different binyanim, and each binyan matches a certain member of [n] and expresses its CSFs.

3

This binyan has four empty slots for four consonants. When a tri-consonantal root interlocks with it, the middle consonant is emphasised (doubled) in what is known as “gemination” (for more on gemination in Arabic, see, e.g., McCarthy 1981; Ratcliffe 1997; Hassan 2002; Ryding 2005, to name a few). This phonological modification can be an example of a post-insertion phonological readjustment.

The Importance of Root Indices: Evidence from Arabic Nouns

165

4. Indices and pluralisation in SA nouns This section investigates pluralisation in SA nouns in order to show how root indices can influence the Morphology. According to Pfau (2009) and Harley (2011), indices can condition the choice of root VIs and meanings in the Vocabulary and the Encyclopedia, respectively. This analysis shows that indices can also condition the activation of some language-specific rules in the Morphology as well, specifically, the rule by which SA templatic plural nouns are formed, termed here as the Templatic Formation Rule (TFR), which cannot be activated if the identity of the root is not revealed to the Morphology. There are two types of plurals in SA, namely, affixal and templatic. In the descriptive literature of SA, affixal plurals are known as “sound,” and templatic plurals are known as “broken.” These names reflect the status of the singular stem when the noun changes to plural. In affixal plurals, the singular stem remains unchanged, and the plurality of the noun is expressed by suffixation, as shown in (6) below. On the other hand, the formation of templatic plurals involves some internal changes to the stems of the singular forms, as shown in (7) below:4 (6)

Affixal Plurals a. ba‫ޝ‬ƫiԦ-u‫ޝ‬na5 researcher-PL.NOM ‘male researchers’ b. ba‫ޝ‬ƫiԦ-a‫ޝ‬tresearcher-F.PL ‘female researchers’ c. t′a‫ޝ‬lib-a‫ޝ‬tstudent-F.PL ‘female students’ d. t′alab-a‫ޝ‬trequest-F.PL ‘requests’

4

The abbreviations used in this analysis are the following: PL – Plural, NOM – Nominative, F – Feminine, M – Masculine, GEND – Gender, and K – Case. 5 The suffix reflects the number and case features of the noun. When the noun is non-nominative, /-u‫ޝ‬na/ is replaced with/-i‫ޝ‬na/.

Chapter Nine

166 (7)

Templatic Plurals a. baƫԦ→ ‘research’ b. t′a‫ޝ‬lib‘student’

→

buƫu‫ޝ‬Ԧ‘researches’ (always feminine) t′ula‫ޝ‬b‘male students’

From a DM viewpoint, the above forms are the phonological representations of the following morphologically modified structures: (8)

Affixal Plurals a. Masculine [[[[¥(index)] n] M] PL, K] b. Feminine [[[¥(index)] n] F, PL]

(9)

Templatic Plurals [[¥(index)] n, GEND, PL]

The above representations show the structural differences between affixal and templatic plurals in SA. On the one hand, it can be seen that [n] in the affixal type is always independent (morphologically unchanged); the phonological representation of [n], which is always a binyan, is the same whether the noun is singular or plural. Pluralisation is expressed by fusing [PL] and [K] when the noun is masculine and by fusing [F] and [PL] when the noun is feminine. Fused morphemes are, then, the slots into which the appropriate suffixes are inserted. The structure of templatic plurals, on the other hand, shows a disturbance to the independency of [n]. Whether a templatic plural is feminine or masculine, [n] is fused with [GEND] and [PL], forming the new morpheme [n, GEND, PL]. When the structure is in the Vocabulary, the new morpheme is phonologically represented by a binyan that is different from the binyan that is inserted into [n] when the noun is singular. Changing the binyan reflects the changes that the [n] morpheme incurs when the noun changes into the plural. It is normal that the binyan that expresses [n] alone is different from the binyan that expresses [n, GEND, PL]. Affixal and templatic forms are both productive, but there are no known reasons as to why or when a noun is pluralised in the affixal or in the templatic form. The plural form that a noun takes is simply unpredictable. However, it can be said that affixal plurals are more straightforward and economical. The formation of this type of plural does not involve any morphological modification to the [n] morpheme of the structure. In comparison, the formation of templatic plurals is more complex and costly, as it involves modifying (fusing) three morphemes.

The Importance of Root Indices: Evidence from Arabic Nouns

167

This potentially makes the affixal the default form of plural in SA, i.e., it does not require any special treatment in the Morphology, and it directs the inquiry towards considering the motivation for forming templatic plurals. Investigating the motivation of templatic plural formation entails investigating the different values of the morphemes involved, specifically, which values of [n] and [GEND] might be the motivation or trigger by which the fusing of these three morphemes is activated. [PL] is an invariable feature in both types of plural, so it surely cannot be the sole motivation. If either of the other two morphemes is a factor, [PL] should constitute the environment in which Fusion is triggered. It is apparent that humanness in [n] is not a factor. Human nouns can take the affixal form, as shown in (6a)–(6c), and the templatic form, as shown in (7b). Non-human nouns, on the other hand, can be affixal, as shown in (6d), and templatic, as shown in (7a). Consequently, it can be initially stated that CSFs cannot be the sole motivation for the formation of templatic plurals. As to gender, neither the affixal nor the templatic is limited to a specific gender. Affixal forms can be masculine, as shown in (6a), or feminine, as shown in (6b)–(6d). Templatic plurals can also be feminine, as shown in (7a), or masculine, as shown in (7b). These observations initially dismiss any hypothesis that the formation of [n, GEND, PL] is internally motivated. If fusing these morphemes is externally motivated, the roots of the structures are potentially the motivating elements. As mentioned earlier, root morphemes are equipped with indices that show the unique identity of the roots in the Vocabulary and the Encyclopedia, enabling these components to provide the appropriate phonological and semantic representations, respectively. Since the indices are capable of influencing these language-specific components, it is possible that they can also influence the Morphology, which is a language-specific component too; that is, when the identity of a certain root is revealed to the Morphology, the TFR is activated, as shown in (10) below. (10)

[n] + [GEND] + [PL] → [n, GEND, PL]

The above TFR results in having the three features bundled in one morpheme, which in turn allows the Vocabulary to provide one binyan to express these features together. This explains why the binyanim of templatic plurals, as opposed to affixal plurals, are different from the binyanim of their singular counterparts. Although this seems to solve the problem, it does not explain why plural structures derived from the root [¥(bƫԦ)] are sometimes pluralised in

168

Chapter Nine

the affixal form, as shown in (6a) and (6b), and sometimes in the templatic form, as shown in (7a) above. Likewise, it does not explain why plural structures derived from the root [¥(t′lb)] can be affixal, as shown in (6c) and (6d), and templatic, as shown in (7b). If the activation of the TFR is solely based on the identity of certain roots, every plural structure derived from these roots should always be templatic, but never affixal. Therefore, what we have here are roots whose structures are pluralised as templatic only in certain environments. For instance, considering the examples in (6) and (7) above, it can be said that plural structures derived from [¥(bƫԦ)] are templatic when their [n] morpheme is inherently specified as [+Human], as seen in (6a) and (6b). Otherwise, as shown in (7a), the structures escape the TFR and default to the affixal form. With the root [¥(t′lb)], however, humanness is not the only factor involved. Plural structures derived from this root are templatic when the noun is human and masculine, as seen in (7b). When [n] is specified as human and the gender is feminine, as seen in (6c), or when [n] is non-human, as seen in (6d), the structure escapes the TFR and defaults to the affixal shape. Thus, it can be claimed that the formation of templatic structures, i.e., fusing [n], [GEND] and [PL], is conditioned by two factors: the values of the fused morphemes (an internal factor) and the identity of the root morpheme (an external factor). This study claims that the internal conditions are stored in the Morphology as more specific versions of the TFR shown in (10). These rules capture the different environments in which plural structures are templatic; the rules are awaiting activation, which is now conditioned by the external factor. As discussed earlier, the identity of roots is stored in their indices, so when a root index does not match the activation condition, the structure escapes the rules and an affixal structure is formed. On the other hand, when the index matches the condition, the TFR creates the morpheme [n, GEND, PL], allowing the insertion of a new binyan into it in the Vocabulary.

5. Conclusion In contrast with Marantz (1995, 1998), this chapter shows that CSFs are not the property of roots. Therefore, they cannot condition the insertion of root VIs in the Vocabulary. Standard Arabic roots can participate in nominal structures that refer to humans, as well as in structures that refer to non-humans. Hence, the humanness feature in SA cannot be associated with the roots from which the nouns are derived. This gives clear evidence that roots, at least in SA, are also devoid of CSFs. This study shows that

The Importance of Root Indices: Evidence from Arabic Nouns

169

CSFs are located in the [n] morphemes of the nominal structures, that is, [n] morphemes are inherently specified for humanness. This entails that [n] is actually a set of different members (Kihm 2005), and each member is specified for being either human or non-human. In the Vocabulary, each member of [n] has a different binyan inserted into it. Binyanim, in this case, not only give roots their nominal classes, but they also express the humanness feature of the whole structure. With roots being devoid of any phonological, semantic and now CSF information, the notion of root indices (Pfau 2009; Harley 2011) seems to account plausibly for how roots are recognised by the Vocabulary and the Encyclopedia. This chapter posits that indices can also help the Morphology determine what modifications should or should not be applied to the structures before Vocabulary Insertion. Affixal and templatic plural nouns in SA were taken as an example to show how the activation of the relevant language-specific rule in the Morphology can be conditioned by the identity of the roots. When indices reveal the identity of their roots to the Morphology, and the revealed identity matches the requirement of the activation, the rule is activated. This results in the creation of the morpheme [n, GEND, PL], into which a single binyan is inserted. This explains how templatic plurals are formed and why their binyanim are richer than and different from the binyanim used in the singular forms of the nouns. It also shows that indices can play a role in the morphological, as well as the phonological and semantic, domains of the derivation process.

References Arad, Maya. 2003. “Locality Constraints on the Interpretation of Roots: The Case of Hebrew Denominal Verbs.” Natural Language and Linguistic Theory 21(4):737–79. —. 2005. Roots and Patterns: Hebrew Morpho-Syntax. Dordrecht: Springer. Aronoff, Mark. 1994. Morphology by Itself: Stems and Inflectional Classes. Cambridge, MA: MIT Press. Embick, David, and Alec Marantz. 2008. “Architecture and Blocking.” Linguistic Inquiry 39(1):1–53. Embick, David, and Rolf Noyer. 2007. “Distributed Morphology and the Syntax/Morphology Interface.” In The Oxford Handbook of Linguistic Interfaces, edited by Gillian Ramchand and Charles Reiss, 289–324. Oxford: OUP.

170

Chapter Nine

Fassi-Fehri, Abdelkader. 2012. Key Features and Parameters in Arabic Grammar. Amsterdam: John Benjamins. Halle, Morris. 1990. “An Approach to Morphology.” Proceedings of NELS 21:150–85. —. 1995. “The Russian Declension: An Illustration of the Theory of Distributed Morphology.” In Perspectives in Phonology, edited by Jennifer Cole and Charles Kisseberth, 321–53. Palo Alto: CSLI, Stanford. —. 1997. “Distributed Morphology: Impoverishment and Fission.” MIT Working Papers in Linguistics 30:425–49. Halle, Morris, and Alec Marantz. 1993. “Distributed Morphology and the Pieces of Inflection.” In The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, edited by Kenneth Locke Hale, Samuel Jay Keyser, and Sylvain Bromberger, 111–76. Cambridge, MA: MIT Press. Halle, Morris, and Alec Marantz. 1994. “Some Key Features of Distributed Morphology.” MIT Working Papers in Linguistics 21:275– 88. Harley, Heidi. 2005. “One-Replacement, Unaccusativity, Acategorical Roots and Bare Phrase Structure.” Harvard Working Papers in Linguistics 11:59–78. Harley, Heidi. 2011. On the Identity of Roots. Tucson: University of Arizona. Harley, Heidi, and Rolf Noyer. 1999. “State-of-the-Article: Distributed Morphology.” Glot International 4(4):3–9. Harley, Heidi, and Rolf Noyer. 2000. “Formal Versus Encyclopedic Properties of Vocabulary: Evidence from Nominalisations.” In The Lexicon-Encyclopedia Interface, edited by Bert Peeters, 349–74. Amsterdam: Elsevier Press. Hassan, Zeki. 2002. “Gemination in Swedish and Arabic with a Particular Reference to the Preceding Vowel Duration: An Instrumental and Comparative Approach.” Proceedings of Fonetik, 81–85. Kihm, Alain. 2005. “Noun Classes, Gender, and the Lexicon-SyntaxMorphology Interfaces: A Comparative Study of Niger-Congo and Romance Languages.” In The Oxford Handbook of Comparative Syntax, edited by Guglielmo Cinque and Richard Kayne, 459–512. Oxford: OUP. Marantz, Alec. 1995. “A Late Note on Late Insertion.” In Explorations in Generative Grammar: A Festschrift for Dong-WheeYang, edited by Young-Suk Kim, Byung-Choon Lee, Kyoung-Jae Lee, Hyun-Kwon Yang, and Jong-Yurl Yoon, 396–413. Seoul: Hanakuk.

The Importance of Root Indices: Evidence from Arabic Nouns

171

—. 1997. “No Escape From Syntax: Don’t Try Morphological Analysis in the Privacy of Your Own Lexicon.” University of Pennsylvania Working Papers in Linguistics 4(2):201–25. —. 1998. “‘Cat’ as a Phrasal Idiom: Consequences of Late Insertion in Distributed Morphology.” Ms., MIT, Cambridge, MA. —. 2001. “Words.” Paper presented at the 20th West Coast Conference on Formal Linguistics at University of California, Santa Cruz. McCarthy, John. 1981. “A Prosodic Theory of Nonconcatenative Morphology.” Linguistic Inquiry 12(3):373–418. Noyer, Rolf. 1997. Features, Positions and Affixes in Autonomous Morphological Structure. New York: Garland. Pfau, Roland. 2009. Grammar As Processor: A Distributed Morphology Account of Spontaneous Speech Errors. Amsterdam: John Benjamins. Ratcliffe, Robert. 1997. “Prosodic Templates in a Word-Based Morphological Analysis of Arabic.” In Perspectives on Arabic Linguistics: Papers from the Annual Symposium on Arabic Linguistics, edited by Mushira Eid and Robert Ratcliffe, 147–72. Amsterdam: John Benjamins. Ravid, Dorit, and Rachel Schiff. 2006. “Roots and Patterns in Hebrew Language Development: Evidence from Written Morphological Analogies.” Reading and Writing 19:789–818. Ryding, Karin. 2005. A Reference Grammar of Modern Standard Arabic. Cambridge: CUP. Watson, Janet. 2006. “Arabic as an Introflecting Language.” In Encyclopedia of Language and Linguistics, edited by Keith Brown, 431–34. Oxford: Elsevier. Yip, Moira. 1988. “Template Morphology and the Direction of Association.” Natural Language and Linguistic Theory 6(4):551–77.

CHAPTER TEN VERBAL CAUSATIVE STRUCTURE FOR POLISH PREFIXED TRANSITIVES IN THE ROOT BASED APPROACH1 ANNA MALICKA-KLEPARSKA

1. Introduction In this text we will concern ourselves with causative verbs in Polish, and more specifically, with the verbs which do not enter causative/anticausative alternation with synthetic anticausatives, but can be diagnosed as causatives nevertheless. We will try to answer the following questions: what structure should they be assigned and what exponents match particular positions in this structure? We will be working within the root based approach, which we hold significantly superior to lexicalist approaches, especially in this area of Polish morphology.

2. Exposition of the data In Polish, there is a vast group of verbs which cannot be claimed to be based on other verbal elements as they do not possess related simpler predicates, yet they contain roots common to adjectives, substantives, as well as complex verbs (see Olszewska 1986; Wróbel 1984; Szymanek 2010). Olszewska (1986) describes them as containing the prefixes u-, na-,

1

I would like to express my gratitude to Marietta Rusinek, Paweá Czerniak, Karolina Drabikowska and Anna PraĪmowska – Doctoral students in a linguistics class in the English Department of the John Paul II Catholic University of Lublin, who helped me to clarify my views upon causativisation.

Verbal Causative Structure for Polish Prefixed Transitives

173

o-, prze-, przy-, roz-, wy-, z/s/ze-2 and za-, as well as the suffix -i-(ü ).3 We will add to this group the verbs formed with nad-, po- and wy-. The list below includes some examples representing particular sub-groups. It is by no means exhaustive, but it mirrors the numerosity of particular patterns: (1)

zestaliü ‘solidify,’ zeĨliü ‘make angry,’ zniechĊciü ‘dissuade,’ zeskaliü ‘petrify,’ zgarbiü ‘bend,’ zeszmaciü ‘crush, colloq.,’ zezwierzĊciü ‘make a beast of,’ zwĊgliü ‘coal,’ streĞciü ‘summarize,’ skawaliü ‘turn into lumps,’ spáyciü ‘make shallow,’ skropliü ‘liquify,’ skoligaciü ‘make related through marriage,’ skróciü ‘skróciü,’ spowinowaciü ‘make related through marriage,’ zawĊĨliü ‘tie into knots,’ zakwefiü ‘put on a vail,’ zapĊtliü ‘loop,’ zagáĊbiü ‘deepen,’ zaprzepaĞciü ‘give up,’ zachwaĞciü ‘infest with weeds,’ zasmĊciü ‘sadden,’ zaskarbiü ‘hoard,’ zagraciü ‘clatter,’ zachĊciü ‘encourage,’ zarybiü ‘put in fish,’ zagrzybiü ‘spread fungi,’ zaĞlubiü ‘wed,’ zaokrągliü ‘make round,’ zapchliü ‘introduce flees,’ uziemiü ‘ground,’ uĞmierciü ‘kill,’ ujednoliciü ‘make even,’ uwznioĞliü ‘elate,’ uwypukliü ‘make tangible,’ upodliü ‘make inferior,’ usidliü ‘catch,’ uproĞciü ‘simplify,’ otorbiü ‘make into a cyst,’ ociepliü ‘make warm,’ okoliü ‘surround,’ ochwaciü ‘nobble a horse,’ powiadomiü ‘send a message,’ poĞwiĊciü ‘sacrifice,’ odczuliü ‘make immune,’ oddaliü ‘estrange,’ odksztaáciü ‘misshape,’ rozanieliü ‘molify,’ rozochociü ‘make jolly,’ rozswawoliü ‘make jolly,’ obramiü ‘frame,’ obmyĞliü ‘think up,’ obrĊbiü ‘hem,’ nasiliü ‘strengthen,’ naĞwietliü ‘irradiate,’ namaĞciü ‘anoint,’ osierociü ‘orphan,’ onieĞmieliü ‘make shy,’ przygarbiü ‘bend,’ przekabaciü ‘turn round, colloq.,’ wydelikaciü ‘make delicate,’ etc., etc.

Various prefixes may co-exist with the same root, e.g.: (2)

naĞwietliü ‘irradiate,’ wyĞwietliü ‘project,’ rozĞwietliü ‘lighten up,’ przeĞwietliü ‘irradiate,’ oĞwietliü ‘lighten up,’ etc.

Particular verbs differ in meaning as prefixes carry with them additional shades of meaning (see, e.g., Szymanek 2010). Causitivizing prefixes coincide in form with the prefixes deriving aspectual distinctions in Polish, however, the class of prefixed verbs which are causative and the class of prefixed verbs which are perfective are not one and the same. Not all prefixed (perfective) verbs are causatives (see (3)). Likewise, not all causatives are perfective (see (16)), so the two classes cannot be treated in

2

For the allomorphic variations involved in the realization of particular prefixes, see, e.g., Gussmann (2007). 3 The material in brackets represents the inflectional suffix of the verb.

174

Chapter Ten

the same way and assigned identical morpho-syntactic structures. The examples in (3) illustrate the aspectual markers on non-causative verbs:4 (3)

robiüI ‘do’ – zrobiüP, ciągnąüI ‘draw’ – pociągnąüP, prowadziüI ‘conduct’ – poprowadziüP, wąchaüI ‘smell’ – zwąchaüP, trąbiüI ‘honk’ – zatrąbiüP, pisaüI ‘write’ – napisaüP, myĞleüI ‘think’ – pomyĞleüP, tuliüI ‘hug’ – przytuliüP, laüI ‘pour’ – przelaüP ‘pour over,’ etc.

The examples above show the distinction between imperfective and perfective prefixed forms of the verbs (see Willim 2006, 187–88); in most cases, the only difference in meaning is the aspect of the form, although additional semantic differentiations may appear as well, contributed by particular prefixes (see, e.g., Szymanek 2010). How do the forms in (1) differ from aspectually marked forms? First and foremost, the verbs in (1), when devoid of a prefix, are not imperfective verbs. Such verbal forms (or other closely related ones) do not exist at all. Consequently, aspectual pairs consisting of a prefixed and non-prefixed verb never arise, although so-called secondary imperfectives appear (see Willim 2006, 196), realized by a suffix: (4)

uĞmierciüP ‘kill’ – uĞmiercaüI ‘kill,’ zawĊĨliüP ‘tie into knots’ – zawĊĨlaüI ‘tie into knots,’ zestaliüP ‘solidify’ – zestalaüI ‘solidify’

At the same time, the aspectually marked forms in (3) are noncausative, according to the criteria specified directly below. Summing up, the class of causative verbs in (1) is prefixed with morphemes which can also perform a perfectivizing function elsewhere. The causatives, however, do not have corresponding simpler imperfectives or synthetic anticausatives, although they contrast with secondary imperfectives. The causative nature of the verbs in (1) will be discussed in the next section.

4

Superscripts refer to imperfective and perfective verb forms.

Verbal Causative Structure for Polish Prefixed Transitives

175

3. Prefixed -i- verbs as causatives Causatives are defined by Levin and Rappaport Hovav (1995, 83) as verbs representing the following event structure: (5)

[x Do-Something] CAUSE [y BECOME STATE]

According to Van Valin (2013), causation consists of two subevents: an action and a result state. This semantic description, while seemingly clear, in particular cases is not that easily applicable. For example, if you have trumpeted the news, is the news in some sense trumpeted?5 Consequently, in most cases, the identity of causatives is ascertained on morpho-syntactic grounds. Causatives are such verbs that enter anticausative alternation (see, e.g., Haspelmath 1993; Haspelmath and Müller-Bradley 2004), i.e., they possess change of state related verbs, whether anticausatives are believed to be based on causatives (e.g., Grimshaw 1982; Chierchia 2004; Levin and Rappaport Hovav 1995; Koontz-Garboden 2009) or the opposite (e.g., Dowty 1979; Lakoff 1968; Van Valin 2013). However, some researchers allow the possibility that causative verbs may have no corresponding anticausatives. For instance, Van Valin (2013) gives the example of dirty as a causative verb, in spite of the fact that it does not involve a corresponding anticausative. Another criterion, separate from the elusive sematic structure and the appearance of anticausative variants, can be adopted from Van Valin (2013), where he shows that the semantic structure of causatives has a formal substantiation. Because causatives are equipped with the semantics of a causing event and a state event separable from each other, causation may be indirect: an intervening instrument may be used for making a change happen. This instrument, in turn, can be mentioned in the causative structure as a subject of the predication, while with other transitives, unievental in nature, no such opportunity exists. In other words, in the case of non-causative transitives, the act is performed by the Actant himself, while with causative transitives – it may be performed by a Causer using some equipment. This becomes clear when we compare the following structures:

5

See Embick (2009) for distinctions in types of states associated with various predicates.

Chapter Ten

176 (6)

a. Causative: John cut bread with a knife. – The knife cut bread. b. Non-causative: John read a book with his spectacles on. – * The spectacles read the book.

The phenomenon can be illustrated for Polish as well: (7)

Jan zwĊgliá stóá podpaáką. ‘John has burnt the table with kindling.’ – Podpaáka zwĊgliáa stóá. ‘The kindling has burnt the table.’

(8)

Jan przeczytaá ksiąĪkĊ w okularach. ‘John has read the book in his glasses.’ – * Okulary przeczytaáy ksiąĪkĊ. ‘Glasses have read the book.’

Additional tests can be adopted from Embick (2009) who discusses various types of states. Uncaused states, that is the ones which result from activities named by transitive non-causative verbs, cannot appear in the constructions appropriate for caused states, e.g., as predicates introduced by the copula byü ‘to be.’ They cannot co-occur with ciągle jeszcze (‘still’) either: (9)

Drzwi są otwarte. ‘Doors are opened.’ (caused state) vs. * WiadomoĞü jest roztrąbiona. (uncaused state) ‘The news is trumpeted.’

(10)

Widziaáam drzwi ciągle jeszcze otwarte. ‘I saw the door still opened.’ (caused state) vs. * Widziaáam wiadomoĞü jeszcze ciągle przeczytaną. ‘I saw the news still read. (uncaused state).

So, apart from a potentially impressionistic semantic analysis, there exist well defined criteria for classifying verbs as causatives: they may possess anticausative counterparts, they may have characteristic morphological marking (depending on the language),6 causation may be seen as conducted by an instrument, and the state resulting from this causation may occupy specific clausal positions. All verbs in (1) count as causatives, according to the above criteria: they have causative semantics, they have characteristic morphological marking – to be described in section 5, they allow instrument subjects, they relate to derived caused states. They, however, do not have corresponding synthetic anticausatives.

6

For instance, English marks causatives in the same way as, e.g., anticausatives (He flattened a piece of paper / A piece of paper flattened), or it does not mark them at all (He broke the vase / The vase broke).

Verbal Causative Structure for Polish Prefixed Transitives

177

Synthetic anticausatives in Polish are created with the suffixes -e-/-ej,7 or -ną- and, as expected, enter the opposition with causatives which are based on the same root: (11)

czerwienieü ‘redden ‘ – czerwieniü ‘make red,’ zielenieü ‘turn green’ – zieleniü ‘make green,’ schnąü ‘dry up’ – suszyü ‘make dry,’ gáuchnąü ‘grow deaf’ – ogáuszyü ‘make deaf,’ miĊknąü ‘grow soft’ – rozmiĊkczyü ‘make soft,’ rzednąü ‘grow more diluted’ – rozrzedziü ‘make more diluted’

The verbs in (1), however, lack such synthetic anticausatives; this is clearly unexpected when we take into consideration their semantics, which subsumes “entering a state,” while the Polish language possesses the appropriate pattern.

4. Prefixed causatives as accomplishments Apart from being causatives, the verbs are also accomplishments, or causative accomplishments, in Van Valin’s (2013) terminology. Accomplishments are telic predicates in which a gradual change of state is encoded. Their telic nature can be decided on the basis of the w godzinĊ/godzinĊ test (‘in an hour/for an hour’),8 while the ‘finish’ test (Vendler 1957) distinguishes them from achievements: (12)

Jan zwĊgliá stóá w godzinĊ. ‘John has burnt the table in an hour.’ (accomplishment) vs. * Jan zwĊgliá stóá godzinĊ. ‘John has burnt the table for an hour.’

(13)

Jan skoĔczyá zwĊglaü stóá. ‘John has finished burning the table in an hour.’ (accomplishment) vs. * Jan skoĔczyá zanurkowaü. ‘John finished diving.’ (achievement)

The tests applied in section 3 and 4 clearly situate verbs in (1) as causative accomplishments. In the next section, we will consider their formal make-up in greater detail. 7

The first allomorph appears, e.g., in the infinitive, while the second one occurs, e.g., in 3rd person present tense forms. 8 See Vendler (1957, 1967), Dowty (1979), for Polish – Willim (2006, 176–77 and ftn. 2).

Chapter Ten

178

5. Morphological make-up of the complex causatives As mentioned at the very beginning, the causatives which we analyze are prefixed with the morphemes which also form aspectually perfective verbs elsewhere. Apart from this characteristic prefixation, they also have the suffix -i-.9 However, the suffix by itself is not enough to ensure that causatives are formed. The examples in (14) below illustrate some cases where the suffix forms a verb, but the verb does not have a causative character. In fact, the suffix marks just the conjugational class of the verb – not its semantics:10 (14)

szeleĞciü ‘rustle,’ chrzĊĞciü ‘grit,’ ziĞciü siĊ ‘come to fruition,’ mĞciü ‘avenge,’ záoĞciü ‘irritate,’ poĞciü ‘fast,’ roĞciü ‘claim rights,’ zazdroĞciü ‘envy,’ nawaliü ‘fail, colloq.,’ strzeliü ‘shoot,’ nagliü ‘incite,’ piliü ‘insist, colloq.,’ kwiliü ‘squeak,’ skomliü ‘whine,’ biadoliü ‘despair,’ dziamoliü ‘nag,’ rzĊpoliü ‘play an instrument badly,’ swawoliü ‘play pranks,’ zezwoliü ‘allow,’ tliü ‘smolder,’ wieĞciü ‘symptomize,’ korciü ‘tantalize,’ kopciü ‘emit smoke,’ kropiü ‘sprinkle with water,’ wróciü ‘come back,’ psociü ‘play pranks,’ Ĩrebiü siĊ ‘bear a colt,’ kociü siĊ ‘bear a kitten,’ przeziĊbiü siĊ ‘to catch a cold,’ chyliü siĊ ‘lean to,’ myliü siĊ ‘err,’ pyliü ‘produce pollen,’ oznajmiü ‘anounce,’ áakomiü siĊ ‘crave for,’ etc., etc.

Verbs in this list represent activities (e.g., szeleĞciü ‘rustle’), subject experience verbs (e.g., zazdroĞciü ‘envy’), states (e.g., wieĞciü ‘symptomize’), anticausatives (e.g., przeziĊbiü siĊ ‘to catch a cold’). Consequently, the suffix cannot be treated as a marker of causativization. Below, we will argue for a particular causative structure representing causative verbs in the root based approach. We will also place morphemes building causatives in specific structural positions reflecting their properties. 9

Olszewska (1986, 80–81) enumerates other suffixal elements realizing causative formations in Polish, e.g., -owaü, -izowaü, but we must limit our data to a subclass that can be analyzed within the limits of this work. 10 In derived imperfectives (see 4), the suffix might be claimed to disappear altogether, but for phonological phenomena which seem to mark its presence (palatalization). We cannot discuss these issues in detail (see Gussmann 2007). The fact remains, however, that the segmental morpheme is no longer there and phonological facts do not univocally point to the presence of the -i- suffix. Much depends on the particular (morpho)phonological approach taken. If the suffix forming secondary imperfective replaces the -i- suffix, then the position of the latter as a causativizing element becomes even more doubtful.

Verbal Causative Structure for Polish Prefixed Transitives

179

6. Causative structure We propose that the causative structure subsumes the schema put forward for change of state verbs by Embick (2009). This structure is bipartite. One node specifies the act which brings about the change of state of the internal argument, while the state and the DP argument constitute the material dominated by the other node. This part reflects the complex bievental structure of causation. At the same time this part is common to anticausatives and causatives. Anticausatives and causatives share causative semantics, as is extensively argued elsewhere.11 This common substructure is in the schema below dominated by the common node v‫މ‬, which represents the predicate with its internal arguments – in this case, a single argument undergoing the aforementioned change of state. Here, causatives and anticausatives part ways: while anticausatives are completed by the middle voice projection (see Alexiadou and Doron 2012), causatives take the active voice projection. The specifier of the active voice phrase is the Causer external argument (see Marantz 1984; Kratzer 1996; Pylkkänen 2008; Alexiadou and Doron 2012). This structure is represented in (15) below: (15)

ȞP Spec DP

Ȟ Ȟ

v‫މ‬ v

¥1

v

STP ST

DP

The position marked as 1 is filled with the most basic morphological material – the root. In the root based approach, roots are devoid of category marking. Consequently, the roots that could appear in (15), in principle, can be shared with adjectives, substantives and other verbs. Embick (2009), however, introduces a distinction between root classes. In 11 Such analyses are presented in, for example, Koontz-Garboden (2009), Beavers and Koontz-Garboden (2013a, 2013b), Malicka-Kleparska (to appear).

180

Chapter Ten

the lexicon, they are marked as good predicates of events, of states or both states and events. The particular roots which are appropriate for the causative derivation are good predicates of events12 (see Embick 2009). Such roots also constitute the building blocks of unaccusatives13 (or anticausatives), which explains why causatives and anticausatives/ unaccusatives are seen as entering the causative/anticausative alternation in lexicalist approaches. We propose that, in (15), prefixes perform the function of heads of the active voice projection, while the suffix -i- enters the derivation under the v node. The suffix, in accordance with its role in the structure, provides for the verbal character of the formation, since roots are a-categorial. As we have mentioned above, -i- appears in a variety of verbal formations. It also decides about the conjugational properties14 so it creates a root-based correspondent of the category of inflectional “stem” in traditional and lexicalist approaches. If this position is filled, for instance, by the -e-/-ejverbalizing affix, the structure can merge the non-active voice head, without a specifier, and anticausative derivatives result (e.g., czerwienieü ‘redden, inf.’ – czerwienieje ‘It reddens’ vs. czerwieniü ‘redden sth., inf.’ – czerwieni ‘It reddens sth.’). Various prefixes (u-, na-, o-, prze-, przy-, roz-, wy-, z/s/ze-, za-, nad-, po- and wy-) constitute active voice heads, as they decide about the introduction of an additional participant into the structure – the causer in the specifier of the active voice projection. This model could suggest that there should be no unprefixed -icausatives in Polish, but it is not borne out by facts, at least at first glance. Such causatives appear, but they have different properties than the prefixed ones. Consider the forms in (16): (16)

12

szkoliü ‘train,’ táuĞciü ‘grease,’ moĞciü ‘pad,’ gromiü ‘give forth thunder,’ bezczeĞciü ‘violate,’ cuciü ‘make sb. to come to,’ mydliü ‘soap,’ Ğcieliü ‘pad,’ mgliü ‘render misty,’ szpeciü ‘uglify,’ haĔbiü ‘violate,’ zdobiü ‘ornament,’ bogaciü ‘enrich,’ etc., etc.

This specification also includes good predicates of events and states. We use the term “unaccusative” for verbs which manifest change of state semantics, but are not clearly morphologically marked as belonging to this group of verbs. Anticausatives are seen as possessing distinctive morphological marking. 14 For details, see Laskowski (1984). 13

Verbal Causative Structure for Polish Prefixed Transitives

181

The verbs in (16) are causatives, like the verbs in (1), which can be verified by the tests which we have already used for the prefixed causatives: (17)

Joanna szkoliáa psa kijem. ‘Joanna has trained the dog with a stick.’ – Kij szkoliá psa skutecznie. ‘The stick trained the dog effectively.’ Pies jest (ciągle jeszcze) szkolony. ‘The dog is (still being) trained.’

However, the unprefixed verbs do not seem to be telic. Consequently, it is not clear whether we can still consider them to be accomplishments, like the ones in (1): (18)

* Joanna szkoliáa psa w godzinĊ. ‘Joanna trained the dog in an hour.’

Apparently, the difference in telicity is due to some particular properties of the prefixes, other things being equal. We believe that telicity, indispensible for the accomplishment interpretation, is contributed by the prefixes (which are, after all, also markers of perfectiveness in Polish).15 When placed in the position of heads of active voice, they both decide the transitivity of the predicate and its telicity. If placed in a position appropriate for aspect marking (e.g., in the verbs in (3)) – they contribute telicity alone, as they are not in the appropriate structural position to decide about the arguments.16 For instance, we may adopt the position taken by McDonald (2008, 149) and assume that the aspectual prefixes in Slavic contribute the feature translating as perfectiveness/telicity directly at the verb level for the verbs in (3):

15

See àazorczyk (2010) for an in depth account of the Polish aspectual system. àazorczyk treats Polish verbs as imperfective in unmarked cases, basing her conclusion on the observation that basic perfectives are few and far between. Prefixes make verbs telic (and, in consequence, perfective). In this text, we follow her analysis, although the problem of aspect in Polish is very complex and widely disputed. For a comprehensive presentation of problems connected with Polish aspect and a different treatment of the phenomenon, see, e.g., Bloch-Trojnar (2013). 16 For proposals about the structural position of aspectual prefixes in Polish verbal structure, see e.g., àazorczyk (2010, 56).

Chapter Ten

182 (19)

…ȞP

Ȟ

VP V

…

Here, is a final subivental feature and – an initial subivental feature, i.e., perfective and imperfective respectively. On the other hand, the verbs in (16), in the head of the active voice position possess zero affixes. A zero affix does not possess any features (telic or atelic). Verbs in Polish are by default marked as atelic so, consequently, those marked by a zero morpheme are atelic as well, hence they cannot be accomplishments. Notice also that the verbs in (16), unlike these in (1), cannot form secondary imperfectives: *szkolaü, *táuĞciaü, *moĞciaü. This is understandable as they are not prefective themselves. In other words, unlike prefixed verbs, non-prefixed causatives are not typical accomplishments. Van Valin’s (2013) terminology does not cover such cases, as the verbs are not causative states (e.g., The dog scares the boy), nor causative active accomplishments (e.g., The sergeant marched the soldiers to the field). We will refer to them as imperfective accomplishments, although their status clearly requires further study. The important thing is that, in some ways, they behave as the prefixed causatives. For instance, both classes require the spelling out of the internal argument, unlike many other unprefixed Polish verbs, which together with their causative semantics makes us think that they reveal the same causative structure as prefixed causatives. The examples below illustrate the property of causatives, both prefixed and unprefixed, which prevents them from appearing in absolute structures, contrasted with other transitives, which can fill such structures relatively easily:17

17

See Malicka-Kleparska (2012).

Verbal Causative Structure for Polish Prefixed Transitives (20)

183

* Joanna zdobiáa i dlatego Ğciana wyglądaáa okazale. ‘Joanna decorated and that is why the wall looked impressive.’ vs. * Jan zwĊgliá i dlatego stóá trzeba byáo wyrzuciü. ‘John singed and that is why the table had to be thrown away.’ vs. Joanna czytaáa i trzeba byáo byü cicho. ‘Joanna was reading and you had to keep quiet.’

We believe that this difference in the distribution of internal arguments results from the properties of the change of state part of the causative structure where the argument is the only element possessing morphological material. The deletion of the internal argument would leave the whole State Phrase (STP) empty of morphological content.

7. Superiority of the root based account and problematic areas The data concerning prefixed -i- causatives in Polish clearly show that the root based approach is superior to the lexicalist approach as a model for describing this sub-system of morpho-syntax. It explains why the causative verbs in (1) do not regularly have synthetic anticausative counterparts – in the lexicalist approach, having such counterparts could be expected, given the semantic and formal affinity of causatives and anticausatives and the fact that one class has to be derived from the other. In our approach, it is possible that the correspondences may appear, as both classes of forms are based on the same roots and they share a subpart of the structure. The forms in (11) show that sometimes the pairs of prefixed causatives and synthetic anticausatives indeed arise. However, the correspondences are not regular, as expected if the forms are derived separately “from scratch.” Significantly, interesting conclusions about the nature of perfectivizing prefixes and the limitations on the absolute uses of causatives can be drawn as well, which are not available in the lexicalist approach. Namely, causative forms and perfective forms may be derived with the same lexical prefixes occupying distinct structural positions. This solution simplifies the content of the lexicon and accounts for the formal identity of the prefixes. Similarly, the structure proposed for causatives, both simple and prefixed, accounts for their object preserving properties, which is a solution beyond the reach of a lexicalist account. Many problems still have to be addressed which go beyond the limits of this text. First of all, the causatives in (1) may possess other non-active counterparts, i.e., dispositional middles, reflexives, reflexive anticausatives,

184

Chapter Ten

or reflexive impersonals. They participate in other ways in which language addresses skipping the causer spell-out in a morpho-syntactic structure or making the causer less prominent by identifying it with the internal argument. In many cases, however, no such structures have been found in the National Corpus of the Polish Language18 – which, of course, does not rule them out as potential forms. Nevertheless, this state of affairs contributes to the general impression of the randomness of causative/noncausative morpho-syntax, which is precisely what is expected in the root based approach – many patterns are possible, but no correlation of particular derivatives need be expected. The examples below illustrate various non-active and reflexive patterns with different verbs from (1) taken from the Corpus: (21)

Cháop zeszmaciá siĊ na naszych oczach doszczĊtnie. (anticausative) ‘The bloke has gone off his rocker in our presence completely.’ Czas przejazdu przez miasto skróci siĊ. (anticausative) ‘The time to cross the city will shrink.’ Salomon spowinowaciá siĊ z faraonem. (reflexive) ‘Solomon became a kin of the Pharaoh.’ Historia zapĊtliáa siĊ. (anticausative) ‘History has turned the full circle.’ ElĪbieta zagáĊbiáa siĊ w nieczystoĞciach aĪ po pachy. (reflexive) ‘ElĪbieta has immersed herself in excrements to her armpits.’ Malec oddala siĊ wzdáuĪ galerii. (reflexive) ‘The boy moves away in the direction of the shopping mall.’ ZachĊca siĊ ludzi do palenia. (impersonal reflexive construction) ‘People are encouraged to smoke.’ Lina zawĊĨla siĊ áatwo. (example mine = A. M-K) (dispositional middle) ‘The rope gets entangled in knots easily.’

The formations are very difficult to distinguish from each other and even a native speaker has to put in some effort to provide the desired construction and meaning, which suggests that the above morpho-syntactic formations may have very similar structures. For instance, they can be formed above the active voice projection, and not within the scope of the change of state substructure. This point, however, requires much more extensive studies, going far beyond the limits of this chapter.

18

The corpus is entered in the references as Przepiórkowski et al. (2012).

Verbal Causative Structure for Polish Prefixed Transitives

185

References Alexiadou, Artemis, and Edit Doron. 2012. “The Syntactic Construction of Two Non-Active Voices: Passive and Middle.” Journal of Linguistics 48:1–34. Beavers, John, and Andrew Koontz-Garboden. 2013a. “In Defense of The Reflexivization Analysis of Anticausativization.” Lingua 131:199–216. —. 2013b. “Complications in Diagnosing Lexical Meaning: A Rejoinder to Horvath and Siloni.” Lingua 134:210–18. Bloch-Trojnar, Maria. 2013. The Mechanics of Transposition. A Study of Action Nominalisations in English, Irish and Polish. Lublin: Wydawnictwo KUL. Chierchia, Gennaro. 2004. “A Semantics for Unaccusatives and Its Syntactic Consequences.” In The Unaccusativity Puzzle. Explorations of the Syntax-Lexicon Interface, edited by Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert, 22–59. Oxford: OUP. Dowty, David. 1979. Word Meaning and Montague Grammar – The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Dordrecht: Reidel. Embick, David. 2009. “Roots, States, and Stative Passives.” Abstract for Root Workshop, University of Stuttgart, June 2009. Grimshaw, Jane. 1982. “On the Lexical Representation of Romance Reflexive Clitics.” In The Mental Representation of Grammatical Relations, edited by Joan Bresnan, 87–148. Cambridge, MA: MIT Press. Gussmann, Edmund. 2007. The Phonology of Polish. Oxford: OUP. Haspelmath, Martin. 1993. “More on The Typology of Inchoative/Causative Verb Alternations.” In Causatives and Transitivity. Studies in Language Companion. Series 23, edited by Bernard Comrie and Maria Polinsky, 87–111. Amsterdam: John Benjamins. Haspelmath, Martin, and Thomas Müller-Bardey. 2004. “Valency change.” In Morphology: A Handbook on Inflection and Word Formation. Vol. 2, edited by Geert Booij, Christian Lehmann, and Joachim Mugdan, 1130–45. Berlin: Mouton de Gruyter. Koontz-Garboden, Andrew. 2009. “Anticausativization.” Natural Language and Linguistic Theory 27:77–138. Kratzer, Angelica. 1996. “Severing the External Argument from Its Verb.” In Phrase Structure and the Lexicon, edited by Johan Rooryck and Laurie Zarig, 109–37. Dordrecht: Kluwer.

186

Chapter Ten

Lakoff, George. 1968. “Some Verbs of Change and Causation.” In Mathematical Linguistics and Automatic Translation, Report NSF-20, edited by Susumu Kuno, 1–27. Cambridge, MA: Harvard University Press. Laskowski, Roman. 1984. “Predykatyw.” [Predicative]. In Morfologia [Morphology], edited by Renata Grzegorczykowa, Roman Laskowski, and Henryk Wróbel, 171–219. Warszawa: PWN. Levin, Beth, and Malka Rappaport Hovav. 1995. “Unaccusativity: At the Syntax-Lexical Semantics Interface.” Linguistic Inquiry Monograph 26. Cambridge, MA: MIT Press. àazorczyk, Agnieszka. 2010. “Decomposing Slavic Aspect: The Role of Aspectual Morphology in Polish and Other Slavic Languages.” PhD diss., University of Southern California. MacDonald, Jonathan. 2008. The Syntactic Nature of Inner Aspect: A Minimalist Perspective. Amsterdam: John Benjamins. Malicka-Kleparska, Anna. to appear. “Polish Anticausative MorphoSyntax: A Case for a Root-Based Model against Lexicalist Reflexivization.” Submitted to New Insights into Slavic Linguistics. —. 2012. “Limitations on Involuntary-state Constructions in Polish: Between Structure and Semantics.” Roczniki Humanistyczne LX: 65– 84. Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge, MA: MIT Press. Olszewska, Teresa. 1986. Causativity as a Linguistic Phenomenon. A Study Based on English and Polish. Lublin: UMCS. Przepiórkowski, Adam, Mirosáaw BaĔko, Rafaá Górski, and Barbara Lewandowska-Tomaszczyk, eds. 2012. Narodowy Korpus JĊzyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN. Pylkkänen, Liina. 2008. “Introducing Arguments.” Linguistic Inquiry Monograph 48. Cambridge, MA: MIT Press. Szymanek, Bogdan. 2010. A Panorama of Polish Word-Formation. Lublin: Wydawnictwo KUL. Van Valin, Robert D. 2013. “Lexical Representation, Co-composition, and Linking Syntax and Semantics.” In Advances in Generative Lexicon Theory, edited by James Pustejovsky, Pierre Bouillon, Hitoshi Isahara, Kyoko Kanzaki, and Chungmin, 67–107. Dordrecht: Springer. Vendler, Zeno. 1957. “Verbs and Times.” The Philosophical Review 66:143–60. —. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press.

Verbal Causative Structure for Polish Prefixed Transitives

187

Willim, Ewa. 2006. Event, Individuation and Countability. A Study with Special Reference to English and Polish. Kraków: Jagiellonian University Press. Wróbel, Henryk. 1984. “Sáowotwórstwo Czasowników.” [Word Formation of Verbs]. In Morfologia [Morphology], edited by Renata Grzegorczykowa, Roman Laskowski, and Henryk Wróbel, 467–511. Warszawa: PWN.

CHAPTER ELEVEN THE LINEARIZATION OF ADJECTIVES IN POLISH NOUN PHRASES: SELECTED SEMANTIC AND PRAGMATIC FACTORS BOĩENA CETNAROWSKA

1. Introduction The present chapter investigates the pre-head and post-head position of attributive adjectives in Polish within the framework of the representational theory postulated in Bouchard (2002, 2009) and applied to Russian in Trugman (2010, 2011). It describes various modes of the semantic composition of adjectival modifiers with their head nouns. While the previous research on Polish adjectival modifiers emphasizes the link between the pre-nominal or post-nominal position of the adjectives and their classifying or qualifying function, as in linia krzywa (Noun + Classifying Adjective) ‘a curve’ vs. krzywa linia (Qualifying Adjective + Noun) ‘a line which happens to be curved’ (cf. Willim 2000; SzymaĔska 2000; Rutkowski and Progovac 2005; Rutkowski 2009), I will focus here on pre-nominal adjectival modifiers with the classifying function. Subtle distinctions between the semantic interpretation of preN and postN classifying adjectives will be discussed. Moreover, I will exemplify the influence of pragmatic factors, such as the contrastive marking of the adjective on its pre-head placement. Finally, it will be shown (on the basis of examples taken mainly from the National Corpus of Polish,1 abbreviated as NKJP) that the speaker/writer can consciously 1

The National Corpus of Polish (i.e., Narodowy Korpus JĊzyka Polskiego) is described in Przepiórkowski et al. (2012). It is available online at http://nkjp.pl.

The Linearization of Adjectives in Polish Noun Phrases

189

exploit the ambiguity between a classifying and a qualifying interpretation of the pre-head attributive adjective. The layout of the chapter is as follows. Section 2 explains the distinction between qualifying and classifying adjectives, illustrating the contrast in their semantic interpretation and placement within noun phrases in Polish. Section 3 presents a brief outline of the analysis of combinations of nouns and adjectives in Polish proposed in Cetnarowska, Pysz, and Trugman (2011a, 2011b). Section 4 exemplifies some semantic differences between noun phrases consisting of N+A combinations and those containing A+N sequences. It also mentions the issue of the noncompositionality (or partial compositionality) of N+A and A+N combinations. Section 5 shows how information structure can determine the pre-head (rather than the post-head) position of CAs2 in Polish (i.e., in the case of the contrastive stress on the adjective). Section 6 discusses yet another factor which may influence the placement of CAs in a Polish noun phrase, namely the speaker’s intention to activate both the classifying and the qualifying senses of a given adjectival form. Conclusions are stated in section 7.

2. The distinction between classifying and qualifying adjectives in Polish Attributive adjectives in Polish are traditionally divided into classifying and qualifying modifiers (see, among others, Jadacka 2005). Classifying adjectives (abbreviated here as CAs) express crucial (classifying) features of entities denoted by their head nouns, and are expected to occur in the post-head position. In contrast, qualifying adjectives (QAs) are placed prenominally and denote less essential (or accidental) properties of the entities denoted by head nouns. This is shown in the examples in (1)–(3)

2

The following abbreviations are used in the chapter: A(dj), adjective; CA, classifying adjective; CLP, Central Linearization Parameter; N, noun; NKJP, National Corpus of Polish; Num, (semantic) Number; QA, qualifying adjective. Abbreviations used in the glosses are as follows: ACC, accusative; ADJ, adjective; DIM, diminutive; GEN, genitive; INF, infinitive; INS, instrumental; N, noun; NOM, nominative; TOP, particle functioning as a topic marker. The abbreviation 3SG.F represents the third person singular feminine form.

Chapter Eleven

190

below, in which the difference between the interpretation of QA+N sequences and N+CA combinations is easy to discern.3 (1)

a. szara foka (QA+N) grey seal ‘a seal which happens to be grey’ b. foka szara (N+CA) seal grey ‘a/the grey seal (as a species or a representative of the species)’

(2)

a. fantastyczny film fantastic film ‘a fantastic (superb) film’

(QA+N)

b. film fantastyczny (N+CA) film fantastic ‘a science fiction or fantasy film’ (3)

a. specjalna szkoáa special school ‘a special, unique school’

(QA+N)

b. szkoáa specjalna (N+CA) school special ‘a school for children with disabilities’

However, some N+A and A+N sequences have roughly the same interpretation, which testifies to the possibility of placing classifying adjectives in the pre-head position (cf. Willim 2000, 2011; Cetnarowska, Pysz, and Trugman 2011a, 2011b).4 (4)

3

a. drukarka laserowa printer laser-ADJ ‘a laser printer’

(N+CA)

More examples of the differences between the meaning of adjectives in classifying and qualifying functions can be found in SzymaĔska (2000), Willim (2000, 2001), Rutkowski and Progovac (2005), Rutkowski (2009) and Cetnarowska, Pysz, and Trugman (2011a, 2011b), among others. 4 This is against the view, expressed by, among others, Rutkowski and Progovac (2005) and Rutkowski (2009), that pre-head adjectival modifiers in Polish are necessarily qualifying ones.

The Linearization of Adjectives in Polish Noun Phrases

(5)

b. laserowa drukarka laser-ADJ printer ‘a laser printer’

(CA+N)

a. dĪinsy tureckie jeans Turkish ‘Turkish jeans’

(N+CA)

b. tureckie dĪinsy Turkish jeans ‘Turkish jeans’

(CA+N)

191

As discussed in Cetnarowska, Pysz, and Trugman (2011a, 2011b), a CA frequently precedes the head noun when there is another postmodifier in the same noun phrase,5 e.g., another CA, a genitive phrase, or a prepositional phrase (as in (6)). (6)

a. laserowa drukarka kolorowa laser-ADJ printer colour-ADJ ‘a colour laser printer’ b. olimpijskie medale Adama Olympic medals-NOM Adam-GEN ‘Adam Maáysz’s Olympic medals’

Maáysza Maáysz-GEN

c. laserowa drukarka ze skanerem laser-ADJ printer with scanner-INS ‘a laser printer with a scanner’ d. Specjalna Szkoáa Podstawowa dla Dzieci z Autyzmem special school primary for children with autism ‘primary school for autistic children’

However, the discussion below will focus on classifying adjectives which occur pre-nominally, although the post-head position, canonical for CAs, is not occupied by another post-modifier. First, a brief overview will be given of the account of adjectival classifying attributes offered in Cetnarowska, Pysz, and Trugman (2011a, 2011b).

5

The influence of various structural factors on the linearization of attributive adjectives in Polish noun phrases is discussed in GĊbka-Wolak (2000) and TopoliĔska (1984).

192

Chapter Eleven

3. Accounting for the distinction between Polish CAs and QAs in Cetnarowska, Pysz, and Trugman (2011a, 2011b) Cetnarowska, Pysz, and Trugman (2011a, 2011b) adopt for Polish the representational theory which Bouchard (1998, 2002) proposed for adjectival modification in French and English. They assume that the interpretation and the placement of adjectives depend primarily on the mode in which attributive adjectives combine with their head nouns, and on the manner in which semantic Number is coded in a given language. In French, for instance, Number is coded on the Determiner (as is argued by Bouchard 1998, 2002). In Polish, as in English, semantic Number is coded on the Noun and it gets morphologically realized on the Noun’s dependents, such as adjectives, possessives and demonstratives. Since Number is coded as a suffix (i.e., an inflectional ending) in Polish, it can have scope only over elements which occur to its left. Consequently, modifiers of nouns are expected to occur in the pre-head position (by virtue of the Number Scope Condition, postulated by Bouchard 1998, 2002), as is exemplified by possessives, demonstratives, and attributive adjectives such as the qualifying ones in (1a, 2a, 3a), and the classifying ones in (4b, 5b). A further component of the representational theory is Bouchard’s assumption that a semantic entry for a common noun consists of a network of functions. These include, among others, a characteristic function f (“which provides the property that interprets the N,” as stated in Kamp and Partee 1995, 131), an indication of the possible world w, a specification of the time interval i, and a variable assignment function g (which determines the truth value of the final formula). As explained by Bouchard (2002, 8), “[t]his network of elements determines the set of things that have the property of being a f in w at i.” Moreover, Bouchard (1998, 2002) asserts that the placement of dependents (i.e., modifiers and arguments) follows from the languagespecific setting of the so-called Central Linearization Parameter (CLP), which determines whether the functor precedes or follows the dependent. The CLP is set for French so that the functor precedes its modifiers and/or arguments, hence adjectives are expected to follow nouns. French attributive adjectives, whose placement can be predicted on the basis of the CLP, modify all the subcomponents (i.e., all the functions) of the head noun. This is the case of adjectives which can be regarded as intersective ones, e.g., ronde ‘round’ in the phrase une table ronde ‘a round table.’ Adjectives which occupy another position in the French noun phrase can modify only a subpart of N, e.g., the characteristic function f, or the time

The Linearization of Adjectives in Polish Noun Phrases

193

interval i at which the characteristic function f holds (e.g., the adjective futur ‘future’ in the phrase le futur président ‘the future president’). This follows from the Principle of Other Value (cf. Bouchard 2009).6 The CLP is set for Polish so that the functor should precede its dependents as, for instance, verbs and prepositions precede their complements in Polish (see Cetnarowska, Pysz, and Trugman 2011a, 2011b). This could lead to the (unwanted) prediction that all Polish intersective modifiers should occur post-nominally while non-intersective ones should be pre-nominal. Some pre-nominal adjectives in Polish are indeed non-intersective and modify a subpart of N, e.g., the adjective faászywy ‘false’ in the phrase faászywy ksiądz ‘a false priest,’ or the adjective byáy ‘former’ in byáy prezydent ‘a former president.’ However, since the element coding semantic Number in Polish is an inflectional ending and it forms a component with the Noun (represented as [N+Num]), pre-nominal adjectives can modify a part of the [N+Num] complex which either corresponds to a subpart of N (as in the case of faászywy ksiądz ‘false priest’), or to the whole of N. The latter situation holds for pre-head qualifying adjectives in (1a, 2a, 3a) and for pre-head classifying adjectives in (4b, 5b). The extension of the QA+N sequence specjalna szkoáa ‘superb/unique school’ in (3a) is an intersection of the sets of schools and the sets of entities described as being unique or special. The CA+N combination laserowa drukarka ‘laser printer’ in (4b) can be, similarly, analysed as having an extension which is the intersection of two sets: the set of printers and the set of objects associated (in one way or another) with a laser. The post-head position of Polish adjectives calls for an additional theoretical explanation since the inflectional Number marker on the noun cannot have scope over elements occurring to its right. Cetnarowska, Pysz, and Trugman (2011a, 2011b) propose that post-head (classifying) adjectives in Polish merge with nouns at a stage where the noun is not atomized (i.e., when the N is not assigned semantic Number). It is the resulting N+CA combination which functions as a complex predicate, referred to as a ‘tight unit’ in Cetnarowska, Pysz, and Trugman (2011a, 2011b) and which is assigned semantic Number as a whole.

6

The Principle of Other Value (in Bouchard 2009) is referred to as the Elsewhere application of the Central Linearization Parameter in earlier works by Bouchard, e.g., in Bouchard (1998, 2002). It predicts for French that pre-head adjectives require some “special” interpretation since they modify a subpart of N.

Chapter Eleven

194

4. Semantic differences between N+CA and CA+N combinations The distinct modes of the semantic composition of pre-head and post-head classifying adjectives in Polish assumed by Cetnarowska, Pysz, and Trugman (2011a, 2011b) imply that N+CA and CA+N combinations can differ in their semantic interpretation. Among others, classifying adjectives which occur in postposition as constituents of N+CA complex predicates, exhibit a closer bond with their head nouns and are often dependent for their interpretation on the meaning of the head. One can compare, in this respect, the semantic contribution of the adjective wodny ‘relating to water’ to the meaning of the N+CA ‘tight units’ such as para wodna ‘water steam,’ áóĪko wodne ‘water bed,’ tor wodny ‘shipping route’ and narty wodne ‘water skis.’ It is not obvious at first sight that there should be any difference in the meaning of N+CA and CA+N combinations, such as those mentioned in (4) and (5), or in the pair given below in (7): (7)

a. radiowóz policyjny squad-car police-ADJ ‘a police car’ b. policyjny radiowóz police-ADJ squad-car ‘a police car’

One potential difference noticeable between (7a) and (7b) is a stylistic one since N+CA sequences are characteristic of (though not confined to) a formal style of speaking and of specialist texts. This is illustrated in (8) below: (8)

a. komisariat policji w Przodkowie otrzymaá od police-station-NOM police-GEN in Przodków received-3SG from UrzĊdu Gminy radiowóz policyjny Polonez, faks, office commune-GEN squad-car-ACC police-ADJ Polonez fax-ACC a takĪe alkotest and also breathalyser-ACC ‘The police station in Przodków received a Polonez police squad car, a fax machine as well as a breathalyser from the Commune Office.’ [NKJP]

The Linearization of Adjectives in Polish Noun Phrases

195

b. Do dyspozycji w gminie przez caáą dobĊ jest takĪe for disposal in commune for whole day is also samochód policyjny. car-NOM police-ADJ ‘A police car is also at (their) disposal in the commune (in the borough) for 24 hours a day.’

The stylistic difference follows from the semantic distinction (between CA+N and N+CA strings). Cetnarowska and Trugman (2012) observe that CAs occurring in N+CA combinations form taxonomic kinds of names. The usage of such names is expected in scientific (and specialist) discourse, rather than in the informal (spoken) variety of Polish. Moreover, the N+CA complex predicates may exhibit a lower degree of compositionality (i.e., of transparency) than the corresponding CA+N sequences. This is exemplified in (9) and (10). Although the N+CA combinations in (9) and (10) are motivated and analysable units (cf. Svensson 2008), they are not fully transparent since, without any previous knowledge of such expressions, a language user cannot fully understand what they mean. (9)

a. lody wáoskie ice cream Italian ‘soft serve ice cream’ b. wáoskie/ polskie/ niemieckie lody Italian/ Polish/ German ice cream ‘ice-cream produced in Italy/Poland/Germany’

(10)

a. strajk wáoski strike Italian ‘work-to-rule, i.e., a type of industrial action in which the strikers follow all the regulations, which results in drastically slower performance’ b. wáoskie/ polskie/ niemieckie strajki Italian/ Polish/ German strikes ‘strikes in Italy/Poland/Germany’

The pre-head ethnic adjective indicates the origin/provenance of the head noun, and thus the A+N string wáoskie lody ‘Italian ice cream’ in (9b) can be interpreted as referring to any type of ice cream produced in Italy. On the other hand, the N+CA sequence lody wáoskie in (9a) denotes a specific type of ice cream, namely soft serve ice cream. Similarly, the N+CA sequence strajk wáoski (lit. strike Italian) ‘work-to-rule’ is not a

Chapter Eleven

196

fully transparent expression and it requires some encyclopaedic knowledge on the part of the speaker and the hearer to be understood fully. In the typology of combinations of nouns and classifying adjectives presented by Cetnarowska, Pysz, and Trugman (2011a, 2011b), N+CA ‘tight units’ are kept distinct from CA+N strings regarded as lexical idioms. The latter type of (idiomatic) CA+N unit is exemplified in (11): (11)

a. niebieski ptak blue bird ‘a loafer’ b. boĪa krówka God-ADJ cow-DIM ‘a ladybird’

The CAs in (11a) and (11b) do not modify the noun predicate but change its characterizing function, i.e., its core meaning. Therefore, the idiom niebieski ptak ‘a loafer’ does not denote a type of a bird, nor does boĪa krówka ‘a ladybird’ denote a type of a little cow. Cetnarowska and Trugman (2012) discuss briefly discontinuous noun phrases in Polish which contain classifying adjectives. It is shown that lexical idioms, such as boĪa krówka ‘a ladybird’ are inseparable, hence the CA cannot ‘move out’ of the CA+N idiomatic predicates. (12)

* Nawet boĪa tu nie przylatuje krówka, even god-ADJ.NOM here not fly-PRES.3SG.F cow-DIM.NOM (nie tylko motyle) (not only butterflies-NOM) *in the reading ‘Even a ladybird doesn't fly in here, not only butterflies’ (from Cetnarowska and Trugman 2012, 148)

Classifying adjectives which occur as constituents of N+CA combinations are not as likely to move out of ‘tight units’ as pre-head (intersective) CAs, yet they can participate in various scrambling phenomena. They sound more acceptable in discontinuous noun phrases when there is an overt contrast provided with another type of the entity denoted by the head noun, as shown in (13).

The Linearization of Adjectives in Polish Noun Phrases (13)

197

a. ?? Nawet fantastycznego nie chciaáa filmu ze mną even fantastic-GEN not wanted-3SG.F film-GEN with me obejrzeü. watch-INF ‘She did not want to watch with me even a science-fiction film.’ b. Nawet fantastycznego nie chciaáa filmu ze mną even fantastic-GEN not wanted-3SG.F film-GEN with me obejrzeü, a nie tylko dokumentalnego. watch-INF, and not only documentary-GEN ‘She did not want to watch with me even a science-fiction film, and not only a documentary.’

In the next section, it will be shown that the presence of contrastive stress can account for the acceptability of the CA+N order in the case of classifying adjectives which form tight units and which would, therefore, not be expected to occur in the pre-head position.

5. Contrastive stress and the placement of CAs It was shown in (10) above that there is a difference between the interpretation of the N+CA tight unit strajk wáoski ‘work to rule’ and the CA+N sequence wáoski strajk ‘Italian strike, i.e., a strike in Italy, or/and by Italian workers,’ as well as between lody wáoskie ‘soft serve ice cream’ and wáoskie lody ‘ice cream produced in Italy.’ However, examples can be found in the National Corpus of Polish where the CA+N sequence is employed with a meaning that is characteristic of the N+CA complex predicate. (14)

a. POLSKA REFORMA, WàOSKI STRAJK. Polish reform Italian strike DziĞ rząd ocenia pierwsze tygodnie weeks today government.NOM assess-3SG first reformy sáuĪby zdrowia. reform-GEN service-GEN health-GEN ‘Polish reform, Italian strike (=work-to-rule). Today the (Polish) government is assessing the first weeks of the national health service reform.’ [NKJP]

198

Chapter Eleven b. Wáoski strajk w prywatnych kamienicach Italian strike in private tenement-houses ‘Italian strike (work-to-rule) in privately-owned tenement houses’ [NKJP] c. AmerykaĔski samochód i wáoskie lody byáy dostĊpne American car and Italian ice-cream were available w ten sam sposób, co urok Liz Taylor, in this same way as charm Liz Taylor-GEN Audrey Hepburn albo Debby Reynolds Audrey Hepburn-GEN or Debby Reynolds-GEN ‘An American car and Italian ice cream (soft serve ice cream) were available in the same way as the charm of Liz Taylor, Audrey Hepburn or Debby Reynolds’ [NKJP]

Although the examples in (14) come from a written corpus (i.e., from a newspaper source), it can be assumed that the two adjectives in the headline, or in the sentences taken from the body of a given newspaper article, bear a contrastive stress. This allows the writer to compare and juxtapose the names of the two nationalities (Polish vs. Italian, American vs. Italian) or the adjective wáoski ‘Italian’ and prywatny ‘privatelyowned.’ Another sentence from NKJP exemplifying the influence of contrastive stress (and parallel syntactic and rhythmic structures) on adjective placement is given in (15):7 (15)

7

Oto trzy armie radiowców oraz kilka dywizji pomniejszych here three armies radiomen-GEN and several divisions smaller stacji w bitwie o PolskĊ B – orĊĪem są konkursowe stations-GEN in battle for Poland B weapon-INS are contest-ADJ nagrody, Ğpiewające gwiazdy, laserowe pokazy, wideoklipy, prizes singing stars laser-ADJ shows videoclips baloniki i cukrowa wata. balloons and sugar-ADJ candy ‘Here are three armies of radio presenters and several divisions of smaller (radio-)stations in the battle for Poland B; the weapons include prizes in (radio) contests, singing stars, laser shows, video clips, balloons and candy floss.’ [NKJP]

Further examples of the influence of information structure on the linearization of adjectives can be found in Cetnarowska (2014).

The Linearization of Adjectives in Polish Noun Phrases

199

The A+N sequences with pre-head classifying adjectives in the above sentence, such as konkursowe nagrody (lit. contest.ADJ prizes), laserowe pokazy (lit. laser.ADJ shows) and cukrowa wata (lit. sugar.ADJ candy) are typically restated as N+A ‘tight units,’ namely nagrody konkursowe (lit. prizes contest.ADJ), pokazy laserowe (lit. shows laser.ADJ) and wata cukrowa (lit. candy sugar.ADJ), when there is no influence from parallel structures or contrastive stress on the placement of adjectives.8

6. Word play and pre-head attributive adjectives At first sight, the example in (16) below appears to illustrate the phenomenon discussed in the previous section, i.e., the occurrence of prenominal CAs in parallel (contrastive) structures in noun phrases. (16)

Specjalna szkoáa pod specjalnym nadzorem. special school under special supervision ‘A school for children with disabilities under special supervision’ Gáówno. Zespóá Szkóá Specjalnych w Gáownie objĊty zostaá specjalnym nadzorem pedagogicznym ze strony áódzkiego kuratorium oĞwiaty. ‘Gáówno. The Special School Unit in Gáówno has been placed under the supervision of the àódĨ schools’ superintendent.’ [NKJP]

Both instances of the pre-head adjective specjalny ‘special’ in the initial (headline) noun phrase in (16) are emphasized. Moreover, while in the phrase specjalny nadzór ‘special supervision’ the pre-head adjective occurs in the qualifying reading, the phrase specjalna szkoáa (lit. special school) is ambiguous between the qualifying and the classifying senses of the pre-head adjective. The difference between the qualifying and the classifying usage of the adjective in question, mentioned in section 2 in (3), is further illustrated in (17).

8

In the full NKJP corpus (searched via the Pelcra search engine), there are 31 occurrences of A+N konkursowe nagrody compared to 34 occurrences of N+A nagrody konkursowe. There are 5 occurrences of A+N laserowe pokazy compared to 46 occurrences of N+A pokazy laserowe; as well as 73 occurrences of A+N cukrowa wata vs. 181 occurrences of N+A wata cukrowa.

200 (17)

Chapter Eleven a. szkoáa specjalna hotelarska school special hotel-ADJ ‘a school for hotel staff for disabled learners’ b. kamerdynerów szkoli specjalna szkoáa w Londynie school-NOM in London butlers-ACC train-3SG special ‘Butlers are trained in a special school in London’ [NKJP]

The first sentence of the main text from the newspaper excerpt in (16), following the headline, makes it clear that what is meant is a school for children with disabilities, not a school which is unique in some other respect. However, when encountering the headline, the reader may at first decide on the qualifying interpretation of the adjective (i.e., ‘special, unique’), and only later he or she may find out that the classifying interpretation is more appropriate. This may be a conscious decision of the writer to make potential readers simultaneously retrieve both senses of the adjective specjalny exemplified in (17). Such recourse to a particular type of word-play, i.e., paronomasia, is characteristic of journalistic language, especially newspaper headlines. Further illustrations of the same phenomenon are given in (18) and (19). In (18a, b), the difference is exemplified between the qualifying usage of the adjective naturalny ‘natural, expected, matter-of-course’ and the classifying usage of the same adjective in the phrase katastrofa naturalna ‘a natural disaster (one resulting from the natural processes of the Earth).’ The noun phrase in (18c) exploits the contrast between the dual senses of the adjective in question (for rhetorical effect), thus giving rise to a pun. (18)

a. naturalna katastrofa (QA+N) natural disaster ‘natural (i.e., usual or expected) disaster’ b. katastrofa naturalna (N+CA) disaster natural ‘natural disaster (i.e., resulting from natural processes of the Earth)’ c. Katrina: nienaturalnie naturalna katastrofa Katrina unnaturally natural disaster ‘(Hurricane) Katrina: unnaturally natural disaster’ [NKJP]

The Linearization of Adjectives in Polish Noun Phrases

201

The ambiguity of the adjective fantastyczny ‘1. fantastic, superb (QA); 2. relating to science fiction or fantasy (CA)’ is visible in (19a) and (19b). (19b) contains a slogan with the name of an art festiwal (Festiwal Fantastyczny Biaáystok), where the occurrence of paronomasia is certainly a conscious choice of the writer. (19)

a. Fantastyczny Gimenez (...) Porządna dawka fantastyki Fantastic Gimenez decent dose fantasy najwyĪszych lotów top-flight ‘Fantastic Gimenez (…) A decent dose of top-flight fantasy’ b. Festiwal Fantastyczny Biaáystok to wielkie wydarzenie festival fantastic Biaáystok TOP great event artystyczne, wspólnym mianownikiem jest szeroko rozumiana artistic common denominator is widely understood fantastyka. fantasy ‘The Festival called ‘Fantastic Biaáystok’ (Fanstok) is a great artistic event, whose common theme is science fiction or fantasy (in the broad understanding of the term)’

In section 3, two distinct modes were postulated in which an adjective combines with the head noun: adjectives either combine with bare N predicates (i.e., non-atomized Ns), giving rise to complex predicates (‘tight units’) consisting of N+CA, or the adjectives merge with [N+Num] complexes (i.e., atomized nouns). It seems that both modes of adjective modification are employed simultaneously in the case of the ambiguous A+N combinations discussed in this section. Cetnarowska and Trugman (2011) suggest that intersective (predicating) adjectives are shifted into subsective (non-predicating) adjectives by some kind of semantic bleaching, which results in a change of entailments. For instance, afrykaĔski ‘African’ in afrykaĔskie dziecko ‘African child’ is an entailing intersective adjective, which predicates an independent property of the head noun (i.e., the property of coming from and living in Africa). The property of Adj holds for each instantiation of the kind denoted by a noun+adjective combination. When a ‘tight unit’ (i.e., a complex predicate N+CA) is formed, the adjective is semantically bleached and non-entailing since the property of Adj holds only for a stereotypical instantiation of the kind. For example, sáoĔ afrykaĔski (lit. elephant African) ‘African elephant’ denotes a natural kind (species) of animal which usually lives in Africa, but some representatives of the species can be found in European or American zoos.

202

Chapter Eleven

In a similar way, it could be argued here that the adjective wáoski ‘Italian’ is entailing and intersective in wáoska ĪywnoĞü ‘Italian food’ or wáoskie lody ‘Italian ice cream,’ whereas it is non-entailing and semantically bleached in the expression lody wáoskie ‘soft serve ice cream.’ While soft serve ice cream sold in Poland (in the 1980s) used to be produced in Italian-made soft cream machines, it is not necessary that soft ice cream should be produced in Italy, by Italians, or with the use of Italian-made equipment. Furthermore, it appears that this type of semantic bleaching of entailing intersective adjectives can be reversed in the case of paronomasia, attested in the newspaper headlines discussed in this section.9

7. Conclusion The present chapter considered the placement and interpretation of attributive adjectives in Polish. While the canonical position of classifying adjectives is the post-head one, cases were investigated of the pre-head occurrence of such adjectives, due to selected semantic and pragmatic factors. The pragmatic factors involved the presence of contrastive stress, which favours the pre-head placement of CAs in parallel “A+N and A+N” sequences. The difference between the semantic interpretation of N+CA complex predicates (‘tight units’) and A+N sequences was illustrated. The partial compositionality of ‘tight units,’ such as strajk wáoski (lit. Italian strike) ‘work-to-rule,’ was demonstrated. It was shown that a semantic factor causing the pre-head position of the CA is sometimes the writer’s intention to exploit the ambiguity between a classifying and a qualifying reading of the attributive adjective in Polish noun+adjective combinations.

9

Putting it in different terms, Cetnarowska and Trugman (2012) propose that intersective adjectives (either CAs or QAs) originate as types and they are shifted into subsective CAs of the type when forming tight units (N+CAs) with bare (non-atomized) nouns. It seems that such subsective CAs undergo a mirror type-shifting operation (cf. Partee 1995), coercing them into intersective adjectives under the influence of the context, e.g., in the examples of newspaper headlines discussed above.

The Linearization of Adjectives in Polish Noun Phrases

203

References Bouchard, Denis. 1998. “The Distribution and Interpretation of Adjectives in French: A Consequence of Bare Phrase Structure.” Probus 10(2):139–83. —. 2002. Adjectives, Number and Interfaces: Why Languages Vary. Oxford: Elsevier Science. —. 2009. “A Solution to the Conceptual Problem of Cartography.” In Alternatives to Cartography, edited by Jeroen van Craenenbroeck, 245–74. Berlin and New York: Mouton de Gruyter. Cetnarowska, BoĪena. 2014. “On Prenominal Classifying Adjectives in Polish.” In Advances in the Syntax of DPs. Structure, Agreement, and Case, edited by Anna Bondaruk, Gréte Dalmi, and Alexander Grosu, 221–46. Amsterdam and Philadelphia: John Benjamins. Cetnarowska, BoĪena, Agnieszka Pysz, and Helen Trugman. 2011a. “Accounting for Some Flexibility in a Rigid Construction.” In Generative Investigations: Syntax, Morphology and Phonology, edited by Piotr BaĔski, Beata àukaszewicz, Monika OpaliĔska, and Joanna Zaleska, 24–47. Newcastle upon Tyne: Cambridge Scholars Publishing. —. 2011b. “Distribution of Classificatory Adjectives and Genitives in Polish NPs.” In On Words and Sounds: A Selection of Papers from the 40th PLM, 2009, edited by Kamila DĊbowska-Kozáowska and Katarzyna Dziubalska-Koáaczyk, 273–303. Newcastle upon Tyne: Cambridge Scholars Publishing. Cetnarowska, BoĪena, and Helen Trugman. 2011. “Falling between the Chairs: Are Classifying Adjective+Noun Complexes Lexical or Syntactic Formations?” Paper presented at GLiP-7, Wrocáaw, December 2–4. —. 2012. “Falling between the Chairs: Are Classifying Adjective+Noun Complexes Lexical or Syntactic Formations?” In Current Issues in Generative Linguistics: Syntax, Semantics and Phonology, edited by Joanna Báaszczak, BoĪena Rozwadowska, and Wojciech Witkowski, 138–54. Wrocáaw: CGLiW. http://www.ifa.uni.wroc.pl/linguistics/GLiW-CIGL2012.pdf/. GĊbka-Wolak, Maágorzata. 2000. Związki Linearne miĊdzy Skáadnikami Grupy Nominalnej we Wspóáczesnym JĊzyku Polskim. ToruĔ: Wydawnictwo UMK. Jadacka, Hanna. 2005. Kultura JĊzyka Polskiego: Fleksja, Sáowotwórstwo, Skáadnia. Warszawa: Wydawnictwo Naukowe PWN.

204

Chapter Eleven

Kamp, Hans, and Barbara Partee. 1995. “Prototype Theory and Compositionality.” Cognition 57:129–91. Partee, Barbara. 1995. “Lexical Semantics and Compositionality.” In Invitation to Cognitive Science, 2nd edition, edited by Daniel Osherson; in Part I: Language, edited by Lila Gleitman and Mark Liberman, 311–60. Cambridge, MA: MIT Press. Przepiórkowski, Adam, Mirosáaw BaĔko, Rafaá L. Górski, and Barbara Lewandowska-Tomaszczyk, eds. 2012. Narodowy Korpus JĊzyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN. Rutkowski, Paweá. 2009. Fraza Przedimkowa w PolszczyĨnie. Warszawa: Wydawnictwo UW, Wydziaá Polonistyki. Rutkowski, Paweá, and Ljiliana Progovac. 2005. “Classification Projection in Polish and Serbian: The Position and Shape of Classifying Adjectives.” In Formal Approaches to Slavic Linguistics: The South Carolina Meeting, edited by Steven Franks, Frank Y. Gladney, and Mila Tasseva-Kurktchieva, 289–99. Ann Arbor, MI: Michigan Slavic Publications. Svensson, Maria Helen. 2008. “A Very Complex Criterion of Fixedness: Non-Compositionality.” In Phraseology: An Interdisciplinary Perspective, edited by Sylviane Granger and Fanny Meunier, 81–93. Amsterdam: John Benjamins. SzymaĔska, Izabela. 2000. “On the Adjectival Modification in Polish Noun Phrases – from the X-bar Theory towards a Functional Explanation.” Acta Universitatis Lodziensis. Folia Linguistica Anglica 2:175–92. TopoliĔska, Zuzanna. 1984. “Skáadnia Grupy Imiennej.” In Gramatyka Wspóáczesnego JĊzyka Polskiego. Skáadnia, edited by Zuzanna TopoliĔska, 301–89. Warszawa: PWN. Trugman, Helen. 2010. “Modifiers of Bare Nouns in Russian.” In Current Issues in Formal Slavic Linguistics, edited by Gerhild Zybatow, Philip Dudchuk, Serge Minor, and Ekaterina Pshehotskaya, 245–70. Frankfurt am Main: Peter Lang. —. 2011. “Instances of Posthead Modification in Russian NPs.” In On Words and Sounds: A Selection of Papers from the 40th PLM, 2009, edited by Kamila DĊbowska-Kozáowska and Katarzyna DziubalskaKoáaczyk, 247–79. Newcastle upon Tyne: Cambridge Scholars Publishing. Willim, Ewa. 2000. “Some Aspects of the Grammar and Interpretation of Adjectival Modification.” In Proceedings of Generative Linguistics in Poland 1, edited by Piotr BaĔski and Adam Przepiórkowski, 156–67. Warszawa: IPIPAN.

The Linearization of Adjectives in Polish Noun Phrases

205

—. 2001. “On NP-Internal Agreement: A Study of Some Adjectival and Nominal Modifiers in Polish.” In Current Issues in Formal Slavic Linguistics, edited by Gerhild Zybatow, Uwe Junghanns, Grit Mehlhorn, and Luka Szucsich, 80–95. Frankfurt am Main: Peter Lang.

CHAPTER TWELVE ONOMASIOLOGICAL STRATEGIES AND POTENTIAL FORMS: IN SEARCH OF THE FEMALE COUNTERPART OF MINISTER IN POLISH MARIA BLOCH-TROJNAR

1. Introduction Not so long ago, we witnessed a heated public debate relating to the formation and the use of forms of address of women holding important political positions, i.e., that of a minister or prime minister. In a popular political chat-show, the host asked his guest, the Minister of Sport and Tourism, Joanna Mucha, which form of address she liked best: the usual pani minister, ministra or ministerka. To the horror of the majority of viewers, commentators and linguists, she said she favoured the term ministra. The only group that took to the new formation were feminist activists, who waved the ideological banner of equality and advocated social change through the medium of language. They held up Germany as an excellent example of where an appropriate bill was passed enforcing the formation and use of female names. The fact that the neologism posáanka ‘female MP’ had entered the Polish language was presented as evidence that the feminisation of Polish was possible. Such was the argument that the Council of the Polish Language felt obliged to issue an official statement which said that

Onomasiological Strategies and Potential Forms

207

forms such as ministerka, premierka or ministra, premiera are well-formed but they are felt by the majority of native speakers to be artificial and awkward, and they cannot enter the language by force.1

The fact that a given form seems awkward, strange, funny or in any way unacceptable is not just a matter of taste, outlook or education. It stems from an unconscious knowledge or linguistic competence with which every language user is equipped. A potential word in morphological analysis is a lexical item which, in functional and semantic terms, can be regarded as the product of a specific Word Formation Rule (WFR) (Allen 1978; Malicka-Kleparska 1985, 1987). For example, forms such as fajtáapowatoĞü ‘clumsiness’ or ekstermalnoĞü ‘extremeness, extremity,’ which are not listed in any standard dictionary (Smóákowa 2004), are potential words on account of the fact that there exists in Polish a WFR generating Nomina Essendi from practically any qualitative noun and the suffix -oĞü is its most productive exponent (Grzegorczykowa, Laskowski, and Wróbel 1999, 416–17). The aim of this chapter is to identify the systemic underpinnings of the grammaticality judgements of speakers, i.e., to demonstrate why in linguistic terms the form (pani) minister is optimal.

2. Why not ministra? The form ministra can be regarded as well-formed if we treat it as a borrowing from Latin. In Latin ministra, -ae ‘female servant, assistant’ is the counterpart of minister, -tri ‘male servant.’ The word minister has got a surprising etymology since it originally denoted someone of lower rank, a servant and was derived from ministrare ‘serve, wait on.’ It was an antonym of magister ‘a superior.’ Thus, there existed male – female doublets such as magister – magistra and minister – ministra. However, in contradistinction to the male form of these nouns, Nomina Feminativa, i.e., ministra and magistra, have never been adopted by Polish speakers. There is no mention of them in dictionaries. A plausible explanation of why the female name has not caught on is because it has no overt exponent (in the guise of an affix), whereby it could be identified. It is formed by means of a paradigmatic derivation or conversion. The 1

The official statement of the Council of the Polish Language was issued on 19 March 2012, and the full text is available on the Council’s website: http://www.rjp.pan.pl.

208

Chapter Twelve

female name is formed by changing the paradigm, as in, for example, fili-us, -i ‘son’ – fili-a, -ae ‘daughter,’ where the functional change is signalled by means of a declension class shift. In the pair minister – ministra, this relationship is not immediately evident on account of vowel alternations in the declension of the masculine form. There is no overt exponent in the nominative singular, i.e., there is a zero morph. The addition of a vowel-initial inflectional ending triggers a vowel-zero alternation in the stem. For example, the genitive form is ministri not *ministeri, which means that the lexical representation of the stem is ministør-. The process of female noun formation involves a change of the declension class, and the resulting form is ministra. Hence, the ending -a is a cumulative inflectional exponent of nominative singular feminine and not a derivational formative sensu stricte. The status of ministra as a potential borrowing must be considered against the backdrop of the general tendencies for borrowing in Polish. Linguistic contact presupposes a lack of balance between the dominating donor language and the recipient language. The excessive influx of barbarisms is generally viewed as a disturbing phenomenon. Despite the Greek etymology of the term (barbaros ‘foreigner’), the Polish language is not flooded with borrowings from Latin or Greek. The knowledge of Classical languages is on the wane, and, if Greek or Latin borrowings are accepted, they are confined to technical terminology. Polish, like many other languages, must now stem the tide of omnipresent anglicisms. The prevailing tendency is to borrow lexical items from English, which currently is the global lingua franca of business, politics and pop-culture. Borrowings undergo adaptation processes, i.e., they are fitted into existing declension and conjugation patterns and serve as bases for further derivation. Agent nouns are no exception to this general tendency. These borrowings are declined in accordance with the masculine declension and act as bases for the derivation of Female names. The affix performing this function is -ka, the default exponent of Nomina Feminativa (Grzegorczykowa, Laskowski, and Wróbel 1999, 422–25). Consider some examples which have become part and parcel of everyday vocabulary: haker – hakerka, trendseter – trendseterka, automaniak – automaniaczka, performer – performerka, biznesmen – biznesmenka, bloger – blogerka or topmodel – topmodelka. Given the tendency to borrow from English, the attempt to introduce a borrowing from Latin seems to be going against the general grain. Let us now consider the possibility of analysing the form ministra as a potential word. Female names in Polish can, in fact, be formed by means of paradigmatic derivation, with the inflectional ending -a in the

Onomasiological Strategies and Potential Forms

209

nominative singular. However, ministra does not fit into the pattern of paradigmatic derivation in Polish. The domain of paradigmatic derivation is restricted to masculine bases following the adjectival mode of declension and terminating in -ny (podwáadny ‘subordinate’ – podwáadna), -owy (ksiĊgowy ‘accountant’ – ksiĊgowa), -alski (spóĨnialski ‘latecomer’ – spóĨnialska) and -ący (przewodniczący ‘chairman’ – przewodnicząca). Other, less numerous nouns include nouns in -arz with an expressive tinge (spryciarz ‘cunning person, crafty bugger’ – spryciara), nouns in -och (táuĞcioch ‘fatso’ – táuĞciocha), nouns in -iĞ (strojniĞ ‘dandy’ – strojnisia ‘dressy woman’) and nouns in -nik (záoĞnik ‘crosspatch’ – záoĞnica ‘shrew’). Morphologically simplex nouns are virtually unattested. The only exceptions include kum ‘godfather’ – kuma ‘godmother’ and markiz ‘marquis’ – markiza ‘marquise’ (Grzegorczykowa, Laskowski, and Wróbel 1999, 424–25). The noun minister does not meet the criteria for paradigmatic derivation. However, we cannot rule out the possibility of the female name being formed with one of the remaining markers of Nomina Feminativa, i.e., -ica, -yni/ini or -ka (cf. Grzegorczykowa, Laskowski, and Wróbel 1999, 422–25).

3. Why not ministerka? The distribution of formal markers is hedged about by various constraints: phonological, morphological, syntactic, semantic and pragmatic (Plag 1999; Bauer 2001; Rainer 2005). The application of paradigmatic derivation is morphologically conditioned since it depends on the presence of specific morphemes in the base. The suffix -ica is productive in the function of deriving the names of animals (lwica ‘lioness,’ tygrysica ‘tigress,’ lisica ‘vixen,’ etc.). *Ministrzyca is unacceptable on account of semantic restrictions and, thus, cannot be regarded as a potential form. The derivatives with the suffix -ini/-yni can be regularly formed from agent nouns terminating in -ca (zwyciĊzca ‘winner’ – zwyciĊĪczyni). Simplex nouns can occasionally act as bases if they terminate in back consonants (czáonek ‘member’ – czáonkini, prorok ‘prophet’ – prorokini). For morpho-phonological reasons, the word *ministrzyni must be excluded from the set of potential words. The productivity of the suffix -ka is virtually unrestricted. Almost any personal masculine noun can give rise to the corresponding female derivative. Both simplex and complex nouns, together with both native and foreign nouns, accept -ka (sąsiad ‘neighbour’ – sąsiadka, roznosiciel

210

Chapter Twelve

‘delivery man’ – roznosicielka, Ğpiewak ‘singer’ – Ğpiewaczka, malarz ‘painter’ – malarka, chemik ‘chemist’ – chemiczka, delegat ‘delegate’ – delegatka, trener ‘coach’ – trenerka). That is why this suffix can be regarded as the elsewhere case. However, three classes of nouns are excluded from the domain of -ka. The first class includes nouns which are selected by other formal markers. Paradigmatic derivation and affixation with -ica and -yni/ini are ordered prior to -ka suffixation in accordance with the Paninian principle. The second class comprises personal masculine nouns which form the female equivalent by rendering the masculine form indeclinable. These include, e.g., minister ‘minister,’ redaktor ‘editor,’ dziekan ‘dean,’ ambasador ‘ambassador’ and nouns ending in -log (psycholog ‘psychologist’) and -a (ortopeda ‘orthopaedist’), which are the focus of our attention. The third group which does not accept the suffix -ka subsumes masculine names which lack female counterparts altogether, i.e., deadjectival personal nouns with the suffix -ec (gáupiec ‘fool’), various simplex nouns (áobuz ‘rogue’) and compound nouns (dusigrosz ‘miser’). Therefore, it is not entirely true, as the leading feminists would have it, that we are supposed to be women when we are secretaries, nurses, teachers, models or prostitutes, and when we are directors, ministers or executives we have to pretend to be men. Feminists bend over backwards to introduce the female equivalents of prestigious titles and high-ranking positions, but are completely oblivious to nouns mentioned in the third group. There are no proposals for female names for the fairly numerous class of nouns terminating in -ec (sprzymierzeniec ‘ally,’ zwyrodnialec ‘degenerate,’ pomyleniec ‘loony, nutter,’ naukowiec ‘scientist’). The reverse dictionary of Polish lists well over a hundred such nouns. There are no proposals for female names for certain simplex nouns and compounds (cieü ‘caretaker,’ alfons ‘pimp,’ leĔ ‘lazybones,’ krwiopijca ‘bloodsucker,’ golibroda ‘barber,’ pasibrzuch ‘pot-bellied man,’ áamistrajk ‘blackleg,’ záoczyĔca ‘malefactor’). We cannot exclude the possibility that, in the course of the diachronic development of the language, the range of applications of a given suffix will expand. In our case, this would boil down to including the second group in the domain of -ka suffixation. Change is the result of a complex interaction of system-internal and external factors. In what follows, we will discuss invariable forms in greater detail and argue why systeminternal factors win over ideological arguments and tip the scales in favour of the form (pani) minister. Reference will be made to some general principles of linguistic organization, such as the tendency towards economy of expression, iconicity and the markedness theory and the existence of a general analytic tendency in Polish (Nagórko 2010, 27–34).

Onomasiological Strategies and Potential Forms

211

4. Why pani minister? One of the key principles underlying the evolution of a linguistic system is the tendency towards economy, which, among others, manifests itself in cases of analogical levelling or the elimination of formal markers which express the same function. Elements which burden the memory and do not make communication more effective are done away with. For example, in conversational style speakers tend to replace the form of the vocative with the nominative case in phrases of the type Janek/Janku zrób to! ‘Do it John!.’ This is not the effect of imitating western European languages which lack vocative forms. Simply, the function of the vocative can be rendered by the appropriate intonation. The fact that the vocative form is retained in the more formal written style seems to corroborate this interpretation. The form pani ministra is dubious since it violates the principle of the least effort, i.e., it expresses the same function periphrastically with the aid of the honorific pronoun pani ‘Mrs, lady’ and morphologically by means of paradigmatic derivation. It has also been observed that the form in question is semantically ambiguous, since the form ministra could be interpreted as the genitive masculine singular of minister, i.e., pani ministra ‘the minister’s lady.’ Another universal property of linguistic structures is iconicity, i.e., the relationship between form and function based on similarity. Iconicity manifests itself in word structure, since words which are more complex semantically are richer in derivational morphemes. Iconicity can be observed in sentence structure since semantic relatedness is bound up with proximity, i.e., words which are semantically related occur closer together. The cases of the violation of this principle are blatantly conspicuous, which seems to be their raison d’être. By breaking the routine, we attract the attention of our interlocutor. Linguists refer to such structures as marked. Markedness can be observed in, for example, clipped forms characteristic of jargon as in impreza ‘party’ – impra, inverted word order, the use of a sophisticated synonym instead of the usual form (maáĪonek ‘husband’ instead of mąĪ ‘husband’). Unmarked structures are characterised by greater frequency. The contrast between marked and unmarked structures was brought into focus by American structuralism, though it originated in the Prague school of phonology and was initially applied to describing the relationship obtaining between certain sounds. For example, in the pair [t] – [d] the voiceless stop is the unmarked member of the pair since in the word final position the contrast between the two sounds is neutralised and the voiceless consonant can represent both, i.e., [kot] may represent kot ‘cat’ and kod ‘code.’ Unmarkedness is

Chapter Twelve

212

associated with ambivalence and a higher frequency of occurrence in texts. The contrast between marked and unmarked structures pervades the linguistic system and can also be observed in morphology. Certain masculine nouns in Polish become indeclinable in order to signal that the referent is female. In the sentence Premier spotkaá siĊ z Minister Sportu i Turystyki ‘The Prime Minister met the Minister of Sport and Tourism,’ the noun in the direct object position refers to a woman. The fact that female gender is signalled in this way does not stem from the speakers’ laziness or reluctance to decline, let alone a conscious attempt to make women linguistically invisible (cf. Karwatowska and SzpyraKozáowska 2005; Szpyra-Kozáowska 2006). Speakers simply decide to take advantage of the functional difference between a declinable and an indeclinable noun. If we compare their paradigms in the singular, as set out in Table 12-1 below, we will observe that the neutralisation of the masculine – feminine contrast obtains only in the nominative. In the remaining cases, there is a formal difference, as in Rozmawiaáem z ministrem ‘I talked to the minister (man)’ and Rozmawiaáem z minister ‘I talked to the minister (woman).’ Table 12-1. The declension of minister (Masculine) and minister (Feminine) in the singular Nominative Genitive Dative Accusative Instrumental Locative Vocative

Masculine minister ministr-a ministr-owi ministr-a ministr-em ministrz-e ministrz-e

Feminine minister minister minister minister minister minister minister

In the position where the contrast is morphologically neutralised it can be expressed by morphosyntactic means, either in NP Agreement on the verb (as in (1a) and (1b)) or in N Agreement on the preceding determiner or adjective (as in (1b) and (1c)).2

2

The following abbreviations have been used: NOM – nominative, GEN – genitive, DAT – dative, SG – singular, MASC – masculine, FEM – feminine, 3 – third person, DIM – diminutive.

Onomasiological Strategies and Potential Forms (1)

213

a. Minister- NOM.SG.MASC podpisaá-3SG.PAST.MASC rozporządzenie. ‘The minister (man) singed the directive.’ Minister- NOM.SG.FEM podpisaáa-3SG.PAST.FEM rozporządzenie. ‘The minister (woman) singed the directive.’ b. Nowa-NOM.SG.FEM minister-NOM.SG.FEM uznaáa-3SG.PAST.FEM, Īe plan jest niewykonalny. ‘The new minister (woman) judged the plan impossible.’ c. ten-SG.MASC inĪynier ‘this engineer (man)’ ta-SG.FEM inĪynier ‘this engineer (woman)’ mój-SG.MASC psycholog ‘my psychologist (man)’ moja-SG.FEM psycholog ‘my psychologist (woman)’

The functional difference manifests itself in cases where the noun refers to a female. It is covert if the noun minister is extracted from the pragmatic and syntactic context. The male counterpart is the unmarked member of the pair since it is used generically and predicatively, e.g., cechy dobrego ministra ‘the qualities of a good minister (man or woman),’ on/ona jest dobrym ministrem-DAT.SG.MASC ‘he/she is a good minister.’ Gender, as understood in linguistic terms, must be kept distinct from its biological meaning. The issue is further complicated by the fact that Gender is used with reference to Natural Gender and Grammatical Gender (Beard 1995, 103–10). The category of Natural Gender characterises nouns referring to sexed beings, so we expect a correlation between semantic features [Male] and [Female] and Natural Gender features [Masculine] and [Feminine], as in, for example, mĊĪczyzna ‘man, +Masculine,’ brat ‘brother, +Masculine’ and kobieta ‘woman, +Feminine,’ siostra ‘sister, +Feminine.’ Grammatical Gender, on the other hand, is another morpholexical category of nouns which refers to the arbitrary lexical subclasses of nouns which “provide the basis for Agreement and for assigning nouns to declensions in languages which possess declensions” (Beard 1995, 104). Thus, Grammatical Gender is also referred to as “Noun Class” (Corbett 1991, 43–49) or “Inflectional Class” (Aronoff 1994, 64–65) and is associated with different sets of declensional endings and different patterns of Agreement. Counter to popular intuition and a close relationship between the two, Natural and Grammatical Gender represent two discrete categories, i.e., the assignment to a particular Noun Class need not correlate with its Natural Gender specification (e.g., +Masculine nouns such as poeta ‘poet,’ idiota ‘idiot’ decline like kobieta ‘woman, +Feminine’) and nouns which do not have

214

Chapter Twelve

Natural Gender (i.e., inanimate nouns) are assigned to Noun Classes (e.g., banan ‘banana, Masculine Declension,’ gruszka ‘pear, Feminine Declension,’ jabáko ‘apple, Neuter Declension’). Since both Natural Gender and Noun Class are lexical categories, i.e., they may be arbitrarily fixed, there can be some mismatches where the semantic feature [+Female] does not correspond to [+Feminine] Natural Gender and Feminine Declension and the semantic feature [+Male] is not bi-uniquely mapped onto [+Masculine] and Masculine Declension, e.g., the noun babsztyl ‘pej. hag, virago’ refers to a woman but is assigned to the masculine declension, the noun ciota ‘pej. faggot’ refers to a man but is assigned to the feminine declension. If we assume the existence of two properties [Masculine] and [Feminine] realizing the morpholexical category of Natural Gender, marked either positively or negatively, the system generates four subclasses, as depicted in column one in Table 12-2 below (cf. Beard 1995, 154–58). According to Beard (1995) the presence of these features in the representation of the lexeme is indicative of the grammatical relevance of the corresponding semantic categories. Nouns with the [+Masculine, íFeminine] or [–Masculine, +Feminine] specification refer exclusively to males or exclusively to females, respectively. Lexemes specified as [íMasculine, –Feminine] refer to “an object whose Gender is grammatically relevant but whose reference is not sex-differentiated” (Beard 1995, 157). Most importantly, the grammatical system also allows for the existence of nouns which can be positively marked for both features and can be used with reference to both sexes. This state of affairs parallels the existence of ambiaspectual verbs in Polish such as awansowaü ‘be promoted’ (Laskowski 1978, 25; Perlin 2010; BlochTrojnar 2013, 232) and collective nouns such as committee, team or government in English, as in, e.g., The committee is/are old. (cf. Corbett 2000, 188–90; Bloch-Trojnar 2012).

Onomasiological Strategies and Potential Forms

215

Table 12-2. The relationship between Natural Gender and Grammatical Gender (Noun Class) in animate nouns in Polish Natural Gender

Grammatical Gender (Noun Class)

[+Masculine, –Feminine]

Masculine declension

[–Masculine, +Feminine]

Feminine declension

[–Masculine, –Feminine]

Neuter declension

[+Masculine, +Feminine]

Masculine declension (unmarked) (Feminine declension – by exception)

Examples mĊĪczyzna ‘man,’ ojciec ‘father,’ ogier ‘stallion’ kobieta ‘woman,’ matka ‘mother,’ klacz ‘mare’ dziecko ‘child,’ ĨrebiĊ ‘colt’ minister ‘minister,’ premier ‘Prime Minister,’ aktor ‘actor,’ nauczyciel ‘teacher’ sierota ‘orphan,’ gĊĞ ‘goose’

This marking allows the derivational rules to differentiate those masculine nouns which can undergo the feminisation rule from those which cannot. Ambigeneric nouns can act as bases for the derivation of Nomina Feminativa, which can be effected with the aid of an overt affix (e.g., nauczyciel [+Masculine, +Feminine, Noun Class: Masculine Declension] ‘teacher’ → nauczycielka [–Masculine, +Feminine, Noun Class: Feminine Declension] ‘teacher (female)’) or by making the input form indeclinable (minister [+Masculine, +Feminine, Noun Class: Masculine Declension] ‘minister’→ minister [–Masculine, +Feminine, 0Noun Class] ‘teacher (female)’). The masculine form is the unmarked member and it is the default form. It is used where the category itself is important but not the actual functional distinction, e.g., nauczyciel can refer to any sexed being who teaches. This would explain why in cases where reference is not made to a particular person but the noun is used generically, masculine forms are also used to describe women. (2)

a. Ona jest kanclerzem-DAT.SG.MASC/ doktorem-DAT.SG.MASC./ redaktorem-DAT.SG.MASC. ‘She is a chancellor/ doctor/ editor.’

216

Chapter Twelve b. Ona jest Ğwietnym-DAT.SG.MASC nauczycielem-DAT.SG.MASC. Ona jest Ğwietną-DAT.SG.FEM nauczycielką-DAT.SG.FEM. ‘She is an excellent teacher.’

Interestingly, in exceptional cases, some personal nouns grammatically classified as feminine can be used with reference to both women and men (sierota ‘orphan,’ skarĪypyta ‘telltale,’ fajtáapa ‘butterfingers,’ osoba ‘person’), as exemplified in (3) below. (3)

a. On/Ona jest sierotą-DAT.SG.FEM. ‘He/She is an orphan.’ b. Straszna-NOM.SG.FEM z niego/niej skarĪypyta-NOM.SG.FEM. ‘He/She is such a telltale.’ c. Straszny-NOM.SG.MASC z niego skarĪypyta-NOM.SG.FEM. ‘He is such a telltale.’

We might hypothesize that ambigeneric nouns are sporadically assigned to the Feminine Declension. Some of them can also serve as bases for the derivation of masculine counterparts, which can be achieved without any formal marking (ta skarĪypyta [+Masculine, +Feminine, Noun Class: Feminine Declension] ‘telltale (female)’ → ten skarĪypyta [+Masculine, íFeminine, Noun Class: Feminine declension]). Note, that the noun in (3c) above declines like kobieta ‘woman, Feminine Declension.’ In cases of conflict, Natural Gender overrides Grammatical Noun Class in Agreement (cf. Beard 1995, 158). Consider the example in (4a) below, where Agreement on the demonstrative and adjective is masculine, whereas the noun declines in accordance with the Feminine Declension. Compare it with (4b). (4)

a. Nie mogĊ znieĞü tego-GEN.SG.MASC strasznego-GEN.SG.MASC skarĪypyty-GEN.SG.FEM. ‘I cannot stand this terrible telltale (man).’ b. Nie mogĊ znieĞü tej-GEN.SG.FEM strasznej-GEN.SG.FEM kobiety-GEN.SG.FEM. ‘I cannot stand this terrible woman.’

The process of turning an ambigeneric noun belonging to the Feminine Declension into a masculine counterpart can exceptionally be effected with the help of affixes. For example, gĊĞ ‘goose’ can be used generically with reference to both sexes and can also give rise to gĊsior ‘gander’

Onomasiological Strategies and Potential Forms

217

(Grzegorczykowa, Laskowski, and Wróbel 1999, 425). Here, in addition to a Natural Gender shift, there is also a change in the Declension Class. These and other similar cases confirm that the mapping from Natural Gender to the Noun/Declension Class need not be one-to-one, and so the two categories must be kept distinct. Having considered markedness relationships, we will present the form pani minister ‘Mrs minister’ against the background of a general tendency to develop analytic structures in Polish (Buttler 1968, 1976; Nagórko 2010, 26). For example, syntactic cases are being replaced with prepositional expressions, e.g., czekaü na list ‘wait for a letter’ is used instead of czekaü listu-GEN ‘await the letter.’ Complex predicates are frequently used instead of simplex verbs, e.g., doznaü oparzeĔ ‘suffer burn’ instead of the simplex reflexive verb oparzyü siĊ ‘get burnt.’ The spread of complex expressions such as woda mineralna ‘mineral water,’ Ğrodki czystoĞci ‘detergents’ can be observed. Periphrastic forms of adjectives gradually supersede forms inflected for the comparative and superlative degree. Analytic constructions take on a certain intellectual aura and are predominantly used in formal and academic contexts. Most importantly, in contradistinction to synthetic forms, they are characterised by semantic precision and clarity. By using analytic forms pan minister and pani minister, not only do we unequivocally express the gender of the referents, but we also convey respect for and the prestige of the positions in question. There is an extra honorific tinge resulting from the use of honorific pronouns. By contrast, suffixed forms are neither so precise nor unambiguous. The suffix -ka is highly polysemous, which may hamper the interpretation of the nouns in (5): (5)

a. gangsterka ‘gangsterism, woman gangster,’ kaskaderka ‘stunt work, female stunt performer,’ fuszerka ‘slapdash work, female botcher’ (names of activities or of women?) b. marynarka ‘jacket, woman sailor,’ pilotka ‘flying cap, female pilot,’ kominiarka ‘balaclava, female chimney sweeper’ (garments or women?) c. drukarka ‘printer,’ kosiarka ‘mower’ (instruments or women?)

218

Chapter Twelve

The suffix -ka (together with its masculine and neuter variants -ek/-ik, -ko) is used to form diminutives, i.e., forms denoting entities of small size (e.g., domek ‘house-DIM,’ bucik ‘shoe-DIM,’ Ĩródeáko ‘spring-DIM’). Diminutives are perceived as informal and are frequently emotionally coloured. The expressive value may be positive (maápka ‘a (nice) little monkey,’ kanapeczka ‘sandwich-DIM, canapé,’ mini-ratka ‘mini-creditrate’) or negative/derogatory (karierka ‘career-DIM,’ gierka ‘game-DIM,’ posadka ‘position-DIM’). In sum, the analytic form is more prestigious and there are no logical arguments to enforce the use of synthetic forms in -ka. On the one hand, feminist circles exert pressure to form female names, which has the undesirable effect of producing informal, jocular, patronizing or depreciative connotations in connection with nouns such as ekoloĪka ‘female ecologist,’ psycholoĪka ‘female psychologist,’ filozofka ‘female philosopher,’ polityczka ‘female politician,’ ministerka ‘female minister.’ On the other hand, there is a widely accepted linguistic norm which says that female gender is signalled by suppressing declension and the optional use of the honorific pronoun. Notably, the noun itself (minister, premier) can refer to either a man or a woman. Reference is made clear in the pragmatic context of the sentence. According to the Corpus of the Polish Language (Przepiórkowski et al. 2012), the form pani minister is used in 903 sentences. Neither the form (pani) ministra nor ministerka are attested. By comparison, the form posáanka ‘female MP,’ which has received a lot of publicity and exposure, is used in 256 sentences. The form pani posáanka appears in 115 sentences. The usual form pani poseá is three times more frequent than the two forms taken together, i.e., it is attested in 1000 sentences. Thus, language users seem willing to maintain the status quo.

5. Summary Ideological arguments must give way to universal linguistic tendencies. These include the tendency towards an economy of expression and the difference between the marked and unmarked pair member. The suppression of inflection is just as effective an exponent of a linguistic contrast as is the use of an overt derivational marker. The form pani minister is in line with the general analytic tendency existing in Polish and the use of an honorific pronoun produces positive connotations relating to prestige and respect. If we assume that the domain of the most productive affix -ka has expanded and the affix has gained in productivity, we may

Onomasiological Strategies and Potential Forms

219

consider the form ministerka a potential word. However, it stands a poor chance of being institutionalised due to the polysemy of the affix and its emotional colouring. The word ministra is not a potential word, since its base (minister) does not meet the conditions for paradigmatic derivation. Its status as a potential derivative from Latin, given the prevailing influence of English on Polish, is also very hard to envisage.

References Allen, Margaret. 1978. “Morphological Investigations.” Unpublished PhD diss., University of Connecticut. Aronoff, Mark. 1994. Morphology by Itself. Stems and Inflectional Classes. Cambridge, MA: MIT Press. Bauer, Laurie. 2001. Morphological Productivity. Cambridge: CUP. Beard, Robert. 1995. Lexeme Morpheme Base Morphology. Albany, NY: SUNY Press. Bloch-Trojnar, Maria. 2012. “The Category of Nominal Number in English and the Inflection Derivation Distinction.” In Sound, Structure and Sense. Studies in Memory of Edmund Gussmann, edited by Eugeniusz Cyran, Bodgan Szymanek, and Henryk Kardela, 29–53. Lublin: Wydawnictwo KUL. —. 2013. The Mechanics of Transposition. A Study of Action Nominalisations in English, Irish and Polish. Lublin: Wydawnictwo KUL. Buttler, Danuta. 1968. “Poáączenia Typu ulec zniszczeniu w JĊzyku Polskim.” Poradnik JĊzykowy 7:349–59. —. 1976. Innowacje Skáadniowe Wspóáczesnej Polszczyzny. Warszawa: Wydawnictwo Naukowe PWN. Corbett, Greville. 1991. Gender. Cambridge: CUP. —. 2000. Number. Cambridge: CUP. Dubisz, Stanisáaw, ed. 2003. Uniwersalny Sáownik JĊzyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN. Grzegorczykowa, Renata, Roman Laskowski, and Henryk Wróbel, eds. 1999. Gramatyka Wspóáczesnego JĊzyka Polskiego. Vol. 2: Morfologia. Warszawa: PWN. Karwatowska, Maágorzata, and Jolanta Szpyra-Kozáowska. 2005. Lingwistyka Páci. On i Ona w JĊzyku Polskim. Lublin: Wydawnictwo UMCS.

220

Chapter Twelve

Laskowski, Roman. 1978. “Aspekt.” In Encyklopedia Wiedzy o JĊzyku Polskim, edited by Stanisáaw UrbaĔczyk, 25–26. Wrocáaw: Ossolineum. Malicka-Kleparska, Anna. 1985. The Conditional Lexicon in Derivational Morphology. A Study of Double Motivation in Polish and English. Lublin: Wydawnictwo KUL. —. 1987. “Potential Forms and Lexicons.” In Rules and the Lexicon. Studies in Word Formation, edited by Edmund Gussmann, 103–20. Lublin: Redakcja Wydawnictw KUL. Nagórko, Alicja. 2010. PodrĊczna Gramatyka JĊzyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN. Perlin, Jacek. 2010. “Ile Jest we Wspóáczesnej PolszczyĨnie Czasowników Dwuaspektowych?” Linguistica Copernicana 1(3):165–71. Plag, Ingo. 1999. Morphological Productivity. Structural Constraints in English Derivation. Berlin and New York: Mouton de Gruyter. Przepiórkowski, Adam, Mirosáaw BaĔko, Rafaá Górski, and Barbara Lewandowska-Tomaszczyk, eds. 2012. Narodowy Korpus JĊzyka Polskiego. Warszawa: Wydawnictwo Naukowe PWN. Rainer, Franz. 2005. “Constraints on Productivity.” In Handbook of WordFormation, edited by Pavol Štekauer and Rochelle Lieber, 335–52. Dordrecht: Springer. Smóákowa, Teresa, ed. 2004. Nowe Sáownictwo Polskie. Materiaáy z Prasy Lat 1993-2000. Kraków: Wydawnictwo LEXIS. Szpyra-Kozáowska, Jolanta. 2006. “Linguistic Invisibility of Women in Polish.” In Culture, Literature, Language, edited by Ryszard Wolny, 191–207. Kielce: Wszechnica ĝwiĊtokrzyska.

PART II: STUDIES IN PHONETICS AND PHONOLOGY

CHAPTER THIRTEEN SCHWA: PHONOLOGICAL CODING VERSUS PHONETIC MANIFESTATION ANNA BLOCH-ROZMEJ

1. Introduction Beyond doubt, whatever phonetic output people produce, their manifestations are delimited by an intricate system of phonological coding driving language production. Numerous studies into language typology reveal that the same phonetic effect can result from different phonological structures, as well as that different language-specific realizations can arise from lexically identical codes. Individual systems encompass both universal principles and laws typical of unique languages. Their interplay underlies the richness of typological patterns. In this chapter, we intend to focus on the interesting behaviour of the mid central vowel schwa in English and French, describing both its phonological representations in the two systems and their effects on the distributional properties of the segment and its licensing potential. The discussion will be couched within the non-linear framework of Government Phonology (Kaye, Lowenstamm, and Vergnaud 1985, 1990; Harris 1994; Gussmann 2007; Cyran 2010; Bloch-Rozmej 2008). The chapter is structured as follows. First, our attention will be focused on the distribution and representation of schwa in English. In particular, attested vowel-zero alternations will be analysed as well as vowel reduction phenomena in which schwa readily participates. It will be postulated that English possesses two kinds of schwa: one that is melodically empty and one that should be represented by means of an active phonological prime. The two types of schwa exhibit different behaviour with respect to phonological processing. Further, they are endowed with differing licensing capacities, which has a serious bearing on the types of immediately adjacent consonant clusters. The discussion will reveal the

Chapter Thirteen

224

importance of segment headlessness in both its phonetic manifestation and phonological behavior. Our findings concerning the nature of the English schwa will be confronted with the French-specific properties of the mid central vowel.

2. Schwa in various positions within phonological domains: structures, manifestations, constraints Gimson (1980, 125) maintains that schwa is one of the most common vowels in English. Schwa can be found in various word positions, as in the words possible [ScVEO], woman [Z8PQ], oblige [EOD,G=], circular [VÆhNM8O], precious [SUH6V], honour [cQ], etc. Thus, it is permitted in the word-initial position and in the word-initial syllable before and after a single consonant or a consonantal cluster. Its presence is attested in wordmedial and word-final contexts as well. A brief overview of the potentially available schwa-containing sites will be presented below, along with a phonological interpretation of the phonetic facts. Our analysis will crucially depend on the notions of proper government and government licensing.1 We shall also employ the three-way division of element status proposed in Bloch-Rozmej (2008). More specifically, primes can enjoy the status of being the segment’s head, an operator, or can remain mere dependents, all three options having a serious impact on the phonological behaviour of a given segment. The English schwa is very often found in an unstressed word-initial syllable: (1)

1

abandon ballon galore macaw

[E±QGQ] [EOXhQ] [JO2h] [PN2h]

We abstain from elaborating on these concepts here. For definitions and examples of the operation of these mechanisms, see Kaye, Lowenstamm, and Vergnaud (1990), Charette (1991), or Harris (1994). Proper government suppresses empty nuclei within phonological domains, whereas government-licensing authorizes the head of the preceding governing domain to discharge its governing responsibility with respect to the potential governee. Only nuclei can be government-licensers.

Schwa: Phonological Coding versus Phonetic Manifestation

225

As the examples in (1) show, this mid-central vowel can occur wordinitially or in a word-initial syllable. It is always followed by a consonant or a consonantal cluster. It does not immediately precede a vowel in any word (Oostendorp 2000, 4) (unless we are dealing with a diphthongal expression). Such a vocalic sequence would have to be separated by an empty onset point. Since sequences of empty positions are disfavoured, we can infer that schwa is, in some way, phonologically similar to an empty slot. Additionally, it can be observed that in the word-final context, the schwa-initial affixes, i.e., >UQ], are never attached to stems which end in a vowel. Thus, it can be concluded that English tries to avoid ‫ۑ‬+V or V+‫ ۑ‬clusters. Given this, we posit that schwa is a melodic expression deprived of a head element.2 Interestingly, in English, a word-initial empty onset is allowed to be licensed by a word-initial empty-headed nuclear expression, i.e., schwa. Hence the availability of such items as ago, enough or allow. It is noteworthy that, within the present model of phonological structure, word domains start with onsets which, in the case of vowel-initial words, are melodically empty. Thus, an onset dominates an empty skeletal slot. Each onset has to be universally licensed by the following nucleus (Central Linearization Parameter). It seems that schwa in English possesses the same onset-licensing potential as stressed vowels (e.g., in answer, empty, eager). Moreover, schwa in English can be manifested phonetically due to proper government failure.3 A lexically empty nucleus acquires melodic content when it fails to be properly governed. According to the Empty Category Principle (Kaye 1990), empty nuclei can acquire phonetic content and very often they are manifested as schwa. Thus, the words tomorrow [tmcU8] and supply [VSOD,] will have the following representations:

2

Government Phonology assumes that segments are composed of elements of which one can enjoy head status, whereas the remaining primes are dependents. 3 Proper government is a special form of government which suppresses empty nuclear positions. It fails to apply when the potential governing nucleus is itself empty or there is an intervening governing domain between the potential governor and governee. Ungoverned empty nuclei acquire phonetic content.

Chapter Thirteen

226 (2) a.

O1

x | t b.

O1

x | s

R | N1 | x | 

O2

R | N1 | x | 

O2

x | m

x | p

R | N2 | x | c

x | O

O3

x | r

R | N3 | x | W

R | N2 | x | a

x | ,

x | 8

In (2a), even though it is possible for the nucleus (N1) to remain empty (due to the following proper governor), it acquires phonetic content. It seems that the melody of schwa residing in N1 is simply lexical. However, in (2b), schwa can be the result of proper government failure because its proper governor, i.e., the nucleus (N2), has to government-license the preceding branching onset [pl]. It is noteworthy that nuclei have to government-license the head of the preceding governing relation (in our case O2) for government to come into effect. Additionally, the occurrence of a governing domain between these two nuclei blocks the application of proper government. In consequence, schwa must be manifested. Another interesting property of schwa is its inability to appear in stressed positions. It appears to be invisible to stress. In English, the word stress is variable but, as far as noun stress in concerned, it falls on the rightmost heavy syllable. If the syllable is light, stress is shifted leftwards (Harris 1994). The syllable containing schwa is skipped during stress assignment. Consequently, the pre-final or antepenultimate syllable is stressed. This property sets schwa apart from the other short vowels in English. All of them, including schwa, are lax, which is represented in the form of headlessness. Yet, taking into consideration the behaviour of schwa with respect to stress placement, there must be some additional factor that distinguishes schwa from the other short vocalic expressions. Bloch-Rozmej (2008) proposes that this characteristic of schwa is due to its representation as a headless melody with a sole element A with the function of a mere dependent. The dependent status of the element A in a

Schwa: Phonological Coding versus Phonetic Manifestation

227

non-headed expression makes schwa phonologically weak, only slightly stronger than an empty skeletal slot. We can thus put forward the view that the system of English will not tolerate such weak segments in the positions of stress. It is also possible to encounter examples in which a word-initial schwa is followed by sC(C) sequences, e.g., escape [VNHLS] and especially [VSH6OL] (Cyran 2003, 293). According to Cyran (2003, 293), such initial contexts require phonetically manifested vowels to follow. The direct licenser of the coda-onset string (i.e., sC) is the following nucleus, e.g., the vowel [e], while the licensing of the preceding schwa is additional or subsidiary. This fact provides additional support for the assumed phonological weakness of schwa in English, which, recall, has been attributed to its representation as a headless segment containing A as a dependent. Being a member of a word-initial melody sequence, schwa can also follow consonantal clusters. This distributional property of the mid central vowel is demonstrated in the examples in (3). (3)

blancmange cravat degradation probatio

[EOPcQ=@ [NUY±W@ [GHJUGH,6Q@ [SUEH,6Q@

As can be observed in (3), schwa shows up after branching onsets. This leads to the conclusion that it can be a government-licenser. The word brazil [br],O] illustrates this situation: (4)

O1

x | b

x | U

R | N1 | x | 

O2

x | z

R | N2 | x | L

O3

x | l

R | N3 | x

Government

In (4), the nucleus N1 government-licenses the head [b] of the preceding governing domain to govern its complement [r]. Summarizing the discussion so far, it has to be underlined that the English schwa should be represented as a single-element headless segment. The sole prime enjoys the status of a dependent. Such a structure

Chapter Thirteen

228

weakens the licensing potential of the melody. Schwa is weaker than the other lax vowels of English. Yet, it is still strong enough to license the preceding governing domains and performs a subsidiary role in licensing the clusters that appear to its right.

3. Schwa and vowel reduction In English, schwa is frequently the result of vowel reduction in unstressed positions. A word-initial schwa alternates with the [¡], [c], [H] and [±] vowels. The relevant examples are presented below: (5)

autumnal phonology telepathic valid

[2hW¡PQO] [IQcOG=L] [WHOS±7LN] [Y±O,G]

autumn phonological telepathy validity

[2hWP] [IcQOcG=LNO] [WOHS7L] [YO,GWL]

As can be seen, full vowels occur in positions which receive primary or secondary stress. However, schwa shows up only in unstressed positions. According to Harris (1994, 109), a schwa-like quality is omnipresent in all vocalic expressions and it is manifested when awarded the head position.4 This means that it does not appear from outside the representation but is present in the phonological structure. Nevertheless, when it occupies a dependent position, it remains silent. The process of weakening, represented in (6), results in the delinking of the elements of the vowels in question. In consequence, the latently present schwa is promoted to the head position. Our analysis introduces an additional modification to Harris’s proposal by resigning from the latently present neutral element and its headed status in the structure of lax vowels. We choose to adopt a three-way division of the element status and represent the lax vowels as empty-headed.5 This can be depicted as follows:

4

Harris proposes that the neutral element present in the melodic structure of the schwa is latently present in all segments but remains mute unless promoted to the head status. 5 Gussmann (2002, 127) postulates that the above words should not be considered as being derived from each other. They have their own lexical representations. This means that schwa possesses its own phonological structure.

Schwa: Phonological Coding versus Phonetic Manifestation (6)

[¡] > []

x | _ | A

x | _ | A

[c] > []

x | U | A

[e] > []

X | _ | A

x | I | A

229

[±] > []

x | _ | A

x | I | A

x | _ | A

It is noteworthy that, in the majority of GP-based analyses, the neutral element has been abandoned. Importantly, elements present in the melodic representation acquire a phonetic manifestation if they are autosegmentally licensed by a given skeletal position. If the neutral element was latently present in the representation, it would be refused such a licence in all expressions in which it would have a status other than the head. The reason for such a refusal remains unknown. Many GP-bound researchers replace the cold vowel/neutral element with the notion of segmental headlessness, as opposed to headedness. Consequently, the elemental headship specifies the phonological strength of a segment. Headed expressions are strong, while headless ones are phonologically weak. According to Cyran (2003, 18), schwa is a representation of the neutral state of articulators and evenly spaced-out formants. In Government Phonology, it may be viewed as the realisation of a neutral element or nothing. Additionally, it is stated that schwa alternating with zero may be composed of no elements, i.e., it is a phonetically interpreted nuclear position without melodic content. Then, the difference between schwa and an empty nucleus is that the former receives a phonetic interpretation and the latter is silent. The structure of the lexical schwa, however, because of its different behaviour and non-alternating character (i.e., the absence of vowel-zero alternations), should be represented as a headless expression with A in the dependent status.

4. The word-medial context: More on the lexical schwa In English, the word-medial schwa can be found between two consonants, as in the examples in (7). (7)

acolyte catapult histamine labyrinth

[±NOD,W] [N±WS¡OW] [K,VWPLhQ] [O±EU,Q7]

Chapter Thirteen

230

In none of these words is schwa an outcome of reduction or proper government failure. This may suggest that it is a lexical schwa – an independent segment with its own unique phonological structure. Importantly, a lexical schwa will not alternate with zero, a situation we find in the case of the word family for instance, where speakers can optionally omit the vowel between [m] and [l]. A word-medial lexical schwa, when preceded by a consonantal cluster, must be interpreted. The relevant data are presented below: (8)

alchemy balcony deprecate vitrify

[±ONPL] [E±ONQL] [GHSUNH,W] [Y,WUID,]

In all the above-mentioned examples, schwa operates as a governmentlicenser of a preceding governing domain. Notice that the consonantal cluster preceding it constitutes a head-final rhymal complement-onset domain (traditional coda-onset). This type of governing relation requires direct licensing coming from the following nuclear position. In English, it seems, this position may not be empty. Thus, it turns out that the lexical schwa represented as a headless mono-elemental segment can serve as a direct government-licenser. It is noteworthy that schwa in English is disallowed in *>PS] and *>1N] clusters. Such a restriction can result either from its internal makeup or a specific parameter imposed by the language. Szigetvári (1994, 193) claims that the clusters [mp] and [1N] occur in English only in monosyllables, after stressed vowels. Bloch-Rozmej (2008, 184) explains that the absence of schwa in these sequences is the result of its inability to transfer the licensing potential onto the clusters in question. Interestingly, schwa may occur before a sonorant followed by a coronal stop or sC(C) clusters. The examples below illustrate this situation: (9)

decadent employment forest

[GHkGQW] [,PSO2,PQW] [f2UVW]

These examples show that it is possible for schwa to license the coronal nasal [n] but it is too weak to license the following bilabial nasal [m] or the velar nasal [1]. It does not seem to provide a sufficient explanation to propose that schwa, as an empty-headed segment, is unable to license [m] and [1] in [PS] and [1N] clusters. The other short headless vowels occur in such positions and do license these clusters, e.g., lamp [O±PS], hunk

Schwa: Phonological Coding versus Phonetic Manifestation

231

[K¡1N]. Cyran (2003, 281) notes that these clusters can be preceded by schwa but only if they are not word-final, e.g., companion [NPS±Q,Q], combine [NPEDLQ]. In addition, [s] followed by a non-coronal stop can also be found in mono-syllabic words, e.g., desk [desk], wisp [w,sp] and, generally, schwa in sequences like *>sp] or *>sk] is not attested. According to Cyran (2003, 284), the restrictions on schwa preceding the coda-onset sequences can be explained as follows. The governing relation between the head of the rhyme and the complement decides which type of coda can be governed by schwa and determines where a full vowel is required. In English, schwa can govern the coda only if it is already strongly governed by the following onset. Such a consonantal string is an easy-to-license cluster. It is a cluster in which the governing relation is stronger because of melodic (homorganicity) or segmental complexity reasons. As a result, the cluster is more integral and easier to license (Cyran 2003, 284). Recall that Bloch-Rozmej (2008, 185) proposes a three-way division of the element status within melodic expressions. The elements which form a segment can be a head (in a headed segment), an operator (in a headless segment), or a dependent (also in a headless segment). Their representations are depicted in (10). (10)

x

A

x | _ | A

x | _ | A

The operator and the dependent are subordinate within melodic structures and the distinction is shown in (10). Some linguistic systems endow their nuclei with an uneven autosegmental licensing potential. Consequently, the operator and the dependent are provided with different licensing capacities, the former being phonologically stronger. If the schwa in [PS] and [1N] clusters were a dependent, it would be too weak to act both as the head of the branching rhyme and the licenser of the preceding onset segment. Schwa as a dependent would be too weak since it is non-headed on the one hand, and has a single prime in the dependent status on the other. Due to its lone elemental resident, its status is the weakest among other elements. This means that the number of melodies it can sustain within the branching rhyme is restricted. This leads to the Branching Rhyme Constraint, according to which a complex rhyme structure requires a stronger nuclear head (Bloch-Rozmej 2008, 186). A mere dependent

Chapter Thirteen

232

may not be the head of a branching rhyme, unless it is given interelemental support. Put differently, the head of a branching rhyme should be attached to a headed element or become strengthened in some other way. Bloch-Rozmej (2008, 186) proposes a bridging relation between the primes of the segments involved in a licensing relation within a branching rhyme. (11)

O

R | N | x | A

x

x | A

licensing

The structure in (11) depicts the operation of the inter-element bridge within which the element A in the melody of the coronal supports the prime of a branching rhyme. Labials and velars do not strengthen the dependent A since they do not possess the A element in their melodic make-up. However, when the head of a branching rhyme becomes stronger, i.e., when it contains the operator prime, it is possible to find V+m/1C structures, as in (12): (12)

sink sunk sump

[sL1N] [V¡1N] [V¡PS]

It can be observed that A as an operator is a stronger licenser than A as a dependent. In effect, it can license the following bilabial nasal [m]. Positions occupied by one dependent, even though they receive melodic content, can behave as phonologically empty. When a segment is emptyheaded, it can be treated as an empty position.

5. The licensing potential of the word-final schwa English allows word-final consonant sequences. There are two types of rhymal complement-onset clusters which are allowed word-finally: homorganic nasal + stop and sonorant + dental clusters:

Schwa: Phonological Coding versus Phonetic Manifestation (13)

damp belt dank land

233

[G±PS] [EHOW] [G±1N] [O±QG]

In English, schwa is able to occur in open syllables and in the word-final positions. Moreover, it can be preceded by coda-onset sequences, which manifests schwa’s government licensing potential. In other words, the vowel can perform the role of the direct government licenser for preceding inter-consonantal governing domains. Consider some illustrative examples in (14). (14)

agenda bartender calendar canker

[G=HQG] [E$hWHQG] [N±OQG] [N±1N]

Clearly, the clusters involved are homorganic, which reveals their integrity. There are no attested examples of the epenthetic schwa in such clusters. Booij (1995, 127–28) maintains that the homorganic consonants which form the rhymal complement-onset domains must be adjacent on the skeletal tier. In other words, the lack of an empty position between consonants excludes the epenthetic schwa. Furthermore, schwa can only precede a particular kind of final rhymal complement-onset sequence. The data illustrating its behaviour are provided below: (15)

achievement coincident emigrant

[W6LhYPQW] [k8,QV,GQW] [HP,JUQW]

As can be observed, the rightmost schwa can be the head of a syllable that is more complex than a CV type. Thus, schwa can be both preceded by a branching onset and followed by a rhymal complement-onset sequence. It cannot head branching rhymes whose rhymal complement is occupied by a non-dental segment. Clearly, the cluster composed of a pair of coronals is the easiest for a schwa to license. Cyran (2003, 117) argues that schwa in English licenses the same structures as full vowels, i.e., coda-onset + equals to coda-onset + a full vowel and branching onset + equals to branching onset + a full vowel, while schwa in Dutch, for instance, licenses the same structures as an empty nucleus. However, if rhymal complement-onset strings formed of non-dental segments are

234

Chapter Thirteen

followed by an empty nucleus, schwa insertion is obligatory. It is possible to find examples in which non-homorganic clusters are represented as two separate onsets split by a nucleus. It is noteworthy that schwa can be both a direct and indirect government-licenser since it can be preceded by both branching onsets and rhymal complement-onset sequences, as in cobra, zebra, agenda or Miranda. In this respect, it enjoys the same phonological strength as other full vowels and emerges as a stronger licenser than a final empty nucleus. Notice the absence of branching onset clusters in the absolute word-final position. In English, a final empty nucleus cannot government-license indirectly.

6. Schwa in French: distribution and phonological potential The French vowel schwa is also called ‘e muet’ (‘silent e’) or ‘unstable e.’ It is the phonetic interpretation of an ungoverned underlying empty nucleus. The distribution of schwa is very restricted. At the leftmost edge of the word, this vowel is permitted only after a consonantal cluster and is deleted if it follows a vowel. It is also possible for schwa to be found word-medially. However, it is ruled out from a word-final position. Importantly, schwa in French is treated differently from other vowels due to its alternations with zero. Phonetically, it is similar to the vowel [^] but their two different realizations and properties suggest that they differ phonologically. According to Anderson (1982, 537), the vowel [^] can be found in stressed syllables, as in the word neuf [n^f] ‘nine,’ while the vowel [] is never found there. Stress in French words falls on the final syllable, but if a schwa occurs in this position, stress is shifted to the penultimate syllable. An exception is the exclamatory que in expressions, e.g., que c’est joli! [NV=2OL@ ‘isn’t it beautiful!’, also in the clitic le in the imperative Prends-le! [SU$ GUO] ‘take it!’ or when emphatically stressed, e.g., in the le grand chef [OJU$ 6(I] ‘the big chief.’ The main difference between both vowels lies in the alternations. The so-called stable [^] does not alternate with any other vowel. This can be illustrated by the word seulette [s^O(t] ‘lonely’ (fem.) in which the vowel [^] is present in unstressed environments, e.g., la seulette [las^O(t] ‘the lonely one’ (fem.) or cette seulette [s(WV^O(t] ‘this lonely one’ (fem.). By contrast, the unstable schwa alternates with zero or with the vowel [(]. In the word pelouse [pluz] ‘lawn,’ schwa occurs between the obstruent [p] and the liquid [l]. Even though it is manifested in the form cette pelouse [s(WSluz]

Schwa: Phonological Coding versus Phonetic Manifestation

235

‘this lawn,’ in la pelouse [lapluz] ‘the lawn,’ it is realised as zero. Let us now look more closely at the occurrence of schwa in different word contexts.

6.1. The left edge of the word In Parisian French, in bisyllabic and polysyllabic words, a schwa in an initial syllable preceded by a single consonant or a branching onset receives phonetic content. This can be demonstrated by the following items (Charette 1988, 82):6 (16)

crevasse frelon premier semestriel

[NUYDV] [frO2 ] [SUPMH] [VPHVWULO]

‘fissure’ ‘hornet’ ‘first’ ‘semestral’

A word-initial syllable is treated in a special way in French. This position is inaccessible to proper government due to the fact that an accented wordfinal nucleus cannot be a proper governor for the syllable-initial empty nucleus. In this case, proper government is not applied and the nucleus must be manifested. However, when a bisyllabic word is preceded by an article, then the syllable is not perceived as initial and proper government can apply as in la chemise [la6PL]] ‘the shirt.’ What is more, schwa never occurs within a branching rhyme. This happens because, within a constituent, a complement must be licensed by a head with phonetic content. When the headless representation of schwa is proposed, the absence of licensing in such cases becomes obvious.

6.2. The word-medial schwa In French, schwa in the word-medial position remains mute if it is followed by a phonetically realized nucleus which is its proper governor. 6

In Québécois French, schwa in bisyllabic words is not realised. However, in words composed of more than two syllables it receives phonetic content. This happens because, in Québécois French, an empty nucleus in a syllable-initial position can be properly governed if its proper governor is the head of the foot. Otherwise, such a nucleus is inaccessible to proper government and hence manifested phonetically.

Chapter Thirteen

236

What is more, when the vowel [] is preceded by a consonant, it does not receive phonetic content. This can be illustrated by the words below (Charette 1991, 29): (17)

brèvement froidement légèreté papeterie

[EUYP$ @ [IUZDGP$ @ [OH=(UWH] [SDS(WUL]

‘shortly’ ‘coldly’ ‘lightness’ ‘stationery’

In all these items, the second nucleus from the right is realized as zero. The representation of the word rejeter [r=WH] ‘to throw back’ illustrates this situation: (18)

O1

x | r

R | N1 | x | 

O2

x | =

R | N2 | x |

O3

x | t

R | N3 | x e

In (18), the phonetically manifested nucleus (N3) properly governs the adjacent nucleus (N2), which consequently remains mute. However, this unrealized nucleus cannot properly govern the preceding nucleus (N1), which must be interpreted as schwa. It is also not possible for the nucleus N3 to properly govern the nucleus N1 because the adjacency condition would be violated. It can be observed that, in a sequence of consecutive empty nuclei found word-medially, a nucleus adjacent to the nucleus with phonetic content can remain mute. Nevertheless, when a word-medial schwa is followed by a ‘h aspiré,’ it must be interpreted. This can be illustrated by the following words (Charette 1991, 88): (19)

va dehors rehausser

[YDG2U@ [URVH]

‘go outside’ ‘to raise again’

In the words in (19), an empty onset (‘h aspiré’) intervenes between the two nuclei (underlined). Here, the ‘h aspiré’ onset does not dominate any segment but only an empty skeletal point. Words which start with an ‘h aspiré’ begin phonetically with a vowel but with an empty onset point in the phonological structure.

Schwa: Phonological Coding versus Phonetic Manifestation

237

Furthermore, when preceded by a consonantal cluster, schwa must receive a phonetic interpretation. The examples listed below support this claim (Charette 1991, 29): (20)

brusquement forgeron pauvrement tristement

[EU,VNP$ @ [I2U=UR ] [SRXYUP$ ] [WULVWP$ ]

‘abruptly’ ‘blacksmith’ ‘poorly’ ‘sadly’

Even though the empty nuclei are properly governed by a following vowel, they are interpreted as schwa due to the requirement of government-licensing (Charette 1988, 208). It must be said that, in French, when proper government is in conflict with government-licensing, the latter is chosen. This happens because the head in a governing domain must govern its complement. Thus, the governor must be governmentlicensed by the following nucleus. Consequently, proper government is not applied and the empty nucleus is realized as schwa which governmentlicenses the head of the preceding cluster. Another interesting case is an empty nucleus followed by a cluster made up of an obstruent and a glide. Depending on where the glide belongs, the nucleus will be manifested or not (this is dialect-specific). If the ‘glide’ is part of a branching nucleus, the preceding empty nucleus can remain mute. Thus, the word cimetière [simtj(r] ‘cemetery’ has the representation depicted in (21) (Charette 1991, 116): (21)

O1

x | s

R | N1 | x | L

O2

x | P

R | N2 | x

O3

x | t

R | N3 /\ xx || i(

O5

x

R | N4 | x

r

(21) shows that, in Parisian French, the [i] belongs to the light diphthong [i(]. There is no intervening domain between the nuclei (N3) and (N2), as a result of which proper government is applied and the nucleus (N2) is interpreted as zero. Apart from this, adjacent empty nuclei found wordmedially and in the sequences of clitics must receive phonetic content. By virtue of this fact, a word such as devenir [dvnir] or [dvnir] ‘to become’ will not be realized as *[dvnir]. However, it is possible to find sequences

Chapter Thirteen

238

of adjacent empty nuclei in French. The relevant examples are presented below (Charette 1991, 199, 201): (22)

a. douze chemises b. quart de litre c. mange de la soupe d. pas de lune

[GX]6PL]] [NDUGOLWU] >P$ =GODVXS] >SDGO\Q]

‘twelve blouses’ ‘quarter of a litre’ ‘eat some soup’ ‘moonless’

In the word douze in (22a), the final empty nucleus is licensed. When followed by chemises, although it is not properly governed, the empty nucleus does not receive phonetic content. In addition, the empty nucleus in the initial syllable of the word chemises is properly governed by the final expressed nucleus [i] and remains mute. On the other hand, in the expression in (22b), the de is preceded by a word ending with an empty nucleus which cannot properly govern the nucleus in de. As a result, it becomes filled with the melody of schwa. The sequence of two adjacent empty nuclei is allowed here because these nuclei belong to two distinct words. In (22c), the word-final nucleus in mange remains mute because it is parametrically licensed. The empty nucleus in de is also manifested as zero due to the fact that the expression de la is treated as one word. Consequently, the word-internal empty nucleus is properly governed by the following nucleus [a]. In pas de lune, the nucleus of de can be unrealized since it is preceded by the word pas whose final nucleus receives an [a] interpretation and can properly govern the nucleus in de.

6.3. The right edge of the word According to Charette (1988, 234), a word-final nucleus can be underlyingly empty in French words. It is subject to parametric variation, depending on whether these empty nuclei are realized as zero or filled with melodic content. French opts for the former mechanism. What is more, word-final empty nuclei, though parametrically licensed, can perform the function of government-licensers. Dell (1973) argues that a lexical schwa can be found word-finally in all feminine adjectives and words ending with an obstruent. Nevertheless, these word-final ‘schwas’ remain mute. The examples below clearly show this situation (Charette 1988, 235):

Schwa: Phonological Coding versus Phonetic Manifestation (23)

a. belle grande lente malade

[E(O] [JU$ G] [O$ W] [PDODG]

‘beautiful’ ‘tall’ ‘slow’ ‘sick’

b. bague mousse tasse viande

[EDJ] [PXV] [W$V] [YM$ G]

‘ring’ ‘moss’ ‘cup’ ‘meat’

239

The words in (23) show that a word-final schwa preceded by a consonant does not receive phonetic content either in the feminine forms in (23a) or in the words ending with an obstruent in (23b). The same situation happens when a word-final schwa follows a consonant cluster. To support this observation, Charette (1991, 122) presents the following examples: (24)

barbe orfèvre pègre verte

[EDUE] [2UI(YU] [S(JU] [Y(UW]

‘beard’ ‘goldsmith’ ‘underworld’ ‘green’

It can be observed in (24) that even if a word-final empty nucleus follows a branching onset or a coda onset, it is not realized phonetically. In these words, a nucleus follows an onset to satisfy universal requirements, i.e., that an onset must be licensed by a following nucleus (Onset Licensing Principle). Thus, the representation of the word autre [otr] ‘other’ can be depicted as follows: (25)

O1

x

R | N1 | x | R

O2

x

x

t

r

R | N2 | x

In (25), the word-final empty nucleus (N2) government-licenses the preceding head of a branching onset to govern its complement. Because the Final Empty Nucleus parameter in French is set on, words can end with a consonant or a consonantal cluster. Even though an empty nucleus is present in the lexical representation, it does not receive phonetic

240

Chapter Thirteen

interpretation. Additionally, a word-final empty nucleus can be a direct or indirect government-licenser of a preceding onset, but this is also subject to parametric variation. In French, a licensed word-final empty nucleus can indirectly government-license a preceding onset to govern its complement within a branching onset or directly in the case of the preceding rhymal complement-onset domain.

7. Conclusion In the foregoing discussion, we have demonstrated that the French vowel schwa is the phonetic realization of an ungoverned empty nucleus. It was shown that a word-initial syllable deserves special treatment and the occurrence of schwa there is due to the preceding consonant or a consonantal cluster which is to be licensed by a full vowel. Also, it was stated that the absence of schwa in the word-medial position is a result of the application of proper government and that its presence is connected with proper government failure. Additionally, it was shown that the French ‘h aspiré’ triggers schwa occurrence word-medially. This follows from the principle which states that a proper governor can properly govern only one empty governee. The French data proved that, even though an empty nucleus is properly governed by a following nucleus, it acquires phonetic interpretation, which happens in words where proper government and government-licensing are in conflict. In such a case, the latter mechanism takes the upper hand. Importantly, the head of a governing domain which precedes the nucleus in question must govern its potential governee. Hence, an empty nucleus will license such a relation either directly or indirectly, depending on the type of relation involved. The examination of the word-final context revealed that even if preceded by a consonant or a consonantal cluster, the final nucleus remains empty, thus performing the role of both a direct and an indirect government-licenser. Schwa in English exhibits different properties than in French. First, it is noteworthy that it can be found in the word-initial position. The data revealed that the English schwa can be an empty nucleus filler when proper government is not applied. However, it will never be found in stressed syllables. This means that it is invisible to stress. The final or penultimate syllable containing schwa is skipped during stress assignment. The presence of schwa after branching onsets allowed us to conclude that it can function as a government-licenser. In addition, it turned out that the two principles, i.e., proper government and government-licensing, can be in conflict in English. The presence of schwa in the word production, for

Schwa: Phonological Coding versus Phonetic Manifestation

241

instance, showed that English chooses government-licensing instead of the proper government of an empty nucleus. Also, schwa is a frequent output of vowel reduction in unstressed positions. Its omnipresent character manifests itself in phonetic realization when the neutral element is promoted to the head position (Harris 1994). Further, the reason for the absence of schwa in *>PS] and *>1N] clusters was explained. It was argued that it is due to the inability of schwa to transfer the licensing potential onto the above-mentioned clusters. In English, schwa can license the rhymal complement only if this position is already strongly governed by the following onset (the so-called easy-to-license cluster). Because of the homorganicity and complexity purposes, such a cluster is easier to license for schwa. Finally, the behaviour of schwa in the word-final position was analysed. The data led to the conclusion that, again, schwa is strong enough to government-license the preceding governing domains. It was pointed out that it can license the same structures as full vowels. Even though English disallows word-final branching onsets, they are attested when followed by schwa.

References Anderson, Stephen. R. 1982. “The Analysis of French Schwa: Or, How to Get Something for Nothing.” Language 5:534–73. Bloch-Rozmej, Anna. 2008. Melody in Government Phonology. Lublin: Wydawnictwo KUL. Charette, Monik. 1988. “Some Constraints on Governing Relations in Phonology.” PhD diss., McGill University. —. 1991. Conditions on Phonological Government. Cambridge: CUP. Cyran, Eugeniusz. 2003/2010. Complexity Scales and Licensing Strength in Phonology. Lublin: Wydawnictwo KUL, Mouton de Gruyter. Dell, Franoois. 1973. Les Règles et les Sons. Introduction à la Phonologie Générative. Paris: Hermann. Gimson, Andrew. C. 1980. Pronunciation of English. London: Hodder Education. Gussmann, Edmund. 2002. Phonology: Analysis and Theory. Cambridge: CUP. —. 2007. The Phonology of Polish. Oxford: OUP. Harris, John. 1994. English Sound Structure. Oxford: Blackwell. Kaye, Jonathan. 1990. “‘Coda’ Licensing.” Phonology 7:301–30.

242

Chapter Thirteen

Kaye, Jonathan, Jean Lowenstamm, and Jean-Roger Vergnaud. 1985. “The Internal Structure of Phonological Elements: A Theory of Charm and Government.” Phonology Yearbook 2:305–28. —. 1990. “Constituent Structure and Government in Phonology.” Phonology 7:193–231. Oostendorp, Marc van. 2000. Phonological Projection: A Theory of Schwa in Optimality Theory. Berlin: Mouton de Gruyter. Szigetvári, Peter. 1994. “Coronality, Velarity and Why They are Special.” The Even Yearbook 1:185–224. Budapest: Eötvös Loránd University.

CHAPTER FOURTEEN THE PHONETICS AND PHONOLOGY OF THE SCOTTISH LATERALS MATEUSZ URBAN AND SàAWOMIR ZDZIEBKO

1. Introduction In an earlier paper (Zdziebko and Urban 2014), we undertook a spectral analysis of the phenomenon of //-vocalisation in Scottish English, based on data from Ayrshire Scottish English (ASE). Our primary aim was to determine the acoustic effects that correspond to this variation. We also suggested a preliminary analysis of the variation in terms of Element Theory. The ultimate aim of the current chapter is to introduce some revision to the said analysis. Based on the results of our acoustic analysis, we determine the representation of the lateral in ASE in terms of Element Theory and check how it compares to the representation of laterals suggested by Backley (2011). Section 2 discusses some problems related to the representation of laterals in generative phonology and summarizes Backley’s position on this issue. In section 3, we point to the lack of convincing phonological evidence in favour of the presence of the element A in English laterals and suggest that the evidence should be looked for in the acoustic properties of the relevant sounds. Section 4 introduces the methodology and the data that were used in the acoustic measurements we conducted. Section 5 presents the results we obtained and compares them with the predictions made by the approach that assumes that the elements A and U form the melodic makeup of []. We conclude that the patterns of the differences in the frequencies we obtained cannot be explained if the representation of [] is {A.U}. In section 6, we put forward an alternative approach to the phonological representation of the Scottish lateral. Section 7 provides concluding remarks.

244

Chapter Fourteen

2. Phonological representation of laterals1 Articulatorily-defined feature systems seem to encounter a major difficulty when trying to come to grips with laterals. The general articulatory bias has been noticeable since the SPE and is clearly visible in Chomsky and Halle’s (1968, 317) description of [+lateral] segments, which are “produced by lowering the mid section of the tongue at both sides or at only one side, thereby allowing the air to flow out of the mouth in the vicinity of the molar teeth.” While Spencer (1984) may be right in claiming that this articulatory description of [+lateral] itself is “fairly obvious” (ibid., 23), one fundamental problem with this feature is its ambiguous correlations with other articulatorily-defined feature values. For example, one of the main strands of controversy concerning the phonological representation of laterals is their unclear status with respect to the feature [±continuant]. This was already recognized in Chomsky and Halle (1968): If the defining characteristic of the stop is taken (as above) as total blockage of air flow, then [l] must be viewed as a continuant and must be distinguished from [r] by the feature of “laterality.” If, on the other hand, the defining characteristic of stops is taken to be blockage of air flow past the primary stricture, then [l] must be included among the stops. The phonological behaviour of [l] in some languages supports somewhat the latter interpretation. … On the other hand, there are other facts in different languages which suggest that [l] is best regarded as a noncontinuant (with the definition of the feature adjusted accordingly). (Chomsky and Halle 1968, 318)

This inherent ambiguity results in an ambivalent analysis of laterals with respect to the feature in question.2 This ultimately leads Holt (2002) to

1

For reasons of space, the following account makes no pretence to exhaustiveness. See Yip (2011) for an excellent up-to-date review of the topic with the author’s own solutions. 2 Based on various sources, Holt (2002, 85–86, fn. 1) gives the following examples of languages with /l/ interpreted [+continuant] (Musey, Zoque, Chipewyan, Frisian), and those where /l/ is considered [–continuant] (Basque, English, Scots English, Gallo-Romance, Korean and Luganda). He later summarizes the analyses of Spanish /l/, which is taken to be [–continuant] by some authors, but [+continuant] by others (see ibid., 90 with references). James W. Harris, who is quoted by Holt as being in the latter group, based on a 1984 article, seems to have

The Phonetics and Phonology of the Scottish Laterals

245

interpret /l/ as both [+continuant] and [–continuant] at the same time, whereas Mielke (2005) argues that distinctive features are emergent rather than universally defined and that phonetically ambiguous segments are likely to exhibit ambivalent phonological behaviour. Thus, in this view, laterals, which share articulatory characteristics with both continuants and noncontinuants (see the quotation above), belong with either one class or the other, depending on the phonological behaviour they exhibit in a given language. This last idea is not entirely remote from the approach taken by the proponents of Element Theory (Harris and Lindsey 1995; Backley 2011), where the presence or absence of a phonological prime depends on the phonological patterns in which a given segment participates. The difference lies in the fact that ambiguities in the phonological makeup of segments are not further motivated by phonetic ambiguities. Furthermore, to the extent that it relies on phonetic argumentation, Element Theory follows the tradition associated with Jakobson, Fant, and Halle (1952) and, more recently, the proponents of Dependency Phonology (Anderson and Ewen 1987), in giving preference to the acoustic properties of the speech signal rather than the articulatory gestures that shape them. Backley’s (2011, 175–84) discussion is the first attempt at a comprehensive cross-linguistic treatment of laterals as a class within Element Theory. He interprets them to be glides, parallel to [w] = {U}, [j] = {I} and [Š] = {A}. What they are supposed to share with the latter is the presence of the element A, which is meant to account for the fact that laterals frequently pattern with rhotics. What makes laterals different from other glides is that they are complex: apart from A, a lateral will feature either I or U. The fact that laterals, like all glides, can only feature the resonance elements I, U and A suggests that in terms of traditional divisions they are prototypically continuants and this corresponds well with their vowel-like spectra. Finally, Element Theory does not preclude the existence of other, non-glide laterals, which may be expressed as A combined with I or U, as well as either  (for noncontinuant laterals) or H (for voiceless laterals). It remains to be added that the vocalization of laterals is analysed by Backley (2011, 179, 182) as a suppression of the element A from the representation. The result is either [w] or [j] depending on whether the original lateral was a dark one (i.e., {U.A}) or a clear or even palatal one (i.e., {I.A}). changed his mind, as in his earlier PhD thesis /l/ was analysed to be [–continuant] (Harris 1969, 43).

246

Chapter Fourteen

3. Our point of departure While the presence of A as an obligatory element in laterals seems to be well justified in many languages (see the evidence adduced by Backley himself), there is little compelling evidence for its presence (or absence) in the representation of these segments in English. Backley (2011, 180) brings up the example of American dialects which have an intrusive [l] after [2«], which is represented as {A.U}. The process of [l] intrusion is then accounted for by the spreading of both resonance elements to the following empty onset position.3 This, however, does not confirm the presence of A in laterals in all dialects. Furthermore, Backley’s (2011, 181) example of the alleged [l] insertion before the nationality suffix -ese [i«]], when it follows stems terminating in [R«] or [8], as in Congolese and Togolese, is hardly a fact about English. Both these words are borrowed from French and the lateral is present in their etymons, cf. congolais, togolais (see OED3 s.v. Congolese and s.v. Togolese). In the absence of convincing phonological evidence pointing to the presence of A, we turn to acoustic phonetics. The fact that laterals are characterized by periodic energy, which makes them similar to vowels, should make it possible to investigate the participation of the particular resonance elements in their overall spectral shape. Therefore, the question that is at the outset of the current chapter is how well Backley’s analysis corresponds with the spectral properties of the segments in question. As a test case, we took the dialect of English spoken in Ayrshire, Scotland. The only phenomenon that may optionally affect laterals in this variety is that of vocalization to a non-syllabic [u] or [w] in a post-vocalic context, but the question still remains open of what element(s) should be taken to undergo suppression. In order to determine this, we decided to investigate changes in formant frequencies between the adjacent vowel and the non-vocalized lateral, and compare them with the formant frequency changes observed in the case of the vowel and the vocoid.

3

For an alternative account of intrusive [r] and [l] in dialects of English expressed in terms of Element Theory, see Bloch-Rozmej (2008, 264–70) and, in similar vein, Kijak (2010).

The Phonetics and Phonology of the Scottish Laterals

247

4. Preliminaries to the study The recordings of Ayrshire Scottish English speakers that provided the material for the current study come from the PAC corpus (La Phonologie de l’Anglais Contemporain: usages, variétés et structure / The Phonology of Contemporary English: usage, varieties and structure) (Carr, Durand, and Pukli 2004). Seven speakers (four male and three female) were determined to exhibit // vocalisation. The study is based on two word lists and a passage of read speech. The method used (see further) meant that the analysis had to be limited to // in the neighbourhood of a stressed nucleus. Dipthongs were disregarded due to their poor representation in the recordings. Of the monophthongs found in Standard Scottish English the following were omitted for similar reasons: /,/, /o«/, /(Ã/, //, /¡/, /2«/ and /a«/. Furthermore, /e«/ and /(«/ were ignored as well because their distribution in the lexical classes is subject to speaker-by-speaker variation. Thus, only those lexical items were subjected to scrutiny in which // preceded or followed a stressed /i/, /‫ݛ‬/, or /‫ܧ‬/. It is the latter context that favours vocalization. The word lists and the passage used to elicit read speech samples featured 14 tokens with contexts promoting vocalization. By way of comparison, a further 19 tokens with non-vocalizing contexts were added to these. The former group of tokens were then classified impressionistically into two subgroups, based on the presence or absence of vocalization. As a second step, measurements were taken of the F1 and F2 frequencies for all 33 instances of the lateral as well as the preceding or following stressed /i/, /‫ݛ‬/, or /‫ܧ‬/.4 This was followed by computation of the changes in formant frequency according to the formula FX(V) – FX([w]/[ܽ]). Finally the measurements were tested for statistical significance by means of a Student’s t-test for correlated or independent samples depending on whether the number of relevant tokens was the same or different for the two variants of the lateral.

5. The acoustic analysis of //-vocalisation In this section we present the results of the measurements of the formant frequency differences between the vowels /L/, /2/ and /Ê/ and the vocalised 4

The measurements of F3 frequencies did not provide us with conclusive results.

Chapter Fourteen

248

and non-vocalised variants of the Scottish lateral. After the presentation of the results for each vowel, we point to possible problems with the assumption that the ‘dark’ variant of the lateral is represented as a ‘complex glide’ comprising the elements A and U, as claimed by Backley (2011). Let us begin with the results for the close front vowel /L/. Table 14-1 presents the results for /L/ and the vocalised variant of the lateral, i.e., /Z/. Table 14-1. Average differences between /L/ and /Z/ syaew syakd syamg syaeh syacc syadm syabw MEAN

F1(/L/)-F1(/Z/) 34.666 -0.4 20.883 17.667 69.5 73 44 37.04

F2(/L/)-F2(/Z/) 1458.8 1454.4 1700.3 1177 984.5 794 891.4 1208.62

As can be seen from the table, the differences in F1 between the vowel and /Z/ never exceed 80 Hz, while the differences in F2 range from 794 to 1700 Hz. The lack of significant change in F1 and the substantial drop in F2 are expected if /L/ is composed of the element I and /Z/ of the element U, because the acoustic effect associated with the element U is the lowering of both F1 and F2, while the acoustic correlates of I are low F1 and high F2. Since the absolute formant values of sounds, and, consequently, the differences in the values of formants of different sounds are relative, the full picture of the acoustic correlates of //-vocalisation is available only when one compares the results from Table 14-1 above with the differences between F1 and F2 of /L/ and the non-vocalised //. These are presented in Table 14-2.

The Phonetics and Phonology of the Scottish Laterals

249

Table 14-2. Average differences between /L/ and // syaew syakd syamg syaeh syacc syadm syabw MEAN

F1(/L/)-F1(//) 10.6 4.25 26 40.75 -26.75 55.16 -37.7 10.33

F2(/L/)-F2(//) 916.5 1081.3 928.4 626.25 397.25 524.17 348.17 688.86

Table 14-2 clearly shows that, on average, the changes in F1 and F2 in the sequence /i/ are smaller than those in /LZ/. However, whereas the differences between the results obtained for F1 in Table 14-1 and Table 14-2 are not statistically significant, with t=1.55 (p=0.086059), the change in F2 is significantly greater in the case of the sequence /LZ/, with t=8.58 (p1.796 p1.796 p