189 96 741KB
English Pages 162 Year 2017
Exploring the Syntax and Semantics of South Asian Languages
Exploring the Syntax and Semantics of South Asian Languages Edited by
Reena Ashem, Gurmeet Kaur and Usha Udaar
Exploring the Syntax and Semantics of South Asian Languages Edited by Reena Ashem, Gurmeet Kaur and Usha Udaar This book first published 2017 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 © 2017 by Reena Ashem, Gurmeet Kaur, Usha Udaar 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): 1-4438-7320-9 ISBN (13): 978-1-4438-7320-8
TABLE OF CONTENTS
Acknowledgement ..................................................................................... vii Chapter One ................................................................................................. 1 Exploring the Syntactic Structures of South Asian Languages— An Introduction Reena Ashem, Gurmeet Kaur and Usha Udaar Chapter Two .............................................................................................. 11 Property Concepts and the Degree Expression in Kannada Sindhu Herur Subramanya Chapter Three ............................................................................................ 37 Semantically Elaborate Categories and Grammaticalization in Malayalam Reshma Jacob and Shreya Mehta Chapter Four .............................................................................................. 55 Aspects in Meeteilon Padmabati Achom Chapter Five .............................................................................................. 73 Semantics of Reduplicative Nominal Quantification in Bangla Nandini Bhattacharya Chapter Six ................................................................................................ 93 The Allomorphs of Genitive and Dative and the Postulation of Grammatical Gender in Malayalam Gouthaman K.J. Chapter Seven.......................................................................................... 113 Understanding Ergative Case Licensing in Haryanavi Usha Udaar
vi
Table of Contents
Chapter Eight ........................................................................................... 129 Dative and Ergative Subject Constructions in Punjabi: Understanding Person Agreement Gurmeet Kaur Index ........................................................................................................ 153
ACKNOWLEDGEMENT
We wish to thank Cambridge Scholars Publishing for their interest in our project and considering it for publication. Thanks also to Victoria Carruthers for guiding us through the entire publication process. Needless to say, this volume would not have materialised without the help and perseverance of all our contributors, who have met our demands untiringly over the last few months. We are also very grateful to our reviewers - Pritha Chandra, K.V. Subbarao, Rajesh Kumar, Mythili Menon, Priyanka Biswas- for their insightful comments on the papers. This volume is a by-product of the Ninth Students’ Conference of Linguistics in India held at the Indian Institute of Technology Delhi. We express our heartfelt gratitude to the entire Linguistics Unit of the Department of Humanities and Social Sciences for their effort in organizing the conference. Without their support, neither the conference nor the volume could have come into being.
CHAPTER ONE EXPLORING THE SYNTACTIC STRUCTURES OF SOUTH ASIAN LANGUAGES— AN INTRODUCTION REENA ASHEM, GURMEET KAUR AND USHA UDAAR1
1. Introduction This volume is a compilation of selected papers presented at the Ninth Students' Conference of Linguistics in India (SCONLI-9), which took place at the Indian Institute of Technology Delhi from 14th to 15th March, 2015. The ninth edition of the conference brought together young researchers from various sub-disciplines of linguistics from all parts of the country. The conference was divided into seven sessions over a period of two days, where research scholars from among the students chaired each session. Papers on various topics including case, agreement, adjectives, finiteness, topic modelling and machine translation were presented by the participants. These issues were illustrated via various South Asian languages such as Malayalam, Tamil, Kannada, Hindi, Magahi, Punjabi, Haryanavi, Braj, Bundeli, Bangla, Assamese, Meiteilon, Sylheti, Badaga, Khoibu and Maring.2 The present volume contains seven papers selected from those presented at the conference.
1
The author names’ have been listed according to the alphabetical order of the last names. 2 There is a divide between linguists with regard to the spelling of the TibetoBurman language spoken by the Meiteis in Manipur. While some linguists use Meiteilon, others prefer Meeteilon. We adopt the former spelling in this paper, except for where the contributors have used the alternative.
2
Chapter One
South Asian languages (henceforth SALs) have attracted a lot of attention in the linguistic literature. Some of it derives from the shared similarities between these languages despite them belonging to different language families. To elaborate, languages in South Asia belong to four different language families—Indo Aryan, Dravidian, Austro-Asiatic and TibetoBurman—but share several linguistic traits among themselves (Abbi, 2012), thereby constituting a “linguistic area” (in the sense of Emeneau 1956, 1980 and Masica 1976). Some of these shared linguistic traits are retroflex sounds, SOV word order, absence of prepositions, morphological reduplication, and complex predicates among others. For a detailed exposition of these traits across various SALs, see Abbi (1991/1992, 2001, 2012). However, it must be noted that the interest in SALs has not been restricted to their typological (un)relatedness, but has extended well beyond into the generative framework. Noted studies on different SALs by scholars such as K.P. Mohanan (1982), Gurtu (1985), Mahajan (1990), Srivastav (1991), T. Mohanan (1994), Jayaseelan (1999, 2001), Kidwai (2000) among others have had an impact on the development of linguistic theory. With investigation of issues on topics ranging across the board— status of primitive categories (nouns, verbs and adjectives), wh-questions, scrambling, clause structure, case and agreement—these studies have used empirical evidence from SALs to ask crucial questions that have helped shape theory. In this volume we focus on three of these topics: (a) status of primitive categories, (b) clausal and nominal structure and (c) case and agreement. Specifically, this volume presents a compilation of papers each of which attempts to investigate one of these three topics. Each paper puts forth novel data from SALs, and provides descriptive-theoretical analysis of the linguistic phenomenon covered. The current volume thereby paves the way to refining our empirical as well as theoretical understanding of the system of language. In the next section, we discuss the importance of said topics in the generative literature, followed by demonstrating how studies on SALs have furthered our understanding of these issues. The final section will focus on the contribution of each paper of the present volume to the same linguistic issues.
2. Relevance of the Topics in the Purview of the Theory and Studies on SALs It is generally assumed that lexical categories like nouns and verbs are universal and found in all languages of the world (see Hale and Keyser,
Exploring the Syntactic Structures of South Asian Languages
3
2003). However, some SALs present a problem for such a claim. Consider the case of Mundari, a Munda language. Peterson (2007) proposes that the language does not have separate noun-verb classes, in that a single word can function as a noun, a verb or an adjective according to the context. A similar proposal has been made for adjectives in Dravidian. Amritavalli (2008) in her study of Kannada, and Jayaseelan (2007) in his work on Malayalam respectively have argued that adjectives are not a primitive lexical category for the two languages under consideration. Specifically, both authors claim that adjectives in these Dravidian languages are derived by incorporation of case markers or postpositions into verbs or nouns. Menon (2014) also argues that there is an absence of the category of adjectives in Malayalam. However, she differs from existing claims by proposing not only that adjectives are not present as a lexical category in the language, but also that they are not derived in the syntax by operations on case markers. The language expresses adjectival meaning via relativization and nominalization. SALs have also broadened discussions on clausal syntax, especially the issue of tense-aspect-mood (TAM) projections. Since the seminal work of Pollock (1989) and its incorporation into the early minimalist developments (cf. Chomsky 1989), the structure of the clause above the VP has become an important topic in syntactic research. In this respect, SALs have been of key interest due to the presence of tenseless languages like Meiteilon, Malayalam, and Kannada. For example, exploring the clausal syntax of Kannada, Amritavalli (2007) suggests that there is no category of T(ense) in the language, such that a clause in Kannada is not a TP, but a MoodP. On the other hand, Kidwai (2010) in her work on another tenseless language, Meiteilon, argues for the presence of a T-like head in the clause structure insofar that this head inherits uninterpretable features from a higher C-like dominating head. Interest in clausal syntax is furthered by the phenomenon of clausal nominalization in SALs. Nominalization is generally understood as the process of “turning something into a noun” (Comrie and Thompson 1985). It is of two types: derived, where a verb acts like a noun phrase; and clausal, where the full clause acts like a noun phrase. Most of the existing literature on nominalization has focused on derivational nominalization, wherein verbs are nominalized to derive nouns and adjectives. However, SALs like Newar and Mongsen Aao have been shown to have clausal nominalization, structurally represented as [clause]NP (DeLancey 1999, 2002 and Genneti et al. 2008). This has raised interesting issues pertaining to how we differentiate clausal from nominal units.
4
Chapter One
On the level of syntactic operations like agreement and case valuation/checking, SALs have been extensively interrogated. In the domain of agreement, two studies that merit our attention are Bhatt (2005) and Chandra (2007). In Chomsky’s system of Agree (2000, 2001), case and agreement are understood to go hand in hand, such that case is a sideeffect of phi feature agreement. A DP which has been case valued (as a free-rider on phi agreement) is no longer eligible to enter into further agreement relations. Long-distance agreement (LDA) constructions in Hindi-Urdu have provided a fertile ground for understanding Agree. Specifically, employing LDA structures in Hindi-Urdu, Bhatt (2005) argues for dissociation between case and agreement. He proposes AGREE, as per which phi feature agreement is possible with a DP which has already been case valued. Also investigating long-distance agreement cases in Hindi-Urdu along with other languages like Tsez, Chandra (2007) questions the status of Agree as a core grammatical operation and instead proposes that all agreement must always take place between sisters, with no agreement taking place in a c-command configuration. Given the rich case morphology attested in SALs, the phenomenon of case has also not gone unnoticed. The literature on case in SALs has raises relevant issues pertaining to the PP status of case-markers and structural configurations that license them. Let us illustrate with a couple of examples. The works of Spencer (2005) and Kidwai (2011) on case markers in Hindi-Urdu have shown that the morphological forms -ne and ko, understood as ergative and accusative case respectively, are not realizations of case but only postpositions that do not project. Further, the discussion of the Hindi-Urdu ergativity has been of key interest to understand if ergative case is an inherent or a structural case. In this respect, employing perfective constructions with complex predicates in the language, Mahajan (2012) has argued that ergative is an inherent case valued on the subject by the v head which hosts the light verb. Not just ergative, but also dative case has been explored by Davison (2003) and Bhatt (2003) for Hindi-Urdu and by Jayaseelan (2004) for Malayalam. Analyzing the dative case as lexically marked, these works have raised questions pertaining to the status of the dative DP in dative subject constructions(sentences in which the logical subject of a clause takes the dative case, rather than the nominative case). To recapitulate, these studies indicate how relevant the investigation of South Asian languages has been to shaping our understanding of various linguistic phenomena including clause structure, status of primitive categories and syntactic operations. Not only have these works refined our
Exploring the Syntactic Structures of South Asian Languages
5
conception of the system of language, but they have also made accessible the special empirical properties of SALs to the general linguistic audience. That being said, the current volume is an attempt to further the discussion on SALs in the generative paradigm. Concretely, the papers in the current volume probe into the three domains under consideration—status of primitive categories, clause structure and syntactic operations—in a range of SALs and attempt descriptive-theoretical analyses in light of the existing literature. The first set of papers by Herur, and by Jacob & Mehta probes the nature and status of lexical items and categories in Kannada and Malayalam respectively. Papers by Achom and Bhattacharya deal with issues like nominalization and clause structure in Meiteilon and Bangla. The final set of papers by Gouthaman, Udaar and Kaur focus on the syntactic and morphological underpinnings of case and agreement in their respective works on Malayalam, Haryanavi and Punjabi. We highlight the key claims of each of these papers in the next section.
3. Contribution of the Present Volume 3.1 Status of Lexical Items/Categories In purview of the current literature on adjectives in Kannada, Sindhu Herur discusses adjectives and comparative constructions in her paper ‘Property Concepts and the Degree Expression in Kannada’. Exploring the underlying nature of property concepts in Kannada, Herur shows that they can be expressed either by nouns or by adjectives in the attributive position. In the predicative position, in contrast, property concepts are expressed by adjectives that are syntactically derived from nouns marked with a dative case. The separate existence of nouns and adjectives is further substantiated by the distribution of heccu, the Q element in comparative constructions in the language. While the element is optional with adjectival comparatives, it is required obligatorily with nominal comparatives. Carrying forward the discussion on lexical categories, Jacob and Mehta add upon Heine and Kuteva’s (2009) work on grammatical categories and extend their analysis to Malayalam in their paper ‘Semantically Elaborate Categories and Grammaticalization in Malayalam’. The paper uses semantic analysis of verbal predicates and demonstrates that similar categories can be drawn in Malayalam through the process of grammaticalization.
6
Chapter One
3.2 General Aspects of Clausal and Nominal Structure Focusing on the structure of the clause above the VP in Meiteilon, the paper ‘Aspects in Meeteilon’ by Padmabati Achom gives an elaborate description of aspectual constructions in the Tibeto-Burman language. Achom argues for the presence of two distinct aspectual heads in Meiteilon- an inner and an outer aspectual head, by providing ample empirical evidence. She claims that the outer aspect head lies above the functional vP and is realised morphologically. On the other hand, the inner aspectual ahead lies inside the lexical VP; it usually remains unrealised. However, certain deictic suffixes in the language help provide the semantics of this aspect head. In a slightly different vein, Nandini Bhattacharya probes into the nominal domain in Bangla by investigating the semantics of nominal quantification in the language. Specifically, the paper titled ‘Semantics of Reduplicative Nominal Quantification in Bangla’, addresses the issue of quantification with a focus on the role of nominal reduplication in encoding implicit quantification in Bangla. Reduplication is an important linguistic feature of South Asian languages, and has received attention in the works of Emeneau (1956) and Abbi (1991) among others. Bhattacharya recounts the earlier findings on the phenomenon at the level of phonology and morphology, and takes the discussion a step further to detail the semantics of implicit quantification, especially in Bangla. The paper investigates key linguistic differences between the reduplicated nominals and other NP expressions in the language. Finally, Bhattacharya offers a formal account of the distributive plurality that is expressed implicitly by such reduplicated nominals.
3.3 Case/Agreement Issues of case and agreement in SALs are investigated in three works in the volume. Genitive and dative case is explored in the paper titled ‘The Allomorphs of Genitive and Dative and the Postulation of Grammatical Gender in Malayalam’, where Gouthaman KJ explores the motivation behind distinct realizations of the two case markers in Malayalam. He presents the syntactic contexts that determine the choice of distinct forms of the case markers. Employing the case-stacking approach, the author posits that the distinct morphological forms of genitive and dative case in the language follow from different sets of nominal features that the Case head agrees with.
Exploring the Syntactic Structures of South Asian Languages
7
Moving from dative-genitive to ergative case, Usha Udaar elaborates upon the case licensing mechanisms in the paper titled ‘Understanding Ergative Case Licensing in Haryanavi’. Focusing on the syntactic nature of ergative case, the paper demonstrates that unlike nominative and accusative cases, the ergative case is neither licensed via phi-feature agreement between a functional head and a nominal nor related to theta assignment by a functional head. Instead, the ergative case in Haryanavi is a dependent case, licensed in the defective perfective aspect domain. The paper, therefore, elaborates upon the debate related to various case modalities existing within the generative literature. In the last paper titled ‘Dative and Ergative Subject Constructions in Punjabi: Understanding Person Agreement’, Gurmeet Kaur argues that person agreement is determined by the structural configuration between the agreeing functional head and the agreement triggering nominal (following Baker, 2008). To arrive at this claim, Kaur explores dative and ergative subject constructions in Punjabi. While the theme of the dative structure triggers full phi agreement on the T head in dative subject constructions, the theme of the ergative structure triggers agreement in number and gender alone. Investigating these structures in detail, Kaur claims that in dative constructions, the theme moves beyond the intervening dative subject and agrees with T in sisterhood, while agreement between the theme DP and v-T in the ergative construction takes place long distance for lack of an A-position for the movement of the theme. These varying structural arrangements correlate with +/- person agreement. In conclusion, we would like to reiterate that the chapters in this volume highlight many interesting and hitherto unnoticed features of SALs. Addressing issues ranging from the status of primitive categories to the working of operations like case valuation and agreement, these papers are a step forward in helping us understand the system of language better.
8
Chapter One
References Abbi, Anvita. “India as a Linguistic Area Revisited”. Language Sciences 13 vol no.2 (1991/1992) —. “Reduplication in South Asian Languages. An Areal, Typological, and Historical Study” (Delhi: Allied publishers, 1991). —. “A Manual of Linguistic Fieldwork and Structure of Indian languages” (Lincom Europa 1992). —. “Languages of India and India as a Linguistic Area”. Accessible at http://www.andamanese.net/Languages%20of%20India%20and%20In dia%20as%20a%20linguistic%20area.pdf. (2012). Amritavalli, Raghavachari and K. A. Jayaseelan. “Finiteness and negation in Dravidian”. In Oxford Handbook of Comparative Syntax, eds. G. Cinque and R. Kayne (2007): 178–220. Amritavalli, Raghavachari. “The origins of functional and lexical categories: Tense-aspect and adjectives in Dravidian”. Nanzan Linguistics 4 (2008): 1–20. Bhatt, Rajesh. “Experiencer Subjects: Topics in the Syntax of the Modern Indo-Aryan Language”. Handout. (Amherst: University of Massachusetts, 2003). Bhatt, Rajesh. “Long distance agreement in Hindi-Urdu”. Natural Language & Linguistic Theory 23 vol no.4 (2005): 757-807. Chandra, Pritha. “(Dis)Agree: Movement and Agreement Reconsidered”. Doctoral Dissertation. (College Park: University of Maryland, 2007). Chomsky, Noam. “Some Notes on Economy of Derivation and Representation”. MIT Working Papers in Linguistics vol no.10 (1989): 43–74. —. “Minimalist inquiries: The framework”. In Step by step: Essays on minimalist syntax in honor of Howard Lasnik, ed. by Martin, R., Michaels, D. and Uriagereka, J. (Cambridge: MIT Press, 2000): 89– 155. —. “Derivation by Phase”. In Ken Hale: A Life in Language, ed. by M. Kenstowicz. (Cambridge: MIT Press, 2001): 1-52. Comrie, Bernard and Sandra A. Thompson. “Lexical nominalization”. In Language typology and Syntactic description, Volume III: Grammatical Categories and the Lexicon, ed. by Timothy Shopen. (Cambridge: Cambridge University Press, 1985): 349–398. Davison, Alice. “Structural Case, Lexical Case, and the Verbal Projection”. In Clause Structure in South Asian Languages, ed. by Dayal, V. and Mahajan, A. (Dordrecht: Kluwer, 2003): 199-226.
Exploring the Syntactic Structures of South Asian Languages
9
DeLancey, Scott. “Relativization in Tibetan”. In Topics in Nepalese Linguistics, ed. by Yogendra P. Yadava and Warren W. Glover. (Kamaladi, Kathmandu: Royal Nepal Academy, 1999): 231–249. —. “Relativization and nominalization in Bodic”. In Proceedings of the 28th Annual Meeting of the Berkeley Linguistics Society, Parasession on Tibeto-Burman and Southeast Asian Linguistics, ed. by Patrick Chew. (Berkeley: Berkeley Linguistic Society, 2002): 55–72. Emeneau, Murray Barnson. “India as a Linguistic Area”. Language 32 (1956): 3-16. —. “Language and Linguistic Area” (Stanford CA: Stanford University Press, 1980). Genneti, Caroline R., Alexander Coupe, Ellen Bartee, Kristine Hildebrandt and You-Jing Lin. “Syntactic Aspects of Nominalization in Five Tibeto-Burman Languages of the Himalayan Area”. Linguistics of the Tibeto-Burman Area 31 vol no. 2 (2008): 97-143. Gurtu, Madhu. “Anaphoric Relations in Hindi and English”. Doctoral dissertation. (Hyderabad: Central Institute of English and Foreign Languages, 1985). Hale, Kenneth and Samuel J. Keyser. “On Argument Structure and the Lexical Expression of Semantic Relations”. In The View from Building 20, ed. by K. Hale and S. J. Keyser. (Cambridge, MIT Press, 1993): 53-109. Jayaseelan, K.A. “The Serial Verb Construction in Malayalam.” In Clause Structure in South Asian Languages, ed. Dayal, V. and Mahajan, A. (Dordrecht: Kluwer, 2004), 152-198. —. “The argument structure of the dative construction.” In Argument Structure, ed. E. Reuland, T. Bhattacharya & G. Spathas. (Amsterdam: John Benjamins, 2007), 37–48. —. “A Focus Phrase above vP,” Proceedings of the Nanzan GLOW, Nanzan University, Nagoya (1999): 195-212. —. “IP-internal Topic and Focus Phrases,” Studia Linguistica 55 (2001): 39-75. Kidwai, Ayesha. XP-Adjunction in Universal Grammar: Scrambling and Binding in Hindi-Urdu. (New York: Oxford University Press, 2000). —. “The cartography of phases.” In Edges, Heads, and Projections: Interface Properties, ed. Di Sciullo, A. and Hill, V. (Amsterdam: John Benjamins, 2010): 233-262. —. “Towards an Adpositional Analysis of Hindi Layer II Case Particles: Some Remarks on -se and -ko.” (Talk given at the Essex JNUKonstanz Workshop on Case and Elicitation Techniques, University of Essex, June 13-14, 2011).
10
Chapter One
Mahajan, Anoop Kumar. “The A/A-bar Distinction and Movement Theory.” Doctoral Dissertation (MIT, 1990). —. “Ergatives, antipassives and the overt light v in Hindi.” Lingua 122:3 (2012): 204-214. Masica, Colin P. “Dening a linguistic area: South Asia.” (Chicago: The University of Chicago, 1976). Menon, Mythili. “Property concepts and the apparent lack of adjectives in Dravidian.” In The Lexicon Syntax Interface: Perspectives from South Asian languages, ed. Chandra, P. and Srishti, R. (Amsterdam: John Benjamins, 2014): 25-52. Mohanan, Karvanuur P. “Grammatical Relations and clause structure in Malayalam.” In The Mental Representation of Grammmatical Relations, ed. Bresnan, J. (Cambridge: MIT Press, 1982): 504-589. Mohanan, Tara. Argument structure in Hindi. (Center for the Study of Language/CSLI, 1994). Peterson, John. “Languages without nouns and verbs? An alternative to lexical classes in Kharia.” In Old and New Perspectives on South Asian Languages.Grammar and Semantics. Papers growing out of the Fifth International Conference on South Asian Linguistics (ICOSAL-5 held at Moscow, Russia, 2003): 274-303. Pollock, Jean-Yves. “Verb movement, Universal Grammar, and the structure of IP.” Linguistic inquiry 20:3 (1989): 365-424. Spencer, Andrew, Miriam Butt, and Tracy Holloway King. “Case in Hindi.” In Proceedings of LFG: 5 (2005): 429-446. Srivastav, Veneeta. “WH Dependencies in Hindi and the Theory of Grammar.” Doctoral Dissertation. (Cornell University, 1991).
CHAPTER TWO PROPERTY CONCEPTS AND THE DEGREE EXPRESSION IN KANNADA1,2 SINDHU HERUR SUBRAMANYA
Abstract This paper connects two threads of research, namely, Property Concept (PC) expressions and the comparative degree construction, both of which are relatively novel for generative work in Kannada, a Dravidian language. I motivate a three-way classification of Property Concept expressions in Kannada based on the facts of attribution and predication. For Kannada, such a classification indicates that Property Concepts are expressed by both adjectives and nouns. The attributive position reveals a broad classification of PCs into two groups: PC adjectives and PC nouns. The PC adjectives form a small, functional class, the rest being PC nouns which form a huge majority. The predicative structures, however, reveal that some of these PC nouns take dative case, and these dative casemarked nouns are syntactically-derived adjectives according to the incorporation account in Amritavalli and Jayaseelan (2003) (henceforth A&J). This phenomenon of PC nouns incorporating into dative case to occur predicatively as adjectives seems unique to Kannada (among her sister Dravidian languages), though a similar construction is noted in Marathi.
1
ø = null, 1= first person, 2= second person, 3= third person, AGR = Agreement, TOP.=Topic, ACC=Accusative Case, DAT=Dative Case, GEN=Genitive Case, LOC=Locative Case, NOM=Nominative Case, F=feminine, M=masculine, N= neutral, SG=singular, PL=plural, Q=quantifier, PostP.=Postposition 2 Kannada, in this study, refers to the standard Mysuru-Bengaluru variety of the language.
12
Chapter Two
By analysing the degree expressions in Kannada I motivate a null degree head in comparative constructions. I analyse heccu on the lines of the Hindi-Urdu zyaadaa (Bhatt 2012) as a Q(uantifier) element and not a degree head. The distribution of heccu, which is optional in adjectival and obligatory in nominal comparatives, provides further support to this three way classification of PCs in Kannada. By considering cross-linguistic data, the study hints at a broader claim that the degree head always selects for a gradable predicate, be it an AP or a QP (and not vice-versa). Keywords: Property Concepts, Kannada, degree expressions, comparatives, Dravidian
1. Introduction This study explores how adjectival meaning is expressed in Kannada and its implication for the structure of the comparative degree construction. Kannada offers interesting insights into how adjectival meaning is expressed and encoded in Dravidian. In this paper, I refer to all terms expressing adjectival meaning by the semantic notion of Property Concept (PC) as in Dixon (1982). Amritavalli and Jayaseelan (2003) note that Dravidian languages have very few ‘true’ adjectives, with adjectival meaning being expressed mostly by nouns. A three-way classification of PCs3 based on predication structures is motivated for Kannada and supports such an observation made in A&J (2003). The first group of PCs consists of a handful of ‘true’ adjectives and we shall refer to them as PC adjectives. These adjectives are few in number and form a closed-class. Therefore, I analyse them to be functional in nature following Cinque (2010; 35-36) wherein the discussion on Yoruba, a language claimed to have only adnominal (attributive) adjectives forming a closed class, supports such a claim. The second group are ‘true-blooded’ nouns which occur predicatively in dative subject constructions, and we shall refer to them as PC nouns. The third group consists of a sub-class of these PC nouns which have dual properties of being ‘adjective-like’ as well as ‘noun-like’. They occur predicatively as adjectives in nominative subject constructions by taking dative case and they also occur as nouns in dative subject constructions. These dative case-marked nouns, which I analyse as syntactically-derived adjectives, seem to be unique to Kannada among
3 I use PCs as a broad term to refer to both nouns and adjectives in Kannada that express a property such as height, anger etc.
Property Concepts and the Degree Expression in Kannada
13
other Dravidian languages. However Marathi, an Indo-Aryan language, has dative case-marked PC nouns in nominative subject constructions. The three way classification of PCs in Kannada receives further support, with the distributional properties of heccu, a Q(uantifier) element (henceforth Q), in comparative constructions. This is the second thread. A brief look at the degree expressions in Kannada, namely, the positive (also referred to as the absolute), the comparative and the superlative degree constructions, leads us to an analysis of the instantiation of the three groups of PCs in comparative constructions in Kannada. I motivate a null degree head for Kannada that always encodes greater than semantics. The novelty of this paper lies in the fact that comparative constructions have not been studied from the perspective of the Property Concept expressions in Kannada. The interesting claims made in this paper are: one, a functional class of adjectives in Kannada; two, a group of syntactically derived adjectives in Kannada; and three, the importance of the degree head selecting only a gradable predicate. The paper, therefore, explores the instantiation of the three groups of PCs in comparative clauses and lays emphasis on the observation that in comparative constructions, the nature of the degree head has universal properties which make it select only a gradable predicate. Hence, the degree head selects for an adjective (which I assume to be inherently gradable) or a quantifier (as quantifiers have gradable properties as well). Hence, the role of heccu as a Q element becomes very clear in comparative constructions. When nouns and verbs participate in comparative constructions, the degree head cannot select for an NP or a VP directly as they are not gradable. Hence the degree head first selects a QP which can now select an NP or a VP. Thus, with nominal and verbal comparatives, heccu is obligatory. However, in the case of APs, the DegP can directly select for an AP and hence heccu is ‘not required’ or is not obligatory. Such a distribution of heccu in comparative constructions re-enforces the three-way classification of PCs motivated for Kannada. Heccu is not required in comparatives with the small group of functional PC adjectives and, very interestingly, is also not required with the syntactically derived adjectives (dative-case marked nouns). Heccu is obligatory with PC nouns.
14
Chapter Two
Recent work on PCs in Dravidian include Balusu (2015)4 who argues that adjectival meaning in Telugu is expressed by composing Property Concept (PC) nouns with the entities they modify, using possessive morpho-syntax. Malayalam has been shown to use verbal and nominal roots which combine with functional architecture to express adjectival meaning (Menon 2013 and Menon & Pancheva 2014). Thus, with Telugu expressing PC states with nouns, and Malayalam using PC roots to do the same, Kannada offers a third interesting perspective wherein apart from PCs expressed by nouns, the language has functional adjectives and syntactically-derived adjectives.
2. Property Concept Expressions in Kannada Let us begin with a discussion of PCs in Kannada and the motivations for a three-way classification. The background data considered for this study is a data set of fifty PCs in Kannada. Initially selected at random, various tests were applied to each PC and three groups emerged. These groups were developed and members were added. Tests of distribution and case marking helped make an initial classification of PCs into two broad groups: PC nouns and PC adjectives.
2.1 Property Concept Expressions in the Attributive Position There are two tests I have employed to ascertain the adjectival status of PCs in Kannada; one, whether they occur in the attributive position of NPs/DPs (Kennedy 1997) and two, whether they take case-marking. In the first part of this two-fold test, if the PC can occur in the attributive position of NPs in its ‘bare’ form and does not require genitive case to attributively modify NPs like nouns do in Kannada, then I proceed to classify them as adjectives. In the second part of the test, PCs, to be classified as adjectives, should not take any case-marking. I found a handful of ‘true’ adjectives which are listed in (1) below. Oѥѥeya ‘good’ being representative of this class of PC adjectives, I refer to this group of PC adjectives as the oѥѥeya group.
4
Balusu (2015) motivates a three-way classification in Telugu PC nouns, and though the three groups are not parallel to the Kannada classification, I owe much of the ideas in this paper to this work.
Property Concepts and the Degree Expression in Kannada
15
(1) Members of the oѥѥeya group of PC adjectives in Kannada: oѥѥeya ‘good’, keҀҀa ‘bad’, cikka ‘small’, doяяa ‘big’, hosa ‘new’, haѥeya ‘old’, yeѥeya ‘tender’ and baяa ‘poor’. Below is the data for oѥѥeya, being representative of its class, for the two mentioned tests. In (2a) below, oѥѥeya ‘good’ occurs in the attributive position of NPs. In (2b) we see that oѥѥeya resists any form of case marking. (2) a. oܿܿeya huܩuga
/ oܿܿeya huܩugi / oܿܿeya huܩug-aru / oܿܿeya
huܩugi-yaru good
boy
/ good
girl
/
good
boy-PL /
good
girl-PL
‘good boy / good girl / good boys / good girls’ b. * oܿܿeya-kke / * oܿܿeya-da
/ * oܿܿeya-dalli / *oܿܿey-vannu / * oܿܿey-u
good-DAT / good-GEN / good-LOC / good-ACC / good-NOM Comparing the data in (2) for oѥѥeya with a PC noun koopa ‘anger’, we see that the opposite holds true. Koopa, by itself, cannot attributively modify a noun as seen in (3a) and it takes case as illustrated in (3b). (3) a. *koopa maatu-gaܿ-u anger word-PL-NOM ‘intended: angry words (lit. *anger words)’ b. koopa-kke / koopa-da / koopa-dalli / koopa-vannu / koopa-vu anger-DAT / anger-GEN / anger-LOC / anger-ACC
/ anger-NOM
Koopa however can occur attributively, taking genitive case as in (4) below. The genitive case helps ‘link’ the two nouns koopa ‘anger’ and maatu-gaѥ-u ‘words’ which otherwise cannot co-occur (as seen above in 3a), as koopa is a noun and cannot attributively modify maatu-gaѥ-u, another noun.
Chapter Two
16
(4) koopa-da anger-GEN
maatu-gaܿ-u5 word-PL-NOM
‘words of anger’ (lit. anger’s words) A list of PC nouns is given below in (5). This group is similar to the –am ending nouns in both Malayalam as in Menon (2013), and in Telugu, discussed in Balusu (2015). PC nouns in Kannada are a mix of Sanskritborrowed nouns and native nouns. The phonology of Kannada ensures nouns end in an open syllable, i.e. vowels such as –a, -u and –e. Kannada does not have the nasal rule for the coda position like the –am ending rule for Malayalam and Telugu. To take an example, koopam ‘anger’ in Malayalam and Telugu is koopa ‘anger’ in Kannada. (5) Some members of the koopa group representing PC nouns in Kannada- koopa ‘anger’, santosha ‘happiness’, sukha ‘contentment’, dukha ‘sadness’, bhaara ‘heaviness’, bhaya ‘fear’, hagura ‘lightness’, sulabha ‘ease’, kaѻҀa ‘difficulty’, teѥuvu ‘thinness’, kobbu ‘arrogance’ and beesara ‘sadness/sulk’. We have at the end of these tests a clear two-way distinction of PCs in Kannada. One group represented by oѥѥeya ‘good’ shows properties of being ‘true’ adjectives, and these are few in number. Let us treat them as a closed, functional class and refer to them as the oѥѥeya group. The other group is the koopa group of nouns. Of the three groups of PCs proposed in this paper, the oѥѥeya group of PC adjectives is the only group to whom it is hard to assign a semantic basis (as can be seen by the list in 1).
5
‘anger’s words’ is not acceptable in English in the intended sense. It maybe paraphrased as ‘words of anger’. Cross-linguistically it is a well-attested fact that possession is expressed with of. English employs two strategies to express possession. One is the genitive case marker ’sand the other is with the genitive PP of. In English, ‘yesterday’s lecture’ expresses a PC. Kannada too has such a construction ‘nenne-ya kacheri’ (yesterday-GEN concert) though it does not have the of-construction. This is a fertile ground to explore in Kannada but it is beyond the scope of this paper.
Property Concepts and the Degree Expression in Kannada
17
2.2 Property Concept Expressions in the Predicative Position This section explores the predication structures of PC adjectives and PC nouns in Kannada. PC adjectives occur in ‘verbless’ copular clauses with an NP NP structure, in predication (discussed in sub-section 2.2.1). The PC nouns split into the koopa group of ‘true-blooded’ nouns which occur predicatively only in dative subject constructions in their bare form and the udda group of PCs. The udda group of PC nouns, exemplified by udda ‘height’, occur as dative-case marked nouns in nominative subject constructions in addition to their ability to occur as bare nouns in dative subject constructions. The former is less marked (unlike general perception about this construction) and can be understood on the lines of John is tall while the latter is more marked and would translate into John has (the) height. Thus the udda group of PCs have properties that make them ‘adjective-like’ as well as ‘noun-like’. The udda group of PCs denote tangible, physical properties such as height, weight, thickness, temperature etc. The koopa group of PC nouns represent psycho-somatic properties. Therefore, there is a very clear semantic basis for this sub-classification of PC nouns in Kannada. 2.2.1 Property Concept Adjectives in the Predicative Position Kannada has been noted to have, at least superficially, verbless clauses with an NP NP structure (Amritavalli 2000; 11; ex I (i)). They are also referred to as ‘nominal clauses’. Hence, a sentence like ‘John is a doctor’ in Kannada has only an NP NP structure as illustrated in (6). (6) John-ø John-NOM
doctor-u. doctor-NOM
‘John is a doctor.’ Oѥѥeya ‘good’ occurs predicatively in such clauses as seen below in (7), followed by a morpheme that encodes number and gender agreement with the subject NP and an invariant 3rd person feature. (7) avan-u oܿܿeya-avanu. / huܩugi-yar-u oܿܿeya-avanu/avaܿu.
oܿܿeya-avaru. / naan-u
Chapter Two
18
he-NOM good-he6/ girl-PL-NOM good-they / I-NOM
good-he/she
‘He is a good person.’ /‘The girls are good people.’ / ‘I am a good person.’ It is also important to note that oѥѥeya cannot occur predicatively in its ‘bare’ form. The construction in (8) below crashes without a suffixal morpheme for the PC adjective oѥѥeya. (8) *avan-u
oܿܿeya
he-NOM good ‘Intended: He is good.’ Keeping the above data in mind, various alternatives are suggested in the following section on the status of these morphemes and the structure of the clause in question. 2.2.1.1 An Analysis of the Predication Structure of Property Concept Adjectives The facts in (7-8) lead us to two ways of analysing what appears to be ‘agreement’, in Kannada, for adjectives in the predicative position. The avanu, avaѥu and avaru morphemes can be analysed as below, either as adjectival agreement (as in 9) or a (pro)nominal (as in 10). At the phrasal level, the two analyses predict different complements to the ‘verbless’ copular clause in Kannada. (9) [[NP avanu] [AP [AgrP avanu] [AP oܿܿeya]]] -- NP AP clause structure (10) [[NP avanu] [NP [FP oܿܿeya] [NP avanu]]] -- NP NP clause structure According to (9), these morphemes would then be analysed as some sort of adjectival agreement in the predicative position (though they do not show agreement in the attributive position). The second analysis (10) would predict an NP NP clause structure where oѥѥeya would be treated as a ‘nominalised adjective’; the avanu morpheme above would be some sort
6 Based on the choice of treatment, the suffixal avanu, avaѥu and avaru can be glossed in two ways. If we treat them as pronouns, we can gloss them as ‘he’, ‘she’ and ‘they’. However, if we choose to treat them as AGR morphemes, we have to gloss them as ‘3MSG / 3FSG / 3PL’. Further along the discussion, I choose to treat them as a dummy pronominal. Hence, the gloss in (7).
Property Concepts and the Degree Expression in Kannada
19
of dummy pronominal head in the sense of Cinque’s (2010) and Baker’s (2008) analyses. Pursuing the analysis in (9) is problematic on many accounts. Primarily, as described in Baker (2008; 61-62), a general rule regarding adjectives is that if in a given language adjectives do not take agreement in the attributive position they do not take agreement in the predicative position either. The better approach would be to analyse avanu, avaѥu and avaru as dummy pronominal heads, as PC adjectives in Kannada cannot occur ‘bare’ in the predicative position. This is the more straight-forward of the two analyses as it is well-known that these morphemes (avanu, avaѥu and avaru) are subject pronominals across Dravidian. In anticipation of the upcoming discussion on the distribution of heccu in comparative constructions, it is interesting to note that the predicative structures of the oѥѥeya group of adjectives do not take an obligatory Q. Hence a clause like avanu oѥѥeya-avanu ‘He good-he’ (as in 7) is interpretable as ‘He is a good man’. The analysis I propose is that oѥѥeya, a functional adjective, is hosted in specifier position of the functional phrase following Cinque (2010; 25; chapter 3; ex 1). This is illustrated in the representation below in (11). Further, oѥѥeya being an adjective, the functional phrase FAP can be directly selected by a DegP. This explains why the clause type under discussion does not require heccu in comparative constructions. When we treat oѥѥeya as an adjective that encodes degree semantics, our data gets a ready explanation. The head of the FAP, the functional phrase hosting the adjective, cannot be empty in Dravidian (for a reason that is unclear at the moment), and hence, we can posit that either the dummy pronominal is inserted here or it moves up from the complement NP. (11)
Chapter Two
20
In the positive degree constructions of the oѥѥeya group of adjectives in (7), the clause avanu oѥѥeya-avanu ‘He good-he’, for example, we said is interpreted as ‘He is a good man’. In comparative degree construction of the same sentence (which we shall see further on in (19)), it is ‘He (x) is a better man than him (y)’. The degree head being null in Kannada, and heccu not being obligatory for comparatives with the oѥѥeya group of PC adjectives, the only addition to the clause structure in (7) is the introduction of the standard of comparison. Thus, in Kannada the comparative construction for the oѥѥeya group of adjectives is literally ‘He is a good man than him.’
2.2.2 Property Concept Nouns in the Predicative Position PC nouns in Kannada occur predicatively in two kinds of copular constructions; one in which the subject is nominative with the PC noun taking dative case and the other in which the subject is dative case-marked and the PC noun occurs in its ‘bare’ form. The udda group of PCs can occur in both these types of constructions, in the former as udda-kke and in the latter as the bare form udda. This is illustrated in the data below in (12a-b). Re-iterating the claim made in the beginning, we shall treat uddakke as a syntactically-derived adjective. (12) a. raama-ø udda-kke idd-aane7[cf. ex 21 and 22 from A & J (2003)] Raama(nom.)
height-DAT
be-3MSG
‘Raama is tall.’ b. raaman-ige Raama-DAT
udda height
id-e be-3N
‘Raama has the height.’
7
This construction is attested in my variety of Kannada and may not be in other varieties of the language. However (as far as my knowledge goes) it is not attested in Tamil, Malayalam, Telugu or Tulu. Also, I would like to thank Ashwini Deo for bringing it to my notice (p.c) that this pair (12a-b) exists in Marathi too. a. Raam unchi-laa aahe. / b. Raam-laa unchi aahe. Ram-NOM height-DAT be-3.sg / Raam-DAT height be-3.sg ‘Ram is tall.’ / ‘Ram has the height.’
Property Concepts and the Degree Expression in Kannada
21
To distinguish these two types of constructions in (12a-b), we can draw a parallel to their respective counterparts in English: Raama is tall and Raama has the height. Just as in English, in Kannada too, Raama has the height is the more marked of the two constructions. However, with the right context, the construction in (12b) is fully acceptable just as it would be in English; for example, in a context like Raama has the height (to join the basketball team). The koopa group of nouns on the other hand occur only in dative subject constructions as in (13b) and cannot occur in constructions such as (13a) with a nominative subject. (13) a. *raama-ø Rama(nom.)
koopa-kke
idd-aane
anger-DAT
be-3MSG
‘intended: Rama is angry.’ b. raama-ige Rama-DAT
koopa
id-e.
anger
be-3NSG
‘Rama has anger.’ Thus, when we compare the predication structures of the udda group in (12) with the koopa group in (13), the motivation for a sub classification of PC nouns in Kannada becomes clear. They result in two groups, the koopa group and the udda group. Such an analysis is strengthened by the fact there is a strong semantic basis to this sub-division among PC nouns in Kannada. As the list in (5) suggests, the koopa group represent psychosomaticproperties such as happiness, sadness, contentment etc. The members of the udda group are listed below in (14) and represent tangible, physical properties such as height, weight, thickness etc. (14) Members of the udda(kke) group of PCs in Kannada- udda ‘height’, saѫѫa ‘thinness’, dappa ‘fatness/thickness’, yetra ‘height’, kuѥѥa ‘shortness’, agala ‘width’, nuѫѫa ‘smoothness’, bisi ‘hotness’, taѫѫa ‘coldness’, mett ‘softness’, gaҀҀi ‘hardness’ etc. Thus, we have arrived at a three-way classification of PCs in Kannada: the oѥѥeya group of functional adjectives, the udda-kke group of syntactically derived adjectives and the koopa group of ‘true’ nouns. Let us now turn
Chapter Two
22
our attention to degree expressions in Kannada and how this three-way classification plays out.
3. The Comparative in Kannada 3.1 An Overview In this section, I examine the instantiation of the three groups of PCs in the comparative degree construction in Kannada. The section also considers crosslinguistic data in English, Telugu and Hindi comparatives to plot a pattern. In this paper I shall assume that all adjectives inherently possess the property of gradability8. Thus I begin with the assumption that a degree head can select for only a gradable predicate (von Stechow 1984; Abney 1987; Kennedy 1997). By gradable predicate, I mean either an AP or a QP. Kennedy (2007; pg4; ex 9) sums up the idea of a gradable predicate in this definition: ‘Gradable predicates map objects onto abstract representations of measurement (SCALES) formalised as a set of values (DEGREES) ordered along some dimension (HEIGHT, WEIGHT, LENGTH etc.).’ For expository purposes, let us consider the three kinds of degree expressions: the positive (also known as absolute) degree construction as in (15a), the comparative degree construction as in (15b), and the superlative degree construction as in (15c) with the intention of ultimately motivating a common structure for the three. (15) a. John is tall. b. John is taller than Bill. c. John is the tallest.
8
This is the standard view on adjectives in the literature. There are various contesting views, one of which is that adjectives are vague predicates and hence a DegP is not involved. However, I do not follow this view here. In addition, Kennedy (1997) mentions adjectives such as dead, octagonal and former as nongradable adjectives. These examples itself are problematic because dead and octagonal can be attributive or predicative but former is purely an attributive adjective. The tests to check for non-gradability include: the adjective cannot take intensifiers or occur in comparative constructions and thus we cannot have *John is very dead or *John is more dead than Bill. In this sense, none of the fifty PCs I have worked on are non-gradable. In Kannada phrases like dead frog (e.g. satta kappe) clearly have relative clause structures.
Property Concepts and the Degree Expression in Kannada
23
3.2 Degree Expression in Kannada Let us begin by listing the degree expressions in Kannada. These sentences using the PC adjective udda-kke ‘height-DAT’9 in (16) below are counterparts for the English data in (15). (16) a. John-ø John-NOM
udda-kke
idd-aane.
height-DAT
be-3MSG
‘John is tall.’ b. John-ø John-NOM
[Bill-ig-inta]
udda-kke
Bill-DAT-Post.P10
idd-aane.
height-DAT
be-3MSG
‘John is taller than Bill.’ (Lit. John is tall than Bill.) c. John-ø John-NOM
[yellar-ig-inta] everyone-DAT-Post.P
udda-kke
idd-aane.
height-DAT
be-3MSG
‘John is the tallest.’ (Lit. John is tall than everyone.) In the Kannada comparative construction, the standard of comparison is introduced by the post-positional morpheme –inta. This is indicated using bracketing in the data above. Let us refer to this complement of the matrix clause as the –inta phrase or the standard phrase (terminology adopted from Schwarzschild 2010). Two important aspects of the above data are striking. One, Kannada does not have a distinct superlative construction at all as seen when we compare (16b) to (16c) above. Looking at the paradigm in the degree expressions in
9 An anonymous reviewer suggested that another PC, other than udda-kke ‘heightDAT’, be used in (16) above where I lay out the three types of degree constructions in Kannada. This was suggested as the reviewer felt that these constructions with udda-kke are marked in Kannada. May I bring to the attention of the reader that dative-case marked nouns such as udda-kke are not marked but part of the standard Mysuru-Bengaluru variety of Kannada. 10 -inta is referred to as a comparative post-position in descriptive grammars of Kannada (Sridhar 1990; 126, 167-168). I have chosen to simply gloss it as a postposition (PostP.).
24
Chapter Two
(16), we notice that the comparative construction in (16b) and the superlative construction in (16c) are the same. In general, superlative constructions predicate over an entire set as a standard of comparison. The overt realisation of yellaru ‘everyone’, as the standard of comparison in superlative constructions in Kannada, is evidence that superlatives indeed predicate over an entire set. The superlative in a language like Kannada is, thus, expressed by the comparative construction. Thus my analysis for the comparative clause in Kannada will encompass the superlative clause too. Two, Kannada does not have a morphological comparative or superlative marker such as the –er and –est morphemes as seen in (16b) and (16c). Keeping in mind the assumption that constructions with degree expressions, be it a positive, comparative or superlative degree expression, contain a degree phrase headed by a degree head, Kannada seems to have a null element heading the degree phrase.
3.3 The Degree Phrase: Important Ideas from Literature The evolution of the degree phrase in the syntax of the comparative construction is interesting. Bresnan (1973), one of the earliest works in this regard, identified and separated the degree head (which she referred to as a ‘determiner’) from a Q head like much. The shift from treating them as determiner-like elements to treating them as degree heads happened with Jackendoff (1977) and then Abney (1987) and Kennedy (1997) among others. Bresnan’s (1973) and Jackendoff’s (1977) analysis was what later came to be referred to as the lexical head approach to the AP with the functional phrase in the [Spec, AP] position. Abney (1987) was one of the first to propose a functional head approach to the extended projection of the AP with the DegP heading the AP. There were thus two major changes in the generative work on APs and comparative constructions which still hold: one being a shift from treating them as ‘determiners’ to treating them as ‘degree’ heads; and two, a shift from a lexical head to a functional head structure. Abney (1987) appears to be one of the first to clearly formalise the relationship between the DegP and the AP in syntactic structure stating, ‘analysing adjective phrases as DegPs’ (Abney 1987; pg189). The work also states clearly that since adjectives are headed by the degree phrase, the degree head can be empty. Abney (1987; 192): ‘If adjective phrases are consistently DegPs, even in cases such as these. The appearance of an empty Deg is not disturbing- I assume an empty degree in all adjective phrases consisting just of an adjective, just as I assume an empty D in all
Property Concepts and the Degree Expression in Kannada
25
noun phrases without an overt determiner. What requires explanation is why an overt degree can appear with these adjectives’. Abney treats quantifiers and adverbs as sub-classes of adjectives differentiating them from adjectives by a feature based system of ±Q and ±Adv. The structure posited in Abney (1987; pg 194; ex 339) is given in (17). I do not follow such an approach in my analysis and provide the representation below for the purposes of the current discussion and an understanding of the background literature. (17a)
(17b)
Kennedy’s (1997; pg 147) proposal emphasised the gradability aspect of adjectives in English as a scalar system based on intervals or extents: ‘Specifically I assume that the extended projection of A is headed by a degree morpheme, i.e., a member of {ø, er/more, less, as, so, too, enough, how, this, that} (where ø is a phonologically null morpheme associated with the positive degree construction)’. More recently, the doctoral dissertation of Bacskai-Atkari (2014:46; chapter 2) also posits a null degree head in positive or absolute degree constructions, and I quote, ‘... in absolute constructions like Mary is tall (for a schoolgirl), the Deg head takes a PP complement (for a schoolgirl) and the Deg head itself is a zero...’. What is crucial to note is that the degree head always indicates greater than semantics. In fact, the degree head in any degree construction always indicates greater than semantics. Hence we need an overt Q to encode less than semantics. Thus, for a comparative or a superlative clause to express less than semantics, a Q head expressing the same becomes crucial. (18) a. John is tall. (always indicates greater than semantics) b. John is taller than Bill. / John is less tall than Bill. c. John is the tallest. / John is the least tall.
Chapter Two
26
These facts bring to the forefront the significance of motivating a DegP and QP layer in comparative clauses.
3.4 The Instantiation of heccu in the Three Groups of Property Concept Comparatives In this sub-section, we will look at the comparative constructions for the three groups of PCs: the oѥѥeya-group, the udda-group and the koopagroup. In addition, this subsection also elaborates on the idea of a null degree head in Kannada and introduces the Q-element heccu glossed as just ‘Q’. (19) John-ø John-NOM
[Bill-ig-inta]
(heccu)
Bill-DAT-PostP.
Q
oܿܿeya-avanu. good-he
‘John is (much) better than Bill.’ (Lit. John is (much) good than Bill.) (20) John-ø John-NOM
[Bill-ig-inta] Bill-DAT-Post.P
(heccu)
udda-kke
idd-aane.
Q
height-DAT
be-3MSG
‘John is (much) taller than Bill.’ (Lit. John is (much) tall than Bill.) (21) John-ige
[Bill-ig-inta]
John-DAT Bill-DAT-PostP.
*(heccu) koopa Q
anger
id-e. be-3NSG
‘John has more anger than Bill.’ Let us consider the data from (19-21). The sentences in (19) and (20) with oѥѥeya and udda-kke are adjectival comparatives. They do not require an obligatory heccu. Koopa, being a ‘true-blooded’ noun, requires heccu obligatorily as seen in (21). A discussion on the ‘obligatoriness’ of a Q element in nominal comparatives, as a robust cross-linguistic generalisation, follows shortly. We now have a neat piece of further evidence with the distributional properties of heccu supporting the threeway classification of PCs in Kannada.
Property Concepts and the Degree Expression in Kannada
27
3.5 The Distribution of heccu: Indicative of a Cross-linguistically Robust Pattern? In Kannada nominal and verbal comparatives, heccu is obligatory. The data below demonstrates this point taking comparative clauses with a mass noun neeru ‘water’ in (22), a count noun makkaѥu ‘children’ in (23) and a verbal comparative in (24). (22) John-ige Bill-ig-inta *(heccu) neeru id-e. / John-ø hattira John-DAT Bill-DAT-PostP.Q water be-3NSG / John-NOM near Bill-ig-inta *(heccu)neeru id-e. Bill-DAT-PostQ water be-3NSG ‘John has more water than Bill.’ (23) John-ige
Bill-ig-inta
*(heccu) makkaܿu idd-aare.
John-DAT Bill-DAT-PostP.
Q
children
be-3PL
‘John has more children than Bill.’ (24) John-ø John-NOM
Bill-ig-inta Bill-DAT-PostP.
*(heccu) Q
aaܩ-utt-aane. play-PROG-3MSG
‘John plays more than Bill.’ Re-stating the facts so far, we can say that heccu is obligatory in nominal (be it a PC nominal, a mass or a count as seen in 21, 22 and 23 respectively) and verbal comparatives (24), and is optional in comparative clauses employing PC adjectives (19 and 20). In Bhatt (2012), Hindi-Urdu has been noted to have adjectival comparatives which do not have any overt degree head or any sort of obligatory Q element (ref to examples 27 a-b). Bhatt (2012) refers to such comparatives as ‘bare comparatives’, a term he owes to Schwarzschild (2010). Let us understand the term ‘bare comparative’. The structure such as the one below in (25) for English represents an overt realisation of the
28
Chapter Two
degree head, while the Japanese comparative in (26) exemplifies what Schwarzschild proposes is a ‘bare comparative’ or a ‘bare (comparative) AP’, i.e. a comparative without a comparative marker. Thus, a bare comparative may refer to a comparative that has no degree phrase or has a degree phrase whose head is null. (25) DegP
DegP
AP
more anxious [Schwarzschild 2010] (26) Taroo-wa Taroo-TOP
[Hanako-yori(mo)] kasikoi. Hanako-than
smart
‘Taroo is smarter than Hanako.’ [ex. 3 in Schwarzschild 2011 taken from data in Bhatt & Takahashi 2011] In reference to defining a ‘bare comparative’ with the Japanese example above, Schwarzschild (2010; pg 89) adds, and I quote here, ‘Whether or not we think there is an empty comparative marker (in the Japanese example) is a matter of analysis, on par with the question of whether some bare plurals have empty determiners or not. And one can assert that a given language has no DPs and still use the term bare plural to describe noun phrases without determiners. Likewise, we may eventually want an analysis of the Japanese AP comparative that makes no use of the Degree Phrase and still we will call (26) a bare comparative.’ What I think Schwarzschild means to convey is that the lack of an overt realisation of a functional head in a particular language does not mean that the functional architecture is not present at all in that language. Thus the idea of a bare comparative is born out of the need to have a universal structure cross-
Property Concepts and the Degree Expression in Kannada
29
linguistically for comparatives in tune with the functional head approach to phrase structure as illustrated in (25) earlier. The Kannada data in (19-20) thus patterns with the Hindi-Urdu data below in (27a-b). We can safely assume a null degree head in both these languages. (27) a. Mina Atif-se (zyaadaa) lambii Mina.F
Atif-Instr.more
hai. [Bhatt 2012; ex 1a and b]
tall.F
be.Prs.Sg
‘Mina is taller than Atif.’ b. Mina Mina.F
Atif-se
(zyaadaa)
Atif-Instr.
bahaadur
more
brave
hai. be.Prs.Sg
‘Mina is braver than Atif.’ However non-adjectival comparatives (nominal and verbal comparatives), as seen below in (27c) which is a nominal comparative, require zyaadaa. In this sense too Kannada patterns with Hindi-Urdu in the sense that nominal and verbal comparatives cannot be ‘bare’. We have already seen this in (21-24) above. c. Mina-ne Atif-se *(zyaadaa) kitaabӁ par.h-ƭ: [Bhatt 2012; ex 6a] Mina-Erg Atif-Instr
more
books
read-Pfv.FPl
‘Mina read more books than Atif.’ These facts hold true for comparative constructions in English as well wherein verbal and nominal comparatives cannot occur without more. Balusu (2015) illustrates the same for Telugu nominal and PC nominal comparatives which take an obligatory ekkuva, a quantifier element. There is clearly a pattern here. (28) a. naa-ku nii-kanTe *(ekkuva) I-DAT
you-than
*(Q)
‘I have more rice than you.’
biyyam
undi.
rice
is
Chapter Two
30
b. naa-ku I-DAT
nii-kanTe *(ekkuva)
carlu
you-than
cars
*(Q)
unnaayi. are
‘I have more cars than you.’ c. naa-ku I-DAT
nii-kanTe you-than
*(ekkuva) *(Q)
koopam
undi.
anger
is
‘I have more anger than you.’ d. neenu I-NOM
nii-kanTe you-than
(*ekkuva) (*Q)
erupu. redness
‘I am redder than you.’ e. naa-ku nii-kanTe *(ekkuva) ettu (*ekkuva) ettu.
undi / neenu
nii-kanTe
I-DAT you-than *(Q) height is / I-NOM you-than (*Q)
height
‘I am taller than you.’ [Balusu (2015); ex 24, 25, 26 & 27] Let us now elaborate on an important assumption made in the discussion so far, which is that nominal comparatives unlike adjectival comparatives require an obligatory QP layer to participate in comparison. Based on this understanding, we can now posit two separate structures for adjectives and nominals in comparative clauses. The AP comes with a DegP heading it, while an NP comes with a QP heading it.
Property Concepts and the Degree Expression in Kannada
31
(29)
In adjectival comparatives in Kannada (and in Hindi-Urdu and Telugu), when the Q element is optional, it provides a big differential reading11 and I suggest that such a QP be posited above the DegP in (29a). In the above schematic representations, we have looked at how the extended domain of an AP and NP (in (29a) and (29b) respectively) vary in a comparative clause.
3.6 An Analysis of heccu Heccu is clearly a Q head and not a degree head. In nominal comparatives in Kannada, be it a PC noun like koopa or any other noun, heccu comes in to ‘rescue’ the construction as the DegP cannot directly select an NP as its complement. The QP heccu, which we can then consider on the lines of much + ø degree head in nominal comparatives in Kannada, provides the needful semantics to keep the nominal comparative clause structure from crashing. We of course do not have an overt proof of such a merger in Kannada between the degree head and the Q, unlike the English more which is putatively –er + much (Bresnan 1973). Thus the clear contrast seen in English between much/many and more does not find a parallel in Kannada as the degree head is always null. But the data makes it clear that we do not really have to doubt heccu being a Q head combining with a null degree head and thus remaining heccu. An evidence to support this line of thought comes from the fact that heccu can occur in non-comparative constructions to simply mean ‘much’, ‘many’ or ‘lots of’.
11 Bhatia, Iyer and Kaur (2013; pg 8-9) claim zyaadaa provides a big differential reading in adjectival comparatives in Hindi-Urdu where zyaadaa is optional.
Chapter Two
32
(30) a. John-ige heccu kelasa id-e. / John hattira heccu ha݅a
id-e.
John-DAT Q work be-3NSG /John near Q money be-3NSG ‘John has lots of work. / John has lots of money.’ b. John-ige John-DAT
heccu
makkaܿu
idd-aare.
Q
children
be-3PL
‘John has many children.’ c. John-ige John-DAT
heccu koopa Q
anger
ide. be-3NSG
‘John has much anger (about something).’ The answer seems straightforward. The degree head in Kannada being null irrespective of the kind of degree expression, heccu reflects no change in form. The difference between nominal and PC adjectival comparatives then boils down to these sentences. (31) a. John is (much) taller than Bill. (Ref. John Bill-ig-inta (heccu) udda-kke iddaane.) b. John has *(more) anger than Bill. (Ref. John-ige Bill-ig-inta *(heccu) koopa ide.) The data in (31) suggests that heccu is not a degree word like more. Heccu is only a quantifier which introduces a measure function needed to map a nominal or a verbal head onto a scale. It should be quite clear by now that nouns and verbs do not inherently possess degree semantics, and hence combining with heccu provides degree semantics licensed by the –intaP in comparative constructions. It should be noted that the data in (30) illustrates that heccu combined with nouns results in a non-comparative reading or in other words only a quantitative-reading. Thus, heccu in the data in (30) is only much/lots of and not more. By positing a null degree head for comparatives, such a degree head’s selection for a QP headed by heccu in nominal and verbal comparatives results in heccu remaining heccu. In (32) below, the null degree head in Kannada is presented by ø. (32) ø + heccu = heccu
Property Concepts and the Degree Expression in Kannada
33
4. Conclusion The paper aimed to combine research on Property Concept (PC) expressions in Kannada with the comparative degree construction. In the first part of this paper, I motivated a three-way classification of PCs which included both adjectives and nouns in Kannada. The attributive position in NPs in Kannada revealed a broad classification of PCs into PC adjectives and PC nouns (the oѥѥeya and the koopa groups respectively). The predicative position revealed a further divide in PC nouns, resulting in the udda-kke group of dative case-marked, syntactically-derived adjectives. For the predicative structure of oѥѥeya and the group of adjectives it represents, we proposed a Cinque (2010) style analysis wherein oѥѥeya was posited in [Spec, FAP], of a functional projection, which in turn selected a NP complement. Such a functional phrase could then be selected by a DegP in comparative constructions. Thus the oѥѥeya group of functional adjectives did not require an obligatory heccu in comparative clauses. Part two of the paper began with a brief introduction to the various kinds of degree constructions. Having noted that the comparative and the superlative constructions are one and the same for languages like Kannada, our analysis needed only to account for the positive and the comparative degree constructions. Kannada did not have any overt realisation of a degree head. The distribution of heccu in comparatives in Kannada revealed that heccu is only a Q element and not a degree word. The koopa group of PC nouns needed an obligatory heccu in comparative clauses whereas the oѥѥeya group of functional adjectives and the udda-kke group of syntactically-derived adjectives did not. Thus, the distribution facts of heccu in comparative constructions supported the three-way classification of PCs in Kannada and provided support for such a categorial classification. A brief comparison of data and relevant literature of comparatives in English, Hindi-Urdu and Telugu revealed an interesting generalisation. Nominal and verbal comparatives require a Q element cross-linguistically to participate in comparative constructions whereas adjectival comparatives do not. Thus, the observation I would like to emphasise through this paper is that the DegP always selects for a gradable predicate. Such a predicate could either be an AP, which I assume has inherent gradable properties, or a QP, for lexical heads such as nouns and verbs which do not have inherent gradable properties.
34
Chapter Two
References Abney, Steven. The English noun phrase in its sentential aspect. (PhD dissertation, Massachusetts Institute of Technology, 1987). Amritavalli, Raghavachari. “Kannada Clause Structure” in The Yearbook of South Asian Language and Linguistics, ed. R. Singh et al. (New Delhi: Sage Publications, 2000), 11-30. Amritavalli, Raghavachari and KA Jayaseelan. “The genesis of syntactic categories and parametric variation” in Generative Grammar in a Broader Perspective: Proceedings of the 4th GLOW in Asia, ed. HangJin Yoon (Seoul: Hancook, 2003), 19- 41. Bacskai-Atkari, Julia. The Syntax of Comparative Constructions: Operators, Ellipsis Phenomena and Functional Left Peripheries. (PhD Dissertation, Universitätsverlag Potsdam, 2014). Baker, Mark. The Syntax of Agreement and Concord. (Cambridge: Cambridge University Press, 2008). Balusu, Rahul. Comparison, predication and lexical semantics of PC nouns in Telugu. Manuscript, 2015. Bhatia, Sakshi, Jyoti Iyer and Gurmeet Kaur. “Comparatives in HindiUrdu: Puzzling over zyaadaa”. Lissim Working Papers 1 (2013): 1528. Bhatt, Rajesh. Many or more: the Hindi-Urdu degree word zyaadaa and the analysis of Bare Comparatives. Manuscript. UMass, Amherst. Bhatt, Rajesh and Shoichi Takahashi.“Reduced and Unreduced Phrasal Comparatives.” Natural Language and Linguistic Theory29 (2011): 581-620. Bresnan, Joan.“The Syntax of the Comparative Clause Construction in English.” Linguistic Inquiry 4 (1973): 275–343. Cinque, Guglielmo.The Syntax of Adjectives- A comparative study. (Cambridge: The MIT Press, 2010). Dixon, Robert M.W. Where have all the adjectives gone? And other essays in semantics and syntax. (Berlin: Mouton, 1982). Kennedy, Christopher. Projecting the adjective: The syntax and semantics of gradability and comparison. (PhD Dissertation.University of California, Santa Cruz, 1997). Kennedy, Christopher. Modes of Comparison. In Papers from the 43rd Annual Meeting of the Chicago Linguistic Society; Volume 1: The Main Session, ed. Malcolm Elliott, James Kirby, Osamu Sawada, Eleni Staraki and Suwon Yoon. (Chicago: Chicago Linguistic Society, 2007b), 139-163.
Property Concepts and the Degree Expression in Kannada
35
Menon, Mythili. The Apparent Lack of Adjectival Category in Malayalam and other Related Languages. In Proceedings of Glow-in-Asia IX 2012: The Main Session, ed. Nobu Goto, Koichi Otaki, Atsushi Sato, Kensuke Takita (Mie University, Japan, 2013), 157-171. Menon, Mythili and Roumyana Pancheva.The grammatical life of property concept roots in Malayalam. In Proceedings of Sinn und Bedeutung, ed. U. Etxeberria, A. Fălău܈, A. Irurtzun, B. Leferman (2013), 289302. Schwarzschild, Roger. Comparative Markers and Standard Markers. In Proceedings of the MIT Workshop on Comparatives, ed. Michael Y. Erlewine and Yasutada Sudo (MIT: MIT Working Papers in Linguistics 69, 2010), 87-105. Sridhar, Shikaripur. Kannada.(London: Routledge, 1990). von Stechow, Arnim. “Comparing theories of comparison.” Journal of Semantics 3.1 (1984), 1-77.
CHAPTER THREE SEMANTICALLY ELABORATE CATEGORIES AND GRAMMATICALIZATION IN MALAYALAM1 RESHMA JACOB AND SHREYA MEHTA
Abstract Semantically elaborate categories (SEC) involve more than one domain simultaneously. For instance, a grammatical category could be related to aspectuality and temporality, and to negation/counterfactuality at the same time. In her paper Grammatical Categories and Linguistic Theory: Elaborateness in Grammar, Tania Kuteva argues grammatical structures can be elaborate not only with respect to their expression but also with respect to their semantics, and introduces the notion of ‘semantically elaborate grammatical categories’. She hypothesizes that the more semantically elaborate a notional category is, i.e. the more specificities and richness of information it involves, the less likely it is for this category to receive a grammaticalized expression across languages. Kuteva proposes five semantically elaborate categories—namely, Apprehensional, Avertives, Frustrated Initiation, Frustrated Completion and Inconsequential—where each of them shows counterfactuality or negates the realization of verb situation. Though all of these categories show counterfactuality, they all differ from each other in terms of the tense, aspect or mood, thus resulting in different realizations of the verb situation. This paper aims to propose and describe the semantically elaborate categories that are found to exist in Malayalam. We will also look into the grammaticalization process in these constructions. We will
1 1=First Person, 2= Second Person, 3=Third Person, ACC=Accusative, AUX=Auxiliary, AVRTV=Avertive, DAT=Dative, DEM=Demonstrative, EXCL=Exclusive, FEM= Feminine, FUT= Future, IMP= Imperative, INF= Infinitive, LOC= Locative, M=Masculine, NOM=Nominative, NEG=Negation, PERF=Perfect, PL=Plural, PROX=Proximitive, PST=Past, SG =Singular
Chapter Three
38
also show that criteria as proposed by Kuteva et al. are met in Malayalam and the same shall be dealt with in detail in the paper. Keywords: Categories.
Grammaticalization, Malayalam, Semantically Elaborate
1. Introduction According to Heine and Kuteva (2007), grammaticalization is “the development from lexical to grammatical forms, and from grammatical to even more grammatical forms”2. Grammaticalization is not just about the individual forms that undergo changes but also about the structure to which they belong. The notion of elaborateness was first introduced by Kuteva and Comrie (2001). Semantically elaborate categories is a recent phenomenon and differs from semantically straightforward categories(SSC). Unlike semantically straightforward categories, semantically elaborate categories involve more than one domain simultaneously. That is, while SSC belongs to one particular functional domain (present tense, past tense, deictic function), SEC contains one single linguistic unit that can show tense, aspect and mood simultaneously. Though all of these categories show counterfactuality, they all differ in tense, aspect or mood giving rise to different realizations of the verb situation. It is interesting to note that all the five categories, i.e. ‘Apprehensional’, ‘Avertive’, ‘Frustrated Initiation’, ‘Frustrated Completion’ and ‘Inconsequential’, can be seen in one language (though not necessarily). This means that despite the fact that they negate the verb situation, the foregrounded degree of verb situation realization and the change in tense, aspect and mood draw the line of separation. This paper is an attempt to look into the semantically elaborate categories in Malayalam, one of the Dravidian languages spoken in Kerala, the southern state of India.
2. Semantically Elaborate Categories in Malayalam Grammatical categories may encode one or more schematic meaning and may involve utilization of a domain for more than one function simultaneously. Tense-Aspect-Mood (TAM) “Grammar of Counter-to-
2
Heine, Bernd and Tania Kuteva.The Genesis of Grammar: A Reconstruction. New York: Oxford Press, 2007
Semantically Elaborate Categories and Grammaticalization in Malayalam 39
Fact” can be divided into at least five distinct sub-portions encoded by grammatical or lexico-grammatical linguistic structures—namely Apprehensional, Avertives, Frustrated Initiation, Frustrated Completion and Inconsequential—where each of them shows counterfactuality or negates the realization of verb situation (Kuteva 2009). We will also look into the “Lest Clause” and “Proximitive” constructions. In Malayalam, we could find all the five semantically elaborate categories which show counterfactuality.
2.1 Apprehensional Apprehensional can be defined as the counterfactuality of an undesirable verb situation. The term ‘apprehensional’ was used by Dixon in the year 1980 and he describes it as an inflexion which specifically marks the verb of a subordinate clause, and denotes an undesirable event which is to be avoided (Kuteva 2009). In an apprehensional construction, it is mandatory to have a minimum of two verbs, which means a verb situation X is featured as necessary for the avoidance of an undesirable verb situation Y. The apprehensional can be understood in terms of:3 x Counterfactuality of foregrounded degree of verb situation realization x Foregrounded degree of verb situation realization: full x resulting degree of verb situation realization: zero x Causality and Undesirability of the verb situation. The apprehensional constructions in Malayalam are given below: 1. ܤvܤn 3M.SG
vi:ݚݚil
po:k-ܤ:tࡧ e
݄ܤ:n
house.LOC go-without 1.SG
nࡧ o:kki watch.PST
“I watched him so that he will not go home”. 2. ravi
viܤݕm kݍܤikk-ܤtࡧ e
Ravi-3M.SG poison eat-without
݄ܤ:n 1.SG
tࡧ ݄݄ܤܩܤu stop.PST
3
Presentation titled “Grammar of Counter-to-fact” presented by Tania Kuteva in JNU (January – April 2014)
Chapter Three
40
“I stopped Ravi from eating the poison”. Thus the apprehensional construction in Malayalam is bi-clausal, where there are two verbs. The verb of the subordinate clause is the undesirable verb situation which is attached by the suffix ‘ܤ:tࡧ e’ which means ‘without’. Moreover, the verb of the main clause stops the undesirable action which could have happened. It is also important to note that the distribution of the second verb is highly restricted; i.e. not every verb can be in the main clause. But only those verbs such as “watch”, “discourage”, “hide”, and “stop”, which have some sort of ‘preventive’ connotation, can be the V2. The apprehensional suffix ‘-ܤ:tࡧ e’ is derived from the word ‘ܤrutࡧ ԥ’ which is a negation4. For example, 3. kܤܿܿܤm paraj-ܤrutࡧ ԥ lie
say- NEG
“Do not lie”. 4. minu
ܤ:
kܤ:r
vܤ:ƾ-ܤ:tࡧ e
݄ܤƾܿܤ
nࡧ irulsܤ:hܤpeܩutࡧ tࡧ i
minu that.DEM car buy-without 1.pl.EXCL discourage.PST “We discouraged Minu from buying that car”. Both the source word and the target suffix co-exist in the language, but they are used in different contexts.
2.2 Lest Clause Lest-clause is another semantically elaborate category which has all the features similar to apprehensional. The term was coined by Austin in 1981. “Lest-clauses basically serve to indicate some situation which the speaker considers to be unpleasant and which should be avoided. They are most commonly used as admonitions or threats but they are also used in
4
Rajaraja Varma, A. R. (11970/1895) Keerala paaniniiyam. NBS, Kottayam.
Semantically Elaborate Categories and Grammaticalization in Malayalam 41
giving warnings”5. The lest-clause constructions that can be seen in Malayalam are: 5. irikkԥillenkil sit.IMP
nࡧ i:
if-not
vi:ݍ.um 2.sg
fall.FUT
“Sit down or you will fall”. 6. satࡧ jܤm pܤrܤ truth tell.IMP
illenkil ninne
݄ܤ:n
kollum
if-not 2.sg.Acc 1.sg.NOM kill.Fut
“Tell the truth lest I will kill you”. The lest-clause construction in Malayalam is also bi-clausal, where the first verb is marked for Imperative mood. The formation of the word ‘illenkil’ meaning ‘lest’, is also interesting. It is a word formed of two different morphemes ‘illa’ meaning “no” and ‘enkil’ meaning ‘if. . .then’. Thus the literal translation of the word ‘illenkil’ would be ‘if not, then’. The lest-clause construction is counterfactual such that it shows causality, undesirability of verb situation and the resulting degree of verb situation realization is zero.
2.3 Avertive Avertive can be defined as the “counterfactuality of realization of imminent, past verb situation where the verb-situation is viewed as a whole.” (Kuteva 2009). It is essentially an aspectual gram, where imminence is the essential requirement. Precisely, it is a linguistic expression used only in past contexts, standing for a verb situation which was on the verge of taking place but did not take place. Non precipitatives (Abbi 1980), a bi-clausal construction, deliver the similar meaning in most of the Indo-Aryan languages. Avertives can be understood in terms of 6:
• Counterfactuality of foregrounded degree of verb situation realization
5
Kuteva Tania. Grammatical categories and Linguistic Theory: Elaborateness in Grammar. In Proceedings of Conference on Language Documentation and Linguistic Theory 2, edited by Peter K. Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells (eds). 13-28. London: SOAS, 2009 6 Kuteva (2009) ibid
Chapter Three
42
• • • • •
Foregrounded degree of verb situation realization: full Resulting degree of verb situation realization: Zero Imminence Pastness Perfectivity (in those languages which perfectivity/imperfectivity distinction).
have
a
The avertive constructions in Malayalam are: 7. ݄ܤ:n ippo vi:݅-e:ne 1SG now fall-AVRTV “I nearly fell down”. 8. enne 1sg-ACC
inࡧ nࡧ ܤle
oru bԥs iܩicc-e:ne
yesterday one bus hit-AVRTV
“I almost got hit by a bus, yesterday”. The avertive construction in Malayalam is mono-clausal. The avertive meaning can be attained by suffixing ‘-ene’ to the verb. In Malayalam, in order to get the avertive meaning, it is mandatory to have a temporal adverb in the sentence. However, if the sentence lacks the temporal adverb, the presence of the suffix ‘-ene’ would give the future perfect mood. For example, 9. ݄ܤ:n 1SG-NOM
po:j-e:ne go-FUT PERF
“I would have gone”. 10. mܩܤutࡧ -ill-ܤ:jirunࡧ nࡧ -enkil ݄ܤ:n tired- NEG-be-if
1.sg.NOM
muݍuvԥn eݍutࡧ ij-e:ne full write-FUT PERF
“Had I not been tired, I would have written everything.”
Semantically Elaborate Categories and Grammaticalization in Malayalam 43
2.4 Proximitive The proximitive can be defined as “a temporal phase located close before the initial boundary of the situation described by the main verb. It indicates a moment shortly before the possible occurrence of the given verbal situation, with no implication that the situation did not actually occur”7. Yet another essential characteristic of the proximitive is that it can be used in both past and non-past contexts. Like avertives, the proximative is a purely aspectual gram, its essential semantic characteristic being imminence. First, we will look into the non-past proximative. 11. mܤhi si:tࡧ ܤje
ܩܤikkܤ:n irikkuv.ܤ
Mahi sita.ACC beat.INF PROX.AUX “Mahi is about to beat to Sita”. 12. ܤvܿܤ
ܤtࡧ ԥ
pܤrܤjܤ:n irikkuv.ܤ
3.fem.sg that.DEM say.INF PROX.AUX “She is about to say that”. In these sentences, the verb situation has not yet started but is about to start very soon. The proximitive meaning of these sentences is given by the grammaticalized verb ‘irikku’ meaning ‘sit’. Moreover, the main verb should be in its infinitive form. It must be noted that the word “irikkuva” does not have its lexical meaning “sitting” in the above examples, but gets grammaticalized and gives the proximitive meaning. The verb /irikkuka/ meaning ‘to sit’ is a highly grammaticalized form in Malayalam that is extended to different contexts so as to get different meanings. Proximitive in past could be understood with the following examples, 13. ݄ܤ:n ܤƾo:ݚݚԥ po:kܤ:n irikku-va:jirunࡧ nࡧ u 1sg
there go.INF PROX –be. PST
“I was about to go”.
7
Kuteva (2009) ibid.
Chapter Three
44
14. mܤhi si:tࡧ ܤ-je
aܩikkܤ:n irikkuv-ܤ:jirunࡧ nࡧ u
Mahi sita.ACC beat.INF PROX-be.PST “Mahi was about to beat Sita.” These sentences are instances of the highly grammaticalized word “irikkuka” meaning ‘to sit’. In these sentences, the word “irikkuka” gives the proximative meaning along with the infinitive form of the main verb. As mentioned earlier, this is the grammaticalized form of the lexical verb ‘sit’. However, in the proximative construction in past, the auxiliary ‘be’ verb is followed by “irunࡧ nࡧ u” which is again a grammaticalized form of the lexical verb ‘sit.PST’ i.e., ‘sat’ so as to give the past meaning. Thus in the same sentence, the word “irkkuka” is grammaticalized in two different ways to give proximative and past meaning. The verb ‘sit’ in Malayalam is also grammaticalized to give the progressive aspect. However, since it does not come under proximative we will not look into it now. It is also important to note that when the main verb is ‘to sit’, then the proximative meaning is given by the grammaticalized form of the lexical word ‘go’, as shown in (15). 15. ݄ܤ:n
irikkܤ:n pokuv-ܤ:jirunࡧ nࡧ u
1.sg.NOM sit.INF go-be.PST “I was about to sit.” This means that the distribution of the word ‘sit’ is restricted in such a way that it cannot occur more than twice in the same sentence, as is illustrated in the following ungrammatical example in (16). 16. *݄ܤ:n
irikkܤ:n irikkuv-ܤjirunࡧ nࡧ u
1.sg.NOM sit.INF
sit-be.PST
The sentence with the lexical verb ‘irikkuka’ will not get the proximitive meaning by the grammaticalized form ‘irikku’ but by ‘po’ meaning ‘go’. However the past meaning is still given by the grammaticalized form of ‘sit.PST’/‘sat’. It is also important that the word “po” meaning ‘go’ is also used to give the past proximative construction. That is, main verb.INF + po + a:jirunࡧ nࡧ u. Similarly, words like ‘come’, ‘begin’ also give proximative meaning in certain contexts. Use of the proximative in the past makes it
Semantically Elaborate Categories and Grammaticalization in Malayalam 45
clear that what is highlighted by the proximative (even when used in the past) is the imminence of the verb situation, its being on the verge of taking place in the past, rather than its counterfactuality.
2.5 Frustrated Initiation Frustrated initiation is a “past verb situation which was about to begin but was frustrated before initiation; even though the verb situation itself might not have been initiated, some prior action indicates the possibility of that verb situation has occurred.” Precisely, a verb situation was about to be entered, and in fact the initial boundary of that situation was touched, but then it was prevented from entering it. The frustrated initiation can be understood in terms of: Counterfactuality of foregrounded degree of verb situation realization, Foregrounded degree of verb situation realization: Initiation, Resulting degree of verb situation realization: Zero, Imminence and Pastness. The frustrated initiation constructions in Malayalam are: 17. irܤcci morij-ܤ݅ܤm-ܤ:jirunࡧ nࡧ u Meat deep fry-should-be-PST “The meat should have been deep fried”. 18. veܿܿܤm tࡧ iܿܤjkk-ܤ݅ܤm-ܤjirunࡧ nࡧ u Water boil – should-be.PST “Water should have boiled.” In these sentences, the action of deep frying the meat and boiling of water did not take place. However, there was some action which happened. That is, the meat must have been fried but the action of deep frying did not take place. The action of ‘deep frying’ the meat has been frustrated before its initiation. Similarly, the water must have been made hot but the action of boiling did not take place. This structure also has two grammaticalized forms. The construction of the frustrated initiation in Malayalam is shown by the root form of the verb along with the modal suffix “-ܤ݅ܤm” and the grammaticalized form of the lexical verb ‘irunࡧ nࡧ u’ meaning ‘sat’ so as to
Chapter Three
46
give the past tense. Here the modal suffix ‘-ܤ݅ܤm’ is grammaticalized from the lexical word ‘ve:݅ܤm’ which means ‘want’.8
2.6 Frustrated Completion According to Kuteva, frustrated completion “is a past verb situation which had begun, but it could not be completed; hence the frequent use of the expression “unsuccessful conative” for the Frustrated Completion structure.”9 It is often expressed by a bi-clausal structure involving the adversative conjunction ‘but’ and the negative particle ‘no/not’. Precisely, the concept of frustrated completion could be understood in terms of:
• • • • •
Counterfactuality of foregrounded degree of verb situation realization Foregrounded degree of verb situation realization: Completion Resulting degree of verb situation realization: Pre-final stage Pastness Imperfectivity of pre-final stage.
The constructions that show frustrated completion in Malayalam are: 19. m݄݄ܤԥ ܤlijܤ:n Snow
tࡧ uܤܩƾƾi
pܤkݕe muݍuvܤnum ܤli݄݄-illܤ
melt.INF start.PST
but
full
melt-NEG
“The snow started to melt but it did not melt completely”. 20. veܿܿܤm tࡧ iܿܤkkܤ:n tࡧ uܤܩƾi, pakݕe ݔԥrikkum tࡧ iܿܤcc-illܤ Water boil.INF start
but
well
boil-NEG
“Water started to boil but it did not boil well.”
8
Mehta, Shreya and Reshma Jacob. “An Enquiry into the Modality of ‘desire’ as expressed by the lexical verb,‘ve:݅ܤm’ i.e. “want” in Malayalam from the Perspective of Grammaticalization Theory”. South Asian Journal of Multidisciplinary Studies Vol 1, Issue 4. (2015): 10 pgs. Accessed on 21st January 2016. url: http://sajms.com/enquiry-modality-desire-expressed-lexical-verb-ve݅am-e-wantmalayalam-perspective-grammaticaliza/ 9 Presentation titled “Grammar of Counter-to-fact” presented by Tania Kuteva in JNU (January – April 2014)
Semantically Elaborate Categories and Grammaticalization in Malayalam 47
The structure of frustrated completion is with the infinitive form of the verb along with adversative conjunction ‘but’ and the negative particle ‘no’. In all these sentences, the action described by the verb has actually started but it could not get completed or it got frustrated in the pre-final stage. Hence, it shows counterfactuality of the past verb situation in its pre-final stage.
2.7 Inconsequential Inconsequential “encodes the lack of completeness, or stability of the expected/wished for results/ consequences of a verb situation that has been realized in the past.”10 Precisely, it is the past verb situation that has taken place but failed in finding the expected result, or the action took place in vain. The concept of inconsequential can be further understood in terms of:
• Counterfactuality of foregrounded degree of verb situation realization • Foregrounded degree of verb situation realization: resultant state after a realized verb situation
• Resulting degree of verb situation realization: Full • Absent/ Incomplete/ Instable resultant state • Pastness. The inconsequential constructions seen in Malayalam are highly contextual and are interesting to look at. It can be explained with an example. Consider the example in (21). 21. ݄ܤ:n
nࡧ okki
1SG-NOM look.PST “I looked”. The example in (21) can mean two different things. One is the surface meaning ‘I looked (at something)’ and the second is the inconsequential meaning which could be understood as ‘I looked, but in vain’. It is interesting to note that the same construction is used to express opposite meaning. However, the inconsequential meaning is contextual, in that it is
10
Ibid
48
Chapter Three
understood when the implied context is known to both speaker and listener. It is also marked by a falling tone. For example, Ravi asked Manu to look for a book in the library. The next day when they meet Ravi asks Manu ‘did you look for the book?’ and if he replies /݄ܤ:n nࡧ okki/ meaning ‘I looked’ with a falling tone, it simply means that he looked for the book but he could not find it. Here, the action of looking for the book is completed but there is no result. That is, the book which was looked for was not found. Hence, the inconsequential meaning is implied only when the context is known to both speaker and listener, and is usually in the form of an answer to the question. Without such a context these are just simple past constructions.
3. Analysis of the Grammaticalized Elements in the Semantically Elaborate Constructions of Malayalam in Terms of Heine and Kuteva Grammaticalization, as mentioned earlier, is the development of a lexical item to a grammatical form and the grammatical form getting more grammaticalized. It is a diachronic process which is synchronically looked upon and can be used in language reconstruction in such a way that grammaticalization provides a more holistic view than that of historical linguistics. Moreover, it is assumed that the process of grammaticalization is based on the interaction of pragmatic, syntactic, morphosyntactic, and phonetic factors. According to Heine and Kuteva, there are four basic parameters of grammaticalization, namely: extension, desemanticization, decategorialization and erosion. In this section, we intend to limit our analysis of these parameters to those grammaticalized words which are found in semantically elaborate construction in Malayalam, as dealt with in the earlier section. The parameters of grammaticalization are:
3.1 Extension This is the rise of new grammatical meanings when linguistic expressions are extended to new contexts. It is also termed context-induced reinterpretation. “Extension has a socio-linguistic, text-pragmatic and a semantic component. The sociolinguistic component concerns the fact that grammaticalization starts with innovation as an individual act. The textpragmatic component involves the extension from a usual context to a new context or set of contexts, and the gradual spread to more general paradigms of contexts and the semantic component finally leads from an existing meaning to another meaning that is evoked or supported by the
Semantically Elaborate Categories and Grammaticalization in Malayalam 49
new context.”11 For example, the lexical word for the verb ‘to sit’ is “irikkuka”. Thus, in the sentence 22. ܤvan 3.sg.Mas
irikkukܤ ܤ:݅ sit.INF AUX
“He is sitting.” The verb “irikkuka” gives its lexical meaning ‘to sit’. However, the use of the same verb is extended in such a way that it gets grammaticalized and gives the proximative meaning. That is, 23. ܤvan
iƾo:ݚݚԥ
vܤrܤ:n
irikkuka ܤ:݅
3.sg.Mas here.DEM come.INF PROX AUX “He is about to come here.” In this sentence, the same word “irikkuka” is used in a different meaning. It follows another infinitival verb which is the main verb and the word “irikkuka” has now got nothing to do with its lexical meaning of ‘to sit’ but is extended in such a way that it gives proximative meaning.
3.2 Desemanticization This is an immediate consequence of extension. That is, “the use of a linguistic expression E in a new context C entails that E loses part of its meaning that is incompatible with C.”12 The concept of desemanticization can be explained with the same example used for extension. The use of the linguistic expression “irikkuka” in a new context of proximative construction entails that the word “irikkuka” meaning ‘to sit’ is incompatible in this context. To be specific, the word “irikkuka” undergoes desemanticization, loses the lexical meaning ‘to sit’ and gives an aspectual meaning of imminence. It is also important to note that all compound verb formations in South Asian languages are a case of desemanticization.
11
Heine, Bernd and Tania Kuteva.The Genesis of Grammar: A Reconstruction. New York: Oxford Press, 2007 12 Ibid
Chapter Three
50
3.3 Decategorialization After desemanticization, the linguistic item tends to lose morphological and syntactic properties characterizing its earlier use. Precisely, decategorialization tends to be accompanied by a gradual loss of morphological and syntactic independence of the linguistic item undergoing grammaticalization. That is: free form > clitic > affix. For example in the construction of frustrated initiation, we have: 24. co:r
ve:v-ܤ݅ܤm-ܤ:jirunࡧ nࡧ u
Rice boil-should-be.PST “Rice should have cooked well”. 25. enikkԥ
oru pܤ:ݚݚԥ ve:݅ܤm
1.sg.DAT one song want “I want a song”. Here, the suffix /-ܤ݅ܤm/ gives the meaning of ‘should’. This widely used suffix in Malayalam has its origin in the lexical word ‘want’ /ve:݅ܤm/. It is decategorialised in such a way that the free form changes to a suffix so as to give the meaning for ‘should’. That is /ve:݅ܤm/ > /-ܤ݅ܤm/.
3.4 Erosion As a result of undergoing grammaticalization, a linguistic expression loses phonetic substance. The phonetic erosion can be loss of phonetic segments, including loss of full syllables, loss of suprasegmental properties such as stress, tone, or intonation, loss of phonetic autonomy or phonetic simplification. However, erosion is not considered to be a mandatory step in the process of grammaticalization and is usually seen in the last stage. In the grammaticalized words in the semantically elaborate constructions, we can see many phonetically eroded words. For example, in the above given example of /ve:݅ܤm/ to /-ܤ݅ܤm/ we can see phonetic erosion.
Semantically Elaborate Categories and Grammaticalization in Malayalam 51
4. Analysis of the Grammaticalized Elements in the Semantically Elaborate Constructions of Malayalam in Terms of Hopper and Traugott According to Hopper (1991), there are five general principles of Grammaticalization, namely: layering, divergence, specialization, persistence and decategorialization. Each one of these will be discussed below in terms of those grammaticalized words that could be seen in the semantically elaborate constructions which were discussed earlier.
4.1 Layering Layering states that it is not necessary for a newly grammaticalized item to replace the older one in the same functional domain. On the other hand, both the items can co-exist. As mentioned earlier the past proximative construction in Malayalam can be expressed by the grammaticalized forms of both ‘go’ and sit’. Though both of these show imminence, one form has not yet replaced the other but they do co-exist. For example, 26. ܤvܿܤ
ܤtࡧ ԥ parajܤ:n
irikku-vܤjirunࡧ nࡧ u
3.sg.FEM that say.INF (sit)PROX-be.PST “She was about to say that.” 27. ܤvܿܤ
ܤtࡧ ԥ pܤrܤjܤ:n
3.sg.FEM that say.INF
po:ku-vܤjirunࡧ nࡧ u (go)PROX-be.PST
“She was about to say that.”
4.2 Divergence This suggests that when an item undergoes grammaticalization, it develops divergent items side by side. For example, the suffix /-ܤ:tࡧ e/ ‘without’ must be a divergent form of the word /ܤrutࡧ ԥ/ ‘do not’, since both of them have a negation implied and are used in different contexts side by side.
4.3 Specialization “As grammaticalization proceeds, the number of possible choices is gradually narrowed down and one specific item, which has become more
52
Chapter Three
general in meaning, is specialized for this particular function with increasing specialization, the chosen item finally becomes obligatory”13 ;i.e., the context forces the use of this element. The same example which has been used here for explaining layering can be used to explain the principle of specialization. As mentioned, both ‘po’ and ‘irikkuka’ are used to give the past proximation. Though both the verbs have semantic correlation the word ‘irikkuka’ is more widely used/specialized so as to show past proximation in Malayalam.
4.4 Persistence This says that after the process of grammaticalization, traces of the original lexical item could still be shown by the newly formed grammaticalized word. For example, the word /ܤrutࡧ ԥ/ has grammaticalized so as to give the suffix /-ܤ:tࡧ e/. Though both the source and target words are used in entirely different contexts, they mean almost the same. That is, the source word here is an independent word which is a negation, whereas the target word is in the form of an affix and is used only as a suffix. The meaning of the source word still persists in the target word.
4.5 Decategorialization According to this principle, the word that undergoes decategorialization loses its categorical autonomy and is downshifted to a secondary category. That is, it loses its morphological and syntactic properties which would identify it as a member of the previous category. As mentioned earlier, the grammaticalization of the word /ve:݅ܤm/ ‘want’ to the modal suffix /ܤ݅ܤm/ is an example. The source words have lost their morphological and syntactic autonomy and have now become an affix. The grammaticalized words in Malayalam which can be found in the semantically elaborate constructionsare confined to the five general principles proposed by Hopper and Traugott.
5. Conclusion This paper on the Semantically Elaborate Categories in Malayalam is based on Kuteva’s analysis of semantic elaborateness. We have proposed that in Malayalam, we have apprehensional, avertives, lest clauses, proximitive,
13 Wischer I. “Grammaticalization”. In Encyclopedia of Language and Linguistics, edited by Keith Brown (ed.). 2-129. New York: Elsevier, 2006.
Semantically Elaborate Categories and Grammaticalization in Malayalam 53
frustrated initiation, frustrated completion and inconsequential (which is contextual) constructions, which show counterfactuality and semantic elaborateness. These categories are placed in increasing order of the counterfactuality of the verb situation realization. Precisely, apprehensional has the lowest degree of verb situation realization in such a way that it stops the undesirable verb situation. Next, we have the avertives where the verb situation was aborted just before the boundary; i.e. avertive has more imminence/proximation than the apprehensional. Next to the avertives are placed frustrated initiation, which has more proximation: i.e. the realization of the verb situation was aborted just before the initiation. This is followed by the frustrated completion in which the realization of the verb situation has already happened but was aborted somewhere in the pre-final stage. The inconsequential is placed last because the verb situation has full realization but the counterfactuality occurs in such a way that the result could not be found. As seen in the earlier sections, each of these has a specific constructive pattern so as to give the intended meaning. It is also important to note that each of the constructions except frustrated completion and inconsequential have grammaticalized elements in them, which makes the discussion on grammaticalization important. Also, based on the analysis of the data in Malayalam, in light of the parameters of grammaticalization put forward by Heine and Kuteva and the general principles of grammaticalization by Hopper and Traugott, it seems that the process of grammaticalization is not a recent phenomenon in Malayalam.
References Heine, Bernd and Tania Kuteva. The Genesis of Grammar: A Reconstruction. (New York: Oxford Press, 2007). Hopper, Paul and Elizabeth Traugott. Grammaticalization. (Cambridge: Cambridge University Press, 2003). Kuteva Tania.Grammatical categories and Linguistic Theory: Elaborateness in Grammar. In Proceedings of Conference on Language Documentation and Linguistic Theory 2, ed. Peter K. Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells.(London: SOAS, 2009): 13-28. Mehta, Shreya and Reshma Jacob. “An Enquiry into the Modality of ‘desire’ as expressed by the lexical verb,‘ve:݅ܤm’ i.e. “want” in Malayalam from the Perspective of Grammaticalization Theory”.
54
Chapter Three
South Asian Journal of Multidisciplinary Studies Vol 1, Issue 4.(2015): 10 pgs. Wischer I. “Grammaticalization”. In Encyclopedia of Language and Linguistics, edited by Keith Brown (ed.). 2-129. New York: Elsevier, 2006.
Other Sources Presentation titled “Grammar of Counter-to-fact” presented by Tania Kuteva in JNU (January – April 2014).
CHAPTER FOUR ASPECTS IN MEETEILON1 PADMABATI ACHOM
Abstract This paper focuses on the nature of aspects in Meeteilon, a Tibeto-Burman language spoken in the north-east part of India, and the contribution of the deictic suffixes towards the aspectual meaning. Aspect is classified into Outer Aspect and Inner Aspect (Travis, 2010). The Outer Aspect is realised morphologically while the Inner Aspect is rarely realised morphologically, and is interpreted depending upon the types of verbs, objects, and its prepositional complement. However, in Meeteilon, the deictic suffixes -khᢌi-, -ru/lu- and –dok- help in providing the meaning of the inner aspect and therefore classify the nature of the eventualities. On the other hand, the deictic suffix -rᖠm/lᖠm- helps in realising the outer aspect. This paper also addresses the problem of the syntactic structure of the inner aspect. Keywords: Outer Aspect, Inner Aspect, deictic, event, sub-event.
1. Introduction The term ‘aspect’ is defined as different ways of viewing the internal temporal constituency of a situation (Comrie, 1976). Tense and Aspect in Meeteilon are often expressed by the same morpheme and, hence, a certain amount of overlap can be observed; it is therefore difficult to tease out the function of Aspect, and separate it from that of Tense.
1
ASP=Aspect, DEIC=Deictic, DEM=Demonstrative, ERG=Ergative, FUT=Future, IND=Indicative, LOC=Locative, PERF =Perfect, PROG=Progressive
Chapter Four
56
(1) ᖠi cᢌk I
rice
cᢌ-re eat-PERF
‘I ate rice/I have eaten rice.’ Given the fact that in some cases, aspect can be uniquely expressed by a morpheme (e.g. the future progressive), tense and aspect are treated as two different categories, and hence, a separate AspP in Meeteilon is projected, inspite of the overlap mentioned. (2) ᖠi cᢌk I
rice
cᢌ-rᖠm-lᖠ-gᖠni eat-DEIC-PERF-FUT
‘I will have eaten rice (when you come).’ Aspect denotes the temporal structure of a situation rather than locates itself in the timeline, unlike Tense. Aspect can be classified into Outer Aspect and Inner Aspect. Outer aspect also known as viewpoint aspect (Smith, 1991) is a grammatical aspect which is realised morphologically. On the other hand, inner aspect, also known as situation aspect (Smith, 1991) or Aktionsart deals with eventualities such as Accomplishment, Achievement, Activity and State (Vendler, 1957). Inner aspect is rarely morphologically realised, and it is interpreted depending upon the types of verbs, objects, and its prepositional complement. To understand the nature of inner and outer aspect better, the paper discusses the difference between the inner aspect and the outer aspect in Section 1.1. The syntax of inner aspect and its relation with outer aspect is discussed in Section 2. Section 3 presents a conflicting view of inner aspect. The syntactic analysis of the aspectual nature in Meeteilon is shown in Section 4, with Section 5 concluding the paper.
1.1 Difference between the Inner Aspect and the Outer Aspect Inner aspect is a phenomenon of language in which the internal temporal mapping, or what is sometimes called the subevent structure of the event, is important. To explicate, predicates are descriptions of events that are further composed of subevents. The structure of subevents is usually expressed in terms of endpoints.
Aspects in Meeteilon
(3) a. John drank a bottle of beer last night.
57
(MacDonald, 2008)
b. John drank beer last night. The sentence in (3a) is a description of an event with an endpoint, which is the moment when the bottle of beer is finished. Thus, (3a) can be considered a completed event. Accordingly, the predicate in (3a) is a telic predicate, while in (3b), the predicate describes an event that may not have an endpoint. The amount of beer drunk is not specified, so the affected object cannot be determined and therefore no endpoint can be specified; this is an atelic predicate. In (3a) the internal argument denotes a specific quantity of beer and the predicate is telic; in (3b) the internal argument does not denote a specific quantity of beer and the predicate is atelic. Now, let us consider some properties of outer aspect, illustrated by the sentences below: (4) a. Jerome was drinking a bottle of beer (when I left the bar). b. Jerome was drinking beer (when I left the bar). (MacDonald, 2008) In the above sentences, neither sentence (4a) nor (4b) describes whether the event has any endpoint irrespective of the nature of the internal argument. One thing that can be noted right away is that outer aspect has a morphological reflection: the verb is in the progressive form due to the presence of the -ing form on the verb, and the auxiliary be is present too. Inner aspect in English on the other hand does not display any morphological effect as in (3). Similar to English, outer aspect in Meeteilon is realised morphologically with the progressive morpheme -ri/-li/-mi attached to the verb. However, in Meeteilon, the deictic morpheme -rᖠm-/-lᖠm- has to be added with the progressive form. This deictic morpheme also conveys some aspectual properties of durative as shown in (5). (5) a. John 1ᖠrᢌ1 beer John yesterday
beer
botᖠl ᖠmᖠ
thᖠk-*(lᖠm)-mi
bottle one
drink-Deic-Prog
‘John was drinking one bottle of beer yesterday (when I left the bar).’
Chapter Four
58
b. John 1ᖠrᢌ1 beer John yesterday
thᖠk-*(lᖠm)-mi
beer
drink-Deic-Prog
‘John was drinking beer yesterday (when I left the bar).’ As in sentences (3), the sentence in (6a) is a description of an event with an endpoint, which is the moment when the bottle of beer is finished. Thus, (6a) can be considered a completed event. Accordingly, the predicate in (6a) is a telic predicate, while in (6b), the predicate describes an event that may not have an endpoint. The amount of beer drunk is not specified, so the affected object cannot be determined and therefore no endpoint can be specified; this is an atelic predicate. (6) a. John 1ᖠrᢌ1
beer
John yesterday beer
botᖠl ᖠmᖠ
thᖠk-i
bottle one
drink-IND
‘John drank one bottle of beer yesterday.’ b. John John
1ᖠrᢌ1
beer
yesterday beer
thᖠk-i drink-IND
‘John drank beer yesterday.’ Keeping in mind that inner aspect corresponds to the structure where subevents are usually expressed in terms of endpoints, I will attach the deictic morpheme –khai- ‘multidirection’ (Hidam, 2013) to the verb and see whether it gives any sub-event to the predicate. (7) keithel-da
bom
po-khᢌi-ri
Market-LOC bomb explode-DEIC-PROG ‘Bomb is exploding at the market.’ The sentence in (7) describes multiple events of bomb explosion one after another. Every single event is completed and is telic. However, the entire event is not completed, and this is expressed by the outer aspect suffix –ri. Here, the deictic -rᖠm/lᖠm- can also be added to convey the outer aspectual property.
Aspects in Meeteilon
(8) keithel-da
bom
59
po-khᢌi-rᖠm-mi
Market-LOC bomb explode-DEIC-DEIC-PROG ‘Bomb was exploding at the market (when I reached there).’ This shows that the deictic suffix -khᢌi- supports the inner aspectual property while the deictic -rᖠm/lᖠm- supports the outer aspectual property in Meeteilon.
2. Syntactic Account of Inner Aspect MacDonald (2008) claims that Object-to-Event (OTE) mapping is syntactically achieved via Agree with an aspectual projection between vP and VP: AspP. He contends that if a [–q] NP Agrees with Asp, the predicate is expected to be atelic; if a [+q] NP Agrees with Asp, the predicate can be telic. Interpretable features are derived from ES (Event Structure) of a predicate which shows that the event is bounded. The beginning of an event is marked by the feature of the predicate; whereas the end to an event is marked by the feature. In this way, ES and the OTE mapping property are treated as independent properties. The four aspectual predicate types (statives, activities, accomplishments and achievements) can be divided in two ways. The first is between stative and eventive predicates (i.e., activities, accomplishments, and achievements). The presence or absence of the aspectual projection AspP in the verb phrase captures this division. Eventive predicates have AspP, as illustrated in (9a), and stative predicates lack AspP, as shown in (9b). (9)
Chapter Four
60
Since in Meeteilon, the deictic -khᢌi- supports the inner aspectual property, deictic phrase (DEICP) can be inserted between AspP and VP for the eventive predicates as below: (10)
2.1 Atelic and Telic The four aspectual predicate types can also be divided between atelic and telic. Activities and statives are atelic and accomplishments and achievements are telic. Activities have the basic syntactic aspectual structure in (11a), and statives the one in (11b) (MacDonald, 2008). (11)
Aspects in Meeteilon
61
It can be seen here that activities have an feature and statives have none, making these predicates atelic. This feature indicates that the event has a beginning. Statives do not have any event features, and therefore they have no subevent structure. Accomplishments have the basic syntactic aspectual structure in (12a) and achievements the one in (12b) (MacDonald, 2008). (12)
Both accomplishments and achievements have an feature and an feature. That is, there is both a beginning and an end to the events that these predicates describe. They are both telic. For accomplishments, time elapses between the beginning and the end of the event. For achievements, no time elapses between the beginning and the end of the event. These distinct interpretations result from the distinct syntactic relations between the event features. The and features of accomplishments are related syntactically through c-command and, consequently, time is interpreted to elapse between the beginning and the end of the event. There is no c-command relation between the and feature of achievements and, consequently, time is not interpreted as elapsing between the beginning and the end of the event. The lack of a c-command relation between the event features of achievements results from the event feature configuration appearing on Asp. A typology of syntactic aspectual predicate types according to this system is summarized in the representation below: (MacDonald, 2008)
Chapter Four
62
(13)
In Meeteilon, the deictic –khai- can be attached only to verbs which have the eventive reading for achievements. (14) meri-nᖠ
jon nu1si-(*khᢌi)
Mary-Erg
State
John love-DEIC
‘Mary loves John.’ (15) ᖠi cᖠt-(*khᢌi)-li
Activity
I walk-DEIC-PROG ‘I am walking.’ (16) bom po-*(khᢌi)-re
Achievement
Bomb explode-DEIC-PERF ‘Bomb (has) exploded.’ (17) un Snow
sᖠu-(*khᢌi)-re
Accomplishment
melt-DEIC-PERF
‘Snow (has) melted.’ However, in Accomplishment, the deictic –dok- ‘outward’ can be attached to the verb as follows:
Aspects in Meeteilon
(18) un
sᖠu-dok-le
Snow
63
Accomplishment
melt-DEIC-PERF
‘Snow (has) melted.’ .
The deictic –lu/ru- ‘motion (near to far)’ (Hidam, 2013) can be attached to the verb to give accomplishment. (19) ᖠi cᖠt-lu-re I
Accomplishment
walk-DEIC-PERF
‘I went/I have gone.’ However, the deictic –lu/ru- cannot be attached with progressive morpheme –ri/li to give activity as in (20). It shows that the outer progressive aspect cannot select a telic event. The deictic suffix –lu/rulike the deictic suffix -khᢌi- gives a telic predicate. (20) *ᖠi I
cᖠt-lu-ri walk-DEIC-PROG
‘I am walking.’ From these sentences discussed above, it can be concluded that the deictic suffixes –lu/-ru,-dok- and -khᢌi- support the inner aspectual property giving telic interpretation. Hence, the syntactic tree structure of MacDonald (2008) can be modified by inserting DEICP in between AspP and VP to express Achievement and Accomplishment in Meeteilon.
Chapter Four
64
(21)
In the above structure (21), for Accomplishment predicate the verb, V, moves to the DEIC giving the telic interpretation. On the other hand, the DEIC will move to Asp to give the telic interpretation for Achievement predicate as in (22). (22)
2.2 Syntactic Account of the Outer Aspect in Relation to the Inner Aspect The relation between the Outer Aspects and the Inner Aspects with respect to the phrase structure are clearly presented with the extended projection of V. Based on Travis (2010), the crucial characteristics of this phrasestructure are based on the following: (i) The VP is layered as in Larson (1987). (ii) Lexical heads within the VP reflect semantic structure as in Hale and Keyser (1993, 2002).
Aspects in Meeteilon
65
(iii) As in Hale and Keyser’s work, a lexical item is a phrasal idiom that spans these heads. (iv) There is a functional category (Aspect) within the layered VP (vP) that is embedded in syntactic representations of the lexical entry. (v) There is a functional category (Event) directly above the VP (vP) that marks the edge of the event. (23)
In the above structure, there are two Vs which are labelled as V1 and V2 based on Larson’s two VP shells. Between these two shells is an inflectional (functional) category, ASP. Above these two shells is another functional category E(vent). Outer Aspect (OASP) takes scope over this entire event (EP). V1 is a lexical category that introduces the external argument, which has a meaning similar to CAUSE. ASP, depending on its feature content, can have a meaning similar to BE/BECOME. V2 introduces the Theme argument and the endpoint of the event, XP. Inner Aspect is an inflectional category within the lexical domain of the VP.
66
Chapter Four
Event Phrase was introduced as a higher parallel to Inner Aspect. Just as Inner Aspect is an event-related category at the edge of V2P, Event is an event-related category at the edge of V1P. The [+/–process] feature is represented in V1 and distinguishes Achievements and States from Accomplishments and Activities, while the [+/–telic] (definite) feature distinguishes Accomplishments and Achievements from Activities and States. This feature appears in ASP of the inner aspect. The material in the [Spec,V2P] does not obtain telicity. A Theme can be an affected object if it moves to [Spec, ASP] or enters into an AGREE relationship with ASP. Outer Aspect can select the shape of its complement. For example, progressive in English selects a durative event. This accounts for why progressive is generally bad when combined with a State or an Achievement.
3. The Conflicting Views MacDonald (2008) claims that an element must be within the domain of AspP (Inner Aspect) if it contributes to the aspectual interpretation of a predicate. Based on this, there are three pieces of evidence that an element is structurally higher than AspP if it cannot contribute to the aspectual interpretation of a predicate. The evidence is as follows: (MacDonald, 2008) a) The lack of aspectual contribution of CAUSE, which introduces the external argument of causer. b) The lack of aspectual contribution of external arguments. c) The lack of aspectual contribution of locational prepositional phrases (adjoined to vP as an adjunct). The lexical head V provides the aspectual meaning to stative verbs such as love, like, and action verbs like go, walk etc. Hence, vP is structurally above the AspP, and AspP is above the VP. Thus, the inner aspect phrase (AspP) is within the vP shell. Note that the syntactic prime CAUSE will be considered to be located at v. However in Travis (2010), V1 is a lexical category that introduces the external argument, which has a meaning similar to CAUSE. The [+/– process] feature is represented in V1 and it distinguishes Achievements and States from Accomplishments and Activities. This contradicts MacDonald’s (2008) first point, which states that there is a lack of
Aspects in Meeteilon
67
aspectual contribution of CAUSE which introduces the external argument of causer. There is a difference between goal PPs and locational PPs, and the difference can be seen in do so constructions as below: (24) a. ??Rahul pushed the box into the hall and Shyam did so into the store. b. Rahul pushed the box in the hall and Shyam did so in the store. From the above examples, the do so facts suggest that location PPs are adjoined to vP while goal PPs are lower in the verb phrase. Thus, Locational PPs are structurally higher than AspP, while goal PPs are lower (MacDonald, 2008). If it is the case that Locational PPs do not contribute to the interpretation of Aspect of a clause, then their placement at the vP level (i.e. outside the AspP) is predicted. However, if Locational PPs also contribute to aspectuality as in (25), then MacDonald’s third requirement needs to be further examined. As it stands, only (b), which is the second requirement of MacDonald’s structure, can stay as a test for positioning of elements outside/inside of the inner aspect. (25) unᖠ-du-nᖠ
*(lᖠimᢌi-dᖠ) kel-le
Leaf-DEM-ERG ground-LOC fell-PERF ‘That leaf fell/has fallen on the ground.’
4. Analysis In Meeteilon, the deictic suffixes -khᢌi- and –lu-/-ru- show the internal nature of the action or event as achievement and accomplishment respectively, and also the deictic suffix -rᖠm-/lᖠm- helps in realising the outer aspect; therefore, all three must be taken into account in terms of their contribution to aspectuality and their positioning. Hence, there will be two positions of DeicP. I will propose that one of the deictic phrases, DeicPO, which gives the outer aspectual meaning, is a category projected between the vP and OAsp. Consider the following sentence in this connection:
Chapter Four
68
(26) ᖠi I
cᢌk rice
cᢌ-rᖠm-mi eat-DEIC-Prog
‘I was eating rice.’ As proposed earlier, the DeicP will project under the Outer Aspect Phrase as in (27). (27)
Aspects in Meeteilon
69
In (27), the verbal root, cᢌ- moves to the inner aspect head, Asp and is interpreted as [-Telic] prior to moving to v, where it is interpreted as [+Process]. From the head, v, it moves to the head DeicO to get the deictic feature of -rᖠm- interpreted and then it moves to the head, OAspP, which scopes over the entire event, and this is interpreted as the progressive of the event activity. Lastly, it moves to the head, T, for tense. The subject, ᖠi is interpreted as the causer, as it is in the cause [Spec, vP] and moves to [Spec,TP] due to EPP. The object cᢌk position is not visible for the computation of telicity. In order for a Theme to be able to measure out a predicate, it must have moved to the [Spec, ASP], and so the object cᢌk is moved to the [Spec, AspP] (Travis, 2010). However, the contradictory functions of head v need to be analysed further to see whether head v lacks in aspectual contribution. I will propose another DeicPI, a category projected between AspP and VP, which gives the inner aspectual meaning. Now, let us consider the following sentence (28) which is syntactically shown in (29). (28) bom Bomb
po-khᢌi-re explode-Deic-Perf
‘Bomb (has) exploded.’
70
Chapter Four
(29)
In the above tree, the verbal root, po-, moves to DeicPI and then moves to inner aspect, ASP head, and is interpreted as [+Telic]; and then it moves to the head, v, and is interpreted as [-Process]. From the head, v, it moves to the head, Deic, and then moves to the head, OASP, which scopes over the entire event, and is interpreted as the perfect of the event achievement.
Aspects in Meeteilon
71
Then it moves to the head, T, for tense. The subject, ᖠi is interpreted as the causer, as it is in the cause [Spec, vP] and moves to [Spec,TP] due to EPP. However, the subject is deleted in the sentence (25), and so it is kept in brackets in the tree. The object bomb position, [Spec, VP], is not visible for the computation of telicity. In order for a Theme to be able to measure out a predicate, it must have moved to the [Spec, ASP], and so the object bomb is moved to the [Spec, ASPP]. The outer deictic phrase, DeicPO and the inner deictic phrase, DeicPI, can occur together to show the nature of the outer and inner aspects respectively. (30) bom Bomb
po-khᢌi-rᖠm-mi explode-DeicI-DeicO-PROG
‘Bomb was exploding (when I reached).’ The sentence in (30) describes the multiple events of bomb explosion one after another. The deictic suffix -khᢌi- conveys that every single event is completed and is telic. However, the entire event is not completed, as is expressed by the outer aspect suffix –mi, and the deictic suffix -rᖠm/lᖠmshows durativity.
5. Conclusion I have shown in this paper that the outer aspect in Meeteilon is marked with the help of the deictic particle -lᖠm/rᖠm-. The inner aspect is indicated with the help of the deictic suffixes –dok and -lu/-ru for accomplishment and the deictic suffix -khᢌi for achievement respectively. I have also discussed the structure of aspects with respect to Travis (2010) and MacDonald (2008). In Travis’s (2010) account, there is a head v which gives [±process] readings. This is in contradiction with MacDonald’s (2008) proposal regarding the v head not contributing to the aspectual meaning. Thus, the mechanism of how inner aspect is realised in the language needs further investigation.
72
Chapter Four
References Achom, Padmabati. “The role of aspect and the structure of VP in Meeteilon.” MPhil diss., University of Delhi, 2014. Comrie, Bernard. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press, 1976. Hale, Kenneth and Keyser, Samuel Jay. “On argument structure and the lexical expression of syntactic relations.” In The view from building 20. Essays in linguistics in honor of Sylvain Bromberger, edited by K. Hale and S.J. Keyser, 53-109. Cambridge, Massachusetts/London, England: The MIT Press, 1993. —. Prolegomenon to a theory of argument structure. Cambridge, Massachusetts/London, England: The MIT Press, 2002. Hidam, Gourashyam Singh. “Space and Tense Entanglement.” Paper presented at the Meeteilon Linguistics Research Studies group, CASL, University of Delhi, 2013. Larson, Richard. “On the double object constructions.” Linguistic Inquiry 19 (1987): 33-91. MacDonald, Jonathan. The Syntactic Nature of Inner Aspect: A minimalist perspective. Amsterdam/Philadelphia: John Benjamins Publishing Company, 2008. Smith, Carlota. The parameter of aspect. Dordrecht: Kluwer Academic Publishers, 1991. Travis, Lisa de Mena. Inner aspect.The articulation of VP. New York: Springer, 2010. Verkuyl, Henk. On the compositional nature of the aspects. Dordrecht: Reidel, 1972.
CHAPTER FIVE SEMANTICS OF REDUPLICATIVE NOMINAL QUANTIFICATION IN BANGLA1 NANDINI BHATTACHARYA
Abstract The formal study of quantification and its logical dimensions has been a widely debated issue in formal semantics as well as in syntax. This is exemplified by Barwise and Cooper’s (1981) arguments on weak and strong quantification, May’s (1981) proposals on the syntax of quantification, Szabolsci’s (2001, 2010) treatment of the syntax and semantics of scope, and the treatment of cardinal and proportional quantification by Partee (2002) etc. However, these arguments have not adequately addressed the phenomena relating to the strategies of encoding quantification in South Asian languages. This paper presents the issue of quantification as encoded in reduplicative nominal quantification in Bangla. In this paper, I examine various reduplicated nominal phrases in Bangla, such as akaše akaše megher khela ‘the play of clouds in different skies’, pэt֍ he pэt֍ he ‘on different routes’, d֍ ike d֍ ike ‘in various directions’. I account for the non-occurrence of reduplicated nominals as conjoined constituents or under the scope of negation. Finally, I put forward a partly formalized account for the distributive plurality that is expressed implicitly by such reduplicated nominals. In conclusion, this paper will highlight the quantification effects of reduplicative nominals in Bangla.
1
NP=noun phrase, 1st pers=first person pronoun, sg=singular, gen=genitive case marker, pres=present tense, prog= progressive, be=copula, loc=locative case marker, cls=classifier, hab=habitual, aux=auxiliary, plu=plural marker, perf= perfect aspect, acc=accusative case marker, loc=locative case marker, Neg=negation, hon=honorific marker, imp= imperative mood, seq=adverb of sequence, 1st self=first person reflexive pronoun, QP=quantifier phrase
74
Chapter Five
Keywords: Quantification, Semantics, Reduplication, Bangla, Implicit Quantification, Quantifiers.
1. Introduction The study of quantification in natual languages has been investigated in formal semantics and philosophy of language for decades by scholars including Montague (1973), May (1977), Barwise and Cooper (1991), Bach (1995), Kennedy (1997), Partee (2002), Balusu (2005), Szabolsci (2010), Dayal (2013) et al. The semantic mapping of quantification and plurality in Bangla has also received notable attention by Bagchi (2011), Biswas (2013), Dayal (2014) et al. In this paper, I draw upon the issue of implicit quantification that can be observed in Bangla and attempt to understand it with respect to some Bangla Plural NPs without overt quantifiers or plural/quantifier markers. Furthermore, in this paper I put forward an observation about the issue of how reduplicated noun phrases in Bangla contribute to add distributive reading and encode implicit unrestrictive quantification. This analysis is significant to shed light on the morpho-semantics of reduplication in Bangla. The data gathered in the study suggests how noun phrases in Bangla are reduplicated to suffice for implicit quantification. Reduplication is an important linguistic feature in South Asian languages (see Emeneau (1956), Abbi (1991), Balusu (2005), Szabolsci (2010)) and can be observed in Bangla, Hindi, Punjabi, Telugu and Malayalam among others. However, for the most part, the study of reduplication in South Asian languages has been limited to the areas of phonology, morphology and to some extent syntax, leaving the semantics of quantifier scope and the semantics of implicit quantification unexplored. By examining reduplication in Bangla, my study provides broad insights into the semantics of reduplication in South Asian languages. In the following sections of this paper, I will examine key linguistic differences between reduplicated nominals and other plural NP expressions in Bangla. In section (2), existing literature on plurality, distributivity and reduplication is summarized to set the stage for the rest of the paper. In section (3) titled ‘Implicit quantification’, I will discuss relevant data showing the occurrence of reduplicative quantifiers in various sentence structures in Bangla along with my observations. In section (4) ‘Semantic Analysis’, I will develop the claim that reduplicative phrases in Bangla contribute significantly to adding additional semantic
Semantics of Reduplicative Nominal Quantification in Bangla
75
features for full interpretation of sentences. I will also provide a partly formalized account of distributivity in this section. Remaining grounds for further research on this issue are mentioned in section (5).
2. Setting the Stage Reduplication has been defined as a morphological process where the root or stem of a word is repeated exactly or repeated with partial modifications. Some of these reduplicated constituents occur as a composite phrase and some of these can be extra-posed. Reduplication is used to perform many grammatical functions in natural language. Reduplication can either be Full Reduplication or Partial Reduplication. Sometimes intonation patterns of the reduplicated roots or stems also alter in the process. There is also phonological reduplication, for example, assimilation etc. Echo reduplication, which is another notable feature of South Asian languages, does not add any alternation to the existing meaning and it is mostly onomatopoeic. Abbi (1992) proposes that there are four semantic features that are represented by nominal reduplication in South Asian languages. These encoded semantic features are distributiveness, emphasis, exclusive and discrete time nominals. Abbi claims that only the distributive feature is found in Bangla nominal reduplications. Distributivity is the only feature that is common to all reduplicated nominals in Indian languages (except Malayalam). These are several instances of reduplication in many South Asian languages, for example, Sindhi verbal reduplication /hԥlԥnde hԥlԥnde/ ‘while walking’, Nepali echo-reduplication /gԥran gurun/ ‘thundering sounds’ etc. Likewise, reduplication is also used in languages spoken in India. The Tibeto-Burman language examples are the following: reduplication of adjectives in /~hoi ~hoi/ ‘good-good’ in Paite and adverb reduplication /thu-n~ thu-n~/ ‘fast-fast’ in Meiteilon. Mizo /olep olep/ 'sticky' is an instance of reduplication, in spite of the fact that the part which is repeated is neither a lexical item nor a part of a lexical item (Abbi, 1990). The examples of reduplication in Dravidian languages are the following: for example, in Telugu, reduplication of numeral /renDu renDu/ ‘two-two’ (Balusu, 2005), in Tamil, /mummuunRu/ ‘three each’. (Abbi, 1991) et al. The instances of reduplication in Indo-Aryan Languages are as follows: Hindi nominal reduplication /ghԥr ghԥr/ ‘in every house’, Punjabi verbal reduplication /cԥlda cԥlda/ ‘while walking’,
76
Chapter Five
Gujarati modifier reduplication /moݚi moݚi/ ‘big’, Oriya modifier reduplication /nuͅ nuͅ/ ‘new new’, Bangla pronominal reduplication /amra amra/ ‘we all’ etc. Bangla is an Indo-Aryan language that is spoken mostly in Eastern India and Bangladesh. There are many dialects of Bangla; this study is based on the Standard variety of Bangla (i.e. Bangla spoken in Kolkata and surrounding areas, used in ordinary communication). The basic constituent order in Bangla is SOV. In Bangla, there is a phenomenon of implicit quantification, where there is no quantifier particle overtly present to encode plurality. In such cases, the quantification is signalled by another linguistic mechanism, i.e. by fully reduplicating the noun phrases. The reduplicated quantifiers occur structurally under the QP node as a conjoined phrase. The phenomenon of reduplication in the language is observed in both the nominal and the verbal domain. Consider the following illustrations: i) quantifier reduplication /kichu kichu/ ‘some of the Ns’ ii) adjective reduplication / lal lal/ ‘red-red’ iii) adverb reduplication, /dhere dhere/ ‘slowly slowly’ iv) nominal reduplication /ghܧre ghܧre / ‘in every house’ v) numeral-classifier reduplication / dࡧ uto dࡧ uto/ ‘two two’ vi) verbal reduplication /bܧlte bܧlte/ ‘while talking’ In this paper, I focus on the nominal reduplication in Bangla and analyse the semantics of these reduplicative NPs in respect to implicit/covert quantification.While describing the reduplication of morphology in Bangla, Abbi (1992) states that, “A variety of, or scattering of objects over a location is expressed by reduplicated nouns”. The cited Bangla example is, / o baݏi baݏi jeto o bhܼkke chaܼto/. [He used to go door to door and beg.] Here, the reduplicated nominal /baݏibaݏi/ encodes distributiveness, which is also signalled by repetition of the English word /door/ in the equivalent sentence. Abbi claims that there is one exception in Bangla, where reduplicated pronouns emphasize the feature of exclusiveness (ibid). The cited example is, / tomra tomra jao/ [you all (only) go]. She asserts that this exclusiveness feature cannot be found in reduplicated nominals in Bangla [ibid]. However, in this paper, more observations on the
Semantics of Reduplicative Nominal Quantification in Bangla
77
morphology of nominal reduplication in Bangla are proposed. These reduplicated NPs are analysed from the perspectives of distributive plurality and implicit quantification. It must be noted that I understand reduplication to be different from syntactic doubling. Syntactic doubling occurs when a constituent, a feature or a phrase, is expressed two or more times within a clause. According to Barbiers (2008), syntactic doubling refers to the phenomenon in which two elements (WH-pronouns or subject pronouns) referring to the same syntactic entity co-occur in one and the same sentence, without making any semantic contribution. Sometimes, these reduplicated syntactic constituents or features are deleted in LF, while the PF retains the original identical copies (Barbiers, 2008). However, I argue that the instances of reduplicated NPs in Bangla imply that these NPs are fully reduplicated and are retained in LF and are pronounced without any intonational modifications. Moreover, the reduplicated NPs in Bangla, indeed, contribute to the semantic interpretation of the propositions by encoding additional semantic features to the predicate. These additional semantic contributions include distributive plurality and implicit quantification. In this sense, they are neither instances of phonological reduplication nor of syntactic doubling. Let me elaborate on the distributive interpretation here. According to Lasersohn (1995), the plural expressions in natural languages may be intuitively characterized as those involving reference to multiple objects or kinds. However, semantic theories differ in how this intuition is worked out formally. Plurality in natural language semantics can be divided into collective or atomic plurality and distributive plurality. Winter and Scha (2015) propose that in the example, ‘Mary and Sue met’, the speaker is conveying a statement about a collection of individuals. Such interpretations are referred to as Collective reading or Collective interpretation. They state that in the example, ‘The girls are smiling’, the speaker claims that each or at least many of the individual girls smiled. Statements of this sort yield Distributive reading or Distributive interpretation. Distributive plurality has been the subject of more in-depth inquiry in contemporary semantics. Distributivity implies dependencies between two quantified variables where the ‘predicate distributes over the collection referred to by its plural argument’. For example, ‘Two girls collided’ encodes collective plurality. ` Collide(a ْ b) However ‘Two girls sneezed’ encodes distributive plurality.
Chapter Five
78
` Sneeze(a) רSneeze(b) There are also plural NPs that are ambiguous between being collective or distributive reading. For example, consider the sentence, ‘Two girls played’. We know this sentence is ambiguous between a distributive reading and a collective reading of the NP /two girls/. ` a. Play (a ْ b) ` b. Play (a) רDANCE(b) The presence of a distributor operator or D-operator to analyse such distributive sentences is also proposed by many semanticists, such as Beghelli and Stowell (1997), Brisson (1995) et al. Brisson (1995) claims that ‘Distributivity’ is a property of the verb phrase or predicate and is represented by a D operator. The function of the D operator is to introduce universal quantification over the plural variable that is introduced by the subject. In the example, ‘The boys ate an apple’, the distributive operator (D) scopes over the existential object ‘an apple’. This is formalised as:`
D
.ate.an apple’(the, boys’)
` x[x[אthe.boys’]ĺ ate.an apple’(x)] I also analyse in the following sections the distributive readings of the implicit quantification in Bangla. This paper observes the mapping of implicit quantification in natural language semantics, in the crosslinguistic arena where the interface between morphology and Formal Semantics becomes prominent.
3. Implicit Quantification The implicit quantification is achieved in Bangla by reduplication of nominals. This reduplication is morphologically (or syntactically) full reduplication, which implicitly implies plurality. Mostly the indefinite common Noun Phrases are reduplicated. In this section, we will look at the occurrence of these reduplicative nominals in different positions and in various sentence types.
Semantics of Reduplicative Nominal Quantification in Bangla
79
3.1. Bangla Reduplicative Nominals in Different Sentence types These reduplicative nominals in Bangla occur both in subject position and object position, and they also occur as locative adjuncts. The different case markings that are generally added to these are locative, genitive, besides bare nominative or accusative. In example (2), the locative case marker /e/ is reduplicated. In Hindi, /ghԥr ghԥr mӁ / the locative marker /mӁ/ is not reduplicated, unlike in Bangla, example (6) /ghܧre ghܧre/. The plural markers in Bangla are also reduplicated in these composite nominal phrases. In example (10), the instrumental case marker /-e/ is reduplicated. In example (7), the plural marker /-ra/ is reduplicated. However in some instances the reflexive pronouns in Bangla are also reduplicated, but it does not encode distributivity; it refers to an atomic individual. For example: 1) se šudࡧ hu ƗmƗr ƗmƗr korejae. se šudࡧ hu
ƗmƗ
ƗmƗ
he only
1st per. –gen. 1st per. –gen. be.pres.prog.
-r
-r
korejae
“He only thinks/goes on about himself.” The following sentences and phrases elucidate the various reduplicated NPs that in turn, implicitly quantify over the variable. For example: 2) akaše akaše megh koreche akaš -e sky
akaš -e
-loc. sky
megh
-loc. cloud
koreche spead.pres.hab.
“There are clouds in different skies.” 3) ami pܧtࡧ he pܧtࡧ he phiri ami pܧtࡧ h -e I
pܧtࡧ h -e
phiri
path -loc.path -loc.roam.pres.
“I roam on many roads.”
Chapter Five
80
4) dࡧ ike dࡧ ike sanai baje dࡧ ik
dࡧ ik
-e
-e
sanai
baje
direction -loc. direction -loc. shehnai play.pres. “Shehnai is being played in many different directions.” 5) “phule phule ށܩole ށܩole bܧhe kiba mridࡧ u bae” ށܩole ށܩole
phul -e phul-e
kiba mridࡧ u bae
bܧhe
flower-loc. flower-loc. sway sway
flow.pres. how slow
breeze
“How! The slow breeze is flowing over the flowers and making them swing to each other.” 6) ei biota ghܧre ghܧre ache ei
boi
ghܧr
-ta
ghܧre -e
-e
ache
this book -cls. house –loc. house -loc. be.pres. “This book is (typically) in every home.” 7) chelera chelera khela korche chele -ra chele
-ra
khela korche
boy -plu. boy
-plu. play
aux.pres.prog.
“Boys are playing (with each other).” 8) dܧݤne dܧݤne melate gelo
dܧݤn
-e
dܧݤn
-e
mela -te
gelo
person. –acc.person -acc fair -loc. go.pres.perf. “Many people have gone to the fair.”
Semantics of Reduplicative Nominal Quantification in Bangla
81
9) gatݚހݕa phule phule dࡧ haka gat ހݕ-ݚa
phul -e
dࡧ haka
phul -e
tree -cls. flower-ins flower-ins. cover.pres. “The tree is covered with flowers.” 10) gatݚހݕar gat ހݕ-ݚa -r
dࡧ ale dࡧ ale notࡧ un patࡧ a hoetހݕe dࡧ ale
dࡧ ale
notࡧ un patࡧ a
bough-clas.–gen. bough bough new
hoetހݕe leave
grow.perf.
“New leaves have grown in different branches of the tree.” These Bangla examples show the reduplicated NP constructions in natural language. It can be observed in these above examples that the reduplicated NPs implicitly imply the plurality without adding any plural morphological marker. In example (3), /pܧtࡧ he pܧtࡧ he / refers to various paths; however the plural marker /gulo/ is not added here to imply plurality. The data also indicates the syntactic doubling of Locative case marking in these composite noun phrases. The reduplicated NPs are spelled out at PF simultaneously (i.e. the case markers/plural markers arenot even deleted at the PF). The repetition of the case markers semantically implies the distribution of locations which is applied to locations available to all accessible worlds. There is also an instance of verb reduplication in Bangla in example (5), / ށܩole ށܩole/ ‘swaying’, which implies repetition of an action, where the reduplicated verb forms quantify over an event/action.
3.2. Other Occurrences of Bangla Reduplicative Nominals In the previous examples, we have seen the reduplicative nominals in declarative sentences. There are other types of syntactic constructions where these reduplicative nominals occur. For example, interrogative sentences, modal sentences, etc. However, in some types of sentence categories, these reduplicative nominals are restricted. Following example sentences (11-16) present scrambling of the reduplication nominal phrase /akaše akaše/ ‘various skies’, which is presented in the previous example (2) and also provide adjacency tests with conjunction and negation. In addition to that, the conditions and constraints for the occurrence of these
Chapter Five
82
reduplicated nominals in Bangla are analysed through semantic interpretations of the reduplicative nominals in various sentence types. 11) * akaše ebܧƾ akaše megh koreche akaš -e
ebܧƾ
sky -loc. and
akaš -e
[Conjoined Reduplicated NPs] koreche
megh
sky -loc. cloud
spead.pres.hab.
“There are clouds in different skies.” 12) *akaše
akaše megh koreni
akaš -e
akaš-e
megh
sky -loc. sky -loc. cloud
[Negation]
koreni spead.pres.hab.Neg.
“There are clouds in different skies.” 13) *megh koreche megh koreche cloud
akaše akaše akaš -e
spead.pres. sky
[Extraposition]
akaš -e
-loc. sky
-loc.
“There are clouds in different skies.” 14) *akaše akaše ki megh koreni ? akaš-e
akaš-e
ki
[Interrogative with Neg]
megh
koreni
sky -loc. sky-loc. what cloud
spead.pres.hab.Neg.
“Are there no clouds in different skies?” 15) akaše akaš -e
akaše ki megh koreche?
[Interrogative]
akaš -e
koreche
ki
megh
sky-loc. sky-loc. what cloud
spead.pres.hab.
“Are there clouds in different skies?”
Semantics of Reduplicative Nominal Quantification in Bangla
16) hܧetࡧ o ei biota ghܧre ghܧre ache hܧetࡧ o
ei boi
-ta
ghܧr
83
[Epistemic Modality] -e
ghܧre
-e
ache
perhaps this book -cls. house –loc. house -loc. be.pres. “Perhaps, this book is in every home (usually).” The above examples suggest that these reduplicative nominals in Bangla can occur in positive tag questions and in some of the modal sentences (i.e. mood of possibility or necessity), as shown in (15) and (16). However, there are structural restrictions over the reduplicative locative NPs. These reduplicative nominals are composite predicates that do not generally occur in wh-sentences, except in tag questions. These reduplicated NPs also cannot occur with conjunction or disjunction, as shown in (11). The adjacency condition is followed when these reduplicated NPs occur in Bangla sentences. Moreover, they do not occur under the scope of negation (12). These reduplicative noun phrases have some other structural feature as well. In the example (16), it can be observed that under the scope of epistemic modal the optional sense of ‘typically’ disappears and the sentence instead implies the reading as ‘possibility’. [see example 6]. Some more notable properties are: (a) the case-marker or plurality marker on the constituent is fully reduplicated, (b) the reduplicative noun-phrases cannot occur as discontinuous constituents (example 11) and (c) they cannot be extra-posed from subject position (example 13).
4. Semantic Analysis According to Szabolsci (2013), compositionality in semantics refers to the notion that “The meaning of a complex expression is a function of the meanings of its parts and how they are put together” [pp. 153]. In keeping with this assumption, the complex components of the reduplicated composite noun phrases yield the complete interpretation, when the meanings of its parts are put together. The first part of this composite noun phrase contains restrictive quantification that relates to a proportional entity/things. In some specific expressions it suggests exclusiveness (exclusive of a kind/class) and definiteness (definite kind/set expression). In some other expressions, it encodes a habitual or generic sense. The reduplicated part of the composite NP gives the full interpretation of the expression, altogether. I argue that it is the second part of this composite
Chapter Five
84
NP that is selected by its identical first part in order to satisfy the distributiveness [+distributive] requirement. 17) akaše megh koreche akaš -e sky
koreche
megh
-loc. cloud
spead.pres.hab.
“There are clouds in the sky.” 18) akaše akaše megh koreche akaš -e
akaš -e
sky -loc. sky -loc.
megh cloud
[Distributive Plurality] koreche spead.pres.hab.
“There are clouds in different skies.” In the first example, the noun phrase, /akaše/ denotes restrictive quantification, where clouds are spread on the proportion of the sky that is visible to the speaker. In contrast, the reduplicative NP, /akaše akaše/ expresses the fact that clouds are spread in different skies that may not be visible to the speaker but are accessible to the speaker’s world. The implicit quantification ranges over the variable sky and the exact quantity is not specified either. The implication here is that the clouds may be spread both in the actual world and in any other possible world. Furthermore, the reduplicated NPs exclude the definiteness feature that is captured by a single locative NP, / ghܧre/ in the following example: 19) ei biota ghܧre ache ei
boi
-ta
ghܧre -e
ache
this book -cls. house -loc. be.pres. “This book is at home.” 20) ei biota ghܧre ghܧre ache ei
boi
-ta
ghܧr
-e
ghܧre -e
ache
this book -cls. house –loc. house -loc. be.pres.
Semantics of Reduplicative Nominal Quantification in Bangla
85
“This book is in every home (usually).” The first example (19) denotes that the book is at the speaker’s home or at someone else’s home (i.e. in a specific home). On the other hand, the second example in (20) implies that the book is usually present at all homes. The quantifier can range over a group of houses in the neighbourhood to (in principle, at least) all the houses in the world. Furthermore, these reduplicative nominals in subject position implicitly capture distributive plurality and sometimes also capture sequence. For example: 21) dܧݤne
dܧݤne
dܧݤn -e
dܧݤn-e
melate
gelo
mela -te
gelo
person –acc. person-acc fair-loc. go.pres.perf. “Many people have gone to the fair.” 22) dࡧ udܧݤn dࡧ udܧݤn kore egie asun dࡧ u-dܧݤn
dࡧ u-dܧݤn
kore egie asun
two person two person
seq. forward come.imp.hon.
“Come forward two by two.”(Imperative) Besides that, there is also the feature of reciprocity that is captured by these reduplicative NPs. The plural marker agreement on the addressee /tࡧ ora/ is also noticeable. In the following example, the reduplicated nominals capture exclusiveness, reciprocity and distributiveness features: 23) tࡧ ora bacchara bacchara khela kor
tࡧ o-ra baccha-ra
baccha-ra
[+exclusive, +collective] khela
you -plu. child -plu. child-plu. play
kor aux.imp.
“Play among yourselves. (only among children).” 24) tࡧ omra nije(ra) tࡧ om -ra
nije -ra
nije(ra) nije -ra
chobitࡧ a ãko [+reflexive] chobitࡧ a
ãko
Chapter Five
86
you –plu. 1st.self -plu. 1st.self-plu. picture
draw.imp.
“You all, draw the pictures yourselves.” 25) bacchara bacchara khela korche baccha -ra
baccha-ra khela
child -plu. child -plu. play
[+reciprocal,+collective] korche aux.prog.
“Children are playing (with each other).” 26) ei kܧthata ei
bacchara bacchara jane
[+distributive]
kܧtha -ta baccha -ra baccha -ra jane
This word-cls.
child-plu.
child-plu. know.pres.hab.
“All (most of) children know this word/saying.” 27) chelera chelera khela korche chele-ra chele-ra
khela
boy-plu. boy-plu. play
[+distributive] korche aux.pres.prog.
“Boys are playing (with each other).” The semantics of these reduplicative nominals is noteworthy for analysis. In Bangla, the reduplicative nominals are fully pronounced, indicating their significance for the semantic interpretations of the sentences. In some other languages for example, Hindi, the case markings on the reduplicative nominals (for example, /ghԥr ghԥr mӁ/ ‘in houses’) do not reduplicate, whereas in Dutch, the identical and non-identical doubling of wh-pronouns can be observed, (for example, /Wie denk je wie ik gezien heb?/ ‘Who do you think I have seen’) [Barbiers, 2008. pp. 14]. Moreover, Barbiers (2008) argues that Swiss German allows reduplication of definite and indefinite articles, (for example, /de vil de schooner garte/ ‘the much the nicer garden’) [pp.5]. Barbiers also cites Norwegian examples of optional syntactic reduplication of proximal and distal demonstratives (for example, /den her-re (her) boka (her)/ ‘here this book here’). In Tibeto Burman languages (for example, in Meiteilon /nupinupi-g~i / ‘only ladies’) the PF copy gets deleted in the LF. However, in Bangla, the
Semantics of Reduplicative Nominal Quantification in Bangla
87
occurrence of reduplicative NPs is not optional, since deletion of one noun alters the meaning of the sentences. Thus, they are semantically more loaded and are an excellent strategy to encode quantification of certain kinds. Plurality in languages is expressed in two ways: (i) collective plurality, and (ii) distributive plurality. For collective plurality, the range of variable is limited or restricted by clausal conditioners. In English, words like marry, couple, together etc. express collective plurality in sentences. Distributive plurality denotes the possible range of the variable over all or most possible worlds. There are many strategies that are used by languages to encode distributivity. In Bangla, as we have already analyzed, distributiveness is captured by reduplicating the nominals with their grammatical markers. However, not all kinds of reduplicative nominals encode distributive plurality. They express the feature of exclusiveness, intensity, reciprocity, reflexivity etc. We can observe them in the previous data. From sentences containing reduplicative nominals that encode distributivity, we get the distributive reading that is not present in their non-reduplicative occurrence. The existing literature on distributivity examines the distributive operator that captures temporal distributivity, event distributivity and to some extent intensification of an action. I would claim that reduplicative nominals in Bangla are strategically used to encode implicit plural quantification and distributive plurality. The general notion of placement of overt or covert plural marker to denote quantification is averted in these expressions. To give a formal account of the semantics of (standard) quantification, the quantifier operators: and are used. The distributive readings are formalized by the existential quantifier operator . Even though the variable x can range over different accessible worlds, the distributive readings denote implicit or covert quantification. They cannot scope over in the universe of discourse. I argue that both of the reduplicative nominals denote similar variables with similar properties and are presented as a united whole or composite constituent. I argue that the first nominal selects another identical copy to express the additional semantic feature of distributive plurality. Therefore, the semantics of distributive plurality in Bangla are expressed by the reduplication of nominals. The analysis of the reduplicative nominals can be formalised as follows:
Chapter Five
88
ghܧre
ache
ghܧre-e
ache
28) ei boita ei boi -ta
this book-cls. house-loc. be.pres. “This book is at home.” Let F= {a1, a2, a3...an } be the total set of houses in the given context. x (xكF)ٿa{(aiאx) ĺ bookȚ(ai)] where i= 1,2,3... BookȚ= a definite book ai can denote the i-th houses or the i-th sub kind of houses. There is a book (i.e. a specific book) that is present or available at each house (usually). 29) chelera chele-ra
chelera
khela korche
chele -ra
khela korche
boy-plu. boy-plu.
play
aux.pres.prog.
“Boys are playing (with each other).” e[ x (xؿBoys)ٿai(aiאx) ĺ Play (ai,e)] where i= 1,2,3... ai can denote the i-th quantity of boys. Here, the reduplicated quantifiers encode the distributive sense of the events. The reduplicative NP /chelera chelera/ corresponds to the distributive /each/ in English. The distributivity is encoded by the reduplicative nominals. The nominals reduplicate to encode the implicit plurality. The distributivity is the property of these reduplicative nominals that quantifies over the predicate. These reduplicative nominals express existential quantification and distributive plurality that is encoded implicitly as a property P of the composite reduplicated NPs.
Semantics of Reduplicative Nominal Quantification in Bangla
89
4. Conclusion To sum up, I have tried to argue that implicit quantification in Bangla reduplicated noun phrases is achieved by reduplicative morphosyntax. The reduplicative nominals express not only distributive plurality but also other semantic features that are essential for full interpretation of the sentence. The quantification is understood without having any overt presence of plural marker or quantifier. Moreover, the distributiveness feature is also implicitly encoded within these reduplicative NPs. Therefore, such implicit quantification is a significant feature of quantification in Bangla.
References Abbi, Anvita. Reduplication in South Asian Languages. An Areal, Topological and Historical Study. (New Delhi: Allied Publishers, 1992). —. “Reduplication in Tibeto Burman Languages of South Asia.” Southeast Asian Studies 28.2 (1990). Bhattacharya, Nandini. “Semantics of Quantification in Bangla”. P-2 Term Paper, Department of Linguistics: University of Delhi. 2014. Barwise, Jon and Cooper, Robin. “Generalized Quantifiers and Natural Language”.Linguistics and Philosophy 4 (1981): 159 -219. Bach, Emmon, Eloise Jelinek, Angelika Kratzer, and Barbara Partee.Quantication in Natural Languages (ed.) (Dordrecht: Kluwer, 1995). Bagchi, Tista. “Quantification, Negation, and Focus: Challenges at the Conceptual-Intentional Semantic Interface”. Paper presented at Department of Linguistics, Aligarh Muslim University, February 2527, 2011. Balusu, Rahul. “Distributive reduplicationin Telugu”.In Proceedings of NELS 36, ed. Christopher Davis, Amy Rose Deal, and Youri Zabbal. (Amherst: GLSA. 2005): 39–53. Barbiers, Sjef. “Microvariation in Syntactic Doubling, an Introduction.” In Microvariation in Syntactic Doubling. Syntax and Semantics36, ed. Barbiers, S. Koeneman, O. Lekakou, M. and van der Ham, M. (Bingley: Emerald. 2008): 1-36. Beghelli, Filippo and Tim Stowell. “Distributivity and Negation: The Syntax of Each and Every”. In Ways of Scope Taking, ed.Anna Szabolsci. (Dordrecht: Kluwer.1997): 71–107.
90
Chapter Five
Biswas, Priyanka. Bangla Associative Plural -ra: A Cross-linguistic comparison with Chinese men and Japanese –tachi”. In Proceedings of the 31st West Coast Conference on Formal Linguistics, ed. by Robert E. Santana-LaBarge.(Somerville, MA:Cascadilla Proceedings Project 2013): 56-65. Brisson, C. “Distributivity, Maximality, and Floating Quantifiers” (PhD Thesis. Rutgers University, 1998). Chierchia, Gennaro and Sally, McConnell-Ginet.Meaning and Grammar: An Introduction to Semantics. (Cambridge: The MIT Press. 2000). Chierchia, Gennaro. “Reference to kinds across languages.” Natural Language Semantics 6. (1998): 339-405. Dayal, Veneeta. “The Syntax of Scope and Quantification.” In Cambridge Handbook of Generative Syntax, ed. Marcel den Dikken. (Cambridge: Cambridge University Press. 2013). Dayal, Veneeta. “Bangla Plural Classifiers”.Language and Linguistics 15.1 (2014): 47-87. Emeneau, Murray Barnson. “India as a Linguistic Area”. Language 32.1. (1956): 3–16. Heim, Irene and Angelika Kratzer. Semantics in Generative Grammar. (Oxford: Wiley-Blackwell.2000). Kennedy, Christopher. “Antecedent-Contained Deletion and the Syntax of Quantification”.Linguistic Inquiry 28.4. (1997): 662-688. Lasersohn, Peter. Plurality, conjunction and events. (Dordrecht: Kluwer Academic Publisher. 1995). Lasersohn, Peter. “Generalized distributivity operators”. Linguistics and Philosophy 21. (1998): 83- 92. May, Robert. “The Grammar of Quantication”. (PhD dissertation, MIT, 1977). McCawley, James D. Everything that Linguists have Always Wanted to Know about Logic* (*but were ashamed to ask). (Chicago: University of Chicago Press. 1981). Montague, Richard. “The proper treatment of quantification in ordinary English”.In Approaches to Natural Language, ed. K.J.J. Hintika, J.M.E. Moravcsik and P. Suppes. (Dordrecht: D.Reidel.1973): 247-270 Portner, Paul and Barbara Hall Partee, ed. “Formal Semantics: The Essential Readings”. (UK: Blackwell Publishing Ltd. 2002). Partee, Barbara H. Compositionality in Formal Semantics. (Oxford/Malden/Victoria:Blackwell Publishing 2004). Szabolsci, Anna. “The Syntax of Scope”. In The Handbook of Contemporary Syntactic Theory, ed. Mark Baltin, and Chris Collins. (MA/ Oxford: Blackwell Publishing.2001): 607-633.
Semantics of Reduplicative Nominal Quantification in Bangla
91
—. Quantification. (Cambridge: Cambridge University Press.2010). —. “Quantifier particles and compositionality”. Appeared at 19th Amsterdam Colloquium paper. (2013). Winter, Yoad. “Functional Quantification”.Research on Language and Computation 2.(2004): 331-363. Winter, Yoad and Remko, Scha. “Plurals”. In Handbook of Contemporary Semantic Theory, ed. Shalom Lappin and Chris Fox. (John Wiley & Sons Ltd: Chichester, UK. 2015): 77-113.
CHAPTER SIX THE ALLOMORPHS OF GENITIVE AND DATIVE AND THE POSTULATION OF GRAMMATICAL GENDER IN MALAYALAM1 GOUTHAMAN K. J.
Abstract In Malayalam, a Dravidian language spoken principally in Kerala in the Southern part of India, the Genitive Case has two different morphological forms. Similarly, the Dative Case also has two different morphological forms. In this paper, the attempt is to explore the motivations behind the different realizations of Genitive Case and Dative Case in Malayalam. The question is whether this allomorphy is phonological or grammatical. I argue that the different morphological forms represent the agreement of the Case head with different sets of Nominal features (singular/plural and masculine/feminine). In other words, each affix, previously considered as a Case morpheme, is understood as a portmanteau of the Case head and the agreement morpheme. However, the anomalous behaviour of Inanimate Nouns, which take either of the two allomorphs in both Genitive Case and Dative Case, is a problem for this analysis. In this regard, I suggest that Malayalam has grammatical gender which is semantically uninterpretable. The motivations for the restriction of this phenomenon of
1
Gen 1= One allomorph of Genitive Case in Malayalam [-inte], Gen 2= Another allomorph of Genitive Case in Malayalam [-uDe], Dat 1=One allomorph of Dative Case in Malayalam, Dat 2=Another allomorph of Dative Case in Malayalam, NP= Noun Phrase, DP= Determiner Phrase, KP= Case Phrase, K=Case Head, Gen= Genitive Case Head, Dat= Dative Case Head, N=Noun, Head of NP, D=Head of DP, Det=(Lexical) Determiner, Spec=Specifier, Infl=Inflection Head, Masc=Masculine Feature, Fem= Feminine feature, Pl=Plural feature,NOM=Nominative Case, GEN=Genitive Case, REL= Relativiser, EX= Existential Copula
Chapter Six
94
allomorphy to Dative Case and Genitive Case will have to be explored separately in further research. Keywords: Genitive Case, Allomorphs, Agreement, Dative Case, Nanosyntax, Grammatical Gender
1. Two Forms of the Genitive Case The Genitive Case in Malayalam has two forms— (in)te and (u)De, which I shall refer to as Genitive 1 (Gen 1) and Genitive 2 (Gen 2). 1) raajav-inte
aana
king-Gen1 elephant ‘The king’s elephant’ 2) RaaNi-yuDe aana queen-Gen2 elephant ‘The queen’s elephant’ Firstly, I explore the possibility of a phonological explanation that the two realizations (-inte and -uDe) are phonologically conditioned allomorphs. On examining data, it appears that the Genitive 2 (-uDe) is restricted to stems ending with vowels /a/, /ܼ/ and /iޝ/ and consonants /r/, /ݐ/ and /ܿ/. Genitive 1 (-inte) seems to be the unmarked option because it appears in all the other cases. Example (2) is an instance of a stem ending with /ܼ/ taking Gen 2. Examples (3) and (4) are stems ending with /a/ and /iޝ/ which take Gen 2, and hence prohibit Gen1. The examples (5) and (6) are instances of stems ending with /ܿ/ and /r/ which take Gen 2. 3) amma-yuDe kuTTi mother- Gen 2 child “mother’s child” 4) tii-yuDe niRam fire- Gen 2 colour
Grammatical Gender in Malayalam
95
“The colour of the fire” 5) paTTi-kaL –uDe lookam dog- Pl –Gen2 world “The world of dogs” 6) ݝeela-kkaar2-uDe
eNNam
work-people –Gen 2 number “The number of servants” We can argue that /a/, /ܼ/ and /iޝ/ are vowels and /r/ and /ܿ/ are vocalic consonants and hence motivate two different phonological environments where Gen 1 cannot appear. However, this phonological explanation is insufficient because of various reasons, which is what I will discuss in the next sub-section.
1.1 Problems with the Phonological Explanation The first empirical problem is regarding stems ending with non-retroflex lateral /l/. These stems take Gen 1. 7) kall –inte
niRam
stone- Gen 1 colour “stone’s colour” Since there is no feature shared by /r/ and /ܿ / that distinguishes /l/ from them, this is a problem. /r/ and /ܿ / do not form a natural class that excludes the non-retroflex lateral /l/.
2
The affix [-kkaar] in this example denotes “people”. The affix is unspecified for gender. Since the nominal denotes a collection of individuals who are either male or female, there is a null plural morph in the DP. When marked for gender with [an] affix or [-i] ending, the nominal becomes singular, in which case an overt affix denoting plurality can be attached above the gender suffix to denote a collection of individuals who are either male (with [-an] suffix) or female (with [-i] ending). This reveals other interesting facts, which will be discussed in section 5.
Chapter Six
96
Secondly, words ending with /r/ and /ܿ/ segments take Gen 1 when these stems are without internal structure. That is, monomorphemic stems ending with /r/ and /ܿ/ segments take Gen 1. Note that the earlier examples (5) and (6), where stems ending with /r/ and /ܿ/ segments prohibited Gen 1 and allowed Gen 2 to surface, were actually words with internal structure. Both the stems had an affix denoting plurality and this crucially distinguishes them from stems in the examples (8) and (9) below. 8) kaaR-inte uDama car- Gen 1 owner “car’s owner” 9) ݝaaL-inte sword- Gen 1
piDi hilt
“sword’s hilt” This is not an exception confined to monosyllabic words either. Stems with more than one syllable, ending with /r/ and /ܿ/ segments, also take Gen 1, if they are monomorphemic. 10) karaL-inte pRacnam liver- Gen 1 problem “problem of the liver" This is where an alternative explanation is required. I argue that it is the [aܿ] suffix (inanimate plural morpheme) and not /ܿ/ segment that triggers Gen 2. Similarly, /r/ ending and vowel ending are also morphological facts. Not all nouns ending with /r/ take Gen 2. Those nouns which end with [-ar] or [-maar] suffix, denoting animate plurality, take Gen 2. Although the vowel-ending (ending /a/, /ܼ/ or /i:/) is semantically uninterpretable when present on inanimate nouns, it denotes feminine Gender when present in the class of human-denoting nouns and with animate non-human nouns which make a masculine/feminine gender distinction. Hence, each of these affixes can be morphologically analysed and broken down into features. This is integral for an alternative explanation of the phenomenon. Before doing that in a formal and
Grammatical Gender in Malayalam
97
systematic manner, let us look into the paradigms of Genitive Case in Malayalam, in pronouns and common nouns.
1.2 The Pronominal Paradigm of Genitive Case Table 1
Singular
Plural
1st Person en-te
2nd Person ni-nte
3rdPerson, Masculine avan-te
3rdPerson, Feminine avaL–uDe
Gen 1 njanga LuDe
Gen1 ninga LuDeG en 2
Gen 1 avanmaaruDe
Gen2 avaLmaar -uDe
avar-uDe
Gen 1 ava-uDe
Gen 2
Gen 2
Gen 2
Gen 2
3rdPerson, Animate
3rdPerson, Inanimate at-inte
Table 1 has the complete paradigm of Pronouns in Malayalam in Genitive Case. The table clearly helps us to make certain generalizations. They are: 11) Gen 1 is prohibited with all the plural pronouns, regardless of the person and gender features of the stem. These pronouns take Gen 2. 12) Gen 1 is prohibited with 3rd Person, Feminine pronoun. Gen 2 is allowed. 13) 1st person singular pronoun, 2nd person singular pronoun, 3rd person singular masculine pronoun and 3rd person singular inanimate pronoun take Gen 1. These pronouns have the default features.
1.3 The Nominal Paradigm of Genitive Case Table 2
Singular
Masculine 1) adhyaapak-an- te Teacher- Masc- Gen 1
Feminine 1) adhyaapika- uDe Teacher- Fem- Gen 2
2) sahodar-an –te Sibling-Masc-Gen 1
2) sahodari- yuDe Sibling- Fem- Gen 2
98
Plural
Chapter Six
1) adhyaapak-an- maar- uDe Teacher- Masc -Pl -Gen2
1) adhyaapika- maar- uDe Teacher- Fem- Pl- Gen2
2) sahodar-an-maar-uDe Sibling-Masc-Pl- Gen2
2) sahodari-maar-uDe Sibling-Fem-Pl-Gen2
Table 2 shows examples of Animate Nouns with gender specifications in the Genitive Case. It shows that the Singular, Masculine Noun takes Gen 1 and the Singular, Feminine Noun takes Gen 23. The Plural forms take Gen 2, regardless of the gender specifications. Table 3 strengthens this generalisation. The animate noun “adhyaapakar” is a collective noun, which denotes a collection of teachers who may be male or female. The semantically interpretable gender is underspecified. Yet, the plural feature triggers Gen 2. Table 3 Animate Singular Plural
adhyaapak- ar- uDe Teacher- Pl- Gen 2
Table 4 poses a problem for the generalizations we made till now. So far, we saw that a particular noun class permits either particular form of Genitive Case, Gen 1 or Gen 2. Nevertheless, Inanimate Nouns (and Nouns denoting non-human animals unspecified for gender) arbitrarily take either Gen 1 or Gen 2, when singular. The nouns in the shaded area exemplify this anomaly.
3
In northern dialects of Malayalam, feminine common Nouns do not take Gen 2. Instead, the vowel undergoes lengthening and takes Gen1. Eg: ammee-nte kuTTi Mother- Gen1 child “Mother’s child” This may be related to the use of [-e] affix in the place of [-uDe] as Gen 2 in northern dialects. Eg: Avar-e kuTTikal They-Gen 2 children “Their children”
Grammatical Gender in Malayalam
99
Table 4
Singular
Plural
Neuter/Inanimate 1) puuݝ- inte Flower- Gen 1 2) meca- uDe Table – Gen 2 1) puukkaLFlower- Pl- Gen2 2) meca- kaL- uDe Table- Pl- Gen 2
uDe
Even here, it is important to note that no particular noun can take both Gen 1 and Gen 2. They are not in free variation: on the other hand, some inanimate nouns take Gen 1 while some others take Gen 2. 14) *puuv- uDe Flower- Gen 2 15) *meca-inte Table- Gen 1 Besides, even in the noun class of inanimate nouns, the plural forms take only Gen 2 and not Gen1. Hence, we can make certain generalizations based on the data regarding common nouns. The generalizations are as follows: 16) In all noun classes, Plural forms take Gen 2 and prohibit Gen1. 17) In noun classes, where gender is specified and is Feminine, Gen 1 is prohibited and Gen 2 is allowed. 18) In noun classes, where gender is specified and is Masculine and Number is singular, Gen 1 is allowed and Gen 2 is prohibited. Comparing (11), (12) and (13) with (16), (17) and (18), we find that we have got a pattern, in which Plural feature and Feminine feature triggers the appearance of Gen 2 and Masculine feature triggers the appearance of Gen 1. However, there are grey areas and irregularities such as Inanimate, Singular Nouns and First Person singular pronoun and Second Person
Chapter Six
100
singular pronoun. I will deal with these problems in a later section. In the next section, syntactic motivations for the allomorphy are explored.
1.4 Analysis: Agreement of K head with the NP Argument The generalizations made in sections 1.2 and 1.3 point towards the possibility of positing a feature-based analysis of the phenomenon of allomorphy in the Genitive Case in Malayalam. Before evaluating the possibility of a feature-based analysis, we need to sketch what the genitive construction in Malayalam looks like. I adopt the proposal of Abney (1987) that nominal expressions are headed by functional categories. According to Abney (1987), when the nominal expression is a possessive construction, the head of the construction is the possessive clitic (‘s). This head is called the D head and the nominal expression a DP. The structure of the DP was of the form: [DP NP [‘s [NP]]]. The NP in the Spec of the D head is the possessor argument and the NP in the complement position is the possessed argument. The motivation for this argument came from observing the similarities between the nominal expression and the sentential expression. In other words, the analysis of a sentence as IP inspired the analysis of the nominal expression as DP. According to the DP hypothesis, it could be argued that the head of a possessive construction is the Genitive Case, which is the Case head K (the head D according to Abney’s DP hypothesis). The structure of the KP would be [KP NP [K [NP]]]. This structure may be used to represent Malayalam possessive constructions, as in (19). 19) [KP John [inte [ pustakam]]] “John’s book” However, whether the whole possessive construction is a KP (with a DP as its complement) or a DP (with a KP in its Spec) is not very relevant for the puzzle I explore here. In Malayalam, the Genitive Case K agrees with the possessor NP and not the possessed NP. For example, in (20), Gen 2 is used, irrespective of the features of the possessed NP, because the possessor NP is feminine. 20) makaL- uDe makan/ makaL daughter-Gen2 son/daughter
Grammatical Gender in Malayalam
101
“daughter’s son”/ “daughter’s daughter” Hence, the agreement of K head with the possessor NP argument, which we are analysing here, is within the domain of KP regardless of the two possibilities, which means we do not have to choose either of the two for our analysis. The Spec-head relation between possessor NP argument and K head is the crucial part. The various analyses of the Subject-Verb agreement motivate a Spec-head relation between the probe and the goal. The nominal expression denoting subject argument is in the Spec of the Infl head (the Tense head). Similarly, we can argue for an agreement between the Nominal expression in the Spec of the K head with the Case head K. This is what I propose as the reason for the allomorphy in Genitive Case in Malayalam. Due to the agreement between the Case head K and the possessor NP argument in its Spec, some features of the nominal expression are copied onto the Case head K. Gen1 seems to be the realization of the Case head K plus Masculine feature copied from NP in [Spec, KP], while Gen 2 seems to be the realization of the Case head K plus Feminine and Plural features copied from NP in [Spec, KP]. However, in an account based on feature-sharing, we can argue that the features of the NP inside KP get copied onto the K head in Malayalam, resulting in two morphological forms, one (Gen 2) corresponding to Plural and Feminine features and the other (Gen 1) corresponding to default or unmarked features. This means that Gen1 appears whenever the NP does not have Plural or feminine features. Gen 1 does not necessarily have to appear with Masculine, Singular NPs. This is exactly what we saw in Malayalam data except for a section of Inanimate Nouns which take Gen 2. The intuition that Feminine feature and Plural feature are marked features compared to Masculine and Singular features is not new. Harley and Ritter (2002) argue that every node denoting a type of feature has a default value. For Number, Singular is the default value and for Animacy, Masculine is the default value. Gen 2 is therefore the option with marked features. Gen 1 is the option with NPs having default features. Thus, a version of the DP hypothesis and an account of feature sharing help us to understand the distribution of two forms of Genitive Case in Malayalam to an extent. In the next section, we examine the two forms of Dative Case and analyse whether the generalisations regarding Dative Case can be explained under the framework I have sketched out.
Chapter Six
102
2. Two Forms of the Dative Case In this section, my arguments about different realizations of the Genitive Case are extended to analyse the allomorphy in the Dative Case. The Dative Case has three realizations in Malayalam— [-kkԥ], [-ykkԥ] and [ԥ]4. It is observed that the distinction between the first two is entirely motivated by phonological factors. The distinction is non-existent in some dialects as well. We can put the first two under Dative 2. The [-ԥ] affix can be called Dative 1. The Dative experiencer sentences below exemplify the two morphological forms of Dative Case. 21) avan-ԥ
pani uNTԥ
He- Dat 1 fever EX “He has fever” 22) avaL-kkԥ pani
uNTԥ
She- Dat 2 fever EX “She has fever”
2.1 The Pronominal Paradigm of Dative Case Table 5 1st Person
2nd Person
3rd Person, Masculine
3rd Person, Feminine
3rd Person, Animate
Singular
eni-kkԥ
avan- ԥ
avaL– kkԥ
Plural
Dat 2 njangaLkkԥ
ninakkԥ Dat 2 ningaLkkԥ
Dat1 avanmaarkkԥ
Dat2 avaLmaarkkԥ
avarkkԥ
Dat2
Dat 2
Dat 2
Dat2
Dat 2
3rd Person, Inanimate at-in-ԥ Gen1Dat1 ava-ykkԥ
Dat 2
4
This paper does not deal with the affix [-ooDԥ], which is argued to be a Second Dative in some accounts of Malayalam. Whether it is Second Dative or Sociative or Comitative does not affect my argument regarding the Dative allomorphy in this paper.
Grammatical Gender in Malayalam
103
The pronominal paradigm of Datives in Malayalam is given above in Table 5. Dat 1 can appear with all the pronouns which take Gen 1 other than 1st person and 2nd person pronouns. In other words, 3rd Person, Masculine pronoun and 3rd Person, Inanimate pronoun take Dat 1. Dat 2 appears with all the plural pronouns and the 3rd Person, Feminine pronoun, exactly like Gen 2. In addition, Dat 2 expands its domain into first person and second person pronouns as well.5
2.2 The Nominal Paradigm of Dative Case An interesting fact is noted on examining the nominal paradigm. The stems which take Dat 2 are identical to the stems which take Gen 2. In other words, Dat 2 appears with all the plural nouns and feminine nouns. Dat 2 also appears with Inanimate Nouns which take Gen 2. Similarly, stems which take Dat 1 are the same as those that take Gen1. These facts are true with respect to the paradigm of common nouns, proper nouns and nominal expressions derived from relative clauses. Table 6 shows data about Animate Common Nouns in Dative Case. Table 6
Singular Plural
Masculine adhyaapak- anԥ Teacher- Masc- Dat 1
Feminine adhyaapika-ykkԥ Teacher- Fem- Dat 2
adhyaapak-an- maar- kkԥ Teacher- Masc- Plural- Dat 2
adhyaapika- maar- kkԥ Teacher- Fem- Plural- Dat2
In addition to the fact that the allomorphs of Dative pattern with allomorphs of Genitive, the data also reveals another interesting fact. Dat 1 can be realized as [-inԥ] in 3rd Person, Inanimate nominal expressions (example (a) in the table below). This involves Case stacking according to the previous study on Datives in Malayalam (Jayaseelan, 2013). This means that [-inԥ] is Dat 1 containing Gen 1. If this is correct, it is interesting how Dat 1 involves stacking while Dat 2 does not, and how Dat 1 contains a part of Gen 1 and not Gen 2. The fact that Dat 1 contains Gen 1 and not Gen 2 strengthens my argument that Dat 1 and Gen 1 appear with the same set of Nominal features.
5
First person pronouns can take Dat 1 in some social dialects.
Chapter Six
104
Table 7 Neuter/Inanimate Singular
Plural
a) puuv- in -ԥ Flower- Gen 1- Dat 1 b) meca- ykkԥ Table – Dat 2 c) puukkaL- kkԥ Flower- Pl- Dat2 d) meca- kaL- kkԥ Table- Pl- Dat 2
Animate
adhyaapak- ar- kkԥ Teacher- Pl- Dat 2
Dat 1 can appear only along with Gen 1 and not with Gen 2 because Dat 1 and Gen 1 are forms that realize agreement of K with the same set of default N features. Dat 2 and Gen 2 are realizations of agreement of K with sets of marked features. These sets need not necessarily match. In fact, the mismatch might be the reason for the use of Dat 2 with 1st and 2nd Person pronouns. Nevertheless, all the four allomorphs correspond to the relevant Case plus a copy of features of the nominal. Hence, we can conclude that the different realizations of Dative Case as well as Genitive Case are motivated by the agreement of the K head with the argument in its Spec; and the regularity of the pattern with respect to both Cases suggests that our feature-sharing account is useful. In the next section, I will attempt to systematise and elaborate the argument by incorporating the machinery of nanosyntax with the feature-sharing account.
3. Nanosyntactic approach towards the Puzzle The intuition of Nanosyntax that lexical items correspond to syntactic subtrees helps us to assume that no feature needs to be spelt out independently. The lexicon can store whole sub-trees. Hence, the K head and the copy of the nominal features of the nominal expression in its Spec can be spelt out together as one lexical entry. The set of nominal features copied onto the K is a copy of the set of nominal features of the NP in its Spec. Hence, the structure [KP [NP N Masc] K] implies that there is a copy of the Masculine feature attached to K, resulting in the structure [KP[NPN Masc] [K Masc]]. Similarly, the structure [KP[NPN Pl Fem Masc] K]
Grammatical Gender in Malayalam
105
implies that the K contains a copy of the set of nominal features {Plural, Feminine, Masculine}.6 In both the cases of Genitives and Datives in Malayalam, there are two competing lexical entries each, corresponding to two sets of nominal features. All the four allomorphs correspond to the relevant Case plus a copy of features of the nominal expression. Given below are the possible syntactic sub-trees corresponding to each allomorph. Gen 1 ļ [K Gen[Masc]] Gen 2ļ [K Gen[[Fem Masc] Plural]] Dat 1ļ [K Dat [Gen[Masc]]] Dat 2ļ [K Dat [Gen[[ Fem Masc] Plural]]] Now, let’s look at the principles that determine the distribution of the allomorphs of Genitive Case. By the Superset Principle, a lexical item gets pronounced even when the syntactic node does not have all the features of the lexical item. The Superset Principle: “A vocabulary item A associated with feature set F can replace a sub-tree X with feature set F’ if and only if F is a superset of F’” (Taraldsen, 2009). Hence, Gen 2 appears when the NP is either Feminine or Plural or both Feminine and Plural. It is not necessary that the NP be both Feminine and Plural for Gen2 to surface. The feature set of the syntactic node K is a subset of the feature set of the lexically stored tree Gen 2. Similarly, the
6
A Noun becomes Masculine by merging with a feature that makes it Masculine. This feature is realized as a suffix when N moves above Masculine. Similarly, Feminine specification is brought about by Noun climbing above both Feminine and Masculine features. In other words, a Noun marked for Feminine contains both Fem and Masc features. This captures the status of Masc as less marked. It is possible that a Masculine Noun, in turn, contains an animate feature and an Animate Noun an inanimate feature and so on. In this way, I suggest a combination of the feature hierarchy of Harley and Ritter (2002) with syntactic operations of Merge and Move.
Although Plural is above the other heads, the surface structure in Malayalam reflects an order in which the Plural suffix follows the gender suffix. This might be because of the pied-piping of Feminine and Masculine nodes along with the Noun before spell-out. This results in the following structure: [KP [NP N [Fem Masc] Plural] K], when the NP is Feminine and Plural.
106
Chapter Six
appearance of Gen 2 with Masculine, Plural Nouns can also be explained, Masculine being a feature contained in the sub-tree corresponding to Gen 2. But, if this is so, Gen 2 must be able to appear even when NP is Masculine, Singular, since [KGen[ Masc]] is also a sub-set of the features corresponding to Gen 2. The Elsewhere Condition (Minimise Junk Principle in Nanosyntax) prevents this. According to this principle, the most specific wins. In other words, since the insertion of Gen2 onto a syntactic node, which does not contain a Feminine or Plural feature, leaves these nodes in the Gen2 unused, it is prevented by the insertion of Gen1, which does not leave any node of the lexical item unused. Gen 1 is prohibited with Feminine and Plural Nouns, because it does not contain those features. Application of the same logic of “Minimise Junk” should derive the insertion of Gen 1 in all Inanimate, singular Nouns. Since Inanimate is the default feature under the Class node (Harley &Ritter, 2002), the node of a K always contains a copy of an inanimate feature, under agreement with NP in its Spec. Hence, the sub-trees corresponding to both Gen 1 and Gen 2 contain the Inanimate node. For insertion, Gen 1 must be the preferred option over Gen 2 since only one node in the Gen 1 (the Masculine feature) is unused by the syntax. However, the facts show that some Inanimate Nouns take Gen 2, even if Gen 1 is the “most specific” option for the K with Inanimate NPs. Thus, this anomaly is a problem for the nanosyntactic analysis as well. The allomorphy in the Dative Case can also be explained in similar terms under the nanosyntactic account. Dat 2 is the unmarked option, because it is the “biggest” and corresponds to [K Dat [Gen[[ Fem Masc] Plural]]]. Hence, it appears whenever the K contains a copy of Feminine feature or Plural feature. But it does not appear with Masculine NPs because then the K contains only the copy of Masculine feature. The most specific lexical item for insertion would then be Dat 1 which corresponds to [K Dat [Gen[Masc]]], in which case there is also the option of spelling out [Gen[Masc]] as Gen 1 and Dat as Dat 1 separately. This is how Case stacking operates. Again, the anomaly of the Inanimate Nouns that take Dat 2 arises. The next section is an attempt to account for this problem.
Grammatical Gender in Malayalam
107
4. The Anomaly of Inanimate Nouns: Gender in Malayalam The class of Inanimate Nouns appears to be an exception to the analyses in sections 1.4 and 3. According to the feature-sharing account, all Inanimate Nouns should take Gen1 and Dat 1, since they are the allomorphs corresponding to default features. Gen 2 and Dat 2 should be allowed only with marked features such as Animate, Feminine and Plural. Yet, some inanimate Nouns take Gen 2 and Dat 2 in Malayalam. As per the nanosyntactic account, the most specific lexical item for inanimate nouns cannot be Gen 2 or Dat 2.
4.1 Phonological Explanation In this section, a phonological explanation of the anomalous class of Inanimate Nouns is attempted. All inanimate nouns ending with consonants take Gen 1, in the case of Genitive Case, and Dat 1, in the case of Dative Case. When the inanimate noun ends with a schwa, the schwa is deleted so that Gen 1 and Dat 1 can be attached to the preceding consonant. When the stem ends with vowels/a/, /ܼ/ and /iޝ/, it takes Gen 2, in the case of Genitive Case, and Dat 2, in the case of Dative Case. Hence, we can argue that /a/, /ܼ/ and /iޝ/ vowels trigger phonological conditioning. Although this analysis based on phonologically conditioned allomorphy seems to have solved the problem, it raises serious questions. First, why is it that the grammatically conditioned allomorphy in other Noun classes and the phonologically conditioned allomorphy in Inanimate Noun class result in the same surface realizations—Gen1 and Gen 2 in the case of Genitive Case and Dat1 and Dat2 in the case of Dative Case? Secondly, why is the vowel ending triggering Gen 2[-uDe] and Dat 2 [-kkԥ], which are very different phonological units? In other words, how do we account for the fact that the same Inanimate Nouns that take Gen 2 take Dat 2? These problems are adequate reasons to abandon the analysis based on phonological conditioning. Therefore, a morphosyntactic explanation of the anomaly of Inanimate Nouns is attempted in the next section.
4.2 Gender in Malayalam In order to develop the morpho-syntactic analysis, firstly, we need to break down the category of Inanimate Nouns in Malayalam. Gender in
108
Chapter Six
Malayalam is considered as semantically interpretable. However, all Animate Nouns are not specified as Masculine or Feminine. All humandenoting nouns and some animal-denoting nouns are specified for Masculine and Feminine. Other animal-denoting nouns, which are not specified for gender distinction, pattern with inanimate nouns in Malayalam despite having semantically interpretable animacy. The distinction between these two different kinds of animal-denoting nouns is shown below. Table 8
Human-denoting Animate Noun Animal-denoting Animate Noun
Masculine
Feminine
a) kuLL-an dwarf-Masc c) kuRukk-an fox-Masc
b) kuLL-athi dwarf- Fem d) kuRukk-athi fox -Fem
Table 8 shows that (a) and (b) behave exactly like (c) and (d) and vice versa. Even the same affixes denoting Masculinity and Femininity are used. Hence, they belong to the same class of nouns. But in the example below, although the word denotes an animate being, the word behaves like an inanimate noun because it is not specified for gender. Furthermore, the stem does not have a counterpart “aan-an” or any such term denoting male elephants in particular. Hence, the generic term for ‘elephant’ can be grouped along with inanimate nouns. Table 9 Animal-denoting Inanimate Noun
Aana elephant
The data, therefore, suggests that more than the semantically interpretable animacy, it is morphologically attested gender specification that categorises Malayalam words. Following this intuition, both humandenoting animate nouns and animal-denoting animate nouns can be put together under the class of nouns that make a morphological gender distinction. Let us assume that this class of nouns has the feature ‘Gender’. All other nouns, whether they denote animals (semantically animate) or things (semantically inanimate), belong to the class that does not make a gender distinction since they lack the feature ‘Gender’. Hence, in addition
Grammatical Gender in Malayalam
109
to the Animate/Inanimate distinction, we have a Gendered/ Non-gendered distinction in Malayalam. I will argue that it is the Gendered/Nongendered distinction which participates in the morphology and accounts for the allomorphy in Case in Malayalam. Since we have gotten rid of the necessity of semantic interpretability, it is possible to postulate semantically uninterpretable Feminine gender or Masculine gender on words belonging to the class that does not make gender distinction. This means that words such as ‘aana’ can mean any member belonging to the class ‘elephant’ (either Male or Female), but such words are grammatically feminine because of the vowel-ending, just as in languages which have attested grammatical gender. This does not mean that ‘aana’ semantically refers to a female elephant. To denote “female elephant”, it will be necessary to say “piDiyaana”, “piDi” being the prefix denoting “female gender”. Similarly, to denote “male elephant”, the prefix “komban”7 denoting “male gender” is used. Also, note that “komban” alone can refer to a “male elephant” while “piDi” alone does not refer to a “female elephant”. Therefore, one way of solving the anomaly is to argue that “Nongendered” Nouns with vowel-ending (ending with /a/, /ܼ/ and /iޝ/ segments) have semantically uninterpretable Feminine gender and those Nouns copy this feminine feature onto the Case head, resulting in the insertion of Gen 2, in the case of Genitive Case, and Dat 2, in the case of Dative Case. Vowels /a/, /ܼ/ and /iޝ/ are used to denote semantically interpretable Feminine Gender in Animate Common Nouns and Proper Nouns. Hence, we can argue that the same suffixes /a/, /ܼ/ and /iޝ/ that denote Feminine gender in “Gendered” Nouns extend their application to the domain of “Non-gendered” nouns. The idea that a language can have two kinds of gender, one semantically interpretable and the other semantically uninterpretable, has been proposed earlier (Fabregas & Perez, 2008). I propose that, in Malayalam, the semantic interpretability of a Masculine feature or a Feminine feature results from the coming together of two different features—the ‘Gender’ feature and the Masculine/Feminine feature. In other words, morphological gender is semantically uninterpretable when present in the class of Non-gendered nouns.
7
The vocabulary item “kombanaana”, semantically denoting “male elephant”, is also marked for Feminine gender due to the vowel ending. Relevant to our discussion is that it takes Gen 2 and Dat 2. But “komban”, meaning “male elephant”, acts as a Masculine Noun and takes Gen 1 and Dat 1.
110
Chapter Six
In order to strengthen my argument regarding the postulation of a ‘Gender’ feature and the Gendered/Non-gendered distinction, it is interesting to look at the interaction of Number with the Animate/Inanimate distinction. Generally, the suffix [-aL] is understood as the Plural suffix with inanimate Nouns while the suffixes [-ar] and [-maar] are considered as Plural suffixes used with animate Nouns. But, in addition to inanimate nouns such as ‘kallԥ’ (stone), ‘puuݝԥ’ (flower) etc., most animal-denoting nouns such as ‘paTTi’ (dog), ‘aana’ (elephant), etc take [-aL] suffix to denote plurality. Even some humandenoting nouns have the option of taking an [-aL] suffix when they denote Feminine gender. E.g.: naDi-kaL (actresses). All that can be generalised about the distribution of the plural suffixes is that those stems which can take [-ar] and [-maar] suffixes are those stems which make a gender distinction. Hence, the interaction of Number is not with Animate/Inanimate distinction, but with Gendered/Non-gendered distinction. “Gendered” Nouns can take [-maar] and include animal denoting nouns “kurukk-an-maar” (foxes) and human-denoting nouns “naDi-maar”. The semantically uninterpretable Feminine feature I proposed earlier cannot take [-maar] suffix (*aana-maar, *paTTi-maar etc.). Hence we can argue that the choice of [-maar] suffix is determined by the presence of ‘Gender’ feature. This supports my analysis that there is a Gendered/Non-gendered distinction in Malayalam nouns. I argue that both classes of nouns—gendered and non-gendered—can have morphological markers that represent Masculine/Feminine distinction in Malayalam. In the class of gendered nouns, the default feature (Masculine) is marked with [-an] suffix while the marked feature (Feminine) is represented variously as [-aL], [-a], [-i] or [-ii] suffixes. In the class of non-gendered nouns, the default feature (Neuter/Masculine) is represented as an [-am] ending or a schwa ending or [-u] ending, while the marked feature (Feminine) is represented variously as [-a], [-i] or [-ii] ending. The copy of the default gender is contained in Gen 1, in the case of Genitive, and in Dat 1, in the case of Dative. The copy of the marked gender is contained in Gen 2, in the case of Genitive, and in Dat 2, in the case of Dative. It is difficult to substantially prove the existence of grammatical gender in Malayalam, which is a language without Subject-Verb agreement. Besides, the absence of recorded grammatical rules in this direction raises
Grammatical Gender in Malayalam
111
doubts, since Malayalam has a rich history of grammar books. However, this explanation is theoretically plausible, and appears to be better than the alternative phonological explanation.
5. Conclusion In this paper, it has been proposed that the allomorphy in Genitive Case and Dative Case in Malayalam is due to the agreement of the Case Head K with number and gender features of the nominal expression. Hence, a particular morphological form of Case is revised as a portmanteau of Case and the copy of nominal features. Under nanosyntactic analysis, the two allomorphs are part of two competing lexical items, each corresponding to a different syntactic sub-tree. In order to account for anomalous behaviour of nouns that are semantically inanimate, I propose the existence of semantically uninterpretable Gender in Malayalam. The crucial question for further research is why the allomorphy is restricted to Datives and Genitives in Malayalam. How does Genitive Case pattern with Dative Case, excluding Nominative, Accusative and Locative? This is a question worth further exploration and this will unravel only if we examine the Case hierarchy in Malayalam. It would also be useful to make a typological study of Datives and Genitives in various languages.
References Abney,Steven. “The English Noun phrase in its sentential aspect.”(PhD diss., MIT, 1987). Caha,Pavel.“The nanosyntax of Case.”(PhD diss., University of Tromso.2009). Fabregas, Antonio and Isabel Perez. “Gender agreement on adverbs in Spanish.” Journal of Portuguese Linguistics 7, (2008): 25-47. Halle, Morris. “Distributed Morphology.Impoverishment and Fission”. In MITWPL 30: Papers at the Interface, ed. Benjamin Bruening, Yoonjung Kang and Martha McGinnis. (Cambridge: MIT Press: 1997): 425-449. Halle, Morris and Alec Marantz. “Distributed Morphology and the Pieces of Inflection.” In The View from Building 20 ed. Kenneth Hale and S. Jay Keyser. (Cambridge: MIT Press, 1993): 111-176. Harley, Heidi and Elizabeth Ritter. “Person and Number in Pronouns: A Feature Geometric Analysis”. Language 78.3 (2002): 482-526.
112
Chapter Six
Jayaseelan, K. “The dative case in the Malayalam verb” In Deep Insights, Broad Perspectives: Essays in Honor of Mamoru Saito. Eds Y. Miyamoto, D. Takahashi, H. Maki, M. Ochi, K. Sugisaki& A. Uchibori (Tokyo: Kaitakusha, 2013):139-166. McIntyre, Andrew. Handout 1: Basic Notions in Morphology inSeminar Issues in Morphology. Available at: //www3.unine.ch/, /www.angl.hu-berlin.de/ ; also at: //www.angl.hu-berlin.de/department/staff/1685901/unterrichts materialien/ intro.morph.pdf Starke, Michal. “Nanosyntax: A short primer to a new approach to language.” Tromso University working papers on Language and Linguistics 36.1 (2009): 1-6. Szabolci,Anna. “The Noun phrase.”(PhD diss., UCLA.1992). Taraldsen, Knut Tarald. “Lexicalising Number and Gender in Lunigiana.” Tromso University working papers on Language and Linguistics.(2009): 113-127.
CHAPTER SEVEN UNDERSTANDING ERGATIVE CASE LICENSING IN HARYANAVI1 USHA UDAAR
Abstract This paper focuses on the issues surrounding multiple case modalities available to a language in the generative framework. I specifically discuss the procedure of ergative case licensing in Haryanavi, a Western IndoAryan language. There has been a considerable amount of debate on the question of ergativity, with analyses ranging from structural to inherent to dependent case in varying languages. Using novel data from Haryanavi, a rather less explored language, I try to consolidate the findings about ergativity in terms of case licensing and the role played by independent factors which bring about a split in the otherwise nominative-accusative system. I utilize appropriate diagnostics to prove that Haryanavi ergative is not licensed via phi-agreement like its nominative counterpart, nor is it a theta-related inherent case assigned alongside the external theta role. The investigation thus sheds light on various theories of case licensing and its interaction with independent factors like clause structure as discussed in the generative literature. Keywords: Dependent case, Ergativity, Inherent case, phi-feature agreement, Structural case
1. Ergativity: A Theoretical Overview The traditional understanding of ergativity calls attention to a classical divide among languages of the world, which are classified as either
1
Erg=Ergative, abs =Absolutive, nom=Nominative, acc=Accusative, hab= Habitual, lv=light verb, m=masculine, f= feminine, sg=singular, pl=plural, perf=perfective aspect, pres=present tense
114
Chapter Seven
ergative-absolutive or nominative-accusative, depending upon how they pair up their transitive subjects. Specifically, the languages that treat their transitive subjects at par with their intransitive subjects to the exclusion of their transitive objects are considered as nominative-accusative. On the other hand, the languages that treat their transitive subjects as different from their intransitive subjects and transitive objects are classified as ergative-absolutive languages (Dixon 1995). Note that the equivalent treatments of subjects and objects are judged on the basis of their case marking, morphological verbal agreement patterns and patterning based on syntactic operations like relativization, control, clausal coordination etc. While not all languages manifest the above mentioned ‘ergative’ characteristics uniformly, it is generally the morphologically overt ergative case which is understood as the common thread binding most ergative languages (Holmer 2009). In this regard, ergativity is a well-discussed yet contentious issue, with a number of analyses proposed for ergative case licensing which have failed cross-linguistic generalization. In light of varied proposals accounting for ergativity, it is now accepted that ergativity may not refer to one syntactic phenomenon across all ergative languages. Consider the analysis of Chukchi ergativity, which is supposed to license a structural ergative case, resulting from phi-agreement between the subject DP and T head (Bobaljik and Branigan 2006). In fact, Chukchi is not an isolated case of structural ergativity. Rezac, Albizu and Etxepare (2014) suggest that the ergative in Basque is a structural case too, which is licensed due to phi-agreement with T and movement of the concerned DP to the spec, TP position. On the other hand, a number of languages are said to manifest inherent ergative case, which is assigned by a defective v head in association with the external theta role for languages like Warlpiri, Hindi-Urdu, Georgian etc (Johns 1992; Woolford 1997; Anand and Nevins 2006). For yet another set of languages including Sakha and Chol, ergative is a configurational case, dependent upon the presence of another DP within the case licensing domain (see Baker & Vinokurowa 2010; Coon 2013). Hence, it becomes imperative to explore the nature of ergative case for each language, especially when dealing with a less-explored language like Haryanavi. Following the above-mentioned discussion, I shall now take up the task of investigating the process of licensing of ergative case in Haryanavi. I will use a number of diagnostics to demonstrate the non-structural status of ergative case, in terms of absence of phi-agreement and consequent structural case licensing on the subjects in perfective constructions in
Understanding Ergative Case Licensing in Haryanavi
115
Haryanavi. Next I illustrate that ergative is not clearly linked to external theta role. I apply varied diagnostics including unaccusatives, unergatives and causative constructions to establish the absence of ergative case on the argument getting internal theta role from vP. Last but not least, I consolidate findings from the previous discussion and make a clear correlation between the presence of ergative case on the subject and availability of another nominal inside the vP domain. Therefore, I provide convincing evidence to support a configurational, dependent case analysis for Haryanavi ergative case. The ensuing discussion in the chapter clarifies the issue of multiple modes of case licensing available to a given language. The analysis for ergativity chiefly underlines the role played by independent factors such as defective functional heads and the process of feature inheritance in bringing about ergativity in an otherwise nominative-accusative system.
2. Haryanavi: An Introduction Haryanavi is a Western Indo-Aryan language (WIAL), mainly spoken in the states of Haryana, Delhi and the western part of Uttar Pradesh. While the language is spoken by a vast population residing in and close to the National Capital Region of Delhi, India, Haryanavi is understood to be one of the westernmost ‘dialects’ of Hindi. There are many similarities between Hindi and Haryanavi at both the lexical and grammatical level. However, the language also demonstrates its own specific peculiarities, especially in the domain of case and agreement. At the grammatical level, Haryanavi is a verb-final language (1) and allows scrambling (2) like its other WIAL counterparts. The default word order of Haryanavi is nominative-accusative and can be observed in (3), where the transitive subject appears unmarked and triggers verbal agreement. In the same construction, the object appears with an overt case morpheme and fails to trigger verbal agreement. (1) tau
philԥm
Old man(sub) movie(obj)
dekhya
kԥrܭ
watch
do.hab.pres
‘The old man watches the movies.’
Chapter Seven
116
(2) philԥm movie(obj)
tau-ܭ
dekhya
kԥrܭ
old man-only(sub)
watch
do.hab.pres
‘Only the old man watches the movies.’ (3) tau
tai-nܭ
bԥlavܭga
Old man.nom old woman.acc
call.m.sg.fut
‘The old man will call the old woman.’ Haryanavi, like many other WIALs, shows aspect-based split-ergativity, i.e. it manifests ergative-absolutive pattern in the perfective aspect. In comparison with (3), the data listed in (4) demonstrates that an overtly case marked transitive subject fails to trigger verbal agreement. The verb, instead, agrees with a case unmarked object. (4) tau-nܭ
tai
Old man.erg old woman.abs
bahԥr
bԥlai
outside
call.f.sg.perf
‘The old man called the old woman outside.’ (5) tai
bahԥr
cԥli
old woman.nom outside go.f.sg.
gyi lv.f.sg.perf
‘The old woman went outside.’ A comparison between constructions given in (3) and (4) reveals the points of differences between imperfective and perfective constructions. We can, additionally, see the ergative-absolutive alignment in Haryanavi in (4)-(5), where the transitive object and intransitive subject show similarities in terms of appearing without an overt case morpheme and triggering verbal agreement, to the exclusion of transitive subject which appears with an overt case morpheme. Given the brief introduction to Haryanavi ergativity, I will now take up the task of understanding the nature of Haryanavi ergative case in the next section.
Understanding Ergative Case Licensing in Haryanavi
117
3. Nature of Haryanavi Ergative Case: Structural versus Inherent versus Dependent The recent discussion in generative literature on the topic of ergativity reveals three major analyses for ergativity. While one form of analysis considers ergative as a structural case licensed by a functional head in certain languages like Basque and Chukchi (Bok-Bennema 1991; Bobaljik and Branigan 2006; Rezac, Albizu, Etxeparre 2014), there is another approach which insists that ergative is instead licensed in association with an external theta by the v head (Woolford 2006; Legate 2008; Aldridge 2004). Finally, there is yet another analysis which de-links ergative from both phi-agreement and theta roles and considers ergative to be a dependent case (Marantz 1991; Coon and Preminger 2012; Laka 2006). According to the dependent case analysis, ergative results from a configurational setting and requires that there should be another nominal present within the same case licensing domain. The following sub-sections are dedicated to ascertaining the nature of ergative case licensing in Haryanavi.
3.1. A Structural Case Analysis? For the structural case analysis, I will essentially try to find out whether there is a connection between ergative case and phi-agreement with the T head. In the case of positive evidence, there is then no structural difference between ergative and nominative case in Haryanavi. To this end, I will utilize several diagnostics including morphological verbal agreement, nonfinite contexts, scope constructions, unaccusative predicates. The example given in (6) demonstrates that the subject fails to trigger verbal agreement in the ergative constructions. Instead, it is the case unmarked object which triggers agreement on the verb. Thus, we have clear evidence suggesting that ergative subject does not enter into phiagreement relation with the T head. (6) tau-nܭ
bԥkri
bandhi
Old man.erg goat.f.sg.abs tie.f.sg.perf ‘The old man tied the goat.’ I will now examine the structural status of ergative by checking the nonfinite constructions in Haryanavi. The underlying idea behind this
Chapter Seven
118
diagnostic is that a structural case is intricately linked to the concerned functional head, which in this case is T. Therefore, the structural ergative case should be absent in a T-less context, if all other conditions of its appearance are met. In (7), the embedded subject appears with the ergative case in the presence of a phi-complete T/finite head. Conversely, ergative case does not appear in the absence of a phi-complete T head in (8). dekhya
(7) tau-nܭ
ԥk tai-nܭ
mussa pԥkԥݏya
Old man.erg see.perf that old woman.erg mouse
catch.perf
‘The old man saw that the old woman caught a mouse.’ (8) tau-nܭ
tai
mussa pԥkԥݏti
dekhi
Old man.erg old woman.abs mouse catch.inf see.perf ‘The old man saw the old woman catching the mouse.’ A cursory investigation of finite versus non-finite context reveals that the ergative case disappears in the absence of a finite T context. However, such an example does not suffice to link ergative with TP due to the fact that the ergative patternin Haryanavi is restricted to the perfective aspect, which is absent for the non-finite embedded clauses. Let us now utilize the diagnostic of scope constructions to compare nominative and ergative subjects. Following Anand and Nevins (2006), I concur that reconstruction across a vP boundary is restricted to DPs which enter into agreement with the heads from which they reconstruct. Thus, nominative subjects, owing to their licensing via TP, can reconstruct across vP boundary and allow surface as well as inverse scope in (9). (9) koe
baܿԥk
hԥr ma݄ԥs-nܭ
dekhܭga
some child.nom every person.acc watch.fut ‘Some child will watch every person.’ [some> every; every > some] (10) kise baܿԥk-nܭ
hԥr ma݄ԥs
dekhya
some child.erg every person.abs watch.perf ‘Some child watched every person.’[some> every; *every > some]
Understanding Ergative Case Licensing in Haryanavi
119
In contrast, the ergative subject in (10) allows only surface scope reading. Therefore, we have clear evidence suggesting that, unlike nominative subject (9), the ergative subject does not phi-agree with the functional head T. Last but not least, we will examine unaccusative constructions. The unaccusative predicates are considered as true intransitives, in the sense that they possess only a single argument which enters into agreement with T and moves to the spec, TP position (see Chomsky 1981, 1995; Collins 1997). The sole argument of unaccusative, thus, realizes nominative case due to phi-feature agreement with T. (11)
Using unaccusative as a diagnostic for the structural status of ergative presupposes that the concerned case should appear with the sole argument as it compulsorily enters into phi-agreement relation with T. (12) tau/*tau-nܭ
bހݚܭya/pԥݏya
Old man.nom/old man.erg sit.m.sg.perf/fall.m.sg.perf ‘The old man sat/died.’ It is evident in (12) that the subject cannot appear with an ergative case morpheme with the unaccusative predicates. Thus, we can recommend that given the diagnostics for structural cases, Haryanavi ergative subjects do not enter into an agreement relationship with the T head. Hence, we conclude that unlike languages like Chukchi and Basque, Haryanavi does not show structural ergative case.
120
Chapter Seven
3.2. Inherent Ergative Case The inherent case analysis ties case licensing on arguments with the theta roles assigned by vP. According to Woolford (1997, 2006), the nonstructural cases can be classified as lexical and inherent cases, the major difference being which theta role an argument gets assigned. Therefore, the lexical case is assigned in association with internal theta role while the argument bearing a case associated with external theta role is said to carry an inherent case (see (13) for details). The forthcoming discussion in this sub-section works on the principle of finding a causal link between ergative case and external theta role. (13)
I will now use suitable diagnostics including unaccusative, unergative, causative constructions to probe whether Haryanavi ergative is a theta related inherent case, licensed by vP in the perfective aspect. Following the discussion, I demonstrate that the ergative case invariably fails to appear on an argument with an internal theta role, even if it raises to the spec, TP position. The ergative appears only on arguments carrying an external theta role in Haryanavi. However, the presence of external theta is not a sufficient condition for triggering of ergativity. Therefore, the inherent case analysis does not seem adequate for the present case. Let us begin with our first diagnostic of unergative constructions. Perlmutter (1978) suggests that unergatives are intransitives that carry a single external theta role. This generalization implies that the sole argument/subject of unergative verbs should be assigned an external theta role. Hence, if ergative case is assigned by the vP to an argument carrying
Understanding Ergative Case Licensing in Haryanavi
121
external theta role, we expect the subject of an unergative construction to surface with ergative case in the perfective aspect in (14). (14) tau-nܭ/*tau
cހƭkya
Old man.erg/old man.nom
sneeze.perf
‘The old man sneezed.’ (15) *tau-nܭ/tau
calya
Old man.erg/old man.nom
walk.perf
‘The old man walked.’ We can observe in the data given above that the subject of unergative verb ‘sneeze’ carried an ergative case in Haryanavi. Therefore, we have evidence stating correlation between external theta role and ergative case in (14). However, the subject of another unergative predicate ‘to walk’ does not bear ergative case in (15). It is noteworthy that Haryanavi behaves like its other WIAL counterpart, Hindi, in restricting its ergativeunergative pattern to a special class of bodily unergative predicates only (see Amritavalli 1979; Richa 2008 for details). Hence, we cannot spot a direct link between external theta role and ergative case for Haryanavi. Moving on to the next diagnostic, I will investigate the domain of unaccusative predicates in Haryanavi. As discussed in the previous section, the unaccusative predicates are regard as true intransitives, carrying a single internal thematic role (Perlmutter 1978). The object DP finally moves to the spec, TP, position for EPP reasons. Given that unaccusative verbs do not license an external theta role, we can observe that ergative case is absent on the subject of the unaccusative predicate ‘to sit’ in (16). (16) tau/ *tau-nܭ
kursi pܭ
bހݚܭya
Old man.nom/old man-erg chair-on sit.perf ‘The old man sat on the chair.’ Finally, we utilize the diagnostic of causative constructions to explore the inherent status of ergative case in Haryanavi. We recognize that causative constructions involve addition of an extra ‘applicative’ layer, where the
Chapter Seven
122
causer/subject DP carries an external theta role. Hence in (17), the subject of the causative intransitive verb like ‘to make sleep’ should get an external agentive theta role. The subject of such construction expectedly manifests ergative case. (17) *tau/tau-nܭ
tai
suva
Old man.nom/old man.erg old woman.acc sleep-cause
di give.f.sg.perf
‘The old man caused the old woman to sleep.’ The discussion about inherent ergative case makes two important points clear: firstly, the ergative case never appears on the arguments carrying internal theta role; secondly, there is no clear association between external theta role and appearance of ergative case morpheme on the subject DP. In lieu of the above-mentioned observations, we can now conclude that Haryanavi does not possess an inherent ergative case. Therefore, we need to further probe the third possibility and find out whether Haryanavi ergative is a dependent case.
3.3. A Dependent Case Analysis Before we begin establishing the dependent case analysis for Haryanavi ergativity, we need to observe that this analysis essentially disproves the causal relationship between appearance of ergative case and phi-agreement as well as assignment of theta roles. In fact, a dependent case is licensed in a special configuration. The idea of a dependent case was first given by Marantz (1991), as a part of the post-syntactic morphological component, where he suggested that ergative (also accusative) case is assigned in a specific position governed by V+I, due to the presence of another DP within the case licensing domain. Accordingly the dependent accusative, in a nominative-accusative language, is assigned to the lower of two nominals when the higher nominal gets a nominative case. On the contrary, the dependent ergative case is assigned to the higher of the two nominals, when the lower nominal is assigned a structural absolutive case in an ergative-absolutive language. The configurational set-up for dependent case analysis is given in (18).
Understanding Ergative Case Licensing in Haryanavi
123
(18)
In line with Marantz (1991) and Baker (2013), Baker and Vinokurova (2010) adopt the idea of dependent ergative case. However, they argue that ergative case assignment takes place in the narrow syntactic component itself, rather than in the post-syntactic morphological component (also see Laka 2006; Coon 2013; Coon and Preminger 2012; Preminger 2012). Baker & Vinokurova further suggest that languages like Sakha consolidate both structural as well as dependent case licensing mechanisms, such that nominative and genitive are licensed structurally, while the ergative and accusative are dependent cases assigned configurationally. I will now try to establish whether Haryanavi ergative is a dependent case. For the configurational system of dependent case licensing, we need to consider only one important factor: the presence of another lower DP in the clause. Detailing upon our findings from the previous sub-sections, we had observed that ergative case does appear in the presence of another DP within the vP domain. I will begin with investigating the domain of transitive verbs in (19): (19) *tau/tau-nܭ
tai
pܭhli bԥriyã dekhi
Old man.nom/Old man.erg old woman.abs first time see.f.sg.perf ‘The old man saw the old woman for the first time.’ The transitive construction in (19) shows the presence of ergative on the subject DP in the presence of the object DP. However, we need to further probe the domain of intransitive predicates to establish a clear correlation between dependent ergative and the presence of another nominal in the case licensing domain.
Chapter Seven
124
In this regard, we have already observed the unaccusative and unergative constructions of Haryanavi in the previous sub-sections. We have seen in (12) and (15) that unaccusatives, which are also considered as true intransitives by the virtue of not allowing an external argument, do not manifest ergative on their subjects. We can assume that this happens due to absence of a dependent configuration in unaccusative constructions. However, we also came across another set of intransitives, namely unergatives, which allow ergative on their subjects in some cases. It is interesting to note that the issue of unergatives has been aptly dealt with in Hale and Keyser’s (1993) work where they suggest that unergatives are underlying transitives, which allow the incorporation of the internal argument into the predicate: see (20) given below. (20)
Therefore, unergatives cannot be considered as true intransitives, as they possess another NP in the object position. Hence, the ergative-unergative constructions in Haryanavi prove to be fertile ground for testing of dependent case analysis. I will now examine the transitive status of Haryanavi unergatives. To this end, I will use the presence of cognate object (21) and adjectival modification (22) to depict that unergatives in this language are indeed underlying transitives. (21) tau-nܭ
T.B.-ki kހãssi
Old man-erg T.B.-gen cough.f.sg
kހãssyi cough.f.sg.perf
‘The old man coughed the cough of T.B. (tuberculosis).’
Understanding Ergative Case Licensing in Haryanavi
(22) tau-nܭ
kԥsutta
Old man-erg strong
125
kހãssya cough.def.perf
‘The old man coughed a strong cough.’ The Haryanavi data given in (21)-(22) give strong indications of the transitivity of its unergative constructions. We can observe in (21) that the unergative constructions do contain a slot for cognate objects, which when overt also trigger verbal agreement. Similarly, the covert internal argument can be modified by use of adjectives in (22). The verbal agreement in such instances seems to be dependent upon the overtness of object DP in such constructions. Landau (2010) also suggests that overt agreement morphology is dependent upon the presence of a D-feature. We assume that the covert status and lack of verbal agreement of object NPs in unergatives can be blamed upon the absence of D-feature, as the objects get further incorporated in the VP. Our main take-away from the discussion on unergatives is that the unergative constructions in Haryanavi are clearly transitives. Therefore, we can now safely posit that the ergative-unergative constructions carry a dependent ergative case by virtue of having a covert object DP for reference in the lower domain in the vP: see (23). (23)
Concluding this section, we have seen that the Haryanavi ergative case is neither structural, nor inherent. Instead, Haryanavi assigns ergative in a dependent configuration which requires the presence of another nominal in
126
Chapter Seven
the vP domain. The dependent case analysis has been attested for transitive as well as underlying transitive constructions.
4. Concluding Remarks This paper discussed a number of modalities allowed for case licensing/assignment in the generative framework. We came across three major types of case licensing mechanism, namely: structural case, inherent case and dependent case. The structural case licensing mechanism involves licensing of case as a result of phi-feature agreement between a DP and a functional head. The inherent case analysis associates some nonstructural cases (not resulting from phi-agreement) with the external theta roles. Finally, dependent case takes into account independent factors like presence of another nominal in the case licensing domain. The investigation of ergative case in Haryanavi proved to be an interesting study as it revealed that it possesses a dependent system for ergative case licensing.
References Aldridge, Edith. “Minimalist analysis of ergativity.” Sophia linguistica: working papers in linguistics 55 (2007): 123-142. Amritavalli, R. “Representation of transitivity in the lexicon.” Linguistic Analysis 5, no. 1 (1979): 71-92. Anand, Pranav and Andrew Nevins. “The locus of ergative case assignment: Evidence from scope.” In Ergativity: Emerging issues, ed. by Alana Johns, Diane Massam, and Juvenal Ndayiragije. (Dordrecht: Kluwer, 2006): 3–25. Baker, Mark C. “On dependent ergative case (in Shipibo) and its derivation by phase.” Linguistic Inquiry (2014): 341-379. Baker, Mark C. and Nadya Vinokurova. “Two modalities of case assignment: Case in Sakha.” Natural Language & Linguistic Theory 28, no. 3 (2010): 593-642. Bobaljik, Jonathan David and Phil Branigan. “Eccentric agreement and multiple case checking.” In Ergativity: Emerging issues, ed. by Alana Johns, Diane Massam, and Juvenal Ndayiragije. (Dordrecht: Kluwer, 2006): 47-77. Bok-Bennema, Reineke. “Case and Agreement in Inuit” (Foris, Dordrecht, 1991). Chomsky, Noam. “A Note on Non-control PRO.” Journal of Linguistic Research 1, no. 4 (1981): 1-11.
Understanding Ergative Case Licensing in Haryanavi
127
—. The minimalist program (Cambridge, MA: MIT press, 1995). Coon, Jessica. Aspects of split ergativity. (Oxford: Oxford University Press, 2013). Coon, Jessica and Omer Preminger. “Taking ‘ergativity’ out of split ergativity: A structural account of aspect and person splits.” LingBuzz/001556 (2012). Dixon, Robert MW. Ergativity. (Cambridge: Cambridge University Press, 1995). Hale, Kenneth and Samuel Jay Keyser. “On argument structure and the lexical expression of syntactic relations.” In The view from Building 20, Essays in Linguistics in Honor of Sylvain Bromberger, ed. K. Hale and S.J. Keyser (Cambridge, MA: MIT Press, 1993):53109. Holmer, Arthur. “The ergativity parameter.” Lund Working Papers in Linguistics 48 (2009): 101-113. Johns, Alana. “Deriving ergativity.” Linguistic Inquiry 23, no. 1 (1992): 57-87. Johnson, Kyle and Satoshi Tomioka. “Lowering and mid-size clauses.” In Reconstruction: Proceedings of the 1997 Tübingen Workshop, Universität Stuttgart and Universität Tübingen, ed. Graham Katz, Shin-Sook Kim, and Heike Winhart (Arbeitspapiere des Sonderforschungsbereichs 340, Bericht Nr. 127, 1997): 185-206. Landau, Idan. “The explicit syntax of implicit arguments.” Linguistic Inquiry 41, no. 3 (2010): 357-388. Marantz, Alec. “Case and licensing.” 1991 paper published in Arguments and case: Explaining Burzio’s generalization, ed. Eric Reuland (Amsterdam: John Benjamins 2000): 11-30. Perlmutter, David M. “Impersonal passives and the unaccusative hypothesis.” In Proceedings of the 4thAnnual meeting of the Berkeley Linguistics Society, ed. Jeri Jaeger, Anthony Woodbury, Farrell Ackerman, Christine Chlarello, Orin Gensler, John Kingston, Eve Sweetser, Henry Thompson, Kenneth Whistler (Berkley, California: University of California, 1978): 157-190. Preminger, Omer. “The absence of an implicit object in unergatives: New and old evidence from Basque.” Lingua 122, no. 3 (2012): 278-288. Rezac, Milan, Pablo Albizu, and Ricardo Etxepare. “The structural ergative of Basque and the theory of Case.” Natural Language & Linguistic Theory 32, no. 4 (2014): 1273-1330. Richa. “Unaccusativity, Unergativity and the Causative Alternation in Hindi: A Minimalist Analysis.”(PhD dissertation, Jawaharlal Nehru University 2008).
128
Chapter Seven
Woolford, Ellen. “Four-way case systems: Ergative, nominative, objective and accusative.” Natural Language & Linguistic Theory 15, no. 1 (1997): 181-227. —. “Lexical case, inherent case, and argument structure.” Linguistic Inquiry 37(2006):111–130.
CHAPTER EIGHT DATIVE AND ERGATIVE SUBJECT CONSTRUCTIONS IN PUNJABI: UNDERSTANDING PERSON AGREEMENT1 GURMEET KAUR Abstract Dative subjects in Punjabi do not prevent the tense head from agreeing in person with the nominative theme DP, unlike in Icelandic where person agreement between the nominative theme and T is banned in the presence of a dative subject (Taraldsen 1995, Sigurðsson 1996 et al.)2. This paper attempts to explain the possibility of person agreement in Punjabi dative subject constructions. I suggest that intervention effects in dative subject constructions in Punjabi are averted by movement of the theme DP past the dative DP, allowing the former to agree with T in a sisterhood configuration. This local arrangement makes 1st/2nd person agreement possible. Further, I compare the dative constructions with ergative structures in the language, where T cannot agree in person with the object. I argue that unlike the dative subject, the ergative DP occupies the edge of vP, preventing the object from moving higher to T. Consequently, the nonpronominal object remains in situ and agrees long distance with T. A pronominal object, by contrast, moves to the edge of vP and receives an adpositional marker, which prevents any further agreement with T. The paper ends with implications of this analysis for person agreement in syntax, in keeping with Baker (2008).
1
Dat=Dative case, Nom=Nominative case, Erg=Ergative case, Gen=Genitive case, 1=First person, 2=Second person, 3=Third person, SG=Singular, PL=Plural, Pres=Present tense, Perf=Perfective aspect, Hab=Habitual aspect, Def=Default agreement, F=Feminine, M=Masculine, Inf=Infinitival, Conj=Conjunctive, Part=Participle, Poss= Possessive, Neg=Negation 2 Punjabi in this paper refers to the variant of the language spoken in Kanpur, Uttar Pradesh.
Chapter Eight
130
Keywords: dative, person agreement, defective intervention, ergative, Punjabi
1. Introduction Dative subject constructions have received a huge amount of attention in literature due to their unusual agreement patterns. This attention has mostly centered on Icelandic, where partial (minus person) agreement obtains between the tense head and the lower theme DP in the presence of a dative subject, as discussed in Taraldsen (1995), Sigurðsson (1996), Chomsky (2000), Boeckx (2000, 2008), Sigurðsson and Holmberg (2008) among others. Consider example (1) for illustration. (1a) Honum
líka
þeir
him.DAT like.3PL they.NOM ‘He likes them.’ (1b)*Honum
líkum
við
him.DAT like.1PL we.NOM Intended: ‘He likes us.’ (1c)*Honum
líkið
þið
him.DAT like.2PL you.NOM.PL Intended: ‘He likes you.’ (Sigurðsson and Holmberg, 2008: 2) The example in (1) is a dative nominative construction from Icelandic (variety A in the sense of Sigurdsson and Holmberg 2008), where only number agreement can obtain between the theme DP and the verb in the presence of a dative subject (1a). Agreement between the verb and the theme DP in (1st/2nd) person results in ungrammaticality, as illustrated in (1b-c). Note that 3rd person is assumed to be minus person (Silverstein 1976, Harley and Ritter 2002, Bejar and Rezac 2003, Anagnostopoulou 2005 among others). Thus, 3rd person agreement in (1a) is not considered as agreement in person, but only in number. A possible explanation to account for the agreement pattern in (1) is provided by defective intervention (à la Chomsky 2000, 2001). In an
Dative and Ergative Subject Constructions in Punjabi
131
Agree based model of phi agreement, defective intervention refers to the blocking of agreement between a probe Į and a matching goal Ȗ in the presence of an intervening argument ȕ, even if it is defective or inactive (Chomsky 2000). Schematically, (2) Į
ȕ
Ȗ
Adopting the notion of defective intervention, lack of person agreement in dative constructions in Icelandic is attributed to the dative DP being an intervener (Bejar and Rezac 2003; Holmberg and Hróarsdóttir 2003 among others). Specifically, it is claimed that the probe on the tense head is split such that there are separate probes for number and person (in keeping with Taraldsen 1995; Chomsky 2000; Anagnostopoulou 2003). The person feature on T probes first and encounters the dative DP. This DP is a quirky dative argument that consumes the person probe on T and moves to spec, TP. With the dative DP having moved to a non-intervening position, the number probe on T can now search and agree with the nominative DP in number. Such a derivation results in partial agreement on T in Icelandic dative nominative constructions. In contrast to the dative constructions in Icelandic, in Punjabi, full phi agreement (+person) obtains between the nominative theme DP and the tense head in dative subject constructions. Consider (3), where the tense auxiliary shows full phi agreement with the theme DP. (3a) karan- nNJNJ Karan-dat
o
pasaƾd
3.sg
like
e be.pres.3.sg
‘Karan likes her/him.’ (3b) karan- nNJNJ Karan-dat
maƭ
pasaƾd
1.sg
like
ãã be.pres.1.sg
‘Karan likes me.’ (3c) karan- nNJNJ Karan-dat
tuu 2.sg
‘Karan likes you.’
pasaƾd like
ͅ be.pres.2.sg
Chapter Eight
132
This paper aims to present and analyze the differential agreement pattern in Punjabi dative nominative constructions. To foretell the analysis, I will argue that full phi agreement between the theme DP and T results from a local agreement relation between the two. This local configuration is made possible by the movement of the theme DP past the dative subject. Specifically, both the dative DP (DP2) and the theme DP (DP1) are base generated within VP, where the dative DP gets case marked by V, while the theme remains unlicensed in the presence of an unaccusative v. Being in the same minimal domain as the inactive dative DP, the theme DP is not prevented from moving to the specifier of TP via spec, vP (assuming it has an edge feature as in Legate 2003). The theme DP and T stand in a sisterhood relation which allows full phi (+person) agreement. Consider (4) for a schematic representation. (4) [TP DP1[TP DP2 [vP (-acc) … [VP tDP2 [ tDP1 V]]]]] The paper also compares dative structures with ergative structures in Punjabi, which permit only partial agreement between the object and the verbal complex. Consider the example in (5), where the unmarked nonpronominal object triggers agreement in number and gender with T. A pronominal object, on the other hand, is obligatorily marked with -nNJNJ, preventing any agreement with T, which subsequently manifests default agreement, as in (6). (5) karan- ne Karan-erg
kuRii
vekhii
girl.f.sg see.perf.f.sg
‘Karan saw the girl.’ (6) karan- ne Karan-erg
maƭ-nNJNJ/tai-nNJNJ
vekhyaa
1.sg-nNJNJ/2.sg-nNJNJ
see.perf.m.sg (def)
‘Karan saw me/you.’ I show that like the dative DP in dative constructions, the ergative DP in ergative structures also satisfies subjecthood and intervention requirements. However, the ergative DP differs from its dative counterpart in that it is base-generated in the specifier of vP. As a result, there is no roll up of the object to a position local enough to the agreeing functional head T. The non-pronominal object in (5) remains in situ and agrees longdistance with T. This is illustrated schematically in (7a). A pronominal
Dative and Ergative Subject Constructions in Punjabi
133
object, on the other hand, by virtue of being referential, raises to a lower specifier of vP. However, it receives an adpositional -nNJNJ in this position. The presence of the PP layer prevents any further agreement between the pronominal object and T3. This is represented in (7b). (7a) [TP [vP(erg) erg subj [VP [ obj V]]]] (7b) [TP [vP(erg) erg subj [vP obj-nNJNJ [VP [ objt V]]]] This paper is organized as follows. Section 2 introduces the dative nominative constructions in Punjabi and shows that the dative DP satisfies subjecthood criteria. Presenting the notion of defective intervention, section 3 illustrates that dative DPs in dative-nominative structures in Punjabi also fulfill the intervener requirements. However, this intervention is overruled by movement of the theme argument past the dative subject, resulting in 1st/2nd person agreement. This analysis is provided in section 4. Section 5 puts forth a comparison with ergative constructions which do not permit person agreement between the object and T. Section 6 discusses the implications of the analysis for person agreement in syntax.
2. The Dative Subject Construction in Punjabi Dative subject constructions are found across South Asian languages including Hindi-Urdu, Kannada, Malayalam as discussed in Sridhar (1979), Mohanan and Mohanan (1990), Bhatt (2003), Subbarao (2012) among others. Punjabi also has dative nominative constructions with experiential predicates (which are unaccusative in the language) such as ‘to like’, ‘to hate’, as in (3), repeated here as (8), where karan is marked with the dative case morphology and maƭ remains unmarked and controls agreement. (8) karan- nNJNJ maƭ Karan-dat
1.sg
pasaƾd
ãã
like
be.pres.1.sg
‘Karan likes me.’
3
Note that Punjabi, like Hindi-Urdu, does not allow arguments marked with overt adpositions to trigger agreement.
Chapter Eight
134
Additionally, the dative DP in (8) satisfies subjecthood requirements, akin to the dative subject constructions in Icelandic. The diagnostics that confirm this claim are: (a) binding of the subject-oriented anaphor, (b) obviation with the pronominal, and (c) control into participial adjuncts (see Mohanan 1994; Kidwai 2000; Anand and Nevins 2006 for discussion of subjecthood diagnostics for Hindi-Urdu). (9) karan-nNJNJi Karan-dati
apniii self.fi
navii
kamiiz
new.f.sg shirt.f.sg
pasaƾd aayii like
come.perf.f.sg
‘Karan liked his new shirt.’ (10) karan-nNJNJi oddii*i/j Karan-dati
navii
kamiiz
pasaƾd
3.sg.f-gen*i/j new.f.sg shirt.f.sg like
aayii
come.perf.f.sg
‘Karan liked his new shirt.’ (11) karan- nNJNJi Karan-dati
kuRii
[PROi
baarii
tõ
vekh
ke]
girl.f.sg PROi window from see.inf conj.part
pasaƾd aayii like
come.perf.f.sg
‘Having seen her through the window, Karan liked the girl.’ Subjects in Punjabi bind the subject-oriented anaphor apnaa. This test is satisfied by the dative DP, which binds the subject-oriented anaphor (fem.) apnii, as shown in (9). Further, as shown in (10), the dative DP obviates coindexation with the pronoun odii, like other subjects in the language. The dative DP can also control into participial adjuncts, as illustrated in (11). The examples in (9)-(11) indicate that the dative DP is the subject of the clause. Thus, there is no difference in subjecthood between the Icelandic dative DP and its counterpart in Punjabi. Probing further into the nature of the dative subject in Punjabi, the next section discusses the status of the dative subject as an intervener in light of the notion of defective intervention (in keeping with Chomsky 2000).
Dative and Ergative Subject Constructions in Punjabi
135
3. Dative Subject in Punjabi: An Intervener So far, we know that dative subject constructions in Punjabi manifest person agreement between the theme DP and the tense head, unlike the dative constructions in Icelandic where person agreement is disallowed. This difference in agreement does not follow from subjecthood status of the dative DPs, which is identical across the two languages. Given these observations, this section attempts to investigate the status of the Punjabi dative subject as an intervener. As mentioned earlier, defective intervention in the sense of Chomsky (2000) refers to the blocking of agreement between a probe and a lower goal in the presence of an inactive or defective goal which c-commands the lower goal. Defective intervention is first applicable at the level of Match, which is a subcomponent of Agree. At this level, an element is eligible to act as an intervener if it has a fully matching or a defective set of phi features. To understand this schematically, in the representation in (12), ȕ can act as the intervener for agreement between Į and Ȗ, irrespective of whether it has a full or an incomplete set of phi features. (12) Į
> ȕ >
Ȗ
In this sense, an element can be taken as matching the probe even if it has only a subset of the features on the probe. For example, consider a probe with a full set of unvalued phi features (person, number and gender). An element with a full phi set or even a defective set (only number and gender features, for example) is a suitable goal to the said probe as long as it lies in the c-command path of the probe. Further, this element will create an intervention effect, irrespective of whether it is valued for case or not. Presence or absence of case value will have an effect only at the level of actual valuation via Agree, where an already case valued element will fail to trigger agreement on the probe despite satisfying the other conditions for agreement. In a nutshell, while the set of features (full or defective) makes an element an intervener at the level of Match, its (missing) case value renders it an intervener at the level of Agree. Inactivity of the intervening goal results in default agreement on the probe. Returning to the dative subject constructions in Punjabi, I now move on to show that the dative subject satisfies the eligibility criteria to be an intervener in the language. Let me illustrate how. With respect to the
136
Chapter Eight
featural specification on the dative DP, it is not incorrect to assume that the dative DP has a full set of phi features with values: it has a person, number and gender feature and each of these features is specified with a value (also see Preminger 2011a). Thus, the number feature on the dative noun-phrase can be valued as singular or plural, its person feature as 1st/2nd/3rd person, and its gender feature as feminine or masculine. This makes it fit to be an intervener at the level of Match. At the level of Agree, however, the facts are more complex. The dative DP is (lexically) case valued and thereby inactive as soon as it is merged in the structure. To elaborate, we know that the predicates selecting dative subjects in Punjabi are noun+verb combinations, pasaƾd aanaa ‘like + come’ for instance. Along with Landau (2010) and Davison (2003), I follow the standard assumption that these N+V complexes are unaccusative (also see Perlmutter 1983, Pesetsky 1995).The dative DP originates in the specifier of the V, where it is lexically case valued. The theme DP, base-generated in the complement of V, remains without case since the unaccusative v is not capable of accusative case valuation. Consider the structure in (13). (13)
As shown in (13), the dative DP receives a lexical case from V in its basegenerated position and is subsequently rendered inactive for any further agreement. Thus, while the dative DP is an intervener, it is inactive for agreement by the probing head. This inactivity of the intervener should lead to default agreement on the probing functional head. However, as the examples in (3) show, the probing head shows full phi agreement with the
Dative and Ergative Subject Constructions in Punjabi
137
theme DP, indicating that the intervention effects of the dative subject are overruled. The next section attempts a derivational account to explain the full phi agreement in dative structures.
4. Dative Constructions: A Derivational Analysis Having shown that the dative DP in the dative subject constructions in Punjabi is a non-agreeing intervener, I now proceed to show how these constructions are derived in the language so as to yield full phi agreement between the theme DP and the agreeing functional head. Since both the dative and the theme DP are co-arguments of the same head (V) as shown already in (13), they are both within the same minimal domain and are equidistant from a higher functional head that might probe into the VP for valuing its phi features. In other words, being equidistant, both the dative and theme DP are rendered equally eligible to move to a position outside the VP domain. I propose that the intervention effects (default agreement) are not attested in the dative structures since the dative DP is placed in the same minimal domain as the theme DP, as a result of which it fails to prevent movement of the theme DP and its consequent agreement with T. Rezac (2008) also discusses ways of averting intervention effects in Icelandic and Basque. To be concrete, he proposes that the partial agreement which takes place in intervention configurations of the kind in (14) can be prevented by movement. Note that in (14), ȕ acts as an intervener for agreement between the probe Į and the lower Ȗ. According to Rezac, intervention effects can be prevented if Ȗ raises past the intervening ȕ, while remaining within the search space of Į, as shown in (15). This movement has a saving effect on agreement, which is now phi complete. (14) Į> ȕ> Ȗ (15) Į> Ȗ> ȕ>Ȗt Employing Rezac‘s (2008) proposal, I provide the representation for the dative constructions in Punjabi in (16):
following
138
Chapter Eight
(16)
In (16), the dative DP and the theme DP are base-generated inside VP. The dative DP in the specifier of VP gets lexical case valued. Since the predicate in question is unaccusative, the theme DP located in the complement of the VP fails to receive any case value at this stage. Upon merger of C-T into the derivational space, T probes to value its unvalueduninterpretable phi set. It first encounters the dative DP, which is a nonagreeing intervener. I propose that, instead of the probe searching further, the lower goal raises to a higher position to agree with T and value its case. Specifically, I assume with Legate (2003) that the unaccusative vP, though incapable of assigning accusative case, has a specifier. The theme DP uses this position to escape the verbal domain and ends up higher than the dative DP. The dative DP does not prevent this movement of the theme DP since both the dative DP and the theme DP are in the same minimal domain (VP) and hence equidistant from the specifier of the higher vP. Further, the theme DP raises further to the specifier of TP so as to agree with it in a sisterhood configuration, and gets nominative case valued in return. I assume with Baker (2008) that 1st/2nd person agreement obtains between T and the theme DP as a result of local agreement.
Dative and Ergative Subject Constructions in Punjabi
139
An implication of this analysis is that the theme DP shows subjecthood properties by virtue of being in spec, TP. Consider the following examples, which show that this prediction holds:4 (17) maƭi
apnei nave
dostãã-nNJNJ
1.sgi
self.pli new.pl
friend.pl-dat
pasaƾd aayaa like
come.perf.m.sg
‘My new friends like me.’ (reading for reflexive) (18) maƭi
odde*i/j
nave
1.sgi 3.m.pl-gen*i/j new.pl
dostãã-nNJNJ friend.pl-dat
pasaƾd aayaa like
come.perf.m.sg
‘His/her new friends like me.’ (reading for pronominals) As already discussed in section 2 of the paper, subjects in the language bind the subject oriented anaphor apnaa, which is attested for the nominative theme DP as shown in example (17). Further, like other subjects in the language, the nominative DP shows obviation with the antisubject oriented pronominal oddaa, as illustrated in (18). As suggested by a reviewer, an additional piece of evidence to show that the theme DP moves to the specifier of TP and is nominative-valued is provided by the wh-form5. Specifically, the theme DP, if present in the wh-form, occurs in its nominative form kaun in the dative structures under consideration. Consider the example in (19): (19) karan- nNJNJ Karan-dat
kaun who.nom
pasaƾd like
e be.pres.3.sg
‘Who does Karan like?’ An unresolved problem pertains to the position of the dative DP. Since the dative argument also shows subjecthood properties in the constructions under discussion, it is possible that it also raises to spec, TP. However, note that while the dative DP manifests most characteristics of a typical
4
The theme DP in dative subject constructions in Hindi-Urdu also behaves like a subject, as has been shown by Bhatt (2003) and Davison (2003). 5 The wh-forms in Punjabi inflect for case. The nominative form of the question word ‘who’ is kaun, while the accusative/dative form is ki-nNJNJ.
Chapter Eight
140
subject as shown earlier in examples (9)-(11), it digresses from subjecthood in one respect. The dative DP can bind possessive pronominals in a structure as shown in (20), unlike other subjects in the language which fail to do so, as demonstrated for the (unmarked) nonnominative subject in the perfective in (21)6. (20) maƭ-nNJNJi 1.sg-dat
meriii
kitaab
pasand
1.sg.f.poss book.f.sg like
e be.pres.3.sg
‘I like my book.’ (21) *maƭi meriii 1.sg
1.sg.f.poss
kitaab book.f.sg
paRhi read.perf.f.sg
‘I read my book.’ Additonally, it fails to control PRO in an infinitival structure, as in (22). (22) *munDe-nNJNJi [PRO*i kuRii pasaƾd aanaa] nayii caaƾdii boy-dat
PRO
girl like come.inf neg
want.hab.f.sg
Intended: ‘The boy does not want to like the girl.’ A possible analysis has been given by Poole (2015) who proposes that T0 (the position associated with subjecthood) is decomposed into two separate functional heads: Pers(pective)0 and B(inding)0, where Pers0 > B0. Pers0 is meant to host PRO, while B0 binds subject-oriented anaphors. A quirky subject that raises only as high as BP can only bind the subjectoriented anaphor. If the subject raises further to PersP, it can both bind the anaphor as well as control PRO. In accordance with this proposal, it is possible that the dative DP in Punjabi raises to BP such that it is able to manifest anaphor binding properties like other subjects in the language, but fails to control PRO in an infinitival structure, as is shown in example (22). Recall, however, that the dative subject can control into participial adjuncts, as was shown previously in (11). Furthermore, it is also unclear
6
Chandra and Kaur (2014) have argued that the unmarked 1st/2nd person subjects in the perfective in Punjabi are neither ergative nor nominative. These subjects manifest mixed characteristics and are licensed by Participant Phrase, a functional head located between the ergative valuing v and the nominative valuing T.
Dative and Ergative Subject Constructions in Punjabi
141
as to how does the dative subject bind the possessive pronominal in (21). Thus, the position of the dative DP in Punjabi dative subject constructions is unclear at the moment and needs more inquiry. To recapitulate, I have shown that despite the shared subjecthod properties of the dative DP, dative subject constructions in Punjabi vary from those in Icelandic7. The dative DP in Punjabi originates in the spec of VP, and receives a lexical dative, becoming inactive for agreement. The theme DP, located in the same minimal domain as the dative DP raises to agree with T in full phi. In contrast, the dative subject in Icelandic is an intervening quirky subject that is both lexically and structurally valued for case. Thus, upon probing by T, it agrees in person with T and moves to the spec of TP. Consequently, T can only agree in number with the lower theme DP. I now proceed to present the ergative constructions in the language, which behave like the Icelandic dative constructions in that they do not manifest person agreement.
5. A Comparison with Ergative Constructions Like Hindi-Urdu, Punjabi shows morphological ergativity in the perfective aspect (Bhatia 1993; Bhatt 2007)8. Thus, subjects of all transitive perfective structures are ergative marked with –ne. Agreement in these ergative constructions is controlled by the object. However, this agreement is either partial, i.e. in number and gender (23a), or default (23b).
7
A reviewer inquires about the difference/similarities between the ECM structures in Icelandic and Punjabi. To my knowledge, languages such as Punjabi and HindiUrdu do not have any ECM structures. With verbs like ‘believe’, the language employs small clauses as shown in (i), where the object gets -nuu marked. However, whether these structures can be analyzed as ECM structures is an issue beyond the scope of this paper. (i) maƭ [karan-nNJNJ syaanaa] mannii ãã 1.sg.nom Karan-nuu smart believe.inf.f.sg be.pres.1.sg ‘I believe Karan to be smart.’ 8 Punjabi also shows person based split ergativity, such that only the 3rd person subject DPs are ergative marked. The 1st/2nd person subject DPs remain unmarked, but are underlying oblique as argued by Chandra and Kaur (2014). As shown in (i) none of the subjects (1st/2nd or 3rd person) trigger agreement on the verbal complex, which instead agrees with the object in number and gender. (i) maƭ/tuu/o-ne roTTii khaaddii 1.sg/2.sg/3.sg-erg bread.f.sg eat.perf.f.sg ‘I/you/he ate bread.' In the present work, I focus only on constructions with 3rd person subjects.
Chapter Eight
142
(23a) karan- ne Karan-erg
kuRii
vekhii
girl.f.sg see.perf.f.sg
‘Karan saw the girl.’ (23b) karan- ne Karan-erg
maƭ-nNJNJ/tai-nNJNJ
vekhyaa
1.sg-nNJNJ/2.sg-nNJNJ
see.perf.m.sg (def)
‘Karan saw me/you.’ In this section, I attempt to compare Punjabi ergative constructions with the dative ones to make two observations. Like its dative counterpart, the ergative argument manifests the properties of a subject. It also satisfies the requirements to be an intervener for agreement between the verbal complex and the object. However, it differs from the dative DP with regard to its position of base-generation. Unlike the dative subject that originates in the specifier of VP, the ergative DP is an external argument that originates in the specifier of vP. This structural position of the ergative DP prevents the movement of the object to a local position vis-àvis T, disallowing agreement in person. To start, the ergative DP in the ergative constructions is shown to be the subject of the clause. Evidence for subjecthood of the ergative DP is provided by the following examples in (24)-(26): (24) karan-nei apniii Karan-ergi self.fi
navii
kamiiz
pasaƾd
new.f.sg shirt.f.sg like
kittii do.perf.f.sg
‘Karan liked his new shirt.’ (25) karan-nei
oddii*i/j
navii
kamiiz
pasaƾd
Karan-ergi 3.sg.f-gen*i/j new.f.sg shirt.f.sg like ‘Karan liked his new shirt.’
kittii do.perf.f.sg
Dative and Ergative Subject Constructions in Punjabi
(26) karan- nei Karan-ergi
kuRii
[PROi baarii
tõ vekh
143
ke]
girl.f.sg PROi window from see.inf conj.part
pasaƾd kittii like
do.perf.f.sg
‘Having seen her through the window, Karan liked the girl.’ In (24), the ergative DP binds the subject-oriented anaphor apnii. It obviates coindexation with the pronominal oddii as shown in (25). In (26), the ergative DP controls into participial adjuncts. The ergative DP satisfies all of these tests, which lends support to its status as the subject of the clause. In this respect, the ergative and the dative DP are alike. Next, the ergative DP is also an intervener for agreement like the dative DP. Given the notion of defective intervention, we know that an element can be an intervener for agreement if it satisfies the following criteria: i. The element in question has a full matching or defective set of phi features as the lower goal and, ii. The element is closer to the probe/landing site than the possible lower goal. Applying these criteria to the ergative DP in the ergative constructions, we note that the said DP satisfies both criteria. The ergative DP has a full set of phi features: person, number and gender. Thus, it can be specified as 1st/2nd or 3rd person, singular or plural and masculine or feminine. Also, it is basegenerated in the specifier of vP, a position closer to the probe on T than that of the object, which is located in the complement of VP. Similar to the dative DP, the ergative DP, while being an intervener, does not trigger agreement on the probe since it is already valued for case and hence inactive. To elaborate, it has generally been assumed in literature that ergative is an inherent case licensed by v to an external argument in its specifier (Legate 2008, 2012 and Woolford 2006). Assuming this conception of the ergative case, I suggest that the subject in ergative structures also gets a theta-related inherent case in its base position upon being merged. This entails that while the ergative DP acts as an intervener, it cannot trigger agreement on T since it is already ergative case valued by v. Consider the structure in (27), where the subject originating in the specifier of a v(erg) head receives a theta-related ergative case.
144
Chapter Eight
(27)
Thus, both the ergative and the dative DPs are interveners by virtue of their matching phi features. However, neither of them can trigger agreement due to already being case valued. Despite these similarities, ergative constructions manifest partial agreement between the object and T, in contrast with the dative structures, which show full phi agreement. I now attempt to provide a possible account for this partial phi agreement. I propose that the object of the ergative structure is trapped inside the vP with no possibility of movement outside the vP due to the intervening ergative subject in the specifier of vP. While the non-pronominal object DP is trapped in the complement of VP, its pronominal counterpart manages to move as high as the lower specifier of vP. This compels the object to agree long-distance with the T head, resulting in minus person agreement. Consider the structure in (28).
Dative and Ergative Subject Constructions in Punjabi
145
(28)
In (28), the subject is base-generated in the specifier of vP and the object originates as a complement of the VP. I assume with Legate (2008, 2012) and Woolford (2006) among others that the subject gets an inherent ergative case in its base position from v upon being merged. Upon the merger of the C-T, the T head probes its c-command domain for a suitable goal. It first encounters the subject, which is already valued for case and hence, unavailable for agreement. The object situated below the ergative subject cannot raise to a local position vis-à-vis the T head since the intermediate specifier position (spec, vP) is occupied by the ergative DP. Depending on its (non)pronominal status, the object can occupy either of the two positions within the vP domain. While the non-pronominal object is situated in the complement of VP, the pronominal object moves to the inner specifier of vP. Precisely, I claim with Kaur (2016) that the nonpronominal object is an NP that originates in the complement of VP and pseudo-incorporates into the V head, in line with Dayal’s (2011) proposal for unmarked objects in Hindi-Urdu. Like their unmarked counterparts in Hindi-Urdu, the unmarked objects in Punjabi also manifest properties of
Chapter Eight
146
pseudo-incorporated nominals. These object arguments (a) are numberneutral, (b) take narrow scope, and (c) fail to take discourse anaphora. For illustration, consider the examples in (29) to (31). (29) mitaa Mita
baccaa child
saambdii
e
look after.prog.f.sg be.pres.3.sg
‘Mita looks after a child/children.’ (30) mitaa baccaa nayii saambegii Mita
child neg
look after.fut.f.sg
‘Mita will not look after a child/children.’ (31) #kal
ikk kuRii aayii.
(Narrow scope only)
karan-ne kuRii vekhii
Yesterday one girl come perf.f.sg. Karan-erg girl see.perf.f.sg ‘A girl came yesterday. Karan saw her.’ The example in (29) has an unmarked singular animate in the direct object position. Despite being specified with a [singular] number specification, the unmarked object can be interpreted as plural, thereby manifesting number neutrality. Next, the unmarked non-pronominal object can never have wide scope over a higher operator (negation for example) in the structure. It only takes narrow scope, as shown in (30), where there is no one particular child that Mita will not look after. Finally, the nonpronominal direct object cannot be used to refer to the entity introduced prior in the discourse, (31). Based on these characteristics, I suggest that the non-pronominal object stays in the complement of VP. Differently from non-pronominal objects, pronominal objects are located outside the VP domain. I claim that pronominal objects, by virtue of being referential, raise from the complement of VP to the inner specifier of vP. Specifically, Diesing (1992) in her seminal study on indefinites postulates the Mapping Hypothesis, as given below in (32).
Dative and Ergative Subject Constructions in Punjabi
147
(32) Mapping Hypothesis i. VP maps into the nuclear scope (the domain of existential closure) ii. IP maps into the restriction (of an operator) The Mapping Hypothesis divides a syntactic tree into two parts with each part associated with a different semantic interpretation. The VP corresponds to the nuclear scope, and the IP provides the restriction. Focusing on the VP, Diesing proposes that VP is the domain of existential closure such that any argument that is quantificational must move out of this domain by LF. Following Diesing (1992), I propose that the pronominal object is forced to move from its base-generated position in syntax. Since the specifier of the transitive vP is occupied by the ergative DP, the object moves and gets tucked into the lower specifier of the vP (in the sense of Richards 1999). In this position, it receives an adpositional – nNJNJ from the v head. This P head licenses the pronominal object, which surfaces as a PP bearing -nNJNJ. Once the object is enclosed inside the PP, it becomes inactive for agreement. Returning to (28), the T head probes and encounters the ergative argument located in the specifier of vP. I suggest that the ergative subject in (28) first raises to the spec of TP, thereby exhibiting subjecthood properties. Once the ergative subject is out of the way, the T head encounters the object. The pronominal object, located in the inner specifier of vP is a PP, which cannot trigger agreement on the probing head. This results in default agreement on T. On the other hand, if the object is a nonpronominal situated in the complement of VP, the T head agrees long distance with it in number and gender9.
9 A reviewer questions the possibility of multiple probes in the language since separate probes for number and person could pose a problem for the analysis posited here. In response, I contend that there is no evidence for split phi probing in the language. The entire verbal complex agrees with the same argument in person, number and gender. However, the features manifested on the verb and the tense head vary such that v exhibits number and gender morphology, while T shows number and person. As for the split phi probing of the kind postulated for Icelandic dative structures, I claim that the approach is untenable for Punjabi. To elaborate, the split phi probing approach assumes that person agreement does not obtain between the object and the said functional head since the person probe on the functional head gets exhausted due to agreement with the higher argument. Only its number probe remains to agree with the object, resulting in (-person)
148
Chapter Eight
To summarize, the absence of person agreement in ergative structures in Punjabi has been argued to follow from the intervening syntactic location of the subject, in concert with the nature of the object.
6. Summary and Implications for Person Agreement This paper has presented dative and ergative constructions in Punjabi. I demonstrated that both dative and ergative DPs in the respective constructions show properties of a subject and an intervener. Notwithstanding these similarities, the agreement patterns in the two constructions differ from one another. While the dative constructions manifest full phi agreement between the theme DP and T, only partial agreement is attested in the ergative constructions. An answer was sought in the structural configurations in which agreement took place. In the dative constructions, the theme agrees with T in sisterhood, while agreement between the object and T in the ergative construction takes place long distance. With Baker (2008), I take these varying structural arrangements to correlate with person agreement. Specifically, 1st/2nd person agreement is possible only when agreement takes place in sisterhood. Long distance agreement between the probe and the goal results in partial (number and gender) agreement only. To elaborate, Baker (2008) postulates a condition called the Structural Condition on Person Agreement (SCOPA), given in (33) to account for the presence or absence of 1st/2nd person agreement morphology on agreeing functional heads. (33) A category F can bear the features +1 or +2 if and only if a projection of F merges with a phrase that has that feature and F is taken as the label of the resulting phrase. According to SCOPA, the occurrence of 1st/2nd person agreement on a functional head X is a consequence of whether or not X is in a local relationship with the 1st/2nd person NP. Thus, 1st/2nd person agreement can take place only if the agreeing 1st/2nd person DP merges with the said functional head. If the DP and the functional head do not end up as sisters and agreement takes place long distance between them, person agreement cannot obtain (although see Preminger 2011b for the need to weaken
agreement. Adopting this approach for the ergative structures in Punjabi would require positing that the ergative DP has a 3rd person feature that agrees with the verbal complex. As far as I understand, there is no evidence to make this assumption. I therefore do not adopt the split-phi probe analysis for dative and ergative structures in Punjabi.
Dative and Ergative Subject Constructions in Punjabi
149
Baker’s SCOPA and for evidence suggesting that person agreement can also obtain at a distance). Note that this formulation of person agreement is different from what is generally assumed for agreement in literature, in the sense of Agree proposed in Chomsky (2000, 2001). In the Agree based system, person agreement is not treated differently from usual agreement, which requires a c-command relationship between the probe and the goal. To elaborate, given a probe P and a goal g, Agree assumes that the entire phi-set on the functional head probes its c-command domain as a unit. Once the unvalued phi-set encounters an appropriate goal, it agrees with it and gets all its phi-features valued, with no distinct structural requirements for valuation of each feature. Given these two differing approaches to person agreement, the analysis proposed in this paper sides with the former approach to agreement, which posits distinct structural requirements on person agreement vis-à-vis number and gender agreement. Person agreement imposes stricter locality constraints and can be obtained only in sisterhood between the agreeing functional head and the goal DP. Number and gender agreement, on the other hand, can take place long distance, as long as the agreeing functional head c-commands the goal DP. This idea is verified by the dative and ergative constructions in Punjabi, where person agreement is attested in the former since agreement takes place in sisterhood. In the ergative structures, where agreement takes place long distance (à la Agree), number and gender agreement obtain without any person agreement. Needless to say, the reason why person agreement imposes stricter requirements on the system needs to be investigated in detail.
References Anand, Pranav and Andrew Nevins. “The locus of ergative case assignment: Evidence from scope,” in Ergativity, ed. A. Johns et al. (Netherlands: Springer, 2006): 3-25. Baker, Maker. The Syntax of Agreement and Concord. (Cambridge: Cambridge University Press, 2008). Bejar, Susana and Milan Rezac. “Person licensing and the derivation of PCC effects,” in Romance Linguistics: Theory and Acquisition, ed. Ana Teresa Pérez-Leroux and Roberge,Y. (Amsterdam: John Benjamins, 2003): 49–6.
150
Chapter Eight
Bhatt, Rajesh. Experiencer Subjects: Topics in the Syntax of the Modern Indo-Aryan Language. Handout. (University of Massachusetts, Amherst, 2003). —. “Long Distance Agreement in Hindi-Urdu,” Natural Language and Linguistic Theory 23 (2005): 757-807. Boeckx, Cedric. “Quirky Agreement,” Studia Linguistica 54 (2000): 354380. —. Aspects of the Syntax of Agreement. (New York: Routledge, 2008). Chandra, Pritha and Gurmeet Kaur. “Two nominative cases in Punjabi?” (Poster presentation at the Formal Approaches to South Asian Languages 4 (FASAL-4), Rutgers University, USA, March, 2014). Chomsky, Noam. “Minimalist Inquiries: the Framework,” in Step by step: Essays on minimalist syntax in honor of Howard Lasnik, ed. Martin, R., Michaels, D. and Uriagereka, J. (Cambridge: MIT Press, 2000): 89–155. —. “Derivation by Phase,” in Ken Hale: A Life in Language, ed. M. Kenstowicz. (Cambridge, MA: MIT Press, 2001):1-52. Davison, Alice. “Structural Case, Lexical Case, and the Verbal Projection,” in Clause Structure in South Asian Languages, ed. Dayal, V. and Mahajan, A. (Dordrecht: Kluwer, 2003): 199-226. Dayal, Veneeta. “Hindi Pseudo-incorporation,” Natural Language and Linguistic Theory 29 (2011):123-167. Diesing, Molly. Indefinites. (Cambridge, MA: MIT Press, 1992). Holmberg, Anders and Thorbjorg Hroarsdottir. “Agreement and Movement in Icelandic Raising Constructions,” Lingua 113 (2003): 997-1019. Kaur, Gurmeet. Person in Punjabi: Investigating Argument and Clitic Licensing. (PhD dissertation, Indian Institute of Technology Delhi, India, 2016). Legate, Julie. “Some Interface Properties of the Phase,” Linguistic Inquiry 34 (2003): 506-515. Mohanan, Tara. Argument structure in Hindi. (Center for the Study of Language, 1994). Kidwai, Ayesha. XP-adjunction in universal grammar: scrambling and binding in Hindi-Urdu, Oxford studies in comparative syntax. (New York: Oxford University Press, 2000). Landau, Idan. The Locative Syntax of Experiencers. (Cambridge: MIT Press, 2010). Perlmutter, David M. “Personal vs. Impersonal Constructions,” Natural Language and Linguistic Theory 1:1 (1983): 141-200. Pesetsky, David. Zero Syntax. (Cambridge: MIT Press, 1995).
Dative and Ergative Subject Constructions in Punjabi
151
Poole, Ethan. “Deconstructing quirky subjects,” (paper presented at the North East Linguistic Society 45, MIT, USA, 2015). Preminger, Omer. Agreement as a Fallible Operation. (PhD dissertation, Massachusetts Institute of Technology, 2011a). —. “Asymmetries between person and number in syntax: a commentary on Baker’s SCOPA.” Natural Language & Linguistic Theory 29:4 (2011b): 917-937 Richards, Norvin. “Featural cyclity and the ordering of multiple specifiers,” in Working Minimalism, ed. Samuel David Epstein and Norbert Hornstein (Cambridge, MA: MIT Press, 1999): 127-158. Sigurðsson, Halldor Á. and Anders Holmberg. “Icelandic Dative Intervention: Person and Number are distinct probes,” in Agreement Restrictions, ed. D’Alessandro, R. et al. (Berlin: Mouton de Gruyter, 2008): 251–279. Sigurðsson, Halldor Á. “Icelandic finite verb agreement,” Working papers in Scandinavian syntax 57 (1996): 1-46. Taraldsen, Knut Tarald. “On Agreement and Nominative Objects in Icelandic,” in Studies in Comparative Germanic Syntax, ed. Haider, Hubert, Susan Olsen and Sten Vikner. (Dordrecht: Kluwer Academic Publishers, 1995): 307-327.
INDEX
Abbi, Anvita, 2, 6, 74-76 Abney, Steven, 22, 24-25, 100 Adjectival Agreement, 18 Agree, 4, 59, 135, 149 AGREE, 4 Apprehensional, 37, 39 Avertive, 37, 41-42 Baker, Mark, 7, 1, 123, 138, 148149 Bangla, 1, 5, 6, 73-74, 76 Bresnan, Joan, 24 Chomsky, Noam, 4, 130-131, 135, 149 Comparative Degree, 11, 22, 33 Copular Clause, 17-18 Dative Subject, 4, 7, 12-13, 17, 129, 130, 133 Decategorialization, 48, 50-52 Defective Intervention, 130-131, 135 Degree Expression, 22-23 Dependent Case, 113, 122-124 Desemanticization, 48, 49 Determiner, 24-25, 100-101 Distributivity, 75, 77-78, 87-88 Divergence, 51 Dixon, Robert, 12, 39, 114 DP Hypothesis, 100-101 Dummy Pronominal, 18-19 Elsewhere Condition, 106 Emeneau, Murray Branson, 2, 6, 74 Erosion, 48, 50 Extension, 48-49 Frustrated Completion, 37-38, 4647, 53 Frustrated Initiation, 37-39, 45, 50, 53 Functional Adjective, 14, 19, 21, 33 Functional Head Approach, 24, 29
Genitive, 6-7, 14-16, 79, 93-94, 9798, 100-105, 107, 109- 111, 123 Gradability, 22, 25 Gradable Predicate, 12-13, 22, 33 Grammatical Gender, 6, 93-94, 109110 Haryanavi, 1, 5, 7, 113-126 Hindi-Urdu, 4, 12, 27, 29, 31, 33, 114, 133-134, 139, 141, 145 Icelandic, 129-131, 134-135, 137, 141, 147 Implicit Quantification, 6, 74, 7678, 84, 89 Incorporation, 3, 11, 124 Inherent Case, 4, 113, 120, 126, 143 Layering, 51-52 Lest Clause, 39-40, 52 Linguistic Area, 2 Malayalam, 1, 3-6, 14, 16, 20, 3748, 50-53, 74-74, 93-94, 97-98, 100-103, 105, 107-112 Marathi, 11, 13, 20 Masica, Colin, 2 Meeteilon, 1, 6, 55-57, 59, 60, 6263, 67, 71 Minimal Domain, 132, 137-138, 141 Minimise Junk Principle, 106 Nanosyntax, 94, 104, 106 Nominal Clause, 17 Nominalised Adjective, 18 Null Degree Head, 12-13, 25-26, 29, 31-32 Persistence 53, 54-55 Person, 99-100, 137-138, 143-144 Person Agreement, 7-8, 145-146, 153-154, Phi-Agreement , 119-120, 120-121, 124-125, 128, 132
154 Phi-Features, 119, 125-126, 132133 Plurality, 6, 77-78, 81-82, 85-86, 88, 90-91, 92 Possession, 135-136 Postpositions, 3-4 Predication, 13-14, 19-20, 23 Primitive Categories, 2, 5-6 Pronominal, 21, 80, 99, 138, 140, 146-147, 150 Property Concept (Pc) Expression, 5, 12, 14-15, 16, 19-20, 22, 28, 35 Proximitive, 41, 45-46, 47, 55 Punjabi, 1, 6-7, 78, 79, 135-139, 141, 143-144, 147-149, 153, 155 Quantification, 6, 77-78, 80-83, 8788, 90-93 Quantifier, 15, 27, 32, 34, 78, 80, 88, 91-93
Index Reduplication 2, 6, 78-80, 82-83, 85, 91 South Asian Languages, 1-2, 5-6, 52, 77-79, 139 Spec-Head Relation, 105 Split Phi Agreement, 154 Structural Case, 120, 122-124, 126, 132 Structural Condition on Person Agreement (SCOPA), 154 Superlative Degree Construction, 15, 25-26 Syntactic Doubling, 81, 85 Syntactically-derived Adjectives, 13, 35 Telugu, 16, 18, 24, 32-33, 35, 78-79 Tense-Aspect-Mood, 3, 41 Unaccusatives, 121, 123, 125-128, 130, 138, 143, 145 Unergatives, 121, 126-127, 130-131