Annual Review of South Asian Languages and Linguistics: 2008 9783110211504, 9783110208290

Citation preview

Annual Review of South Asian Languages and Linguistics 2008

≥

Trends in Linguistics Studies and Monographs 209

Editors

Walter Bisang Hans Henrich Hock (main editor for this volume)

Werner Winter

Mouton de Gruyter Berlin · New York

Annual Review of South Asian Languages and Linguistics 2008

edited by

Rajendra Singh

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

ISBN 978-3-11-020829-0 ISSN 1861-4302 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. ” Copyright 2008 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Christopher Schneider, Laufen. Typesetting: Asco Typesetters, Hong Kong. Printed in Germany.

Contents

Editorial preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

General Contributions Transparency and arbitrariness in Natural Language: Some empirical issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Probal Dasgupta

3

Reduplication and ‘echo words’ in Hindi/Urdu. . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Annie Montaut No smoke without fire: Invisible agent constructions in South Asian languages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Prashant Pardeshi A pragmatic account of the Hindi presumptive . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Ghanshyam Sharma

Special Contributions: The Syntactic Typology of Dravidian: A Debate What’s so subversive about Dravidian? Revisiting finiteness in Dravidian syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Sanford B. Steever Dravidian syntactic typology: A reply to Steever . . . . . . . . . . . . . . . . . . . . . . . . . 163 Hans Henrich Hock

Regional Reports North America . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Tej K. Bhatia Linguistic publications in Bengali (2000–2008): A brief review . . . . . . . . . . 217 Niladri Sekhar Dash

vi Contents South Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Rajend Mesthrie Linguistic publications in Punjabi (2000–2008): A brief survey . . . . . . . . . . 243 Joga Singh

Reviews Gregory D. S. Anderson – The Munda verb: Typological perspective . . . 265 by Ayesha Kidwai Joseph Bayer, Tanmoy Bhattacharya and M. T. Hany Babu (eds.) – Linguistic theory and South Asian languages: Essays in honour of K. A. Jayaseelan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 by Anjum P. Saleemi Colin P. Masica (ed.) – Old and new perspectives on South Asian languages: Grammar and semantics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 by Ashwini Deo Ghanshyam Sharma – Il dizionario di Hindi: Dizionario Hindi-Italiano, Italiano-Hindi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 by Alice Davison

Dialogue Bengali Determiner Phrase revisited: A response to Dasgupta and Ghosh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 by Shishir Bhattacharja The study of ‘Indian English’: What is it good for?. . . . . . . . . . . . . . . . . . . . . . . 307 by Lionel Wee

Appendices Announcements: The Gyandeep prize/Housekeeping . . . . . . . . . . . . . . . . . . . 321 Notes on contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323

Editorial Preface

Annual Review of South Asian Languages and Linguistics (ARSALL) is devoted to bringing out what is currently being explored in South Asian linguistics and in the study of South Asian languages in general. South Asia is home to a wide variety of languages, structurally and typologically quite diverse, and has often served as a catalyst and testing ground for theories of various kinds. Although linguists working on South Asia have made significant contributions to our understanding of language, society, and language in society, and their numbers have grown considerably in the recent past, until recently there was no internationally recognized forum for the exchange of ideas amongst them or for the articulation of new ideas and approaches grounded in the study of South Asian languages. The Yearbook of South Asian Languages and Linguistics, of which this annual is a direct descendant, played that role during the last decade, but in 2007 we decided to go a bit further and incorporate a slightly modified form of such a forum into Trends in Linguistics. This is the second issue of ARSALL as part of the series Trends in Linguistics: Studies and Monographs. Each volume of this annual has five major sections: i.

General Contributions, consisting of selected open submissions that focus on important themes and provide various viewpoints.

ii.

Special Contributions, consisting of inter-related or easily relatable, invited contributions on important issues, ranging from the narrowly grammatical to the wide-scope socio-linguistic/socio-political. This section will in e¤ect constitute a mini-symposium, albeit in the written form, on the issue chosen for a given year. It will serve the function of familiarizing the reader with current thinking on issues seen as salient in the study of South Asian languages.

iii. Reports, consisting of reports from around the world on research on South Asian languages.

viii Editorial Preface iv. Reviews and Abstracts, consisting of reviews of important books and monographs and abstracts of doctoral theses. v.

Dialogue, consisting of a forum for the discussion of earlier work, preferably previously published in this annual, comments, reports on research activities, and conference announcements.

Other than excellence and non-isolationism, ARSALL has no theoretical agenda and no thematic priorities. The first, general section of this, the second, issue of ARSALL contains four contributions: Dasgupta’s Transparency and Arbitrariness in Natural Language, Montaut’s Reduplication and Echo Words in Hindi/Urdu, Pardeshi’s Invisible Agent Constructions in South Asian Languages, and Sharma’s A Pragmatic Account of the Hindi Presumptive. The Special Contributions section is dedicated to a debate (between Sanford Steever and Hans Henrich Hock) on the syntactic typology of Dravidian. As North America, like Europe, continues to be a major center for the study of South Asian languages, we thought it was appropriate to have our Regional Reports section publish a report on North American research on South Asian languages in this issue of ARSLL. The first issue, readers may recall, contained a report on European research on South Asian languages. It also contains a report from South Africa and, following the tradition inaugurated in the 2004 issue of The Yearbook, reports on recent linguistic publications in two South Asian languages, Bangla (Dash) and Punjabi (Singh). The Review section of this issue contains reviews of Anderson’s recent important book on the Munda verb, Bayer et al’s Festschrift for Jayaseelan, Masica’s Old and new perspectives on South Asian languages, and Sharma’s Italian-Hindi-Italian dictionary. These have been written by Kidwai of J.N.U., Delhi, Saleemi of G.C.U., Lahore, Deo of Yale, and Davison of the University of Iowa. I am particularly happy to note that the Dialogue section of this issue contains responses to two pieces in ARSALL-2007. These contributions establish that ARSALL has in fact become the forum I wanted it to be. I am grateful to Dr. Ursula Kleinhenz and Wolfgang Konwitschny of Mouton de Gruyter and Dr. Shishir Bhattacharja of the University of Dhaka for help far beyond the call of duty in the preparation of this issue. Rajendra Singh

General Contributions

Transparency and Arbitrariness in Natural Language: Some Empirical Issues

Probal Dasgupta

The distinction between nonarbitrariness and transparency has seldom been brought to bear on empirical issues. On the basis of noun phrase data from Bangla (Bengali), the present study, working in the substantivist framework, argues that – faced with a gap in the paradigm (definite human nominals cannot mimic the inanimate definiteness format) – Bangla throws up a minimally arbitrary language-particular format as well as a UG-inspired transparent format to fill the gap. Neither of these blocks the other, suggesting that nonarbitrariness must be distinguished from transparency – and associated with a discourse-focused social science of language and a grammar-focused natural science of language, respectively. 1. Preliminaries Linguists expect to deal with general patterns in the grammar and with special facts in the lexicon. This expectation does not translate at once into strategies for the proper allocation of grammatical and lexical descriptive resources. On the way to such strategies, we may usefully examine the concepts of transparency and arbitrariness. The present discussion focuses on the conceptual level, in dialogue with empirical material; some earlier writings on these issues were discussed – on the basis of proposals close to those articulated here – by Dasgupta, Ford and Singh (2000: chapter 1). In the abstract context of linguistic theory, simple signs such as French chien ‘dog’ or Bangla (a.k.a. Bengali) kukur ‘dog’ are termed ‘‘arbitrary’’ because no biological or other foundation underwrites their concrete forms. Sliding from the ‘‘they are ungrounded’’ version to the ‘‘they constitute the ground’’ version of this doctrine, some linguists view kukur as unmotivated, as carrying no clues, while the ‘‘relatively motivated’’ kukurer ‘dog’s’ invites comparison with kukur ‘dog’, beRaler ‘cat’s’ and beRal ‘cat’. In such a perspective, arbitrariness and motivation count as natural opposites.

4 Probal Dasgupta What awaits serious articulation is the relation between ‘‘arbitrary/ motivated’’ and the distinct concept pair ‘‘transparent/ opaque’’. Transparency refers to the undistorted compositionality of an utterance. A construction is compositional if no opaque barrier within it (such as a world-creating predicate or modal operator) distorts or fragments the cumulation of part-interpretations assembling the interpretation of the whole. The view that relatively motivated signs like kukurer ‘dog’s’ must count as composite signs may lead its proponents to fuse the two concept pairs as follows. A language is anchored in a basic vocabulary consisting of simple (entirely arbitrary/ unmotivated) signs. Every relatively motivated sign is a composite sign, a construction composed of simple or composite signs. The patterns of the composition phenomena of a language are exhaustively describable in terms of rules. Rules specify opacity factor e¤ects where necessary and implement transparent compositionality elsewhere. Rules and phenomena pattern in principled ways that often lend themselves to maximally general description and, at higher levels of analysis, to explanation and more. We shall use the term ‘‘formalist linguistics’’ for work done on the basis of the package just outlined or of views closely resembling these. The substantivist alternative developed in Dasgupta, Ford & Singh (2000), Dasgupta (1993, forthcoming) and related writings approaches language phenomena in terms of a non-structuralist UG (universal grammar) account of the human language faculty and a non-ethnographizing historical characterization of the human capacity for society. The generalities of a macrolinguistic social science and those of a microlinguistic psychological science meet at the specifically patterned phenomena of language. Substantivist studies systematically co-articulate these patterns with grammar’s maximization of transparency and with the social dimension’s tra‰c of arbitrariness. The goal of the present study is to contextualize the notions of transparency and arbitrariness in this enterprise. The formalist fusion of relative motivatedness and relative transparency creates something in the nature of a paradox; bringing substantivist considerations to bear on this issue may help open the domain up for further inquiry. A formalist should a priori expect all relatively motivated forms to be equally easy to learn, since composition comes for free in UG. However, forms exhibiting more and more structure, though this brings with it an increase in motivatedness, actually become harder, not easier, for L1 acquisition in childhood or L2 acquisition in adult life. We shall approach the

Transparency and Arbitrariness in Natural Language

5

matter by first presenting some substantivist considerations about relative di‰culty without reference to arbitrariness or transparency. On that basis, we return to these conceptual issues, and, in section 2, we bring some empirical material to bear on the discussion. Relative di‰culty distinguishing the subregions of a person’s knowledge of language pertains to issues of heterogeneous access to one’s cognition. The overall framework of substantivist linguistics has been developing a distinctive emphasis on cognitive heterogeneity ever since the earliest careful proposals of this type emerged in the seventies. Substantivism can be usefully construed as a family of responses to Tesnie`re’s Problem. While no passage in Tesnie`re 1959 articulates it directly, the strand of inquiry that runs through his text invites the following formulation: ‘‘What grammatical and lexical principles, in general and in particular, underwrite the linguistic equivalence between certain compact expressions such as nagaram in Sanskrit and certain di¤use expressions like to the city in English?’’ Following Tesnie`re, substantivists use translation as the fundamental analytic metaphor to tackle a family of questions built around this one. Furthermore, again following Tesnie`re, interlexical research in the substantivist framework uses Esperanto as the medium of lexical content representation; for justification and discussion, see Dasgupta 2007. The central proposal of substantivism is that a linguistic representation constitutively involves simultaneous multiple characterizations. Relations of translation intuitively validated by relevant speakers hold between these co-characterizations. Only some of these relations – a class to be empirically delineated – lend themselves to formally exact description. To understand the point of such a perspective not just for obvious grammar-society interface phenomena or for cross-language alignment studies but within linguistics proper, consider the multiple genitive problem in English. Speakers can say and understand the consequences of the destruction of the city and the consequences of the city’s destruction, but not the city’s destruction’s consequences. However, formalist methods standard in linguistics – that treat a phrase structure representation, say, as amenable to separate validation independent of its translation relations with the entity’s other representations – leave us with an unhelpful competenceperformance binary. Formalist assumptions force us to attribute the asymmetry either to the grammar – o¤ering a formal derivation of the asterisk on *the city’s destruction’s consequences – or to some performance factor, discoverable through psycholinguistic ingenuity.

6 Probal Dasgupta In contrast, a substantivist expects to be able to deal with the contrast between ill-formed nominals like the city’s destruction’s consequences and such acceptable cases as John’s father’s brother’s wife in terms of specific properties of the words father, brother, wife contrasting with those of city, destruction, consequence, which make the city’s destruction’s consequences count as so hard that one cannot say it. Words, in the substantivist visualization of a lexicon/ encyclopaedia, count simultaneously as lexical entries interacting with the grammar in microlinguistics and as encyclopaedic entries anchored in discourse or macrolinguistics – another case of translation managing co-characterizations. To return to English multiple genitives, substantivist preferences lead us to seek a solution that includes a psycholinguistic account based on the properties of the sets of words, a lexicogrammatical description, and a translation connecting the two analyses. To summarize, formalism’s strategy is to look for a grammatical answer that will trivialize the psycholinguistics, or as a second choice to find a psycholinguistic account that removes the problem from the grammar. But such elimination procedures strike substantivists as missing the point. The guiding intuition of substantivism is that A and B (here psycholinguistics and grammar) get to share the work, and translation oversees the sharing. Within this framework visualizing language in terms of grammatically and discursively co-specified cognition, substantivists deploy particular resources for the study of the di¤erential accessibility of subregions of such knowledge – one example of cognitive heterogeneity. Translation can gloss necessitarianistically as ‘in a necessitarianistic way’ (inexactly, for this is a feature of lexical glossing), necessitarianistic as ‘typical of necessitarians’, necessitarian as ‘someone who believes in necessity’, and necessity as ‘the quality of being necessary’. The weight of these glosses marks necessitarianistically as more di‰cult than happily. The logic of glosses leads substantivism in a direction quite distinct from the logic of morphemic decomposition, which in one version or another inspires formalists, and for which even happy counts as bimorphemic, given that hapless and happy subtend an imaginary noun hap! This example stands in for a thousand words stating just why, for the morphology module, substantivism adopts WWM, Whole Word Morphology, as articulated in Ford, Singh & Martohardjono 1997 and Singh & Starosta 2003. Substantivists have no doctrinal reason to doubt that happy is a maximally accessible word needing no gloss.

Transparency and Arbitrariness in Natural Language

7

The substantivist study of relative di‰culty as an instance of cognitive heterogeneity goes back to the marked-unmarked distinction. Access to unmarked defaults is costless. In contrast, the action of producing or comprehending a marked item involves some computational cost for the mind. It is this notion of access cost (a familiar, if insu‰ciently understood, aspect of the theory of markedness) that enables us to state that necessitarianistically is in general harder to say and to understand than the word happily, or that Who does John say Sue thinks Tom believes Jane will marry is in general less accessible, for production or comprehension, than the sentence Who does Tom believe Jane will marry. Does the substantivist story about these sentences mimic the one about words, though? The concrete procedures of translation that help construct substantivist descriptions of relative di‰culty are varied; lexical glossing is only one of them. The choice of implementations flows from the conceptualization of the modules and their interrelations, a domain where substantivist inquiry has been assembling a richer set of questions than Tesnie`re had access to. For instance, commenting on the morphology-syntax boundary, one text embodying the decision to adopt WWM as substantivism’s characterization of morphology – Dasgupta, Ford & Singh (2000) – makes the constitutive point that there is no syntaxless, infinitely compact language (not even among the polysynthetic languages) that makes every utterance take the form of a polysynthetic megaword. Turning to the syntax-discourse boundary, we may add that there is no discourseless, infinitely syntactic language – not even among those who read Proust with pleasure – that makes multisentential discourses vanish and invariably turns each text into one huge sentence. For a closer look at the crucial role of discourse as a site of multiple validation in the earliest articulation of substantivism by Bhartrihari as a counterpoint to Panini’s formalism, see Dasgupta (2008). Beyond these points of entry lie certain noteworthy aspects of the syntax-discourse interface that the methods of substantivism compel us to place on the agenda. Here is one example. Imagine a language Z in which the sentence embedding system distinguishes once-embedders, like the particle that in John knows that Susan is coming, an example of single embedding, from twice-embedders, such as the particle ZRAT in the double embedding example John knows that Bill thinks ZRAT Susan is coming, and even from thrice-embedders, such as the particle SHRAT in the triple embedding case Tony says that John knows ZRAT Bill thinks SHRAT Susan is coming.

8 Probal Dasgupta Now, note that languages such as Z do not exist. All known embedding devices are once-embedders. The substantivist response to this fact is to propose that syntax on its own handles only single embedding; all multiple embedding is jointly managed by syntax and discourse. One way to execute this proposal is a bimodular ‘‘Recompose’’ operation that, given John knows this and Bill thinks that Susan is coming as input, yields the syntactic object John knows that Bill thinks that Susan is coming plus a specified translation relation with the input sentence pair. On such a view, every case of multiple embedding counts as translated – not glossed in the lexical sense – and thus as computably more di‰cult than sentences lacking such structure. If we need a sophisticated account of relative di‰culty anyway, and specific apparatus for such modules as morphology, syntax and discourse, then the way ahead is to deploy particular devices diversifying what had once been seen as a single, homogeneous type of work – the work that the classical theory of relative motivation was designed to do. The arbitrary as a whole, the socially conventionalized lexical and encyclopaedic knowledge that a child cannot innately inherit and that varies across societies, engages in intricate and di¤erentiated kinds of interplay with linguistic principles. As is suggested in Dasgupta (forthcoming), the macrolinguistic social science of what we are here calling ‘‘arbitrary’’ material and the microlinguistic psychological (natural) science focused on ‘‘transparency’’ manage their interface at the lexicon, which abbreviates the social encyclopaedia and at the same time encapsulates the grammar. To give a brief example of what the interplay looks like in real linguistics, consider the unacceptable German sentence *Es wurde heute von Pradip nicht gestorben ‘*It was not died by Pradip today’ and its acceptable Bangla counterpart prodiper aj mOra holo na (literally an impersonal passive translating the German sentence, but conveying the sense of ‘Pradip did not get around to dying today’). A substantivist account of such material stresses the option of claiming that UG bans impersonal passives for agentless, truly unaccusative verbs, but that societies attribute agency variably. One is then able to conjecture that discourse-nourished encyclopaedic knowledge makes it easy for speakers of Bangla at least in the ad hoc space of a particular conversation to construe ‘dying’ as an action by an agent, while the social/ discursive background of conversations in German make such a construal far-fetched or unavailable. It is of course more helpful to develop these ideas in the context of an extended example. Bangla o¤ers two domains of interest to choose from;

Transparency and Arbitrariness in Natural Language

9

for the domain of classification format phenomena, taken up in section 2, this study proposes that a transparent UG stop-gap coexists with a language particular alternative, creating a problem for formalistic accounts based on an absolutized notion of blocking. Another domain – a¤ective context cognate object nominalizations – is considered in Dasgupta (2006), where it is argued that the Word Formation Strategy that cognate object nominalizations encourage us to postulate shows that the demarcation of the lexicon/ grammar boundary must be based on the notion of transparency rather than arbitrariness.

2. Classification formats in Bangla We begin with some elementary facts about Bangla nominals. Bangla displays noun classification phenomena often described in terms of ‘classifiers’ – which would be a fair description for languages like Bahasa Indonesia, where such a typical expression as tiga ekor gajah, literally ‘three tail elephant’, for ‘three elephants’, does use a clearly independent classifier word ekor (lit. ‘tail’). The relevant phenomena in Bangla, however, as the following examples involving numerals indicate, invite description in terms of classification formats, not distinct ‘classifiers’, to rehearse a point made in more detail by Dasgupta & Ghosh (2007): (1)

a. EkTa meye jabe. one.Gnl girl go.Fut ‘One girl will go.’

b. Ekjon meye jabe. one.Hum girl go.Fut ‘One girl will go.’

(2)

a. duTo meye jabe. two.Gnl girl go.Fut ‘Two girls will go.’

b. dujon meye jabe. two.Hum girl go.Fut ‘Two girls will go.’

(3)

a. duTo ghOr khali ache. two.Gnl room vacant Cop ‘Two rooms are vacant.’ b. *dujon ghOr two.Hum room

(4)

a. *duTo two.Gnl

bhOdromohila lady

10 Probal Dasgupta b. dujon bhOdromohila jaben. two.Hum lady go.Fut.Hon ‘Two ladies will go.’ The noun meye ‘girl’ can occur in Bangla either with a ‘‘general’’ numeral such as EkTa ‘one.Gnl’, duTo ‘two.Gnl’, or with a ‘‘human’’ numeral such as Ekjon ‘one.Hum’, dujon ‘two.Hum’. Some nouns are more selective. Thus, ghOr ‘room’ cannot take a ‘‘human’’ numeral, hence the starred status of (3b). In contrast, bhOdromohila ‘lady’ never occurs with a ‘‘general’’ numeral, which is why (4a) is ill-formed. A detail that will matter later in the discussion appears at the verb: the honorific future form jaben ‘will go’ at (4b) contrasts with the non-honorific jabe in (1) and (2). Classification formats are also available at Det, as in (5), or N, as in (6) (the glosses ‘‘NuanIndiv’’ and ‘‘NuanColl’’, for nuanced individuation and nuanced collectivity, label poorly understood feature matrices): (5)

(6)

a. konTa which.Gnl ‘which one?’

b. konjon which.Hum ‘which one?’

c. konkhana which.Inan ‘which one?’ (inanimate)

d. kongulo which.Coll ‘which ones?’

e. konTi which.NuanIndiv ‘which one?’

f. konguli which.NuanColl ‘which ones?’

ei meyeTa this girl.Gnl ‘this girl’

N and Det cannot, however, compete with Numerals and other Quantifiers as far as classificatory richness is concerned: (7)

a. EkTa one.Gnl ‘a wall’

deyal wall

b. Ekjon bhOdrolok one.Hum gentleman ‘a gentleman’

Transparency and Arbitrariness in Natural Language

c. Ekkhana camoc one.Inan spoon ‘a spoon’

d. kOtokgulo ciruni couple.Coll comb ‘a couple of combs’

e. kOyekTi gan a.few.NuanIndiv song ‘a few songs’

f. Onekguli many.NuanColl ‘many songs’

g. EtoTuku this.much.Dimin ‘so little tea’

h. Ekgacha laThi one.Quirky stick ‘a (walking) stick’

i. EkpaTi one.Quirky ‘a shoe’

ca tea

juto shoe

k. EtoTa this.much.Extent ‘so much co¤ee’

11

gan song

j. Onekkhani SomOy much.Expanse time ‘a lot of time’ kOfi co¤ee

Det does not appear in such forms as *kongacha, *konpaTi, *konkhani ‘which.Quirky, which.Quirky, which.Expanse’ (where ‘‘Quirky’’ signals the extremely specialized lexical selection associating certain classification formats with certain nouns) and cannot switch on the ‘Extent’ meaning to make konTa mean ‘what quantity’ along the lines of (7k). This exercise helps us to choose between a ‘classifier morpheme’ view of the matter and the WWM approach on empirical grounds. A morphemebased analysis must assign clear and distinct feature compositions separately to a Det/ Num/ Q/ N base and to a Classifier a‰x morpheme. We have shown elsewhere (Dasgupta, in press) that even the simple grouping of the common ‘Classifier morphs’ into ‘Classifier morphemes’ is an unfeasible project. No proposal postulating morphemes like Ta, Ti, gulo, guli can achieve coherence in the face of the facts considered in Dasgupta (in press). It is argued in Dasgupta & Ghosh (2007) that the segmentability assumption accepted in Dasgupta (in press) for argument’s sake, that a ‘Classifier morph’ can be separated from a ‘base’ in featural terms, is unsustainable. While we do not have the space to rehearse those arguments here, it is important to look at the classification formats a noun can exhibit in order to mark definite/specific readings (specific with a demonstrative and definite elsewhere), now that we have agreed not to recognize a classifier a‰x. No noun appears in a human /Xjon/ format, as shown in (8), but the /XTa/

12 Probal Dasgupta format is widely used for singulars and /Xgulo/ for plurals, see (9), while /Xkhana/ marks inanimate singulars, as in (10): (8)

a. *meyejon girl.Hum ‘the girl’

b. *bhOdromohilajon lady.Hum ‘the lady’

c. *upacarjojon vice-chancellor.Hum ‘the vice-chancellor’ (9)

a. meyeTa, *meyekhana girl.Gnl, *girl.Inan ‘the girl’

b. meyegulo girl.Coll ‘the girls’

c. ei meyeTa this girl.Gnl ‘this girl’

d. ei meyegulo this girl.Coll ‘these girls’

(10) a. camocTa, spoon.Gnl, ‘the spoon’

camockhana spoon.Inan

c. ei camocTa/ this spoon.Gnl/ ‘this spoon’

camockhana spoon.Inan

b. camocgulo spoon.Coll ‘the spoons’ d. ei camocgulo this spoon.Coll ‘these spoons’

e. DimTa, Dimkhana egg.Gnl, egg.Inan ‘the egg’

f. Dimgulo egg.Coll ‘the eggs’

g. ei DimTa/ Dimkhana this egg.Gnl/ egg.Inan ‘this egg’

h. ei Dimgulo this egg.Coll ‘these eggs’

The interaction between classification format exponence and the noun, the numeral/ quantifier and the determiner is not the focus of the present study; one account of the tra‰c is provided in Dasgupta & Ghosh (2007). Su‰ce it to say that a given Bangla nominal structure carries classification features only at one of the three possible sites Det, Num and N. The richest set of formatting options is available at the numeral or quantifier, clearly this phenomenon’s centre of gravity in the nominal syntagm.

Transparency and Arbitrariness in Natural Language

13

We turn now to issues related to the way verbs agree with nominals on the honorificity axis, a point briefly touched upon in the context of (4b). The phrase ‘these five students’ translates two distinct Bangla phrases: (11) a. ei this

paMcjon chatro five.Hum student

b. ei this

paMcTa chatro five.Gnl student

The numeral paMcjon in (11a), positively specified for a Human feature, contrasts with (11b)’s numeral paMcTa, whose feature composition is as general as a numeral will allow it to be. Numerals appear either skeletally, when we count Ek dui tin car paMc ‘one two three four five’, or in this format that carries classification features. The present analysis describes (11a,b) in terms of Word Formation Strategies (see Dasgupta & Ghosh 2007 for details omitted here). The relevant strategies can be stated as (12) and (13): (12) WFS for Human Numerals [X]Num $ [Xjon]Num; Cla; Hum (13) WFS for General Numerals [X]Num $ [XTa]Num; Cla; Gnl Bangla verbs agree with their subject for Person and Honorificity. While a pronoun, as in (14), must formally commit itself to an Honorificity value (Intimate, nonHon, or Hon), a noun is, within limits, free to refer to individuals of varying degrees of honour, as shown in (15): (14) ‘You will go tomorrow’, three variants: a. tui kal jabi. you.Intim tomorrow go.Fut.Intim b. tumi kal jabe. you.nHon tomorrow go.Fut.nHon c. apni you.Hon

kal jaben. tomorrow go.Fut.Hon

(15) ‘My student will go tomorrow’, two variants: a. amar chatro my student

kal jabe. tomorrow go.Fut.3p.nHon

14 Probal Dasgupta b. amar chatro kal jaben. my student tomorrow go.Fut.Hon Grammatically, any noun can take either Hon or nonHon agreement. When the noun means ‘baby’ or ‘goat’, Hon agreement signals irony. If the noun means ‘president’ or ‘queen’, nonHon agreement indicates a speaker’s intention of expressing disrespect. Such deviations do not jeopardize grammaticality. Pronouns, however, are committed to a specific feature value and insist on appropriate agreement. If this requirement is not met, the results are neither ironic nor disrespectful, but sharply ungrammatical – compare (14) with the following: (16) *tumi you.nHon

kal jaben. tomorrow go.Fut.Hon

(17) *apni you.Hon

kal jabe. tomorrow go.Fut.nHon

Particular nouns have no lexically specified absolute Hon values. Formally the freely assigned Hon value a given nominal phrase carries triggers agreement. Does the noun control this Hon value? (18) a. ei paMcjon this five.Hum

chatro student

kal jabe. tomorrow go.Fut.3p.nHon

b. ei paMcjon chatro kal jaben. this five.Hum student tomorrow go.Fut.3p.Hon ‘These five students will go tomorrow.’ (19) a. ei paMcTa chatro kal jabe. this five.Gnl student tomorrow go.Fut.3p.nHon ‘These five students will go tomorrow.’ b. *ei this

paMcTa chatro five.Gnl student

kal tomorrow

jaben. go.Fut.3p.Hon

Human classification features carried by the numeral are compatible with both nonHon and Hon agreement, we find at (18a,b). But (19) shows that a General numeral triggers nonHon agreement, sharply excluding Hon.

Transparency and Arbitrariness in Natural Language

15

What form should the proper description of this contrast take? This question represents one aspect of the Bangla Honorificity Agreement Problem. We have seen at (15) that a noun can in principle accept both values of Hon. That (18b) contrasts with (19b) shows that the General feature matrix resists Hon agreement, whereas the Human feature composition of the nominal phrase in (18b) permits it. Just what needs to be said to describe both this fact and the rigidity of pronouns? We would expect the exploration of properties of particular nouns to improve our understanding of these matters. If we had the space to consider nominal subcategory formats of the type studied by Ghosh (2006) here, the reader would have access to a wider data base, but our conclusions would stand; we thus refer the reader to the valuable paper by Ghosh and, for the moment, simply consider the properties of particular nouns like upacarjo ‘vice-chancellor (university president, rector)’: (20) a. ??ei dujon upacarjo kal jabe. this two.Hum VC tomorrow go.Fut.3p.nHon ‘These two vice-chancellors will go tomorrow.’ b. ei dujon upacarjo kal jaben. this two.Hum VC tomorrow go.Fut.3p.Hon ‘These two vice-chancellors will go tomorrow.’ (21) a. ei duTo upacarjo kal jabe. this two.Gnl VC tomorrow go.Fut.3p.nHon ‘These two vice-chancellors will go tomorrow.’ b. *ei this

duTo two.Gnl

upacarjo kal jaben. VC tomorrow go.Fut.3p.Hon

While chatro ‘student’ is a neutral noun, upacarjo ‘vice-chancellor’ carries an Hon expectation, which (20b) meets. If a speaker intends disrespect, one expects the o¤ensive use of language to go all the way, as in (21a), where the General (rather than Human) features of the numeral ‘two’ and the nonHon agreement on the verb ‘will go’ both express disrespect. (20a) is so puzzling as to sound like an error; the choice of the Human format for the numeral, together with the pragmatic default of honour for vicechancellors, leads us to expect an Hon verb, but we get the puzzlingly nonHon verb jabe instead. To rescue (20a), we can imagine it to be uttered by a senior figure who is so far above all vice-chancellors in status that s/he can

16 Probal Dasgupta a¤ord to use nonHon verbs for them casually, but who wishes to avoid blatant disrespect and thus sticks to the Human format. The need to imagine such a special viewpoint for (20a) is what makes it nearly uninterpretable. Now, suppose you are a disrespectful speaker and would use (21a). You would then say ‘The vice-chancellor will go tomorrow’ in the singular as (22). If you wish to show normal respect as in (20b), however, your choices are (23a,b). The Nuanced Individuation form, (23a), carries mild irony. The unformatted noun in (23b) can be diagnosed as a case of UG imposing a transparent stopgap in a niche left unoccupied by the arbitrary logic of classification formats in the particular grammar of Bangla. For some evidence for our diagnosis, note that even mild pejoration at duTi contradicts honour at the verb so severely as to nearly star (24a): (22) upacarjoTa kal VC.Gnl tomorrow

jabe. go.Fut.3p.nHon

(23) a. upacarjoTi kal VC.NuanIndiv tomorrow b. upacarjo kal VC tomorrow (24) a. ??upacarjo-duTi VC-two.NuanIndiv

jaben. go.Fut.3p.Hon

jaben. go.Fut.3p.Hon kal jaben. tomorrow go.Fut.3p.Hon

b. upacarjo-dujon kal VC-two.Hum tomorrow

jaben. go.Fut.3p.Hon

c. upacarjo-duTo kal jabe. VC-two.Gnl tomorrow go.Fut.3p.nHon ‘The two vice-chancellors will go tomorrow.’ In other words, speakers choose between the respect-preserving default (24b) and the overtly disrespectful alternative (24c), and have no use for (24a), except perhaps to convey extreme irony. But (23b) sounds normal and (23a) comes out as an only slightly ironic variant. What does this indicate? Our reading is that the UG default at (23b) and Bangla’s Nuanced Individuation format /NTi/ at (23a) must be stepping in to fill a languageparticular system gap. Revisiting (8a–c) helps identify the gap in question

Transparency and Arbitrariness in Natural Language

17

(the format /Njon/ is starred); but the point of interest is the availability of two fillers, not the gap itself. The language-particular system o¤ers a limited extension of /NTi/ at (23a) (limited in that (24) makes /NumTi/’s nonparticipation evident); UG o¤ers the option that an unformatted Hon noun can take on the definiteness features (see Dasgupta & Ghosh 2007 for details) normally associated with a classification formatted noun; neither of the fillers blocks the other. Throughout the discussion it has been clear that what are often treated as language particular quirks actually count as quirks in the discursive space of the culture in which Bangla is used. One cannot usefully separate expressions encoding respect in the language from the typical actions of expressing respect by its speakers. Taking this inseparability of the expressions from the expressing seriously is tantamount to accepting the substantivist claim that the arbitrary, word-carried particular realities of a language are discourse-anchored realities in a state of interplay with UG. How does this interplay work, though? More concretely, the question is: how are we to make sense of the fact that, in a context where resources have to be stretched to meet unusual needs, what the arbitrary or particular face of Bangla has to o¤er by way of a minimal stretching of the logic of classification formats in order to fill the gap that we first noted at (8) – namely, form (23a) – neither blocks nor gets blocked by (23b), the form that the transparent or UG-welcoming face of the language o¤ers as a filler for the same gap? To see that this fact contradicts the formalist doctrines, we need to first imagine, along formalist lines, a single dimension ranging from ‘‘most arbitrary/opaque’’ to ‘‘most motivated/transparent’’, conflating the concept pair ‘‘arbitrary/motivated’’ with the pair ‘‘opaque/transparent’’. We then note that such a fusion of opacity with arbitrariness entails the prediction that, of the two choices (23a, b), one would block the other. (A stronger, formalist-doctrine-focused variant of the prediction would say that only (23a) should count as well-formed, for (23b) flies in the face of the requirement in Bangla that a definite nominal should use a classification format on the noun.) Comparing this prediction, in either the weak or the strong variant, with the facts, we observe that the facts of (23) clearly disconfirm this prediction. Thus, the availability of (23a, b) confirms the substantivist conjecture that arbitrariness, a matter of provincial, discourseparticular quirks in a given language, needs to be distinguished from the UG concept of transparency.

18 Probal Dasgupta Perhaps more needs to be said. Once formalist colleagues place (23a, b) in what they see as a defensible theoretical framework, we will be in a position to say more, if need be.

Acknowledgements The research reported here was enabled in part by an Esperantic Studies Foundation travel grant in 2008.

References Dasgupta, Probal 1993 The Otherness of English: India’s Auntie Tongue Syndrome. New Delhi/London/Thousand Oaks: Sage. Dasgupta, Probal 2006 Bikiron, aakoronpokkho aar kaayaapokkho. Sahitya Parishat Patrika 113 (1–2): 147–57. Dasgupta, Probal 2007 Remarks on lexical kernel archiving as a cognitive science enterprise. South Asian Language Review 15 (2): 19–39. Dasgupta, Probal 2008 Knowledge and language in classical Indian linguistics: some observations. In Forms of Knowledge in India: Critical Revaluations, Suresh Raval, G. M. Mehta, Sitanshu Yashaschandra (eds.), 89–104. Delhi: Pencraft. Dasgupta, Probal In press Blocking: a reexamination. In No More Phemes: Explorations in Whole Word Morphology, Rajendra Singh (ed.), Berlin/New York: Mouton de Gruyter. Dasgupta, Probal ForthCharacterizations in linguistic science. In Science, Literature and coming Aesthetics, Amiya Dev (ed.). New Delhi: Centre for Studies in Civilizations. Dasgupta, Probal, Alan Ford, and Rajendra Singh 2000 After Etymology: Towards a Substantivist Linguistics. Muenchen: Lincom Europa. Dasgupta, Probal, and Rajat Ghosh 2007 The nominal left periphery in Bangla and Asamiya. In Annual Review of South Asian Languages and Linguistics 2007, Rajendra Singh (ed.), 3–29. Berlin/New York: Mouton de Gruyter.

Transparency and Arbitrariness in Natural Language

19

Ford, Alan, Rajendra Singh, and Gita Martohardjono 1997 Pace Panini. New York: Peter Lang. Ghosh, Sanjukta 2006 Honorificity-marking words of Bangla and Hindi: Classifiers or not? Bhashacintana 1.21–27. Singh, Rajendra, and Stanley Starosta (eds.) 2003 Explorations in Seamless Morphology. New Delhi/London/ Thousand Oaks: Sage. Tesnie`re, Lucien 1959 Ele´ments de Syntaxe Structurale. Paris: Klincksieck.

Reduplication and ‘echo words’ in Hindi/Urdu

Annie Montaut

The aim of this paper is to enquire into the various meanings of reduplication as a linguistic operation, and not as a merely stylistic or expressive device. The theoretical frame is Antoine Culioli’s ‘e´nonciative’ linguistics (notion and located occurrence, notional domain and boundary); context and intersubjectivity are taken into account as much as possible. The first section deals with total reduplication, within the nominal, verbal and adjectival category: it shows that reduplication on an occurrence modifies the relation between the reduplicated term and the term syntactically associated to it by denying the occurrence any specific stable value. It thus modifies the scheme of individuation of the notion (its actualization into an occurrence). The second section, dealing with partial reduplication or echo constructions, whether formed with a v- substitution to the initial consonant or with other forms of alliteration, shows that it modifies the notion itself by de-centring it, and reshapes it by taking into account various forms of heterogeneity, particularly the conflicting viewpoints of speaker and hearer.

Introduction Reduplication is a pan-Indian phenomenon regularly cited as one of the dozen features accounting for the consistency of the South Asian linguistic area1. It is however more often quoted than really analysed. Within the Indian area, studies on reduplication have emphasized its structural importance in respect to the linguistic area (from Emeneau 1969, the most solid study, to Abbi 1992) or listed its various forms and meanings in a given language (Abbi 1980). Singh’s 2005 contribution is the first one to give a detailed theory-based analysis of its morphology, formulating a nearly exhaustive set of explicit morphological rules for reduplication processes in Hindi/Urdu. He also associates reduplication to echo-constructions or partial reduplication and to semantic pairs, followed by Montaut (2008). As for the semantics, the most current hypothesis is the thesis of iconicity

22 Annie Montaut (Kouwenberg 2003), with the most interesting discussions bearing on the problems raised by various meanings apparently non iconic (Kyomi 1995). Do these three types represent a same operation (with distinct actualizations) or distinct operations? Here is an attempt to answer this question for the first two types of reduplication. In the first section, I will show that R (total reduplication: F-F) works on the occurrences of the notion: R is the trace of an operation which prevents singling out and locating any given occurrence; in the second section, I will show that the echo-construction (F-F 0 ) modifies the notion itself, which no longer remains centred, whether its traces in R involve the regular v- alteration or some other kind of alliteration. The terms ‘‘notion’’ and ‘‘occurrence’’ as used here belong to the theoretical framework of Culioli (1990a, 1990b, 1999). A notion or notional domain (Culioli 1990b: 181) ‘‘can be defined as a complex of physico-cultural representations with no extensional properties’’ (it is a purely qualitative categorization, purely intentional, for instance ‘‘dog’’). A notional domain has a centre (in X, what is typically X), and a boundary which delimits its Interior (I) from its Exterior (E). The centre of the notion ‘‘dog’’ for instance is a dog fully conforming to the properties usually associated with it, what we can truly call a dog. ‘‘To construct the extension of the notion is to construct its occurrences’’ (a dog, the dog, this dog, many dogs, etc.), which are ‘‘distributed in relation to the organizing centre of the domain’’ (an occurrence is then locatable: absolute value, referring to the type, is attached to the centre of the domain, whereas relative values decrease as ‘‘you move away from the centre’’). Constructing the occurrences is the basic scheme of individuation of a notion (it amounts to constructing the extension of the notion), and it consists in an operation of quantification together with qualitative sub-categorization. In the construction of occurrences, the basic operation is that of extraction: ‘‘ascribing an existential status to a situated (located) occurrence of a notion’’, extraction ‘‘brings into existence an individuated occurrence that has no other distinguishing feature than the fact that it has been singled out’’ (Culioli 1990b: 182)2. Other operations in the construction of the occurrence involve re-identification (pinpointing: ‘‘this dog which we are referring to, the same dog’’) and scanning. Scanning means that you have to scan the whole notional domain without finding a possible stable location (‘‘any dog, which dog’’). A notional domain may be represented as homogeneous (typical values: really p) or containing non-typical values (not really p, verging on p 0 or non p) and so including heterogeneity (Culioli retains ‘alterity’ for French ‘‘alterity’’).

Reduplication and ‘echo words’ in Hindi/Urdu

23

What follows shows that reduplication modifies the scheme of individuation of the notion (integral reduplication) or the notion itself (partial reduplication): it is obviously far more than a stylistic device or a ‘‘way of speaking’’, a categorization which implicitly denies R the status of linguistic category and make it an exotic phenomena. The two recently published collections (Hurch 2005 and Kouwenberg 2003) provide the reader with an important mass of data, yet do not always give the appropriate contextualisation for fully understanding the meanings of the data presented. The aim of this paper is to enquire into the various meanings of reduplication as a linguistic operation, and not as a merely stylistic or expressive device, with appropriate contextualisation. In Hindi, reduplication provides for an important part of the lexicon, both verbal and nominal3, as well as for grammatical structures (distribution, iteration): it belongs to the core of the language, if we hold language to be the regulated organization of given lexical material. It also provides many ‘‘manners of speaking’’, ‘‘stylistic or expressive uses’’, which do not obey easily recognizable constraints and are all the more di‰cult to grasp since they present great variation even between users of the same language, and involve the speaker’s subjectivity. The paper will deal with the two main areas of reduplication: total or integral reduplication (R), where the whole unit (F) is reduplicated in the same form (F) (R¼F-F), and echo constructions or partial reduplication (E), where the first unit (F) is altered in the second occurrence (F 0 ) in a more or less systematic way (R¼F-F 0 ). 1. Total Re´duplication: Non-centering of the occurrence The reduplication (R) of an entity modifies the relation of this entity with one or several of the other constituents in the statement: for a noun, R modifies the relation of this noun with the predicate, for an adjective, R modifies the relation between the noun and the adjective, for a verb (always a dependent one when reduplicated in Hindi), R modifies the relation of the dependent predication with the main predication. 1.1. Nouns and numerals Distribution is the most frequent meaning, often considered as prototypical for the nominal class. In its restricted meaning (for each X, n Y), it however

24 Annie Montaut occurs only with numerals, where R involves more than one relation with the other constituents, which makes it more complex even if it is perceived as more basic.

1.1.1. The typically distributive meaning: Numeral-numeral nom Apart from the iteration of the process for each occurrence of the beneficiary in (1a), ‘‘give one X (to¤ee) to each Y (child)’’, hence the possible commutation of (1a) with statements having the indefinite har ‘each’ (2a), the reduplication of the numeral acts on the scheme of the individuation of Y (n occurrences of ‘‘child’’) as well as that of X (n occurrences of ‘‘to¤ee’’)4: (1)

a. baccoN ko ek-ek taˆf ˆı do child-p dat one-one to¤ee give ‘give a to¤ee to each child, one to¤ee per child’ b. baccoN ko ek keji tafiyaˆN do child-p dat one kilo to¤ees give ‘give one kilo to¤ees to the children’

(2)

a. har bacce ko ek taˆf ˆı each child-s dat one to¤ee ‘give a to¤ee to each child’

do give

b. bacce ko ek taˆf ˆı do child-s dat one to¤ee give ‘give a to¤ee to the child’ (definite occurrence) (1a) shows that the beneficiary is the class of children, morphologically plural, whereas (2a) refers to this same class by a singular, ‘‘each child’’. On the one hand, we cannot set a definite referential value for ‘‘a child’’, so that there is no locatable occurrence which we may construct, and on the other hand plurality as constructed by the reduplication of the numeral in (1a) is distinct from plurality as an homogeneous group, in (1b) for instance with the morphological plural, by the fact that each unit is isolated and individuated as a beneficiary (hence the equivalence between (1a) and (2a)). In (1a), the beneficiary is characterized as a non-global plurality which is formed by the exhaustive collection of all the distinct singularities within the set, with no possibility of selecting any of them. The reduplication of

Reduplication and ‘echo words’ in Hindi/Urdu

25

the numeral acts as a variable which makes it necessary to scan the whole set of occurrences without being ever able to stop on any specific occurrence5, exactly as does the quantifier har ‘each’.

1.1.2. The «listing» e¤ect: noun or pronoun in the singular The reduplication of singular relatives or interrogatives gives the meaning ‘‘each element, with no exception’’, and suggests a complete series which, again, constructs a plurality made of n singularities, in a non cumulative and non interchangeable way, so that there is no single occurrence we can pick up and locate, and we have to go through the whole set of occurrences – a typical case of scanning too: (3)

a. tum kahaˆN kahaˆN gae? tumne kyaˆ kyaˆ dekhaˆ? you where where went you-erg what R saw? ‘where did you go?’ ‘what did you see?’ (give a list of all and every place) b. jo-jo aˆegaˆ use bataˆnaˆ ki maiN who who will-come 3s-dat say that 1s ek ghaNTe baˆd aˆuˆNgıˆ 1 hour after come-fut ‘say to whoever will come (to all and every visitor) that I will come back in one hour’

The reduplication of singular nouns, which often creates intensive meanings or even amounts to presenting the entity as an extreme, can be explained in the same way: intensiveness in (4a) results from the construction of an exhaustive series, with all its elements collected one by one, hence the e¤ect of an integral hair-rising; in (4b) ‘‘know’’ is predicated not about an occurrence but about a set of occurrences (each of them being considered as a singular occurrence), which tends to mean that its validity is above any contingency; as for the meaning ‘‘even’’, it results from the improbability of the relation between predicate and noun (know / child). (4)

a. uskaˆ rom-rom tharraˆ uThaˆ his hair-ms-hair-ms rise get up-aor-ms ‘each of his hair rose up /his hair rose up all over his body’

26 Annie Montaut b. baccaˆ baccaˆ jaˆntaˆ hai child-ms child-ms know pres-3ms ‘the last boy is aware/ even a child knows that, every child including the last one knows that’ Here we construct a set which is distinct from the ordinary (homogeneous) plural by the fact that each constituent retains its singularity and is not fused into a global whole, and at the same time it cannot be located in isolation. This distinction between a set of individualities and a global atomic purality, two di¤erent meanings of plural, has been worked out in FassiFehri and Vinet (2001). In (4) as well as (1a) both plurals are of the first type; but in (4) we construct plural out of singular, whereas in (1a), ‘‘give a to¤ee to the kids’’ (¼ to each of them), we construct singular out of plural, since we reconstruct the beneficiary, out of a homogeneous plural, as unique for every to¤ee distributed.

1.1.3. Reduplication of plural nouns It is less common, and even less frequently mentioned in the relevant literature, with the meaning ‘‘exclusiveness’’ or ‘‘restrictiveness’’. Reduplication of plural nouns constructs the notional domain ( p) in relation to its complementary p 0 (non p or other than p), a meaning which can be reinforced by the exclusive particle hıˆ: (5)

a. yahaˆN mahilaˆeN-mahilaˆeN baiTheNgıˆ here women-women will-seat ‘here only women / women and only women will seat’ (context: there are too rooms, one for men, one for ladies) b. bookmarkoN-bukmaˆrkoN meN hıˆ baˆt hotıˆ calıˆ gaıˆ bookmarks-bookmarks in just speech be went ‘the conversation went on exclusively by means of bookmarks’ (two lovers strictly looked after by the girl’s family: M. Joshi, K)

The operation in fact always deals with a set of occurrences and not with the notional domain. The statement (5a) is meaningful only within a context where the set has been selected in a paradigm where it is opposed to the other elements of the paradigm. Within a context of segregation of women

Reduplication and ‘echo words’ in Hindi/Urdu

27

vs men: with reference to the meeting hall, the reduplication of ‘‘women’’ means that the opposition women / men is disqualified in order to focus and homogenize on ‘‘women’’ (‘‘women-women’’ meaning ‘‘women and not other-than-women’’: p and not p 0 other than p, p in relation to p 0 other than p). In (5b), whereas in the beginning the lovers used various devices to communicate (the bookmarks being only one of these), now we focus on and homogenize ‘‘bookmarks’’ (irrespective of other communication devices), which amounts to disqualifying the other devices previously resorted to. The restrictive meaning (restriction to the set X, exclusion of other than X) is associated with contexts with a previous segregation. In (5a), such a context relies on, apart from the institutionalisation of sexual segregation, the announcement of a meeting concerning women. In (5b), where the narrative context alone can fulfil such a segregating function, hıˆ ‘only, just’ is required to block the distributive meaning (‘‘in each and every bookmarker’’). The meaning involved here, di¤erential qualification of a set, is less grammaticized in the language than the distributive meaning since we may add the restrictive particle hıˆ after the reduplicated form (mahilaˆeN hıˆ mahilaˆeN ), whereas har ‘each’ and reduplication cannot cumulate.

1.2. Reduplication of the verb: Iteration of the process In the verbal class, only nonfinite verbal forms can reduplicate with the pattern F-F6. Various occurrences of reduplicated participles (imperfect or present/perfective or past, conjunctive participles (V-kar) are grammaticized in various types of iteration, the typical meaning of non-centring when processes are concerned. Since reduplicated participles are already dependent verbal forms, the occurrence of the process cannot be localized but by the main verb: R cannot be, as it is in the nominal category, responsible for the non-localized, non-stabilized status of the occurrence, in need of localization. R indeed a¤ects the occurrence in need of localization in such a way as to multiply it into n similar occurrences, none of which is the best (definite) value, but all of them construct a series which fragments the process (or make the state into a durative) and draws its meaning from the relation with the main finite verb. Iteration within the verbal category and distribution within the nominal category are thus symmetrical. In this way, with an action or event main verb as in (6a), not to speak ‘‘eating-eating’’

28 Annie Montaut means that each word/statement is associated with an occurrence of eating, hence the illusion of more simultaneity7, and in (6b), the reduplication of the past stative participle ‘‘slept-slept’’ with main verb ‘‘die’’ means that at some moment in this state (sleep) he died, hence the appearance, here again, of a perfect concomitance. In (6c) the reduplicated conjunctive participle Tahal-Tahalkar multiplies the occurrences of wandering, so that the process may appear more imprecise (non-telic) but this indefiniteness comes from the semantics of the verb. In (6d), the iteration of n occurrences of ‘‘laugh’’ adds a meaning of intensity, here again a side-e¤ect of the basic operation of de-centring by scanning, while (6e–f ), with two action processes, displays the basic e¤ect of R when non iterative, that is, giving temporal width to the dependent process (no one single locatable occurrence): (6)

a. khaˆte-khaˆte mat bolo khaˆte (hue) mat bolo eating-eating neg speak-imper eating been neg speak! ‘do not speak while eating / don’t speak when eating’ b. soye-soye mar gayaˆ slept-slept die went he died in his sleep

/

???soye mar gayaˆ slept die went ???in his sleep he died’

c. maiNne Tahal-Tahalkar saˆraˆ din kaˆTaˆ 1s-erg wander-wander-Cp whole day cut ‘I spent the whole day wandering (here and there, in various places)’ d. usne haNs-haNs-kar puˆrıˆ kahaˆnıˆ sunaˆˆı 3s-erg laugh-laugh-CP entire story told ‘he told the whole story laughing (a lot, at many times)’ e. maˆrg meN calte-calte aˆj maˆN se ek savaˆl puˆchhaˆ street in walking-walking today mother to one question asked ‘today, while walking on the road, I ask Mother a question’ f. jaˆte-jaˆte ve kahte going-going 3p said ‘he used to say while going’ (from Santapt, Chauhan like (6e)) Both the following series exhibit a specific relation between reduplicated participle (conjunctive, accomplished or unaccomplished) and main verb. If the main verb represents a transient state and the dependent participle an action, the relation is causal (cause-e¤ect: cf. Abbi 1980), which amounts

Reduplication and ‘echo words’ in Hindi/Urdu

29

to representing a series of iterations leading to a result (successful accumulation) such as in (7): (7)

a. yah caˆdar dhul-dhulkar phaT gaıˆ this sheet wash-wash-cp tear went ‘this sheet got torn by/following repeated washings’ b. tumhaˆrıˆ shikaˆyat sunte-sunte (sun-sunkar) main your complaint hearing-hearing (-cp) I ‘I was fed up listening to your complaint’

uˆb gayaˆ thaˆ bore go ppft

c. vahaˆN akele baiThe-baiThe uˆb gayaˆ here alone sat-sat bore went ‘he got fed up of sitting here (inactive)’ d. rote-rote uskıˆ aˆNkheN suˆjh gaıˆN crying-crying his eyes swell went ‘his eyes got swollen by (repeatedly, thoroughly) crying’ If the main verb represents avoidance (« be-saved/escape », « remain, stay ») and the participle action or event, reduplication means that the accumulation of occurrences does not reach the normally expected result, hence the lacunar value in (8): (8)

a. vah girte-girte bac gaya 3s falling-falling escape went ‘he almost fell (failed to fall)’ b. yah baˆt hoNThoN par aˆ-aˆkar rah gaıˆ this thing lips on come-come-conj.part stay went ‘I was about to say this thing but did not’ c. baˆrish hote-hote rah gaıˆ rain being-being stay went ‘it almost rained (but did not)/it was about to rain but did not’

In statements like (7) where the relation is between a dependent action verb and a main verb expressing a transient state, reduplication is necessary for the cause-e¤ect meaning (9a), and a non-reduplicated participle will produce simple concomitance between both processes (9b). Moreover, a nonreduplicated participle not only fails to produce the avoidance meaning in statements like (8), but it is non-grammatical with main verb meaning

30 Annie Montaut ‘‘escape’’ (10a) while the reduplicated participle with an action main verb means concomitance (10b): (9)

a. sıˆtaˆ kaˆm karte (hue) thak gaıˆ Sita work doing (been) got tired ‘Sita got tired when working b. kaˆm karte karte thak gaıˆ work doing doing got tired got tired of/by working’

(10) a. ???vah girte bac gayaˆ 3s falling escape went ‘while falling he screamed’ b.

girte girte zor se cillaˆyaˆ falling falling strong screamed

Non-centring is responsible for the special meanings of (7–8). A reduplicated dependent process in relation with a state (or change of state) main verb respectively may either entail a result if repeated or on the contrary drag on without any result. In other terms, R tends to make the dependent process more autonomous from the main verb, which modifies the simply temporal concomitance. Needless to add that scholars claiming for the iconic interpretation of reduplication have granted a central role to distributive and iterative meanings, similarly to the plural meanings in languages which display it. ‘‘Twice is meaningful’’: if one assumes that ‘repeat’ always amounts to ‘‘say more’’, distribution, iteration and intensity which are often correlated (4, 6d) are obviously in conformity with this intuition. Fragmentation and dissemination, as well as avoidance, to the extent that such meanings point to nonsingle-time processes, have also been claimed to be indirectly iconic (Kouwenberg & LaCharite´ 2001, 2005). However it is quite clear that, even with such ‘‘prototypical’’ meanings, the supplement supposedly added by R is only the trace of an operation which does not amount to say more but modifies the relation between N and V or dependent V and main V. The meanings of reduplication observed above in examples (1a), (3), (4), (6), (7), (8) such as distribution, list effect, iteration, do not amount to saying more but to conceive di¤erently the

Reduplication and ‘echo words’ in Hindi/Urdu

31

relation between the reduplicated entity and the constituents with which it is syntactically associated in the statement. The case of the adjective is even more revealing, since the meanings of R are more proliferating.

1.3. Reduplicated adjectives: Degree, expressivity or neutralisation of the di¤erential property? Within the frame of iconicity, high degree (intensity: ‘‘much, very, quite, completely A’’) derives quite naturally from the postulate ‘twice means more’. Low degree, as well as medium degree, more and more commented with the growing presentation of data and descriptions, needs on the contrary some justification8. Such a justification is proposed in a clever argumentation (Kouwenberg 2003, Kouwenberg & LaCharite´ 2005) by means of fragmentation and dissemination, forms of discontinuity in their own right, which parallels distribution as a form of discontinuity. A colour which is not represented as plain and saturated but appears in the form of spots, stripes or scattered zones, that is, in discontinuity, represents the ‘‘lacunary’’ meaning, and by extension the low degree. The animal with black spots is then what links the animal completely black or very black to the animal not really black. But here again, the Hindi data display a series of meanings including many other values than these two polar cases, a series which besides rarely display the real high degree, but which shows that reduplication of A, like N or V, a¤ects the relation between the syntactically associated constituent, in the matter N and A. AA N modifies in a systematic way the relation A-N, namely the attribution of the property A to the noun N, with the various meanings resulting from the di¤erent nature of nouns (discrete or compact) and the context which actualizes AN as an occurrence (with or without preconstruction, as a mode of presence, as the construction of the property)9. As opposed to the simple adjective, the reduplicated adjective is not descriptive10. Normally an adjective qualifies the noun by conferring to it a di¤erential property (which makes it descriptive): lambaˆ laRkaˆ ‘tall boy’, or lambe laRke ‘tall boys’, refers to a boy or a set of boys with tallness as a distinctive property (as compared to other boys, small and medium-sized); being tall here is a di¤erential property. The reduplicated expressions, on the contrary, lambaˆ-lambaˆ laRkaˆ, as well as lambe-lambe laRke, do not identify

32 Annie Montaut the noun as being qualified by the property ‘‘being tall’’ in contrast with other possible properties, but suggests either that the relation boy-tall is already pre-constructed (the noun is pre-identified by the property ‘being tall’ and reduplication stands for a valuation of this tallness, in a subjective appreciation11) or that it has a distributive meaning – each of the boys is tall, the only commented meaning for plural in the relevant literature.

1.3.1. ‘Intensity’ and ‘high degree’: Surface e¤ects of various operations Existing descriptions are mainly compatible with the iconic function of R. But before studying the most interesting examples of adjectival reduplication, it is worth noticing that none of the generalities commonly found in existing descriptions holds against counter-examples: for instance reduplicated adjectives with plural nouns should always be distributive12, and reduplicated adjectives with singular nouns should be intensive (high degree: ‘‘very much’’), while reduplicated adjectives of colour or taste should always mean approximation or low degree (‘‘almost, more or less’’). In reality, the reduplicated adjective with a plural noun rarely has a distributive meaning: choTe-choTe bacche (small-small children) for instance rarely refers to a set of children where each of them is small, although it of course may do so in some contexts, but rather expresses that the speaker is in an empathic relation with the children, who are not particularly tiny besides, but, as little children, suggest a¤ectionate thoughts. It may simply be the plural of the singular reduplicated expression. As for the example (11), it is a conventional portrait of beauty and not a cartoon-like negative description as would suggest the standard interpretation of the reduplicated colour adjective in the low degree (if you take baˆl ‘hair’ as a collective singular) and of the reduplicated ‘‘big’’ with a plural noun as distributive; the alternative interpretation of the reduplicated colour adjective, with a plural baˆl ‘hair’, would be equally displaced within the general tone of such a portrait, which is appreciative for all Hindi speakers. (11) uske baˆl kaˆle-kaˆle the, uskıˆ aˆNkheN baRıˆ-baRıˆ thıˆN her hair black-black were, her eyes big-big were ‘she had very (dark) black hair, large (attractive) eyes’ ???she had blackish hair / each of her hair was black and each of her eyes was big

Reduplication and ‘echo words’ in Hindi/Urdu

33

As for (12a), R is certainly ‘‘intensive’’, but is not equivalent to high degree, as shown by the unnatural character of reduplication for expressing excess, according to most speakers (12c): (12) a. yah rahaˆ tumhaˆraˆ kok, pıˆ lo, ThaNDaˆ-ThaNDaˆ hai here is your coke, drink take, cold-cold is ‘here is your coke, drink it, it is nicely cold’ (not ‘‘very cold’’) b. yah lo tumhaˆrıˆ caˆy. Garam-garam hai, this take your tea hot-hot is, ‘take your tea. It is nicely hot, drink it’

piyo drink

c. yah lo tumhaˆrıˆ caˆy. Garam / bahut garam / this take your tea hot / very hot / ??garam-garam hai, abhıˆ mat pio ??hot-hot is right now neg drink ‘here is your tea, it is very hot, don’t drink it now’ In (12), R does not correspond to high degree but to the optimal degree, the ideal temperature for a good tea according to the speaker and addressee. ‘‘Ideal’’ means that the degree of hotness is conform to the expectations of the drinkers on the basis of the speaker’s (and hearer’s) cultural habits and knowledge. ‘‘Very hot’’ is descriptive: it measures a degree and intends to be objective, and can be contrasted with comparative (X is hot but Y is hotter). ‘‘Nicely hot’’ does not allow such a contrast (*X garam-garam hai par Y zyada garam(-garam) hai) because is does not measure the degree. It is not descriptive and that is the reason why it sounds somewhat like a subjective appreciation. Similarly, vegetable and fruit sellers in markets, when they advertise their goods, systematically reduplicate the adjective (freshfresh vegetable, fresh-fresh news, hot-hot pakaure, etc.). Quality is emphasized, but rather for its adequateness to the customer’s expectations than for its objective degree. In such examples, the noun (compact) is the bearer of the property and its relation with the adjective has already been constructed, freshness or hotness being pre-requisite qualities in the given contexts. A simple adjective would simply indicate that the tea is neither cold nor lukewarm, but R neutralizes the feature ‘‘di¤erential’’ in the property in order to emphasize its manifestation: conformity with the typical ideal of good tea (nicely hot, but precisely not too hot) sets the value, shared par the subject S and hearer as imagined by S. Similarly, the pakaure are not described as hot in contrast with cold, and the vegetable are not described

34 Annie Montaut as fresh rather than rotten or dry, but as hot-pakaure and as fresh-vegetable whose quality is commented as ideal. There is no room for a di¤erent evaluation which could exclude the property (hot, fresh), the speaker does not a priori allows the possibility for the hearer to conceive the entity with another property. This explains that the property is represented as saturated, in conformity with what the speaker imagines concerning the hearer’s expectations, and that such constructions trigger empathy with the hearer13. As for the colour of grass in (13), it is neither greenish nor pale green, it is fully green and saturated, in conformity with the Bollywood stereotype of pleasant scenes, even if the noun is in the singular (supposed to automatically shift to the low degree with R): (13) peRoN kıˆ chaˆNv meN kuch yugal taruˆN harıˆ-harıˆ mulaˆyam trees of shadow in some couple young green-green tender ghaˆs par cahalqadmıˆ kar rahe the grass on stroll were doing ‘some youths in couple were strolling under the shadow of the trees, on the green tender grass’ (Himanshu Joshi, Yatharth) Here reduplication grants the property both an appreciative character and saturation. Subjective appreciation accommodates qualitative variation, which is only possible because the property has previously been preconstructed as homogeneous in contrast to the various choices maintained as other possibilities when there is no R. For instance when a gardener wishing to grow a lawn and selecting his plants among various colours, will ask for green (or dark green, or pale green, eventually yellow, etc.) in using the simple adjective (pointing to basic heterogeneity: colour as a differential property), and not R. In a similar configuration, a ‘‘blue-blue sky’’ is most often interpreted as ‘‘quite blue’’, ‘‘really blue’’, ‘‘very pure’’: (14) yah nilaˆ-nıˆlaˆ aˆkaˆsh dekhkar mujhe pahaˆR meN bachpan kıˆ yaˆd aˆˆı this blue-blue sky see-CP 1s-dat mountain in childhood of memory came ‘seeing such a blue sky I remembered my childhood in the mountains’ The pre-construction here is not cultural but situational: the speaker does not describe the sky with its colour as he discovers it when stepping outdoors, but dreams over the associations he can relate to the blue sky which

Reduplication and ‘echo words’ in Hindi/Urdu

35

he is confronted with, as such. The colour of the space of the sky (bearer of the property) is already made homogeneous. In contrast, a speaker for weather forecast cannot describe the sky by using reduplication, even if he wishes to emphasize its perfect blueness and clarity, because what he aims at is communicating information on the colour of tomorrow sky (neither grey nor black nor covered), with no pre-construction. Similarly a peasant who says ‘‘if the sky is blue tomorrow (the weather is fine) I will begin the crop’’, cannot use R because the anticipated blue of the sky is only one possibility among others and retains its meaning of di¤erential property. More generally, a sky which is blue (not grey, not black) is expressed by the simple adjective, whereas a blue-sky (which may be particularly blue, evocative, or else) is expressed with R14.

1.3.2. ‘‘Low degree’’ and attenuation: Di¤erent operations according to the semantics of the noun (discrete, compact, dense) In a general way, the attenuative (approximation, diminution) meaning is related to the mode of presence of an entity. The notion of ‘mode of presence’ relates to a particular stand of the speaker: tell the world (a given entity of the world) such as he is confronted to it (and not in a descriptive, analytical, way), hence the a‰nities with verbs of perception. What is crucial is not the fact that the adjective refers to singular, nor even that it expresses a colour or a taste, but its relation with the noun. In the singular (15a) as well as in the plural (15b), the property ‘‘blue’’ assumes the attenuative meaning ‘‘bluish’’ with a noun of the category ‘dense’ (mountain crests, water-stream of a mountain river) in a relation adjectivenoun pertaining to the mode of presence. (15) a. nice nıˆlaˆ-nıˆlaˆ paˆnıˆ bahaˆ rahaˆ thaˆ below blue-blue water flow prog impft ‘bluish water was running below’ b. saˆmne nıˆle-nıˆle pahaˆr kıˆ rekhaˆ dikhaˆˆı de rahıˆ thıˆ in-front blue-blue montains of line be-seen prog impft ‘the bluish line of the mountains could be seen in front’ Such examples as (15) can be contrasted with the ‘‘truly blue sky’’ of (14) which implied a pre-constructed relation, whereas (15) pertains to direct

36 Annie Montaut perception15. In the series (16), we may similarly explain the attenuative meaning (16a) by the association to a noun categorized as discrete (paper) of a property which is presented as a mode of presence (yellow-yellow): without any verb of perception, narration itself constraints the viewpoint of an observer in the process of getting aware of what is visible around him. In contrast, in (16b), at a wall paper sender, the selection of a roll of paper chosen for its di¤erential property (yellow, and not blue or green) requires the simple (non-R) adjective ( pıˆlaˆ-vaˆlaˆ: ‘‘the yellow one’’)16; only if the hearer (Culioli’s co-e´nonciateur: S1) answers by repeating the already selected colour, reduplication can occur, since it presents the mode of construction of the property (the colour yellow-yellow itself ), with an homogeneization on ‘yellow’, so that the meaning is saturation (bright yellow, frankly yellow). (16) a. ek puraˆnıˆ ciTThıˆ paRıˆ thıˆ, pıˆle-pıˆle kaˆgaz par caˆr shabd likhe the an old letter fall had been, yellow-yellow on 4 words written were ‘an old letter was lying on the floor, four words were written on the yellowed (yellowish) paper’ b. A. pıˆlaˆ-vaˆlaˆ lo ! – B. yah pıˆlaˆ-pıˆlaˆ rang kamre meN acchaˆ nahıˆN lagegaˆ yellow-that take ! this yellow-yellow colour room in good neg willseem ‘A. take the yellow one ! B. this bright yellow won’t fit in the room’ Besides, there are statements which allow both interpretations, like (16d) which associates the reduplication of ‘‘yellow’’ to the noun ‘‘mango’’ in a nursery rime for children and may have the reading ‘‘intensive’’ or ‘‘attenuative’’. Interpreted as a direct perception (dekho ‘look’), the property is constructed as a ‘mode of presence’ and means ‘‘greenish yellow’’, ‘‘yellowish’’ (particularly since these mangos are fresh and not extra-ripe); interpreted as generic and because of that in disjunction from any specific perception (‘‘the king of fruit’’), the property is saturated and means ‘‘truly yellow’’, ‘‘a nice /deep yellow’’. (16) c. dekho, kitnaˆ aˆm rasıˆlaˆ / look, how-many mango juicy / lagtaˆ kitnaˆ taˆzaˆ hai / seems how-much fresh is /

chilkaˆ uskaˆ pıˆlaˆ-pıˆlaˆ skin its yellow-yellow aˆm phaloN kaˆ raˆjaˆ hai mango fruits of king is

Reduplication and ‘echo words’ in Hindi/Urdu

37

‘look, how juicy is the mango/ its skin deep yellow (yellowish) / how fresh it looks / mango is the king of fruits’’ Whatever the meaning, approximation or saturation, it flows from the neutralization of the di¤erential property in the adjective. That is why reduplication cannot be correlated or contrasted with the simple adjective for the same quality as seen in (16d)17: (16) d. *yah langRaˆ aˆm khaTTaˆ hai, par vah dashaharıˆ aˆm khaTTaˆ-haTTaˆ hai this langra mango sour is, but that dashahari mango sour-sour is (*this Langra mango is sour but that Dashahari is sourish) An adjective (simple) conveying a di¤erential property cannot correlate with an adjective conveying the neutralization of a di¤erential property. Finally, it is noteworthy that low degree is not exclusively restricted to colours and taste, since the reduplication of an adjective, whether or not referring to colour/taste, is compatible with the approximation su‰x -saˆ18. Certainly, the reduplication of a colour or taste adjective is always attenuative with this su‰x: pilaˆ-pilaˆ-saˆ ujvaˆlaˆ (yellow-yellow-like brightness), ‘a vaguely yellow brightness, a pale luminosity’, kaˆlaˆ-kaˆlaˆ-saˆ kapRaˆ (blackblack-like cloth) ‘blackish garment’. But other adjectives in similar contexts (-saˆ) may also have, if not systematically, the low degree meaning. bholaˆbholaˆ saˆ laRkaˆ, simple-simple-saˆ boy ‘a rather naı¨ve boy’, Tarch kıˆ marıˆmarıˆ-sıˆ roshnıˆ, torchlight of dying-dying-saˆ light ‘quasi dying /agonizing light of the torchlight’, but simTaˆ-simTaˆ-saˆ baccaˆ shrank-shrank-saˆ child, ‘(totally) curled over child’. It is then very clear that integral reduplication, whether of nouns, verbs or adjectives, is not only a ‘‘way of speaking’’ to be treated as a stylistic or expressive device, nor is it, as claimed by the theory of iconicity, a manner of saying more. Certainly intensive meanings and lacunar meanings (the latter verging on augmentation and multiplication by means of dissemination and fragmentation) are very frequent. But we have seen that these meanings can be accounted for by an operation of non-centring which acts on the relation NV, NA or V2V1 respectively. This operation has nothing to do with iconicity (or very indirectly), but it modifies the scheme of individuation of the notion into an occurrence.

38 Annie Montaut 2. Echo-constructions: De-centring the notion It has been shown above that integral reduplication amounts to taking into account the n occurrences of the reduplicated term. As a non-centring device allowing for qualitative variation (each occurrence retains its singularity), it rules out the possibility of assigning a specific definite value to the occurrence and redefines the relation between the reduplicated term and another term in the statement: noun and verb, participle (dependent verb) and main verb, adjective and noun, with the typical meanings of distribution and iteration but also with a whole series of di¤erent meanings. The echo construction (a phonetic alteration of F into F 0 ) bears on the notion: it works as an extension of the notional domain (‘‘N and other similar things’’). It is on the domain and not on the occurrence that the echo construction operates as a de-centering process: it introduces here ‘‘alterity’’ (heterogeneity) by simultaneously taking into account several viewpoints on the notion associated to the reduplicated term. These heterogeneous viewpoints correspond to a particular structuration of the notional domain in di¤erent zones, each of them defining a distinct mode (zone) of the notion (and a viewpoint on it): Interior, noted below I (really p: the viewpoint is centred), boundary, noted I-E (not really p: the viewpoint is de-centred, but compatible with I), and Exterior, noted E (other than p: the viewpoint is de-centred and not compatible with I)19. Echo constructions mean that, parallel to the centred viewpoint, a de-centred viewpoint is taken into account: this second viewpoint is constructed either as coextensive to the first one, or as opposed to it (as a deconstruction of the viewpoint which is centered), or it can also represent an alternative with no possible choice. It results from this that many usages of the v- alteration, not mentioned in the relevant literature, tend not to create a mere extension or approximation of the notional domain but to create parody, or derogatory meaning. An instance of the mere extension of the notional domain is the classical caˆy-vaˆy (tea-echo), ‘‘tea and other eatable and drinkable’’, shaˆdıˆ-vaˆdıˆ (marriage-echo) ‘marriage and so on’, pen-ven ‘pen and the like’. The ‘‘semantic supplement’’ glossed by ‘‘etc.’’, ‘‘and so on’’, ‘‘and all’’, ‘‘e tutti quanti’’, ‘‘and the like’’ in classical descriptions such as Abbi (1980), show that we get out of the centred domain (tea which is really tea, what we can call tea, tea-tea) and include the neighbouring notional domain or stay at the margins of the domain: tea and other drinkable or eatable which can

Reduplication and ‘echo words’ in Hindi/Urdu

39

be associated to the ritual afternoon or morning tea, pen and other necessary stu¤ used for writing20.

2.1. Form of the canonical constructions in vThe first consonant of a mono- or poly-syllabic word is replaced by v-: shaˆdi (mariage)-vaˆdıˆ ‘marriage etc.’, caˆy(tea)-vaˆy, ‘tea etc.’, paRhnaˆvaRhnaˆ ‘read etc.’, if we retain temporarily the standard translation as given in the relevant literature for such expressions. When the vowel in the first syllable is rounded, the initial consonant disappears: ghoRaˆ ‘horse’ ghoRaˆ-oRaˆ, ‘horse etc.’. When there is no initial consonant in F, F 0 is formed by adjunction and not substitution: aˆtmaˆ-vaˆtmaˆ ‘soul etc.’. If the word begins with a consonant cluster, the second too is retained: kraˆntivraˆnti ‘revolution-etc.’, prem-vrem ‘love-etc.’. We can then set the following rule: for F ¼ C-, F 0 ¼ v-; for F ¼ Co/u-, F 0 ¼ o/u-; for F ¼ V-, F 0 ¼ vV-. Both units F and F 0 vary when the word allows variation (number, case for instance): pakauRaˆ-vakauRaˆ ‘vegetable fried preparation etc.’ has the plural form pakauRe-vakauRe, since singular masculine nouns in -aˆ inflect to -e in the plural, and laRkıˆ-vaRkıˆ ‘girl-etc.’ substitutes the plural ending -iyaˆN to the singular ending -ıˆ: laRkiyaˆN-vaRkiyaˆN. English words are reducplicated under the same conditions: pen-ven ‘pen etc.’, Taim-vaim ‘time-etc’, noTis-voTis ‘notice etc.’, and if required inflect according to the native Hindi system (plural feminin for instance, -ıˆ > -iyaˆn: paˆrTiyaˆn-vaˆrTiyaˆn). The variation is similar for Persian words (shaˆdıˆ-vaˆdıˆ ‘marriage’), Arabic words (qismat-vismat ‘destiny’, talaˆq-valaˆq ‘divorce’) and Sanskrit words (karma-varma ‘fruit of action’). All the categories of speech are freely derivable with this strategy, be it verbs, nouns, adjectives, adverbs. Such a phenomena is omnipresent in all the so-called ‘‘dialects’’ or regional varieties of Hindi, although it often displays a consonant di¤erent from the v- used in Standard Hindi: In Panjabi and Panjabi-ized Hindi for instance sh- is used to derive F 0 (matlab-shatlab ‘‘signification’’, with some of such formations quasi lexicalized (gap-conversation- shap, ‘gossiping, talking’); in the Pahari (mountain) speeches, h- or ph- is used with the same function (lenıˆn-henıˆn, ruˆs-huˆs, ishk-phishk ‘love-etc.21.

40 Annie Montaut 2.2. Basic semantics of the construction: de-centring and extension of the notional domain In grammars, the only meaning mentioned for such constructions is the extension of the notional domain. The echo reduplication modifies the notional domain by including neighbouring zones and defining thus a new inclusive or associative domain, a process clearly emphasized by the terminology adopted by Parkvall (‘‘associative reduplication’’: 2003: 27). This extension amounts to introduce ‘alterity’ (‘‘and other things’’) within the notion, by associating to the Interior (pen in (17a), tea in (17b), marriage above, etc.) something located at the limit between Interior and Exterior of the notional domain, on the boundary I-E (other items than pen, tea, marriage stricto sensu, etc.: other but related items). The association is suggested by the context: within a context where a schoolboy asks a friend if he has taken pen-ven when leaving for school, the domain resulting from taking I-E into account besides I will include exercise-book, pencils, rubbers, whereas within a context where somebody searches his pockets to note down a phone number on his agenda, the notional domain is more limited (pencil, stylo, ink pen). (17) a. mere paˆs koıˆ pen-ven nahıˆN of-me near indef pen-echo neg ‘I don’t have anything to write’ Asking a visiting friend the following: (17) b. tum caˆy-vaˆy piyoge? 2 tea-echo drink-fut-2 ‘will you drink something?’ amounts to asking him if he will have something to drink, tea, co¤ee, cold drink or any other related thing, and an answer such as ‘‘No, I rather have co¤ee’’ would be at least strange, whereas it is perfectly natural for a similar question asked with the simple noun (F: caˆy) instead of echo (F-F 0 : chaˆy-vaˆy). In the above two examples, the echo construction redefines the notion F such that F is only one element of a paradigm in a wider notional domain, and the other elements, which remain implicit (hence the fuzzy character,

Reduplication and ‘echo words’ in Hindi/Urdu

41

vagueness, often referred to), may further in the exchange, be either selected instead of F or added to F in a cumulative way. As for the semantic area of F, it may be vague (17c) or precise (17a–b). (17) c. khaˆnaˆ-vaˆnaˆ vahıˆN ho saktaˆ hai eat-echo there-only be can pres-3s ‘we can have food and everything there (we’ll find everything there for meal)’ Here the co-existence of Interior and boundary (I-E) is cumulative, its signals that we are not restricted to I but associate I-E to I, with an emphasis on I in I-E (the boundary is represented as related to the Interior of the notion). Since the non centring on I (F: tea, pen, food), because of the association of I-E to I, imports a fuzzy supplement of neighbouring notions (regulated by the specific context or the cultural habits of the speakers), the meanings may exhibit quite considerable variation. Similarly, the use of current technical terms in English may encapsulate a whole process whose details are not fully or exactly known but are roughly pointed to by the notion F. The echo construction in these contexts stands for an open global idea of F, all the more open since English acts as a screen which may hide by its opacity various unknown items annexed to the notion22: (17) d. vivaˆh kar lenaˆ koıˆ aˆsaˆn kaˆm nahıˆN hai, aur aisaˆ bhıˆ nahıˆN hai ki adaˆlat meN gae to bas shaˆdıˆ ho gaıˆ. notis-votis bhıˆ to denaˆ paRtaˆ hai ‘get married is not an easy thing, don’t think it is enough you go to the court and that’s it, you are married (lit. such is not the case that you go to the tribunal then enough, marriage happened). It is also compulsory to give notice and the like’ (noTis-voTis: there are papers to sign / a whole administrative procedure to follow, including the publication of pre-marriage notice) On verbal bases too, the echo formation amounts in (17e) to extending the domain of the notion from the typical meaning of sajaˆnaˆ ‘get dressed, decorated and prepared’ to the neighbouring meanings ‘‘festive atmosphere and devotion’’, with the additional connotations of a¤ectionate (‘‘nicely, fondly’’) attitude towards the divine couple Shiv-Parvati made by the devotee:

42 Annie Montaut (17) e. [maˆn] miTTıˆ mangaˆkar usse shiv-paˆrvatıˆ banaˆtıˆ, kele ke patte se sajaˆ-vajaˆkar puˆjaˆ kartıˆ . . . ‘mother asked for earth and made an image of Shiv-Parvati out of it, she fondly decorated it with banana leaves and did her puja (did her ritual o¤ering to the deity)’ 2.3. Polemical use of the relation between F and F 0 : De-centring and disqualification of F by F 0 Very often, such an extension by associating I-E to the Interior of the notion is used for polemical and derogatory aims, in the same way as other languages may use expressions like ‘‘and everything’’, ‘‘and all this crap’’, ‘‘et cetera’’23. In (18a), the game of cards may behave as one paradigmatic element within a wider configuration including implicit other elements such as dice, carromboard or even khabbaddi, but the contextual interpretation (a mother fed up with her child’s laziness) is essentially depreciative. The echo mainly marks that A speaker disapproves of B speaker and blames him for doing or saying something (F) which is not good (F 0 ). In (18b), even if the reading of the echo-expression on ‘‘but’’ as the construction of a wider set of refusals or escapes, the most obvious meaning of the structure is the polemical intent. Similarly (18c) reduplicates ‘‘time’’ with an echo formation which refers to the inner state of the speaker and not of the addressee. Obviously the speaker is not wishing to extend the notion, already vague and wide enough to include every temporal location, duration or leisure; rather, he simply aims at manifesting to the hearer that he is crossed. Similarly in (18d), the speaker, a young man just presenting his foreign girl friend to his mother who serves the food in foreign newly brought plates, blames his mother for this unwelcome initiative: he creates an echo on the adjective ‘‘new’’, not so much to discard the new plates (which he incidentally himself brought home) but to suggest that he is upset by this way of welcoming the girl, i.e., as a foreigner. (18) a. taˆsh-vaˆsh khelne kıˆ bajaˆy apnaˆ kaˆm khatm karo card-echo play instead refl work finish do ‘instead of losing your time in playing cards (or similar stupid games) / these damned cards, you should finish your work’ b. koıˆ lekin-vekin nahıˆN indef but-echo neg ‘there is no ‘but’, stop escaping’

Reduplication and ‘echo words’ in Hindi/Urdu

43

c. mere paˆs Taim-vaim nahıˆN hai of-me near time(Engl)-echo neg est ‘I have no time to lose (stop hassling me please)’ d. kripayaˆ jab tak main huˆN nayaˆ-vayaˆ nahıˆN niklegaˆ, go-out-fut please till I am new-echo neg jismeN roz khaˆte haiN usıˆ meN khaˆeNge in-which everyday eat that in will-eat ‘please, till I am here, do not take out the new plates, we will eat in the plates we use everyday’ The statement in (18b) quotes a previously uttered ‘‘but’’ in order to disqualify it, but (18a) and (18c) simply refer to a notion which is new in the context, new but presented as triggering disagreement: the v- echo that modifies the original form F in F-F 0 betrays the altercation of two di¤erent viewpoints at odds, A trying to invalidate B’s supposed idea on the matter. On ‘‘cards’’, the echo adds a negative comment from A about the game, on ‘‘time’’, it comments not the notion itself but B’s assumption that A has time and is free. Similarly ‘‘new’’ in (18d) is more a critique of the mother’s clumsy behaviour and a manifestation of the speaker’s irritation at it than a critique of novelty or new plates. Various meanings result from this dynamics of altercation: some of them directly bear on the notion (parody, depreciation), others bear on the adequateness of the notion in the situational context, others on the relation of the addressee with the notion (his own interpretation of the notion). 2.3.1. Depreciative parody within a polemical goal: I vs I-E Proper nouns F echoed by F 0 have most of the time a derogatory e¤ect like any deformation of names in various languages. The polemical charge embedded in the echo construction is not related to a paradigmatic treatment within a wider set of neighbouring notions of which they would represent one of the possible examples. In this way, when an elderly counterrevolutionary or non concerned youngsters utter the name of Lenine with the alteration lenin-venin (lenin-henin in Pahari), the name Russia (ruˆs) distorted into ruˆs-huˆs, addressing a young fiery communist24, they only wish to communicate their hatred and dislike towards what is associated with both names. The name distorted in this way is presented as between inverted comas, as if A was quoting B, with the comas referring to A’s

44 Annie Montaut viewpoint as a disqualification of B’s viewpoint. You name these persons as good, whereas I think that they are nobodies or pests. ‘‘Don’t bother me with your Lenin /your Russians’’ could then be a possible translation. The reason why the echo disqualifies the simple term (F: Interior) is that F 0 (I-E) verges on E and is in contrast to I pour the speaker A (F 0 henin: more a rascal than a hero as posed by F): the notion Lenin (I) is de-centred towards its boundary and the boundary seen from the outside of the domain (more a rascal, E than a hero, I). Here the boundary, added to the notion by the echo, acts as a pole of ‘alterity’ (heterogeneity). The speaker A (So, ‘‘e´nonciateur’’) opposes F 0 to F uttered by the speaker B (or what A internalizes as B: S1 ‘‘co-e´nonciateur’’). The opposition of I-E (F 0 ) to I (F) often makes the implicit appreciation of B on F appear as positive (at least the appreciation that A attributes to B). De-centring here refers to A’s wish to diverge from B by introducing a markedly di¤erent viewpoint on F. Here in I-E, E is emphasized, whereas in (17) I was emphasized; the boundary I-E becomes a place for confrontation between both subjects. Common nouns, particularly learned or abstract words, are often used with echo constructions in colloquial exchange in a similar intent of parody and polemical requalification, particularly when they represent the quotation of a previous utterance. The quoting speaker (A) opposes the interpretation explicitly or implicitly proposed by B, by opposing I-E to I, thus de-centring the initial notion. This type of de-centring grants the notion a quasi metalinguistic status, such as in (19a), where B has previously justified the strange behaviour of his friend by love (ishq) and A questions this use of ‘‘love’’, and (19c), where A, an illiterate villager, decodes in the term ‘‘private tuition’’ boh a way to approach the girl and a pedantic sign (English) of the new urban class. (19) a. – kyaˆ huaˆ isko ? – ishq. – ishk-phishk to ham jaˆnte nahıˆN bhaˆˆı interr fut to-him ? – love. – love-echo top we know not brother ‘– what is happening to him ? – love. – love and what so, we don’t un derstand, brother (we know nothing about all this crap)’ b. ‘vah lekhak hai’. lekhak-vekhak hai, yah sab unkıˆ samajh meN aˆtaˆ na hıˆN thaˆ ‘he writer-echo is’. writer-echo is, this all his mind in come not impft ‘ ‘He is a writer’. He could be a so-called writer, that was meaningless for her’

Reduplication and ‘echo words’ in Hindi/Urdu

45

c. yah tumhaˆraˆ Tyuˆshan-hyuˆshan ghar ke bhıˆtar nahıˆN hogaˆ this your tuition-echo home inside not will-be ‘no way you introduce at home your (trick of the) tuition’ Whether the speaker who distorts the signifier of a notion knows or does not know the precise meaning of it is not relevant. The echo construction F-F 0 , which amounts here to set F in a quasi metalinguistic status (a word selected for comment by A), indicates that A rejects the positive viewpoint that he attributes to B, and he rejects it by re-qualifying negatively B’s notion of F25. A at the same time emphasizes the added symbolical value of the word for F (noble word, poetic word, foreign word, technique word, etc.) and he rejects it: the echo makes explicit the positive connotations of the notion (even if not intended by B), such as the romantic halo of love supposed to justify all misbehaviours, the respectable status of writers, the safeness and professionalism of private tuitions), and he reduces to nil these positive connotations. Echo reduplication of ‘‘tuition’’ here, in a context of quotation (‘‘your’’), signals less the incomprehension or rejection of the English term by a villager than the strong refusal of the very idea of tuition, very well understood by A (since it would allow the young boy to get close to the girl). What is added by the distorting echo is the aggressive charge: ‘‘you can go to hell, you and your private tuitions’’. Here again, the aggressiveness comes from the opposition, by A, of I-E to I, I being the notion as A thinks B interprets it. Significantly, the syntactic context is always negative in these polemical uses which disqualifies the notion, or the way B uses the notion ( justification for crazy acts out of love, magnification in presenting the visitor, strategy for infiltration). What justifies such a rejection is the disqualification (or negative requalification) of the term, a disqualification obtained by opposing I and I-E. That it is not cumulative (not I þ I-E) like in (17) is made clear by example (20). In (20), the notion ‘‘pandit’’ can in no way be extended by the echo to other connected notions, since it is used as an appellative – a term of address traditionally used by women in this family for the men (father and son) – and ‘‘pandit’’ is the only possible term as a traditional appellative in the context. A young servant is quarrelling with her husband, also working as a servant in the same family, about how to call the young master, who does not like to be called ‘‘pandit’’ since he rejects the traditional appellatives and culture altogether.

46 Annie Montaut (20) unheN paNDit-vaNDit mat kahaˆ karo, unheN acchaˆ nahıˆN lagtaˆ 3p-acc pandit-echo neg call freq-imper 3p-dat good neg seems ‘stop calling him Pandit (don’t tag him a pandit), he doesn’t like it’ The young woman protests against the designation pandit for the young master, who prefers to be called saˆhab, « sir ». While doing so with echo construction, she does not construct any notional extension where ‘‘pandit’’ would represent the most typical of the traditional appellative terms. She does not ironize either on the general designation of the learned Brahmins by the word ‘‘pandit’’, or on its adequateness as an appellative for other people – she keeps calling the old Master ‘‘pandit’’. But she refuses, in agreement with the young master himself, to use a term he does not like, inadequate for this only reason. Moreover, by doing so, she opposes her husband, a servant but a traditionalist too. The echo construction makes fun of the husband’s concept of appellatives. The conflict between the two interpretations, that of B (S1) the husband, and that of A (So) the wife, is about the interpretation of both speech-act participants regarding the validity of the designation ‘‘pandit’’ for the young master: you think it is a good one, I think it is absurd. With the echo formation, at the same time I make your viewpoint explicit and I invalidate it as absurd. I oppose I-E, which I construct on the notion ‘‘pandit’’, to the Interior which you stick to because you are panditaized26. The contrast between two conflicting viewpoints is sometimes explicit in the context, as for instance in (21), where a young activist, pressed by his uncle to write a thesis in order to escape the police, clearly compares the two ways of living a political involvement: action side by side with the oppressed and intellectual research (risarc). E (risarc-visarc) simply states that the speaker disqualifies the notion as inadequate to his expectations by comparing it to the other option: (21) vamnecchaˆ par haˆvıˆ hotıˆ merıˆ kraˆntıˆcchaˆ, risarc-visarc ke lie zaraˆ bhıˆ utsaˆh na thıˆ, maiN kisaˆnoN yaˆ mazduˆroN ke bıˆch jaˆkar kaˆm karnaˆ chaˆhtaˆ thaˆ ‘my fire for revolution was dominated by a fire for radical left, without the least enthusiasm for research-echo, I wanted to go and work with the peasants and workers’ The unit visarc (F 0 :I-E) is opposed to risarc (F:I) as A’s (So’s) conception (the true revolutionary must live with the workers and peasants) to the

Reduplication and ‘echo words’ in Hindi/Urdu

47

conception of B (S1) as imagined by A (for the uncle, research is the good choice for an intellectual revolutionary).

2.3.2. ‘‘Pedagogic’’ requalification of the notion: I but also I-E In contrast with the previous cases, here there is no assumption by A of the pole of ‘alterity’ corresponding to the boundary I-E. In a non-polemical context, the echo construction, which stages the discrepancy between two viewpoints on the notion, can be used as a dissuasive strategy without necessarily involving the devaluation of the basic notion F. In the case of revolution, a term used a first time with echo and a second time without echo, it is obvious that the speaker (the uncle, in the same scenario as the previous example) has nothing against revolution and defends the objective concept of it (second occurrence). But he also takes into account the disqualification of the term among the conservative folk and the local power, and it is this disqualification that he confronts his nephew with (first occurrence): you should realize that a negative connotation is associated with the term among most of the people (rather than the interpretation ‘‘revolution and other contestation discourses’’). (22) a. I understand you very well, I too have been young and communist lekin kuch din kraˆnti-vraˆnti kıˆ baˆt mat kıˆjie, but some days revolution-echo of speech neg do kraˆnti (*vraˆnti) kıˆ hıˆ khaˆtir ke lie revolution (*echo) of only interest for ‘but forget for sometimes your revolutionary big talk (do not speak of revolution-echo), in the interest of revolution itself (echo is impossible)’ With F-F 0 , the speaker quotes what corresponds to the knowledge shared by him and his hearer about ‘‘revolution’’ as an experience of his nephew: it may be linked to a halo of dreams and activities, but the main connotation is the blame and rejection in the dominant public opinion. A here simulates the viewpoint of others, those hostile to revolution. The second occurrence of the term (F) reflects the speaker’s own opinion: it rules out the echo F 0 because the term here refers to the Interior of the notion (‘act for the benefit of revolution’).

48 Annie Montaut In a similar context, in (22), A tries to convince B to adopt a less risky behaviour, now in love matters; A begins to explain the fatal consequences of sentimentality (bhaˆvnaˆ) in a relatively technical and solemn language, then he draws the conclusion of this sketch describing the unavoidable ruin awaiting the lover, and for that he uses the usual word for love ( prem) with E ( prem-vrem); the echo conveys in the form of connotation the meaning of what has been explicitly demonstrated in the previous sentence: (22) b. bhaˆvnaˆ kıˆ bhaˆvnaˆ karne se vaˆsnaˆ paidaˆ hotıˆ hai aur vaˆsnaˆ se andhaˆ huaˆ vyakti sahıˆ maˆrg dekh nahıˆN paˆtaˆ hai . . . islie bhavuktaˆ se khud duˆr raho . . . ab jaˆo . . . thoRe din ke lie prem-vrem bhuˆlaˆ kar saˆraˆ dhyaˆn paRhaˆˆı par lagaˆo ‘being in love with love (by feeling the feeling) necessarily ends up with lust and a person blinded by lust is unable to see the right path . . . this is why you should keep away from emotionality . . . now go . . . forget this silly business of love (love-echo) and devote your whole attention to your studies’ The echo construction on prem ‘‘love’’ is a simple summary and translation of the argument previously stated (without reduplication), but it is meant to have a stronger e¤ect on the hearer. It marks a shift in the discourse from the pompous sti¤ness of high rationalized language and general truths to colloquial and personalized exchange in the everyday register27. The first part of the speech, with its quasi scientific rigor and aloofness does not involve the speaker nor hearer’s subjectivity, can remain distant for the hearer, an alien discourse not specially intended for him, whereas the reduplicated term results from the notion such as constructed by S1 or the hearer B (F prem) according to So or A, and it is this construction that So denies with I-E (F 0 : vrem).

2.4. (Re)construction of the notion as a plurality of viewpoints, I et I-E being in disjunction 2.4.1. Disjunction in a negative syntactic context In all the previous contexts, negative too, the speaker aimed at denying or ridiculize the hearer’s viewpoint on F (parody), or at obtaining from the

Reduplication and ‘echo words’ in Hindi/Urdu

49

hearer that he adopts a di¤erent viewpoint (pedagogical intent). Heterogeneity (‘alterity’) resulted from the co-existence of two diverging viewpoints, one of which being strongly qualified as bad. Less often, and still in a negative context, the echo formation conveys the simple disjunction of two viewpoints on the notion, without any value judgement. In (23a), a dying agnostic tells his last wishes to his best friends, who are believers, and whom he entrusts for the execution of his wishes after his death. He uses the echo construction on the notion aˆtmaˆ (soul) in a non-derogatory meaning; the construction may be considered at best associative (soul and other metaphysical or spiritual notions), but it mainly conveys that the speaker acknowledges a double view point on the matter: I don’t believe, you do, each one his opinion and I respect yours as I plead you to respect mine. (23) a. maiN aˆtmaˆ-vaˆtmaˆ par vishvaˆs nahıˆN I soul-echo on trust neg kartaˆ, aˆp log karte haiN . . . do, you people do pres ‘I don’t believe in soul (God or such things), you people you do . . .’ In (23a) the speaker takes B’s viewpoint into account, although it di¤ers from his own, and he does not pass any judgement about the legitimacy of any viewpoint. In (23b), the notion ThaND (cold) is not requalified nor extended (to the general discomfort related by sleeping on the floor??) since a parallel is given with the warmth which overcomes cold, but it kind of quotes the fear expressed by the speaker’s auntie (aisıˆ: ‘such, of this type, as it had been told’) and the echo construction signals that, contrary to the aunt’s fear, there has been no feeling of cold. (23) b. bichone par sone lagaˆ. Mujhe aisıˆ koıˆ ThaND-vaND bhıˆ nahıˆN lagıˆ. bedding on sleep began 1s-dat such some cold-echo even neg felt Shaˆyad nıˆche biche kraˆntikaˆrıˆ saˆhitya se uThtıˆ garmaˆhaT kıˆ kripaˆ thıˆ yah ‘I fell asleep on the mat on the floor. I did not feel the slightest cold. May-be that was because of the heat raising from the revolutionary literature lying on the floor’ ‘Alterity’ (heterogeneity) here results from the contrast between what is really felt (no cold) and what was expected (that cold she was fearing). In

50 Annie Montaut both cases, although (23b) and not (23a) grants the notion a possibility of internal variation, both conflicting viewpoints of A and B are maintained, and the echo construction suggests that the notion is constructed in reference to B’s viewpoint.

2.4.2. Disjunction in a positive context Finally, in positive contexts, the echo formation has most often positive connotations which contrast with the negative (or neutral) qualification supposed to be that of speaker B. The following examples may help grasping at such meanings. They all belong to the same scenario: two Indians settled in Paris, from Madhya Pradesh, about forty years old, one has just lost his job and is depressed, the other is a good friend who tries to support him in finding the best solutions to overcome depression: (24) a. paˆrTıˆ-vaˆrTıˆ do, bhıˆR-bhaˆR milaˆe28, pakauRe-vakauRe banaˆeN, party-echo give, crowd-echo meet-caus, fried-dip-echo make ‘give a big party, meet lots of people, we will prepare vegetable fried-dips’ b. biyar-viyar ho jaˆe . . . laRkıˆ-vaRkıˆ pakRe na! bier-echo be-go-subj . . . girl-echo grasp-subj tag ‘there will be bier (it will be nice) . . . and you’ll find a girlfriend / if you could manage to grasp at some girl (s.e. that would be the good thing for you)’ Biyar-viyar (beer), paˆrTıˆ-vaˆrTıˆ (party), pakauRe-vakauRe (dip-fried vegetable), laRkıˆ-vaRkıˆ (girl), occur in a positive context, with imperative or potential predicates which express the supporting attitude of A towards B (suggestion, friend’s advice). There is no extension of the notional domain to other notions in the same paradigm (not ‘beer and/or other alcoholised beverages such as whisky, rum, etc.’, not ‘party or any such festive meeting’, not ‘pakauRaˆ or any such salted fried dish such as saˆmosaˆ’). But, contrary to the enumeration of simple (non-reduplicated) terms, which would present a neutral catalogue of solutions, the idea of beer as suggested by A to B includes euphoric and bountiful connotations, the idea of dip-fried suggests that there will be plenty, nicely flavoured, that of the girl, that she is both anonymous and attractive. Whereas the simple enumeration (X,Y,Z) can be specified (Gold bier, nicely fried pakaure, pakaure

Reduplication and ‘echo words’ in Hindi/Urdu

51

fried in ghee, a blue-eye girl), the enumeration of E (F-F 0 ¼v-) rules out such specifications, but displays to the subject imagination a whole spectrum of unspecified qualitative variations which A invites B to share with him. Echo reduplication means that viewpoints are multiple: these viewpoints are not in conflict but simply mean that each of them (you and I) may find the appropriate item, what he needs and wishes. If each of these notions, whether compact or discrete, in their echo form, is presented in (24) as the ‘‘good’’ thing to do in the given situation (fight against depression), that is, trigger the reluctant hearer’s adhesion29, by suggesting he modifies his (initially neutral) viewpoint on the notion, it is largely due to the discursive context and the verbal mood. Yet if such serendipity is possible, it is due to the introduction, in the notional domain, of positive connotations in accordance with the cultural stereotypes shared by the speakers. The feature F 0 , its ‘‘alterity’’, comes from the added inner variety which gives assurance of getting the good item, in contrast with the definite occurrence (F, in the singular) or the homogeneous plural (F in the plural). Non-centred, the notional domain becomes variegated and diverse enough to suit every wish. In a slightly aggressive context, if for instance A tries to get rid of B who asks for something to eat with too much insistence, the same constructions F-F 0 v- take a di¤erent meaning, again resulting from inner variation: (25) a. are, koıˆ saNDvic-vaNDvic khaˆ lo, mujhe tang mat karo hey, some sandwich-echo eat take, 1s-acc bother neg do ‘oh just get any sandwich whatever, and stop bothering me’ b. koıˆ laRkıˆ-vaRkıˆ DhuˆNDh lo, aur shikaˆyat karnaˆ band karo some girl-echo look-for take, and complaint do stop do ‘you manage to fish any girl and stop complaining’ What is obtained in the echo formations of (25) is a requalification of the notion so as to make all possible actualisations equivalent: any kind of sandwich will do, cheese, chicken or salad, any kind of girl will do, slim or fat, tall or small, there is choice enough for you to be pleased and stop bothering me. 2.5. On the predicative notion: Demultiplying the process When the echo reduplication a¤ects a predicative notion (verbal basis in a finite syntagma), the result is not a derogatory parody. In appearance, there

52 Annie Montaut is an extension of the notional domain, with a fragmentation of the process, without its successive occurrences being equivalent to each other. Speakers refer to this modification in the meaning by describing the process as less precise, less definite, kind of fuzzy or blurred. The verb moRnaˆ, which means ‘‘to turn’’, gets with the echo (moRnaˆ voRnaˆ), the meaning of to globally change direction, in one or several turns, without referring to one definite occurrence (like a single turn), or even to a precise orientation (you may go in a zigzag manner). In (26a) two friends in a car have taken by mistake a single way in the wrong way, the driver gets upset, the passenger tries to cool him down: (26) a. bas, aˆge kahıˆN moR-voR lenaˆ, enough, ahead somewhere turn-echo take, is saRak se nikleNge this street out will-go ‘no problem, all we need is to take any turn somewhere and we will get out of this street’ Whereas echo is ruled out if we want to tell the driver that he must turn (at the next crossing, turn left), and similarly to prevent misdirection (turn not here but at the next turn), it is appropriate to propose or describe a somewhat groping trajectory, with a clear objective (get out of here and change direction) but improvised means for doing so. We may have to turn several times or only once, the directives may be not precisely located, just try whenever you get the chance. Similarly, the echo on verb paRhnaˆ ‘‘read/ study’’ would have no meaning in reference to a localised definite process (such as ‘‘read this poem aloud to me’’), but in (26b) it means that the reader is kind of eclectic, interrupts his activity, jumps from one item to another, one readable material to another, takes all and every chance to read with no specific aim (the speaker here is a servant who just learned how to read)30. The habitual aspect (frequentative) may be responsible for the fragmentation of the predicative notion, but echo is responsible for the inner diversification of it: (26) b. roz kuch paRh-vaRh liyaˆ kartıˆ huˆN everyday something read-echo take frequentative pres-1s ‘I use to (manage to) read something or other everyday (when ever I find time, a review, book, booklet, etc.)’

Reduplication and ‘echo words’ in Hindi/Urdu

53

In (26a) and (26b) the notion gets infused with inner di¤erenciation and is no longer presented as homogeneous and centered as it is with the non reduplicated stem: echo makes it multiple, each possible occurrence di¤ering from others in quality. In the same way as plural singularity can be constructed in the nominal class, with inner di¤erentiation (‘‘the plural left’’)31, here in the verbal class echo adds qualitative plurality to the notion,32 including typical values as well as non typical ones (not really read, not frankly turn). Cumulative and lacunary functions are both present here (cf. Kabore´ 1998).

2.6. Other formations with echo or alliteration Alliterating formations (F 0 does not exist as an independent word) are made mainly by modifying the radical vowel: dekh-daˆkh ‘see/look’ on dekh ‘see/look’, puˆch-paˆch ‘inquire’, on puˆch ‘ask’; the vowel -aˆ- is practically always substituted to any other vowel (e > aˆ, i > aˆ, u > aˆ). In some cases, the vowel alteration occurs with a consonant modification (kabhıˆkabhaˆr ‘occasionally’, on kabhıˆ ‘once, ever’), and in other cases the first consonant is suppressed (aˆs-paˆs ‘around’, on paˆs ‘near’). Some adverbs are lexicalized in standard Hindi in their echo form, as aˆspaˆs or aˆr-paˆr both meaning ‘around’ ( paˆr ‘beyond’, paˆs ‘near’): they involve an extension of the notional domain (truly close þ not really close; truly across þ not really across). But most of the alliterating formations are found in colloquial speech, such as kabhıˆ-kabhaˆr (on kabhıˆ ‘once, ever’) ‘occasionally, sometimes not so often’, with more dissemination than the standard kabhıˆ-kabhıˆ ‘sometimes’. Example (24) above includes a noun (bhıˆR ‘crowd’), whose distorting alliteration (bhıˆR-bhaˆR), like the v- echo, adds a plural qualification to the notion: many various people, people of all kinds. But most of the time this type of alliteration is found on verbal basis, in non finite forms (favouring monosyllabic items?). In a process (dekh-‘see’-daˆkh, puˆch-‘ask’-paˆch, bec-‘sell’-baˆc), the di¤erence with the simple form of the verb relates to inner plurality, similarly to the standard echo (F-v-) formation. The process may be completed in several times ((27a), a statement addressed to a friend who is anxious about the location of the appointment place), or presented as a re-examination (second visit to a flat, that the hearer hesitates renting in (27b)), or presented as the final result of a long process (27c), or even as anticipated and feared ((27d), where two lovers

54 Annie Montaut meet secretly). Such a representation of the process means that its occurrences are potentially multiple but there will be a final or good occurrence (although not known in advance). (27) a. kisıˆ se puˆch-paˆch lenaˆ somebody from ask-echo take ‘you just ask to anybody (it is not a big thing, you will easily find somebody to tell you)’ b. dekh-daˆkh lenaˆ zaraˆ dhyaˆn se see-echo take little attention with ‘look (at it) thoroughly, quite well, pay attention to everything’ ( just need to go back and visit again to confirm that yes, you should take it) c. sab kuch bec-baˆc-kar vilaˆyat calaˆ gaya all sell-echo-cp foreign leave went [after father’s death he] ‘sold everything out and left for England’ d. kisıˆne hameN dekh-daˆkh liyaˆ to badnaˆm hogaˆ someone us-acc see-echo took then bad-name will-be ‘if anybody happens to see us (catch us) we will be dishonoured’ The -aˆ formations specifically favour the representation of a process as finalized, although its realization may be di‰cult or hazardous. Hence the definite character of the process, however groping the previous stages of realisation, hence the use of these forms in contexts where A wishes to stop B’s hesitations. The alliteration on bhuˆlnaˆ ‘forget’ is a good illustration: whereas forget is usually a non intentional and spontaneous process, without any display of stages leading to the result, the expression bhuˆl-bhaˆl kar, which is quasi lexicalized, means ‘put a final stop to something, a definite end to some thought, empty one’s mind of something’. The process may be deliberate (‘‘you should take this out of your mind, do manage to get over and forget everything’’) and may also result from absolute unconsciousness, but in the latter case unconsciousness is represented as hardly conceivable (‘‘he forgot his own family, his native land, as if all that no longer existed for him’’). The non-reduplicated expressions bhuˆlkar ‘forgettingly, by mistake’, and bhuˆlkar bhıˆ ‘even by mistake’, constructed on the notion as a homogeneous centred one, do not convey any particular value judgement and involve no teleonomy, but the reduplicated expression bhuˆl-bhaˆlkar, constructed on the notion as heterogeneous, emphasizes the radical character

Reduplication and ‘echo words’ in Hindi/Urdu

55

of the result, presented as allowing no coming back, a superlative forgetfulness in relation to all various forms of forgetfulness included in the notion.33 If echo formations may be in some way iconic (the distortion of the signifier produces a distortion in the signified, making it approximate or derogatory), here again it is but a gross appreciation of the phenomenon. The phenomenon once analysed, shows that we systematically deal with a process of de-centering of the notion, which plays on the Interior and Exterior of the notion to reshape the contours of the notional domain, most often by contrasting the speaker’s viewpoint with the hearer’s one (I-E is adjoined to I in a cumulated or opposed way, or in disjunction).

Conclusion Although numerous points raised in this study still need further exploration, and notional reduplication should be taken into account in the same perspective34, it is obvious that reduplication in Hindi/Urdu is a linguistic category in its own right: it encodes an operation which can be analyzed. Far from being the icon of excess (the ‘‘more’’ we say in reduplication being responsible for the meanings of intensity, distribution, iteration), convertible into ‘‘less’’ and lacunar meanings by way of fragmentation, it operates in a systematic way. Integral reduplication R (F¼F 0 ) modifies the scheme of individuation of the notion by opposing the centring of the occurrence: it modifies the relation between the reduplicated term and the constituent(s) to which it is syntactically associated (relation Noun – Verb, dependent verb – main Verb, Adjective – Noun), according to the category of the reduplicated term. Modifying the scheme of individuation may involve the subjective interaction of the speaker and hearer’s viewpoints. Partial or alliterating reduplication E (F 0 is an altered form of F) modifies the notion itself by introducing a pole of heterogeneity (‘alterity’) I-E, which means that not only the centred notion I is taken into account, but its modified form too (I-E), and this account may be of a cumulative, contrastive or disjunctive type. Interior is emphasized (weight on I), and the operation involves almost always the confrontation of two distinct viewpoints. There is nothing ‘‘stylistic’’ there and still it is true that R, in its tree main types, is far more generalized at the informal and colloquial level (apart from the grammatical uses of total R) than in the formal or objective discourse. The reason for these discursive preferences is that R, in all its non-grammatical

56 Annie Montaut uses, involves the confrontation of two viewpoints (speaker’s and hearer’s viewpoints, So and S1), a confrontation which is not necessarily aggressive and may rely on adjustment: it is then natural that R occurs with more frequency in the space of intersubjectivity and dialogue.

Notes 1. Masica 1992, Emeneau 1980, along with the phonological opposition of cerebral to dental consonants, the rigid word order SOV, postpositions, anteposition of genitive complements, anteposition of adjectives and determinative relatives, morphologically related causatives, transitive and intransitives, verbo-nominal predicates, compound verbs, marked definite or human objects, dative subjects, etc. 2. ‘‘The extracted occurrence is not just any occurrence, but it is endowed with a di¤erential property that stabilizes it as being this occurrence’’ (Culioli 1990b: 183). Extraction involves quantification (it relates the occurrence to a definite portion of space in the domain), whereas pin-pointing also involves qualification, in a preponderant way. Culioli 1999 gives more details and analyses on the operations involved in constructing the domain and its scheme of individuation, but with no section in English. 3. From the clearly onomatopeic formations (khaT-khaT ‘toc toc’, khaˆy-khaˆy, saˆy-saˆy ‘sound of the wind in the trees’, kal-kal ‘sound of running water’, taptap ‘sound of drippling water’, Dam-Dam ‘drumming’), to terms in relation with sensorial register, which rely on a correspondence between audition and other senses (cham-cham ‘twinkling’, jham-jham ‘glittering’, gich-gich ‘sticky’). Derived verbs are also more or less iconic (dhakdhakaˆnaˆ ‘palpitate’, hinhinaˆnaˆ ‘neigh’, khaTkhaTaˆnaˆ ‘knok on the door’). Such onomatopeics occur with or without derivational su‰x. 4. In the transcription of Hindi, the sign ^ on a vowel indicates length, capitals indicates retroflex consonants except for N which indicates nasalization. The abbreviations are the usual ones: dat for dative, erg, ergative, etc. 5. Culioli 1990: 183. ‘‘Scanning (French ‘parcours’) consists in running over the whole domain without being willing or able to pick out one distinguished value’’ (to scan: French ‘parcourir’). 6. Except of course when simple repetition is involved, emphasizing what the speakers says in the same way as oral stress (baRhtaˆ calaˆ gayaˆ, baRhtaˆ hıˆ calaˆ gayaˆ, ‘‘[I] kept going ahead, kept going ahead’’, in conclusion of chapter 6 in Tyaˆg Patr. Similarly the commonly used salutation jıˆte raho, jıˆte raho, ‘‘keep alive, keep alive’’, the insistant karuˆNgaˆ, karuˆNgaˆ, ‘‘I will do, I will do’’, etc. 7. Simultaneity is only apparent for strictly transient processes also, such as ‘‘reach’’: us kuˆce tak pahuNcte-pahuNcte maiN aˆj bhıˆ thoRıˆ der ke lie raˆste kıˆ

Reduplication and ‘echo words’ in Hindi/Urdu

8.

9.

10. 11. 12.

13.

57

un dukaˆnoN par rukuˆNgaˆ (that lane till reaching-reaching I today too little time for road of those shops on will-stop) ‘today also I will stop for a moment while (until) reaching that lane at the shops on my way’ (K. B. Vaid, Guzraˆ huaˆ zamaˆna). Reduplication of a strictly transitory process gives it a temporal thickness, beyond a strict stabilization, as is even more obvious with a main verb in the inceptive: sir joRne kıˆ baˆrıˆ aˆte aˆte vah mehnat vyarth lagne lagıˆ (Gitanjali Shri, Mai) (head joining of turn coming coming this e¤ort useless seem began) ‘when the time came for joining the head (to the body of the doll made by the little girl with cloth) all this labour began to appear meaningless’ (¼ by the time it came to joining the head. . . .). Quite frequent in creoles and contact languages (Moravsick 1978, Kouwenberg 2003), where it can display various sub-meanings such as diminutive, attenuation, approximation, the middle or low degree has come to be considered as prototypical as well as the high degree (Chaudenson 1974, Kyomi 1995): Chaudenson (1974: 367) gives examples in the Creole of Re´union Island such as en zafer ruz ruz ‘something reddish’, fay-fay ‘slightly tired, weak’, besides examples of high degree such as en bel-bel koson ‘a huge pig’. See also in Mauritius Creole, li met en rob ruz-ruz ‘she wears a reddish dress’, zot res dan en kaz malang-malang ‘they live in a house which is a bit dirty’, the second constituent alone being stressed (Baker 2003: 214). In this view, ambivalence itself becomes the prototypical meaning. This distinction of various plans of variation is borrowed from Denis Paillard, in his study of reduplication in Khmer (International Conference on Identity: L’Identite´, Universite´ de Tours, 29–30 novembre 2008, To be published in the Proceedings). Similarly, in French, the qualifying adjective in the left position (pre-nominal) has no descriptive properties, contrary to the post-nominal one (right position). S the subject grades the property as optimal according to his personal taste, « nicely » tall, which is not necessarily very tall. As in: is galıˆ meN sirf bare-baRe ghar haiN (this street in only big-big houses are) ‘there are only big houses in this street, every house is big’’ or is galıˆ meN nıˆle-nıˆle ghar haiN ‘every house is blue in this street’. Note that in the absence of sirf ‘‘only’’, the last sentence will be preferably interpreted as ‘‘there are bluish houses in the street’’, which means that sirf ‘‘only’’ is responsible for the distributive meaning. ‘‘Optimal’’ is of course specified by the context. In this way, a negative or derogatory context will associate the reduplicated adjective to an unpleasant emphasis on quality (‘‘optimally boring’’) with often ironical interpretation: maiN to baRe-baRe logoN kıˆ baRıˆ-baRıˆ baˆtoN se tang aˆ gayaˆ I top big-big people of big-big talks of bore come went ‘I became tired of the tall tales of big-shots’

58 Annie Montaut 14. Even discovered as a new experience a ‘‘blue sky’’ can be represented as conveying a preconstruct: itnaˆ gahraˆ nıˆlaˆ-nıˆlaˆ aˆsmaˆn maiNne pahlıˆ baˆr dekhaˆ so deep blue-blue sky I-erg first time saw ‘it was the first time I saw such a blue sky’

15.

16. 17.

18.

19.

20.

21.

The preconstructed blueness is emphasized (itnaˆ), saturated (gahraˆ) and marvellous or astonishing, that is to say appealing the the subjectivity of the speaker (with R). Like the oft quoted examples (without context) such as ‘‘greenish sari’’ (harıˆ harıˆ saˆRıˆ ), ‘‘bluish cap’’ (nilıˆ nilıˆ topıˆ ): in order to be interpreted with the meaning of low degree, these statements have to relate to direct perception. It seems that the mode of presence is crucial here, more than the character of the noun (dense, compact, discrete) since the cap like the sari rather belong to the category ‘‘discrete’’. The su‰x -vaˆlaˆ, which, in this context, indicates selection, rules out reduplication: *pıˆlaˆ-pıˆlaˆ vaˆlaˆ. Example from Abbi (1980: 107), who also gives example of the possible occurrence of the same structure with di¤erent qualities: yah aˆm khaTTaˆ hai, par vo aˆm miThaˆ-mıˆThaˆ hai (this mango sour is, but that mango sweet-sweet is), ‘this mango is sourish, but that one is sweetish’. However, the explanation given in Abbi (antonymic semantic features ‘‘exact’’ for the simple adjective vs ‘‘inexact’’ for R), is not confirmed by other devices of approximation, which can correlate with simple (‘exact’) adjectives: vah aˆm thoRaˆ-saˆ khaTTaˆ hai, par yah vaˆlaˆ ekdam khaTTaˆ hai (that mango somewhat sour is but this one really sour is) ‘that mango is somewhat sour but this one is really sour’. This su‰x, like the full form jaisaˆ, ‘like’, ‘similar to’, is a de-centring device (it may also, particularly with dimensional and quantifying adjectives, result in a high degree interpretation: bahut-se, ‘‘really many’’), but it bears only on the adjective and not, like the reduplicated structures, on the relation between adjective and noun. In what follows I, I-E or E stand for such viewpoints on the notion as they allow for di¤erent ways of taking it into account. These viewpoints do not necessarily correspond to di¤erent speakers and their subjective positions. The notions and notations of, I-E, E are borrowed from the theory of the notion and notional domain in Antoine Culioli (1999) as well as the notation of So for the subjective position of the speaker and S1 for the subjective position of the hearer, who can be an abstraction internalized by So. This type of « associative » reduplication are found in Atlantic Creoles and Parkvall (2003) relates this presence to the Indian substratum (kume-bime ‘to eat and all’). Pahari (« mountain ») varieties include mainly Garhwali and Kumaoni. ishk transcribes the native prononciation of ishq. This type of echo is even pan-

Reduplication and ‘echo words’ in Hindi/Urdu

22.

23.

24.

25. 26. 27. 28.

59

indian (Emeneau 1980), with various consonants used for the first consonant in F 0 , such as g- in Telugu ( puli-guli « flower »). Which may trigger suspicion for the ignorant as is the case in (17d): ‘‘Beware, it is more than a simple advice you will have to give to marry in the court, do not imagine it is a simple thing, it is not as easy as you fancy’’. The context is the following: a young boy is fed up with the obstacles raised by the family which do not want to let him marry the girl he loves, and he discloses his intention of a civil marriage in front of his uncle. (17d) is the uncle’s answer. (cf. infra). We may assume that the distortion on the signifyer is by nature derogatory, and so come back to the iconic virtue of reduplication, but what follows shows that such is not always the case. This example (Pahari: /h/ est la consonne initiale de F 0 ) is drawn from a long dialogue in a contemporary novel of Manohar Shyam Joshi (Kyaˆp, onomatopeic title, 2000) dont d’autres contextes sont cite´s plus loin (exemples 19, 21, 22, 23b). Or in (19b) the narrator rejects the viewpoint that A attributes to B (since the sequence is in the indirect reportive style). Upendranaˆth Ashk, Kiskıˆ baˆt (one act play from the mid twentieth century). A function often performed by code-switching (Hindi/English). Which can be contrast with the derogatory meaning of the same echo in a less optimistic, and non inter-subjective context (because of the relation with the predicate), as: maiNne yah sab choR diyaˆ, paˆrTiyoN-vaˆrTiyoN meN jaˆnaˆ band kar diyaˆ 1s-erg this all quit gave, party-echo(v-) in go stop do gave ‘I dropped all this, stopped going in parties and such foolishness’

29.

30. 31. 32. 33. 34.

In (24)a, the alliterative reduplication parallels the v-construction, in the same meaning. But there are also purely descriptive contexts, aiming only at conveying a euphoric ambiance, by calling up (with E) connotations culturally associated to the typical ideal party (heaps of flavored pakauras, lots of bier, etc.). It suggests a real party, in conformity with everybody’s expectations (‘‘as you can imagine’’). Example from the contemporary novel of Krishna Baldev Vaid, Naukaraˆnıˆ kıˆ Daˆyrıˆ (Diary of a servant, 2000). Example commented in Maria Jarrega’s PhD (2000), ‘‘la gauche plurielle’’. Favoured by the use of verbal vector lenaˆ, which not only turns the process towards the subject (auto-benefactive) but makes it perfective too. Hence the unacceptability of *bhuˆl-bhaˆlkar bhıˆ in the meaning of ‘‘by mistake’’. The fact that it involves two notions (and not one, nor one occurrence) makes it both similar to and di¤erent from E and di¤erent (no variability of emphasis on I or E). I hope to deal with the topic in the form of a note in the next issue of this annual. It is dealt with in French in Montaut in press.

60 Annie Montaut References Abbi, Anvita 1980 Semantic Grammar of Hindi, A Study in Reduplication, Delhi, Bahri Publications. 1992 Reduplication in South Asian Languages. An Areal, Typological and Historical Study. Delhi: Allied Publishers. Baker, Peter 2003 Reduplication in Mauritian Creole with notes on Reduplication in Reunion Creole, In Kouwenberg 2003: 211–18. Chaudenson, Robert 1974 Le Lexique du parler cre´ole de la Re´union. Paris: Champion. Culioli, Antoine 1990a The Concept of Notional Domain. In Pour une linguistique de l’e´nonciation. Ope´rations et repre´sentations, Vol. 1. Gap: Ophrys: 67–82. 1990b Representations, referential processes and Regulation. In Pour une linguistiquede l’e´nonciation. Ope´rations et repre´sentations, Vol. 1. Gap: Ophrys: 177–214. 1999 Pour une linguistique de l’e´nonciation. Domaine notionnel, Vol. 3. Gap: Ophrys. Emeneau, Muray B. 1969 Onomatopoeics in the Indian Linguistic Area. Language 45-2: 294– 99. Fassi-Fehri, A. and Vinet, M. T. 2001 Distribution of Number and Classifier in Arabic and Chinese and Parametrization. Linguistic Research 9-1. IERA Publications, pp. 6– 51. Hurch, Bernard (ed.) 2005 Studies in Reduplication, Berlin/New-York: Mouton de Gruyter. Jarrega, Maria 2000 Le roˆle du pluriel dans la construction des syntagmes nominaux en franc¸ais contemporain. The`se de doctorat en sciences du langage. Paris X Nanterre. Kabore´, Raphael 1998 La Re´duplication. Faits de Langue 11-12, Les langues d’Afrique subsaharienne, 359–376. Kouwenberg Sylvia and Darle`ne LaCharite´ 2001 The iconic interpretation of reduplication: Evidence from Caribbean Creole languages. European Journal of English Studies 15, 59–80. Kouwenberg Sylvia and Darle`ne LaCharite´ 2005 Less is More: Evidence from Diminutive Reduplication in Caribean Creole Languages. In Hurch (ed.), 533–45.

Reduplication and ‘echo words’ in Hindi/Urdu

61

Kouwenberg Sylvia (ed.) 2003 Twice as Meaningful: Reduplication in Pidgins, Creoles and other contact languages. Londres: Batlebridge. Kyomi, S. 1995 A New Approach to Reduplication: a Semantic Study of Noun and Verb Reduplication in Malayo-Polynesian Languages. Linguistics 33: 1145–1167. Montaut, Annie in press Re´duplication et constructions en e´cho en hindi/ourdou. Faits de Langue, Les Cahiers 1 in press. Montaut, Annie 2007 Formes et valeurs de la reduplication totale en hindi/ourdou. Faits de Langue 29, La Re´duplication, pp. 175–190. Montaut, Annie 2004 Hindi Grammar. Mu¨nchen: Lincom-Europa. Moravcsik, Edith 1978 Reduplication. In Universals of human language. Volume 3: Word structure, Greenberg Joseph H. (ed.), 297–33. Stanford: Stanford University Press. Parkvall, Michael 2003 Reduplication in Atlantic Creoles. In Kouwenberg (ed.), 19–36. Singh, Rajendra 2005 Reduplication in Modern Hindi and the Theory of Reduplication, In Hurch (ed.), 263–81.

No Smoke without Fire: Invisible Agent Constructions in South Asian Languages*

Prashant Pardeshi

One of the contributions of the Japanese linguistic tradition to the field of linguistics is the notion of the contrast between SURU (DO) type expressions versus NARU (BECOME) type expressions. SURU (DO)-expressions (such as those involving verbs like kimeru ‘‘to decide’’, mitukeru ‘‘to find out’’, etc.) can overtly encode an agent while NARU (BECOME)-expressions (such as those involving verbs like kimaru ‘‘to be decided’’, mitukaru ‘‘to be found’’, etc.) cannot despite the fact that the event in question could not have been realized without the involvement of an agent. In this paper, I will focus on the NARU (BECOME)-expressions or invisible agent constructions in four South Asian languages (Marathi, Hindi, Telugu, and Tamil) and demonstrate that the SURU (DO) versus NARU (BECOME) contrast is also a characteristic feature of South Asian languages. Further, I will argue that NARU (BECOME)-expressions constitute a kind of quasi-passive construction with the highest degree of agent-defocusing (cf. Shibatani 1985) and that while NARU (BECOME)-expressions show semantic a‰nity with the canonical passive constructions both di¤er from each other morphosyntactically. I will also argue for maintaining a distinction between the NARU (BECOME)-expressions which imply external agency (e.g. hatyaa hoNe [murder become] ‘be murdered’) and the so-called spontaneous expressions which are typically void of agency (e.g. nidhan hoNe [death become] ‘die’).

1. Introduction One of the contributions of the Japanese linguistic tradition to the field of linguistics is the notion of the contrast between SURU (DO) type expressions versus NARU (BECOME) type expressions [Alfonso (1966), Kunihiro (1974), Teramura (1976/1993), Monane & Rogers (1977), Ikegami (1981, 1991), Hinds (1986), inter alia]. SURU (DO)-expressions (such as

64 Prashant Pardeshi those involving verbs like kimeru ‘‘to decide’’, mitukeru ‘‘to find out’’, tateru ‘‘to build’’, tukamaeru ‘‘to catch’’, tutaeru ‘‘to convey’’, todokeru ‘‘to deliver’’, tasukeru ‘‘to help/to save’’, tukuru ‘‘to prepare’’, etc.) can overtly encode an agent while NARU (BECOME)-expressions (such as those involving verbs like kimaru ‘‘to be decided’’, mitukaru ‘‘to be found’’, tatu ‘‘to be built’’, tukamaru ‘‘to be caught’’, tutawaru ‘‘to be conveyed’’, todoku ‘‘to be delivered’’, tasukaru ‘‘to be helped/saved’’, dekiru ‘‘to be ready’’ etc.) cannot despite the fact that the event in question could not have been realized without the involvement of an agent. In this sense, NARU (BECOME)expressions are ‘‘invisible agent’’ construction. SURU (DO)-expressions are said to focus on the ‘‘action’’ of the agent while NARU (BECOME)expressions focus on the ‘‘result’’ of the agent’s action. While both types express semantically transitive situations featuring two participants, the former are syntactically transitive (two-place predicates) while the latter are syntactically intransitive (one-place predicates). In the Indian grammatical tradition NARU (BECOME)-expressions [ie, invisible agent constructions] have received a cursory treatment (see Damle 1911 and Pandharipande 1997 for Marathi; Bhatia 1987 and Kachru 2006 for Hindi; Sridhar 1990 for Kannada). In this paper, I will focus on the NARU (BECOME)-expressions or invisible agent constructions in four South Asian languages, namely Marathi, Hindi, Telugu and Tamil, and demonstrate that the SURU (DO) versus NARU (BECOME) contrast is also a characteristic feature of these South Asian languages. Further, I will argue that NARU (BECOME)-expressions constitute a kind of quasipassive construction with the highest degree of agent-defocusing (cf. Shibatani 1985) and that while NARU (BECOME)-expressions show semantic a‰nity with the canonical passive constructions both di¤er from each other morphosyntactically. I will also argue for maintaining a distinction between the NARU (BECOME)-expressions which imply external agency (e.g. hatyaa hoNe [murder become] ‘be murdered’) and the so-called spontaneous expressions which are typically void of agency (e.g. nidhan hoNe [death become] ‘die’). The organization of the paper is as follows. In section 2, I will o¤er a brief summary of past research on the SURU (DO) versus NARU (BECOME) dichotomy. Against this backdrop, I will describe the NARU (BECOME)-constructions in South Asian languages (Section 3). In section 4, I will argue that NARU (BECOME)-constructions are in fact a kind of passive construction and contrast them with canonical passive

Invisible Agent Constructions in South Asian Languages

65

constructions. Further, I will draw a contrast between NARU (BECOME)constructions and the so-called spontaneous constructions and argue for distinguishing them. Finally, in section 5, I will summarize my findings and present conclusions.

2. Previous studies on the suru (do) vs. naru (become) contrast The SURU (DO)-expressions versus NARU (BECOME)-expressions dichotomy has been initially proposed in the context of contrastive studies of Japanese and English. Alfonso (1966/1974/1980: 884–86) makes the following observation on the contrast between the use of a transitive versus intransitive verb in Japanese and English. In SPONTANEOUS expressions, an INTRANSITIVE verb is used when the speaker’s thought is focused on THE OBJECTIVE FACT, a TRANSITIVE verb is used when the speaker’s thought is focused on the AGENT OR THE AGENT’S ACTIVITY. This is one case in which English speakers and Japanese speakers di¤er in their habitual approach to situations. In English, for example, it does not make much di¤erence whether a person chooses to say: ‘‘They DECIDED that . . .’’ ‘‘We SAW him there.’’ ‘‘They ELECTED him . . .’’ ‘‘He BROKE the window.’’

or or or or

‘‘IT WAS DECIDED that . . .’’ ‘‘HE WAS SEEN there.’’ ‘‘HE WAS ELECTED . . .’’ ‘‘The window WAS BROKEN by him.’’

We do not mean to say that the English speaker does not see any di¤erence in meaning between the expressions in the left-hand column above and their respective passive expressions in the right-hand column. Surely, there is a difference in meaning. But the English speaker does not have any constitutional preference for one form over the other, no ingrained inclination towards the use of one form rather than the other. The Japanese speaker, however, DOES have a preference, and DOES tend to [use] one form rather than the other.

More recently, Jacobsen (1992: 106–108) in discussing the di¤erence between Japanese and English cites the following examples. The ‘‘preferred’’ English versions are given in (a) while the ‘‘preferred’’ Japanese versions are given in (b):1

66 Prashant Pardeshi (1)

a. Have you found an apartment yet? b. apaato wa moo mitukarimasita apartment TOP yet be-found-POL-PAST ‘‘Has an apartment been found yet?’’

(2)

ka? Q

a. Did someone by any chance turn in a lady’s watch? b. hyotte shite huzinyoo no tokei ga todoite-iru by any chance lady’s GEN watch NOM arrive-PERF desyoo ka? perhaps Q ‘‘Has a lady’s watch by any chance been turned in (to you)?’’

(3)

a. (Watching someone fishing) Oh! He caught one! b. a! tureta (Alfonso, 1974) EXC be-caught-PAST ‘‘Oh! (One) was caught!’’

Jacobsen (1992: 106) remarks on this di¤erence as follows: Japanese exhibits a ‘‘constitutional’’ preference for result-oriented expression, and it is common to encounter situations where a transitive expression would be preferred in English, but the corresponding expression in Japanese involves an intransitive construction which, at least in its English translation, is passive in meaning. . . . Although not passive in their morphology, the Japanese versions bear a semantic a‰nity to passives because the presence of an agent giving rise to the result is in each case clear from contextual and common sense factors. There are, however, equally many non-agentive intransitive constructions where a passive treatment is clearly inappropriate.

Jacobsen discusses the non-agentive intransitive construction which lacks a passive interpretation (or implication of agent) citing the following examples. (4)

kinoo gakkoo de tyotto omosiroi ziken ga yesterday school LOC quite interesting incident NOM okotta be-PAST ‘‘Quite an interesting incident occurred at school yesterday.’’

Invisible Agent Constructions in South Asian Languages

67

(5)

sora ga harete-iru to huzisan ga mieru sky NOM clear-up-PERF COND Mt. Fuji NOM be visible ‘‘When the sky is clear, Mt. Fuji is visible.’’

(6)

syoowa tennoo no si to tomo ni Showa emperor GEN death COM with DAT zidai ga hazimatta era NOM begin-PAST ‘‘A new era began with the death of Emperor Showa.’’

atarasii new

The underlined noun phrases in (4)–(6) are subjects of the intransitive verbs in question and semantically they function as objects of corresponding transitive verbs. Jacobsen refers to these noun phrases as ‘‘semantic objects’’. In examples (1)–(6) a semantic object is placed in the role of syntactic subject to convey the meaning that an event has come about apart from the intentional involvement on the part of the subject. According to Jacobsen, examples such as these are neutral as to the question of whether an entity independent from the subject is responsible for bringing about the event in question, although such an interpretation (passive interpretation) may be imposed by certain contextual or common sense factors. Pointing out that there is no suitable term in Western grammar to refer to the general class of such constructions Jacobsen adopts the term zihatu ‘‘spontaneous’’ from traditional Japanese grammar. By doing this, Jacobsen essentially clubs agent-implying intransitive construction in (1)–(3) together with the nonagent implying intransitive constructions given in (4)–(6) on the basis of their shared similarity of placing a semantic object in the role of syntactic subject. In this paper I will focus on agent-implying intransitive constructions [invisible agent constructions like those in (1)–(3)] as well as non-agent implying intransitive constructions [agent-less constructions such as those in (4)–(6)] in four South Asian languages, viz. Marathi and Hindi (IndoAryan) and Telugu and Tamil (Dravidian). Drawing a contrast between agent-implying passive constructions and agent-implying intransitive constructions, I will argue that the latter should be treated as quasi-passive terming them as ‘‘BECOME passive’’ constructions. Further, contra Jacobsen, I will argue that a distinction should be made between the agentimplying intransitive constructions (BECOME passives in my proposal)

68 Prashant Pardeshi and non-agent implying intransitive constructions (for which I will reserve the term ‘‘spontaneous construction’’).

3. Invisible agent intransitive constructions in South Asian languages In South Asian languages invisible agent intransitive constructions (aka agent-implying intransitive constructions) are widely used. Such constructions involve morphologically simplex intransitive verbs as well as morphologically complex intransitive verbs comprising {Action Noun/Predicative Adjective implying agency þ light verb}. In most cases the light verb is ‘‘BECOME’’ but in some cases other light verbs (such as ‘‘stay’’, ‘‘come’’, etc.) are used. The verb BECOME is used as a cover term for the verbs occupying the light verb slot. Morphologically simplex agent-implying intransitive verbs are few in number while morphologically complex agentimplying intransitive verbs form a substantial class. Typical examples in Marathi of both types are illustrated below. (7)

Morphologically simplex agent-implying intransitive verbs in Marathi TharNe (be decide), pohotsaNe (be delivered/reach), saapaDNe (be found), etc.

(8)

Morphologically complex agent-implying intransitive verbs in Marathi {aTak/shikshaa/karvaai/hakaalpaTTi/padonnati}þhoNe {arrest/punishment/action/expulsion/promotion}þBECOME {stuti/nindaa/apmaan/Tikaa/satkaar}þhoNe {praise/criticism/insult/criticism/felicitation}þBECOME {niwaD/nemNuk/badli/padaanvati/pramoshan}þhoNe {selection/appointment/transfer/demotion/promotion}þBECOME {radda/raakhiv/nirNay/Tharaav/badal/suTkaa}þhoNe {cancelled/reserved/decision/change/release}þBECOME ubh{aa/i/e} hoNe/rahaaNe (lit. upright become/stay, ‘‘be built, have come up’’)

A few representative full sentence examples of the verbs in (7) and (8) are given below.

Invisible Agent Constructions in South Asian Languages

(9)

69

shukravaari sandhyaakaaLi 5 waadztaa nighaaytse friday evening 5 O’clock leave.of Tharale got decided ‘‘It was decided to leave at 5 O’clock on Friday.’’ (my3m.blogspot.com/2005_08_01_archive.html)

(10) bomb-sphoT khaTalyaat sanjay datta-laa sahaa warshaa-chi bomb-blast case.in Sanjay Dutt-to six years-of.F sakta-madzuri-ci shikshaa dzhaali hard-labour-of.F punishment.F become.F ‘‘In the bombing case Sanjay Dutt was given a punishment of 6 years of hard labour.’’ (mr.upakram.org/node/621)

(11) varaLi si phes yethe saanDapaaNi samudraat soDaNaaryaa WaraliSea Face here waste water sea.in leaving gaTaaraacyaa mukhaa-war-tsa don Tolejanga imarati drainage.GEN.Obl face-on-EMPH two sky scraper buildings ubhyaa raahikyaa aahet stand stayed be.PRES.PL ‘‘At Worli Sea Face, two skyscraper buildings have come up right in front of a drainage which pours waste water into the sea.’’ (www.esakal.com/esakal/06042007/Mumbai07A8674C47.htm)

Similar expressions are widely attested in other South Asian languages as well. Due to space limitations I will confine myself to two Indo-Aryan languages (Marathi and Hindi) and two Dravidian languages (Telugu and Tamil) and show parallels across these languages in their use of a particular morphologically complex agent-implying intransitive expression {arrestþ BECOME}. This expression has a transitive counterpart {arrestþDO} as well as a passive counterpart {arrestþDOþPASSIVE MARKER}. I will present examples of all three of these related expressions from Hindi, Marathi, Telugu and Tamil. In each set the first example is a transitive, the second one is a passive, and the third one is an agent-implying intransitive (BECOME-type) expression.

70 Prashant Pardeshi Hindi (12) amit anil-ko budhwaar-ko bhaanDup-ke mangatdaas Amit Anil-to Wednesday-on Bhandup-of Mangatdas peTrol pamp-ke paas-se siiniar inspecTar nitiin alaknure petrol pump-of near-from senior inspector Nitin Alaknure satiish mayakar aur sudhiir dalawii kii Tiim-ne Satish Mayakar and Sudhir Dalwi of team-ERG giraf Taar kiyaa thaa arrest.M.SG. do.M.SG was.M.Sg ‘‘A team of senior inspectors comprised of Nitin Alaknure, Satish Mayakar, and Sudhir Dalwi had arrested Amit and Anil on Wednesday near the Mangatdas petrol pump in Bhandup.’’ (http://navbharattimes.indiatimes.com/articleshow/2454534.cms) (13) sudhiir dalawii ke anusaar amit bhogale ko 2003-mE ek Sudhir Dalwi according to Amit Bhogle to 2003-in one marDar kes-mE giraftaar kiyaa gayaa murder case-in arrest.M.SG. do.M.SG go.M.SG.PST thaa was.M.Sg ‘‘According to Sudhir Dalwi, Amit Bhogle was arrested in a murder case in 2003.’’ (http://navbharattimes.indiatimes.com/articleshow/2454534.cms) (14) sanyog-se 2007 kii shuruwaata-mE amit maaraamaari ke accident-by 2007 of beginning-in Amit fighting of ek maamale-mE phir giraftaar huaa One incident-in once again arrest.M.SG became.M.SG ‘‘Accidently, in the beginning of 2007, in one of the fighting incidents, once again Amit got arrested.’’ (http://navbharattimes.indiatimes.com/articleshow/2454534.cms) Marathi (15) polis-aan-ni sahaa-dzaNaan-cyaa ToLi-laa police-PL ERG six-people-Gen.M gang.M-ACC keli did.F.Sg ‘‘The police arrested a gang of six people.’’

aTak arrest.F

Invisible Agent Constructions in South Asian Languages

(16) polis-aan-kaDun sahaa-dzaNaan-cyaa ToLi-laa police-PL-by six-people-Gen.M gang.M-ACC keli geli did.F.Sg went.F.Sg. ‘‘A gang of six people was arrested by the police.’’ (17) sahaa-dzaNaan-cyaa ToLi-laa six-people-Gen.M gang.M-ACC ‘‘A gang of six people was arrested.’’

aTak arrest.F

Telugu (18) pooliisulu anil-ni arresT police.PL Anil-ACC arrest ‘‘Police arrested Anil.’’

cesaaru did.PL

(19) anil pooliisula-ce arresT ceyabaDDaaDu Anil police.PL-by arrest did.PASS.PST.SG ‘‘Anil was arrested by police.’’ (20) anil arresT ayyaDu Anil arrest became.SG ‘‘Anil was arrested.’’ Tamil (21) polisaar anil-ai kaithu seithnaar police Anil-to arrest did ‘‘Police arrested Anil.’’ (22) anil polisaar-aal kaithu seiyappaTTaar Anil police-by arrest did.PASS ‘‘Anil was arrested by police.’’ (23) anil kaithu aanaar Anil.M arrest became ‘‘Anil was arrested.’’

71

aTak arrest.F

dzhaali becsme.F.SG.

72 Prashant Pardeshi Having looked at the scope of the invisible agent intransitive constructions (aka agent-implying BECOME-type intransitive expressions) in Hindi, Marathi, Telugu and Tamil, I will compare them with canonical passive constructions and drawing on data from Marathi I will point out the similarities as well as di¤erences between the two types. Based on the similarities they share with the passive construction, I will argue that invisible agent intransitive constructions should be treated as a quasi-passive and should be christened the ‘‘BECOME passive’’. Further, I will contrast these invisible agent intransitive constructions (or in my terms ‘‘BECOME passive’’ constructions) with the non-agent implying intransitive constructions (the so-called canonical spontaneous constructions) and pace Jacobsen (op. cit.) I will argue for making a distinction between the two and reserving the term ‘‘spontaneous’’ construction for only the non-agent implying intransitive constructions.

4. Invisible agent intransitive construction vs. canonical passive construction In many language of the world the passive construction is formed by using an auxiliary verb and many of them use more than one auxiliary verb as a marker of the passive [for example, German, Dutch, Swedish, Finnish, Latvian, Polish, among others, use BE and BECOME, English recruits BE and GET, while Bengali uses GO and BECOME as passive marker auxiliaries (cf. Siewierska 1984: 126 for more details)]. Marathi also is a case in point and uses both GO and COME as passive marker auxiliaries. These two passives di¤er in meaning but these di¤erences are not the issue at hand, hence I will not go into the details here (for discussion see Pardeshi 2000a, 2000b, and forthcoming). I will use GO passive as a representative of Marathi passive and contrast it with the BECOME-type invisible agent intransitive construction. Consider the following pairs of Marathi examples where the (a) versions are BECOME-type invisible agent intransitive expressions and the (b) versions are GO passives. (24) a. pantapradhaanaan-ci hatyaa dzhaali prime minister-GEN.F assassination.F become.PF.F Lit. ‘‘Prime Minister’s murder became.’’ ‘‘The Prime Minister was murdered.’’

Invisible Agent Constructions in South Asian Languages

73

b. (atirekyaa-kaDun) pantapradhaanaan-ci hatyaa extremist-by prime minister-GEN.F assassination.F keli geli do.PstPart.F go.PF.F ‘‘The Prime Minister was murdered by the extremist.’’ (25) a. raam-ci pakshaa-tun hakaalpaTTi dzhaali Ram-GEN.F party-from expulsion.F become-PF-F Lit. ‘‘Ram’s expulsion from the party became.’’ ‘‘Ram was expelled from the party.’’ b. (pakShashresThin-kaDun) raam-ci pakshaa-tun Party high command-by Ram-GEN.F party-from hakaalpaTTi keli geli expulsion.F do.PstPart.F go.PF.F ‘‘Ram was expelled from the party by the party high command.’’ (26) a. raam-ci badli dzhaali Ram-GEN.F transfer.F become.PF.F Lit. Ram’s transfer became. ‘‘Ram was transferred.’’ b. (variShTha adhikaaryaa-kaDun) raam-ci badli senior o‰cer-by Ram-GEN.F transfer.F keli geli do.PstPrt.F go.PF.F ‘‘Ram was transferred by the senior o‰cer.’’ (27) a. ek mahin-yaat caukashi-tsaa ahawaal tayaar one months-in investigation-of report.M ready dzhaalaa become.PF.M Lit. ‘‘The investigative report became ready in one month.’’ b. ek mahin-yaat caukashi-tsaa ahawaal tayaar one months-in investigation-of report.M ready kelaa gelaa do.PstPart.M go.PF.M ‘‘The investigative report was made ready in one month.’’

74 Prashant Pardeshi The pairs of sentences in (24) through (27) are semantically alike in that both contain a ‘‘non-agent’’ subject. They di¤er from each other, however, in terms of overt encoding of the agent: BECOME-type invisible agent intransitive construction do not permit the presence of an agent while GO passive constructions are felicitous in the presence of the agent (albeit oblique) as shown in example (28) below. (28) a. (*shaasanaa-kaDun) simenT udyog-aat yaa purwi government-by cement industry-in this before jaahiir dzhaalelyaa kshamataa weLe-war announce become.Pst.Prt.F.PL capacities.F.PL time-on ubhyaa dzhaalyaa upright become.PF.F.PL ‘‘The (production) capacities announced before have come up (*by the government).’’ (http://www.esakal.com/esakal/11122007/ Arthvishwa2518B2ED03.htm) b. (shaasanaa-kaDun) simenT udyog-aat yaa purwi government-by cement industry-in this before jaahiir dzhaalelyaa kshamataa weLe-war announce become.Pst.Prt.F.PL capacities.F.PL time-on ubhyaa kelyaa gelyaa upright do.PstPart.F.PL go.PF.M.PL ‘‘The (production) capacities announced before have been built by the government.’’ The transitive active counterpart corresponding to (28b) is given in (29) below wherein the agent appears as an obligatory argument (subject) which cannot be omitted. (29) shaasanaa-ne simenT udyog-aat yaa purwi jaahiir government-ERG cement industry-in this before announce dzhaalelyaa kshamataa weLe-war ubhyaa become.Pst.Prt.F.PL capacities.F.PL time-on upright kelyaa do.PF.F.PL ‘‘The government built the (production) capacities announced before.’’

Invisible Agent Constructions in South Asian Languages

75

On comparing the BECOME-type invisible agent intransitive construction in (28a) with the active transitive construction in (29) it is clear that: (i) functionally, (28a) defocuses the agent noun phrase; (ii) semantically, the valency of the situation expressed by the verb in (28a) is two which is the same as for its active counterpart in (29); (iii) syntactically, the valency of the verb in (28a) as compared to its active counterpart in (29) is less by one; and (iv) morphologically, the predicate in (28a) is distinct from its corresponding active counterpart in (29). From these facts it is clear that the BECOME-type invisible agent intransitive construction in (28a) satisfies all the criteria given below for defining the passive prototype as proposed by Shibatani: (30) The Passive Prototype: Shibatani (1985: 837) a. Primary function: Defocusing of agent. b. Semantic properties: (i) Semantic Valence: Predicate (agent, patient) (ii) Subject is a¤ected. c. Syntactic properties: (i) Syntactic encoding: agent ¼ Ø (not encoded), patient ¼ subject (ii) Valence of P[redicate]: Active ¼ P/n, Passive ¼ P/n
1 d. Morphological property: Active ¼ P, Passive ¼ P [þpassive] In view of this correspondence I regard the BECOME-type invisible agent intransitive construction as a bona fide quasi-passive construction and call it the BECOME passive. The BECOME-type invisible agent intransitive construction or the BECOME passive in my terms has been analyzed as an ‘‘impersonal passive’’ by Sridhar (1990: 215) and by Pandharipande. In a very brief discussion Pandharipande (1997: 399) observes: This construction most commonly lacks the surface subject/agent and the verb is invariably hoNe ‘to happen’; which agrees with the (semantic) direct object.

Sridhar contrasts the ‘‘impersonal passive’’ (BECOME passive in my terms) with the active construction citing the following examples. Example (31) is an active clause while (32) is an ‘‘impersonal passive’’.

76 Prashant Pardeshi (31) ya:ro: i: nirNayavannu khaNDisidaru someone this resolution-acc denounce-pst-3pl-hum Someone denounced this resolution. (32) i: nirNayavannu khaNDisala:yitu this resolution-acc reprove-inf-become-3sn This resolution was reproved. As for the ‘‘impersonal passive’’ in Kannada (example (32)) Sridhar (1990: 214) remarks as follows: The impersonal passive can never have an overt subject constituent. It has a direct object, marked by the accusative. (Note that the accusative ending, though optional elsewhere with non-human objects, occurs fairly regularly in this construction, undoubtedly to ensure that the first noun phrase is not interpreted as the subject.) The verb appears in the infinitive form (with the – al ending), followed by the auxiliary (or the stative copula) –ag:u ‘become’. The auxiliary is inflected for the tense and aspect, but carries the invariable agreement feature of third singular neuter, apparently the ‘‘unmarked’’ agreement form. This construction though still ‘‘literary’’ or ‘‘formal’’ and almost never used in (informal) spoken styles, is used much more frequently than the personal passive.

It should be added that there is di¤erence of opinion between Sridhar and me in the interpretation of the term ‘‘subject’’ in the case of impersonal passive. Sridhar argues that the first noun phrase in the impersonal passive construction (underlying object of the corresponding active counterpart) is not a subject but an object since it retains (sometimes) its accusative marking and does not agree with the verb. Additional examples provided by Sridhar (op. cit.: 215–216) show that the accusative case retention in the case of impersonal passive is optional and that the verb agreement in the case of personal as well as impersonal passive is identical. (33) mane kaTTalpaDuttide house build-inf-pass-prog-n.pst-3sn The house is being built. (34) mane (yannu) kaTTala:guttide house (acc) build-inf-become-prog-n.pst-3sn The house is being built.

Invisible Agent Constructions in South Asian Languages

77

(35) raste agala ma:DalpaTTide road wide make-inf-pass-pf-n.pst-3sn The road has been widened. (36) raste agala ma:Dala:gide road wide make-inf-become-pf-n.pst-3sn The road has been widened. I treat the underlying ‘‘objects’’ of the corresponding active counterparts appearing in the impersonal passive construction (BECOME passive in my analysis) as ‘‘subject’’ in that it is the sole argument of which the predication is made. Unlike in Kannada, these noun phrases control verb agreement in Marathi as shown in (24a)–(27a). Intuitively speaking, in Marathi also, the usage frequency of BECOME passives (impersonal passive) seems to be higher than the GO passive (personal passive). To draw a more definitive conclusion an empirical investigation is needed. There are expressions which at a first glance appear to be the counterexamples to my claim that BECOME passive do not permit the overt presence of an agent phrase. Note the following examples.2 (37) tyaacyaa-kaDun apekshit kaamgiri dzhaaleli he-by expected performance.F become.PstPrt.F naahi NEG Lit. Expected performance has not been achieved by him. He has not performed up to the expected level. (origin.maharashtratimes.indiatimes.com/articleshow/2462179.cms) (38) erwii ase sharthi-tse prayatna tyaacyaa-kaDun other times like die hard-of e¤orts.M him-by dzhaale naahit become.PF.M NEG Other times such die hard e¤orts have not been done by him. (www.esakal.com/esakal/05312007/SportsB2C9E2EA69.htm) A closer look reveals however that the expressions in (37) and (38) di¤er from the BECOME passive in that they obligatorily contain an agent. Recall that, BECOME passives (invisible agent intransitive construction)

78 Prashant Pardeshi do not permit presence of an agent (cf. (28a)). I treat the constructions in (37) and (38) as ‘‘potential construction’’ wherein the agent’s (in)ability is depicted and as such the presence of the agent is obligatory. In this respect, the potential construction di¤ers from the passive, wherein the presence of an agent is optional (see Pardeshi 2000b for more details). To sum up, semantically, BECOME passive [NARU (BECOME)expressions/ invisible agent intransitive constructions] as well as the canonical passive expressions perform agent-defocusing: BECOME passive defocus (or erase) the agent altogether while canonical passive expressions defocus the agent partially. Syntactically, BECOME passive di¤er from canonical passive expressions in that BECOME passive do not permit the overt presence of an agent while canonical passives permit (at least optionally) the overt expression of an agent. Furthermore, BECOME passive are morphologically simple (unmarked) while passive expressions are morphologically complex (marked). Having demonstrated the similarities and di¤erences between the BECOME passive and the canonical GO passive, I will now move on to the issue of how to di¤erentiate the agent-implying BECOME passive (BECOME-type invisible agent intransitive construction) from the agentless ‘‘spontaneous’’ construction. Recall that Jacobsen (1992) does not make a distinction between these two and refers to both of them collectively as the ‘‘spontaneous’’ construction. In South Asian languages the BECOME passives are isomorphic with the {Spontaneous Process Noun / Predicative Adjective void of agency þ BECOME} construction which conveys the spontaneous emergence of a state of a¤airs. Due to space constraints I will discuss only one example here in order to demonstrate the di¤erence between the agent implying BECOME passive and the agent-less ‘‘spontaneous’’ construction. (39) pantapradhaanaan-ci hatyaa dzhaali prime minister-GEN.F murder.F become.PF.F Lit. Prime Minister’s murder became. The Prime Minister was murdered. (40) pantapradhaanaan-tsa nidhan dzhaala prime minister-GEN.N death.N become.PF.N Lit. Prime Minister’s death became. The Prime Minister died.

Invisible Agent Constructions in South Asian Languages

79

Both (39) and (40) describe the death of the Prime Minister and they show a formal resemblance. However, semantically they di¤er crucially in that while the former typically involves an external agent, the latter is devoid of agency. Owing to this, the BECOME-type invisible agent intransitive construction or {Action Noun / Predicative Adjective implying agency þ BECOME} construction in (39) possess an active transitive counterpart, namely, {Action Noun / Predicative Adjective implying agency þ DO} construction while the {Spontaneous Process Noun/Predicative Adjective void of agency þ BECOME} construction in (40) lacks it. The following examples attest to this fact. (41) (atirekyaa-ne) pantapradhaanaan-ci hatyaa keli extremist-ERG prime minister-GEN.F assassination.F do.PF.F The extremist assassinated the Prime Minister. (42) (*atirekyaa-ne) pantapradhaanaan-tsa extremist-ERG prime minister-GEN.N The extremist killed the Prime Minister.

nidhan death.N

kela do.PF N

The BECOME-type invisible agent intransitive construction or {Action Noun / Predicative Adjective implying agency þ BECOME} construction is a ‘‘passive’’ construction (BECOME passive) while the {Spontaneous Process Noun / Predicative Adjective devoid of agency þ BECOME} is an agent-less ‘‘spontaneous’’ construction. To reiterate, the former possesses a transitive active counterpart while the latter lacks it. Although they formally look alike, they should be distinguished in the face of the foregoing di¤erences.

5. Summary & conclusions In this paper I demonstrate that the SURU (DO) versus NARU (BECOME) contrast is one of the characteristic features of South Asian languages. Further, drawing on data from Hindi, Marathi, Kannada, Telugu, and Tamil, I argue that the invisible agent construction or NARU (BECOME)-expressions constitute a kind of quasi-passive construction with a high degree of agent-defocusing and that semantically as well as syntactically they contrast with canonical passive constructions. I christen

80 Prashant Pardeshi the NARU (BECOME)-expressions as the BECOME passive and demonstrate that while they show a close semantic a‰nity to canonical passive expressions in terms of agent-defocusing or concealing of agent the two di¤er from each other syntactically in that the BECOME passive does not permit the overt presence of an agent while canonical passives do (albeit as an oblique argument). Further, morphologically, NARU (BECOME)expressions are unmarked while passive expressions are marked. Finally, pace Jacobsen (1992), I argue for making a distinction between the BECOME passive or {Action Noun/Predicative Adjective implying agency þ BECOME} construction and the {Spontaneous Process Noun/Predicative Adjective devoid of agency þ BECOME} construction or ‘‘spontaneous’’ construction. In conclusion, the robust presence of NARU (BECOME)type constructions in the South Asian languages under discussion suggests that South Asian languages lean more towards the NARU (BECOME) end of the SURU (DO) versus NARU (BECOME)-expression preference continuum.

Notes * I would like to thank Prof. Peter Hook, Dr. Takayuki Touno and two anonymous reviewers of ARSALL for their invaluable comments and criticisms on both the content and the form of the paper. Thanks are due to Siriappareddy Tamalampudi and Verrappan Veerasureshkumar for providing data on Telugu and Tamil respectively and to Prof. K. V. Subbarao for cross-checking the Telugu data. The usual disclaimer applies. The transliteration systems adopted for Japanese and South Asian languages are the ones widely used in linguistic literature on these languages. 1. The abbreviation used in the glosses are as follows: ACC – accusative; LOC – locative; PAST/Pst/pst – past; COM – commutative; M/m – masculine; PRES – present; DAT – dative; N/n – neuter; prog – progressive; ERG – ergative; NE – Negative; PstPerfpast – perfect; EMPH – emphaticparticle; Obl – oblique; TOP – topic; F/f – feminine; PASS – passive; Q – question particle; GEN – genitive; PF – perfect; 1 – first person; hum – human; PL/pl – plural; 2 – second person; inf – infinitve; POL – polite; 3 – third person 2. I would like to thank Prof. Peter Hook for bringing these examples to my attention.

Invisible Agent Constructions in South Asian Languages

81

References Alfonso, Anthony 1980 Japanese Language Patterns. Tokyo: Sophia University L.L. Center of Applied Linguistics. Bhatia, Tej 1987 A history of the Hindi grammatical tradition. Leiden: E.J. Brill. Damle, Moro 1911 Shastriya Marathi Vyakaran (A Scientific Grammar of Marathi). Pune: Deshmukh ani company. Hinds, John 1986 Situation vs. Person focus. Tokyo: Kurosio Publishers. Ikegami, Yoshihiko 1981 ‘‘Suru’’ to ‘‘Naru’’ no Gengogaku (The Linguistics of ‘‘Do’’ and ‘‘Be come’’). Tokyo: Taishukan. Ikegami, Yoshihiko 1991 ‘‘DO-language’’ and ‘‘BECOME-language’’: Two contrasting types of linguistic representations. In The Empire of Signs: Semiotic essays on Japanese culture, Yoshihiko Ikegami (ed.), 285–326. Amsterdam: John Benjamins. Jacobsen, Wesley 1992 The Transitive Structure of Events in Japanese. Tokyo: Kurosio Publishers. Kachru, Yamuna 2007 Hindi. Amsterdam: John Benjamins. Kunihiro, Tetsuya 1974 Ningen chuushin to jyoukyou chuushin – nichiei hyougen kouzou no hikaku [Person-centeredness and Situation-centeredness – A contrastive study of the structure of English and Japanese expressions]. Eigo Seinen, Feb. issue: 688–690. Monane, Tazuko and Lawrence Rogers 1977 Cognitive features of Japanese language and culture and their implication for language teaching. In Proceedings of the UH-HATJ conference on Japanese language and linguistics, John Hinds (ed), 129–37. Honolulu: Univ. of Hawaii.. Pandharipande, Rajeshwari 1997 Marathi. London: Routledge. Pardeshi, Prashant 2000a Transitivity and voice: a Marathi-Japanese contrastive perspective. Ph.D. diss., Kobe University, Japan. 2000b The Passive and Related Constructions in Marathi. The Yearbook of South Asian Languages and Linguistics 2000: 147–171.

82 Prashant Pardeshi To appear

A Tale of Three Passives in Marathi – A Glimpse into the Subtle Interplay between Agent Defocusing and Intentionality. In CLS 42. Chicago: Chicago Linguistic Society. Siewierska, Anna 1984 Passive: A Comparative Linguistic Analysis. London: Routledge Kegan & Paul. Sridhar, S. N. 1990 Kannada. London: Routledge. Shibatani, Masayoshi 1985 Passive and Related Constructions. Language 61 (4): 821–848. Teramura, Hideo 1976 ‘‘Naru’’ hyougen to ‘‘suru’’ hyougen – nichiei ‘‘tai’’ hyougen no hikaku (‘‘Become’’ and ‘‘do’’ expressions – a comparison of voice expression in Japanese and English). In Nihongo to Nihongo Kyouiku: Moji-hyougenhen (Japanese and Japanese Pedagogy: Writing and Expression). Tokyo: National Language Research Institute. 49–68. [Reproduced in 1993 in Teramura Hideo Ronbunshu II – GengogakuNihongokyouikuhen (Collection of Papers by Hideo Teramura, Vol. II, Linguistics and Japanese Language Pedagogy Related Papers), Tokyo: Kurosio Publishers.]

A Pragmatic Account of the Hindi Presumptive

Ghanshyam Sharma

Taking Hindi presumptive modality as a case study, this paper argues that modality is detrimental in an overall organization of the tense-aspect system in languages in which these categories are marked, although only tense and aspect categories have attracted much attention in linguistic research. The paper analyzes di¤erent usages of the future form of the Hindi verb hona¯ ‘to be’ – both as a simple and an auxiliary verb – to ascertain its real modal meanings. Contrary to the widely-held view, this verb form may very rarely refer to situations which are to hold at a point in time subsequent to speech time, as the term ‘future’ may require. The paper claims that a better understanding of the role played by modality can open up a new chapter in the pragmatics of the tense-aspect system. The absence or presence of one of the tense-aspect categories in a language cannot be simply considered to be just a matter of chance. Rather, it should be an indicator of di¤erent pragmatic factors which are operational in the tense-aspect system of that language.

1. Why presumptive?1 – A heuristic Natural-language utterances consist, among others, of statements made by the speaker about entities, things, situations, phenomena, happenings or any type of human experience by which she is surrounded, which she has to confront and to which she wants to refer in order to say something significant in a natural-language communication setting. These statements, for the most part, are propositions which are employed by the speaker to assert the truth or falsity of the real properties reported therein. When the speaker engages in any human communication, she is supposed to assert the truth of the proposition, not just for the sake of making an assertion, but also to signal her agreement with it.2 Furthermore, by agreeing with the truthconditions3 of the proposition, the speaker inevitably subscribes to the modal meanings of which the proposition will be a vehicle. The modal

84 Ghanshyam Sharma meanings, therefore, are simply the speaker’s commitment to the truthfulness of the proposition.4 In other words, at times she is in a stronger position to assert the truthfulness of the proposition (for example, in the case of making an utterance of a proposition such as ‘This is [a picture of ] an elephant’,5 seeing the picture of an elephant in figure 1), for she possesses epistemic elements which are based on her first-hand knowledge.

Figure 1.

Natural-language utterances, in addition, are made up of those propositions whose truthfulness cannot be substantiated by elements of knowledge possessed by the speaker. In such cases, the speaker finds herself not in a position to make any irrefutable claim about the truthfulness of the proposition (for example, in the case of uttering ‘This is [a picture of ] an elephant’, upon seeing the picture of a creature in figure 2), as any utterance of it is going to be based solely on her belief, rather than on some direct knowledge of the facts reported in it.

Figure 2.

Thus when viewing the picture of an elephant in figure 1, the speaker will be in a stronger position to make an utterance comprising the proposition ‘This is [a picture of ] an elephant’, because the proposition will turn out to be fully corroborated by the evidence which is based on her ‘knowledge’. The proposition, in addition, will be endowed with a covert modal meaning, namely ‘necessity’, as symbolized in (1a), i.e., ‘This is necessarily [a picture of ] an elephant’. The same, however, could not be said about an utterance of the same proposition, ‘This is [a picture of ] an elephant’, if the speaker should make one upon seeing the picture in figure 2, given that all the information she possesses would negate its veracity. An elephant, for example, for all she knows, possesses neither horns nor a sharp-edged hump

A Pragmatic Account of the Hindi Presumptive

85

nor a wagging tail. What she will be left with in such a situation is her sole belief, which would need to be substantiated by epistemic modal elements such as ‘necessity’ or ‘possibility’. On the basis of her knowledge of other animals which may bear some resemblance to the strange creature in figure 2, she would probably be in a position to make an utterance which comprises a proposition such as ‘Possibly/Probably this is [a picture of ] an elephant.’ One should, however, notice that in making such an utterance she will be in a doubtful epistemic state, and, consequently, not able either to assert the proposition, ‘This is [a picture of ] an elephant’, or to negate it, ‘This is not [a picture of ] an elephant’. Hence it will be considered to be based on her belief substantiated by epistemic ‘possibility’ only. Many natural languages, like Hindi, employ a subjunctive mood to express such epistemic possibility beliefs, as symbolized in (1d). There can be another epistemic state in which the speaker’s utterance of the above mentioned proposition can be thought to carry her belief substantiated by a modal ‘necessity’ rather than modal ‘possibility’. For example, being aware of possible physical di¤erences that one might encounter among di¤erent races of elephants which inhabit geographically distant places on the globe, she could probably also make an utterance consisting of a proposition such as ‘This must be [a picture of ] an elephant’. Although such an utterance, as the one in the previous case, would still be based on the speaker’s belief rather than knowledge, it will, unlike the previous case, carry an epistemic ‘necessity’ instead of an epistemic ‘possibility’, as symbolized in (1b). In other words, although some of the physical distinctions displayed by the strange creature in figure 2 will not allow her to associate it definitively with an elephant, its many other characteristics that resemble those of an elephant, rather than that of any other known creature, will provide her with reasonable evidence or assumptions lending probability to her belief to link it to the elephant – the only creature with similar characteristics that she is aware of! A ‘presumptive’, therefore, can be conceived of as a type of epistemic modality6 which is employed by the speaker to convey her ‘necessary beliefs’. There is a fundamental di¤erence between the epistemic state of a subjunctive and that of a presumptive which, alas, has not been clearly maintained in grammar books. The presumptive, as a consequence, can be seen as being classified, together with the subjunctive, as a mood or a tense which is supposed to express doubts, uncertainty, etc.7 A subjunctive mood, as pointed out above, is employed to convey the epistemic state which has its origin in the ‘possibility’ of a ‘belief ’. A presumptive,

86 Ghanshyam Sharma on the other hand, carries the speaker’s ‘necessary’ beliefs. Along the lines of this heuristic process of combining di¤erent epistemic modals, it seems reasonable to propose yet another epistemic category which can be considered to be based on the speaker’s knowledge and to carry epistemic ‘possibility’ (i.e., knowledge of a possibility). In order to account for the di¤erent epistemic states explored above through our heuristic method, we used a slightly revised8 version of the notation popularized by Hintikka (1962). Following the symbolism, the overall picture of epistemic modality can be represented through di¤erent possible combinations of epistemic modal elements in the following manner: (1)

a. Kjp b. Bjp c. K‚p d. B‚p

The epistemic categories formalized in (1a) and (1c) are cases of utterances made by the speaker on the basis of her ‘knowledge’. They can be deemed to belong to the same type from this point of view. The di¤erence between them can be seen in terms of their modal meanings only: the first one carries modal ‘necessity’, the second modal ‘possibility’. The semantical ‘knowledge’ of the speaker in the case of (1a) forms the basis for expressing epistemic ‘necessity’. In other words, for all she knows, the proposition p is necessarily true, and it is not possible that not-p (i.e., F‚Fp). The category in (1c), although based on the speaker’s ‘knowledge’, is meant to carry the meaning of modal ‘possibility’ instead of ‘necessity’. It can be conceived of as an epistemic category which means that, for all the speaker knows, the facts reported in p are possibly true. The category, therefore, is employed to indicate an anomalous situation, as one cannot know things which are inexistent, yet possibly true. In this sense, it seems to be a case of alethic modality rather than of epistemic category.9 This modal meaning is concerned with eventualities and thus not normally grammaticalized in natural languages. In order to understand the real application of this modal category we could think of situations such as the following one: ‘I know that it is possible that it will rain in Venice tomorrow’. Needless to say, such an utterance ends up in the category of beliefs since one cannot, at the time of utterance, know that it will rain in Venice tomorrow. If we have to stick to

A Pragmatic Account of the Hindi Presumptive

87

the real meaning of ‘possible knowledge’, we will have to imagine a modal situation such as this one: ‘I know that [it is possible that] it may rain in Venice anytime’ which is an example of an eventuality, and not a modal possibility of rain in Venice at a particular time on a particular day. The epistemic categories abbreviated in (1b) and (1d), on the other hand, are based on speaker’s belief rather than her ‘knowledge’. The category in (1b), for example, means that, her no-knowledge of p notwithstanding, the speaker nonetheless believes that necessarily p and thus her belief is that it is not possible not-p (i.e., F‚Fp). The category in (1d), on the contrary, is employed by the speaker to express a belief based on ‘possibility’, and thus means that, given that she does not know it to be the case, she believes that possibly p, and that it is not necessarily not-p (i.e., FjFp). In conclusion, then, the presumptive can be considered to be a type of epistemic modality – the category represented in (1b) – i.e., Bjp. Some languages grammaticalize this category either morphologically or by other grammatical devices, while others make use of modals, as is the case with the English language. It is, however, an indispensable means of expressing the speaker’s ‘necessary beliefs’, which are di¤erent from ‘necessary knowledge’ (i.e., Kjp), on the one hand, and from ‘possible belief ’ (i.e., B‚p), on the other.

2. Presumptive: A tense or a type of modality? Verb forms in a language may express di¤erent qualities, among which tense, mood and aspect (TMA hereafter) are the most inter-related since they all concern the event or action denoted by the verb. However, there is no known example of a language in which TMA qualities are encoded – distinctly and universally – throughout the structure of the language. Some languages have grammatical devices to mark one category but not the other. Other languages employ the same grammatical device to mark two or even three categories together. Most of the studies on TMA categories have, consequently, diverged on di¤erent levels because in the absence of one the others can be conceived of as the marker of the default meaning of the first. Among these, since time is an essential part of human cognition, tense is obviously the most widely discussed and commonly known notion in language. The present paper argues that, although grammatical tense seems to occupy the most important hierarchical position among the three,

88 Ghanshyam Sharma as it is related to the most evident category – time – it is modality (or grammatical mood for that matter) which is responsible for an overall organization of TMA categories in the system of a language. According to Comrie, tense is deictic and ‘‘relates the time of the situation referred to some other time’’, i.e., to a deictic centre, usually the time of speaking (1976: 1–2). If we stick to the original insight provided by Reichenbach (1947: 287–298) that tense and aspect involve three time references (Event time, Speech time and Reference time), three simple tenses can be represented in the following manner:10 (2)

a. Past tense

E,R—S

b. Present tense

E,R,S

c. Future tense

S—E,R

This obviously and overtly simple division between di¤erent tenses, however, is not incontrovertible. Firstly, if we stick to the category of pure tense, we can theoretically have three types of tenses only; the past, the present and the future, as all other tense types are obtainable solely through an inevitable combination of the above three along with other grammatical aspects. Furthermore, even these three seemingly pure types of tenses are hardly ever employed without some inherent elements of aspect, since, as mentioned above, they inevitably acquire a default aspectual interpretation in the absence of any evident aspect marker. For example, a past tense verb form, even without a perfective marker, by default, tends to acquire a perfective-aspect interpretation if it does not carry any imperfective marker. Similarly, a verb form without a perfective-aspect marker tends to acquire a default ‘imperfective’ interpretation, even if there is no marker of imperfectivity attached to it. Or else, a pure aspectual category may by default give an interpretation similar to tense. The so-called aoristic past in Hindi, for instance, is a tense with no marker of past tense at all.11 It acquires by default a past-tense interpretation since, in the absence of a tense marker, a perfective aspect refers to a situation which obviously should have taken place prior to the time of speech. Secondly, whereas past and present tenses are concerned with events or actions that have either taken place already or take place contemporaneously with the utterance, the future tense is concerned with an event or action which is subsequent to the present moment or yet to happen.12 This characteristic makes the former two completely

A Pragmatic Account of the Hindi Presumptive

89

distinct from the latter: the situations denoted by past and present concern real events or actions and, therefore, are based on the speaker’s epistemic knowledge, whereas the situation denoted by a future tense concerns actions or events that may or may not happen at the time referred to in the sentence, and hence it originates from the speaker’s beliefs only. It is because of this special nature of the future tense that its real status has been questioned in di¤erent languages, including English.13 As stated above, the TMA categories are inter-related and, therefore, none of them can be considered independent of the others, since the presence or absence of one category is destined to give rise by default to interpretations which can be linked to other categories. We argue, nonetheless, that an overall organization of TMA elements in a natural language is determined by modality, although not all languages mark it overtly. The presumptive, therefore, is not a category of tense but a type of modality, and is employed by the speaker to convey ‘epistemic necessity’ based on her ‘belief ’ rather than her ‘knowledge’. The future form of the auxiliary which is generally employed to express this modality14 has no future time reference at all in these situations. In other words, whereas a future tense may induce an implicature15 that the situation described in it does not hold at the time of speaking, the future form used to express presumptive modality does not carry any such implicature at all. It indicates situations in which events or actions are either contemporaneous with or have already taken place at the time of speech. As a consequence, we believe that modal elements have to be incorporated into any model that intends to analyze tense and aspect. The Reichenbachian model has been elaborated and revised by di¤erent authors (Hornstein 1990: 117; Giorgi, and Pianesi 1999: 29) to include di¤erent combinations of tense and aspect.16 As will be discussed in the following sections, there are four tense-aspect combinations in Hindi which can express presumptive modality. In order to give an account of their modal meaning, it is necessary to add the modal element to the Reichenbachian notation (Reichenbach 1970: 290). It is possible to incorporate the necessary belief (i.e., Bjp) element in the Reichenbachian notation which intends to describe tenses by representing three points, namely, point of speech (S), point of the event (E) and point of reference (R). With this added modal element, four tenseaspect combinations in Hindi will have the following form: (3)

a. Presumptive without an aspect

Bj: S,R,E

b. Habitual Presumptive

Bj: S,R,E

90 Ghanshyam Sharma c. Progressive Presumptive

Bj:

S,R,E

d. Perfective presumptive

Bj:

E—S,R

According to this formalism, (3a), (3b) and (3c) all stand for ‘‘the speaker believes that the event necessarily referred to a point in time which is contemporaneous with speech time.’’ The main di¤erence among these three can be thought to be aspectual only. (3d) indicates that ‘‘the speaker believes that necessarily the action or event reported has come to an end at speech time.’’

3. Presumptive modality in Hindi Hindi belongs to the group of languages which exhibit TMA categories quite systematically. It employs verbal auxiliaries to mark tenses, and encodes grammatical aspects through di¤erent morphological devices such as aspectual su‰xes on the root or an extra verbal item. The presumptive modality in Hindi is expressed by the future form17 of the auxiliary hona¯ ‘to be’. Although, it has been classified as a marker of the future tense by some grammarians,18 when employed as a helping verb, it generally refers to situations which have either taken place already or are to be held at a time contemporaneous with speech time. When used as either a single verb without an aspect or with a marker of progressive aspect, it can in addition have a reference to a time subsequent to the speech time. While analyzing the Hindi presumptive, we will try to show that modality should serve as a basis for understanding the logic behind the systematic organization of TMA system in Hindi.

3.1. Presumptive without an aspect As stated above, one of the most common functions of the future form of the Hindi verb hona¯ ‘to be’ can be seen in the domain of presumptive modality. In such cases, it is employed to refer to a situation that is to hold at a point in time which is not subsequent to, but contemporaneous with the moment of speech. For example, the situations referred to in (4) and (5) may hold at any point in time, i.e., at a point in time either prior

A Pragmatic Account of the Hindi Presumptive

91

to, or contemporaneous with, or subsequent to speech time. When referring to a prior time it means that the state of a¤airs reported in it was true at the time of speech prior to speech time. The sentence in (4) may thus, depending on the time reference, carry the following modal meaning: although the speaker does not know the real age of Ramu, she, nonetheless, believes that Ramu must be twenty years old at a point in time in the past, the present or the future. Similarly, in (5) the speaker does not know, but believes that on the basis of her assumptions Ramu must then be/ have been in the kitchen in the past, now or in the future. In terms of modality all these cases, however, carry the same epistemic necessity based on the speaker’s belief. (4)

ra¯mu kı¯ umr [tab/ ab/ kal ] bı¯s sa¯l Ramu-M of-F age-F [thenpast / now/ tomorrow] twenty years hogı¯. FUT-F-SG19 ‘Ramu must be twenty years old [then (past)/ now/ tomorrow].’

(5)

ra¯mu [tab/ ab/ kal ] venis me˜ hoga¯. Ramu-M [thenpast / now/ tomorrow] Venice in FUT-M-SG ‘Ramu must be in Venice [then (past)/ now/ tomorrow].’

3.2. Presumptive with aspect In addition to the above mentioned usages as a simple verb, the future form of hona¯ ‘to be’ is also employed as an auxiliary verb to express presumptive modality in combination with di¤erent aspectual markers attached to the root of the verb. The exact number of aspects in Hindi has been controversial, as authors have not kept the di¤erence between grammatical aspect and lexical aspect (or Aktionsart) clear, and have consequently tried to classify even ‘telic’ and ‘atelic’ aspects as grammatical aspects.20 For the purpose of a discussion on modality in examples with aspectual markers in Hindi, we do not consider it necessary to enter into the debate to settle the exact number of aspects in Hindi. For the present purpose it will su‰ce to analyze some of the aspects only. The presumptive in Hindi can be seen in imperfective (both habitual and progressive), and perfective aspects, as well as some other less commonly discussed aspects.21

92 Ghanshyam Sharma 3.2.1. Presumptive with imperfective aspect Generally, all ‘non-perfective’ aspects can be included in the category of imperfective, even though they are not imperfective in the same sense. Along these lines, habitual, progressive, continuative, and frequentative may all come under the umbrella term ‘imperfective’, although they cannot be considered to carry the same ‘imperfectivity’ in the same manner. Habitual aspect is considered an imperfective aspect since it expresses the occurrence of an event or state as a characteristic of a period of time. In other words, it indicates a situation in which actions or events take place as a daily routine. Progressive aspect is a continuous aspect that expresses processes, not states. The continuative can be considered as a subcategory of the progressive, indicating states rather than processes. It may also indicate a continuous state of a process. The frequentative, on the other hand, indicates a frequency of actions completed on di¤erent occasions rather than a continuous action. Both habitual and frequentative, therefore, should be considered imperfective in the sense that they indicate a series of di¤erent actions rather than just one single action. 3.2.1.1. Habitual aspect This class of examples has rightly been described as ‘presumptive habitual’ by some authors22 since the imperfective form of the main verb in these examples marks a habitual aspect, whereas the future form of the auxiliary hona¯, ‘to be’, which follows the main verb, carries presumptive modality, as in (6). The imperfective aspect in this case is indicated by the su‰x -t on the root of the verb. The future form of the auxiliary, however, can only have reference to a point in time which is either prior to or contemporaneous with speech time. It can never refer to a situation with future reference, as will be clear from the double question marks before the time adverbial ‘next year’ which makes it pragmatically anomalous: (6)

ra¯mu [ pichle sa¯l/ a¯jkal/ ??agle sa¯l ] bahut a¯m Ramu-M [last year/ these days/ next year] much mango kha¯ta¯ hoga¯. eat-IMPFV-M-SG FUT-M-SG ‘Ramu must be eating a lot of mango [last year/ these days/ ??next year].’

A Pragmatic Account of the Hindi Presumptive

93

3.2.1.2. Progressive aspect Progressive (or continuous23) is another class of imperfective aspect in Hindi, and is marked by the root of rahna¯ ‘to stay’. Presumptive modality can exhibit this aspect by attaching the future form of the auxiliary hona¯ ‘to be’ to it. However, unlike the presumptive in habitual aspect, the presumptive progressive can refer to all three kinds of situations. It may refer to situations whose state of a¤airs were true at a point in time prior to, are true at a point contemporaneous with and will be true subsequent to the moment of speech, as in (7), although with certain verbs its reference to a future point in time is anomalous, as in (8). (7)

ra¯mu [tab/ ab/ kal ] venis me˜ ghu¯m Ramu-M [thenpast / now/ tomorrow] Venice in walk raha¯ hoga¯. PROG-M-SG FUT-M-SG ‘Ramu [must have been walking then/ must be walking now/ must be walking tomorrow] in Venice.’

(8)

ra¯mu [un dino˜/ a¯jkal/ ??agle sa¯l ] bha¯rat me˜ bahut Ramu-M [those days/ nowadays/ next year] India in much a¯m kha¯ raha¯ hoga¯. mango eat PROG-M-SG FUT-M-SG ‘Ramu [must have been eating a lot of mango last year/ must be eating a lot of mango these days/ ??must be eating a lot of mango next year] in India.’

3.2.1.3. Continuative aspect The term ‘continuative’ is used to refer to a kind of verbal aspect which indicates a continuous state of action. In this sense it belongs to the category of imperfective rather than perfective as it does not indicate completion of the action, although it does not indicate an un-ending state of the action either. The imperfective participle of the verb is followed by the continuative auxiliary rahna¯ ‘to stay’ and the future form of hona¯ ‘to be’ which marks the presumptive modality. Also in this aspect the presumptive marker may not refer to situations whose state of a¤airs will be true at a time subsequent to the speech time.

94 Ghanshyam Sharma (9)

ra¯mu [un dino˜/ a¯jkal/ ??agle sa¯l ] das baje tak Ramu-M [those days/ nowadays/ next year] ten o’clock until sota¯ rahta¯ hoga¯. sleep-IMPFV-M-SG CONT-M-SG FUT-M-SG ‘Ramu must be used to continue sleeping until 10 o’clock those days/ these days/ ??next year.’

3.2.1.4. Frequentative Presumptive modality can be expressed with a frequentative aspect as well. In this case, the frequentative aspect seems to belong to the territory of the habitual and thus by nature to be an imperfective. In certain contexts frequentative and habitual can be used interchangeably, although they di¤er in meaning quite significantly. A verb expressing the frequentative meaning, karna¯ ‘to do’, follows the perfective participle of the main verb. Given that it is an imperfective, the frequentative marker can have all aspects except perfective.24 Its use in presumptive modality, however, is restricted to past and present reference only. Its reference to a future state of a¤airs is anomalous, as will be clear from (10): (10) ra¯mu [un dino˜/ a¯jkal/ ??agle sa¯l ] roz Ramu-M [those days/ nowadays/ next year] everyday ghumne ja¯ya¯ karta¯ hoga¯. stroll-OBL go-PFV-M-SG FREQ-IMPFV-M-SG FUT-M-SG ‘Ramu [those days/ nowadays/ ??next year] must be used to going to the park daily.’

3.2.2. Presumptive with perfective aspect The presumptive modality can frequently be seen in examples showing a perfective aspect. A perfective aspect in Hindi is obtained by a -V su‰xed on the verb.25 This vowel displays concord according to the gender and number. In the sense of presumptive modality, the perfective aspect may refer to situations or events which either have taken place prior to, or take place contemporaneously with, speech time. However, with presumptive meaning it is not generally employed to indicate those actions or events which are to hold at a future point in time, as can be seen in (11), where a

A Pragmatic Account of the Hindi Presumptive

95

time reference ‘tomorrow’ makes the sentence anomalous, if not totally ungrammatical. An augmented perfective aspect in some cases requires the use of a compound verb instead of a simple verb which is obtained by adding a vector verb to the root of the main verb. This vector verb carries elements of verbal concord, as is the case in (12). Here, however, it can describe situations which hold at a point in time subsequent to speech time. (11) ra¯mu [kal/ a¯j/ ??kal ] zaru¯r venis Ramu-M [yesterday/ today/ tomorrow] certainly Venice gaya¯ hoga¯. go-PFV-M-SG FUT-M-SG ‘Ramu must have gone to Venice [yesterday/ today/ ??tomorrow].’26 (12) ra¯mu ne [tab tak/ ab tak/ tab tak] apna¯ Ramu-M ERG [by thenpast / by now/ by thenfuture ] his-REFL ka¯m pu¯ra¯ kar liya¯ hoga¯. job-M complete do take-PFV-M-SG FUT-M-SG ‘Ramu must have finished his job by [then (past)/ now/ then (future)].’

3.3. Presumptives in the passive voice The presumptive modality can be seen throughout the verbal paradigm in Hindi, including the passive voice. In the passive voice it can exhibit the habitual, progressive and perfective aspect, as in (13), (14) and (15), but not the frequentative and continuative, as both indicate the state of the speaker rather than an action. It may also appear in a passive27 verb with inceptive aspect, as in (16). It may express presumptive modality referring to situation in the past and present but not future. Its use with the progressive aspect, however, can have a future reference, as can be seen in (14). (13) [un dino˜/ a¯jkal/ ??agle sa¯l ] bha¯rat me˜ bahut a¯m [those days/ nowadays/ next year] India in much mango kha¯e ja¯te ho˜ge. eat-PFV-M-PL PASV-IMPFV-M-PL FUT-M-PL ‘A lot of mangos must be eaten (habitually) in India [those days/ nowadays/ ??next year].’

96 Ghanshyam Sharma (14) [tab/ ab/ tab] rasoı¯ me˜ da¯l [thenpast / now/ thenfuture ] kitchen in lentils-F-SG bana¯¯ı ja¯ rahı¯ hogı¯. cook-PFV-F-SG PASV-ROOT PROG-F-SG FUT-F-SG ‘In the kitchen lentils must be under preparation [then (past)/ now/ then (future)].’ (15) [kal/ a¯j/ ??kal ] da¯l bana¯yı¯ [yesterday/ today/ tomorrow] lentils-F-SG cook-PFV-F-SG gayı¯ hogı¯. PASV-PFV-F-SG FUT-F-SG ‘Lentils must have been cooked [yesterday/ today/ ??tomorrow].’ (16) [tab/ ab/ ??tab] su¯raj dikhne [thenpast / now/ thenfuture ] sun-M-SG appear-OBL laga¯ hoga¯ INCE-PRF-M-SG FUT-M-SG ‘The sun must have begun to be visible then (past)/ now/ ??then (future).’

3.4. Presumptive in rhetorical usages The presumptive modality is also frequently used in those expressions which can be classified as ‘rhetorical’ usages. In such cases the future form of hona ‘to be’ does not refer to a future point in time but rather to a general situation that might hold in the present. We believe that such utterances are fundamentally based on a type of conditional argumentation. Roughly, its reasoning can be paraphrased in the following way: ‘although the speaker does not believe that jp, even if it is the case that jp, then q. Needless to say, the utterance is based on a modal meaning. (17) hoga¯ vah paise va¯la¯, mujhe kya¯ FUT-M-SG he rich I-DAT-OBL what ‘He may be rich, it doesn’t concern me.’

lena¯-dena¯. concern-M-SG

As stated above, this usage of the future form of hona¯ ‘to be’ cannot have any reference to the situation which holds true at a moment subsequent to

A Pragmatic Account of the Hindi Presumptive

97

speech time. It can therefore be yet further proof that the form has been wrongly classified in the class of future tense.

3.5. Presumptive with a reference to past, present and future As discussed above the future form of hona¯ ‘to be’ is the marker of presumptive modality in Hindi, and generally refers to those situations in which actions or events have either taken place already or are under way. The same form with di¤erent aspectual meanings can also be employed to refer to situations in which events or actions have a reference to past time, i.e., a point in time prior to speech time. Only in cases in which this form is employed either as simple verb or in association with the progressive aspect, can it refer to situations in which actions or events may have reference to a point in time subsequent to the moment of speech. Such use, however, should be judged as a case of presumptive modality rather than future tense, as its truth-conditions cannot be verifiable. The fact that the future form of hona¯ ‘to be’ can be employed to refer to situations in which the actions or events take place at a moment di¤erent from the subsequent should provide yet further proof of its being a marker of modality, rather than of the future tense. The ample data discussed above will su‰ce to prove that the future form of hona¯ ‘to be’ is employed in Hindi mainly to express presumptive modality, rather than to indicate a future situation in which actions or events hold at a point in time that is subsequent to the time of speech. Even when it is used to indicate actions or events likely to take place at a future point in time, it carries the speaker’s ‘necessary beliefs’ which are either in anticipation of events to come or based on the assumption that the action referred to in the utterance will necessarily be carried out by the agent – a clear case of ‘necessary belief ’. A comprehensive picture of various modal usages of the future form of the Hindi verb hona¯ ‘to be’ is presented in the following table 1. It becomes evident from the distributional situation of the future form of hona¯ ‘to be’, as shown in table 1, that it can very rarely be employed to refer to situations in which actions or events are to hold at a point in time subsequent to speech time, although it can appear in the progressive aspect and in cases where it is used as a simple rather than an auxiliary verb. However, even in those cases where it does have a future reference, it carries an epistemic belief of the speaker. Furthermore, its combination with the

98 Ghanshyam Sharma Table 1. Classification of various uses of the future form of the Hindi verb hona¯ ‘to be’ according to the di¤erent points in time they may refer to. The occurrence of it in the future with progressive aspect is limited to certain verbs only. The asterisc indicates its inapplicability to certain verbs. Verb hona¯

Past

Present

Future

Time of the action/event

E,R—S

E,S,R

S—R,E

Active voice (1) As a main verb (2) With habitual aspect (3) With progressive aspect (4) With continuative aspect (5) With frequentative aspect (6) With Perfective aspect

Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes Yes

Yes No Yes* No No No

Passive voice (1) As a main verb (2) With habitual aspect (3) With Progressive aspect (4) With continuative aspect (5) With Frequentative aspect (6) With Perfective aspect

– Yes Yes Yes Yes Yes

– Yes Yes Yes Yes Yes

– No Yes No No No

perfective aspect in Hindi is not to be confused with the future perfect of English. From the discussion above we can conclude that the future form of the Hindi verb hona¯ ‘to be’ is the marker of presumptive modality rather than of the future tense.

4. Scalar meaning of presumptive modality As has been demonstrated elsewhere (Sharma 2002), Hindi auxiliaries, in addition to relating situations indicating events or actions to a deictic center, also carry modal meanings which are compositional. Recollect that three out of four epistemic categories established through our heuristic argument in (1), namely Kjp, Bjp, and B‚p, are systematically encoded, respectively by the hœ, ho and hoga¯ forms of the Hindi auxiliary, whereas the category K‚p, i.e., ‘possible knowledge’, can be considered to be an alethic category, and is conveyed by the modal verb sakna¯ ‘be able to, can’. We argue that such forms of auxiliary are employed in Hindi primarily to encode

A Pragmatic Account of the Hindi Presumptive

99

epistemic modalities. Further research into the pragmatics of the Hindi TMA system can reveal many complexities which remain unresolved. For a better understanding of the TMA system in Hindi, we believe, the above mentioned epistemic categories should also be included in glosses, as in (18), (19), (20) and (21), as the di¤erent forms of Hindi auxiliary does not always refer to a point in time they are considered to refer: (18) vah cor {hœ/ ho/ hoga¯} he thief {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘He {is/ may be/ must be} a thief.’ (19) vah roz naha¯ta¯ he everyday shower-IMPFV-M-SG {hœ/ ho/ hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘He {takes/ may be taking/ must be taking} a shower everyday.’ (20) vah ab so raha¯ he now sleep PROG-M-SG {hœ/ ho/ hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘He {is/ may be/ must be} sleeping now.’ (21) usne a¯j a¯m kha¯ya¯ he-OBL-ERG today mango-M-SG eat-PFV-M-SG {hœ/ ho/ hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘He {has/ may have/ must have} eaten a mango today.’ Now, if we look into the inter-relationship that holds among the three forms, we notice that they express their internal scalar positions as well. By mapping their hierarchical modal meanings on the traditional square of opposition, it is possible to understand their internal organization. For instance, on the Square of Opposition28 in figure 3, the hœ form of the auxiliary should occupy corner A, whereas the ho form occupies corner I, as they are meant to carry modal necessity (i.e., Kjp) and modal possibility (i.e., B‚p), respectively. Their scalar position on the pragmatic scale is detrimental in the implicatures they are deemed to induce. The meaning of

100 Ghanshyam Sharma

Figure 3. On the traditional square of opposition A and E are stronger than their counterpart I and O. A entails I, and both are on the positive (A¤Irmo) side, whereas E entails O, and both are on the negative (nEgO) side. A and O, and E and I are contradictory (meaning they cannot both be true, nor can they both be false). The relation between I and O has been controversial over the ages, but has found its proper use in pragmatics now.

a scalar linguistic item A in a linguistic context C depends not only on what A means both universally (i.e., in all linguistic context) and specifically (say, in context C ), but also upon what another scalar item, say B, could have meant had it been used instead (Horn 1989, Levinson 2000, Hirschberg 1991). Scalar implicatures can play a fundamental role in characterizing modal meanings of tense and aspect.29 It is therefore possible to understand the pragmatic scalar meaning these categories are supposed to implicate conversationally. On the square of opposition the corner A items are stronger than the corner I items. In fact, I is entailed by A. Thus, making an utterance of an item from corner I induces an implicature which is tantamount to the modal meaning, namely, that the speaker does not know, but believes that it is not necessarily the case that p. For instance, by selecting a weaker term from the pragmatic scale that occupies the corner I, e.g. form ho of the auxiliary, the speaker implicates that she is not in a position to make an utterance of the stronger item, i.e. the form hœ which sits on the corner A. If she had known the state of a¤airs reported in the utterance with certainty, she would have made an utterance using the item from corner A. As will be shown later, such an implicature is cancelable by attaching either a canceling or a suspending phrase. If we turn now to the presumptives, we notice that their epistemic position comes between indicatives and subjunctives, since they carry modal necessity (i.e., jp) rather than modal possibility (i.e., ‚p), as in (22b), while on the other

A Pragmatic Account of the Hindi Presumptive

101

hand, they are based on the speaker’s ‘belief ’ rather than ‘knowledge’, as can be seen in (22a). On the scale of semantic strength presumptives are therefore weaker than indicatives, but stronger than subjunctives. Their respective positions on the pragmatic scale can be explained in the following manner: (22) a. 3Kjp, Bjp4 b. 3Bjp, B‚p4 As can be seen from the square of opposition in figure 3, indicatives (i.e., jp) sit on corner A, whereas subjunctives (i.e., B‚p) on corner I. The same hierarchical epistemic relationship can be established between presumptives and subjunctives, although they both are based on ‘belief ’ rather than ‘knowledge’. Presumptives carry epistemic necessity (i.e., jp), thus must occupy corner A, whereas subjunctives occupy corner I, as in (22b). As far as the relationship between indicatives and presumptives is concerned, we notice that, although from the point of view of their epistemic strength they seem to occupy the same position A on the semantic scale, as both carry epistemic necessity, from the point of view of ‘knowledge’ and ‘belief ’ elements, presumptives are on the lower level I with respect to indicatives, since indicatives are on the higher A, as in figure 4. It has been argued that natural language categories may or may not bear a close resemblance to their logical counterparts, thus making it necessary to use the logical tools with caution to deal with complexities of natural language category. Hence the combination of modal categories becomes essential to show the strength these categories seem to possess on the semantic scale: 3Kjp, Bjp, B‚p4

Figure 4.

102 Ghanshyam Sharma In order to ascertain that neither indicatives nor presumptives carry epistemic possibilities, it is su‰cient to observe the pragmatic anomalies such categories are likely to have in the presence of an epistemic possiblity marker, such as ‘it is possible that’ in English. For instance, the Hindi examples (23) through (26) all contain an element of epistemic possibilty, i.e. the phrase ho sakta¯ hœ ki ‘it is possible that, possibly’. We notice that this added element renders the use of hœ and hoga¯ auxiliaries anomalous, since none of them expresses epistemic possibility – a semantic quality required by the former. As an utterance must carry the same epistemic modal meaning universally, hœ and hoga¯ auxiliaries come into conflict with the epistemic element ‘it is possible that’. The auxiliary which carries epistemic possibility, namely ho, is the only possible option in this context:30 (23) ho sakta¯ hœ ki vah cor it is possible that he thief {??hœ/ ho/??hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘Perhaps he {??is/ may be/ ??must be} a thief.’ (24) ho sakta¯ hœ ki vah roz naha¯ta¯ it is possible that he everyday shower-IMPFV-M-SG {??hœ/ ho/??hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘It is possible that he {??takes/ may be taking/ ??must be taking} a shower everyday.’ (25) ho sakta¯ hœ ki vah ab so raha¯ it is possible that he now sleep PROG-M-SG {??hœ/ ho/ ??hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘It is possible that he {??is/ may be/ ??must be} sleeping now.’ (26) ho sakta¯ hœ ki usne a¯j a¯m it is possible that he-OBL-ERG today mango-M-SG kha¯ya¯ eat-PFV-M-SG {??hœ/ ho/ ??hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG

A Pragmatic Account of the Hindi Presumptive

103

‘It is possible that he {??has/ may have/ ??must have} eaten a mango today.’ Having seen the semantic acceptability of the three forms in the presence of an epistemic possibility element, let us now try to check their semantic position with an added epistemic element which carries a modal ‘necessity’. Now, if we add a Hindi phrase such as mœ˜ ja¯nta¯ hu˜¯ ki, ‘I know that’, to an utterance, we notice that only the hœ form of the auxilairy is fully acceptable – as it is in a harmonious relationship with the epistemic necessity element in the added phrase. The semantic situation of the form hoga¯, however, is ambiguous as, on the one hand, it carries a ‘belief ’ element which assigns it a lower position on the semantic scale, while, on the other, it carries a modal necessity element that puts it on a higher position. For instance, the examples in (27), (28), (29) and (30) contain an added epistemic necessity element and therefore accept the hœ form in all contexts. The use of auxiliary ho (which expresses B‚p) would make the sentence semantically anomalous. The use of the form hoga¯, however, will show a lower degree of unacceptability because, as said above, its position can be considered ambiguous as it carries a necessary belief. The ambiguity in the situation of the hoga¯ form can disappear, if we substitute the ‘necessary knowledge’ element (mœ˜ ja¯nta¯ hu˜¯ ki) with a ‘necessary belief ’ element (mœ˜ ma¯nta¯ hu¯˜ ki). (27) mœ˜ ja¯nta¯ hu¯˜ ki vah cor I know PRES-1-SG-Kj that he thief {hœ/ ??ho/?hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘I know that he {is/ ??may be/ ?must be} a thief.’ (28) mœ˜ ja¯nta¯ hu¯˜ ki vah roz I know PRES-1-SG-Kj that he everyday naha¯ta¯ shower-IMPFV-M-SG {hœ/ ??ho/ ?hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘I know that he {takes/ ??may be taking/ ?must be taking} a shower everyday.’

104 Ghanshyam Sharma (29) mœ˜ ja¯nta¯ hu˜¯ ki vah ab so raha¯ I know PRES-1-SG-Kj that he now sleep PROG-M-SG {hœ/ ??ho/ ?hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘I know that he {is/ ??may be/ ?must be} sleeping now.’ (30) mœ˜ ja¯nta¯ hu˜¯ ki usne a¯j I know PRES-1-SG-Kj that he-OBL-ERG today a¯m kha¯ya¯ mango-M-SG eat-PFV-M-SG {hœ/ ??ho/ ?hoga¯} {Kj-PRES-3-SG/ B‚-SUBJ-3-SG/ Bj-FUT-3-M-SG} ‘I know that he {has/ ??may have/ ?must have} eaten a mango today.’ Having established the scalar positions of the three forms of the Hindi auxiliary hona¯ ‘to be’ through examples above, let us now turn to the question of what kind of scalar implicature – a class of generalized conversational implicature based on Gricean maxims of Quantity (Grice: 1989: 26) – the so-called future form of hona¯ may induce conversationally. As has been demonstrated by Horn (1989) and Levinson (2000), O corner elements on the square of opposition are the main elements responsible for inducing implicatures in a conversational setting. However, in order to ascertain how the selection of this form may induce implicatures in a conversational setting, it would su‰ce just to check whether the implicatures we ascribe to its scalar position are cancelable or not. As mentioned above, the relation between indicative and presumptive can be seen from the point of view of their scalar position in that the first contains elements of ‘necessary knowledge’, the second of ‘necessary belief ’. In the following example we will, by adding a ‘canceling’ phrase, try to see whether the implicature which we claim the future form of the Hindi auxiliary carries, is cancelable or not. As is clear from the examples in (31), (32), (33) and (34), an added canceling phrase containing a stronger element from corner A (stronger hœ ¼ Kjp and weaker hoga¯ ¼ Bjp) evaporates the implicature, i.e., ‘‘the speaker does not know that p’’. As can be seen from the following examples, the implicature induced by the weaker form disappears once the stronger element on the scale is added to the phrase. The type of resoning in these examples is similar to the following: ‘‘not only Weaker, Stronger’’.

105

A Pragmatic Account of the Hindi Presumptive

(31) vah cor hoga¯ nahı˜¯, he thief Bj-FUT-3-M-SG NEG, ‘He IS, not must be, a thief.’

cor thief

hœ Kj-PRES-3-SG

(32) vah roz naha¯ta¯ hoga¯ he evryday shower-IMPFV-M-SG Bj-FUT-3-M-SG naha¯ta¯ hœ shower-IMPFV-M-SG Kj-PRES-3-SG ‘He does take, not must be taking, a shower everyday.’ (33) vah ab so raha¯ hoga¯ he now sleep PROG-M-SG Bj-FUT-3-M-SG raha¯ hœ PROG-M-SG Kj-PRES-3-SG ‘He IS, not must be, sleeping now.’ (34) usne a¯m kha¯ya¯ he-OBL-ERG mango-M-SG eat-PFV-M-SG nahı¯˜, kha¯ya¯ hœ NEG eat-PFV-M-SG Kj-PRES-3-SG ‘He HAS, not must have, eaten a mango.’

nahı˜¯, NEG

nahı˜¯, NEG

so sleep

hoga¯ Bj-FUT-3-M-SG

The fact that the hoga¯ form (i.e., Bjp) is weaker than the hœ form (i.e., Kjp) can also be established by applying another technique. It has been suggested that in certain cases, by attaching a suspending phrase, it is possible to ascertain whether the implicature induced by any utterance is cancelable or not. ‘‘The rationale, of course, is that because implicatures unlike entailments are defeasible, it is possible to assert the contrary, or raise its possibility, without any sense of contradiction.’’ (Levinson 2000: 81). In the examples (35), (36), (37) and (38), for instance, the suspending element ya¯ kahe˜ . . . ‘rather . . .’ reveals similar characteristic of the forms of auxiliary in question: the weaker form hoga¯ is likely to induces an implicature which is defeasible by adding a stronger element, i.e., the form hœ from the scale. The implicature, as in the previous case, is the same, i.e., the speaker does not know that necessarily p. (35) vah cor hoga¯, ya¯ kahe˜ cor he thief Bj-FUT-3-M-SG rather thief ‘He must be or rather is a thief.’

hœ Kj-PRES-3-SG

106 Ghanshyam Sharma (36) vah roz naha¯ta¯ hoga¯, ya¯ kahe˜ he everyday shower-IMPFV-M-SG Bj-FUT-3-M-SG rather naha¯ta¯ hœ shower-IMPFV-M-SG Kj-PRES-3-SG ‘He must be taking, or rather does take a shower everyday.’ (37) vah ab so raha¯ hoga¯, he now sleep PROG-M-SG Bj-FUT-3-M-SG so raha¯ hœ sleep PROG-M-SG Kj-PRES-3-SG ‘He must be, or rather is sleeping now.’ (38) usne a¯m kha¯ya¯ he-OBL-ERG mango-M-SG eat-PFV-M-SG ya¯ kahe˜ kha¯ya¯ hœ rather eat-PFV-M-SG Kj-PRES-3-SG ‘He must have, or rather has eaten a mango.’

ya¯ kahe˜ rather

hoga¯, Bj-FUT-3-M-SG

The meanings of tenses should, therefore, be explained in terms of modality since they alone do not seem to have any independent role to play in the TMA system other than just relating the situation to a deictic center, i.e., speech time. All other meanings attributed to them are obtainable either through their combination with aspectual elements, which are exhibited in the sentence in various morphological devices, or through inherent modal meanings.

5. Conclusion At the outset of this paper we began by providing a heuristic argument to define presumptive modality as a distinct category in the domain of the TMA system. It has been demonstrated that there is a need to treat ‘presumptive’ as a separate category of epistemic modality, rather than to enlist all cases of presumptive modality under a single tense category, namely the ‘future’. For a clear understanding of its exact role in the TMA system, it is important to survey ways in which world languages express it. Some languages exhibit presumptive modality through modals, while others try to encode it morphologically, using various future verb forms. It has been

A Pragmatic Account of the Hindi Presumptive

107

shown that the various usages of the future form of the Hindi verb hona¯ ‘to be’ mostly indicate situations in which actions or events are to hold at a time either prior to or contemporaneous with speech time, but generally not at a time subsequent to speech time, as the term ‘future’ may suggest. It is, therefore, not accurate to classify such usages under the umbrella term ‘future’. If we take a closer look at the tense-aspect system of the Hindi language and consider the role ‘modality’ overtly plays in assigning meaning to its tenses and aspects, we notice that the distribution of the di¤erent tense categories is determined by modality. Modality, thus, turns out to be the sole factor in establishing the scalar meanings overtly carried over by tenses. Following this line of research, it has been demonstrated that the so-called indicatives (IND) tenses sit above all others on the pragmatic scale of tenses, as they carry epistemic necessity. The presumptives (PRESUM) come next and the subjunctives (SUBJ) on the third level on the scale. Roughly put, they must have the following scalar positions: 3IND, PRESUM, SUBJ4. This assignment of scalar position might explain all implicatures these categories may induce which are based on their scalar meaning only. For example, selection of PRESUM by the speaker is likely to implicate that she is not in a position to make an utterance using the strongest element from the scale, namely IND. By making an utterance containing PRESUM, the speaker implicates a generalized conversational implicature that she does not know (i.e FKp), but nonetheless believes to be necessarily true (i.e., Bjp). It has been shown that the implicature induced by the selection of this category is cancelable. The same could be argued along these lines for subjunctives. The paper has argued for the inclusion of modality as the basis for an overall organization of tense-aspect system of a natural language. This might open up new horizons for pragmatic research into the tense-aspect system of a language.

Notes 1. ‘Presumptive’, an adjective, is being used here as a noun, and a technical term. By ‘presumptive’ we mean that grammatical element attached to the main verb which carries modal ‘epistemic necessity’. It is also called ‘deductive’ or ‘assumptive’ (Palmer 2001: 28) The phenomenon referred to by this term in Hindi

108 Ghanshyam Sharma

2.

3. 4. 5. 6.

7. 8.

9.

10.

is classified by some authors as the ‘presumptive tense’, although the tense marker in such cases – the future form of hona¯ ‘to be’ – does not always refer to situations in which actions or events hold at a point in time subsequent to speech time. Some authors, however, rightly classify it as a presumptive marker of modality (Montaut 2004: 128). This is the common idea behind the performative theory of truth as developed by Strawson. According to him, ‘‘to say a statement is true is not to make a statement about a statement, but rather to perform the act of agreeing with, accepting, or endorsing a statement. When one says ‘It’s true that it’s raining,’ one asserts no more than ‘It’s raining.’ The function of [the statement] ‘It’s true that . . .’ is to agree with, accept, or endorse the statement that ‘it’s raining.’ ’’ As generally conceived in model-theoretic semantics. See Palmer (2001) for a detailed discussion on the topic. Or, more elaborately, ‘This is a necessary fact that this is [a picture of ] an elephant.’ The term ‘mood’ has been used in linguistics to refer to di¤erent sentence types (declarative, imperative, interrogative, subjunctive, optative, etc.) which is a part of the wider and more basic term ‘modality’. ‘Modality’, therefore, seems to be an encompassing term with a tilt towards deep ontological questions. We consider that ‘modality’ has to do with modal notions which are the basis of every human utterance, and, therefore, semantics of tense and aspect can be fully understood in terms of modality only. For example, Sharma (1983: 101–110) considers presumptive as a class of subjunctive mood rather than a separate category. In the symbolism developed by Hintikka (1962: 10) ‘‘Ka ’’ nd ‘‘Ba ’’ are formal counterparts of the words ‘‘a knows that’’ and ‘‘a believes that’’, respectively. In epistemic logic, the modal operator j and ‚ are also used as counterparts of the two. However, we believe that to give a full account of the di¤erent epistemic (and alethic modal) states encountered through our heuristic method in section 1, we need to use a combination of the two elements. Thus, through combination four epistemic states can be obtained, namely ‘necessary knowledge’ (Kjp), ‘necessary belief ’ (Bj), ‘possible knowledge’ (K‚p) and ‘possible belief ’ (B‚p). In logic, a distinction is made between alethic and epistemic categories, although they are often expressed in English using the same words. The category in (1c), i.e., K‚p, is anomalous in the sense that it might be both alethic and epistemic: ‘‘It is possible that it will rain in Venice tomorrow because it is possible for rain to take place anywhere anytime’’ or ‘‘For all I know, it is possible for rain to take place tomorrow’’. The first being an alethic category, the second an epistemic. In Reichenbachean formalism, the letters E, S and R are used to indicate Event Time, Speech Time and Reference Time, respectively. Event time is a point on

A Pragmatic Account of the Hindi Presumptive

11. 12. 13.

14. 15.

16.

109

the time line at which the event reported in a sentence takes place, Speech Time is a point at which speaker utters the sentence, and Reference Time is a reference internal to the situation in which an event occurs (for example at 6 o’clock, on Sunday, before he left, etc.). The dash line, ‘—’, stands for the time line. E—S, therefore means that the event time is prior to the speech time. A comma between the letters indicates that the points on the time line represented by capital letters are contemporaneous. For a detailed discussion on the topic, see for example, Reichenbach (1947: 287–298), Comrie (1985), Hornstein (1990), Giorgi and Pianesi (1997: 27). On the topic of aorist tense in Hindi see Davison (2002: 27–57) and Montaut (2004: 103–110). See Comrie (1985: 44) and Dahl (1985: 103) for a description of the distinct epistemological nature of the future tense. Some linguists claim that there are only two tenses in English: non-past (present) and past (indicated by ablaut or ending in -ed). According to them there is no future tense in English – only a modal will expressing future. Others consider will a future marker and add two more tenses: future (obtained through modal will ) and future-in-past (obtained through the past form would ). Comrie (1985) maintains that there are two uses of the future in English: one makes a clear prediction about some future state of a¤airs whose truth can be tested at the future time whereas the second is by nature modal. We believe that, apart from this questionable situation of the future tense in English, the epistemic foundation of the future tense in general is totally di¤erent from that of past and non-past tenses which can be understood in terms of modality only. Many languages, such as Hindi/Urdu and Italian, make use of the future to express this modality. See also Palmer (2001: 104). The term coined by Grice in the fifties (see 1989), and further elaborated and modified by, among others, Gazdar (1979), Horn (1989) and Levinson (2000), to refer to that part of meaning which, roughly put, is ‘implicated’ by the utterer through ‘what he has said’. Giorgi and Pianesi (1999: 29) provides the following inventory of the tenses: Present Past Future Present Perfect Future perfect Past Perfect Future in past Proximate future Distant future

: : : : : : : : :

(S,R) (R—S) (S—R) (S,R) (S—R) (R—S) (R—S) (S,R) (S—R)












(R,E) (E,R) (R,E) (E—R) (E—R) (E—R) (R—E) (R—E) (R—E)

S,R,E E,R—S S—R,E E—S,R E—R—S S,R—E S—R—E

17. For the sake of simplicity, we will be calling this form ‘the future form’ throughout this paper. However, as it only very rarely refers to a future point

110 Ghanshyam Sharma in time, it should be considered as a marker of presumptive modality rather than a ‘future form’. 18. For instance, the combinations of the Hindi auxiliary hona¯ ‘to be’ with imperfective, continuous and perfective aspects are classified by MacGregor (1977: 29) as imperfective, continuous and perfective future, respectively. 19. We have tried to gloss only those items which are essential to explain the point in discussion. In providing glosses, the following abbreviations are used: ?? ¼ pragmatically anomalous; 1 ¼ first person; 3 ¼ third person, CMPL ¼ completive; CONT ¼ continuative; DUR ¼ durative; EMP ¼ emphatic; F ¼ feminine; FREQ ¼ frequentative; FUT ¼ future; IMPFV ¼imperfective; INCE ¼ inceptive; IND ¼ indicative; M ¼ masculine; OBL ¼oblique; PASV ¼ passive; PFV ¼ perfective; PRES ¼ present; PRESUM¼ presumptive; PROG ¼ progressive; SUBJ ¼ subjunctive 20. Despite the widely-held view about the necessity to make a distinction between grammatical aspect and lexical aspect (Aktionsart), ‘‘recently, as scholars have come to appreciate the inter-relation between viewpoint and situation structure, use of the term has broadened to include the temporal properties of the situations themselves, internal events or Aktionsart.’’ (Smith 1991: 3) Consequently, some authors on Hindi grammar have made proposals to include telic in the class of grammatical aspects: ‘‘. . . in addition to the two traditionally recognized aspect categories, the perfective and the imperfective, a third category of grammatical aspect, the telic, must be distinguished on distributional grounds.’’ (Agha 1998: 126). 21. Comrie (1976: 25) classifies aspect in the following manner: (A) Perfective (B) Imperfective (B1) Habitual (B2) Continuous (B2a) Non-progressive (B2b) Progressive 22. Masica (1991: 292), Shapiro (1989: 55), Kachru (2004: 146). 23. It is possible to find di¤erences between ‘progressive’ and ‘continuous’ in some languages – the first refers to a dynamic quality of action, whereas the second to a state of the agent. However, despite some minor di¤erences between their aspectual meanings, the two terms have been used interchangeably in Hindi grammatical tradition to refer to the same aspect – some preferring ‘continuous’ (Masica and McGregor), others ‘progressive’ (Kachru, Montaut and Shapiro). Some authors (e.g., Kellogg 1965: 261), however, prefer to consider this construction under compound verbs.

A Pragmatic Account of the Hindi Presumptive

111

24. Kachru predicts that it ‘‘. . . occurs only in its imperfect participle form . . .’’ (2004: 154). The fact is that, being itself an imperfective tense, it cannot have a perfective marker. Consider the following example in which it does not have an imperfective participle: 25. kal se vah roz mandir ja¯ya¯ tomorrow from he everyday temple go-PFV-M-SG karega¯ FREQ-FUT-M-SG ‘From tomorrow he will be visiting the temple everyday.’ 26. Some authors prefer to describe the phenomenon slightly di¤erently. According to them the perfective participle in Hindi is obtained by adding a zero morpheme rather than an -a¯. See for example Masica (1991: 293). 27. A Hindi sentence having the perfective aspect and the future auxiliary, like the example in (11), is mistakenly considered to have the same meaning as in an English sentence such as ‘‘Ramu will have gone to Venice tomorrow’’. To get a sentence in Hindi which could have the similar meaning to that of the English future perfect, it is necessary to use a completive aspect, as can be seen in the following example. ra¯mu¯ tab tak venis ja¯ cuka¯ Ramu then by Venice go CMPL-PFV-M-S ‘Ramu will have gone to Venice by then.’

hoga¯ FUT-M-S

28. Other terms to indicate the verb with the same characteristics are anti-transitive and original passive. 29. We do not enter into the details of the Square of Opposition as it has been widely discussed in logic for over two millennia. Its use in describing scalar implicature is due to work by Horn (1989). For an excellent discussion on the topic see Levinson (2000: 64–72). 30. The corners on the Square of opposition are presented in the literature in the following manner (Horn 2004): A ¼ jp $ F‚ Fp I ¼ ‚p $ Fj Fp E ¼ F‚p $ j Fp O ¼ Fjp $ ‚p

Ex(Fx ! Gx) bx(Fx & Gx) Ex(Fx ! FGx) bx(Fx & FGx)

E.g. All/every F is G E.g. Some F is/are G E.g. No F is G E.g. Not every F is G, Some F is not G

31. Although ho is the only possible auxiliary to be utilized to express epistemic possibility in Hindi, due to some noticeable influence of the English language into the use of the Hindi subjunctive, it is possible to encounter the hœ form, especially in the Hindi register of speakers who are also fluent in English.

112 Ghanshyam Sharma References Agha, Asif 1998

Form and function in Urdu-Hindi verb inflection. The Yearbook of South Asian Languages and Linguistics: 105–133. Comrie, Bernard 1976 Aspect. Cambridge: Cambridge University Press. 1985 Tense. Cambridge: Cambridge University Press. ¨ sten Dahl, O 1985 Tense and Aspect System. Oxford: Basil Blackwell. Davison, Alice 2002 Agreement features and projections of TENSE and ASPECT. In The Yearbook of South Asian Languages and Linguistics, Rajendra Singh (ed.), 27–57. New Delhi/Thousand Oaks/London: Sage Publications. Gazdar, Gerald 1979 Pragmatics: Implicature, Presupposition and Logical Form. New York: Academic Press. Giorgi, Alessandra, and Fabio Pianesi 1997 Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Grice, H. Paul 1989 Studies in the Way of Words. Cambridge, MA: Harvard University Press. Hintikka, Jakko 1962 Knowledge and Belief. Ithaca, NY: Cornell University Press. Hirschberg, Julia 1991 A Theory of Scalar Implicature. New York: Garland Publishing, Inc. Horn, Laurence R. 1989 A Natural History of Negation. Chicago: The University of Chicago Press. 2004 Implicature. In The Handbook of Pragmatics, Laurence R. Horn and Gregory Ward (eds.), Oxford: Blackwell Publishing. Hornstein, Norbert 1990 As Time goes by. Cambridge, MA: The MIT Press. Kachru, Yamuna 2006 Hindi. Amsterdam: John Benjamins Publishing Company. Kellogg, S. H. 1965 Reprint. A Grammar of the Hindi Language. London: Routledge & Kegan Paul Ltd. Original edition, 1875, 2 nd edition, 1893. Masica, Colin 1991 The Indo-Aryan Languages. Cambridge: Cambridge University Press.

A Pragmatic Account of the Hindi Presumptive

113

McGregor, R. Stuart 1995 Reprint. Outline of Hindi Grammar. 3 rd ed. Oxford: Oxford University Press. Original edition, 1972. Montaut, Annie 2004 A Grammar of Hindi. Munich: Lincom Europa. Olsen, Mari Broman 1999 A Semantic and Pragmatic Model of Lexical and Grammatical Aspect. New York: Garland Publishing, Inc. Palmer, Frank R. 2001 Mood and Modality. 2 nd ed. Cambridge: Cambridge University Press. Reichenbach, Hans 1980 Reprint. Elements of symbolic logic. New York: Dover Publications. Original edition, London: Macmillan, 1947. Shapiro, Michael C. 1989 A Primer of Modern Standard Hindi. Delhi: Motilal Banarsidass. Sharma, Aryendra 1983 Reprint. A Basic Grammar of Modern Hindi. 4 th ed. New Delhi: Central Hindi Directorate, Government of India. Original Edition, 1958. Sharma, Ghanshyam 2000 On the modal meanings of the subjunctive in Hindi. The Yearbook of South Asian Languages and Linguistics: 177–198. Smith, Carlotta S. 1991 The Parameter of Aspect. Dordrecht/Boston/London: Kluwer Academic Publishers. van der Auwera, Johan 1996 Modality: the three-layered scalar square. Journal of Semantics 13: 181–195. van der Auwera, Johan, and Vladimir Plungian 1998 On modality’s semantic map. Linguistic Typology 2: 79–124.

Special Contributions: The Syntactic Typology of Dravidian: A Debate

What’s So Subversive about Dravidian? Revisiting Finiteness in Dravidian Syntax

Sanford B. Steever

Two analyses of finiteness in Dravidian syntax are compared, one based on constituent structure and dominance relations, the other on constructions and precedence relations. Syntactic evidence reveals that the latter approach, the Strict OV Restriction, fails to provide an adequate basis for either the descriptive analysis or the historical development of finite structures in the Dravidian languages. It is rejected in favor of the former analysis, one grounded in Dravidian structures, processes and diachronic trajectories. The failure of the Strict OV Restriction to explain Dravidian syntactic phenomena disqualifies it as an argument for or against grammatical convergence in the South Asian linguistic area.

1. Introduction Early generative grammar penetrated the analysis of South Asian languages, particularly the Dravidian family, only slowly and with mixed results. One reason is that analogues to the general transformations in such languages as English which had helped to win converts could be found only with di‰culty or not at all. One instance is the rule S0 $ S0 [S1 and S2 ]S0 : it permits the factoring of ‘Sam runs and Julie dances’ into the two well-formed sentences ‘Sam runs’ and ‘Julie dances’; conversely, it licenses the combination of ‘Sam runs,’ ‘and’ and ‘Julie dances’ into a well-formed complex sentence. Students of the Dravidian languages know how rare correlates to such common English phenomena are; where found, they prove to be marked or unnatural within the context of Dravidian grammar. Brute-force application of this conjunction rule to Tamil yields unpromising results. On the one hand, factoring the complex sentence ra
ttiri po
y pakal vantatu ‘night departed and day came’ into its two parts, *ra
ttiri po
y *‘night departing’ and pakal va-nt-atu ‘day came’, gives rise to at least one ill-

118 Sanford B. Steever formed sentence. On the other hand, using the closest thing to a conjunction, the clitic ¼um ‘all, every’, to combine two well-formed simple sentences, ra
ttiri po
n atu ‘night departed’ and pakal vantatu ‘day came’, into *ra
ttiri po
n at¼um pakal vantatu ‘night departed and day came’, also results in an ungrammatical structure. Also, though a grammarian might detect functional similarities between Tamil of en ru ‘saying’ and English ‘that’ in matrix-complement structures (see below), he or she will find no formal parallel: ‘that’ is subordinating conjunction, en ru a nonfinite verb form. The analysis of complex sentences in Dravidian during the early phase of generative grammar produced nothing to rival Bloch’s (1946) earlier work. To close that gap I spent a year in India, studying the interaction of morphology and syntax in complex sentences, and published a monograph on a set of specific constructions with multiple finite predicates (Steever 1988), the Serial Verb Formation (SVF). The introduction to that monograph sketched the general architecture of Dravidian syntax with special attention paid to where finite and nonfinite predicates could occur. Their distribution was specified by a rule that limited the number and position of finite predicates in the sentence over against their more numerous nonfinite counterparts. By virtue of its occurrence throughout the family, that rule was projected to Proto-Dravidian (PDr). It laid the foundations for the monograph’s central topic, the SVF, a family of constructions with multiply congruent finite verbs. In a series of publications beginning in 1987, Hans H. Hock (hereafter HHH) undertook a two-part critique my work on finiteness in Dravidian. In the first part, he questioned whether my rule provided an adequate analysis of finiteness, suggesting that it did not. In the second, he introduced an alternative to fill the void which he believed was opened by the first part: the Strict OV Restriction, or SOVR. In his program, the SOVR serves two purposes: as a description of Dravidian syntax and as an argument against the Dravidian Subversion Hypothesis, or DSH (Hock 2001). HHH characterized the DSH as the claim that when the Indo-Aryan languages came into contact with the Dravidian languages in South Asia, they were influenced by and gradually evolved to resemble their Dravidian neighbors. Finding this model of convergence unpersuasive, HHH took an unusual step to disprove it: he argued that syntactic structures in early Dravidian and early Indo-Aryan were both governed by the SOVR, and di¤ered so little from each other that Indo-Aryan grammar could not have been influenced by Dravidian. Acceptance of the SOVR also promised to confer

What’s So Subversive about Dravidian?

119

collateral benefits: it was supposed to characterize finiteness not only in the Dravidian and Indo-Aryan languages, but in others as far-flung as Burushaski, Luisen˜o and Quechua, bringing them all under a single typology. Why HHH co-opted my work for his critique of the DSH is unclear. As a study of historical Dravidian morphosyntax, Steever (1988) dealt with structures and processes internal to that family, taking no position on whether they exerted influence on other languages. Indeed, that monograph suggested that contact with Indo-Aryan languages may have prompted certain changes in the North Dravidian subgroup of languages, accelerating its divergence from a common Dravidian template.1 HHH’s appropriation of my work in the campaign against the DSH invites scrutiny of his critique of, and his alternative to, my proposal to determine whether the former is logically justified and the latter descriptively adequate. Close examination shows that his program fares poorly on both counts. The SOVR fails to do justice to Dravidian structures, makes false predictions about their grammatical analysis and dismantles obvious generalizations about their syntax. This essay identifies serious flaws in his model which disqualify it as a description of finiteness in Dravidian syntax and, consequently, as a premise in the argument against the DSH. The study of the shortcomings in his program brings to light the requirements for an adequate theory of finiteness in Dravidian, and shows that those requirements are already met by the analysis of finiteness in Steever (1988), but not by the SOVR. Section 2 summarizes my original analysis, section 3 reviews and rebuts HHH’s criticism of that work, section 4 critically examines the SOVR as an analysis of Dravidian syntax and section 5 reflects on the typology underlying the SOVR.

2. Finiteness in the Dravidian sentence Steever (1988) analyzes the role that finiteness plays in the Dravidian languages, primarily to analyze what is called the Serial Verb Formation. Finiteness is posited as a syntactic property of the sentence that is interpreted morphologically: certain morphological forms serve to mark the end of a sentence, helping thereby to delineate its constituent structure. Dravidian morphology distinguishes primarily between words and clitic particles (identified here by the boundary marker ¼). Two primary parts of speech, noun and verb are distinguished; beyond these two, consensus breaks down.

120 Sanford B. Steever Dravidian languages generally lack such parts of speech as adverbs, complementizers and conjunctions;2 furthermore, the grammatical functions traditionally assigned to them generally fall to verbs in Dravidian. As against such languages as English or French, the distribution of finite predicates in Dravidian languages is highly restricted, though not as restricted as Hock’s (2001) Strict OV Restriction would lead us to believe (see below). In my original monograph, the rule governing the distribution of finite and nonfinite predicates is approached in four steps, culminating in rule (103) in Steever (1988:111–12), reproduced here as rule (1). The phenomena that simultaneously motivate and exemplify each clause of this rule are presented and discussed here one after the other; for a fuller exposition, consult Steever (1987a, 1988).3 (1)

a. Each Dravidian sentence has n þ 1 functionally finite predicates, where n stands for the number of finite predicate embedders (FPEs) and 1 for the Root Domain of Finite Predicates (RDFP). b. The functionally finite predicate in the RDFP occurs highest and rightmost in the root clause, c-commands all other predicates and is itself c-commanded by none. Each FPE induces a dependent domain of finite predicates (DDFP). The functionally finite predicate in each DDFP occurs highest and rightmost in the domain defined by the FPE, c-commands all other predicates in that domain and is itself c-commanded by none. c. Each functionally finite predicate may be accompanied by indefinitely many formally finite verbs, if it belongs to the set of expressions that govern the serial verb formation (SVF). d. Each formally finite verb occurs in a construction, the head of which is a functionally finite verb. Its morphological interpretation is assigned by that functionally finite verb; it incorporates tense, mood or negation, and a personal ending congruent with the functionally finite verb’s. e. All other predicates are nonfinite, functionally and formally.

In (1a) finiteness is taken to range over the sentence, which in Dravidian may be simple or complex. A simple sentence typically consists of a subject and a predicate. The subject may appear in the nominative case or, in specified instances, the dative; the predicate may be a verb or a predicate nominal. All four implied combinations of subject and predicate occur

What’s So Subversive about Dravidian?

121

throughout Dravidian. The examples below are from Modern Tamil; their Ancient Tamil counterparts may be found in Steever (2004:1034–36). Restrictions of space compel me to refer readers to Steever (1987a, 1988) for examples from the other languages and subgroups. (2)

a. avan that.man-nom ‘He came.’

va.n-ta
n . come-pst-3sm

b. avan nalla man itan . that.man-nom good man-nom ‘He is a good man.’ c. avan -ukku can kı
tam pitikkum. that.man-dat music-nom like-fut-3sn ‘He likes music.’ d. avan -ukku oru that.man-dat one ‘He has a son.’

makan . son-nom

All sentences in (2) conform to rule (1). Since there is but one predicate in each sentence, there is exactly one finite predicate under (1a). To test the rule’s scope, therefore, we must examine complex sentences. Complex sentences consist of two or more clauses, each with its own predicate. As Dravidian scholars have observed, nonfinite verb forms figure prominently in the formation of complex sentences: the infinitive occurs in (3a–c), the conjunctive in (4a–d), the conditional in (5) and the adnominal in (6a–c). Note that these forms appear in a variety of di¤erent constructions: matrixcomplement structures (3b, 3c, 4c, 4d), coordinate structures (4a, 4b), relative clauses (6a) and adverbial clauses (3a, 6c), among others. Note also most nonfinite forms are not dedicated to a single, specified context or construction. (3)

a.

S0 [S1 [kamala


b.

S0 [na
n

pa
t-a] S2 [caro
ja
Kamala-nom sing-inf Saroja-nom ‘As Kamala sang, Saroja danced.’ S1 [avan ai

I-nom he-acc ‘I told him to come.’

var-a.c]S1 come-inf

a
t-in -a
l ]S2 ]S0 . dance-pst-3sf

con -n -e
n ]S0 . tell-pst-1s

122 Sanford B. Steever

(4)

c.

S0 [S1 [aval

pa
t-a]S1 she-nom sing-inf ‘I heard her singing.’

a.

S0 [S1 [ra
ttiri

b.

S0 [S1 [maz ai

c.

S0 [S1 [na
n

na
n ke
t-t-e
n ]S0 . I-nom hear-pst-1s

po
-y]S1 S2 [ pakal va-nt-atu]S2 ]S0 night-nom go-cf daylight-nom come-pst-3sn ‘Night departed and daylight came.’

pey-tu]S1 S2 [veyil ati.t-tu]S2 rain-nom rain-cf sunshine-nom beat-cf to
n r-i.y-atu]]S0 S3 [va
n avil rainbow-nom appear-pst-3sn ‘It rained, the sun shone, and a rainbow appeared.’ col.l-i ]S1 aval ke
t.k-a.v I-nom say-cf she-nom listen-inf ‘She didn’t listen to what I was saying.’

illai ]S0 . ind-neg

d. na
n avalai ka
r o
tt-i.p pa
r-tt-e
n . I-nom she-acc car drive-cf see-pst-1s ‘I saw her driving a car.’ (5)

S0 [S1 [makan

(6)

a.

poy co.n -n .a
l ]S1 amma
ati.p-p-a
l ]S0 . son-nom lie-nom tell-cnd mother-nom beat-fut-3f ‘If a son lies, his mother will beat him.’ S0 [NP [S1 [u

ne
rru va.n-t-a]S1 man itan -ai.p] na
n yesterday come-pst-adn man-acc I-nom

pa
r.t-t-e
n ]S0 . see-pst-1s ‘I saw the man who came yesterday.’ b.

S0 [NP [S1 [na
n

u po
-kir-a]S1 go-prs-adn

u
r] town-nom

I-nom ta
n ]S0 . indeed ‘The town that I am going to is Madurai.’ c.

S0 [N [S1 [avan

va.n-t-a]S1 po
tu] na
n he-nom come-pst-adn time I-nom ‘I was reading at the time/when he came.’

maturai Madurai-nom

pati.t-t-e
n ]S0 . read-pst-1s

What’s So Subversive about Dravidian?

123

Inspection of these sentences reveals that at this stage in the exposition, complex sentences have exactly one finite predicate, with all other predicates being nonfinite verbs. That lone finite predicate occurs highest (and rightmost) in the syntactic structure of the sentence, c-commands all other predicates and is itself c-commanded by none. This is called the Root Domain of Finite Predicates, or RDFP. This data set is adequately covered by clauses (1a and e). One crucial observation needs to be stressed at this juncture: predicate nominals, as in (2d and b) and (6b), exhibit exactly the same distribution as finite verbs (Steever 1988:2–3). Although nonfinite forms prove quite versatile in complex sentences, the grammatical fragment compiled so far contains several functional gaps. From what we have seen, it permits the formation of indirect discourse, as in (3b), but not direct discourse (see Steever 2005:153–59). Further, certain verbs, such as the Tamil permissive varala
m ‘(one) may come’, lack nonfinite forms and so cannot appear outside the RDFP. Moreover, relying as it does on nonfinite verbs to form complex sentences, this fragment also prohibits predicate nominals from appearing outside the RDFP, because as nouns they lack the requisite (nonfinite) verb morphology (Steever 1988:2– 3). The Dravidian languages provide two sets of devices to overcome these and similar functional obstacles in forming complex sentences. First, certain verbs may embed finite predicates, noun or verb; the two most common across the family are reflexes of *en - ‘say’ (DEDR 868) and *a
ka- ‘become’ (DEDR 333). Others, such as Tamil po
la ‘resemble’, also occur in this capacity but are not universal. The term ‘‘embed’’ and its derivatives, as used here, equally describe one clause that is joined to another as a conjunct in a coordinate structure, and one that is subordinated to a main clause in a matrix-complement structure.4 Observe that in (7a) the conjunctive of en - ‘say’ embeds a subordinate sentence with the predicate nominal nallavan ‘good man’ under a verb of propositional attitude in the main clause; in (7b) the conditional of a
ka- ‘become’ embeds the finite verb vanta
n ‘he came’ in the protasis of a conditional proposition; and in (7c) the infinitive of po
la ‘resemble’ embeds the finite verb varum ‘it will come’. Their Ancient Tamil counterparts in (8a–c) confirm that in the earliest stages of the language these three already possessed the capacity to embed finite predicates. (7)

a.

S0 [na
n

S1 [avan

nallavan ]S1 I-nom he-nom good.man-nom ‘I think that he is a good man.’

en ru say-cf

nin aikkire
n ]S0 . think-prs-1s

124 Sanford B. Steever

(8)

b.

S0 [S1 [avan

vanta
n ]S1 a
n a
l na
n he-nom come-pst-3sm become-cnd I-nom pa
rkkavillai ]S0 . see-ind-neg ‘Although he came, I didn’t see him.’

avan ai.p he-acc

c.

S0 [S1 [mazai

varum]S1 po
la.t teriyum]S0 . rain-nom come-fut-3sn resemble-inf be.clear-prs-3sn ‘It looks as if it will rain.’

a.

makan ya
n t(u) ula
n ¼o
]S1 en a you-gen son-nom where be-npst-3sm¼int say-inf vin avuti ]S0 ask-npst-2s ‘You ask, ‘‘Where is your son?’’ ’

b.

S0 [S1 [ayar

emar]S1 a
n a
l aytti.y-e
m cowherd-nom our.kin-nom become-cnd cowherdþf-1p ya
m]S0 . we-1p-inc-nom ‘If (it is the case that) our kin are cowherds, then we are female cowherds.’

c.

S0 [S1 [ival

S0 [S1 [nin

an an ku urran ai ]S1 po
ri ]S0 . she-gen a¿iction su¤er-npst-2s resemble-npst-2s ‘You appear as if you are su¤ering her a¿iction.’

The exact inventory of Finite Predicate Embedding Verbs (FPEVs) varies from language to language (see Steever 1988:25–29), permitting new members to enter into this class on occasion, but reflexes of *en - ‘say’ and *a
ka‘become’ functioning as FPEVs are remarkably robust and stable throughout the twenty languages of the South (SDr), South-Central (SCDr) and Central Dravidian (CDr) subgroups. Each FPEV induces a Dependent Domain of Finite Predicates, or DDFP, which hosts a single functionally finite predicate. This leads to the specification in clauses (1a and b): the number of finite predicates in the Dravidian sentence equals n þ 1, where n represents the number of instances of FPEVs and 1 represents the verb in the RDFP. Each FPEV induces a DDFP which hosts a single finite predicate. All other predicates – all of them morphologically verbs – are nonfinite, under (1e).

What’s So Subversive about Dravidian?

125

However, verbs are not the only operators that may embed finite predicates; certain postclitic particles may do so as well. Two of the most common are ¼e
‘even, and’ and ¼o
‘or, any, whether’. In (9a) the clitic ¼o
joins a subordinate clause to a main clause in a dubitative construction; in (9b) the clitic ¼e
subordinates a sentence to an NP in an upper clause, forming a relative-correlative clause.5 (9)

pa
tukira
l ]S1 ¼o
en akku.t teriyavillai ]S0 . she-nom sing-prs-3sf¼or I-dat know-ind-neg ‘I don’t know whether she is singing.’

a.

S0 [S1 [aval

b.

S0 [S1 [ne
rru

vanta
n ]S1 ¼e
avan -ai yesterday come-pst-3sm¼even he-acc ‘I saw him who arrived yesterday.’

na
n pa
rtte
n ]S0 . I-nom see-pst 1s

Finite Predicate Embedding Clitics, or FPECs, occur in a broad variety of constructions, including conditionals (Kolami ¼te), dubitatives, indirect questions, relative-correlatives clauses and reported speech (Modern Spoken Tamil ¼nu). Generalizing FPEVs and FPECs as members of a single set of Finite Predicate Embedders (FPEs) leads to the final formulation for (1a, b and e). The number of finite predicates in the Dravidian sentence equals n þ 1, where n equals the number of occurrences of FPEs and 1 represents the predicate that appears in the RDFP. Each FPE induces a DDFP which hosts one functionally finite predicate. All other predicates are nonfinite verbs.6

2.1. Serial verb formations Even with the distribution of finite and nonfinite predicates as stipulated in (1a, b and e), the Dravidian languages present further examples of finite verb forms where they would otherwise not be expected to occur: the Serial Verb Formation, or SVF. Since this is the subject of an entire monograph (Steever 1988), only a brief summary is provided here. In SVFs the personal endings of all the verbs in the construction are congruent and vary directly with the verb that occupies the local domain of finite predicates (DFP), that is, with a predicate that is designated finite under clauses (1a and b). This discovery led me to propose the crucial, systematic di¤erentiation between ‘‘functionally finite’’ predicates, those that stand in DFPs, and ‘‘formally finite’’ predicates, those that do not.7

126 Sanford B. Steever Example (10a), from Ancient Tamil (SDr), illustrates a periphrastic negative verb (see Steever 2004:1030) in which two finite verb forms – the first formally finite, the second functionally finite – combine to make a single auxiliary compound verb (Steever 2005:39–40). Note that the endings on both are congruent. Example (10b), from Pengo (SCDr), exemplifies a balance compound verb (Steever 1988:41, pass.) in which two finite verbs combine to make a single predication. Note again the congruence between the endings on both forms. Example (10c), from Konekor Gadaba (CDr), presents a lexical compound verb (Steever 1988:6) in which two synonymous finite verbs combine to signal an intensive compound verb. As before, their tense markings and personal endings are congruent. Finally, example (10d), from Kurux (NDr), illustrates another intensive compound verb in which the endings of both finite forms coincide. In all four examples, the grammatical endings of the first (formally finite) verb are determined by the endings of the second (functionally finite) verb, which stands in either an RDFP or a DDFP. In e¤ect, SVFs use repeated finite verb inflections to mark concord among the members of the compound, thereby delineating its boundaries. (10) a. celve
m alle
m. go-fut-1p become-neg-1p ‘We will not go.’ b. narpur eba ke ju
t-n-ah]S de S [rat-n-ah world.of.men there prt drop-fut-1p descend-fut-1p prt icar. say-pst-3p ‘‘They said they would take it down to the world of men.’’ c. mu
den su
d i nette ved da udla
kuda. rabbit-acc see-cf dog-nom run-prs-3sn chase-prs-3sn ‘On seeing the rabbit, the dog chases after it.’ d. ra
j narainas indira
ga
ndhin haraba
cas Raj Narayan-nom Indira Gandhi-acc defeat-pst-3s ciccas. give-pst-3s ‘Raj Narayan soundly defeated Indira Gandhi.’ Incorporating these and similar data into rule (1) led me to propose (1c and d) and place them after the first two clauses. The clauses in (1) are structured in a deliberate order: the first two, (1a and b), collectively define functional

What’s So Subversive about Dravidian?

127

finiteness and precede the next two, (1c and d), which together define formal finiteness. This order secures the observation that within Dravidian, formal finiteness presupposes functional finiteness. The rule also moves from the marked to the unmarked case: (1a) stipulates that the Dravidian sentence must have at least one finite predicate while (1e) requires that all other predicates, subject only to the exceptions in (1b–d), be nonfinite, viz. nonfinite verbs. The relative markedness of marked finite and unmarked nonfinite forms thus gives explicit shape to the pretheoretical observation that, all things being equal, nonfinite forms are more numerous and have a broader distribution within the Dravidian sentence than their finite counterparts.

2.3. Historical changes in the expression of finiteness Based on the distribution of the various sentence types and operators throughout the four subgroups, rule (1) was reconstructed to PDr. As outlined in Steever (1988), certain diachronic changes overtook the daughter languages and subgroups, leading to a di¤erentiation in how they express the rule. In Malayalam (SDr), for example, the distinction between functionally and formally finite verbs has been neutralized so that functionally finite predicates no longer mark person, number or gender. Some languages, such as Konda and Kurux, use a process of A‰x-Truncation (Steever 1988:71, 101; Steever 1993:111–17) to erode the personal endings on certain formally finite verbs, thereby easing the structural tension implicit between functional and formal finiteness. By and large, both SDr and SCDr, which include the majority of languages, retain a robust system of FPEs and SVFs, strongly suggesting the reconstruction of (1) to the earliest stages of Dravidian.8 The only noteworthy variation concerns the exact list of FPEs for each daughter language. CDr also preserves the core content of (1); although that subgroup preserves fewer FPECs (but note the Kolami clitic ¼na ‘because’ in [11]) than SDr or SCDr, it still preserves a vibrant system of inherited FPEVs and SVFs. (11)

ma
ri S1 [ pod uttin]S1 ¼na tamme they-nom then sun-nom set-pst-3s¼because self-gen beknet pa
v o
leng la
gtar]S0 . mother-nom path wait-inf begin-pst-3pm ‘Then they, because the sun had set, began to wait for their mother on the path.’

S0 [avr

128 Sanford B. Steever The NDr languages show two developments that impact rule (1). First, so many predicates have migrated into the class of FPEVs that nonfinite verb forms are correspondingly more restricted in the formation of complex sentences. Even so, within a language like Malto (Steever 1998:36, 383) vestiges of the earlier system are seen with FPEs that originate from Dravidian sources. The second development concerns the fate of FPECs in NDr. Under the impact of neighboring Indo-Aryan and Iranian languages, NDr languages have borrowed such words as aur ‘and’, hole ‘if ’, je ‘that’ ki ‘or, whether, that’ and nole ‘if not’ from Bangla, Hindi and Oriya. These function words, including conjunctions and complementizers, subcategorize finite complements and have e¤ectively superseded the set of inherited Dravidian clitics.9 What results is a system of complementation that looks remarkably like English sentential complementation (but without relative pronouns). Even so, a virtual counterpart of rule (1) remains intact in the grammar of these languages because NDr still retains, first, a distinction between finite and nonfinite verbs and, second, a well-defined set of SVFs. Since SVFs occur in all four subgroups and since they are defined by (1c and d), they presuppose functional finiteness defined by (1a and b); accordingly, rule (1) is projected to the protolanguage in its entirety.

3. Hock’s critique The first part of HHH’s program questions whether rule (1) accurately describes and explains the distribution of finite and nonfinite predicates in the Dravidian sentence; the second part introduces his alternative, the Strict OV Restriction, in rule (19) below. These two, interrelated parts are reviewed in sections 3 and 4, respectively. Both tasks are compromised by a lack of firsthand knowledge of the Dravidian languages, their grammars and the specialist literature. On these grounds, Steever (1993:63) voiced concern over Hock’s (1987, 1988) critique and alternative; it seemed axiomatic that particularly in historical linguistics, there can be no substitute for a direct knowledge of the languages involved and their grammatical systems. Nearly fifteen years after introducing the SOVR and not addressing these gaps, Hock (2001:93) replied that what matters is the linguistic evidence. Such a response skirts an important issue: what counts in linguistic analysis is the accurate description and interpretation of the language evidence. Linguists

What’s So Subversive about Dravidian?

129

interpret data by referring them to the grammatical system in which they originate and from which they receive an analysis. Without a practical grasp of their grammars, HHH is not equipped to interpret Dravidian data sui generis but must look elsewhere instead, often to English, Sanskrit or an abstract typology, for a grammatical interpretation. As the discussion below shows, his program on finiteness does not locate or interpret Dravidian syntactic evidence within the matrix of Dravidian grammar. Because neither his critique of my original proposal nor his alternative to it is firmly grounded in the structures and rules of Dravidian grammar, the overall program lacks substance and persuasiveness.

3.1. Word order In a brief overview of Dravidian languages, Steever (1988:7) remarked in passing that Dravidian syntax may be characterized as rigidly SOV. Hock (2001:66–67) interpreted this to mean that the word order of major constituents was crucial to my analysis of finiteness; elsewhere (Hock 1988:217), he chided me for not producing a treatise on word-order phenomena, and then took the position that any deviation from a strict sequence of SubjectObject-Verb somehow invalidates rule (1).10 It does not. First, the existence of surface permutations in word order does not automatically invalidate the claim that the unmarked word order in Dravidian is SOV. Second and more importantly, the claim that a language belongs to the strict SOV type is not primarily about word order: in work on implicational universals (McCawley 1982:159–75, Comrie 1988), it pertains to the correlations that hold between the order of major constituents in a clause and other dimensions of linguistic structure. A strict SOV language, for example, exhibits not just a dominant Subject-Object-Verb word order, it also has postpositions not prepositions; main verbs precede auxiliaries; complement clauses precede matrix clauses; relative clauses precede their heads; genitives precede their head nouns; syntactic structures branch to the left, not the right; Gapping takes place to the left; etc.11 The underlying generalization is that all these dimensions exhibit the order of operand-operator throughout; at this level of abstraction, the Dravidian languages are robustly SOV.12 And because the opposition of finite versus nonfinite is not of the same order as the distinction between operator and operand, finiteness cannot automatically be considered a parameter on par with the other

130 Sanford B. Steever dimensions of language structure conventionally correlated with word order. In empirical terms, though Tamil does allow some freedom of permutation in the surface order of constituents, it observes all the harmonic correlations noted above. Even Ancient Tamil (Steever 2004:1034) confirms the existence of these correlations in the earliest attested Dravidian language despite obvious word-order variations in the poetic corpus. Two points bear keeping in mind: first, nothing in my analysis depends on correlating word order with where finite predicates occur; second, HHH o¤ers no justification that it should do so. But by insisting on such a correlation, he ascribes to precedence relations an importance absent from Steever (1988): rule (1) specifically utilizes the language of dominance relations (see McCawley 1988:39), such as ‘‘highest’’ and ‘‘c-command.’’ It should be stressed at this juncture that all mention of linear order could be removed from rule (1) with no loss of descriptive power, leaving a rule that stipulates just dominance relations. To recover the precedence relations implied by these tree structures, the hierarchical constituent structure is simply ‘‘linearized’’ through the application of the independent word-order conventions of Tamil to the abstract tree structures. For the time being, the term ‘‘rightmost’’ is retained in (1b) as a heuristic for readers to locate finite predicates in strings of words. Thus departures from strict SOV word order do not constitute counterevidence to rule (1). Finally and related more to rhetoric than to linguistic arguments, insisting on a spurious correlation between finiteness and word order allows HHH to take the stray mention of strict SOV word order in Dravidian and graft on to it his Strict OV Restriction, thereby giving the impression that the SOVR is somehow the continuation of an established scholarly tradition rather than something of his own invention (see section 3.5).

3.3. Extraposition Further in this vein, Hock (1988:217, 1989:402) holds that the existence of Extraposition, broadly conceived, compromises my theory of finite predicates, once again supposing that linearity and word order are essential to the analysis. They are not. Steever (1988:16–17) presented data that nonfinite verbs may be moved rightwards over finite verbs under specified conditions without compromising the analysis of where finite predicates may occur. Indeed, such examples were explicitly intended to show that

What’s So Subversive about Dravidian?

131

dominance relations prevail over – and ultimately dispense with the need for – precedence relations in the formulation of rule (1).

3.4. Adjacency Focused still on linear order, Hock (1988:217; 1989:403) opines that adjacency is crucial to my analysis of finiteness, insinuating that the existence of constructions with nonadjacent finite predicates must compromise it. But as examples in Steever (1988) already show, adjacency is not central to my analysis or to the formulation of (1), which requires only c-command. Such a minor issue, were it germane, could easily be resolved through the deft placement of nonessential variables in the relevant rules and structures.13 Adjacency is thus but another red herring in the critique of (1); however, as seen below in section 4.2, it proves to be a genuine problem for HHH’s alternative. It is also curious that when promoting that alternative HHH has nothing to say about how it correlates finiteness with linear word order, Extraposition or adjacency; what apparently is a defect in my analysis is not in his.

3.5. The strict OV restriction Hock (1988:212; 1989:398; 2001:66, 90) claims that Dravidian languages exhibit the ‘‘. . . well-known ‘Strict OV Restriction’ (SOVR) that sentences have only one finite verb . . .’’ Perhaps mistaking the preliminary, didactic exposition of (1) in Steever (1987b) for accepted tradition, HHH assumes that (1a) represents a historically earlier stage of the Dravidian languages, so that primitive Dravidian allowed but one finite verb per sentence, without exception. However, in view of abundant counterevidence, no scholar has seriously or consistently held that any Dravidian language exhibits such a rule synchronically or that it represents a diachronically prior stage in the family’s history. Copious well-formed sentences with multiple finite predicates show this claim to be observationally inadequate.14 In denying that (1) applies to Ancient Tamil, HHH e¤ectively claims that the language lacked sentences with multiple finite predicates, which flies in the face of the available evidence, such as (8a–c) above. Thus despite strong evidence to the contrary, he persists in the idea that something called the SOVR describes the distribution of finite predicates in the Dravidian sentence. In

132 Sanford B. Steever light of section 3.1, the choice of Strict OV Restriction as a label is unfortunate since it suggests that finiteness is essentially correlated with word order, when these two vary independently.

3.6. Quotatives and quotitivals Hock (1987, 2001) seeks to restrict the scope of (1) by claiming that constructions involving direct discourse are not subject to it, but to some other facet of grammar. The criticism is misguided in two ways. First, verbs that elicit the use of FPEVs include more than verba dicendi; they include verbs of propositional attitude (7a) and perception (7b–c), among others, that subcategorize finite sentential complements. Second, the majority of FPEVs, for example Tamil a
ka ‘become’ (7b) and po
la ‘resemble’ (7c), though not involved in reported speech, exhibit the same distribution vis-a`-vis embedded finite predicates as do structures with the FPEV en a ‘say’. Accordingly, all of them belong together. Without explanation, HHH’s reanalysis in (19d) artificially segregates ‘‘quotative’’ FPEVs from their nonquotative counterparts, discarding as spurious a generalization about their shared syntax. Hock (2001:90) makes the unusual claim that in reported speech with two finite verbs, the two component parts, the quoting material and the quoted material, constitute two separate sentences. This requires us to treat ordinary direct discourse as a discontinuous structure in which the quoted material is inserted as a parenthetical expression between the subject and the verb of the quoting material. Consider the Ancient Tamil example in (12). On HHH’s theory, the reported speech in S1 must be a parenthetical expression transported into the middle of another sentence, S0 , the reporting speech, in e¤ect separating the final finite verb vin avuti ‘you ask’ from its dependent clause cirril narru
n parri ‘(while) grasping the pillar of (my) hut’. A far simpler analysis, one consistent with rule (1), is to treat (12) as an ordinary instance of the (center) embedding of a sentential object within an SOV main clause, a phenomenon well established in the syntax of SOV languages. (12)

S0 [cirril

narru
n parri makan S1 [nin small.house pillar grasp-cf your son ulan ¼o
]S1 en a vin avuti . . . ]S0 be-npst-3sm¼int say-inf ask-npst-2s

ya
n t(u) where

What’s So Subversive about Dravidian?

133

‘(While) grasping the pillar of my hut, you ask, ‘‘Where is your son’’?’ Hock also overlooks ample evidence for the syntactic relation between subordinate and matrix clauses when both have finite predicates, e.g., the possibility of Raising the subject of a lower clause into a main clause (Steever 1981b:247–76). Further evidence that reported speech is instantiated by a matrix-complement clause, and not asyndetic parataxis, is presented in section 4.4. HHH then pursues a digression on the quotitival, a construction in which the quoted material is both preceded and followed by quoting expressions, encircling the quote, as it were, with verbal ‘‘quotation marks.’’ In his attempt to separate the quoted and quoting material of reported speech into two or perhaps even three separate sentences, he appeals to parallels from Homeric Greek, neglecting pertinent structures closer to home.15 Within Dravidian, so-called quotitivals appear only in three languages to the south of the Dravidian-Indo-Aryan frontier, Kuvi, Kolami and Parji; in none of them is it a dominant or unmarked embedding strategy. In Parji texts, for example, only four instances occur, and they are marked over against the inherited Dravidian quotative construction based on reflexes of *en ‘say’, which is five times more frequent. HHH claims (2001:87, n.d.) the quotitival occurs in NDr languages, when in fact no language in that subgroup has it; Kolami and Parji belong to CDr, Kuvi to SCDr. Importantly, no quotitivals occur in the older languages, such as Ancient Tamil. They are thus not retentions, but innovations in the three languages, and represent a transition from a robust SOV to a modified SOV template (Steever 1998:36) under the impact of language contact. (13) a.

S0 [S1 [u
nin

b.

S0 [ . . .aru

and kipam]S1 he-acc kill-HRT ‘They plan to kill him.’

eni bicar ca
jomer]S0 . say-cf plan do-prs-3pm

pokrad ki S1 [cenovar ciric pu
barto and say-fut-3m that Saturday Pubar-loc mand ey a
t a
y]S1 eni ku
yurad ]S0 . festival market be say-cf call-fut-3m ‘And he says, ‘‘The festival market at Pubar will be on Saturday,’’ saying he calls.’

134 Sanford B. Steever c.

S0 [a
r

awd nar je S1 [nı
m they-nom say-prs-3p that you-nom ondratar]S1 ]S0 . take-pst-2p ‘They say that you took the pot of rice.’

tiql rice

lotan pot-acc

Within Dravidian, the quotitival represents a transitional stage from the inherited Dravidian construction in the Parji example (13)a, where the quoting verb follows the quoted material, through a ‘‘quotitival’’ stage, where quoting material brackets the quoted material fore and aft in the Parji example (13b), to a final stage in some NDr and CDr languages, where the quoting verb now exclusively precedes the quoted material in imitation of Indo-Aryan, as in the Malto example (13c).16

3.7. Relative-correlative structures Hock (1988:225; 1989:403–404; 2001:70–73) faults rule (1) for licensing the reconstruction of finite relative-correlative clauses to Proto-Dravidian. Taking the traditional Indological stance (see section 5), he holds that early Dravidian lacked finite relative-correlative clauses with FPECs, and if that is so, then (1) must be flawed. Example (14c) illustrates how the Modern Tamil ¼o
relative-correlative strategy uses the clitic ¼o
‘or, any’ to embed (14a), along with its finite verb, under the NP oru paiyan ‘a boy’ that appears in (14b). It contrasts with the relative clause in (14d), which is formed by the unmarked adnominal strategy. (14) a. oru paiyan ne
rru va.n-t-a
n . one boy-nom yesterday come-pst-3sm ‘A boy came yesterday.’ b. oru paiyan -ai.p pa
rtte
n . one boy-acc see-pst-1s ‘I saw the boy.’ c. enta.p paiyan ne
rru va.n-t-a
n ¼o
anta.p which boy-nom yesterday come-pst-3sm¼or that paiyan ai.pi pa
rtte
n . boy-acc see-pst-1s ‘I saw the boy who came yesterday.’

What’s So Subversive about Dravidian?

135

d. ne
rru va.n-t-a oru paiyan -ai.p pa
rtte
n . yesterday come-pst-adn one boy-acc see-pst-1s ‘I saw the boy who came yesterday.’ In support of his claim that older Dravidian lacked such relativecorrelatives, HHH appeals to Ramasamy (1981), an 18-page paper focused on Modern Tamil with just 11 examples from earlier eras. Insightful as Ramasamy’s article is, it is not a definitive historical study of Tamil Relative Clause Formation (RCF). What Ramasamy, following Annamalai (1969, 1997), shows is that all stages of Tamil employ several morphosyntactic strategies for RCF, including adnominal, conditional and correlative strategies (with and without FPEs), to which may also be added tagquestion strategies (Steever 1988:31, 1998:152).17 Just on the basis of cognate structures from Tamil, Toda, Kannada, Telugu and Konda, representing SDr and SCDr, the ¼o
relative-correlative may be projected to at least the period before these two subgroups split, at a minimum time depth of 1500 BCE (see section 3.10.2). Even if, counterfactually, no finite relative-correlatives were attested in Ancient Tamil texts, the conclusion that the language lacked them would be an argumentum ex silencio; internal reconstruction and comparative methodology still permit their reconstruction to a time when Indo-Aryan influence was negligible or nonexistent.18 Pursuing his argument, HHH turns Lehmann’s (1989:349, pass.) statement that clitics are obligatory in Modern Tamil finite relativecorrelatives on its head in an attempt to claim that they were absent in the ancient language (Hock 2001:71). The ¼e
relative-correlative clause in (15), from purana
n u
ru 301:7–8, is clear counterevidence. Not only does it resemble the ¼o
correlative in all details up to the identity of the clitic, it is also the direct ancestor of the ¼e
relative-correlative clause of Modern Tamil, as in (9b).19 (15) en ai na
l tan kum num po
r¼e
an ai how.many day last-npst-3sn your war-nom¼even that.many na
l eriya
r erital ya
van . . . day battle-neg-3p-nom battle-vn how ‘How can one battle those for even one day when their army remains in camp every day and will not fight?’ i.e., ‘On which day can our king fight those whose army remains in camp every day and will not (come to the battle-field to) fight?’

136 Sanford B. Steever 3.8. The dubitative and allied constructions An equally serious shortcoming in his treatment of relative-correlatives and their reconstruction to PDr is the fact that Hock (1987, 1988, 1989:403; 2001) ignores other Dravidian structures with the same gross syntactic features, most prominently the dubitative construction. Such homologous structures tend to fall into a blind spot because even though they exhibit the same overall structure as the ¼o
correlative, they lack Indo-Aryan counterparts to which their synchronic analysis could be referred or in which their historical source might be sought. Two clauses, each with a finite verb, are joined by the clitic ¼o
‘or, any’ to create a matrixcomplement structure. Besides (9a) above, pertinent examples appear in Steever (1988): (38b), (39b), (40b), (41b). Consider also the set of Telugu sentences (Krishnamurti and Gwynn 1985:360–61) reproduced below in (16) which exhibit the same overall syntactic structure as dubitative and relative-correlative constructions. (16) a.

eppud u wasta
d ]S1 ¼o
na
ku he-nom what.time come-fut-3sm¼or I-dat teliyadu]S0 . be.known-neg-3sn ‘I do not know when he will come.’

S0 [S1 [a
yana

b.

kottatam elag]S1 ¼o
wa
d iki pocket-p strike-vn what.manner¼or he-dat telusu]S0 . be.known-3sn ‘He knows how to pick pockets.’

c.

e
m ceyya
l ]S1 ¼o
a
lo
cincu.]S0 hereafter what do-prm¼or think-imp ‘Think (about) what should be done hereafter.’

S0 [S1 [ je
bulu

S0 [S1 [ika

These constructions are widespread and unremarkable in Dravidian grammar, and their distribution throughout Dravidian clearly warrants their reconstruction to the early stages of the family. In arguing against the reconstruction of finite relative-correlatives, HHH once again dismantles a significant generalization by dividing into multiple, arbitrary constructions, syntactic structures which belong together as far as their relation to finite

What’s So Subversive about Dravidian?

137

predicates is concerned. Thus little confidence may be placed in his analysis of these complex structures or in his claim that rule (1) is flawed because it is consistent with relative-correlatives and their reconstruction.

3.9. The serial verb formation Hock’s (1988:229) claim that neither his critique of, nor his alternative to, rule (1) a¤ects my analysis of the Serial Verb Formation in the Dravidian languages – the core subject of my 1988 monograph – is disingenuous since under that analysis, formal finiteness, the defining characteristic of the SVF, presupposes functional finiteness. So to dispose of (1a and b), as HHH’s alternative in section 4 demands, is in e¤ect to vitiate the analysis of the SVF in (1c and d), how it interacts with functional finiteness in Dravidian and why it is so striking within this family.

3.10. The SOVR’s problems with Dravidian linguistics 3.10.1. Geographic and Genetic Subgrouping A number of ga¤es in HHH’s critique may be directly traced to an unfamiliarity with the Dravidian languages or the specialist literature. Scholars currently divide the languages into four genetically related subgroups (Steever 1998, Krishnamurti 2003): South, South-Central, Central and North.20 To support his claim that the ‘‘northern languages’’ are both typologically closer to Indo-Aryan and more nearly representative of PDr syntax than the others, Hock (1988:211–212) must abandon the accepted classification and divide the languages by rough geographic terms into a Northern and a Southern subgroup. This gives rise to two illusions: first, that each subgroup has an equal number of languages so that each one carries equal weight in decisions concerning reconstruction to the protolanguage and, second, that the quality, quantity and antiquity of data is equal between the two. Clearly, this is not so. For example, isoglosses support only three NDr languages, Brahui, Kurux and Malto; none is attested before the modern era. So the split between Northern Dravidian and the other languages is not half and half, as Hock’s schema would have us believe, but only 3:23 (or 3:26 if we include the ancient forms of Tamil, Kannada and Telugu as separate

138 Sanford B. Steever languages). Tamil has a richly attested history spanning 2200 years while records of Malto first appeared less than 200 years ago. As noted in section 3.6, HHH misclassifies Kolami, Kuvi and Parji as NDr. Hock (2001:71) mistakenly holds that Koraga, spoken in the South Kanara district of Karnataka, belongs in NDr and urges us to accept that its presence among SDr languages constitutes both a survival from earliest Dravidian and proof that his ‘‘Northern Dravidian,’’ along with its putative syntactic patterns, once spanned a broader geographic range than today. This would, he believes, give more weight to his claim that finite predicates in early Dravidian and early Indo-Aryan were governed by the SOVR. However, closer inspection reveals that Koraga is SDr. Bhat’s (1971:3) suggestion that it shared certain features with NDr languages rests on assertions which HHH lacked the specialist expertise to confirm or refute. For example, the claim that Koraga’s use of -n- as a future tense marker places it in the NDr subgroup because Kurux also uses it, runs afoul of the fact that SCDr languages, such as Kui, also use it (Steever 1993:35–68), as do certain constructions in Ancient Tamil (SDr) (Steever 1993:101). It is therefore a retention from PDr, not an innovation within NDr. Similarly, Bhat’s isogloss of not using a plural su‰x for irrational nouns to join Koraga with Kurux in the NDr subgroup simply ignores the fact that the same situation also holds in Tamil. For this and other reasons, Steever (1998) and Krishnamurti (2003) continue to catalogue Koraga where it belongs, in SDr (to the extent it is an independent language).21 In another misidentification, Hock (2001:71) calls Konda a SDr language when it is SCDr. It is surprising, to say the least, that HHH’s alternative subgrouping disregards all the previous scholarly work on isoglosses used to support the accepted subgrouping simply to prop up his alternative. These and similar di‰culties cast doubt on his arguments based on subgrouping, particularly that the SOVR (see section 5) once had a wider geographic range and represents a more archaic stage of Dravidian.

3.10.2. Relative chronology of Dravidian and Indo-Aryan Hock (2001:65) suggests that when comparing Dravidian and Indo-Aryan syntax, precedence should be accorded to Indo-Aryan since the earliest attested Dravidian appears one thousand years later than the earliest Indo-Aryan of the Rig Veda. We should recall, however, that no written

What’s So Subversive about Dravidian?

139

attestations of Vedic date back to 1200 BCE; that date is an extrapolation based on comparison with other Indo-European subgroups, but an extrapolation nonetheless. (And, the epigraphic evidence suggests that Ancient Tamil may well have appeared in inscriptions before Vedic did.) Whatever the case, if we can extrapolate for Indo-Aryan, we can do the same for Dravidian: it has been argued that Proto-Dravidian existed as a single language as late as 4000 BCE, and that South Dravidian, the last branch to split o¤, took shape around 1500 BCE (Zvelebil 1970, Steever 1998:11). If so, Indo-Aryan can hardly be said to have exerted any significant influence on features commonly shared by the SDr, SCDr and CDr languages, whose common ancestor well predates 1500 BCE. In the event, Hock reverses his principle of ‘‘the earlier the attestation, the weightier the evidence’’ when discussing subgroups within Dravidian, ascribing to his ‘‘Northern Dravidian’’ subgroup greater importance in the reconstruction of PDr syntactic patterns than ‘‘Southern Dravidian,’’ whose earliest attestations predate those of Northern Dravidian by about two millennia. Finally, Hock (2001:90) opines that rule (1) applies only to his ‘‘Southern’’ subgroup, and then only in the modern period. This stance suggests that despite prolonged and intense interaction with surrounding IndoAryan and, in the case of Brahui, Iranian languages, NDr retains an archaic system enshrined in (19) below while (1) is an innovation in 20 out of the 23 modern languages. In doing so, HHH implies that the ancient forms of Tamil, Kannada and Telugu also reflect this older system. This flatly ignores pertinent evidence from Ancient Tamil and Ancient Kannada in Steever (1988); Lehmann’s (1994:143–148) illustration that the system of FPEs operates in Ancient as well as in Modern Tamil; and the fact that (1) was explicitly projected to PDr because it has remained largely intact in the SDr, SCDr and CDr subgroups. Playing fast and loose with subgrouping and relative chronology serves as little more than a red herring to artificially limit the true scope of (1) in order to claim this such a limitation is something that (19) overcomes.

3.10.3. Misanalysis of Dravidian data The absence of a practical command over any Dravidian language leads HHH into numerous mistakes in identifying and analyzing specific forms and structures that appear in his critique and alternative, gravely

140 Sanford B. Steever jeopardizing their credibility. For instance, Hock (2001:70) claims that his example (7), reproduced here as (17), contains a nonfinite relativecorrelative clause when it contains no relative clause whatsoever, correlative or otherwise. Ignorant of Tamil grammar, he misanalyses (17) as a coordinate structure with three conjuncts, mistranslating it as, ‘I did that, finished (it), and (then) went away’; the sequence ceytu mutittu ‘finished doing’ is an auxiliary compound verb whose members function as a single predicate (Steever 2005:235–37); (17) thus has just two clauses. (17) na
n atai.c ceytu mutittu.p po
n e
n . I-nom it-acc do-cf finish-cf go-pst-1s ‘I finished that and went away.’ Hock (1988) presents a series of examples which he claims rule (1) cannot adequately handle, and which therefore necessitates his alternative in (19). However, he misidentifies the structure of most of these putative counterexamples, so they prove not to be genuine counterevidence for rule (1), but merely red herrings. For example, Hock (1988:221) mistakes a lexical compound verb (Steever 2005:80) for a coordinate sentence with multiple conjuncts: his Kuvi example (15b), d e
kitesi vatesi, is an SVF that means ‘he carries’, not ‘he holds and he goes’ (cf. Tamil kon tu vara ‘take’). His putative counterexamples (15c and d) contain multiple separate sentences in a discourse, not instances of single sentences with multiple clauses. Since the scope of rule (1) is explicitly restricted to the sentence in clause (1a), not a discourse, it cannot be falsified by instances of parataxis. Hock’s (1988) misanalysis of examples in his (18a–g) is particularly egregious: he presents strings with multiple finite predicates, attributing their occurrence to the presence of the clitic ¼um ‘and, all’, as he attempts to build a bridge to an analysis based on coordination (see section 4.2). But he misses an important fact: while ¼um may attach to a wide variety of forms, it does not attach to finite verbs. Yet he sees in it a homologue to Sanskrit ca ‘and’ which is ‘‘enclitic on the non-verbal first constituents of conjoined sentences (Hock 1988:222),’’ a restriction irrelevant to Dravidian ¼um. He produces examples with multiple finite predicates, claiming they ‘‘cannot be accounted for as licensed by embedding predicates or clitics . . . (Hock 1988:233).’’ In (18a) below, HHH attributes the presence of two finite predicates to the presence of the clitic ¼um in the second sentence. Even if the two finite predicates were part of a single sentence, a point not transparent

What’s So Subversive about Dravidian?

141

or readily defended, HHH fails to recognize that alla is the negative stem of the verb ka ‘become’, one of the two primary FPEVs in Dravidian, so that each occurrence of this verb licenses the presence of a finite predicate.22 The multiple reflexes of this FPEV also license the multiple finite predicates in HHH’s Kannada example (17b). The purported instance of conjunction in Hock’s (1988:222) example (17c) is between two nonpredicate NPs, not two verbs, and so cannot falsify a rule that governs finite verbs and predicate nominals. Consider next his treatment of (18b) below (¼ HHH’s 18d), which contains two separate sentences, each with its own finite verb. HHH’s English translation combines two separate Tamil sentences into one, then performs Conjunction Reduction on the result; this loose English translation can in no way constitute an acceptable analysis of the Tamil original.23 His examples (18d, e and g) display instances of discourse with multiple sentences – examples of parataxis, not coordinate sentences. The Toda example in (18c) below is particularly instructive: it contains three sentences in a row, each with its own finite verb. Strangely, HHH o¤ers no grounds for ignoring the periods that Emeneau (1984:67–68) took care to place after each finite verb to mark the end of its sentence. Here HHH is analyzing the syntax of Dravidian structures in terms of their English translations, not the grammars that actually generate them. Finally, HHH’s example (18g), from Malto (NDr), o¤ers no support for his position because it lacks any reflex of, or counterpart to, the clitic ¼um ‘and, or’. (18) a. itu mey alla atu.v¼um this.one-nom truth-nom become-neg-3sn that.one-nom¼also mey alla. truth-nom become-neg-3sn ‘This one is not true. That one is also not true.’ b. man itar az inta
rkal. kutiraikal¼um man-pl-nom perish-pst-3Np horse-pl-nom¼also az intan a. perish-pst-3np ‘The men perished. The horses also perished.’ ß m c. twı
¼ ıßtspini. e
r¼m batsˇpini. funeral¼also put-pst-1s bu¤alo¼also caught-pst-1s ko
ß ¼m bi odsˇpini. funeral¼also help perform-pst-1s

142 Sanford B. Steever ‘‘I have made the two-day funeral. I have also caught the bu¤alo. I have also helped (you) perform the funeral.’’ HHH fails to recognize that ¼um ‘and, or, also’ has several functions: it does not invariably act as a conjunction, sometimes it serves as a quantifier, and sometimes as a discourse particle meaning ‘also’, as in (18b). In the Tamil sentence ya
r¼um vara.villai ‘no one came’, for example, ¼um serves as a quantifier: it implies the presence of no (other) conjunct. Inspection of the putative counterexamples in this section and those discussed in section 3.6 reveals a serious underlying problem in HHH’s critique and alternative: he provides no consistent means for distinguishing between the clauses of a complex sentence and a string of independent sentences in a discourse related only by parataxis. In (18) he treats clear instances of parataxis as examples of syntactically coordinated sentences while in (12) and (13) he treats ordinary instances of matrix-complement sentences as examples of two syntactically separate sentences, with one parenthetically inserted in another. Without grounding his program in a reliable, independent concept of sentencehood, he cannot say whether (19) ranges over single sentences or stretches of discourse. If we assume that rule (19) applies just to single sentences, then arguments in the next section show that it fails to make correct predictions about instances of reported speech. By contrast, rule (1) is explicitly formulated to range over the sentence; once we eliminate HHH’s cases of parataxis, it makes all the correct predictions. The preceding discussion suggests that without a working knowledge of any Dravidian grammar and relying wholly on secondary sources and transliterated examples, HHH is hard pressed to identify even the most basic forms in Dravidian sentences or to match them to their proper grammatical functions. This e¤ectively frustrates his attempts to embed his analysis of Dravidian finiteness within the matrix of Dravidian grammar. None of the supposed counterevidence disproves rule (1) because it is irrelevant and fails to test that rule’s parameters. As it invalidates neither rule (1) nor the analysis of Dravidian syntax which that rule represents, what other purpose might it serve? By casting doubt on rule (1), the critique appears to function as a proof of HHH’s alternative, suggesting that a critical review of one proposal may legitimately serve as a positive demonstration in favor of anothr. But even if the critique were valid, it would still not secure the SOVR’s superior descriptive and explanatory powers; those have to be earned independently. Arguments in the next section show that his

What’s So Subversive about Dravidian?

143

alternative fails to do so, raising many more problems than it solves and solving far fewer problems than rule (1).

4. The strict OV restriction Hock’s alternative treatment of the distribution of finite verbs in Dravidian (Hock 2001:70–71, 90) is grounded in his critique of rule (1); as a consequence, that alternative inherits many of the flaws brought to light in the previous section’s rebuttal. The Strict OV Restriction (Hock 2001:90) is given explicit form below in rule (19), where letters have been added to identify the individual clauses. (19) ‘‘[a]One finite verb is permitted per sentence. [b]However, in conjoined structures, each of the conjoined clauses is treated as a sentence in its own right and is therefore permitted to have its own finite verb. [c]Relative-correlatives syntactically are conjoined structures, permitting each of the two sub-clauses to have its own finite verb. [d]In addition, cited discourse is treated as syntactically independent and thus entitled to its own finite verb(s); it may be marked by a quotative marker to signal that it is exempt from finite-verb restrictions. [e]Elsewhere, i.e., in structures with syntactic embedding, only one finite verb is permitted per sentence, occurring in the ‘matrix’ clause; other verbs have to appear in non-finite shape (participles, absolutives, infinitives, or verbal nouns).’’ Even superficial comparison of rules (1) and (19) reveals several striking contrasts. The clauses in (1) are ordered hierarchically, those in (19) are arranged sequentially; (1) makes crucial use of FPEs while (19) dispenses with them. Rule (1) analyzes the distribution of finite predicates, both noun and verb, in terms of their position in a constituent tree structure whereas (19) analyzes the distribution of finite verbs by the kinds of constructions in which they appear, ranging from the general, e.g., coordinate structures, to the specific, e.g., relative-correlative clauses. Critical examination of (19) in this section reveals that it is neither observationally nor descriptively adequate for the syntactic phenomena covered by rule (1), or even for the subset of cases to which HHH confines himself, and is thus to be rejected as a principle of Dravidian sentential architecture.

144 Sanford B. Steever Rule (19) strings together disparate topics, and its five clauses are poorly organized; (19e), for example, redundantly repeats information in (19a). Clause (19e) enumerates some but not all the nonfinite verb forms found in Dravidian, omitting, for instance, the conditional. The three interior clauses list exceptions to (19a): coordinate structures, relative-correlatives and reported speech. Clause (19c) claims that relative-correlatives are exempt from (19a) because each of its parts is a conjunct in a coordinate structure and so is e¤ectively accommodated by (19b), which grants each conjunct its own finite verb. Clause (19d) claims that reported speech (¼HHH’s cited discourse) is exempt from (19a) because even though it resembles a single complex sentence, it actually represents two syntactically separate and distinct sentences. No explanation is attempted of what underlying property these three constructions might share in common, to the exclusion of all else in Dravidian syntax, that shields them from the operation of (19a). The rule’s problems inhere not only in what it includes but also in what it omits. Clause (19c) ignores several constructions with the same overall structure as relative-correlatives formed with the clitic ¼o
‘any, or, whether’, including those in (9a) and (16). While clause (19d) exempts reported speech from (19a)’s restriction of one finite verb per sentence, it ignores homologous structures with nonquotative verbs such as a
ka ‘become’ and po
la ‘resemble’. Moreover, rule (19) pointedly excludes two important provisions for the analysis of finiteness that were incorporated into rule (1). First, it disregards the role predicate nominals serve in the construction of the Dravidian sentence. Second, it does not di¤erentiate between functional and formal finiteness, and therefore provides no basis for analyzing SVFs. Though it is customary to adduce examples and provide syntactic arguments to support one’s analysis, HHH largely foregoes an empirical demonstration that (19) accurately describes any of the languages, Dravidian or Indo-Aryan, whose structure and history he is glossing. For example, we are not provided with a consistent means of distinguishing between internally complex single sentences and multiple sentences in a discourse that are related by parataxis. We are asked instead to accept that the criticism of rule (1) somehow validates (19). At no point, however, are we actually shown how (19) is derived from, applied to or confirmed by attested Dravidian examples. Attempts in this section to apply rule (19) to even a modest corpus of syntactic structures reveal it to be descriptively inaccurate and explanatorily inadequate for the Dravidian languages.

What’s So Subversive about Dravidian?

145

4.1. Predicate nominals The single most egregious omission in rule (19) is its failure to accommodate predicate nominals, which in Dravidian sentences have the same distribution as finite verbs (Steever 1988:2–3); (19) thus ignores the possibility of their ever functioning as finite predicates. The rule claims, in e¤ect, that only nonfinite verbs, but not predicate nominals, may ever appear in embedded structures, coordinate or subordinate. Moreover, even after replacing the term ‘‘finite verb’’ with ‘‘finite predicate,’’ the revised rule (19) still makes false predictions about Dravidian sentential architecture. Thus amended, it falsely predicts that within a Dravidian sentence, a predicate nominal may appear tout court in any conjunct of a coordinate sentence; however, except for the matrix clause in a complex sentence, Dravidian predicate nominals are always embedded by an FPE, a concept that the program behind (19) does not support. Empirically then, the lack of a provision for predicate nominals renders (19) incapable of describing onehalf of the basic sentence types of Dravidian, as well as the many embedded structures, coordinate and subordinate, in which they are routinely found to appear. These synchronic shortcomings aside, HHH’s model suggests no diachronic pathways along which predicate nominals might have developed to become integrated into Dravidian grammar as finite predicates.

4.2. Coordinate (and subordinate) structures Strictly construed, clause (19b) fails to limit coordinate structures to conjoined sentences, counterfactually predicting that any conjunct in any conjoined structure, e.g., conjoined NPs, could host a finite verb. Such a prediction is falsified by the available evidence. However easily such a ga¤e may be remedied, (19b) faces far more serious challenges. With two specific exceptions, relative-correlative (19c) and quotative (19d) constructions, clause (19b) places the entire burden of accounting for the distribution of multiple finite verbs on coordinate structures. By the same token, it e¤ectively eliminates the possibility of finite predicates ever occurring in subordinate structures.24 The condition that finite verbs appear in coordinate structures cannot be considered a su‰cient condition since Dravidian languages exhibit many coordinate structures, such as (4a and b), which despite multiple

146 Sanford B. Steever conjuncts, have but one finite verb. Nor can it be considered a necessary condition since finite verbs do occur – and regularly so – in subordinate structures, even in those not explicitly exempted by clauses (19c and d). The Kurux sentence in (20) contains a verb of propositional attitude which takes as its object a subordinate sentential complement with its own finite verb. Insisting that (20) is a coordinate structure, in conformity with (19b), yields a bizarre, involuted coordinate structure in which one of the two conjuncts, S1 , is shoved inside the other, S0 , apparently as a parenthetical, and suspiciously in a position where a (sentential) object is expected. (20)

S0 [e
m

S1 [a
s

nalakh nanos]S1 ba
ckam]S0 . we-nom he-nom work do-fut-3sm expect-pst-1p ‘We expected him to do the work.’

Despite HHH’s assertion that they are not syntactically embedded structures, quotative constructions with reflexes of *en ‘say’ exhibit the grammatical properties of matrix-subordinate structures and thus disprove (19b)’s predictions, as do parallel structures that use such FPEs as *a
ka ‘become’ or po
la ‘resemble’, which cannot reasonably be considered quotatives at all. A number of quotative constructions, for example, commonly allow Raising to Object Position (Steever 1981b). To treat quotative constructions as instances of parataxis, in the attempt to assimilate Dravidian structures to the SOVR, serves only to generate questionable, unwieldy structural analyses wherein the entire expression consists of two independent sentences, one of which (the quoted material) is wholly contained within the other (the quoting material).25 Rule (19b) thus forces us to ascribe to these structures a grammatical analysis they demonstrably do not possess. Whereas section 3.4 showed that adjacency plays no significant role in the model of finiteness in rule (1), it is crucially involved in the SOVR in (19b) since that clause recognizes only linear order as relevant to the assignment of finiteness. Under clause (19b), a verb is finite by virtue its position within a string of linearly ordered conjuncts in coordinate structures, and nothing more. This makes it nigh impossible for two finite verbs to occur adjacent to one other because extraneous material from their individual conjuncts is bound to intervene. Thus, in S0 [S1 [S-X-O-FV]S1 and S2 [S-X-OFV]S2 and S3 [S-X-O-FV]S3 ]S0 , as interpreted under (19b), the multiple tokens of finite verbs (FV) cannot stand adjacent to one another because such material as subject (S), object (O) or some other expression (X) may

What’s So Subversive about Dravidian?

147

intervene. This artifact of (19b) rules out the possibility of many ordinary matrix-complement structures, such as those found in (8c) and (20). Ironically, the only construction that might meet the structural description implied in (19b) is the SVF, a construction that HHH excludes altogether from the scope of rule (19). Besides these synchronic shortcomings, nothing in HHH’s discussion of (19b) provides a diachronic mechanism for the transition from a system that relies on coordination and adjacency for finiteness to one that does not, as in the Dravidian languages as actually attested. HHH’s insistence on incorporating adjacency in the theory of Dravidian finiteness thus runs afoul of these examples, and is as misplaced in his alternative as it is in his critique of rule (1). At this point it seems germane to recall that (19) has nothing to say about such topics as word order or Extraposition in Dravidian, putative shortcomings for which (1) was criticized.

4.3. Supernumerary analyses A further infelicity of (19) is that it imposes radically di¤erent grammatical analyses on constructions that may be formed through two di¤erent morphosyntactic strategies, one using a finite, the other a nonfinite verb. Consider the multiple analyses it imposes on conditional and relative clause constructions. Note first the pair of Tamil conditional sentences, realized by di¤erent strategies: in (21a) the verb in the protasis assumes a nonfinite conditional verb form (vanta
l ‘if [he] comes’) while in (21b) the corresponding protasis consists of a finite predicate and the conditional form of the FPE en a ‘say’. Linking finiteness with coordination in (19b and e) forces us to analyze the protasis of (21a) as subordinate to its apodosis just because it contains a nonfinite verb. But by the same token, it treats the protasis in (21b), because it contains a finite predicate, as coordinate to its apodosis.26 Both are clearly instances of matrix-complement structures, regardless of the specific strategy their construction employs. That (19) generates supernumerary and incorrect structural analyses undermines its claim to superior descriptive power. (21) a. avan vanta
l na
n he-nom come-cnd I-nom ‘I will beat him if he comes’

avan ai he-acc

atippe
n . beat-fut-1s

148 Sanford B. Steever b. avan varukira
n en ra
l na
n avan ai atippe
n . he-nom come-prs-3sm say-cnd I-nom he-acc beat-fut-1s ‘I will beat him if he comes.’ Further evidence from conditional sentences reveals that, far from representing the pristine state of PDr syntax, as HHH claims, the distribution of finite predicates in even the NDr languages fails to conform to (19b), seen already in example (20). This rule incorrectly describes the Kurux conditional sentences in (22) whose protasis and apodosis both have finite forms, imposing on them a coordinate structure when the protasis is actually subordinated to, not coordinated with, the apodosis. (22) a. a
s kalos hole he-nom go-fut-3sm if ‘If he goes, I will go, too.’

e
n I-nom

hu˜ too

kalon. go-fut-1s

b. a
s xessan xoya kalos annuhu˜ a
d he-nom paddy reap-inf go-fut-3sm even.if she-nom kalo. go-fut-3sf ‘Even if he goes to harvest paddy, she will not go.’

mal not

Hock (2001:69) tries to finesse this problem by asserting without demonstration that early Dravidian had what he calls ‘‘non-embedded finite subordination,’’ as exemplified by relative-correlatives which, according to (19c), constitute two conjuncts in a coordinate structure. What sort of syntactic concept non-embedded finite subordination may be, and what justification HHH has for introducing such an ill-defined concept into syntactic argumentation, are questions that scholars might validly raise but will find unanswered. The attempt to explicate it as a form of ‘‘pragmatic embedding,’’ which apparently contrasts with syntactic embedding, succeeds only as a circular explanation of one undefined concept by another. Turning to Relative Clause Formation, contrary to what SOVR predicts, the possibility of Backwards Pronominalization within Tamil ¼o
relative-correlative clauses – in fact, within all relative clauses, regardless of the strategy used to form them – demonstrates that the preceding clause is subordinate to, not coordinate with, the following one. Backwards Pronominalization permits a pronoun to precede its antecedent just in case that pronoun stands in a clause that is lower to, and c-commanded by,

What’s So Subversive about Dravidian?

149

its antecedent. In (23a) the reflexive pronoun tan ‘self ’ is coreferential with and precedes its antecedent appa
‘father’, something that would be impossible if they occurred in two conjuncts of a coordinate structure. Such commonplace sentences indicate that the reflexive pronoun occupies a clause lower than its antecedent. And the same pattern holds for all relative clauses in Tamil, regardless of the strategy used to realize them, finite or not; one thus finds Backwards Pronominalization into relative clauses formed on the conditional (23b) and even the ultra-Dravidian adnominal (23c) strategy. (23) a.

S0 [S1 [anta.k

ka
lattil ya
r-ya
r tan i a
fisukku that time-loc who-nom-who-nom self-gen o‰ce-dat vanta
rkal ]S1 ¼o
avarkal-ai appa
i cantippa
r.]S0 come-pst-pl¼or they-p-acc father-nom meet-fut-3h ‘In those days, father would meet with whoever came to his o‰ce.’

b.

ka
lattil ya
r tan i a
fisukku that time-loc who-nom self-gen o‰ce-dat vanta
l¼um]S1 avarkal-ai en kal appa
i come-cnd¼and they-pl-acc our-gen father-nom cantippa
r.]S0 meet-fut-3h ‘In those days father would meet whoever came to his o‰ce.’

c.

S0 [NP [S1 [tan i

S0 [S1 [anta.k

a
fisukku vanta]S1 a
tkal-ai]NP self-gen o‰ce-dat come-adn person-p-acc appa
i cantippa
r.]S0 father-nom meet-fut-3h ‘Father would meet those people who came to his o‰ce.’

HHH’s analysis of Dravidian relative-correlatives in (19b) explicitly and against all available evidence construes them as coordinate structures, thus generating false structural analyses. The arguments in this section expose a defect in HHH’s analysis: it directly correlates finiteness with specific constructions, such as coordinate or relative clause structures, rather than with the positions a finite predicate may occupy in a tree structure. In so doing, it brings into contact two dimensions of linguistic structure that do not directly interact and generates a set of flawed structural descriptions. Rule (1), because it avoids this spurious correlation, does not give rise to the

150 Sanford B. Steever problems uncovered here, but accurately predicts the distribution of finite and nonfinite predicates in all the sentences of the data set. Apart from these synchronic errors, HHH’s alternative o¤ers no explanation for the historical development of ‘‘coordinated’’ conditional and relative-correlative structures into the subordinated structures actually attested throughout Dravidian. Finally, since areal considerations play a role in his arguments against the DSH, we note that Kachru (1987:488) describes the two clauses of relative-correlative clauses in Hindi-Urdu as main and subordinate clauses, not as two conjuncts of a larger coordinate structure. So, anticipating the typological argument HHH wishes to advance below in section 5, we may fairly wonder whether the SOVR accurately describes relative-correlatives in Indo-Aryan, much less Dravidian.

4.4. Reported speech Section 3.6 rebutted HHH’s claim that in Dravidian reported speech, the quoting speech and the quoted speech constitute two independent sentences related in a linear string only by parataxis. Evidence from Steever (2002) further demonstrates that in Tamil reported speech, be it direct or indirect discourse, the quoted speech behaves syntactically like the subordinate clause of a matrix-complement structure; the quoting speech, like its corresponding main clause. The criticism just leveled against HHH’s analysis of conditional sentences such as (21) applies with equal force to his analysis of direct discourse: the SOVR requires that to appear in its final form, quoted speech must be inserted within the quoting speech as a parenthetical expression – and coincidentally in just the position where a (sentential) object is expected on independent grounds. The distinction between direct and indirect discourse is mediated in part by the syntactic process of Clause Union, which applies within the confines of a single sentence, not across a stretch of two or more independent sentences in discourse. It may apply to the two clauses of the reported speech to varying degrees, creating a cline of indirect discourse in which the verb in the quoted material assumes a nonfinite form. So while Clause Union does not apply in the derivation of the direct discourse in (25a), it does in the derivations of the indirect counterparts in (24b and c), with the result that the nominative-case subject of the lower clause is raised into the main clause and appears in the accusative case while the lower-clause verb assumes nonfinite forms.

What’s So Subversive about Dravidian?

(24) a.

S0 [avan

va
]S1 S1 [nı
he-nom you-nom come-imp ‘He said, ‘‘You come!’’ ’

b.

S0 [avan

c.

S0 [avan

en ru saying-cf

151

con n a
n ]S0 . say-pst-3sm

en n ai vara.c con n a
n ]S0 . he-nom I-acc come-inf say-pst-3sm ‘He told me/said to me to come.’ en n ai varum pati con n a
n ]S0 . he-nom I-acc come-fut-adn manner say-pst-3sm ‘He told me to come’

To separate direct discourse into two independent sentences and assert that their relation is one of asyndetic parataxis, while simultaneously claiming indirect discourse involves but a single sentence with a matrix-complement structure, as the SOVR does, generates untenable analyses contradicted by the available syntactic evidence. Nor, as section 3.6 already showed, can appeal to a ‘‘quotitival’’ explain how a paratactic, ‘‘coordinate’’ direct discourse historically evolves into the one-sentence, matrix-subordinate structure actually observed in Dravidian. Not only does the SOVR fail to capture the gross syntactic structure of reported speech, it fails to accurately predict the distribution of finite predicates in ordinary Dravidian sentences. Take the brace of Tamil sentences in (24a–c) above, each of which contains the quotative verb con n a
n ‘he said, told’. Under (19c), the presence of this quotative verb should trigger the presence of a finite predicate in the quoted speech, when in fact it does so only in (24a). Clause (19c) thus incorrectly predicts that (24b and c) should have finite predicates in the quoted speech when both of them exhibit nonfinite verb forms. Rule (1), by contrast, accurately predicts the distribution of finite predicates in (24a–c) because only (24a) contains a FPE, viz. en ru ‘saying’. This highlights another contrast between the two rules. Rule (1) invokes only formal dimensions of tree structure whereas (19) mixes formal and functional criteria, with (19d) relying on an implicit lexicographic classification that distinguishes quotative verbs, or verba dicendi, from all other kinds of verbs. Rule (1) makes no such distinction; what unites forms of Tamil en a ‘say’, a
ka ‘become’ and po
la ‘resemble’ is not membership in an underlyingly unified lexicographic class, but the syntactic fact that none imposes case marking on it objects. Thus, appeal to quotative verbs in (19d) is neither su‰cient nor necessary in determining the distribution of finite predicates in the Dravidian sentence.

152 Sanford B. Steever 4.5. Serial verb formations Rule (19) fails to draw the distinction between functional and formal finiteness that is crucial to the original analysis of the Serial Verb Formation. HHH’s analysis appears not to recognize that formal finiteness, the key to analyzing the SVF, is defined in clauses (1c and d) over against functional finiteness, which is defined in clauses (1a and b). But formal finiteness presupposes functional finiteness; therefore, the presence of SVFs in any Dravidian language would imply that it has a means of defining functional finiteness. NDr SVFs, in the shape of auxiliary verb constructions, fail to conform to (19b) because as Steever (2005:14–18, 99–90, 127–30) shows, auxiliary verb constructions neither originate nor end up in coordinate structures. Descriptive compound and echo compound verbs, present throughout Dravidian, fail grammatical analysis as coordinate structures, as well. Yet these constructions regularly occur in NDr with multiple finite verbs, in stark contrast to what the SOVR predicts is possible. Thus in failing to provide a basis for distinguishing between functional and formal finiteness, and therefore failing to provide for SVFs, the SOVR and rule (19) e¤ectively define into existence a class of exceptions that they can neither explain nor explain away.

4.6. Diachronic issues The SOVR, as expressed in (19), sheds no light on the diachronic side of the equation, o¤ering no explanation of how HHH’s alternative version of PDr developed into the grammatical systems actually found in Dravidian. Leaving aside the considerable structural problem posed by predicate nominals, how do finite verbs, which under (19b) can occur only one to a conjunct in a coordinate structure, finally come to appear in subordinate clauses? Moreover, how do 20 out 23 languages innovate the same system of FPEs and SVFs described in rule (1) by choosing from all of the grammatical and lexical devices available the exact same set of FPEVs and FPECs? And how is this massively coincidental innovation, beyond all probability, superior to a hypothesis of retention from a common ancestor? Can concepts as murky as ‘‘pragmatic subordination’’ and ‘‘non-embedded subordination’’ explain such linguistic changes? Appeal to such notions, absent any serious engagement with Dravidian syntax, must be deemed circular and self-serving.

What’s So Subversive about Dravidian?

153

4.7. The shortcomings of the SOVR The SOVR, as expressed in (19), incorrectly describes the syntax of finiteness in Dravidian and makes false predictions about many di¤erent structures. First, in pegging all its explanatory potential to the linear order of conjuncts in coordinate sentences, it cannot accommodate finite predicates in subordinate clauses: (7a–c), (8a–c), (9a–b), (11a), (12), (13a–c), (14c), (15), (16a–c), (20), (21b), (22a–b), (23a) and (24a) above are all counterexamples. Second, it wholly neglects the fact that predicate nominals have the same distribution as finite verbs in Dravidian, and thus fails to describe one-half of the basic sentence types of the languages, as well as constructions in which predicate nominals may be embedded, such as (7a), (8b), (15), (16b). Accordingly, it ignores the fact that when embedded in either subordinate or coordinate structures, such as the four just noted, predicate nominals are always embedded by FPEs, a concept with no standing in HHH’s alternative. Third, in dispensing with FPEs, the SOVR ignores the fact that in the vast majority of the languages, FPEs are limited to a set of about a half-dozen elements, and that throughout the family that core of FPEs consists of cognate forms. Fourth, it fails to recognize that nonquotative FPEVs, such as *a
ka ‘become’, have the same distribution with respect to embedded finite predicates as do quotative constructions based on reflexes of *en ‘say’, as in (7a–c), (8c), (13)a. In e¤ect, (19) dismantles that generalization by insisting on radically di¤erent analyses for these structures. Fifth, it distorts the di¤erences between finite and nonfinite strategies used to form relative clauses, conditional clauses and other kinds of dependent clauses, implying, for example, that a relative clause with a nonfinite verb form, such as (23b and c), is a matrix-subordinate structure while a relative-correlative with a finite verb form such as (23a) is a coordinate structure. The SOVR thus generates supernumerary analyses of these structures, at least half of which are automatically false. Sixth, it fails to di¤erentiate between functional and formal finiteness, a necessary precondition for the analysis of serial verb formations, thereby creating a class of unexplained exceptions. Seventh and perhaps most crucially, (19) describes no actual Dravidian language while (1) straightforwardly describes the vast majority of the languages, 20 out of 23, or, if one lists the ancient languages separately, 23 out of 26. As regards the three NDr outliers, (1) may be appropriately modified to accommodate their special developments by modifying the set of FPEs they possess. Lastly, apart from its descriptive

154 Sanford B. Steever shortcomings, rule (19) o¤ers no basis for the diachronic development of finite and nonfinite predicates in the Dravidian languages while rule (1) does provide a template for diachronic change through the order and relative markedness of its individual constituent clauses. Though this review of the SOVR has o¤ered a few suggestions for improving the formulation of rule (19), such as replacing the term ‘‘verb’’ by the term ‘‘predicate’’ in (19a), the bulk of the evidence makes it clear that the SOVR cannot either in its present form or in any likely revision be reconciled with the Dravidian data, but represents an analytic cul-de-sac. The very formulation of the SOVR in rule (19) requires, but fatally lacks, independent means for (i) identifying a sentence and distinguishing it from a string of sentences paratactically related in a discourse or (ii) for identifying complex sentences and distinguishing between subordinate and coordinate structures. Indeed, a major stumbling block of (19) is that it makes finiteness vary directly with individual constructions, such as coordinate and relative-correlative clause structures, rather than correlating finiteness with general tree topology, as is done in rule (1). The failure of rule (19) to accurately capture the distribution of finite and nonfinite predicates in the Dravidian languages stems in large part from an indi¤erence towards basic facts of Dravidian morphology and syntax. That attitude compromises the attempts of HHH’s program at syntactic analysis of the Dravidian languages. Indeed, the program behind (19) manifests little commitment to analyzing these languages on their own terms. As a result, it fails to systematically correlate the Dravidian data with their grammars; it is opportunistic in its choice of data; and it imposes demonstrably flawed analyses on syntactic structures. All these shortcomings suggest that this program is not primarily – and perhaps not remotely – interested in analyzing the Dravidian languages, but perhaps in shielding Indo-Aryan from the possibility of Dravidian influence and in serving a larger typological cause. Since, however, the SOVR does not seriously engage or accurately describe the sentential architecture of any Dravidian language from either a synchronic or a diachronic perspective, it cannot function as a general principle of Dravidian grammar. As a result, it cannot be reasonably expected to serve as a premise in HHH’s argument against the Dravidian Subversion Hypothesis. Rule (1), by contrast, avoids all the pitfalls of the SOVR and rule (19) in describing and analyzing the interaction of finiteness and syntax in the Dravidian languages. Comparison of how the two rules are applied to

What’s So Subversive about Dravidian?

155

Dravidian sentences reveals two fundamental errors that undermine the formulation of rule (19), but not rule (1). First, rule (1) does not link finiteness to coordination or subordination, but operates generally over all kinds of structures. Second, it does not link finiteness to specific constructions, but to the position that a predicate occupies within a constituent tree structure. In avoiding the serious misanalyses committed by rule (19) and in correctly predicting the position of finite and nonfinite predicate, along with their number, rule (1) continues to o¤er a superior platform for the syntactic analysis of Dravidian sentence structure.

5. Dravidian subversion and the SOVR Su‰cient evidence has now accumulated to evaluate the core claim of HHH’s program, viz., that the Strict OV Restriction, because it governed the distribution of finite and nonfinite predicates in both early Dravidian and early Indo-Aryan, eliminates recourse to the Dravidian Subversion Hypothesis as an explanation for the areal convergence of syntactic patterns in South Asia. Simply put, the thorough failure of the SOVR to describe any stage of Dravidian syntax invalidates it as a key premise of that syllogism. The DSH, whatever its merits may be, remains una¤ected by HHH’s argument. This failure naturally brings into question the SOVR’s utility as a typological principle, one which is potentially available for the analysis of natural languages. As noted above, it runs afoul of general structural and functional considerations, which undermine its viability as a typological explanation. First, on general structural grounds there seems to be no necessary correlation between word order and the positions that finite and nonfinite predicates may occupy in a sentence. The opposition of finite versus nonfinite is simply not of the same order as those dimensions of language structure which are conventionally associated with word order variation, e.g., prepositions versus postpostions, relative clauses preceding versus following their heads. Second and on functional grounds, languages with embedded predicate nominals, finite verbs and direct discourse, as are actually found in all stages of Dravidian, pose a serious challenge to the SOVR. Such languages have managed to generate these embeddings, even though the SOVR predicts they should not be able to and, in making such predications, rules out an entire class of attested languages. We should thus

156 Sanford B. Steever seriously question the SOVR as a typological explanation since at this stage of our knowledge, we lack firm evidence that it describes the actual workings of finiteness in any language. As a matter of procedure, any typological principle assumes the prior existence of descriptions of many languages from which it is abstracted in the course of linguistic analysis. But if the SOVR does not emerge from the grammatical analysis of the Dravidian languages, where does it come from? It seems to me to reflect two strands of linguistic thought. Hints of the first appear in the SOVR’s treatment of Dravidian complex sentences as a string of conjoined structures with what we today call a ‘‘flat’’ tree structure with no or little embedding. Here, as in rule (19), parataxis and simple coordination overshadow complex sentence formation and subordination. Such an approach recalls nothing so much as the 19 th century ideology of linguistic stadialism which contrasted ‘‘primitive’’ languages with ‘‘advanced’’ (or ‘‘cultivated’’ or ‘‘civilized’’) ones. Under this pre-scientific classification, primitive languages lacked such cultivated structures as subordinate clauses, indirect discourse or complex embedding, all of which were taken to be the ornaments of the more civilized languages. At some level, such attitudes clearly infiltrated Indology and informed its treatment of South Asian languages so that 19 th century scholars might unashamedly claim that in comparison with refined Sanskrit, the Dravidian languages had nothing as advanced as the elegant relative-correlative construction with its classically balanced periods. Further to the historical assumptions of this ideology, any such structure where found in Dravidian must have been borrowed from Indo-Aryan, and could not have arisen independently from the resources of the more primitive family. Modern linguistics is supposed to have relegated such notions to antiquarian study; however, in arguing that Proto-Dravidian could not have had such linguistically advanced structures as finite subordinate clauses, and in denying that the Dravidian languages could have more than one relative clause-forming strategy, the SOVR appears to be resurrecting a crude, stadialistic typology from the 19 th century, one wisely abandoned in the early 20 th . The second strand of thought implicitly supporting the SOVR appears to derive from attempts in the 1960s and 1970s to force Indic languages into a Procrustean mold of classical generative notational schemes based on the superficial morphosyntax of European languages. This approach held that a rule such as ‘S0 $ [S1 and S2 ]’, where both constituent sentences could have a finite verb, is part and parcel of Universal Grammar and must

What’s So Subversive about Dravidian?

157

apply tout court to the Dravidian languages. In point of fact, (19b) e¤ectively incorporates this position. Generative grammar has since laid to rest the idea that the gross surface combinatoric properties of Western European languages constitute a universal template for all human languages, and so should the analysis of Dravidian syntax. In an unexpected aside, HHH (n.d.) finally observes, ‘‘the phenomenon of serial verbs is of no direct relevance to South Asian convergence.’’ We may at last agree, but with an important proviso. It must be stressed that within the context of Dravidian grammar, the analysis of the SVF necessarily rests on the prior analysis of formal and functional finiteness in these languages. Accordingly, the analysis of SVFs embedded in (1c and d) presupposes the analysis of functional finiteness in (1 a and b). If we take HHH at his word, then from this it follows that the entire analysis of Dravidian finiteness as embodied in rule (1) is of no direct relevance to HHH’s proofs for or against convergence in South Asia. As for the bearing of the Dravidian Subversion Hypothesis on the study of finiteness, the only thing that can be viewed as subversive in this episode, or so it seems to me, is the distorted description and analysis that Dravidian syntactic structures have received under the SOVR. What has been shown here is that the family of analyses implied by the SOVR leads to irreparably flawed descriptions and demonstrably false explanations, and so must be rejected as a basis for the further analysis of Dravidian syntax. Future research on the role of finiteness in Dravidian sentential architecture, including intensive syntactic analysis of the individual languages, will make better headway if grounded in rule (1) and the analysis of finiteness, complex sentence formation and the distinction between coordination and subordination underlying it. After 20 years, that proposal still seems to me to o¤er the greatest promise for exploring how finiteness is utilized in the grammars of the Dravidian languages and how it has changed over time.

Notes * I owe a debt of gratitude to E. Annamalai for discussing with me relative and relative-correlative clauses in Ancient Tamil and Modern Tamil, particularly in terms of their relation to subordinate structures. Thanks also go to two anonymous referees. As a refutation of an alternative proposal, this essay may in certain passages strike some readers as polemical as the tone adapts itself to the logical substance of the linguistic argumentation used. I have done my best to

158 Sanford B. Steever

1.

2. 3.

4.

5.

6.

7. 8. 9.

minimize this. Any remaining shortcomings, factual or rhetorical, are my own. The following abbreviations are used in the interlinear glosses: acc Accusative case, adn Adnominal form, cf Conjunctive form, cnd Conditional form, dat Dative case, f Feminine, fut Future tense, gen Genitive, h Honorific, hrt Hortative, imp Imperative, inc Inclusive, ind Indicative, inf Infinitive, int Interrogative, loc Locative, m Masculine, n Neuter, Neg Negative, nom Nominative case, NPST Nonpast tense, P Plural, prm Permissive, prt Particle, prs Present tense, pst Past tense, s Singular, vn Verbal noun, 1 First person, 2 Second person, 3 Third person, - Morpheme boundary, ¼ Clitic boundary. Nor have I been an ardent proponent of the DSH, if only because my work has focused on the Dravidian languages and their historical development, not the evolution of the South Asian linguistic area as a whole. Further, I have in several instances written against brute-force borrowing as a mechanism in grammatical change (Steever 1981a:61–63; Steever 1987a:114–115; Steever 1988:110; Steever 1993:24), favoring instead language-internal pathways for language change of the kind for which Stampe and Donegan (2004) so compellingly argue. Where found in individual languages, they tend to be marked, e.g. Kannada mattu ‘and’ (see Bhat 1978:543, pass.), or borrowed, as in Kurux. I have updated terminology, for example, replacing ‘command’ by ‘ccommand’. Further, while clauses (a and b) and (c and d) could be collapsed into single clauses, they remain separate here to maintain comparability with the original. This use of ‘embedding’ also applies to nonfinite forms such as the conjunctive, which may be used in coordinate structures (4a) and matrix-subordinate structures (4d). Certain clitics are general between quantifiers and conjunctions; from a semantic perspective, the two di¤er as they apply to a definite description or a set description. However, not all uses of clitics are confined to this generalized notion of quantification; as seen below, some signal emphasis or other pragmatic notions. Though it might appear strange that a grammatical rule can predict the number of finite predicates in a sentence, recall that it is correlated with a set of perhaps a half dozen elements in closed classes that can be enumerated within the given language. In general, both the set of clitics and the set of verb bases in Dravidian are closed. Steever (1988) presents certain refinements which do not a¤ect the present discussion. The modern South Dravidian languages have done away with most, but not all SVFs. Note that in (22a) below, the Bangla form hole ‘if (it) becomes)’ has replaced inherited Dravidian FPEs as represented by Tamil a
n a
l ‘if it becomes’,

What’s So Subversive about Dravidian?

10.

11. 12. 13.

14.

15.

16. 17.

159

Telugu ayite
‘id.’ Since Bangla hole cannot be analyzed into its constituent parts by Kurux word-formation rules, it is borrowed wholesale into the language as an independent word meaning ‘if ’, incidentally helping to establish a new word class of complementizers and conjunctions. Similar changes may be seen in some CDr and SCDr languages on the Dravidian-Indo-Aryan frontier, though not as thoroughly as in NDr. Having studied the Dravidian languages for over 35 years, I am well aware of the flexibility of Dravidian word order; however, Steever (1988) was never intended as a treatise on word order, as deserving as that topic is for book-length treatment. HHH neglects the fact that such harmonic correlations are noted in the sentence after I mention ‘rigid SOV’ word order (Steever 1988:7). This term is now preferred over strict SOV so as not to confuse readers with HHH’s SOVR. Hock (2001:83) overlooks how the category of mur reccam (Steever 1988:50– 52), a concept from traditional Tamil grammar relevant to my analysis of the SVF, bears on the issue of adjacency. A mur reccam is a finite verb form that appears where, on general grounds, a nonfinite verb form is expected. In traditional grammar, it is a variety of eccam, namely a verb form that anticipates a following word, noun or verb, which stands either adjacent or at a remove. Traditional grammarians distinguish the eccam and the form it anticipates from the two parts of compounds such as kol-ya
n ai ‘killer elephant’ by virtue of that fact that additional material may come between a eccam and what it anticipates but not between the members of a compound (Steever 2005:67). In (4a), for example, the conjunctive form po
y ‘going’ is a vin aiyeccam ‘a (nonfinite) verb that anticipates another verb; and it is separated from the following verb vantatu ‘it came’ by the NP pakal ‘daylight’. Thus, any eccam, including a mur reccam, and its following head may – but need not – be separated by intervening material. Steever (1987b) speaks loosely of a constraint that early Dravidian permitted but one finite per sentence, but simplified the exposition for a generalist audience. For clarification, the more precise formulation in Steever (1988) should prevail in specialist uses. HHH overlooks evidence from just the other side of the Dravidian-Indo-Aryan frontier, where quotitivals are found in both Oriya and Marathi (see Bayer 1991:17, 32). There they are viewed as innovations within the grammars of these two languages based on a contact situation, and not as retentions from earlier stages of Indo-Aryan. See Bayer (1991) for an insightful proposal as to how a grammar might accommodate the analysis of the two competing strategies implied by the quotitival. HHH ignores the fact that several of the older relative-correlative clauses Ramasamy (1981:375–77) cites actually employ FPEs, e.g., Ramasamy’s (1) uses a

160 Sanford B. Steever

18.

19.

20.

21.

22.

23.

24.

25.

26.

form of a
ka ‘become’ while (4) uses en ‘say’. Several, e.g., (6) and (9), use the (nonfinite) conditional strategy. For example, the ki-echo Formation (Steever 1993:148–57) is not documented until the literature of the Viras´aiva movement in Kannada and Telugu in the late medieval or early modern period, yet its reconstruction to earliest Dravidian (minus, perhaps, NDr) is guaranteed by its distribution throughout the family. The paucity of clitics in relative-correlatives in certain styles of text may be due to several factors. First, as a marked strategy, it would be unsuitable for poetic discourse, which favors terser, less marked forms such as the adnominal or the conditional strategy. Second, particularly in inscriptions of the medieval period, clitics may have been omitted in imitation of what was perceived as a more ‘‘elegant’’ Indo-Aryan style. The four subgroups are SDr: Tamil, Malayalam, Irula, Toda, Kota, Badaga, Kodagu, Kannada, Tulu; SCDr: Telugu, Gondi, Konda, Kui, Kuvi, Pengo, Manda: CDr: Gadaba, Kolami, Naiki, Parji; and NDr: Kurux, Malto, Brahui. This means scholars should reject as fanciful Hock’s (n.d.) collateral proposal that speakers of Koraga migrated south into Karnataka from a NDr homeland somewhere in Bihar or Jharkand. Examples of this verb stem already appear in Steever (1988:24), including examples (26b) and (27b) there, along with an explicit statement about its allomorphy. In fact, Conjunction Reduction is not available to these Tamil structures. From the time of tolka
ppiyam it has been realized that the conjunction of uyartin ai ‘human’ and ahrin ai ‘nonhuman’ subjects is highly problematic, if not ungrammatical. Since the two exceptions to (19b) are classified as forms of coordinate structures or as two independent sentences, no appeal can be made to them to explain the embedding of finite verbs in subordinate clauses. As in the case of relative-correlative structures, demonstrated below, the availability of Backwards Pronominalization from the second (superordinate) into the first (subordinate) clause is proof that the construction as a whole is best construed as an instance of subordination, not coordination. The possibility of Backwards Pronominalization from the apodosis into the protasis precludes the analysis of any conditional construction as a coordinate structure.

References Agesthialingom, S. and R. Rajasekharan Nair (eds.) 1981 Dravidian syntax. Annamalainagar: Department of Linguistics, Annamalai University.

What’s So Subversive about Dravidian?

161

Bayer, Josef 2001 Two grammars in one: Sentential complements and complemen tizers in Bengali and other South Asian languages. The Yearbook of South Asian Languages and Linguistics, 2001: 11–36. Bhat, D. N. S. 1971 The Koraga language. Poona: Deccan College Postgraduate and Research Institute. 1978 kannad a va
kyagalu: antarı
ka racane mattu ardhavya
vasthe. Mysore: Geetha Book House. Bloch, Jules 1946 Structure grammaticale des langues dravidiennes. Paris: AdrienMaisonneuve. Comrie, Bernard (ed.) 1987 The world’s major languages. London: Croom Helm. 1988 Language universals and linguistic typology. Second edition. Chicago: University of Chicago Press. Donegan, Patricia and David Stampe 2004 Rhythm and the Synthetic Drift of Munda. The Yearbook of South Asian Languages and Linguistics, 2004: 3–23. Emeneau, M. B. 1984 Toda grammar and texts. Philadelphia: American Philosophical Society. Krishnamurti, Bh. 2003 The Dravidian languages. Cambridge: Cambridge University Press. Hock, Hans H. 1988 Finiteness in Dravidian: Review Article. Studies in the Linguistic Sciences 18 (2): 211–233. 1989 Review of The serial verb formation in the Dravidian languages. Language 65 (2): 398–405. 2001 Typology vs. Convergence: The Issue of Dravidian/Indo-Aryan Syntactic Similarities Revisited. The Yearbook of South Asian Languages and Linguistics, 2001: 63–99. (n.d.) The problem of time in South Asian convergence. www.ciil.org/ announcement_MBEprogramme Kachru, Yamuna 1987 Hind-Urdu. In Comrie (ed.) 1987: 470–89. Krishnamurti, Bh. and J. P. L. Gwynn 1985 A grammar of Modern Telugu. Delhi: Oxford University Press. Lehmann, Thomas 1994 Grammatik des Alttamil. Stuttgart: Franz Steiner Verlag. McCawley, James D. 1982 Thirty million theories of grammar. Chicago: University of Chicago Press.

162 Sanford B. Steever 1988

The syntactic phenomena of English. Chicago: University of Chicago Press. Ramasamy, K. 1981 Correlative relative clauses in Tamil. In S. Agesthialingom and R. Rajasekharan Nair (eds.) 1982, pp. 363–380. Steever, Sanford B. 1981a Selected papers on Tamil and Dravidian linguistics. Madurai: Muttu Patippakam. 1981b The pragmatic exploitation of raising to object position in Tamil. In S. Agesthialingom and R. Rajasekharan Nair (eds.) 1982: 247–276. 1987a Remarks on Dravidian complementation. In Studies in the Linguistic Sciences 17(1): 104–119. 1987b Tamil and the Dravidian languages. In Comrie (ed.): 725–746. 1988 The serial verb formation in the Dravidian languages. Delhi: Motilal Banarsidass. 1993 Analysis to synthesis: The development of complex verb morphology in the Dravidian languages. New York: Oxford University Press. 1998 The Dravidian languages. London: Routledge. 2002 Direct and indirect discourse in Tamil. In Reported discourse, A meeting ground for di¤erent linguistic domains, Tom Guelde mann and Manfred von Roncador (eds.), 91–108. Amsterdam: John Benjamins Publishing Company. 2004 Old Tamil. In Roger Woodard (ed.), The Cambridge encyclope dia of the world’s ancient languages, pp. 1015–1040. Cambridge: Cambridge University Press. 2005 The Tamil auxiliary verb system. London: Routledge. Zvelebil, Kamil 1970 Comparative Dravidian phonology. The Hague: Mouton.

Dravidian Syntactic Typology: A Reply to Steever

Hans Henrich Hock

Part of this reply to Steever’s ‘‘What’s so subversive . . .’’ is to correct misinterpretations and misrepresentations of my earlier work in that paper, to clarify some of my ideas, and to correct certain mistakes on my part. More important are two other parts. One of these demonstrates that, pace Steever, relative-correlatives with finite predicates in the relative clause but no postRC clitic must be admitted for the oldest attestations of Dravidian, the geographically northern languages, Koraga, and (optionally) Modern Malayalam. Post-RC clitics, thus, are a southern Dravidian innovation which crosses genetic subgroups. Further consequences of this finding are discussed. Secondly, I map out areas where I believe neither Steever’s account nor mine has yielded adequate insights and where further research is needed. These areas include sentence or clause conjunction and the issue of multiple finite verbs, the regional distribution of Quotatives vs. Quotativals in Dravidian, the prehistory of Dravidian relative-correlatives, and the classification of Koraga within the Dravidian family.

1. Introduction In 1988, Sanford B. Steever (henceforth SBS) published an important monograph on The serial verb formation in the Dravidian languages. As I stated in my review article later the same year (Hock 1988a), Although its title may suggest something quite di¤erent, the monograph under review presents the most systematic investigation to date into the question of finiteness in Dravidian . . . [Steever’s] conclusions provide significant challenges not only for comparative Dravidian linguistics, but for general syntactic typology and for the issue of South Asian convergence.

My review article focused on SBS’s arguments and their empirical foundation in a typological and crosslinguistic framework; fuller discussion of this

164 Hans Henrich Hock alternative account and its implications was left for later publications. The earliest of these is Hock 2001, with major focus on South Asian convergence, an issue further explored in Hock 2005a. Finally, Hock 2005b, which apparently was not available to SBS, discusses in fuller detail the typological implications of SBS’s and my di¤erent accounts, in addition to modifying some of my earlier claims. Throughout, I accepted the appropriateness of SBS’s observations for what I call the modern, geographically ‘‘southern’’ Dravidian languages. What I did question was the empirical adequacy of some of his claims for the oldest attested ancestors of these languages, as well as many of the modern geographically ‘‘northern’’ languages.1 Most important is the fact that in these languages relative-correlative constructions with finite predicates do not exhibit post-RC clitics, contrary to SBS’s claim that these are required in all of Dravidian. See example (1), where (1a–c) illustrate the absence of such clitics in Old ‘‘southern’’ and in ‘‘northern’’ Dravidian languages, and (1d) shows the modern ‘‘southern’’ pattern that conforms to SBS’s theory.2 Along the way I found that Koraga, located far to the south, but claimed to be possibly related to North Dravidian (Bhat 1971), behaves like a ‘‘northern’’ language, which naturally raises interesting questions about the historical developments responsible for this behavior. (1)

a. [e-var‹i nall-avar a¯t
avar]RC [a-var‹i nall-ai]CC which-place good-pl.3m men-pl.3m that-place good-sg.2 (Pur
a 183.3–4; Old Tamil; from Th. Lehmann 1998) ‘You (are) good in a place where the men are good.’ b. [ya¯tontu maha¯ra¯janiyo¯gam]RC [atu a-van
n
am]CC what.n-one maharaja-order that.n.that-manner (Old Malayalam; from Pillai 1973: 165) ‘What is the king’s order, (let) that (be done) in that manner.’ c. [ed et ked
avar d
olnen]RC where where side fall-sg.3m.dur [ad at ked
avar go
lner]CC there there side beat-pl.3.m.dur (Kolami; Emeneau 1961: 168, #172) ‘On whichever side he fell, on that side they beat him.’ d. [ya¯r who.nom

an ke¯ mutalil there first

vantu come.conv

ceru-v-a¯r]RC ¼o¯ arrive.fut.pl.3 ¼ Clitic

Dravidian Syntactic Typology: A Reply to Steever

165

[avar t
iket
t
u van kala¯m]CC they.nom.pl ticket.nom buy.permissive (Modern Tamil; from Annamalai & Steever 1998) ‘Let whoever reaches there first buy the tickets.’ In an earlier publication (Steever 1993), SBS objected to my alternative proposal, claiming that I do not have expert knowledge in Dravidian linguistics. My reply (Hock 2001: fn. 9) was that this is beside the point and that what matters is the linguistic evidence. SBS’s article ‘‘What’s so subversive . . . ?’’ [WSS] evidently is intended as a response to this reply. In spite of what SBS himself acknowledges as its ‘‘polemical’’ tone in the first footnote, WSS serves a useful purpose by initiating a debate, as regards both the empirical evidence and the interpretation of that evidence. I am grateful to him for agreeing to my participation in this debate, and I thank the editor, Rajendra Singh, for making the debate possible. The issues are indeed important and can be briefly summarized as follows. Should we accept SBS’s claim that a single syntactic typology holds true for all of Dravidian, both chronologically and geographically? Or should we accept evidence that suggests diachronic change and synchronic variation? SBS himself has accepted change and variation for at least one phenomenon, namely the ‘‘Serial Verb Construction’’ (SVC), which has been (nearly) eliminated in modern ‘‘southern’’ Dravidian languages but continues in the more ‘‘northern’’, tribal languages. Should we treat Dravidian syntactic typology as a phenomenon limited to, or relevant for, only Dravidian? Or should we attempt to place it in a larger, crosslinguistic typological context? Is the syntactic typology developed in traditional Dravidian scholarship, on the basis of Middle and Modern Tamil, appropriate for other varieties of Dravidian, such as Old Tamil or Kolami? Or should the typology of these languages be an empirical question, to be addressed by the evidence internal to these other varieties? In trying to reconstruct Proto-Dravidian syntax and syntactic typology, is a ‘‘tree model’’ the only appropriate approach? Or do we need to supplement that approach by a ‘‘wave model’’, which allows for innovations to spread across the di¤erent branches of Dravidian?

166 Hans Henrich Hock Before turning to these issues, it is unfortunately necessary to briefly address the fact that, as the presentation in WSS suggests, SBS has misunderstood or misinterpreted many important elements of my earlier publications. To set the record straight and to give a proper foundation for what follows, I present in section 2 a brief clarification of terminology which SBS misinterpreted and in section 3 a brief historical outline of my earlier work and the motivations for that work (which go beyond the question of Dravidian/Indo-Aryan contact). In the interest of keeping this paper focused on what I consider the important issues, I will forgo a detailed discussion of SBS’s misinterpretations and misrepresentations.3 The core of my paper begins with Section 4, which provides a discussion of robust evidence against the empirical appropriateness of SBS’s account for Old Dravidian and the modern ‘‘northern’’ languages with regard to relative-correlative constructions. In the process I show that Dravidian typology has indeed been subject to change. In Section 5, I discuss issues of Dravidian syntax for which my account needs to be reconsidered or revised. Issues that I believe are still unresolved and require further investigation are discussed in Section 6. Section 7 presents conclusions and implications for further research.

2. Clarification of some important terminological issues To judge by his WSS presentation, SBS has had di‰culties understanding some crucial terminological conventions that I have been following in my earlier publications (and which are also employed in this paper). It is therefore important to briefly clarify these terminological issues.

2.1. ‘‘Northern’’ vs. ‘‘southern’’ languages Throughout his paper, SBS fails to understand an important distinction that I make between an areal or geographic classification of the modern Dravidian languages and the traditionally established genetic subcategorization of Dravidian. My areal classification distinguishes (broadly speaking) the more ‘‘northern’’ from the more ‘‘southern’’ languages, based on which subtype of relative-correlative constructions they o¤er. This classification also largely correlates with the preservation of SBS’s Serial Verb

Dravidian Syntactic Typology: A Reply to Steever

167

construction (in the ‘‘northern’’ languages) and the general loss of that construction (in the ‘‘southern’’ languages). The classification is not intended to supersede the traditional genetic classification into South Dravidian (or SD I), South Central Dravidian (SD II), Central Dravidian, and North Dravidian – a classification which, pace SBS, I am well aware of, but which does not correlate with the distribution of di¤erent types of relativecorrelative structures.

2.2. Embedding, subordination, and related issues SBS likewise appears to be at a loss regarding my use of terms such as ‘‘embedding’’ and ‘‘subordination’’, probably because like many others, SBS considers ‘‘embedding’’ and ‘‘subordination’’ to be synonymous. In my publications I have restricted the term ‘‘embedding’’ to structures that are center-embedded or ‘‘physically inserted’’ into their matrix structures, such as English postnominal relative clauses. As a consequence, I characterize as ‘‘non-embedded’’ the conjoined or adjoined relative clauses of relativecorrelative constructions, which are not ‘‘embedded’’ into their matrix clauses. By contrast, I use the term ‘‘subordination’’ to refer to the pragmatic status of a clause in relationship to another one, and hence I may talk about ‘‘pragmatic subordination’’. This terminological use is informed by the fact that clauses may be in a pragmatically subordinate relationship to another clause without being syntactically dependent. (For instance, in traditional Modern German, causal clauses introduced by weil have DC word order, but causal clauses introduced by denn have MC word order; see Hock 1989 for further discussion.) The usefulness of these terminological distinctions can be illustrated by way of common Dravidian syntax. The relative clauses in relativecorrelatives are not embedded, but pragmatically they are subordinate. Converbs, relative participles, etc. are normally embedded (although converbs can be right- or left-dislocated); and the same holds true for Cited Discourse, at least in ‘‘southern’’ Dravidian languages. However, actions indicated by converbs may be pragmatically both subordinate (‘after X did this . . .’) and coordinate (‘X did this and . . .’); and Cited Direct Discourse can hardly be considered subordinate, since its internal syntax is entirely independent from that of the matrix clause.

168 Hans Henrich Hock 3. An outline of my work from Hock 1975 to Hock 2005a,b SBS mistakenly assumes that the motivation of my work is merely to counter the common view that early Indo-Aryan/Sanskrit underwent strong substratum influence (or ‘‘subversion’’) by Dravidian; and this mistaken belief is even reflected in the title of WSS. In fact, however, my work on the Finiteness or Strict OV Constraint in Dravidian grows out of a much broader interest in several di¤erent, but intersecting issues. One of these concerns Proto-Indo-European reconstruction and typology, especially the controversy regarding word order and relative clause formation (see the diametrically opposed views in W. P. Lehmann 1974 and Friedrich 1975). I accepted Watkins’s evidence (1976) that PIE relative constructions were of the relative-correlative type4 and, based on the work by Andrews (1875/1985), Downing (1978), and Chr. Lehmann (1980), argued that relative-correlatives are typologically common in SOV languages; see Hock 1986: 319, 618–620 (a publication that appeared two years before Steever 1988). It is in this context that I first referred to Dravidian as having ‘recently been shown to have an inherited RC strategy of this type’. I further attempted to account for the development of early Germanic relative-clause syntax in terms of prosodic and then syntactic rebracketing of relative correlatives inherited from PIE; see the brief discussion in Hock 1986: 341–344, followed by fuller presentations in Hock 1988b and 1991. A second concern has been the common assumption of prehistoric substratum influence of Dravidian on Sanskrit/Indo-Aryan; see Hock 1975 and 1984 for early discussions. In this context I was especially interested in publications by Lakshmi Bai (1985) and especially Ramasamy (1981) which argued that relative-correlatives are inherited in Dravidian, and not borrowings from Indo-Aryan. The existence of such structures in Dravidian raised important questions both about the traditional view of Dravidian typology, which claimed that only one finite verb can occur in a complex sentence (see e.g. (2)), and the view that Dravidian and pre-Sanskrit IndoAryan/Indo-European had very di¤erent typologies. (2)

In every sentence there is but one finite verb, which is the last word in the sentence, and the seat of government; and all the verbs which express subordinate actions or circumstances, whether antecedent or contemporaneous, assume an indeterminate, continuative character,

Dravidian Syntactic Typology: A Reply to Steever

169

as verbal participles or gerundials, without the need of conjunctions or copulatives of any kind; so that the sense (and more or less the time also) waits in suspense for the authoritative decision of the final governing verb. (Caldwell 1913: 488) I was therefore intrigued and delighted by SBS’s 1988 monograph, which proposed an interesting answer to the typological question (see Hock 1988a, 1989b). Moreover, I felt that with some adjustments, SBS’s account made it possible to account for certain crosslinguistic tendencies in SOV languages, not only as regards the tendency to employ relative-correlatives, but also quotatives, and a range of non-finite subordinating devices. At the same time, for Dravidian languages other than the modern ‘‘southern’’ varieties, I found that his account runs into problems. (See Section 3 below.) The issue of typological di¤erences between Dravidian and Indo-Aryan/ Indo-European and their implications for the claimed subversion of IndoAryan by Dravidian was taken up much later, in Hock 2001, with a summary and discussion of further implications in Hock 2005b. What made SBS’s account especially interesting for me was that, with some modifications, it seemed to make possible a new, typologically grounded account for Sanskrit complex syntax (Hock 1989), especially that of relative-correlatives, an issue that I had been working on even before reading Steever 1988. Eventually, I was able to extend this account to the issue of PIE complex syntax (Hock 1991). In the case of Sanskrit relative-correlatives, I found strong evidence for conjunction, rather than adjunction. For instance, in example (3), the relative clause in the middle contains two relative pronouns, one of which correlates with a correlative pronoun in the preceding clause, the other with one in the following clause. Clearly, under no reasonable syntactic approach could the relative clause be assumed to be simultaneously adjoined to two di¤erent correlative clauses; only a conjunction analysis is able to account for such structures.5 (3)

[sa¯i vai daivı¯ va¯g]CC that.nom ptcle divine speech.nom [yaya¯i yad yadj eva vadati]RC which.instr which-which.acc. ptcle speak.pres.sg.3 [tad tadj bhavati] (BAU 1.3.27) that-that.nom become.pres sg.3.

170 Hans Henrich Hock ‘Divine speech (is) thati (faculty) whateverj one speaks by means of whichi , thatj comes about.’ Q ‘Divine speech is a faculty such that if you speak something by means of it, that comes about.’ The fact that a conjoined analysis thus seems appropriate for Sanskrit led me to assume the same analysis for relative-correlatives in other languages that have them – including Dravidian –, as well as to try to account for SBS’s Serial Verbs as conjoined (at least in origin). I began having some doubts in Hock 2001, where I referred to Dravidian relative-correlatives as ‘non-embedded and finite (quasi-coordinate)’. I further hedged my claims in Hock 2005b, where I stated that a ‘cautious approach’ would characterize relative-correlatives merely as ‘‘non-embedded’’ (i.e. as adjuncts). This, I now believe, is in fact the correct approach for Dravidian and many other languages with relative-correlatives, where counterparts to (3) would be ungrammatical. More important, in Hock 2005b I attempted to provide a typological motivation for the historical developments in Dravidian relativecorrelatives, citing parallels from Burushaski and Basque. In short, my interest in relative-correlatives and other typological tendencies of SOV languages extends far beyond the question of Dravidian subversion of Sanskrit/Indo-Aryan. To characterize it as being confined to or motivated by the latter question is not particularly helpful.6

4. Relative-correlatives: Robust evidence for change As mentioned already, SBS claims that the relative clauses of Dravidian relative-correlatives must be followed by a clitic particle (commonly ¼o¯, but also ¼e¯, or in some languages, yet other particles) and that it is this particle that permits the relative clause to have its own finite predicate. Examples supporting this claim can indeed be found in many modern, geographically ‘‘southern’’ languages, as in (4). Other languages with structures of this sort include the South Dravidian (or SD I) languages Malayalam, Kannada, Toda, and Tulu (–(n)a¯, beside conditional7 –da), as well as in South Central Dravidian (SD II) Telugu and Kon d a. (4)

a. [ya¯r who.nom

an ke¯ mutalil vantu there first come.conv.

ceru-v-a¯r]RC ¼o¯ arrive.fut.pl.3 ¼ Clitic

Dravidian Syntactic Typology: A Reply to Steever

171

[avar t
iket
t
u van kala¯m]CC they.nom.pl. ticket.nom buy.permissive ‘Let whoever reaches there first buy the tickets.’ (Modern Tamil; from Annamalai & Steever 1998) b. [embe nı¯ is
t
am kinid]RC ¼o where you-sg. desire do-sg.2.fut¼Clitic [ba¯n idffla]cc there put.impve ‘Put it wherever you want to.’ (Kon d a; Krishnamurti 1969) However, as first noted in Hock 1988a/1989b, the ‘‘northern’’ Dravidian languages have relative-correlatives without post-RC clitics, as in (5). (5)

a. [ece ka¯r
do puni a¯nat]RC how much time.loc. full-moon become.fut.sg.3.nonmasc. [a¯sen parbu tinaka]CC we.incl festival eat.verbal noun ‘At whatever time the moon is full, (at that time) we must eat the feast.’ (Pengo; Burrow & Bhattacarya 1970, Story 5, #3) b. [emba¯si e¯ barha˜yan
i payinesi]RC who.sg.m this wild-pig.acc.kill.npst.sg.3m [e¯van
aki na¯ ma¯n ga bı¯ha kifflı˜]CC he.dat.sg.m. I daughter marriage do.npst.sg.1 ‘I shall give my daughter in marriage to whoever kills the wild pigs.’ (Kuvi; Israel 1979: 261) c. [ed et ked
avar d
olnen]RC where where side fall.dur.sg.3m [ad at ked
avar go
lner]CC there there side beat.dur.pl.3.m Emeneau: ‘On whichever side he fell, on that side they beat him.’ (Kolami; Emeneau 1961: 168, #172; see also ibid. 105) d. [a¯t cenmor where go.pl2.polite.pres

ale]RC address ptcle

172 Hans Henrich Hock [a¯n bele immod
i verran]CC I emph.pcle you.with go.sg1fut ‘Wherever you go, I will go with you.’ (Parji; Burrow & Bhattacharya, 1953: 114, #77) e. [endran nı¯m ne¯or]RC what-acc. you-pl.nom. ask-fut.pl.2.m. [a¯din nima¯ ge¯ ci’on]CC that-acc. you-pl.dat. give-fut.sg.1 ‘Whatever you will ask, I will give to you.’ (Kurukh; Hahn, 1911: 111) Steever (1988: 35) claims that structures of this sort result from Indo-Aryan influence; and the fact that these languages are geographically closer to Indo-Aryan and show other evidence for Indo-Aryan influence (e.g. coordinating conjunctions such as Kurukh awr) might be considered to support this claim. In their behavior, however, the relative-correlatives of these languages di¤er considerably from their Indo-Aryan counterparts: a. They are always generic, while Indo-Aryan relative-correlatives can also be specific b. As in all other Dravidian languages, relative clauses can only precede their correlatives, while Indo-Aryan clauses can have both the order RC-CC and CC-RC. Moreover, as Steever notes, ‘‘northern’’ Dravidian languages are archaic in terms of retaining Serial Verbs, while the ‘‘southern’’ languages have generally lost them. This makes possible the assumption that the relativecorrelatives of the ‘‘northern’’ languages are likewise an archaism. Further evidence supports this interpretation. First, Malayalam optionally permits clitic-less relative clauses in RC-CC structures, as in (6a). Second, as noted in Hock 2001, Koraga relative clauses are not followed by clitics; see (6b).8 Third, as I first observed in Hock 1988/1989b, the oldest stages of the ‘‘southern’’ Dravidian languages Tamil and Malayalam likewise have clitic-less relative clauses; see (7ab); and as it turns out, Old Kannada can be added; see (7c).9 (6)

a. [e¯t-oruvan dro¯ham ceyyunnu]RC which-one.m evil do.pres

Dravidian Syntactic Typology: A Reply to Steever

173

[avan pa¯pi a¯kunnu]CC he sinner become.pres. ‘He who does evil becomes a sinner.’ (Malayalam; Asher & Kumari 1997: 53, with the comment ‘Note that, as in this case, the particle –oo is sometimes omitted from the first clause.’)10 b. [anna pet
ige kala po¯yerega e¯ti¯ jana ra¯keri¯]RC my box take go.conv how many people flee.pst.pl.3 [a¯ti¯ janakana sipa¯yli¯ bara guddu so much people.dat.(?) soldier come.conv beating pe¯d
od
u]CC put.assertive.sg.3.nonmasc Bhat: ‘Let as many soldiers as there are people who took away my box come and beat them.’ (Koraga; Bhat 1971: 61, §1; similarly p. 60, §8) (7)

a. [e-var‹i nall-avar a¯t
avar]RC which-place good.pl.3m men.pl.3m [a-var‹i nall-ai]CC that-place good.sg.2 ‘You (are) good in a place where the men are good.’ Lit.: ‘At which place men are the good ones, at that place you are good.’ (Old Tamil; Th. Lehmann 1984: 94) b. [ya¯tontu maha¯ra¯janiyo¯gam]RC [atu a-van
n
am]CC what.n-one maharaja-order that.n. in-that-manner ‘What is the king’s order, (let) that (be done) in that manner.’ (Old Malayalam; Pillai 1973: 165) c. [a¯val
pativrate.y.]RC who.nom.sg.f. faithful-to-husband [aval
¼e ma¯nye]CC that.nom.sg.f.¼emph.ptcle venerable ‘Who is faithful to her husband is venerable.’ (Old Kannada; Kittel 1903: 330; similar structures passim)

While for Modern Malayalam it would be a priori possible to assume IndoAryan influence, it is equally possible that it is an archaism, continuing the

174 Hans Henrich Hock clitic-less pattern in (7b). For Koraga, socially isolated in the South, IndoAryan influence is hardly likely, especially given that the two more ‘‘metropolitan’’ languages that have been influencing it, Tulu and Kannada, do not have clitic-less relative clauses. Most important, of course, is the evidence of the Old Dravidian languages. For Old Tamil, Th. Lehmann (1998: 94) specifically states, regarding example (7a), ‘In this correlative construction a finite clause is directly embedded under a head NP without any embedding device; later in Modern Tamil, a correlative clause must be embedded with a clitic such as ¼o¯ or ¼e¯.’ Given this evidence, I have felt justified in assuming that early Dravidian did not have relative clauses followed by clitics, that the occurrence of cliticless structures in Modern Dravidian is an archaism, and – most important for present purposes – that post-RC clitics in the modern ‘‘southern’’ languages are a (regional) innovation. SBS counters my interpretation with several arguments (WSS §3.7). First, he claims that RCs with following clitics must be reconstructed for Proto-Dravidian: Just on the basis of cognate structures from Tamil, Toda, Kannada, Telugu and Kon d a, representing SDr and SCDr, the ¼o¯ relative-correlative may be projected to at least the period before these two subgroups split, at a minimum time depth of 1500 BCE . . .

This argument would hold only if post-RC clitics were in fact attested in Old Tamil, Old Malayalam, and Old Kannada (an issue that will be addressed presently). Since they are not, the presence of post-RC clitics in modern ‘‘southern’’ languages must be explained di¤erently, as a relatively late, regional innovation, spreading across – and ignoring – genetically established subgroups, such as South Dravidian (SD I) and South Central Dravidian (SD II). In fact, other innovations have similarly spread across the subgroups. This is the case for changes a¤ecting alveolar stops and retroflex nasals and laterals (Krishnamurti 2003: 177). The case is similar for the complex changes and variations involving the development of c > s > h > fl. The core area seems to be SD I plus Telugu; but the change also includes all of SD II, as well as part of Central Dravidian (Kolami, Naiki, and Gadaba); see Krishnamurti 2003: 498 with 122–123. In morphology, we similarly find that the perfective participle –(c)ci of SD II is also found in

Dravidian Syntactic Typology: A Reply to Steever

175

part of Central Dravidian (Parji, Ollari, and Gadaba); Krishnamurti 2003: 499. SBS continues (WSS §3.7) as follows: Pursuing his argument, HHH turns Lehmann’s (1989: 349, pass.) statement that clitics are obligatory in Modern Tamil finite relative-correlatives on its head in an attempt to claim that they were absent in the ancient language . . . The ¼e¯ relative-correlative clause in (15) [cited here as (8)], from purana¯n u¯ru 301: 7–8, is clear counterevidence. Not only does it resemble the ¼o¯ correlative in all details up to the identity of the clitic, it is also the direct ancestor of the ¼e¯ relative-correlative clause of Modern Tamil . . . (WSS, §3.7)

(8)

en ai na¯l
tan kum num po¯r¼e¯ how.many day last-NPST-3SN your war-NOM¼EVEN an ai na¯l
eriya¯r erital ya¯van
. . . that.many day battle-NEG-3P-NOM battle-VN how Steever: ‘How can one battle those for even one day when their army remains in camp every day and will not fight?’ i.e., ‘On which day can our king fight those whose army remains in camp every day and will not (come to the battle-field to) fight?’ (SBS’s glossing)

Even a priori, SBS’s example is of dubious cogency, for the clitic ¼e¯ here could just as well be in the scope of the preceding NP (num po¯r) as in the scope of the entire (putative) relative clause; compare the en.k ka¯talo¯r¼e¯ of example (9) below, where the clitic attaches to an MC noun phrase. What is more, it is di‰cult to see a connection between Steever’s translation (or better, interpretation) and the glosses of the individual elements of the utterance. A better interpretation would seem to be ‘How is there a fighting against those who don’t fight for even so many days as your war lasts?’ But this doesn’t make much sense either. Thomas Lehmann (p.c. 2008) informs me that most commentaries take the passage to consist of two questions, with the first part to be translated as ‘How long will your war last?’ and the second part as ‘During these days, where will be the battle against those who do not attack?’ In fact, Thomas Lehmann (p.c. 2008) points out that many putative examples of Old Tamil relative-correlatives, cited by Ramasamy (1981), are

176 Hans Henrich Hock problematic. This includes the example that I cited in 1988a and 1989b as evidence for clitic-less RCs, see (9). (9)

[min n it
ai nat
unkun kat
ai.p-peyal va¯t
ai ekka¯l varuvat(u) en ri]RC what-time come-npst.3sg.n. say-npst.2sg [akka¯l varuva[r] en.k ka¯talo¯r¼e¯]CC that-time come-npst.pl.3hon. my lover¼Clitic Ramasamy’s transl.: ‘My lover will come at the time when you say that the cold wind will set in.’

As Lehmann notes, the form en ri creates problems for my analysis (and the implicit analysis by Ramasamy) and makes it unlikely that what precedes is a relative clause. This favors the interpretation by most of the commentaries of (9) as containing a question ‘At what time do you say will the north wind come?’, followed by an answer ‘At that time my lover will come.’11 After eliminating all problematic examples, Lehmann comes to the conclusion that unambiguous instances of relative-correlatives are extremely rare in Old Tamil, amounting to about five (or ten, if Post-Sangam, Late Old Tamil is included). More significant for present purposes, Lehmann notes that all unambiguous Old Tamil examples have clitic-less relative clauses. To judge by the data in Kittel (1903), the same holds true for Old Kannada. There is thus no evidence for SBS’s claim that post-RC clitics are a feature of all of Dravidian, at all of its chronological stages. Rather, the evidence favors my view, that post-RC clitics are a relatively late, regional innovation in (most of ) ‘‘southern’’ Dravidian.12,13 In Hock 2005b, I attempted to account for this innovation in terms of typological pressure. The assumption is that in Old Dravidian and in the Modern Dravidian languages that have relative-correlatives without postRC clitics, the finite verbs (or predicates) of relative clauses are exempt from the Finiteness Constraint by the fact that they are not embedded into their ‘‘matrix’’ clauses.14 While this exemption is signaled only negatively in these languages, by non-embedding, the modern ‘‘southern’’ languages have introduced a more robust signal, namely the post-RC clitic, whose function is similar to that of the Quotative marker. Parallels for this development are found in Burushaski (Berger 1988) and Basque (Hualde & Ortiz de Urbina 2003); and since publishing Hock 2005b I have come across

Dravidian Syntactic Typology: A Reply to Steever

177

a third typological parallel, in Nagamese and Mongsen Ao (see Coupe 2007). Let me conclude this section by summing up my major findings: a. The relative clauses of Old Tamil, Old Malayalam, and Old Kannada are (left-)adjoined to their correlative clauses b. The same holds true for ‘‘northern’’ Dravidian languages, as well as Koraga and (optionally) Modern Malayalam c. While SBS’s typology accounts for the modern ‘‘southern’’ Dravidian languages that have post-RC clitics, it fails to account for the other languages; a di¤erent account, with (left-)adjunction, is required for these d. The use of post-RC clitics in most of modern ‘‘southern’’ Dravidian is an innovation, which can be typologically motivated and has outside parallels e. The innovation is a regional one that cuts across the boundaries of Dravidian sub-families f. The assumption that a single typology (the one postulated by SBS) holds for all of Dravidian, at all times, cannot be sustained; rather, as to be expected in natural languages, the typology has been subject to diachronic change and regional variation.

5. Earlier claims of mine that need to be reconsidered As noted in the preceding section, my interpretation of the passage in (9) needs to be reconsidered. Similarly, I noted in section 2 that my earlier interpretation of Dravidian relative clauses without post-RC clitic as conjoined needs to be reconsidered and that an adjoined analysis is more appropriate. In this section I briefly discuss several other issues where my earlier interpretations and claims need to be reconsidered. One of these concerns the fact that I usually talk about Finiteness Restrictions in terms of verbs rather than predicates in general. Similarly, in my enumeration of non-finite subordination devices I failed to explicitly include conditionals; note however that in Hock 1988a and 1989b I explicitly set aside Old Tamil relative-correlatives with conditionals in the relative clause, precisely because conditionals are non-finite. At any rate, neither of these two problems has any major e¤ect on the relevant arguments.

178 Hans Henrich Hock More serious is my failure to consider embedded structures with finite verbs followed – and licensed – by predicates other than quotative en -. All I can say in my defense is that my focus on quotative marking has been motivated by my crosslinguistic typological interests, as well as my interest in the question of South Asian convergence. Note, however, that for the modern ‘‘southern’’ Dravidian languages and, by implication, for Old Tamil, I accept SBS’s account that assumes that embedded structures of this sort are licensed by what he calls FPEPs. It is only for the modern ‘‘northern’’ languages, as well as Koraga, that I o¤ered a di¤erent, speculative account in Hock 2001 (This issue is addressed in section 6.2.)

6. Issues that require detailed further investigation In this section I take up several issues which, I believe, have not been adequately accounted for either by SBS or by me and which, therefore, deserve further detailed investigation.

6.1. Serial Verbs, multiple finite verbs, and coordination Throughout his paper, especially in §3.10.3, SBS attacks as uninformed my attempt to deal with the issue of multiple finite verbs, including Serial Verbs, in simple and complex sentences (other than relative-correlatives and structures embedded by means of SBS’s ‘‘FPEs’’). As it turns out, my earlier views on this issue were colored by my assumption that the notion of ‘‘conjunction’’ is appropriate for dealing with the distribution of finite verbs or predicates in Dravidian, but as noted in Section 4, I no longer hold this position as regards relative-correlatives. So, instead of trying to defend the details of my earlier position against SBS’s objections, I believe it is more productive to focus on the general issues that I still consider important and that require further investigation. These issues are of two sorts. First, there is the general question of syntactic (rather than merely pragmatic15) conjunction or coordination and its possible relationship to the Serial Verb construction. Second, there is the more specific question as to whether the ‘‘Standard’’ analysis in Tamil grammar is appropriate for Old Tamil or, more important, for other Dravidian languages, especially the ‘‘northern’’ ones.

Dravidian Syntactic Typology: A Reply to Steever

179

By way of background, it is useful to outline the relevant assumptions in the ‘‘Standard’’ analysis, based on Tamil. The most important premise is the idea that, outside relative-correlatives and structures embedded by means of FPEs, only one finite predicate can occur in each sentence. The occurrence of apparent exceptions in Serial Verbs (10a), as well as in Old Tamil sentences such as (10b–d) is accounted for by the notion mur
r
eccam, a term which can be glossed as ‘a finite verb in the function of a non-finite verb’. (See Steever 1988: 50–52, Th. Lehmann 1994: 128–130.) (10) a. cel-v-e¯m all-e¯m go.npst.pl.1 become.neg.pl.1 ‘We will not go.’ Lit. ‘We will go, will not do’ (or the like) (Old Tamil; Steever 1988: 42) b. . . .

pa¯yal per-e¯n . . . tuyar ur‹a-pp-e¯n sleep get.neg.sg.1 sorrow su¤er.npst.sg.1 ‘I su¤er sorrow without getting sleep.’ Lit. ‘I do not sleep . . . I su¤er sorrow.’ (Old Tamil; Th. Lehmann 1994: 129)

c. . . .

pa¯t
-in -ai celin ir‹ai peru-ku-v-ai sing.pst.sg.2 go.cond. jewel get.npst.sg.2 ‘If you sing (and) go, you will receive jewelry.’ Lit. ‘You sang, if you go, you will receive jewelry.’ (Old Tamil: Th. Lehmann 1994: 130)

d. . . .

maliv-an -am marukk-i . . . nir‹al be-full.npst.euph.pl.1 wander.conv. shade iru-nt-an -am remain.pst.euph.pl.1 ‘. . . (we) were filled (with desire), wandered (around), and remained in the shade.’ Lit. ‘(we) are filled (with desire), wandering, remained in the shade.’ (Old Tamil; Th. Lehmann 1994: 129)

The second major assumption is that the postpositive coordinating conjunction –um cannot be used to coordinate finite verbs or predicates and, by extension, cannot be use to coordinate clauses within a sentence. Moreover, it

180 Hans Henrich Hock is assumed that coordinating –um must follow each of the coordinated elements and thus is di¤erent from –um ‘also, even’, which can follow just one constituent.16 (See e.g. Beythan 1943: 154–155.) Even a cursory glance at the two first examples in (10) will reveal a difference. In the case of (10a) there can be no reasonable doubt about the two verbs belonging together in a single sentence or predication. In (10b), by contrast, it is not at all clear why the proposition cannot be interpreted as two coordinated sentences or clauses. Similarly, in (10c) it would be possible to have a coordinate interpretation, ‘You have sung; if you go (now), you’ll get jewelry’. In some cases, however, such as (10d), such an interpretation becomes more di‰cult, since the non-past tense on the first finite verb (maliv-an -am) would have to be interpreted as either present or future, an interpretation that would clash with the past tense of the putative Main Verb (iru-nt-an -am). If, on the other hand, iru-nt-an -am is interpreted as equivalent to a non-finite converb, tense presumably is neutralized and hence there is no tense clash. Moreover, as Th. Lehmann notes, in the nonpast, the ‘‘euphonic’’ su‰x –an - is limited to mur
r
eccam-use (1994: 129). Even so, questions remain. This is most obviously the case for structures such as (10bc), where we might wonder whether the mur
r
eccam interpretation imputes to Old Tamil a typological perspective that may be appropriate for later Tamil – which no longer has such a use of finite verbs – but not for the early language. Even in cases such as (10d) we might toy with the idea that the use of the su‰x –an - simply makes the finite verb tenseless and hence avoids tense clash in a syntactically coordinate construction. As it turns out, there is ample evidence suggesting that a reexamination of the ‘‘Standard’’ Tamil-based analysis of multiple finite verbs and syntactic coordination may be appropriate. Let us start with SBS’s Serial Verbs. While in structures such as (10a), repeated as (11a), the first verb can be said to be subordinate to the second, in that negation dominates the construction, it is more di‰cult to make such a claim for ‘‘Balance Verb’’ structures such as (11b); for here the two verbs in question are pragmatically coordinate, and neither dominates the other. (11) a. cel-v-e¯m all-e¯m go.npst.pl.1 become.neg.pl.1 ‘We will not go.’ Lit. ‘We will go, will not do’ (or the like)

Dravidian Syntactic Typology: A Reply to Steever

181

b. . . .

vizu aya kulur pot
i tinad un
ad all that crane bird eat.npst.sg.3n. drink.npst.sg.3n. ‘. . . that crane consumes everything.’ Lit. ‘that crane eats, drinks everything’ (Kon d a; Steever 1988: 70)

As SBS himself realized (1998b: 296), ‘certain examples in Andres’ corpus [of Gond i] suggest that complex sentences can be formed paratactically . . .’, with multiple finite verbs; see (12). SBS adds that ‘[s]uch exceptions to the rules governing the distribution of finite predicates are limited; the devices that permit them are easily enumerated . . .’ (296–297), but does not further pursue the issue. (12) undi natttur’torom undi pen’torom undi put
t
ul’torom one blood.gen.pl.1 one god.gen.pl.1 one o¤spring.gen.pl.1 SBS’s translation: ‘We are of one blood, one god and one ancestral line.’ Finally, there are structures such as those in (13), which suggest that the phenomenon of complex sentences with multiple finite predicates may be even more widespread. (13) a. buga buga icat u¯mtat say.pst.sg.3n fly.pst.sg.3n. rayla mato kuckatat rayla tree.loc. settle.pst.sg.3n. ‘It made (the sound) ‘‘bugabuga’’, flew about, (and) settled on a Rayla tree.’ (Pengo; Steever 1988: 78) b. man itar ar‹inta¯rkal
man.pl.nom.m. perish.pst.pl.3rational kutiraikal
(-um) ar‹intan a horse.pl.nom.n.(-&) perish.pst.pl.3n. ‘Men perished and horses perished.’ ¼ ‘Men and horses perished.’ (Tamil; Beythan 1943: 179) c. twı¨- m e¯r-m batsˇpini. ¨ıt
s
pini. two-day-funeral-& put.pst.sg.1 bu¤alo-& catch.pst.sg.1

182 Hans Henrich Hock ko¯r
-m bı¯ yodsˇpini funeral-& help perform.pst.sg.1 Emeneau: ‘I have made the two-day funeral, and have caught the male bu¤alo, and have helped you perform the funeral.’ (Toda; Emeneau 1984: 680) SBS classifies structures like (13a) as ‘co-ordinate verb constructions’ (1988: 77), Serial Verb structures he defines as ‘most closely approximat[ing] . . . the compound verb construction’ (1988: 39). However, this is stretching the definition of ‘‘Serial Verb’’ to a degree that leaves me to wonder whether it simply serves to explain these structures away. As for (13b), SBS analyzes it as two sentences and translates the –um of the second part as ‘also’ (WSS §3.10.3). This ignores the fact that, as Beythan (1943: 197) observes, the structure serves to avoid gender agreement between masculine/rational and neuter/irrational and hence takes the place of the single-sentence proposition ‘Men and horses perished’.17 If, then, this is a coordinate construction, we may have to recognize that –um can be used to conjoin clauses. Concerning (13c), SBS argues that my interpretation ignores the punctuation in Emeneau’s example, which suggests that we are dealing with a series of separate sentences (WSS §3.10.3), and he translates the –m as ‘also’. However, Emeneau states that –m here is ‘added to the subjects or objects of verbs in order to coordinate sentences’ (1984: 68; my emphasis), and his translation treats the passage as a single proposition, as indicated by its punctuation. I suspect that Emeneau’s punctuation of the Toda passage reflects his theory of Dravidian syntax which is more clearly laid out in Emeneau 1961: 105, where he defines predications (i.e. sentences) as containing one finite verb. This, of course, reflects the ‘‘Standard’’ Tamil-based account; but again, it is legitimate to ask whether this account is appropriate for languages other than (Middle or Modern) Tamil – or even for that language (see example (13b)). As I see it, the important question to ask is this: Is it appropriate to interpret the structures in (13) and others like them as paratactic or as conjoined sentences, just because the ‘‘Standard’’ view is that, with certain exceptions such as Serial Verbs (as well as relative-correlatives and structures embedded by means of FPEs), a sentence contains only one finite predicate? Or should we try to find some independent criteria that make it possible to distinguish between sentential and clausal coordination? If we adopt the

Dravidian Syntactic Typology: A Reply to Steever

183

former view, there is really no way to falsify the analysis; any structure that would cause problems for the view would automatically be interpreted as containing separate sentences. Adopting the latter view at the least would potentially stimulate further research, perhaps in terms of the prosodic organization of such utterances. In this regard it is interesting that in his Koraga texts, Bhat (1971) does not employ periods, but rather uses slashes and that, moreover, these slashes do not coincide with the sentential organization called for by the ‘‘Standard’’ analysis; see e.g. (14). Although Bhat does not clarify his use of slashes, I suspect that they are intended to indicate some kind of prosodic ‘‘chunking’’. (14) naramani ru:pa:nd
¯i / mide miya¯dara po¯no onas human-form is-shaped then bathe-conv. go-sg.3habit. meal ma¯nno banno / make-sg.3habit. come-sg.3habit. Bhat: ‘And a human form shape gets formed. After that he goes to take a bath, takes his meal and comes back.’ (Koraga; Bhat 1971: 62)

6.2. Quotatives and quotativals Starting with Hock 1982 (§17) and 1988 (§4.3) I have noted that Direct Discourse (DD) in the ‘‘northern’’ languages is not embedded by means of a post-DD Quotative marker, as in the ‘‘southern’’ languages, but may be left unmarked altogether, or preceded and/or followed by verbs of speaking (SPEAK). In Hock 2001 I introduced the term ‘‘Quotatival’’ for structures of this sort which lack ‘the fixity of full quotative construction’ (2001: §1). Unfortunately, at the beginning of the detailed discussion of these structures (§4.2.1), I referred to these constructions as employing ‘a preposed finite verb of speaking (optionally preceded by a particle meaning ‘thus’), and a second, postposed finite verb of speaking’. It is only later in the section that I added the remark that ‘(i)n addition to structures with the double finite construction, the northern Dravidian languages, plus Koraga, o¤er a great variety of other structures – another sign of lack of fixity.’18 In his discussion of the issue (WSS §3.6), SBS focuses exclusively on the preliminary definition of Quotativals as employing finite verbs of speaking both before and after DD, without considering the broader issue, namely

184 Hans Henrich Hock the lack of fixity. As a consequence, his response to my earlier arguments is incomplete, in addition to being problematic on other grounds. At the same time, he does raise important questions that deserve consideration. What is at issue is that in Hock 2001 (§4.2.3) I proposed a ‘speculative account of how the southern quotative can be derived from the quotatival double finite construction of northern Dravidian’, where the operative word is ‘‘speculative’’. I was fully aware that this account had problems, especially since it had to assume that the change took place prior to even the earliest stages of the ‘‘southern’’ languages, i.e. at least 2000 years before the attestation of the ‘‘northern’’ languages. Moreover, I noted the problem that even in the Quotatival constructions of the ‘‘northern’’ languages, the most commonly used verb of speaking is *en -, the same verb that is used in the Quotative marker of (most of ) the ‘‘southern’’ languages. In my account, then, the ‘‘northern’’ Quotatival structures are archaism, and the ‘‘southern’’ Quotatives innovations. Focusing on structures with finite verbs of speaking both before and after DD [‘‘embracing’’ structures for short], SBS instead proposes that these are innovations, a transitional stage from the inherited Dravidian construction . . . , where the quoting verb follows the quoted material, through a ‘‘quot[a]tival’’ stage, where quoting material brackets the quoted material fore and aft . . . , to a final stage in some NDr and CDr languages, where the quoting verb now exclusively precedes the quoted material in imitation of Indo-Aryan . . . (WSS §3.6)

In this context he points to similar embracing structures in Oriya and Marathi, on ‘the other side of the Dravidian-Indo-Aryan frontier’, with reference to Bayer 1991: 17, 32, as a parallel development (WSS, note 15), with the comment that I ‘overlooked’ this parallel. Further, he claims that the ‘so-called quot[a]tivals appear only in three languages to the south of the Dravidian-Indo-Aryan frontier, Kuvi, Kolami and Parji; in none of them is it a dominant or unmarked embedding strategy.’ (WSS §3.6) Even taken at face value, with its limitation to embracing constructions, SBS’s account is problematic. First, it ignores the fact that Koraga, far to the south, also has constructions of this type; and as noted in Section 4, Koraga is unlikely to have been subject to Indo-Aryan influence. Second, the kinds of embracing constructions that Bayer refers to are di¤erent from the ‘‘northern’’ Dravidian ones in question, for they constitute a

Dravidian Syntactic Typology: A Reply to Steever

185

compromise between two di¤erent marking strategies, one with a pre-DD complementizer (ki or je/jo), the other with a post-DD Quotative marker based on a verb of speaking. As noted in Hock 1982, such compromise structures are also found in Vedic Sanskrit (in the form yatha¯ [relative adverb ‘so that’] DD iti [Quotative marker]). The embracing structures of the ‘‘northern’’ Dravidian languages cannot be explained in this way, as compromises between a complementizer or relative adverb and a Quotative marker. Third, the elements involved in these embracing constructions normally do not include a non-finite Quotative marker, but usually are both finite verbs of speaking. Most important, embracing structures are only a subset of the ‘‘northern’’ Dravidian Quotativals and, moreover, the larger phenomenon of Quotativals is not as limited as SBS believes. Before pursuing this issue further, it is useful to provide a brief survey of the variety of devices employed in ‘‘northern’’ Dravidian languages to set o¤ DD; see the Kuvi examples in (15) with references to Israel 1979.19 (15) a. SPEAK-fin. þ DD þ SPEAK-nonfin. þ SPEAK-fin. . . . bur
ha on
pitesi ke [DD] old man think-sg.3m.pst narr.ptcle injihı˜ ke on
pitesi ke en .conv ptcle. think-sg.3m.pst narr.ptcle ‘The old man thought ‘‘DD’’ (saying he thought).’ (Story 6, #5) b. SAY-fin. þ DD þ fl . . . ele ice [DD] fl thus say-sg.3.nonmasc.pst ‘. . . she said ‘‘DD’’.’ (Story 1, #9; a rare type in most languages) c. DD þ SPEAK-fin. . . . bur
hi [DD] ice old woman say-sg.3.nonmasc.pst ‘The old woman said ‘‘DD’’.’ (Story 1, #5) d. DD þ SPEAK-nonfin. þ SPEAK-fin. [DD] injihı˜ rı¯skatesi en .conv ask-sg.3m.pst

186 Hans Henrich Hock ‘ ‘‘DD’’ (saying) he asked.’ (Story 2, #17; a type which is found occasionally) e. No SPEAK either before or after DD fl [DD] fl (Story 6, #54–55) (That is, direct discourse without preceding or following verb; a rare type in Kuvi, but common in Parji and Koraga dialogues.) The frequency with which these di¤erent structures occur in di¤erent languages varies. For Kolami, Emeneau (1961: 113–114) observes that a verb of speaking (most commonly from the root en -) precedes DD 35 times, follows 10 times, and embraces DD 9 times (with a demonstrative adverb intervening in two cases). Statistics for other languages, such as Kuvi, Parji, and Koraga, are still a desideratum. To attribute this variety of Quotatival constructions to Indo-Aryan influence would be di‰cult, unless it could be shown that neighboring IndoAryan dialects o¤er the same variety (and even then, it would be di‰cult to establish the direction of influence). Moreover, it would be necessary to explain how Koraga, far removed from Indo-Aryan, might have acquired similar Quotatival structures through contact. At this point, then, the best conclusion would be to accept Quotativals as indigenous to ‘‘northern’’ Dravidian, as well as Koraga. This, however, does not necessarily prove my speculative claim of 2001 that Quotativals are an archaism and the ‘‘Standard’’ Quotatives of the ‘‘southern’’ languages are an innovation. It is at least possible that the languages with Quotativals lost an original Quotative strategy and that their Quotativals are an innovation.20 As stated in Hock 1982, the latter account would gain in plausibility if the post-DD particle ay of Malto can be demonstrated to be related to the verb en - which is the basis for the ordinary Quotative marker of the ‘‘southern’’ languages. Moreover, there is the di‰cult issue of how to syntactically account for the Quotativals of ‘‘northern’’ Dravidian and Koraga. For the ‘‘southern’’ languages, of course, SBS’s account works well. But the ‘‘northern’’ languages evidently don’t need an FPE to embed cited discourse. But is this embedding a discourse phenomenon or can it be dealt with as a matter of complex-sentence syntax? As in the case of multiple finite verbs, then, there is ample room for further research.

Dravidian Syntactic Typology: A Reply to Steever

187

6.3. Relative-correlatives revisited In the course of discussing the relative-correlatives of Old Tamil with Thomas Lehmann, I learned that clear examples of relative-correlatives not employing (non-finite) conditionals can be found only for nominal predicates, but not for verbal ones. In this context, the following remark by Lehmann, regarding example (7a) above, acquires special significance As Ramasamy (1981) shows, the correlative strategy is used where a variety of constraints prohibit the use of the adnominal strategy. It is thus an integral part of Tamil syntax. . . . Note that the correlative strategy must be used [in this example] because the subordinate predicate a¯t
avar ‘men’, as a predicate nominal, lacks the adnominal form, a verb form, used in the adnominal strategy.

While Lehmann here focuses on the lack of a relative-participle formation for the nominal predicate a¯t
avar [read nallavar ?], it seems equally true that nominal predicates lack a conditional. This suggests a possible explanation for the apparent asymmetry between relative-correlatives with nominal vs. verbal predicates: Verbal predicates can appear in (non-finite) conditional form and thus need not be finite; nominal predicates, lacking conditional forms, can only appear in finite (i.e., unmarked) form. If correct, this would raise interesting questions for the (pre-)history of Dravidian relative-correlatives, which deserve further investigation. For the time being, it is worth noting that unlike Old Tamil, Old Kannada does have finite verbal predicates in relative-correlatives without postRC clitics; see (16). However, given the later attestation of Old Kannada compared to Old Tamil, it is unclear whether this structure should be considered an archaism or an innovation. (16) [basavan embaþ aks
ara triyavan u osedu cittadol
u Basava en
.rel.pple aksara-triplet.acc. joyful mind.loc. ormeyum dhya¯nisuvan u a¯van]RC once¼concessive contemplate.npst.sg.3m who.nom.sg.m. [avana vadanasa¯rasa.v¼e neleya þ a¯g(u)-irpudu]CC that.gen.sg. face-lotus¼emph.ptcle steady become-be.npst.sg.3.n. ‘Who contemplates the three syllables Ba-sa-va(n) once with joyful mind his lotus face indeed will remain steady.’

188 Hans Henrich Hock 6.4. The position of Koraga within Dravidian SBS argues against Bhat’s suggestion that Koraga may be a North Dravidian language in origin and my tentative acceptance of this claim (WSS §3.10.1, with Note 21). Rather, like Krishnamurti (2003: passim) he classifies it as a South Dravidian (SD I) language. In fact, Krishnamurti repeatedly claims that Koraga is a recent o¤-shoot of Tulu and dismisses what I consider Bhat’s strongest argument for a‰liation with North Dravidian, namely the use of a past-tense su‰x with velar stop. According to Krishnamurti, ‘this is an innovation in Koraga of giving a past meaning to a nonpast su‰x just like Kota, which uses – p- (< *-pp), a non-past marker, as a past marker.’ (2003: 300) The latter argument, of course, is speculative at best; and the many features in which Koraga agrees with (or is similar to) Tulu as well as Kannada, can be explained in terms of the fact that, as Bhat (1971) has demonstrated, Koraga has undergone extensive influence by these two languages, especially as regards its lexicon. What neither Krishnamurti nor SBS seems to have appreciated is that the agreement system of Koraga is very di¤erent from that of Tulu, Kannada, or the other South Dravidian (SD I) languages. While the latter make a distinction in the third person singular between masculine, feminine, and neuter,21 Koraga only distinguishes masculine from non-masculine (¼ feminine and [human]); see e.g. Bhat 1967 for Tulu and Bhat 1971 for Koraga. In this respect, Koraga agrees with the Type III gender system of North Dravidian and Telugu (for which see Krishnamurti 2003: 207–212), a fact that might be considered to further support a North Dravidian a‰liation. (In fact, Bhat (1971: 3) takes it that way.) However, since Krishnamurti plausibly argues that this is the original Dravidian gender system and since North Dravidian and Telugu preserve that system independently, it is certainly possible that the occurrence of this system in Koraga is likewise an independent preservation. Nevertheless, both in its past-tense marking and in its gender system, it would be wrong to claim a special a‰liation of Koraga with Tulu (or Kannada). In fact, if it is a South Dravidian (SD I) language at all, the presence of these two features would highly mark Koraga as a maverick. Add to this the fact that Koraga, like the more ‘‘northern’’ languages, has relative-correlatives without post-RC clitics (see Section 4) as well as

Dravidian Syntactic Typology: A Reply to Steever

189

Quotativals (Section 6.2), and the case for considering Koraga as a maverick becomes even stronger. (Neither of these two features was considered either by Krishnamurti or by SBS.) Does this ‘‘prove’’ a ‘‘northern’’ or North Dravidian a‰liation of Koraga? Perhaps not; a larger amount of evidence might be needed. However, one thing should be clear: The assumption that Koraga is a South Dravidian (SD I) language, closely related to Tulu, is in need of reconsideration. Here too, then, further research is needed.

7. Conclusions and implications for further research We have come a long way since the days when Caldwell could characterize Dravidian syntactic typology as permitting only one finite verb per sentence (see (2) above), with implicit acknowledgment that quotative constructions may have their own finite verbs (but without attempting to propose a principled account for this), and with relative-correlatives commonly attributed to Indo-Aryan influence (see the incomplete catalogue of bibliographical references in Hock 1988a). Important contributions to overcoming this highly simplified view were made by Lakshmi Bai (1985) and especially Ramasamy (1981), who convincingly argued that relative-correlatives are indigenous to Dravidian. But, as noted earlier, the major break-through came with Steever 1988, which made a first and thorough attempt to account for all the apparent violations of the single-finite-verb constraint, including not only quotative constructions and other similar structures licensed by other FPEs, and relative-correlatives, but also the question of Serial Verbs and the related issue of mur
r
eccam. The fact that I have questioned the empirical appropriateness of some of his claims for Old Tamil, Old Malayalam, Old Kannada, the ‘‘northern’’ tribal languages, Koraga, and (to some extent) Modern Malayalam should not distract from the significance of SBS’s work. In fact, as I noted in Hock 1988a, it is precisely SBS’s work that inspired me to attempt a more comprehensive investigation of these languages and to propose alternative ways of accounting for their syntactic behavior. It is unfortunate that my attempt has evidently rankled SBS and led him to write his ‘‘polemical’’ attack. I hope that my response will go some way toward initiating a less passionate exchange of ideas.

190 Hans Henrich Hock In comparison to other, more ‘‘privileged’’ language families, including Indo-Aryan, the Dravidian languages and their history have attracted much fewer scholars, especially in the area of pan-Dravidian and diachronic syntax. This is unfortunate, since linguistic subfields benefit from the vigorous, sometimes ‘‘agonistic’’ exchange of ideas that comes with a larger number of scholarly participants. It is also unfortunate because, as I hope to have shown, there is ample room for further detailed research – in the area of sentential or clausal coordination, the issue of di¤erent strategies for discourse or sentence embedding of direct speech, the history and development of relative-correlatives in early Dravidian, or even in the genetic and areal classification of Dravidian languages. Beyond raising these questions, I have in this paper provided further support and arguments for the view that the syntax of relative-correlatives of early Dravidian, the ‘‘northern’’ languages, Koraga, and (optionally) Modern Malayalam is di¤erent from that proposed by SBS; at the same time, his account works admirably well for the modern ‘‘southern’’ languages that have post-RC clitics. This makes it possible to answer some of the questions raised in Section 1 above. As far as relative-correlatives are concerned, there clearly has been change and we therefore need to accept that chronologically and geographically di¤erent varieties of Dravidian do di¤er in their typology. The typology of relative-correlatives in early Dravidian, the ‘‘northern’’ languages, etc. turns out to be more similar to that of most other languages with relative-correlatives. As argued in Hock 2005b, the simple fact of nonembedding (in the sense that I use the term) may be su‰cient to ‘‘shield’’ the finite verbs (or predicates) of relative clauses in such structures from the Finiteness Constraint. At the same time, the addition of a post-RC clitic marker likewise has outside parallels, in Burushaski, Nagamese/Mongsen Ao, and Basque, and for these parallels it is likewise possible to suggest an explanation, namely that such markers provide a more ‘‘robust’’ shield against the Finiteness Constraint. At any rate, as far as relative-correlatives go, Dravidian is not an island of exceptionalism but conforms to (di¤erent) crosslinguistic tendencies. Regarding the question of whether the ‘‘Standard’’ model of Dravidian syntactic typology, based largely on Middle and Modern Tamil is appropriate for other varieties, the answer must be in the negative as far as relative-correlatives are concerned. For other phenomena, such as the question of the embedding of cited discourse, the jury is, I believe, still out.

Dravidian Syntactic Typology: A Reply to Steever

191

Finally, the case of the relative-correlatives confirms what has already been known in the areas of phonology and morphology (see the references in Section 4 above), namely that innovations can spread across the di¤erent branches of Dravidian. As in all other natural-language families, then, we must recognize the relevance of both the ‘‘tree’’ and the ‘‘wave’’ model.

Abbreviations & acc CC cond conv dat dur emph euph f fut gen habit hon impve incl

‘and’ (coordinator) accusative correlative clause conditional converb (or absolutive) dative durative emphatic ‘‘euphonic’’ su‰x feminine future genitive habitual honorific imperative inclusive (first plural)

loc m n narr neg nom nonmasc npst pl pres pst ptcle RC rel.pple sg

locative masculine neuter narrative negative nominative nonmasculine nonpast plural present past particle relative clause relative participle singular

Notes 1. See Section 2 on this issue. 2. Glossing, transcription, and syntactic bracketing are adjusted for consistency. 3. One of the most egregious misinterpretations or misrepresentations occurs in section 3.10.3: The absence of a practical command over any Dravidian language leads HHH into numerous mistakes in identifying specific forms and structures that appear in his critique and alternative. For instance, Hock (2001: 70) claims that his example (7), reproduced here as (17), contains a nonfinite relative-correlative clause when it contains no relative clause whatsoever, correlative or otherwise. Not knowing Tamil grammar, he misanalyses (17) as a coordinate structure with three conjuncts, mistranslating it as, ‘I did that, finished (it), and (then) went away’; the sequence ceytu mut
ittu.p ‘finished doing’ is an auxiliary

192 Hans Henrich Hock compound verb whose members function as a single verb (Steever 2005: 235– 37); (17) has just two clauses. (17) na¯n atai.c ceytu mut
ittu.p I-NOM it-ACC do-CF finish-CF ‘I finished that and went away.’

po¯n e¯n go-PST-1S

Contrast the original formulation of my example [7]. Clearly, I did not claim that it ‘‘contains a nonfinite relative-correlative clause’’. Nor did I ‘‘misanalyse’’ the example ‘‘as a coordinate structure with three conjuncts’’. Rather, as the ‘‘Lit(eral)’’ interpretation shows, I analyzed it ( just like SBS) as containing two non-finite verbs, which I labeled ‘‘abs(olutive)’’, plus one finite verb. The only thing that I might be accused of is not interpreting ceytu mut
ittu.p as a single, compound verb, but it is an open question whether that is the only possible interpretation. [7] na¯n atai-c ceytu mut
ittu-p po¯n e¯n I-sg.nom. that-sg.acc.n do-abs. finish-abs. go-sg.1.pst ‘After I had done that and finished I went away.’ Or: ‘I did that, finished, and (then) went away.’ Lit. ‘Having done that, having finished, I went away.’

(Tamil)

4. See also Chr. Lehmann 1980 and Hettrich 1988 (with references to earlier literature). 5. Interestingly, similar structures are still possible in Modern Hindi, as in the example below. This should put to rest SBS’s statement that ‘we may fairly wonder whether the SOVR [i.e., the conjoined analysis] accurately describes relative-correlatives in Indo-Aryan . . .’ (WSS, §4.3) hai]CC [yah vah ra¯nı¯i this that queen be.pres.sg.3 jo naukra¯nı¯j kar rahı¯ thı¯]RC [jisi -kı¯ seva¯ who.obl.-gen. service who.nom.servant do AUX was.sg.3fem. bha¯g gayı¯]CC [vahj hı¯ that EMPH flee go.pst.sg.3fem. ‘This is that queeni whosei service which servantj was doing, that onej indeed fled.’ Q ‘This is the queen whose servant, who was doing her service, fled.’ 6. Neither is SBS’s insinuation that my approach is somehow akin to 19 th -century ‘‘stadialism’’, with its notions of ‘‘primitiveness’’ (of Dravidian) vs. ‘‘advanced’’ character of Indo-Aryan (WSS, §5). In fact, like SBS and Lakshmi Bai and Ramasamy before him, I believe the evidence favors the view that relativecorrelatives are indigenous Dravidian constructions and not the result of IndoAryan influence. Where I disagree with SBS is in the analysis of early Dravidian and modern ‘‘northern’’ etc. relative-correlatives, and in the belief that,

Dravidian Syntactic Typology: A Reply to Steever

193

like all natural languages, the Dravidian languages are subject to linguistic change. 7. Since in most of the Dravidian languages conditionals are non-finite, relativecorrelatives with conditionals in the relative clause are excluded from the discussion of relative clauses with finite predicates, as they have been in all of my earlier publications. 8. An exception is found in the Onti-dialect text, where a relative clause is followed by clitic –(n)a; see below. This, however, is the same marker as the one employed by Tulu, and given the extensive influence of Tulu on Koraga, its appearance here can be plausibly attributed to Tulu influence. ¼na [od
e od
e pukkun]RC where-where go.pst.sg3m. Particle [a:yagi¯ oppunna ponnu sikkanaka]CC he.sg.dat agreeable woman find.pst.neg. Bhat: ‘Wherever he went he did not get a woman who will be agreeable to him.’ (Bhat 1971: 64, §1) 9. I am grateful to Nalini Rau Murthy and Dr. M. N. Sankarshana Murthy for helping with the Old Kannada examples. 10. I am indebted to K. V. Subbarao for alerting me to this kind of structure in Malayalam. 11. SBS considers the final en ri of the (putative) relative clause to function as an (FPE) element licensing the preceding finite predicate (WSS, note 17), but this ignores that en ri is itself a finite predicate. 12. Even if we were to accept SBS’s objection (WSS, §3.7) that basing my conclusion on the absence of post-RC clitics in Old Dravidian is an ‘argumentum ex silen[t]io’, we would still be left with the fact that Old Dravidian did permit relative-correlatives without post-RC clitics and that SBS’s typological claim fails to account for these constructions. 13. A further argument introduced by SBS in favor of his view that post-clausal clitics are an inherited feature of Dravidian syntax involves the existence of dubitative constructions such as the following in modern languages such as Tamil, Telugu, and Kon d a. pa¯t
ukira¯l
]S1 ¼o¯ en akku.t teriyavillai ]S0 she-nom sing-prs-3sf¼or I-dat know-ind-neg ‘I don’t know whether she is singing.’ (WSS, ex. (9a); SBS’s analysis and glossing)

S0 [S1 [aval


Th. Lehmann (p.c. 2008) plausibly suggests that these structures arise through deletion of the complementizer, the quotative marker (Mod. Tam. en ru), and notes that he has not found any such structures in Old Tamil. This too, then, may be an innovation.

194 Hans Henrich Hock 14. The same would hold true for the large number of other SOV languages with Finiteness Constraints but also with non-embedded, clitic-less relative clauses. 15. This limitation is intended to exclude from consideration the frequently available pragmatic interpretation of converbs as coordinated with their main verbs, as in the second gloss of the Tamil example below. na¯n
atai-c ceytu mutóittu-p pon
e¯n
I-sg.nom that-sg.acc.n do-abs. finish-abs. go-sg.1.pst ‘After I had done that and finished/after I had finished doing that I went away.’ Or: ‘I did that, finished/I finished doing that, and (then) went away.’ Lit. ‘Having done that, having finished, I went away.’ 16. I ignore other uses of –um, which are of no relevance in this context. 17. However, in Note 23, SBS gratuitously comments on the fact that ‘‘conjunction reduction’’ is not possible in structures with such mixed gender antecedents. 18. Quotativals of this sort are not uncommon crosslinguistically. Consider e.g. English expressions such as He says/says he, as well as I says/says I, which occur frequently in colloquial narratives to indicate that what precedes or follows is cited discourse. As shown by the paradigmatically frozen says, there has been a certain degree of grammaticalization; however, other verbs of speaking may be used, and the relative order of says and cited discourse is not fixed. 19. Here as elsewhere, SPEAK includes verbs such as ‘ask’ or ‘think’. 20. However, this would entail either that the Quotativals of Koraga must be considered an independent innovation or that, contrary to SBS’s view, the hypothesis postulating a northern origin of Koraga is to be accepted. 21. With the well-known exception of Toda and Modern Malayalam, which lose all gender distinctions in the verb.

References Andrews, Avery D. III. 1975 Studies in the Syntax of Relative and Comparative Clauses. Cam(1985) bridge: MIT dissertation. (Reprinted in ‘lightly retouched version’, 1985, New York & London: Garland.) Annamalai, E. & Sanford B. Steever 1998 Tamil. In: Steever (ed.), 100–128. Asher, Ronald E. & Kumari, T. C. 1997 Malayalam (Descriptive Grammars). London: Routledge.

Dravidian Syntactic Typology: A Reply to Steever

195

Bayer, Josef 2001 Two Grammars in One: Sentential complements and complementizers in Bengali and other South Asian languages. The Yearbook of South Asian Languages and Linguistics 2001: 11–36. Berger, Hermann 1998 Die Burushaski-Sprache von Hunza und Nager, 3 volumes. Wiesbaden: Harrassowitz. Beythan, Hermann 1943 Praktische Grammatik der Tamilsprache. Leipzig: Harrassowitz. Bhat, D. N. S[hankara] 1971 The Koraga Language. Linguistic Survey of India Series, 7. Poona (Pune): Deccan College. Bhat, D. N. Shankara 1967 Descriptive analysis of Tulu. (Building Centenary and Silver Jubilee Series, 15.) Pune: Deccan College. Burrow, Thomas and S. Bhattacarya 1970 The Pengo Language: Grammar, texts, and vocabulary. Oxford: Clarendon Press. Burrow, Thomas, and S. Bhattacharya 1953 The Parji Language: A Dravidian language of Bastar. Hertford: Stephen Austin & Sons. Caldwell, Robert 1913 A Comparative Grammar of the Dravidian or South-Indian Family of Languages. 3 rd edition, revised and edited by J. L. Wyatt and T. Ramakrishna Pillai. Reprinted 1974, New Delhi: Oriental Books Reprint Corporation. Coupe, A. R. 2007 Converging patterns of clause linkage in Nagaland. In New Challenges in Typology: Broadening the horizons and redefining the foundations, Matti Miestamo and Bernhard Wa¨chli (eds.), 129–136. Berlin/ New York: Mouton de Gruyter. Das, A. Sisir Kumar 1973 Structure of Malto. Annamalainagar: Annamalai University, Department of Linguistics. Downing, Bruce T. 1978 Some universals of relative clause structure. In Universals of Human Language, J. H. Greenberg (ed.), 4: 375–418. Stanford, CA: University Press. Emeneau, Murray B. 1961 Kolami: A Dravidian language. Annamalainagar: Annamalai University Publications in Linguistics.

196 Hans Henrich Hock Emeneau, Murray B. 1984 Toda Grammar and Texts. Philadelphia: American Philosophical Society. Friedrich, Paul 1975 Proto-Indo-European Syntax: The order of meaningful elements. (Journal of Indo-European Studies, Monograph 1.) Hahn, Ferd 1911 Kurukh Grammar. Calcutta: Bengal Secretariat Press. (Repr. Delhi: Mittal [no date].) Herring, Susan 2000 Poeticality and word order in Old Tamil. In Textual Parameters in Older Languages, ed. by S. Herring, P. van Reenen & L. Schoesler (eds.), 197–236. Amsterdam: Benjamins. Hettrich, Heinrich 1988 Untersuchungen zur Hypotaxe im Vedischen. Berlin: de Gruyter. Hock, Hans Henrich 1975 Substratum Influence on (Rig-Vedic) Sanskrit? Studies in the Linguistic Sciences 5 (2): 76–125. 1982 The Sanskrit Quotative: A historical and comparative study. Studies in the Linguistic Sciences 12 (2): 39–85. 1984 (Pre-)Rig-Vedic Convergence of Indo-Aryan with Dravidian? Another look at the evidence. Studies in the Linguistic Sciences 14(1): 89–107. 1986 Principles of historical linguistics. Berlin: Mouton de Gruyter. 1988a Review article: Finiteness in Dravidian. (Steever (1988): The serial verb formation in the Dravidian languages). Studies in the Linguistic Sciences 18(2): 211–231. 1988b Rebracketing and Relative Clauses in Old English. In Germanic Linguistics II: Papers from the Second Symposium on Germanic Linguistics, University of Illinois at Urbana-Champaign, 3–4 October 1986, Elmer H. Antonsen and Hans Henrich Hock (eds.), 35–54. Bloomington: Indiana University Linguistics Club. 1989a Conjoined We Stand: Theoretical implications of Sanskrit relative clauses. Studies in the Linguistic Sciences 19 (1): 93–126. 1989b Review of Steever (1988): The Serial Verb Formation in the Dravidian Languages. Language 65: 398–405. 1991 On the Origin and Development of Relative Causes in Early Germanic, with Special Emphasis on Beowulf. In Stæfcræft: Studies in Germanic linguistics, Elmer H. Antonsen and Hans Henrich Hock (eds.), 55–89. Amsterdam: Benjamins, 1991. 1992 Reconstruction and Syntactic Typology: A plea for a di¤erent approach. In Explanation in Historical Linguistics, G. W. Davis & G. K. Iverson (eds.), 105–121. Amsterdam/Philadelphia: Benjamins.

Dravidian Syntactic Typology: A Reply to Steever 2001

197

Typology vs. Convergence: The issue of Dravidian/Indo-Aryan syntactic similarities revisited. The Yearbook of South Asian Languages and Linguistics 2001: 63–99. 2005a The Problem of Time in South Asian Convergence. In Proceedings of the Murray B. Emeneau Seminar, Central Institute of Indian Languages, Mysore, 2005b Electronic publication at http://www.ciil.org/ Main/Announcement/MBE_Programme/frameset1.htm. 2005b How Strict is Strict OV? A Family of Typological Constraints with Focus on South Asia. Yearbook of South Asian Languages and Linguistics 2005: 145–164. Hualde, Jose´ Ignacio, and Jon Ortiz de Urbina (eds.) 2003 A Grammar of Basque. Berlin/New York: Mouton de Gruyter. Israel, M. 1979 A Grammar of the Kuvi Language (with Texts and Vocabulary). Trivandrum (Tiruvanantapuram): Dravidian Linguistics Association. Kittel, F. 1903 A Grammar of the Kannad
a Language in English: Comprising the three dialects of the language (ancient, mediæval and modern). Mangalore: Basel Mission Book and Tract Depository. Krishnamurti, Bhadriraju 1969 Kon
d
a or Ku bi: A Dravidian language. Hyderabad: Tribal Cultural Research & Training Institute. Krishnamurti, Bhadriraju 2003 The Dravidian Languages. Cambridge: University Press. Lakshmi Bai, B. 1985 Some notes on correlative constructions in Dravidian. In For Gordon H. Fairbanks, V. Z. Acson & R. L. Leed (eds.), 181–190. Honolulu: University of Hawaii Press. Lehmann, Christian 1980 Der indogermanische *kwi/kwo-Relativsatz im typologischen Vergleich. In Linguistic Reconstruction and Indo-European Syntax: Proceedings of the Colloquium of the ‘Indogermanische Gesellschaft’, University of Pavia, 6–7 September 1979, P. Ramat et al. (eds.), 155–69. Amsterdam: Benjamins. Lehmann, Christian 1984 Der Relativsatz: Typologie seiner Strukturen, Theorie seiner Funktionen, Kompendium seiner Grammatik. Tu¨bingen: Narr. Lehmann, Thomas 1994 Grammatik des Alttamil unter besonderer Beru¨cksichtigung der Can kam-Texte des Dichters Kapilar. Stuttgart: Frans Steiner Verlag. Lehmann, Thomas 1998 Old Tamil. In Steever (ed.), 75–99.

198 Hans Henrich Hock Lehmann, Winfred P. 1974 Proto-Indo-European Syntax. Austin: University of Texas Press. Mahapatra, B. P. 1979 Malto: An ethnosemantic study. Mysore: Central Institute of Indian Languages. Pillai, P. V. 1973 Early Malayalam Prose: A study. Trivandrum: University of Kerala. Ramasamy, K. 1981 Correlative relative clauses in Tamil. In Dravidian Syntax, S. Agesthialingom & N. Rajasekharan Nair (eds.), 363–380. Annamalainagar: Annamalai University Publications in Linguistics, 73. Steever, Sanford B. 1987 Tamil and the Dravidian Languages. In The World’s Major Languages, B. Comrie (ed.), 725–46. New York: Oxford University Press. Steever, Sanford B. 1988 The Serial Verb Formation in the Dravidian Languages. Delhi: Motilal Banarsidass. Steever, Sanford B. 1993 Analysis to Synthesis: The development of complex verb morphology in the Dravidian languages. New York: Oxford University Press. Steever, Sanford B. (ed.) 1998 The Dravidian Languages. London/New York: Routledge. Steever, Sanford B. 1998a Introduction. In Steever (ed.), 1–39. Steever, Sanford B. 1998b Gond i. In: Steever (ed.), 270–297. Watkins, Calvert 1976 Towards Proto-Indo-European Syntax: Problems and pseudoproblems. In Papers from the Parasession on Diachronic Syntax, 305–26. Chicago: Linguistic Society.

Regional Reports

North America

Tej K. Bhatia

Introduction The forces of globalization are shaping language practices around the globe in a manner which is unprecedented in the history of human communication. This brief survey, in part, attempts to identify the ways in which the research on South Asian languages and linguistics in North America has responded to this new context and challenges posed by it. This brief survey aims to shed light on the emerging new patterns of research on the theory and practice of South Asian languages and linguistics in the region. This is accomplished by focusing on published and unpublished material produced since 2004. A cursory examination of the research can best be characterized by the term, ‘interface’. The interface of theory and practice manifests itself in a wide variety of ways which transcend the traditional boundaries and branches of linguistics. For this reason, it is imperative to point out at the outset that the traditional categorization of research as presented in this report is neither absolute nor water tight. For example, the traditional boundaries between syntax, psycholinguistics and sociolinguistics are gradually becoming blurred in linguistics. This is also the trend witnessed in research on South Asia. The first part of the report is classified into the following four broad sections: (1) Theoretical Linguistics; (2) Psycho-and Neuro-linguistics; (3) Sociolinguistics and Applied Linguistics; (4) South Asian Language Pedagogy; the second part (Annotated references) summarizes some recently completed relevant dissertations. Although an attempt has been made to present representative trends, not all the works and the trends since 2004 are included here. A critical appraisal of the trends also falls outside the scope of this brief report.

202 Tej K. Bhatia I. Book, articles and research tools 1. Theoretical linguistics In the area of phonetics and phonology, the schwa-syncope rule, nasalization and syllable structure continue to draw attention from a number of different theoretical and disciplinary perspectives. The schwa-deletion rule has intrigued phoneticians, phonologists and sociolinguists alike for approximately the past two-decades. Building on the body of work since 1970s, Narasimhan, Sproat and Kiraz (2004) focuses on the variable deletion of schwa within single and multimorphemic words in Hindi. Combining the findings and the corpus of prior studies with a new high-frequency compound word data-base, the authors propose an integrated model for textto-speech synthesis involving the schwa-syncope rule. Shosted (2006) deals with the issue of the emergence of Nasal coda. Nasal Coda emergence is the process that leads to the development of a nasal coda which may, or may not have, been present in an earlier form of a spoken language. Based on an aerodynamic comparative study of Carioca Brazilian Portuguese, Hindi and French, the study concludes that the lowered velum and/or raised tongue body lead to the development of coda obstruent on nasal vowels. An exhaustive treatment of the syllable structure of Maithili is presented within the framework of the Optimality Theory in Mishra (2007), while the syllabic structure and the phonotactic patterns of Hindi are examined by Kumar (2006), employing three distinct approaches. Turning to morphology, Singh (2004) presents a treatment of reduplication in modern Hindi which extends the scope of his earlier work on Whole Word Morphology, which includes a monograph (co-authored with Agnihotri), discussed in my earlier report, on Hindi morphology. As regards syntax, semantics and universal grammar, topics such as word order variability, the tense-aspect agreement, direct and indirect subjects, light verbs, finiteness, and learnability continue to draw attention from a variety of theoretical and typological perspectives. Bhatt (2005) argues that Chomskyan notion of Agree (1998, 1999, and 2001) needs to be revisited in order to account for Long Distance Agreement and Object agreement in Hindi-Urdu. Long Distance Agreement involves a verb agreeing with a constituent inside the verb’s clausal complement. The paper presents a new analysis of the phenomenon which favors dissociation between case and agreement because Head agrees with an

North America

203

argument, but does not take its case value. Bhatt and Dayal (2007) presents a non-antisymmetric account of rightward scrambling in Hindi-Urdu which argues against the antisymmetric treatment of the phenomenon presented by Mahajan (1997) and Simpson and Bhattacharya (2003). Their new account is proposed in terms of rightward remnant-VP movement. Manetta (2006) carries out a comparative investigation of Kashmiri and HindiUrdu involving wh-movement and wh-explitive constructions. Systematic di¤erences in the two languages are captured in terms of the properties of the phase-defining heads C and v. The analysis shows that the specifier position of vP plays the same role often attributed to the specifier position of CP in wh-movement. Deo and Sharma (2006) present a typology of ergative marking and agreement in Indo-Aryan languages in order to demonstrate a progressive loss of ergativity marking with varying degree in di¤erent systems. The two mechanisms leading to the loss of marking are attributed to the loss of an overt subject marking in nouns and the loss of marked agreement in verbs. Based on the analysis of dialect variation in Marathi, the paper shows that the dialect typology of Marathi mirrors the cross linguistic typology, but with a limit imposed by the parent language, old Marathi. The universal subhierarchies of markedness are motivated by the Optimality Theory framework. Concerning typological issues is also the nagging problem of ‘‘Strict OV’’ typology. Hock (2005) proposes a new framework for the discussion of ‘‘Strict OV’’ word order typology in order to present a unified account of contact and convergence in South Asian languages. Vasishth and Lewis (2006) exemplifies the interface of syntactic and psycholinguistic research in South Asian linguistics. The study begins with the robust locality principle that is the proximity between arguments and verbs as an indicator of sentence-processing di‰culty and then moves on to examine the validity of non-locality approaches in processing. For this purpose, two self-paced reading experiments involving Hindi were set up. Their results demonstrate that verb (head) and argument distance is an important, but not su‰cient, condition for human sentence parsing. Examining the claims of the linguistically-driven theories concerning locality and antilocality e¤ects on sentence processing and memory, the paper argues that an integrated account of locality, anticipation and interference best characterizes the general constraints on memory as evidenced by Anderson et al. (2004). Light verbs/Complex predicates in Hindi-Urdu are accounted for in Davison (2005). Her study argues that in complex predicates the nominal

204 Tej K. Bhatia argument functions as the syntactic object of the V in which the argument structure of N and V are merged. Gair (2005) unravels some uncommon distribution features of embedded verb forms and their subjects in Sihala which, in turn, challenge the theory of finiteness and control. Based on the evidence provided from the di¤erential processing of semantic vs. syntactic cues during sentence processing in Hindi and Japanese, Phillips (2006) challenges the existing statistical models of learnability. The paper argues that statistical models of learning fall short on the following three grounds: complexity, consistency and causality. Let me conclude this section with Kachru (2006) which is an authoritative grammar of Hindi. The work demonstrates how an expert linguist can distill and elegantly integrate the insights of modern linguistic theory and traditional grammar. In addition to providing a sociolinguistic account of Hindi, a detailed account of the phonetics, phonology, morphology, syntax, information and discourse structure is presented. This work provides an excellent reference tool for linguists and non-linguists alike.

2. Psycho-and neuro-linguistic studies 2.1. First and second language acquisition The role of negative evidence, first language influence, and parents’ grammar are notable topics of research on language acquisition. Lakshamanan (2006) presents the development of verbal inflection, mainly of tense and agreement, by two children acquiring Tamil prior to the age of 24 months. The data obtained is intriguing since it may lead to the conclusion that children’s knowledge of verbal inflection di¤ers substantively from that of an adult. However, alternative analyses point to the reverse conclusion, even though at the surface, a child’s production does not mimic the features of adult grammar. The research presents critical evidence for grappling with the issue of children’s early knowledge of the grammatical properties as it relates to adult’s knowledge. The role of interference in language acquisition is the subject of the following three studies. Wiltshire and Harnsberger (2006) illustrates how the mother tongue (L1) influences the sound pattern of Indian English (IE) for the two groups of speakers – Gujarati MT and Tamil MT. The acoustic analysis of IE spoken by Gujarati and Tamil MT speakers proficient in

North America

205

English demonstrates that the two groups varied due to transfer from their native language. The transfer e¤ects became apparent in Gujarati English back vowels, Tamil English rhotics, and the proportion of rising vs. falling pitch accents in Gujarati English intonation. A study by Nestor and Singh (2005) deals with the di¤erence in language processing of phonetic variation by English-speaking monolingual children (age 2:9–3:10) and HindiEnglish bilingual children (age 2:9–3:10). The children were introduced to words with phonetic variation, which is phonemic in Hindi while nonphonemic in English. In this experimental study, children were given familiar and unfamiliar words with vs. without variation involving aspiration in voiced and unvoiced stops. The study reveals that when asked to assign non-words to either of the two languages, bilingual children correctly perceived phonetic variation relevant for Hindi and insignificant for English. Similarly, a comparative study by Pruitt, Jenkins and Strange (2006) illustrates that allophonic variation can be detrimental to the acquisition of a new phonemic contrast. For instance, the acquisition of Hindi dental vs. retroflex phonemic contrast by speakers of American English is di‰cult despite the existence of dental and retroflex allophones in (American) English. In contrast, it is relatively easier for Japanese learners to acquire Hindi contrast. Like Hindi, Japanese exhibits a phonemic contrast between dental /d/ and the flapped /r/ (sometimes pronounced as a retroflex). The pattern of the acquisition of the dental vs. retroflex Hindi contrast by Japanese and American learners of Hindi shows that Japanese learners were uniformly more successful than American learners in the identification of the contrast during the three phases of testing – pre-testing, training testing and posttesting. The training session was carried out by means of a computerinteractive program.

2.2. Neurolinguistic studies Two works by Adamson (2005) and Adamson and Hillige (2005) aim to explore the hemispheric asymmetry in English-Urdu bilinguals for processing printed material. Six experiments for the two visual fields – right visual field/left hemisphere [RVF/LH] and left visual field/right hemisphere [LVF/RH] or both visual fields simultaneously (BVF) – were designed to determine the role of orthographical properties in di¤erential hemispheric processing during reading. Experiments 1–4 showed a RVF/LH advantage

206 Tej K. Bhatia for processing Urdu words and pronounceable non-words regardless of whether the letters were joined. The qualitative analysis of errors showed that non-joined letters were processed more sequentially than joined letters. Experiments 5–6 showed an overall RVF/LH advantage, but not an absolute advantage. The findings reveal that qualitative di¤erences in hemispheric processing as the function of language-specific scanning direction, orthographic-phonological mapping and morphology in addition to an individual language proficiency. These new findings are qualitatively similar to the findings of processing studies devoted to Hebrew and Arabic processing. However, the findings are distinct from English processing.

3. Sociolinguistics and applied linguistics A large body of research devoted to sociolinguistics and applied linguistics is associated with the various facets of bilingualism/multilingualism and South Asian Englishes. The topics, which have received significant attention, include media discourse, popular cultures, social factors in language maintenance and shift, and language of religion. Two handbooks, The Handbook of World Englishes (Kachru, Kachru and Nelson 2006) and The Handbook of Bilingualism (Bhatia and Ritchie 2004; revised edition 2006) provide a window on the expanding scope of South Asian sociolinguistics research with global perspectives. The Handbook of World Englishes presents a state-of-the art research on World Englishes with the following two main objectives: one, ‘to revisit the proliferation of terminologies and concepts articulating the global uses of Englishes’; and two, to present the global profile of the English language. South Asia Englishes weaves through practically every chapter in the handbook. This work contains several contributions from North American scholars. The Handbook of Bilingualism presents a comprehensive account of the diversity and complexity of the field of bilingualism and multilingualism. Four chapters (3, 13, 20, 29) are particularly noteworthy for their focus on South Asia, encompassing topics such as the following: bilingual aphasia (Tamil), socio-psychological motivations for code-mixing, media, and a case study of South Asian bilingualism. With the largest number of users in Asia, it is timely and urgent to explore the phenomenon of Asian Englishes. Kachru (2005) and Kachru and Nelson (2006) fill this gap by providing a comprehensive treatment of

North America

207

English with intersecting yet di¤erent perspectives. While research on Asian Englishes is the main focus of the former work, the latter is grounded in the global context of English with a pedagogical orientation. Issues such as spread, acculturation, functions and evolution of English in Asia are addressed in both works. Concentrating on Indian English, Singh (2007) and Bhatt (2007) also address some of the same issues, though from a somewhat di¤erent point of view. Bhatt provides an optimality theoretic account of the syntax of Indian English and Singh discusses the theoretical problems thrown up by the existence of varieties such as Indian English. A new body of research on code mixing and code switching is not only a welcome addition to the empirical scope of the research on the topic, but also underscores key issues. Examples of such issues are socio-psychological determinants of language choice, the distinction between matrix language and peripheral language, universal constraints on the grammar of mixed utterances and the acquisition of mixed systems, among others. Needless to say, the phenomenon of code mixing and code switching is also critical to understand the psycholinguistic complexity of speech production and speech processing among bilinguals. Abeywickrama (2007) aims at filling a gap in research on code-mixing and code-switching by focusing on the interactional pattern of code-mixing instead of the grammar of code-mixing. Following a discourse analytical approach, the paper reveals that in conversations, code-switching between Sinhala and English exhibits a ‘new cultural pattern’ of discourse in Sri Lanka. Kachru (2006a) uncovers yet another aspect of Hinglish in Indian popular culture. Her work reveals that the salient function of mixing English in popular Hindi songs is to create humor and parody. How does one represent American super-heroes through Indian languages in recent South Asian comic books? Bhatia (2006) attempts to answer this question by analyzing creative linguistic strategies, with a particular focus on the representation of super-heroes. The strategy involves the mixing of two languages – English and Sanskrit – which are perceived as super languages in the context of South Asia. The scalar semantics and psycholinguistic processing represent the other two notable dimensions of language mixing in comic books. The exploration of the sociolinguistics of language and religion marks a new mile stone in the work of Omoniyi and Fishman (2006). The work presents a theoretical and methodological conceptualization of the sociology of language and religion. Pandharipande (2006) has taken a major

208 Tej K. Bhatia step in this direction to uncover the complex dimensions of language choice and language spread in the multi-cultural and multi-religious setting of India. The issue of language choice in advertising discourse aimed at rural India is explored in Bhatia (2007). This work illustrates how the forces of top-down and bottom-up globalization are radically changing the mode of rural communication and thus redefining the qualitative nature of Indian bilingualism. Hook and Zia (2005) present an account of Impersonal Causative Expressions (ICE) in the Gilgit dialect of Shina, spoken in the Northern Areas of Pakistan, and which has recently been adopted as a standard for radio broadcasting. One of the most interesting findings is the use of feminine in Shina as a default gender in ICE. In addition to transfer and other factors outlined in section 2.0 above, socio-psychological factors such as learners’ beliefs also play an important role in language acquisition and language processing. Sider (2004) examines the role of self-perception in language attrition and retrieval among bilingual children. This study aims at gaining insights into the self-perception of expatriate children whose first language is English and who acquired Hindi as a second language while their parents worked in India. Data on children’s reporting of di‰culties with the retrieval of Hindi were obtained by employing semi-structured interviews. Children’s narratives reveal the following inventory of self-perception of language retrieval and language attrition: Indian ‘environment’ (e.g. socialization in Indian environment); personal motivation; time duration of non-usage of the L2; their own attitudes toward the L2; perceived social desirability of the L2; and the lack of development of Hindi literacy. The role of diasporic media, particularly Bollywood (Hindi) films, in maintenance of Hindi among North American Asian Indian teenagers is identified in Shankar (2004). While Hindi films provide a much-needed range of language input (e.g. from scripted dialogues to social registers and varieties), the study reveals that the teenagers employ the topic Hindi films in domains ranging from humor, flirting to various interactive social interactions including conflict resolution.

4. Corpus linguistics, digital reference tools and language pedagogy As mentioned in my last report, the September 11, 2001 terrorist attack marked a new turning point in the teaching, learning and research of South

North America

209

Asian languages in North America. The establishment of the Hindi-Urdu flagship center at the University of Texas at Austin in 2007 opens a new chapter in the language pedagogy and research on Hindi-Urdu. Other significant developments in this area are as follows: Digital dictionaries of South Asia (DDSA) are a major undertaking of the University of Chicago in collaboration with Columbia and the Triangle South Asia consortium in North Carolina. The aim of this ambitious project is to ‘create and disseminate electronic dictionaries for the languages of South Asia.’ The current project includes dictionaries of thirty-three major and minor languages of South Asia. Eighteen more languages will be added in the near future, thus bringing the total to fifty one languages. When completed, the corpus of form and usage of South Asian languages will serve as important tool for linguistic, sociolinguistic and second language learning research. For more detail, see: http://dsal.uchicago.edu/dictionaries Another notable attempt in corpus linguistics is by Lise (2005). It presents a corpus of 1844 Indic words which constitute an integral part of English/Creole of Trinidad and Tabago. Nearly all words are attributed to Bhojpuri, which served as a source of leveling of languages brought by indentured Indians who arrived in Trinidad soon after the abolishment of slavery. The corpus classified into eighteen socio-cultural domains. In its April 2006 meeting of South Asian Language Teachers Association, the need for classification and organization of the existing digital material for teaching and learning Hindi was underscored in addition to the identification and development of methodological frameworks for teaching SA languages. A notable attempt in the area of language pedagogy is the development of Hindi standards for foreign and heritage language learning for K-16 levels by a national team of twelve educators of Hindi from various U.S. schools and colleges. The Hindi project, funded by the National Collaborative Board and the South Asia Language Resource Center, is also creating an additional set of Hindi Standards for beginner, intermediate and advanced level college learners. The Hindi Standards are based on the generic standards and are organized around five goal areas: Communication, Cultures, Connections, Comparisons and Communities. In short, this project proposes guidelines to a variety of educators at the level of individual classroom, school district or university in developing standards-based Hindi curricula, teaching materials and tests. A draft version of the Hindi standards

210 Tej K. Bhatia will be available by the end of March 2008. For more information, please contact Dr. Vijay Gambhir of the University of Pennsylvania.

5. Conclusion The research on South Asian languages and linguistics continues to gain vitality and diversity in terms of its scope, theoretical and applied orientation. It receives its impetus and dynamics from interdisciplinary and empirical research. It continues to respond to the challenging of globalization and international education on one hand and theoretical and empirical research on the other.

II. Annotated references: Selected doctoral dissertations Aier, Deepa. 2005. Cross-Linguistic Comparison of Negative Evidence in English, Hindi, and Tamil. Tempe: Arizona State University. The role of input in the form of Negative Evidence (NE) is critical for children to learn from their errors while acquiring their first language. NE which is provided by an adult (usually parents), can take various forms, ranging from hints, clarifications, recasting, elaboration to explicit correction. This dissertation focuses on both qualitative and quantitative aspects of NE in three languages: English, Hindi and Tamil. This experimental cross-linguistic study shows that NE provided by mothers to children varied on both qualitative and quantitative grounds. While English-speaking mothers provided cues for children errors approximately 25% of the time, Hindi- and Tamil- speaking mothers provided such an input to children approximately 48% of the time. In addition, the study finds qualitative (stemming from structural complexities) and cultural di¤erences in NE. Recasting, hints and corrections were the main types of NE employed across the three language groups. Ayres, Alyssa. 2004. Speaking like a state: Nationalism, language, and the case of Pakistan. Chicago: University of Chicago. This dissertation examines the sociology of language in Pakistan, particularly from the historical perspective of language and nationalism. Three

North America

211

narratives are presented to grapple with the complex, conflicted and puzzling history of Pakistan. The first narrative deals with Urdu, while the second with Punjabi. The Urdu language movement has its roots in the complex history of nationalism. The Punjabi language movement is essentially grounded in aesthetics (Punjabiyat) and is, thus, more abstract in nature. The third narrative deals with the comparative role that language plays in nationalism. The dissertation claims that the language is a by-product of national consciousness/identity rather than vice versa. This poses a challenge for the powerful theories of nationalism, which assume the centrality of language, literacy and communicative practices for the core of national consciousness. Choi, Hansook. 2006. Prosodically induced acoustic variation in English, Korean and Hindi. Urbana-Champaign: University of Illinois. English, Korean and Hindi speech production experiments are carried out in order to isolate the determinants of acoustic variation, as witnessed in the following two cases: (i) laryngeal phonological contrast among stop consonants, and supralaryngeal among vowels; and (ii) the role of systemic and prosodic factors. The study shows that accentual e¤ect was uniformly shared in acoustic variation across the three languages, while the boundary e¤ect turned out to be the main reason for language specific acoustic variation – Hindi exhibiting the initial strengthening of vowels without enhancement of the preceding consonants, and Korean with a more peripheral distribution in final position. Hong, Sungok. 2006. On subject and indirect subject construction in Hindi. Madison: University of Wisconsin. This dissertation takes a fresh new look at the notion of subjecthood by examining various indirect subject constructions, such as the dative and the instrumental subject construction. The dissertation reveals that the notion of subject still poses a challenge for linguistics on a cross-linguistic and intra-language basis. The findings include the following: (i) a multifactor approach is needed to come to grips with subjecthood in Hindi and in that sense lending support to Keenan (1976) hierarchy; (ii) from the viewpoint of Li and Thompson (1976) typological distinctions, Hindi meets most features of a topic-prominent languages as do languages such as Korean; (iii)

212 Tej K. Bhatia Dative subjects fall into Dowty’s Proto-patient rather than ‘Proto-agent’ category. Kumar, Aman. 2006. Aspects of Hindi syllabic structure. Ann Arbor: University of Michigan. This dissertation provides a systematic and exhaustive treatment of consonant clusters in word initial, word medial and word final position. In order to examine the validity of Hindi phonotactic patterns, the following three approaches are examined: the sonority-based approach, the word-edge based approach, and the CVX (where X can either be a consonant or a vowel) approach. The approaches are based on mechanisms such as extrametricality, resyllabification, sonority, articulatory feature theory and notion of complex sounds. The analysis reveals that the CVX theory can o¤ers a simple, yet powerful, treatment of the phonotactic patterns of Hindi, as it posits the maximal syllable length as CVX. The other two approaches which posit larger syllable structures and predict more complex medial syllabic structures are in a sense uneconomical and unsuitable for Hindi. Lam, Margaret Yui-Ki Kong. 2005. The acquisition of grammatical skills and its relation to reading comprehension in ESL students. Toronto: University of Tornoto The two models of second language acquisition – The Parallel model and the Interactive model – are examined for their suitability and significance for explaining the acquisition of grammatical skills by learners of English as a second language (ESL). The role of grammatical skills, vocabulary knowledge and phonological processing skills in reading were examined in the following two groups of children in grade (1–3): English mother tongue children (EL1) group and ESL group representing Cantonese and Punjabi MT children. The results lend strong support, but not categorical support, to the Parallel model, as individual and group variation was also observed. While vocabulary knowledge was a significant common determinant for grammatical skills in both EL1 and ESL children, the ESL group varied in terms of phonological awareness which turned out to be more significant for Cantonese-speaking ESL learners than Punjabi-speaking ESL learners. The research underscores the importance of phonological and vocabulary development as a critical measure for intervention and identification of ESL students at risk.

North America

213

Mishra, Mithilesh. 2007. The syllabic structure and stress pattern of the Maithili language (India, Nepal). Ph.D dissertation. Urbana-Champaign: University of Illinois. This dissertation provides an account of the syllable structure and stress pattern of Maithili within the framework of the Optimality Theory. It shows that while the consonantal inventory of Maithili is similar to that of other Indo-Aryan languages found in India and Nepal, the vocalic system is di¤erent as vowel length contrast has been replaced by the tense and lax contrast. The mechanism of the Syllabic structure constraints and the Metrical constraints in Maithili receive a particular focus in this work. The findings lend support to claim of the Optimality Theory that just as di¤erent ranking of universal constraints render typological di¤erence among languages, the di¤erent ranking of language-specific constraints yield variation in a language.

References Abeywickrama, Priyanvada 2007 Do we codeswitch or codemix in Sri Lanka? Journal of Multicultural Discourse 2 (1): 63–77. Adamson, Maheen Mausoof 2005 Hemispheric di¤erences in Urdu. Ph.D dissertation. LA: University of Southern California. Adamson, Maheen Mausoof and Joseph B. Hillige 2005 Hemispheric di¤erences for identification of words and nonwords in Urdu-English bilinguals. Neuropsychology 20 (2): 232– 248. Anderson, John R., Michael Byrne, Scott Douglass, Christian Lebiere and Yulin Qin 2004 An integrated theory of mind. Psychological Review 111 (4): 1036– 60. Bhatia, Tej K. 2007 Advertising and marketing in rural India. New Delhi: Macmillan India. 2006 Super-heroes to super languages: American popular culture through South Asian language comics. World Englishes 25 (2): 279–298. Bhatia, Tej K. and William C. Ritchie (eds.) 2004; Handbook of Bilingualism. Oxford: Blackwell Publishing 2006

214 Tej K. Bhatia Bhatt, Rajesh 2005 Long distance agreement in Hindi-Urdu. Natural Language and Linguistic Theory 23 (4): 757–807. Bhatt, Rajesh and Veneeta Dayal 2007 Rightward scrambling as rightward remnant movement. Linguistic Inquiry 38 (2): 287–301. Bhatt, Rakesh 2007 On the Native/Non-native Distinction. Annual Review of South Asian Languages and Linguistics: 2007. 53–70. Deo, Ashwani and Devyani, Sharma 2006 Typological variation in the ergative morphology of Indo-Aryan languages. Linguistic Typology 10 (3): 369–418. Davison, Alice 2005 Phrasal predicates: How N combines with V in Hindi/Urdu. The Yearbook of South Asian Languages and Linguistics 2005: 83–115. Gair, James W. 2005 Some aspects of finiteness and control in Sinhala. The Yearbook of South Asian Languages and Linguistics 2005: 117–143. Hock, Hans 2005 How strict is Strict OV? A family of typological constraints with focus on South Asia. The Yearbook of South Asian Languages and Linguistics 2005: 145–164. Hook, Peter and Mohammad Amin Zia 2005 Searching for goodness: A study of sensory and other impersonal causative expressions in the Shina of Gilgit. The Yearbook of South Asian Languages and Linguistics 2005: 165–188. Kachru, Braj 2005 Asian Englishes: Beyond the cannon. Hong Kong: Hong Kong University Press. Kachru, Braj, Yamuna Kachru and Cecil Nelson 2006 Handbook of World Englishes. Oxford: Blackwell Publishing. Kachru, Yamuna 2006 Hindi. Amsterdam: John Benjamin Publishing Company. 2006a Mixers lyricing in Hinglish: blending and fusion in Indian pop culture. World Englishes 25 (2): 223–233. Kachru, Yamuna and Cecil Nelson 2006 World Englishes in Asian context. Hong Kong: Hong Kong University Press. Mahajan, Anoop Kumar 1997 Rightward movement. In Rightward movement, Dorothee Beerman, David LeBlanc, and Henk van Riemsdiik (eds.), 185–213. Amsterdam: John Benjamins.

North America

215

Lakshamanan, Usha 2006 Assessing linguistic competence: Verbal inflection in child Tamil. Language Assessment Quarterly 3 (2): 171–205. Manetta, Emily Walker 2006 Peripheries in Kashmiri and Hindi-Urdu. Ph.D. diss., Department of Linguistics. Santa Cruz: University of California. Narasimhan, Bhuvana, Richard Sproat and George Kiraz 2004 Schwa-deletion in Hindi text-to-speech synthesis. International Journal of Speech Technology 7 (4): 391–333. Nestor, Sarah and Leher Singh 2005 E¤ects of phonetic variation on early word learning. Proceedings of the Annual Boston University Conference on Language Development 29 (2): 401–412. Omoniyi, Tope and Joshua A. Fishman (eds.) 2006 Explorations in the sociology of language and religion. Amsterdam: John Benjamins Publishing Company. Pandharipande, Rajeshwari 2006 Ideology, authority, and language choice: Language of religion in South Asia. In Omoniyi and Fishman (ed.), 141–164. Phillips, Colin 2006 Three benchmarks for distributional approaches to natural language syntax. In Znautiini, Rafaella, Hetor Campos, Elena Herburger and Paul Portner (eds.), Crosslinguistic research in syntax and semantics: Negation, tense and clausal architecture, 13–23. Washington, D. C: Georgetown University Press. Pruitt, John, James Jenkins and Winfred Strange 2006 Training the perception of Hindi dental and retroflex stops by native speakers of American English and Japanese. The Journal of the Acoustical Society of America 119 (3): 1684–1696. Shankar, Shalina 2004 Reel to real: Desi teen’s linguistic engagements with Bolloywood. Pragmatics 14 (2–3): 317–335. Shosted, Ryan K. 2006 Vocalic context as a condition for nasal coda emergence: Aerodynamic evidence. Journal of the International Phonetic Association 36 (1): 39–58. Sider, Steve R. 2004 Growing up overseas: Perception of second language attrition and retrieval amongst expatriate children in India. Canadian Journal of Aplied Linguistics/Revue canadienne de linguistique applique´ 7 (2): 117–137.

216 Tej K. Bhatia Simpson, Andrew and Tanmoy Bhattacharya 2003 Obligatory overt wh-movement in a wh-in-stu language. Linguistic Inquiry 34: 127–142. Singh, Rajendra 2007 Nature, Structure, and Status of Indian English, Annual Review of South Asian Languages and Linguistics: 2007. 31–44. 2004 Reduplication in Modern Hindi and the theory of reduplication. The Yearbook of South Asian Languages and Linguistics 2003: 155–170. Vasishth, Sharavan and Richard L. Lewis 2006 Argument-head distance and processing complexity: Explaining both locality and antilocality e¤ects. Language 82.4: 767–794. Wiltshire, Caroline and James Harnsberger 2006 The influence of Gujarati and Tamil L1 on Indian English: a preliminary study. World Englishes 25 (1): 91–104.

Linguistic Publications in Bengali (2000–2008): A Brief Review

Niladri Sekhar Dash

1. Introduction In the following pages I present a brief sketch of Bengali publications in linguistics during the last few years (2000–2008) as a continuation of the surveys made by earlier scholars (cf. Nath 2001). Since it was not possible to refer to all works published during this period, I report on publications that are relevant for and seem central to this survey. Due to their indispensable reference value, I have, however, included a few publications in English, and made occasional references to some earlier publications. The areas I cover are general linguistics, spelling, morphology, syntax and lexicology, semantics, language teaching, lexicography, dialect study, sociolinguistics, stylistics, and corpus linguistics. 2. General linguistics Descriptive study of the Bengali language and linguistics seem to be quite popular, perhaps because these areas are seen as not requiring any kind of specialization. It is indeed a tough task to search through many amateurish exercises to identify work that has some pertinence. To start with general Bengali linguistics, quick reference may be made to those publications that focus on preservation and planning of Bengali (Singh 2004), general features and properties of the language (Bhattacharya 2000), evolution of language and its speakers (Sur 2001), features and techniques of linguistics for studying the language (Huq 2001), general issues of the language and its linguistics (Huq 2002), scope of linguistics (Musa 2002), the general theories and principles of modern linguistics applicable to Bengali (Bandyopadhyay 2005), texture and form of modern Bengali (Das 2006), and factors important for acquiring good foundation in Bengali (Das 2000, Singh 2000).

218 Niladri Sekhar Dash We also come across some anthologies where various issues of Bengali language and linguistics are addressed. While some of them refer to the love for and discomfort with Bengali (Sarkar 2003), relationship between language and its speakers (Sarkar 2005), and thoughts of Bengali intellectuals about the language (Sarkar 2006), others refer to trends of present linguistic thought and activity in Bengali (Bandyopadhyay 2002), problems and future of the language (Bandyopadhyay 2004), the present status and position of Bengali and its dialects (Bhaumick 2003), the use and abuse of Bengali (Das 2002), and dangers and prospects of Bengali in future sociocultural diversity of the region (Chakarbarty 2006, Ghosh 2007). Some e¤orts are devoted to distinctions between Bengali speech and writing (Ghosh 2007), issues related to the standardization of technical and scientific terms in Bengali (Huda 2005), the interfaces between the language and literature (Bhattacharya 2005), and the relations between the brain and the language (Bhaumick 2002).

3. Bengali spelling The Bengali spelling system has been one of the most sensitive issues for various reasons for generations, possibly because it is related to the linguistic identity and solidarity of Bengali speakers. Over the years several analyses and investigations, propositions and criticisms have often been put forward by some scholars. As there are inconsistencies in Bengali spelling, people face problems with regard to correctness of spelling, caused from debatable propositions made by experts. Though there are some proposals for standardization (cf. Sarkar et al. 2003, Mukhopadhyay 2004, Biswas 2005, Chakrabarti 2006, Mukhopadhyay 2007), the problem still remains unresolved as reflected in the observations made by some other scholars (Bhattacharya 2003, Chattopadhyay 2004, Saha 2004, Khan 2005, Nath 2004, Chakrabarti 2005, Dash 2004, Mandal 2004, Mukhopadhyay 2005, and Dash 2006). We come across an attempt for studying empirically the patterns of variation of spelling found in written Bengali texts for the first time in Dash (2004). It opens up avenues for theoretical and practical debates regarding standardization and acceptance of Bengali spelling (Dash 2006). Although this study leans towards investigating chronological development of ideas and thoughts regarding the modification and the standardization of Bengali

Linguistic Publications in Bengali (2000–2008): A Brief Review

219

spelling forwarded by scholars over the centuries (Bhattacharya 2006), it finds the proposal for new spelling (Basu 2002) non-realistic for it tries to go against the age-old norms and practices of the spelling system the native Bengali speakers are used to.

4. Morphology, syntax and lexicology Numerous publications address several complex issues related to Bengali morphology, syntax and lexicology within traditional linguistics. While some scholars try to highlight the anarchy observed in the modern Bengali grammars (Bandyopadhyay 2004), others try to record the history of birth and growth of Bengali grammar (Das 2000), and others try to discuss the general properties and features of the modern Bengali grammar solely from a synchronic perspective (Sarkar 2006). The generative model of the grammatical formalism is also an area of interest for some Bengali scholars. Following this scheme, they try to show how di¤erent Bengali phrases are constituted and arranged (Chakrabarti 2004) and how theories and principles of generative syntax are applicable to Bengali sentences (Chakrabarti 2000, Chakraborty 2006). Others try to explain how the Bengali conjunctive verbs play crucial role in expressing modal di¤erences in past tense (Dasgupta 2000) and how they vary in usage in accordance with variations of mood of speakers (Dasgupta 2005). Although serious discussion on Bengali lexicology has been neglected for years, e¤orts are now being made for enlisting the most popular modern Bengali words (Chaudhuri 2000) to define patterns of error in their use in written texts (Chaudhuri 2002). Following historical lexicology, some have tried to trace the history and origin of some of the Bengali words (Pal 2004, Biswas 2006), while others have reflected on the extralinguistic factors that are associated with some Bengali words of native and foreign origin (Chaki 2002, Kabiraj 2006). Some have also tried to analyze the form and nature of onomatopoeic words used in modern Bengali by compiling a dictionary of such words (Dakshi 2001).

5. Semantics Barring a few attempts made in earlier years, we hardly come across any work of recent past where modern theories and approaches to semantic

220 Niladri Sekhar Dash studies have been used with reference to Bengali. The traditional theories of semantics are, however addressed in a few publications together with evidences gathered from the Bengali texts (Das 1995). This trend will continue in future unless serious attempts are made to exploit this area with new theories and models. The availability of electronic corpus of written Bengali texts has, however, inspired some scholars to investigate lexical polysemy and homonymy in Bengali (Dash 2003). The analysis of empirical examples opens up possibilities for finding out linguistic factors (e.g., contexts, situations, registers, motives, etc.) that appear important (Dash 2004) for designing computerapplicable model for understanding the nature of lexical polysemy and for extracting the contextual senses of the polysemous words from written Bengali texts (Dash 2007: 139–178).

6. Language teaching Teaching of Bengali is a highly captivating area of linguistic exercise for some Bengali scholars. With close reference to the age-old pedagogic systems they have proposed utilizing Bengali linguistic resources for this purpose or suggested for adapting new strategies for teaching Bengali to learners. To mention a few, while some have tried to explore psychological intricacies involved in learning Bengali (Chakrabarty and Bhattacharya 2005), and the nature of morality-based language education imparted to the learners as reflected in Bengali prose texts (Khastagir 2005), others have tried to estimate the advantages and disadvantages of course materials and curriculum recommended for the secondary level (Sen 2004). A notable variation in teaching Bengali is observed in publications that attempt to prescribe the ways and means for s´uddha (i.e. correct) and good Bengali writing (Chaki 2007), correct ways of speaking it (Sarkar 2006), methods for proper Bengali writing (Mamud 2005), and guidelines for the use and the abuse of Bengali both in speech and writing (Lahiri, Sharif, Rahman, Kaium and Anisujjaman 2005). Strategic change in approach for adopting new techniques for teaching Bengali is recommended in a recent work (Dash 2007) in which the author has argued for utilizing the Bengali language corpus as a resource for direct class-room teaching as well as for developing language teaching materials including syllabuses, textbooks, grammars, basic and graded vocabularies,

Linguistic Publications in Bengali (2000–2008): A Brief Review

221

and dictionaries, etc. It has been argued that the Corpus-Assisted Language Teaching (CALT) technique proposed here has several advantages both at the primary and advanced level of language teaching in which the learners are allowed to access the language databases processed by and preserved in computers for enhancing their all-round linguistic skills.

7. Lexicography Dictionary compilation seems to be a particularly productive area. Within last few years, we have come across the publication of several monolingual Bengali dictionaries covering various areas of our interest. For instance, compiling dictionary of Bengali slang and jargons has been an area of keen interest for many scholars (Roy Chaudhury 2001, Bhattacharya 2002, Ray 2004, and Basu 2005). Some others have compiled dictionary of English words used in Bengali (Musa and Elias 2002), dictionary of pseudo names (Ghosh 2007), dictionary of foreign words and terms used in Bengali texts (Dasgupta and Das 2006), dictionary of person names (Maiti 2005, Pal 2006), dictionary of quotations (Mitra 2006), dictionary of linguistics (Dakshi 2003), dictionary of Bengali language and literature (Bhattacharya 2006). Attempts have also been made for compiling general reference dictionaries in Bengali (Mukhopadhyay 2005) as well as dictionaries for Bengali spellings (Sarkar et al. 2003, Mukhopadhyay 2004, and Chakrabarti 2006). We have come across a work, which presents a short history of the chronological development of dictionary making in Bengali covering 18 th –20 th century. It includes ‘introductions’ of some early Bengali dictionaries to provide some ideas how earlier scholars realized the importance of dictionary in Bengali and how they made e¤orts to excel in the work (Mishra 2000). Despite so much interest in and enthusiasm for dictionary compilation, no one has yet tried to deal with the theoretical and technical issues related to lexicography, although a few sporadic attempts have been made to highlight some rudimentary issues related to this (cf. Bhaumik 1999). It is disappointing to refer to a recent attempt that fails miserably to focus systematically on the contents and methodologies of a few Bengali dictionaries as well as fails to highlight how and where these dictionaries excel, di¤er or falter (Bandyopadhyay 2005).

222 Niladri Sekhar Dash 8. Sociolinguistic studies Sociolinguistics, due to its panoramic scope, diverse aspects, and profound appeal, is one of the most captivating areas of linguistic research among the Bengali scholars. Although serious sociolinguistic research into the social fabric of the Bengali language started nearly a few decades ago, most of the studies carried out in the recent past concentrate only on a few aspects of Bengali language and life, relating, in particular, to the patterns of use of Bengali proverbs and idiomatic expressions, slang, and gender-based use of language. We come across publications that focus on use of proverbs in children riddles and rhymes (De 2002), form, content and context of use of proverbs in general (Kabiraj 2002), and socio-cultural interfaces operating behind the use of idiomatic and proverbial expressions in Bengali and other Indian languages (Dutta Majumder 2004). While some works are compilation of idioms and proverbs used in language, others add necessary sociolinguistic interpretations to their works (Dutta 2004, Basak 2007). These works are useful in the sense that they contribute in understanding the inner life of the Bengali speech community. That Bengali people are equally skilled in using slang can be measured from some recent studies (Goswami 2000, Ray 2004). These publications not only include exhaustive lists of slang and lexical items related to cants and taboos, they also investigate the socio-cultural and psychological factors operating behind the use of such linguistic items in regular Bengali speech and writing. Also, some borrowed lexical items related to these areas are curiously investigated to draw inferences about the wider etymological spectrum of the slang used in the language (Basu 2005). Following the widely accepted norms used for identifying sex-based language, some scholars have attempted to show how Bengali women use specific words and terms hardly used by their male counter parts (Basu Dutta 2000, Bhattacharya 2000). Also they show how many gender-based forms and terms are stored and used in Bengali folk songs, riddles, rhymes and proverbs, which directly define the role of gender and power in the use of language of the speakers. Bengali is, unfortunately, also inflicted with ‘sexism’ as male and female members of the society try to maintain di¤erent relationships to the language they use in the daily course of their living (Chakraborty and Bhunia 2007).

Linguistic Publications in Bengali (2000–2008): A Brief Review

223

Some sociolinguistic publications deal with the language of the people living in fringe areas (Barua 2000), method of studying folk language and culture (Chaudhury 2007), and terminology databases used in folkloristics (Chaudhury 2007). Some others refer to various issues of sociolinguistics in the form of text and reference books written for college and university students and people especially interested in these areas (Humayun 2001, Bandyopadhyay 2003).

9. Dialect studies Bengali has several dialects spoken in various geographical areas of West Bengal and Bangladesh. Also, dialectal varieties of the language are spoken in other parts crossing the barriers of state and national boundaries. With so many dialectal varieties in Bengali, it is quite normal that people will be interested to focus on the uniqueness reflected in each variety as well as to highlight how the varieties relate to each other and to the standard one (Bhaumick 2003). With a background of many studies (professional and amateurish) of dialects of Bengali it becomes di‰cult to refer to the salient points of each study within the frame of present discussion. Therefore, the best option open to me is to refer to a few studies considered representative to the general discussion of dialects of the language. The study of dialect and culture of the Shersabadiya speech community of the district of Maldah, West Bengal (Karim 2001), the dialect of the Rarha region of Bardwan, West Bengal (Ray 2001), the dialect of Diamond Harbour region in South 24 Parganas, West Bengal (Hardar 2002), and the dialect of Midnapore, West Bengal (Pal 2003) are worth mentioning for their detailed coverage of linguistic features observed in the dialects. Across the border of the state, we come across studies about Bengali dialects spoken in Jharkhand (Ghoshal 2004) and in Barak Valley of North Kachar district of Assam (Chakrabarty 2005). These are worth mentioning for their systematic attempt for recording linguistic features of the dialects. The investigators have addressed, within traditional frame, issues related to phonology, morphology, semantics, lexicology, proverbs, phrasal units, and syntax, etc. Over the border of the country, we come across an interesting study relating to semantic closeness of Bengali and Nepalese vocabulary against the background of close cultural and linguistic interfaces underlying between the languages and people (Yash 2005). Also, we have come across

224 Niladri Sekhar Dash a study which shows how sentences of the dialect of Barishal district di¤er from the standard Bengali sentences used in Bangladesh (Mallick 2006). Since these studies are conducted in traditional frameworks, we discern several limitations in selection of dialect communities, planning in dialect surveys, preparation of questionnaire, training of field workers, selection of informants, process of interviews, manner of samples collection, method of data compilation, and technique of data processing, analysis and inference drawing. These limitations lead us to argue for incorporating corpus-based approach accepted as an authentic method of dialect research in the present global context (Dash 2007: 101–138).

10. Stylistics Recently Bengali scholars have turned their attention to stylistics to address some major issues of stylistics with close reference to Bengali texts – both prose and poetry (Chakrabarti 2003). Although stylistics is not yet explored in full, attempts are made to highlight various important areas of stylistics relating to history of prose writings in Bengali (Mukhopadhyay 2000, Bishi and Dutta 2006), use and impact of symbolism and metaphors in Bengali literary works (Ray 2007), conceptual interfaces between language and style (Bhattacharya 2006), and inherent nature of rhetoric and prosody in Bengali poetic works (Das 2001). Theoretical issues of stylistics are not ignored altogether. While some focus on the rudimentary issues of stylistics (Ray 2006), others concentrate on major principles, theories, and models of stylistics used in analysis and interpretation of literary texts (Majumdar 2006, Majumdar 2007), and some others try to show how various stylistic strategies are deployed in formation of Bengali prose texts (Dasgupta 2006) as well as in the language used in broadcasting and telecasting (Dash 2007). The analysis of style of composition of individual authors also receives attention from some scholars. While some tend to use statistical methods to calculate frequency of use of linguistic-stylistic properties in the writings of Rabindranath Tagore, the Nobel Laureate (Mallick and Nara (2000), others identify the salient stylistic aspects in the poetic writings of the same author (De 2006). Similar attempt is made to find the most notable characteristic stylistic features in prose texts of Iswarchandra Vidyasagar, the Father of Bengal Renaissance (Dutta 2006).

Linguistic Publications in Bengali (2000–2008): A Brief Review

225

11. Corpus linguistics The corpus-based study of Modern Bengali is initiated in Dash (2002). With a goal for popularizing this, Dash publishes regularly (Dash 2002, Dash 2003, Dash 2004, Dash 2005), where he argues for the study of the Bengali language with data, information and examples directly obtained from corpora designed with wide representative samples of texts of regular use of the language (Dash 2006). To argue how corpus-based approach is e¤ective and useful in applied linguistics, Dash (2007) shows the utility of the Bengali corpus in language teaching, dialectology, dictionary making, word-sense disambiguation, and machine translation. Data and information procured from Bengali corpus contribute towards solving problems related to Bengali pronunciation and spelling (Dash 2006) and information about the frequency of use of Bengali graphemes in written texts contributes in designing language teaching text books and reference materials (Dash 2008). Since only a few such studies are carried out in Bengali, we need to start serious work in this direction for developing electronic dictionaries, thesauruses, grammars books, graded vocabularies, and language teaching materials. These goals can be achieved only when we succeed to combine Bengali corpora and computer to devise intelligent systems to address the needs of the time. For instance, if we can develop a system for measuring readability of the Bengali texts, we can have better understanding of the language to develop fruitful teaching texts materials and linguistic resources for the new generations as well as for the non-native learners.

12. Conclusion This report is but a brief survey of the linguistic research publications in Bengali language during the first few years of this millennium. I am painfully aware of the fact that it fails to refer to each and every publication in the language during the period it covers, since there has been no e¤ort to systematically document linguistic publications in Bengali, as it is done for Bengali literary works. In spite of such limitations, the present survey carries some significance, since it tries to highlight the general trends of studies in Bengali language and linguistics with a focus on its future direction.

226 Niladri Sekhar Dash With optimism I have been observing an unprecedented upsurge for serious research work in Bengali during last few years, which is gathering momentum with involvement of the new generation of Bengali scholars, who are coming into this area due to new openings and opportunities opened up by information and language technology. Also, the applicationoriented linguistic research and development work are attracting large number of scholars to delve into the inner texture of the language and its properties to devise systems and software useful for the Bengali speech community. Such a paradigm shift in attitude in linguistic research and application in Bengali is bound to yield new range of publications, which may be recorded in the same manner by experts as a contribution towards the growth of information about and knowledge of the Bengali language.

References Bandyopadhyay, Debaprasad (ed.) 2004 Banglar Mukh o Shuyorer Mangsha (The Face of Bengal and Pork). Ababhas. Vol. 4 (3). Bandyopadhyay, Sandip (ed.) 2002 Bangalir Bhasachinta (Linguistic Thoughts of Bengalis). Kolkata: Progressive Publishers. 2003 Bangalir Bhasachinta: Samajbhasa (Linguistic Thoughts of Bengalis: Sociolinguistics). Kolkata: Progressive Publishers. Bandyopadhyay, Satya Ranjan 2004 Bangla byakarane arajakata (Anarchy in Bengali grammar). Ababhas 4(1–2): 74–87. Bandyopadhyay, Sucharita 2005 Adhunik Bangla Bhasatattva (Modern Bengali Linguistics). Kolkata: Pyapirus. Barua, Nihar 2000 Prantabasir Jhuli (The Bag of the Outsiders). Kolkata: Stree. Basak, Sudeshna 2007 Banglar Prabad (Proverbs of Bengal). Kolkata: Ananda Publishers. Basu, Abhra 2005 Bangla Slang: Samiksa o Abhidhan (Bengali Slang: Survey and Dictionary). Kolkata: Papyrus. Basu, Mridul Kanti 2002 Natun Bangla Banan (New Bengali Spelling). Kolkata: Mother Publishing.

Linguistic Publications in Bengali (2000–2008): A Brief Review

227

Basu Dutta, Sharmila 2000 Banglay Meyeder Bhasa (Language of Women in Bengali). Kolkata: Prama. Bhattacharya, Bitashok 2006 Bangla Bhasa o Sahityer Abhidhan (Dictionary of Bengali Language and Literature). Kolkata: Banishilpa. Bhattacharya, Mitali 2006 Bangla Banan Chintar Bibabrtan (Evolution of Thoughts on Bengali Spelling). Kolkata: Parul Prakashani. Bhattacharya, Nandadulal 2002 Bicchiri Shabder Abhidhan (Dictionary of Ugly Words). Kolkata: Ranar Prakashan. Bhattacharya, Subhas 2000 Bangalir Bhasa (Language of the Bengalis). Kolkata: Ananda Publishers. 2005 Bhasa Sahitya o Rabindranath (Language, Literature and Rabindranath). Kolkata: Nandanik. 2006 Bhasa o Shaili (Language and Style). Kolkata: Bangiya Sahitya Samsad. Bhattacharya, Sumita 2003 Bangla banan prasange (in context of Bengali spelling). Sandhitsa. (November): 56–64. Bhattacharya, Sutapa 2000 Meyeli Path (Feminine Texts). Kolkata: Pustak Bipani. Bhaumick, Nripen 2002 Bhasa o Mastiska (Language and Brain). Kolkata: Dip Prakashan. Bhaumick, Tapas (ed.) 1999 Abhidhan Sankhya (Dictionary Number). Kolkata: Korak Sahitya Patrika. 2003 Bisay: Bangla Bhasa (Issue: Bengali Language). Kolkata: Korak Sahitya Patrika. Bishi, Pramathanath and Bijitkumar Dutta (eds.) 2006 Bangla Gadyer Padanka (On the Footsteps of Bengali Prose). Kolkata: Mitra-Ghosh Publishers. Biswas, Naren 2005 Bangla banan o uchharan: samasya o samadhaner upay (Bengali spelling & pronunciation: problems and solutions), Banglabhasa Matribhasa. Kolkata: Aikatan Gabesanapatra: 79–86. Biswas, Nitish 2006 Samayer Shabdamala (Words of Time). Kolkata: Aikatan Gabesana. Chaki, Jyoti Busan 2002 Bagarthakautuki (Fun in Word Meaning). Kolkata: Ananda.

228 Niladri Sekhar Dash 2007

Shuddha Lekho Bhalo Lekho (Write correctly, write Good). Kolkata: Mitra-Ghosh Publishers. Chakrabarti, Bhabani Shankar 2006 Bangla bhasar bipad: bartaman o bhabisyat (Danger of Bengali language: present and future). Ujagar 2(1): 49–54. Chakrabarti, Biplab (ed.) 2003 Shaili: Chinatacharcha (Views and Thoughts on Stylistics). Kolkata: Ratnabali. Chakrabarti, Jagannath 2005 Barak Upatyakar Anchalik Bangla Bhasar Abhidhan o Bhasatattva (Dictionary and Linguistics of Local Bengali Dialect of Barack Valley). Haphlong: Bhasa-Sanskriti Akademi, Asam. Chakrabarti, Pranab Kumar, and Nrisimha Kumar Bhattacharya 2005 Shiksaner Manastattva (The Psychology of Education). Kolkata: Paschimbanga Rajya Pustak Parsad. Chakrabarti, Rabi 2005 Banglabhasar aitihya o banan sanskar: kothay chaleche pabhu (Tradition and Bengali spelling reform: where it is going my Lord). In Bangla Banan Bangla Bhasa 2005, R. Chakrabarti and K. Khan (eds.), 59–88. Kolkata: Kaurab Prakashani. Chakrabarti, Rabi and Kalim Khan 2005 Bangla Banan Bangla Bhasa (Bengali Spelling and Bengali Language) Kolkata: Kaurab Prakashani. Chakrabarti, Sailen 2006 Natun Bangla Banan Abhidhan (Dictionary of Modern Bengali Spelling). Kolkata: Dev Sahitya Kutir. Chakrabarti, Uday Kumar 2000 Sambartani Bangla Byakaran (Generative Bengali Grammar). Kolkata: Shri Aurobindo Publication. 2004 Bangla Padaguccher Sangathan (Structure of Bengali Phrases). Kolkata: Dey’s Publishing. Chakraborty, Nilima 2006 Bangla Bhasa o Chamskitattva (Bengali Language and Chomskyan Theories). Kolkata: Indus Publishers. Chakraborty, Rajib and Aditi Bhunia 2007 Does sexist language really exist: an assessment of the debate over sexist-non-sexist language in the context of Bengali. In, Rainbow of Linguistics, Vol. II, 2007, Niladri Sekhar Dash, Probal Dasgupta, and Pabitra Sarkar (eds.), 29–43. Kolkata: T. Media Publication. Chattopadhyay, Krishna 2004 Banan laiya gujabe kan diben na (Do not believe in rumors of spelling). Ababhas 4(1–2): 154–159.

Linguistic Publications in Bengali (2000–2008): A Brief Review

229

Chaudhuri, Bidyut Baran 2000 Janapriyatama Bangla shabda (Most popular Bengali words). Desh. 5 th August. Pp. 53–57. 2002 Bangla shabdabhuler calchitra (Patterns of error in Bengali words). Desh. 18 th May No: 31–39. Chaudhury, Dulal 2007 Lokasanskriti: Paribhasa (Folkloristics: Technical Terms). Kolkata: Pustak Bipani. 2007 Lokasanskriti: Samiksa Bijnan (Folkloristics: Survey) Kolkata: Pustak Bipani. Dakshi, Alibha 2001 Bangla Dhvanyatmak Shabda (Bengali Onomatopoeic Words). Kolkata: Subarnarepha. 2003 Bangla Bhasabijnan Abhidhan (Dictionary of Bengali Linguistics) Kolkata: Sanskrit Pustak Bhandar. Dash, Bhabesh (ed.) 2007 Sampracharer Bhasa o Bhangi (Language of Advertisement). Kolkata: Gangchil. Das, Nirmal 2000 Bangla Bhasar Byakaran o Tar Kramabikas (Bengali Grammar and it’s Evolution). Kolkata: Rabindrabharati University Press. 2006 Bhasa Paricched (Articles on Language). Kolkata: Sahityalok. Das, Ramaprasad 1995 Shabda o Artha: Shabdarther Darshan (Words and Meanings: Philosophy of Word Meaning). Kolkata: Ananda Publishers. 2000 Bhasar Baniyad: Kichu Prasanga (Foundation of Language: Some Issues). Kolkata: Paschimbanga Bangla Akademi. 2002 Kathar Karma o Apakarma (Use and Abuse of Language). Kolkata: Paschimbanga Bangla Akademi. Das, Uttam 2001 Bangla Chander Antahprakriti (Internal Nature of Bengali Prosody). Baruipur: Mahadiganta. Dasgupta, Ajay and Mrinal Kanti Das 2006 Banglay Atithi Shabder Abhidhan (Dictionary of Guest Words in Bengali). Kolkata: Punascha. Dasgupta, Dhiman 2006 Gadyagathan (Formation of Prose). Kolkata: Banishilpa. Dasgupta, Probal 2000 Otiter tin mejaj ar shikaronshonggi jojok-kriya (Conjunctive verbs to refer to the variations of mood in past tense). In Samaj Sangskriti Itihas 2000, R. Chakrabarti, K. Chakrabarti, A. Banerji (eds.), 52– 59. Kolkata: Ananda Publishers.

230 Niladri Sekhar Dash 2005

Mejajer herpher ar Bangla kriyapad (Variations of mood and Bengali verb). Anustup 39(4): 105–115. Dash, Niladri Sekhar 2002 Bhasansha bhasatattva: bhasabijnaner natun diganta (Corpus linguistics: a new horizon of linguistics). Ababhas 2(2): 36–62. 2003 bahudharthak shabder prakrita pratibeshik artha nirupan (Exploring contextual sense of polysemous words). Sahitya Parisat Patrika 109(1–2): 145–178. 2004 Upabhasa carcay prak-baidyutin bhasa-namunar byabahar (Use of pre-electronic corpus in dialectology). Sahitya Mela 27 (1): 31– 43. 2004 Bhasangsha nirman (Developing a corpus). Bangla Boi 61: 1–4. 2004 Bangla likhita Bhasangshe bananer byabhar baichitrya (Variation of spelling in written Bengali corpus). Sahitya Parisat Patrika 110(4): 166–192. 2004 Shabder artha o pratibesh (Word meanings and contexts). Bhasa 7(1): 80–95. 2005 Upabhasa: svatantrer sankat o svaruper sandhan (Dialect: crisis and search for its identity). Sahitya Parisad Patrika 112(1–2): 37– 48. 2005 Bhasa-prayukti: bharater calcitra o banglar mukh (Language technology: the Indian scenario and the face of Bengal). Bangla Boi 68: 1–3. 2005 Bhasa-prayukti o Bangla Bhasa (Language technology and Bengali language). Computer Jagat. April–May 2005. 2006 Bahurupi Bangla Banan (Multifaceted Bengali Spelling). Kolkata: Dakshabharati. 2006 Bhasangsha Samgraha o Adhunik Bhasabijnan (Language Corpora and Modern Linguistics). Kolkata: Dakshabharati. 2006 Sadharan Bangla likhita bhasangsha nirman: paddhati o prakaran (Process of designing a general written Bengali corpus). Vinirman 3(4): 35–62. 2006 Lokabhasar gabesanay bhashangsher prasangikata (Relevance of corpora in dialectology). Loukik 1(2): 32–51. 2006 Baidyutin Bangla abhidhan nirman paddhati (Method of designing an electronic Bengali dictionary). Computer Jagat. January– December 2006. 2006 Bhasa-prayukti: kaler kallol kimba pragatir pathanirdeshika (Language technology: call of the time or direction for the future). Image 2(1): 28–47. 2007 Baidyutin abhidhaner byabahargata subidha (Functional advantages of electronic dictionary). Vinirman 5(1): 19–28.

Linguistic Publications in Bengali (2000–2008): A Brief Review 2007

231

Bangla bhasar baidyutin shabdabhandar nirman (Developing an electronic lexicon in Bengali). Computer Jagat 3(1): 15–44. Calcutta Book Fair Number. 2007 Bangla bhasa prakriyakaran (Bengali language processing). Computer Jagat. February–December 2007. 2007 Language Corpora and Applied Linguistics. Kolkata: Sahitya Samsad. 2008 Bangla Barnamala: Byabahar-bhittvik Bishlesan (Usage-based Analysis of Bengali Orthographic Symbols). Kolkata: Sahitya Samsad (In Press). De, Sushil Kumar 2002 Bangla Prabad:Chara o Calti Katha. (Bengali Idioms: Children Rhymes and Popular Lore). 5 th Edition. Kolkata: A. Mukherji. De, Ujjwal Kumar 2006 Rabindrakabita: Bhasaritigata Paryabeksan (Poems of Tagore: Linguistic-stylistic Analysis). Kolkata: Pustak Bipani. Dutta Majumdar, Sabita 2004 Prabad Prabachan (Idioms and Proverbs). Kolkata: Thema. Dutta, Bibhabasu 2006 Bidyasagarer Gadyashaili (Prose-style of Vidyasagar). Kolkata: Pustak Bipani. Dutta, Kalyani 2004 Prabadmala (Benglai Idioms). Kolkata: Thema. Ghosh, Dipankar (ed.) 2007 Bhasa-Bhabna: Unish-Bish Shatak (Linguistic Thoughts: 19 th and 20 th Century). Kolkata: Paschimbanga Bangla Akademi. Ghosh, Harsa Bardhan 2007 Chadmanamer Abhidhan (Dictionary Pseudo Names). Kolkata: Dip Prakashan. Ghosh, Nityapriya (ed.) 2007 Mukher Katha Lekhar Bhasa (Spoken Words and Written Language). Kolkata: Bookfront Publication Forum. Ghosh, Saurindra Kumar 2006 Bangalir Jati Parichay (Identity of Bengali). Kolkata: Sahityalok. Ghoshal, Chanda 2004 Jharkhandi Banglar Jharganyi Rup (Local Form of Jharkhandi Bengali). Kolkata: Lokasanskriti O Adivasi Sanskriti Kendra. Goswami, Satrajit 2000 Bangla Akathya Bhasa o Shabdakos (Bengali Slang and Dictionary). Kolkata: Joydurga Library. Haldar, Balaichand 2002 Diamond Harbar Anchaler Bangla Upabhasa (Dialect of Diamond Harbor). Kolkata: Pustak Bipani.

232 Niladri Sekhar Dash Huq, Anamul 2001 Adhunik Bhasatattver Svarup o Prayog (Nature and use of Modern Linguistics). Dhaka: New Shikha Prakashani. Huda, Nurul 2005 Bangla paribhasa pramitakaran prasanga (Standardizing of Bengali terminology). Banglabhasa Matribhasa. Kolkata: Aikatan: 87– 89. Humayun, Rajib 2001 Samajbhasabijnan (Sociolinguistics). Dhaka: Agami Prakashan Huq, Muhammad Daniul 2002 Bhasabijnaner Katha (Introduction to Linguistics). Dhaka: Mowla Brothers. Kabiraj, Bijay 2002 Bangla Bagdhara Prasanga Prayog (Context and Use of Bengali Proverbs). Kolkata: Punascha. 2006 Shabdamadhuri (Beauty of Words). Kolkata: Punascha. Karim, Mir Rejaul 2001 Shersabadiya Sampradayer Bhasa o Sanskriti (Language and Culture of Shersabadiya Community). Kolkata: Pustak Bipani. Khan, Kalim 2005 Banglabhasar banan samskar: agun phele anch!? (Reformation of Bengali spelling: playing with fire!?). In Bangla Banan Bangla Bhasa, R. Chakrabarti and K. Khan (eds.), 13–58. Kolkata: Kaurab Prakashani. Khastagir, Ashis 2005 Bangla Gadye Nitisiksa (Moral Education in Bengali Prose). Kolkata: Pustak Bipani. Lahiri, S., A. Sharif, D. Rahman, A. Kaium, M. Anisujjaman 2005 Bangla Bhasar Prayog o Apaprayog (Use and Abuse of Bengali Language). Dhaka: Bangla Academy. Maiti, Chitya Ranjan 2005 Bangla Namkaran Abhidhan (Naming Dictionary in Bengali). Kolkata: Dev Sahitya Kutir. Majumdar, Abhijit 2006 Chandimangal: Shailibicar (Chandimangal: Stylistic Analysis). Kolkata: Dey’s Publishing. 2007 Shailibijnan o Adhunik Sahityatattva (Stylistics and Modern Literary Theories). Kolkata: Dey’s Publishing. Mallick, Bhakti Prasad and Tusuki Nara (ed.) 2000 Sheslekha: Bhasatattvik-sankhyatattvik Bishlesan (Sheslekha: Linguistic-Statistical Analysis). Kolkata: Paschimbanga Bangla Akademi.

Linguistic Publications in Bengali (2000–2008): A Brief Review

233

Mallick, Nur Muhammad 2006 Barishaler upabhasay bakyatattva (Syntax of Barishali dialect). Bangla Akademi Patrika (19–20): 198–219. Mamud, Hasan 2005 Bangla Lekhar Niyamkanun (Rules of Bengali Writing). Dhaka: Bangla Academy. Mandal, Kalyan 2004 Ucchrinkhal Bangla banan: shrinkhal-bhanga shrinkhala (Indisciplined Bengali spelling: system breaking system). Thirabijuri 4(7): 97–101. Mishra, Saraswati 2000 Bangla Abhidhaner Kramabikash (Evolution of the Bengali Dictionary). Kolkata: Pustak Bipani. Mitra, Dilip Kumar 2006 Uddhriti Abhidhan (Dictionary of Quotations). Kolkata: Parul Prakashani. Mukhopadhyay, Ashoke 2004 Samsad Banan Abhidhan (Samsad Spelling Dictionary). Revised and enlarged 2 nd Edition. Kolkata: Sahitya Samsad. 2005 Banglay yuktavarna o kichu prasangik prashna (Bengali cluster and some related questions). Bangla Bai: 1–3. Mukhopadhyay, Amitabha 2005 Sahaj Bangla Abhidhan (Easy Bengali Dictionary). Kolkata: MitraGhosh. 2007 Bananer Hatbai (Handbook of Spelling). Kolkata: Mitra-Ghosh. Mukhopadhyay, Arun Kumar 2000 Bangla Gadyaritir Itihas (History of the Bengali Prose Style). Kolkata: Dey’s Publishing. Musa, Monsur 2002 Bhasachinta: Prasanga o Paridhi (Language Thoughts: Context and Scope). Dhaka: Bangla Academy. Musa, Monsur and Monoara Elias 2003 Banglay Pracalita Inreji Shabder Abhidhan (Dictionary of English Words in Bengali). Dhaka: Mouli Prokashoni. Nath, Mrinal 2001 Bish shataker bhasatattva o bhasacharcha (Linguistics and language studies in the 20 th century). Parikatha 4(1): 125–162. 2004 Duhkhini barnamala (Unfortunate Bengali script). Ebang Mushayera 11(1): 62–73. Pal, Animesh Kanti 2003 Medinipurer kathyabhasa ba upabhasa (Dialect of Midnapore). In Bisay Bangla Bhasa, T. Bhaumik (ed.), 393–397. Kolkata: Korak Sahitya Patrika.

234 Niladri Sekhar Dash 2004

Bhasabijnan o Bangla (Linguistics and Bengali). Kolkata: Bama Pustakalaya.

Pal, Sukumar 2006 Bangla Byaktinam Abhidhan (Dictionary of Bengali Person Names). Kolkata: D.M. Library. Ray, Ajit (ed.) 2004 Shahar: A Special Issue on Bengali Slang. Vol. 29. No. 1. 2004 Chotoloker Shabdalok (Language of the underworld). Shahar 29(1): 25–39. Ray, Ajay 2007 Adhunik Bangla Sahitye Rupak o Pratik (Metaphor and Symbols in Modern Bengali Literature). Kolkata: The Printobooks. Ray, Apurba Kumar 2006 Shailibijnan (Stylistics). Kolkata: Dey’s Publishing. Ray, Bhaba 2001 Rahrer Lokabhasa o Shabdakos (Dialect and Dictionary of Rahra). Kolkata: Dipayan. Roy Chaudhury, Manas Kumar (ed.) 2001 Bangla Ashista Shabder Abhidhan (Dictionary of Bengali Slang). Kolkata: Abhijata Prakashani. Saha, Apurba Kumar (ed.) 2004 Bangla Banan: Chapan Utor (Bengali Spelling: Points and Counterpoints). Thirabijuri Patrika 4 (7). Sarkar, Pabitra 2003 Bhasaprem Bhasabirodh (Language Love and Language Conflict). Kolkata: Dey’s Publishing. 2005 Kathasutra (String of Words). Kolkata: Mitra O Ghosh. 2006 Bangla Byakaran Prasanga (Issues of Bengali Grammar). Kolkata: Dey’s Publishing. 2006 Bhasamanan o Bangali Manisa (Language Thoughts and Bengali Intellectuals). Kolkata: Punascha. Sarkar, P., A. Mukhopadhyay and P. K. Dasgupta (eds.) 2003 Akademi Banan Abhidhan (Akademi Spelling Dictionary). 4 th Edition. Kolkata: Paschimbanga Bangla Akademi. Sen, Subhadra Kumar 2004 Natun pathyakram prasange (About new syllabus). Ababhas 4(1–2): 18–28. Singh, Udaya Narayana 2000 Bhasa, bhasi, bhasiki (Language, speakers, and speaking style). Bahubacan 2(1): 33–57 & 2(2): 305–317. 2004 Bhasa-sanraksan o bhasa parikalpana prasange (In the context of language preservation and planning). Bhasa 7(1): 23–33.

Linguistic Publications in Bengali (2000–2008): A Brief Review Sur, Atul 2001

235

Bangla o Bangalir Bibabartan (Evolution of Bengali language and Speakers). 3 rd Edition. Kolkata: Sahityalok. Yash, Subodh Kumar 2005 Sanskritik Preksapate Bangla o Nepali Shabdartha Samiksa (Survey of Bengali and Nepali Word Meanings at the Backdrop of Cultural Fusion). Kolkata: Pustak Bipani.

South Africa

Rajend Mesthrie

Since my last report of 1998 South Asian languages cannot be said to feature prominently in South African linguistic research. This is partly due to the ongoing language shift in the country’s Indian community and partly due to the state of linguistic research with specialist researchers in Indian languages being few and far between. Overall, there has been a shift to more applied linguistic than descriptive linguistic work in the country as a whole, driven by the country’s developmental and educational needs. This report will cover both Indian languages as well as the ‘Indian’ variety of English that is gradually replacing them. It will also cover Arabic, which is an important language in the Indian community’s repertoire (as well as that of people identified as ‘Malay’). Studies of three Indian languages in diaspora appeared in this period. Usha Desai, who had written an M.A. thesis on the external history and sociology of Gujarati in the province of Natal (now known as KwaZuluNatal) in 1992, extended this work at PhD level in 1997 to cover the other provinces of the country where the language exists in reasonably significant numbers. As in her earlier work the themes of maintenance and shift in diaspora are prominent. The same themes occur in Varijaskshi Prabhakaran’s PhD thesis of 1991 on the history and sociocultural change found in Telugu in South Africa. Two articles ensuing from the thesis research can be found in the South African journal Alternation (Prabhakaran 1997) and the international journal Multilingua (Prabhakaran 1998a). Prabhakaran’s work serves to highlight the history of Telugu and its speakers in their own right, rather than as a language that has always played second fiddle to the major Dravidian language of South Africa, Tamil. Her work contains discussions of loanwords from Tamil and other languages of South Africa. Two articles on Tamil itself appeared in the period under review: both by Rajend Mesthrie, who had worked in the 1990s in conjunction with Bavanathan Pillay, a schoolteacher and native speaker of the South African form of the language. Mesthrie (2007a) provides a bird’s-eye view of South African Tamil

238 Rajend Mesthrie in the South African journal Language Matters, detailing the lexical changes in Tamil consequent upon its plantation history in KwaZulu-Natal province. These changes include loanwords from a number of South African languages like Zulu (especially pidgin Zulu or Fanakalo), English and Bhojpuri. The article suggests that Tamil in South Africa did not undergo as many changes as had Bhojpuri. The latter is a koine´, the former not. The reasons for this di¤erence can be found in Mesthrie (2007b), an article in the Indian journal Oriental Anthropologist on the dialect origins of South African Tamil. By examining the ships’ lists of the 19 th and early 20 th centuries Mesthrie shows that the overwhelming majority of Tamil-speaking migrants (85.5%) came from what Indian linguists have called the Northern dialect area, i.e. from North and South Arcot and Chingleput. The South African variety therefore has close similarities to the Northern dialect area, especially to the dialect of the district of North Arcot, the district producing the most migrants. This influence shows up mainly in lexical items and some phonetic realisations, but apparently not in the syntax. On the other hand, very few traces of the defining characteristics of the other Indian dialects of Tamil (central, eastern and western) can be found in South Africa. This contrasts strongly with Bhojpuri in South Africa, whose speakers emanated from a much wider area, and formed a koine´ that shows characteristics from many dialects and languages, chiefly Bhojpuri, Awadhi, Magahi and Hindi. Three articles giving an overview of Indian languages in South Africa appeared in the period under review. Mesthrie (2002a) provides a chapterlength overview of the 140 year history of Indian languages in South Africa. Prabhakaran (1998b) wrote a chapter on Indian languages in KwaZuluNatal, confirming the state of advanced language shift noted by earlier researchers. She also details some adherence to the ethnic mother tongues in terms of their use in religious and formal cultural domains (e.g. weddings) and in the number of students studying them at school. In 1996 there were 37 000 students studying one of Gujarati, Hindi, Tamil, Telugu or Urdu in the province’s schools. The other article entitled ‘Language shift, cultural change and identity retention’ (Mesthrie 2007c) examines the crucial period of the 1960s, exactly one hundred years since the first arrival of indentured Indians in Natal, for clues to why language shift started to occur in this decade. Mesthrie stresses the role of youth culture amongst urban Indians, who slowly began to build a common Indian South African identity, above allegiance to an ethnic home language. Cultural change resulted from western influences in music and dress at the beginning of the era of globalisation

South Africa

239

and was reinforced by developments in housing and education in South Africa. Two book-length studies pertaining to Arabic in South Africa’s Indian communities appeared. The first, edited by Mohamed (1997) is a collection of articles pertaining to the history and methodology of Arabic teaching in South Africa (amongst its Muslim communities of Indian and Malay backgrounds). The second is a historian’s account (Jeppie 2007) of the Arabic Study Circle of Durban, a group formed in the 1950s of mainly Gujarati Muslims dedicated to the study of Arabic for religious purposes. Both books are valuable for their insights into language teaching and study within a particular sociohistorical milieu. The flip side of the decline of Indian languages is the rise of a distinctive English in the community. Four dissertations on aspects of South African Indian English appeared in the period under review. Shelley Delbridge (2006) completed an M.A. dissertation on phonetic aspiration in South African Indian English within a sociolinguistic framework, showing that accommodation between North and South Indian speakers had occurred to a large extent by the 1980s. Sharita Bharuthram (2001) discusses politeness in the Indian community as expressed in English. Articles appearing from the research include Bharuthram (2003) and Bharuthram & De Kadt (2003). Lisa Wiebesiek (2007) examines the non-inversion of questions and lack of do-support in South African Indian English in terms of principles and parameters theory. Her data comes from students at the University of KwaZulu-Natal campus. Leela Pienaar (2007) reports on a corpus of 60 000 words that she compiled on the basis of informal interviews with speakers of South African Indian English at the Rhodes University campus. She confirms that younger educated speakers have slightly di¤erent norms from the speakers interviewed by Mesthrie in the late 1980s. However, many of the core acrolectal and upper-mesolectal features have stabilised. Pienaar’s corpus, which will be made available as part of a Corpus of Educated South African English being developed at Rhodes University, is a valuable resource tool for future researchers. Mesthrie has continued his studies of South African Indian English, with two socio-historical overviews (Mesthrie 2002b, c), articles on its phonetics (Mesthrie 2004a) and morphosyntax (Mesthrie 2004b) as well as accounts of young children’s syntax (2003), of parodies of the dialect in the 1940s (Mesthrie 2002d) and comparisons with representations in more serious literature emanating from within the community (Mesthrie 2005).

240 Rajend Mesthrie A study of the L2 interlanguage Zulu of Indian speakers was made by Mesthrie (2007d), contrasting it with the features of pidgin Zulu (Fanakalo) as spoken by Indians. References Bharuthram, Sharita 2001 Politeness: The case of apologies and requests: An intergenerational cross-sex study in the Hindu sector of the South African Indian English speaking community. M.A. diss., Department of Linguistics, University of Natal, Durban. 2003 Politeness phenomena in the Hindu sector of the South African Indian English speaking community. Journal of Pragmatics 35(10–11): 1523–44. Bharuthram, Sharita and Elizabeth De Kadt 2003 The value placed on politeness by men and women in the Hindu sector of the South African Indian English-speaking community. South African Journal of Linguistics & Applied Language Studies 21(3): 87– 102. Delbridge, Shelley 2006 Aspiration in South African Indian English – emerging phonological norms in new dialect formation. M.A. thesis, Department of Linguistics, University of Cape Town. Desai, Usha 1997 Investigation of factors influencing maintenance and shift of the Gujarati language in South Africa. PhD thesis, Department of Indian Languages, University of Durban-Westville. Jeppie, Shamil 2007 Language, Identity, Modernity – the Arabic Study Circle of Durban. Cape Town: HSRC Press. Mesthrie, Rajend 2002a Language Change, survival, decline: Indian languages in South Africa. In Language in South Africa, R. Mesthrie (ed.), 161–176. Cambridge: Cambridge University Press. 2002b Building a new English dialect: South African Indian English and the history of Englishes. In The History of English: Alternative Perspectives, P. Trudgill and R. Watts (eds.), 111–33. London: Routledge. 2002c From L2 to L1: Indian South African English. In Language in South Africa, R. Mesthrie (ed.), 339–355. Cambridge: Cambridge University Press. 2002d Mock languages and symbolic power: the South African radio series ‘Applesammy and Naidoo’. World Englishes 21(1): 99–112.

South Africa 2003

241

Children in language shift: the syntax of fifth-generation South African Indian English speakers. South African Journal of Linguistics and Applied Language Studies 21(3): 119–26. 2004a Indian South African English: phonology. In A Handbook of Varieties of English vol 1: Phonology, E. Schneider, K. Burridge, B. Kortmann, R. Mesthrie and C. Upton (eds.), 953–63. Berlin: Mouton de Gruyter. 2004b Indian South African English: morphology and syntax. A Handbook of Varieties of English vol 2: Morphology and Syntax, Kortmann, K. Burridge, R. Mesthrie, E. Schneider and C. Upton (eds.), 974–92. Berlin: Mouton de Gruyter. 2005 Assessing representations of South African Indian English in writing: an application of Variation theory. Language Variation and Change 17(3): 303–326. 2007a A bird’s eye view of South African Tamil. Language Matters 38(2): 179–94. 2007b The origins of colloquial South African Tamil. Oriental Anthropolo gist 7(1): 17–38. 2007c Language shift, cultural change and identity retention: Indian South Africans in the 1960s and beyond. South African Historical Journal 57: 134–152. 2007d Di¤erentiating pidgin from early interlanguage: pidgin Nguni (Fanakalo) versus second language varieties of Xhosa and Zulu. Southern African Linguistics and Applied Language Studies 25(1): 75–89. Mohamed, Yasien (ed.) 1997 The Teaching of Arabic in South Africa: History and Methodology. Cape Town: University of the Western Cape. Pienaar, Leelavathie 2007 Towards a corpus of Indian South African English: an investigation of lexical and syntactic features in a spoken corpus of contemporary ISAE. M.A. thesis, Department of Linguistics, Rhodes University. Prabhakaran, Varijakshi 1997 Social stratification in South African Telugu. Alternation 4(2): 162– 185. 1998a Multilingualism and language shift in South Africa: the case of Telugu, an Indian language. Multilingua 17(2/3): 297–319. 1998b Indian Languages in KwaZulu-Natal. Multilingualism in a Multicultura Context, Guus Extra & Jeanne Martens (eds.), 75–90. Tilburg: Tilburg University Press. Wiebesiek, Lisa 2007 Addressing the ‘Standard English’ debate in South Africa: the case of South African Indian English. M.A. thesis, Department of Linguistics, University of KwaZulu-Natal.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey

Joga Singh

1. Introduction The Punjabi language is spoken by approximately 120 million speakers and is about the 10 th most widely spoken language of the world. But, surprisingly, it is one of those languages which have been explored very meagerly in terms of the modern linguistic idiom. However, there have been still quite a number of writings published in Punjabi during the present decade which provide some analysis, raise some theoretical point or raise some pertinent question about the Punjabi language or linguistics in general. This article briefly introduces these writings. Surely, not all of them can be characterized as very professional studies. But this, in fact, is the nature of Punjabi linguistic scene that there is an extreme paucity of linguists trained in the most recent techniques and theories of linguistic analysis. Also, not all the periodicals mentioned here are of professional linguistics nature. But there are some literary type of journals which do publish articles which have an academic relevance. I have included these journals too for churning out the writings included here. Though, it is di‰cult to list a large number of entries under a particular rubric, I have divided these works into categories of General Linguistics, Phonology, Morphology, Syntax, Semantics, Semiotics, Historical Linguistics, Lexicography, Dialectology, Teaching of Punjabi, Language Technologies, and Language Policy/Planning. The order I discuss them in is entirely extrinsic. Further, only the works published in Punjabi are included in the report. Also, the report cannot be claimed to be all inclusive as a few write ups might have been missed. This is the first report on the linguistic research published in Punjabi. I hope this becomes a regular feature for Punjabi research.

244 Joga Singh 2. General Linguistics Dhaliwal (2002) and Saksena (2001) provide a general introduction to linguistics. Sangha (2001b) and Harkirat Singh (2004) provide a general introduction to the structure of Punjabi language. Harkirat Singh (2004) provides information on the formation of Punjabi language. Jassal (2006b) and Kang (2001) provide information on Punjabi linguists and de Saussure. Khaira (2002), Harkirat Singh (2003a), Joshi (2001a), Joga Singh (2004) and J. S. Puar (2001a, 2001b) are other works commenting on the linguistic studies and issues in Punjabi. Parmjit Singh (2005) describes how linguistic model can help analyzing the communication models. Thind (2004) is collection of articles sketching Prem Prakash Singh Dhaliwal’s life and his contribution to Punjabi linguistics.

3. Phonology This is one of the areas on which very few write ups are available. Sukhjit Kaur (2001) enumerates Punjabi suprasegmentals. Updesh Kaur (2001) discusses some phonetic properties of Punjabi word structure. Updesh Kaur (2006b) is an elaborate study of Punjabi phonology. Shastri (2001) discusses sound change in Punjabi while Gurmukh Singh (2001) advocates the suitability of Gurmukhi script for the Punjabi language.

4. Morphology Agnihotri (2001b) discovers the non-segmental a‰xes in Punjabi. Agnihotri (2001c) surveys the indeclinable categories in Punjabi. Agnihotri (2003) is a description of word formation processes in Punjabi language. Brar (2006) points out towards the creativity of morphological processes. Cheema (2002) discusses the bases of classification of words. Dhaliwal (2004b) introduces basic concepts of morphology and then provides descriptive morphological analysis of major Punjabi word classes. Kanwaljit Kaur (2001) discusses the potential of word coinage in Punjabi while Kanwaljit Kaur (2007) is a descriptive account of Punjabi pronouns. Sidhu (2002b) describes the linguistic structure of one composition of Guru Granth, the Sikh scripture.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 245

5. Syntax Syntax has relatively drawn a good attention of Punjabi linguists. Joga Singh (2003c) brings out strong evidence to refute the theoretical position that word order (in the sense of phrase order) is free in Punjabi. Boota Brar (2003d) and Joga Singh (2002) provide brief introduction to Transformational Generative grammar. The latter, perhaps, is one of the best introductions to the theoretical foundations of Chomskyan syntax. Boota Brar (2003) and Cheema (2001) discuss the present stage of Punjabi syntactic studies while Bhupinder Kaur (2001) is a discussion on old Punjabi grammars. Agnihotri (2007), Boota Brar (2001), Cheema (2003a, 2003b, 2007), Dhandli (2003), Jassal (2002), S. S. Joshi (2001c) and Sewak Singh (2002) are other works devoted to various aspects of Punjabi syntax.

6. Semantics Sidhu (2006b), Sidhu and Ramanprit Kaur (2003b), and Sewak Singh (2004) describe semantics of some part or the other of the Sikh scripture.

7. Semiotics The Linguistics department at Punjabi University, Patiala (India) was founded by an internationally known semiotician Harjeet Singh Gill and it has spawned a number of semiotic studies of Punjabi language and culture. The tradition is still continued by the students of the Patiala Department. Harjit Gill (2001) contributes to understanding the basic semiological functions of language. Bajwa (2001) points out the neglect of semiotic significance of word-culture. Jeet Joshi (2004), Parmjit Sidhu (2001b), and Parmjit Sidhu and Ramanprit Kaur (2005a) describe semiotics of some folk forms. Parmjit Sidhu (2002a) describes the semiotic structure of a particular novel. The Sikh scripture and its language is a topic of quite a number of studies. Boota Brar (2003a, 2003c) and Parmjit Sidhu and Ramanprit Kaur (2003a, 2004a, 2004b) are works in this vein.

246 Joga Singh 8. Historical Linguistics Brar (2004b, 2005c), Cheema (2003c), Dhaliwal (2004a), Jassal (2004b), and Jasbir Kaur (2001) explore the origin and development of Punjabi language. Gupta (2001), Kadri (2001), Pritam Singh (2006), and Gulwant Singh (2001) are discussions on the relations of Punjabi language with one language or the other.

9. Lexicography Bahri (2001), Dhot (2001), Jeet S. Joshi (2003), S. S. Joshi (2001b), Kapoor (2004), Kumari (2001), Machwe (2001), Ryal (2001, 2002), Shan (2001), Atamjit Singh (2000), Bakhshish Singh (2006), Davindr Singh (2002), Harkirat Singh (2001), Onkar Singh (2001b), Sital (2001), Suman Preet (2005) and Tanvi (2005) discuss general issues pertaining to the practice of lexicography and comment on Punjabi lexicographic practice. The monolingual general purpose Punjabi dictionary published during this period is Hasija (2001b). Hasija (2001a), S. S. Joshi (2002), and Major Gurmukh Singh (2002) are the bilingual dictionaries. Sekhon (2007) is a dictionary of the Malwai dialect of Punjabi and Bakhshish Singh (2002) is a Punjabi-English dictionary of Punjabi idioms.

10. Dialectology Updesh Kaur (2006a) gives an account of the major dialects of Punjabi while Harkirat Singh (2000, 2003b), provide introductions to the Majhi and Multani dialects, respectively.

11. Stylistics Linguistic analysis of literary works is the most widely explored are after language policy/planning. Agnihotri (2001a), Boota Brar (2004a, 2005a, 2005b), Iqbal Dhillon (2002), Baldev R. Gupta (2000), Jassal (2003, 2004a, 2006a, 2006c, 2007a, 2007b), Palwinder Kaur (2001a, 2001b), Sangha (2001a, 2003a, 2003b, 2004, 2006), Sangha and Palwinder Kaur (2003),

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 247

Parmjit Sidhu (2006a), Parmjit Sidhu and Ramanprit Kaur (2004b, 2005c), Prem Prakash Singh (2005) and Joga Singh and Poonam Dhillon (2007) are works of this nature.

12. Teaching of Punjabi Agnihotri (2001d), Bala (2001), Bining (2005), S. S. Joshi (2003a), Ranjit Kaur (2007), Sangha and Palwinder Kaur (2001) and Agyajit Singh (2003) discuss issues such as the teaching of Gurmukhi script, teaching of various aspects of Punjabi language structure, and the teaching of Punjabi in di¤erent regions of India.

13. Language technologies Johl (2003) and Lehl and Bhatti (2003) are discussions by computer scientists relating to technological challenges facing Punjabi language while Parmjit Sidhu (2003a) and Onkar Singh (2001a) are similar discussions by language experts.

14. Language policy/planning This aspect has been most written about. One reason being the relative nontechnicality of the area but more important than this is the concern the Punjabi academia shows for the dilution in the status of Punjabi language vis-a`vis the English language. There are a number of studies discussing the state and place of Punjabi language in the contemporary Indian Punjab and the Indian context. These include Bala (2001), Jagjit Brar (2001), Dhingra (2001), Darshan Gill (2005b), S. S. Joshi (2000, 2003c), Dhanwant Kaur (2003), Madhopuri (2001), Sandhu (2005b), Sangha (2007), Amanpal Sara (2000), Gurmel Sidhu (2005), Parmjit Sidhu (2003b), Gurnam Singh (2000), Pritam Singh (2003), and Ishar S. Tangh (2000). There are a number of studies describing the state and place of Punjabi in other countries. These are – Gill (2005a) and Dhanwant Kaur (2003) for the scene in di¤erent countries around the world and Bining (2001) for Canada, Shivcharan Gill (2005) for Great Britain, Dharam Singh (2001), Kulwinder Singh (2001),

248 Joga Singh Thind (2003) and Zaman (2003) for the scene in Pakistan. The rising number of English medium schools and total paralysation of the Punjabi medium government schools has drawn a particular attention of serious linguistic scholarship. Joga Singh (2003a) provides a survey of the major research done internationally on the issue of medium of instruction. The study also discusses the place of mother tongue in the life of an individual and a society and thus strongly argues against English being used as medium of instruction. The study foretells the frightening social, educational, and political consequences of this unscientific policy. S. S. Joshi (2003b), Nirankari (2001), Sharma (2003), Joga Singh (2003b), and Surjeet Singh (2003) are the other studies devoted to the issue. Balwinder Kaur (2001), Sandhu (2001), and Onkar Singh (2003) delineate what needs to be done for the development of Punjabi language. Bhardwaj (2005) and Noor (2005) take up the issue of the suitability of particular script for the Punjabi language. Joga Singh and Gurbax Singh (2004) discuss the social forces responsible for the formation of Punjabi language in the sense of sociology of a particular language.

15. Conclusion What one can conclude from the survey of the research published in Punjabi on Punjabi during the period under consideration is that there is a discernible incremental growth in writing on various linguistic aspects of Punjabi language. Although rigorous formal analyses of Punjabi are relatively rare, many articles of semi-formal nature have been published. Language policy/planning and stylistics seem to be the fields which have occupied more space than the others. Syntax and the language of the Sikh scripture are the other relatively salient areas. However, when compared with the immediately earlier period, there does seem to be not only a noteworthy increase in publications in Punjabi on Punjabi language but also a tendency towards making them more rigorous and formal.

References Agnihotri, Ved 2001a kavita te vaartak: vaak viontak nikheRa (Poetry and prose: Syntactic dinstinction). Parvachan 4: 22–30.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 249 2001b

panjaabi vich akhanDi vadhetraan di dhuniviont (Phonology of nonsegmental panjabi a‰xes). Panjabi Darpan 47: 113–120. 2001c panjaabi vich nipaat shabdaavli: ik jaaija (Indeclinables in Punjabi: An appraisal). In Bhaasha Vigiaan: Sankalp te Dishaawaan, (Linguistics: Concepts and dimensions), J. S. Puar (ed.), 90–97. Jallandhar: Punjabi Bhasha Academy. 2001d hindi bhaashi ate panjaabi sur parbandh (Punjabi tone system and the Hindi speakers). In Bhaasha Vigiaan: Sankalp te Dishaawaan (Linguistics: Concepts and dimensions), J. S. Puar (ed.), 188–201. Jallandhar: Punjabi Bhasha Academy. 2003 shabad rachna diiaan vidhiian (Processes of word formation). Parvachan 10: 33–39. 2006 panjaabi bhaasha da vachan parbandh (Number system of the Punjabi language). Samdarshi 91: 33–38. 2007 naanv vaakansh di baNtar (Structure of the noun phrase). Parvachan 29: 21–25. Bahri, Ujjal S. 2001 panjaabi koshkaari vich Dogri (Dogri in Panjabi lexicography). Panjabi Dunian 51 (10): 70–84. Bajwa, Ranjit S. 2001 ikkviin sadi vich shabad sabhiaachaar nu darpesh chuNautiaan te samassiaanvaan (The challenges and problems facing word-culture in the twenty first century). Khoj Darpan 47: 137–142. Bala, Madhu 2001 panjaabi bhaasha nav-parpekh de sandarbh vich (The Punjabi language in the new context). Khoj Darpan 47: 143–150. 2003 hariaaNe vich panjaabi bhaasha di sathiti ate adhiaapan (The state and teaching of Panjabi language in Haryana). Alochana 200: 135– 138. Bhardwaj, Mangat R. 2005 maamla phirkaaprasti da (The issue of communalism). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 20–31. Ludhiana: Chetna Parkashan. Bhatti, H. S. and Gurpreet S. Lehl 2003 panjaabi bhaasha da kampiuTrikaran (Computerisation of Panjabi language). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 129–138. Patiala: Punjabi University. Bining, Sadhu S. 2001 kaneDa vich panjaabi bhaasha te is da bhavikkh (Panjabi language and its future in Canada). SirjNa 120: 45–66.

250 Joga Singh 2005

british kolambia (kaneDa) vich panjaabi di paRhaai (The teaching of Punjabi language in British Columbia). In Panjaabi Phaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 70–89. Ludhiana: Chetna Parkashan. Brar, Boota S. 2001 panjaabi prerNaarthak vaak: nav parpekh (Punjabi causative sentences: New perspective). Khoj Darpan 47: 97–112. 2003a japuji di chihn jugat (Semiotic structure of Japuji). Khoj Patrika 58: 46–52. 2003b panjaabi viaakraN: sathiti te sambhaavnaavaan (Panjabi graamar: Its state and the prospects). Alochana 200: 80–86. 2003c salok fariid: chihn parbandh (Salok Farid: semiotic organisation). Khoj Patrika 58: 201–204. 2003d rupaantri sirjNaatmak viaakraN (Transformational Generative Grammar). Alochana 200: 95–113. 2004a mohn singh di kaav bhaasha (The language of Mohan Singh’s poetry). Khoj Patrika 60: 189–198. 2004b Panjaabi Bhaasha: Sarot te Saruup (Form and origin of Punjabi language). Amritsar: Waris Shah Foundation. 2005a guru granth sahib da bhashaai saruup: punar chintan (Liguistic design of Guru Granth Sahib: A reappraisal). In Siri Guru Granth Sahib: Vibhin Pasaar. (Sri Guru Granth Sahib: Various dimensions), Ravinder Bhathal and Surjit Singh (ed.), 131–135. Ludhiana: Punjabi Sahit Academy. 2005b panjaabi alochana di bhaasha (The language of Punjabi criticism). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 103–115. Ludhiana: Chetna Parkashan. 2005c panjaabi bhaasha da ikk mileniam (A millennium of Punjabi language). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context) Darshan Gill (ed.), 97–102. Ludhiana: Chetna Parkashan. Brar, Jagjit 2001 pratiuttar: panjaabi bhaasha di Thuk (The majesty of Panjabi language: A rejoinder). SirjNa 122 (42): 70–76. 2006 shabad-shaastar: shabdaan di lishkor (Morphology: The shine of words). SirjNa 140: 54–62. Cheema, Baldev S. 2001 panjaabi viaakraN sahit da mulaankaN (Evaluation of literature on Punjabi grammar). Khoj Darpan 47: 41–50. 2002 shabad shreNiaan – vargikaran da adhaar (Word classes: Bases for classification). Bhaakha Sanjam 22: 81–88.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 251 2003a

panjaabi vaak parbandh: baNtar te kaaraj (Punjabi sentence system: Structure and function). Patiala: Punjabi University. 2003b puraatan janamsaakhi di viaakraN – kujh pakkh (The grammar of Puratan Janamsakhi – some aspects). In Paath DrishTti (Conceiving a text), Balwinder Kaur Brar (ed), 189–197. Patiala: Punjabi University. 2003c lipi ate gurmukhi de vikaas paRaavaan de sidhaantak adhaar ate ajoke ruup vich guru angad dev da yogdaan (Script and the theoretical foundations of the stages of development of Gurmukhi script and Guru Angad Dev’s contribution in the modern form). Nanak Parkash Patrika 59: 86–93. 2006 siri guru granth sahib: sankalan ate sampaadan jugtaan (Sri Guru Granth Sahib: Compilation and editing techniques). Alochana 211: 60–74. 2007 akaalaki upvaak: sankalp, baNtar ate vichran sthaan (Non-finite clauses: concept, structure, and distribution). Khoj Patrika 56: 251– 255. Dhaliwal, Prem Prakash S. 2002 Sidhaantak Bhaasha Vigiaan (Theoretical linguistics). Patiala: Madan Publications. 2004a Panjaabi Bhaasha da Janam te Vikaas (The origin and development of Panjabi language). Patiala: Madan Publications. 2004b Ruup Vigiaan ate Panjaabi Shabad Rachna (Morphology and Punjabi word formation). Patiala: Madan Publications. Dhandli, Daljit K. 2003 puaadhi viaakraN de kujh aihm paihlu (Some salient aspects of Puadhi syntax). Alochana 200: 114–117. Dhillon, Poonam, and Joga Singh 2007 paash ate paatar di paRnaaviin varton (The pronominal usage of Pash and Patar). Khoj Patrika 56: 245–250. Dhillon, Iqbal S. 2002 Chhand-Shaastar: Navin Dhunivigiaan Vidhi (The science of meter: new phonetic method). Delhi: National Book Shop. Dhingra, Parmjit Singh 2001 adhunakikarn te panjaabi (Punjabi and modernization). Khoj Darpan 47: 151–164. Dhot, Harnek Singh 2001 koshkaari: ik itihaasak sandarbh (Lexicography: A historical context). Punjabi Dunian 55: 192–200. Gill, Darshan (ed.) 2005a Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context). Ludhiana: Chetna Parkashan.

252 Joga Singh 2005b

panjaabi bhaasha di sathiti (State of Punjabi language). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed), 7–9. Ludhiana: Chetna Parkashan. Gill, Harjit Singh 2001 bhashaai sanchaar de panj uchaaraN ate sabhiachaarak chihnvigiaan (Five articulations of linguistic communication and cultural semiotics). Khoj Darpan 47: 1–8. Gill, Shivcharan S. 2005 bartaania vich panjaabi da drishTmaan khetar (The perceived space of Punjabi in Britain). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 90–96. Ludhiana: Chetna Parkashan. Gupta, Baldev Raj 2000 bhaasha vigiaan di drishTi vich siri guru angad dev ji di baNi (Sri Guru Angad Dev’s Bani from the linguistic perspective). Nanak Parkash Patrika 8: 114–121. Gupta, Dharmendar K. 2001 vaidik bhaasha te panjaabi (Punjabi and the vedic language). Punjabi Dunian 51: 57–70. Hasija, Madan Lal 2001a Hindi-Panjaabi Kosh (Hindi-Punjabi dictionary). Patiala: Bhaasha Vibhag, Punjab. 2001b Panjaabi Bhaasha da Kosh (Dictionary of Punjabi Language). Patiala: Bhasha Vibhag, Punjab. 2001c PramaaNik Panjaabi Kosh (Standard Punjabi dictionary). Patiala: Bhasha Vibhag, Punjab. Jassal, Kanwaljit 2002 Panjaabi Viaakarn de Kujh Pakkh (Some aspects of Punjabi grammar). Amritsar: Ravi Sahit Parkashan. 2003 kailaashpuri diaan kahaaniaan: samaaj bhaasha vigiaanak adhian (Stories of Kailash Puri: Sociolinguistic study). Khoj Darpan 25: 45– 51. 2004a panjaabi ishtihaaraan di bhaasha (Language of Punjabi posters). Khoj Darpan 27: 171–178. 2004b panjaabi bhaasha de sarot te baNtar: ik jaaiza (Sources and Structure of Punjabi Language: An overview. In DaakTar Prem Prakaash Singh Samriti Granth (Festschrift for Dr. Prem Prakash Singh), K. S. Thind (ed.), 190–192. Prince George: South Asian Review. 2005 bhaTTaan di baaNi: sanchaar jugtaan (Bani of the Bhatt’s: Communication strategies). Nanak Parkash Patrika 1: 103–109. 2006a kaurav sabha: paaThaatmak partaan (Textual layers of Kaurav Sabha). In Kaurav Sabha Diaan Paratan (Layers of Kaurav Sabha), H. S. Bhatia (ed.), 157–162. Ludhiana: Lahore Book Shop.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 253 2006b

Panjaabi Bhaasha Chintak (Punjabi language thinkers). Amritsar: Ravi Sahit Parkashan. 2006c siddh goshTi diaan bhashaai jugtaan (Linguistic strategies in Sidh Goshti). In Sidh GoshTti da Bahupakkhi Adhian (Multidimensional analysis of Sidh Gosht), Shashi Bala (ed.), 240–252. Amritsar: Guru Nanak Dev University. 2007a bhagat ravi daas di baaNi: bhaashagat adhian (Bani of Bhagat Ravi Das: Linguistic study). Samkali Sahit 48: 31–36. 2007b jarnail singh sekha rachit naval bhagauDa: sanchaar jugtaan (Jarnail Singh Sekha’s novel Bhagauda: Communication strategies). In Panjaabi Saiht da Sabhiaachaarak Parsang (The cultural context of Punjabi literature), Harcharan Singh Bedi (ed.), 111–117. Amritsar: Guru Nanak Dev University. Johl, Janmeja S. 2003 panjaabi bhaasha nu takniiki chuNautiaan te sambhaavi hall (The technological challanges facing Punjabi language and the possible solutions). Alochana 200: 70–79. Joshi, Jeet S. 2003 koshkaari da mahattav ate panjaabi koshkaari (Importance of lexicography and the Punjabi lexicography). Alochana 200: 118–124. 2004 Panjaabi Bhaasha ate Lokdhaara (Folklore and Punjabi language). Amritsar: Waris Shah Foundation. Joshi Shiv S. 2000 panjaabi bhaasha, sahit, chaNautiaan ate sambhaavnaavaan (Punjabi language, literature, the challenges and the prospects). Punjabi Duniaan 51 (10): 109–114. 2001a panjaabi baare bhaasha-vigiaanak adhian (Linguistic studies in Punjabi). Khoj Darpan 47: 21–40. 2001b panjaabi koshkaari da takniiki pakkh ton vishleshaN ate sujha (An analysis of technical aspects of Panjabi lexicography and the suggestions). Panjabi Dunian 51 (10): 22–29. 2001c panjaabi viaakarNaan vich prashanvaachak vaak (Interrogative sentences in Punjabi grammars). Khoj Darpan 47: 91–96. 2002 Sanskrit–Panjaabi Kosh (Sanskrit–Punjabi dictionary). Patiala: Punjabi University. 2003a globali parpekh vich panjaabi bhaasha da adhiaapan (Teaching of Punjabi language in the global context). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 144–150. Patiala: Punjabi University. 2003b muDhali sikhia da maadhiam (The medium of basic education). Alochana 200: 70–79.

254 Joga Singh 2003c

panjaab vich panjaabi bhaasha di sathiti (The state of Panjabi language in Punjab). Alochana 200: 22–30. Kadri, Seyyad M. Z. 2001 urdu te panjaabi (Punjabi and Urdu). Punjabi Dunian 51: 181–191. Kang, Gulzar 2001 adhunak bhaasha vigiaan da pitaama sosiuur (Saussure: the father of modern linguistics). Samdarshi 18 (6): 71–74. Kapoor, Navratan 2004 Panjaabi Koshkaari: Tulnaatmak VishleshaN (Punjabi lexicography: a comparative study). Patiala: Punjabi Publications. Kaur, Balwinder 2001 panjaabi bhaasha da vikaas (Development of Punjabi language). Punjabi Dunia 51: 41–51. Kaur, Bhupinder 2001 muDhle panjaabi viaakarNaan vich kiria (The verb in the first Punjabi grammars). Khoj Darpan 47: 81–90.

Kaur, Dhanwant (ed.) 2003

Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Phaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)). Patiala: Punjabi University.

Kaur, Jasbir 2001 Panjaabi Bhaasha da Vikaas (The development of Punjabi language). Jallandhar: Kuknas Parkashan. Kaur, Kanwaljit 2001 panjaabi shabad baNtar: sathiti te sambhaavnaavaan (Punjabi word structure: the state and the prospects). Khoj Darpan 47: 67–80. 2007 Panjaabi PaRnaav: ik SarvekhaN (Punjabi pronouns: A Survey). Amritsar: Ravi Sahit Parkashan. Kaur, Palwinder 2001a vigiaapan bhaasha: sarokar ate sanrachana (The language of advertising: concerns and structure). In bhaashaa vigiaan: sankalp te dishaawaan (Linguistics: concepts and dimensions), J. S. Puar (ed.), 202–212. Jallandhar: Punjabi Bhasha Academy. 2001b vigiaapan bhaasha: sarokar, saruup ate sanrachna (The language of advertising: concerns, nature and structure). Alochana 198: 57–69. 2003 Vigiaapankaari ate Panjaabi Bhaasha (Advertising and Punjabi language). Jallandhar: New Book Company. Kaur, Palwinder, and Sukhwinder S. Sangha 2001 gurmukhi lipi da adhiaapan (Teaching of Gurmukhi script). Khoj Darpan 47: 195–212. 2003 vigiaapan bhaasha (Language of advertising). Alochana 200: 24–45.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 255 Kaur, Ramanprit, and Parmjit S. Sidhu 2003a bhagat baaNi di simiotiks (The semiotics of Bhagat Bani). Nanak Parkash Patrika 2: 144–155. 2003b mahalla paihla diaan vaaraan: arthvigiaanak jugat (The ballads of the First Guru: The semantic structure). Khoj Patrika 58: 97–107. 2004a aasa di vaar da chihn vigiaanak adhian (Semiotic study of Asa di Var). In Aasa di Vaar: Bahupakkhi Adhian (Asa di Var: Multidimensional studies), Rai Jasbir Singh (ed.), 47–58. Amritsar: Guru Nanak Dev University. 2004b bhaTTaan de svaiie: paaTh-vigiaanak vishleshan (Text linguistic analysis of Bhats’ Svaiye). Nanak Parkash Patrika 1: 55–67. 2004c shabad guru di simiotiks (The semiotics of Shabad Guru). In Guru Granth Sahib de Vividh Parpekh. (The various dimensions of Guru Granth Sahib), J. S. Dhillon (ed.), 171–186. Amritsar: Guru Nanak Dev University. 2005a janeu sanskaar: chihn vigiaanak vishleshaN (Semiotic study of the ritual of Janeu). Khoj Darpan 56: 65–74. 2005b siri guru granth sahib vich bhashaai saanjh (Linguistic a‰nity in Sri Guru Granth Sahib). In Sundar Parteek – Siri Guru Granth Sahib (The beautiful symbol – Guru Granth Sahib), Bhai Kirpal Singh (ed.), 258–272. New Delhi: Gobind Sadan. 2005c siri guru granth sahib di bhashaai vilakkhaNta (The linguistic uniqueness of Sri Guru Granth Sahib’s language). In Siri Guru Granth Sahib: Vibhin Pasaar (Guru Granth Sahib: Various dimensions), Ravinder Bhathal (ed.), 116–25. Ludhiana: Punjabi Sahit Academy. Kaur, Ranjit 2007 Kaarji Panjaabi (Functional Punjabi). Amritsar: Ravi Sahit Parkashan. Kaur, Sukhjit 2001 Panjaabi bhaasha adhiin vichardiiaan akhanDi dhuniiaan (The nonsegmental sounds of Panjabi language). Panjabi Dunian 3: 52–58. Kaur, Updesh 2001 panjaabi shabdaan da dhuniaatmak pakkh (Phonetic aspect of Punjabi words). Khoj Darpan 47: 121–130. 2006a Panjaabi Bhaasha de Ruup (Forms of Punjabi language). Delhi: Bala Ji O¤set. 2006b Panjaabi Dhuni Jugat (Phonology of Punjabi language). Jallandhar: Deepak Publishers. Khaira, Bhupinder Singh 2002 panjaabi bhaasha-vigiaan de sarokaar te chuNautiaan (The concerns and challenges before Punjabi linguistics). Khoj Patrika 56: 237–244.

256 Joga Singh Kumari, Nirmal 2001 panjaabi koshkaari ikk adhian (Punjabi lexicography: a study). Panjabi Dunian 51 (10): 58–69. Lehl, Gurpreet S., and H. S. Bhatti 2003 panjaabi bhaasha da kampiuTrikaran (Computerisation of Panjabi language). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 129–138. Patiala: Punjabi University. Machwe, Prabhakar 2001 bhaarat vich dobhaashi shabadkosh: samassiaanvaan te sambhaavnaavaan (Bilingual dictionaries in India: problems and the prospects). Punjabi Dunian 51: 47–57. Madhopuri, Balbir Singh 2001 dilli vic panjaabi da saDa mishan puura ho gia (Our Punjabi mission in Delhi is fulfilled). PreetlaRi 68 (5): 32–36. Nirankari, Maan Singh 2001 panjaabi naal angrezi bhaasha di sikhia (The teaching of English along with Panjabi). PreetlaRi 68 (5): 39–42. Noor, Satinder S. 2005 panjaabi te gurmukhi da masla (The issue of Punjabi and Gurmukhi). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 16–19. Ludhiana: Chetna Parkashan. Puar, Joginder Singh (ed.) 2001a Bhaasha Vigiaan Sankalp te Dishaavaan (Linguistics: Concepts and dimensions). Jallandhar: Punjabi Bhasha Academy. Puar, Joginder Singh 2001b bhaasha vigiaan, viaakarn ate bhaasha adhian (Linguistics, grammar and language studies). Khoj Darpan 47: 9–20. Ryal, Gurcharan Singh 2001 koshkaari ik ati gambhir jimmevaari vaala kaaraj (Lexicogrphy: An act of immense responsibility). Punjabi Dunia 55: 201–208. 2002 moh shabad de pichhokaR vich (The background behind the shabad moh). Khoj Patrika 56: 218–220. Saksena, Babu Ram 2001 Bhaasha Vigiaan (Linguistics). Patiala: Bhasha Vibhag, Punjab. Sandhu, Gurpal S. 2001 panjaabi bhaasha da vikaas, samassiaanvaan te sarokaar (Development of Punjabi language: Problems and the issues). Alochana 195: 67–80.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 257 2005

ajoke bhaarat vich panjaabi bhaasha di sathiti (The state of Punjabi language in contemporary India). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 32–38. Ludhiana: Chetna Parkashan. Sangha, Sukhwinder S. 2001a vigiaapani bhaasha, sarokaar te sanrachna (Language of advertising: The concerns and the structure). In Bhaasha Vigaian: Sankalp te Dishaawaan (Linguistics: Concepts and dimensions), J. S. Puar (ed.), 202–212. Jallandhar: Punjabi Bhasha Academy. 2001b Panjaabi Bhaasha Vigiaan (Punjabi linguistics). Jallandhar: Punjabi Bhasha Academy. 2003a Sahit Bhaasha ate Bhaasha Vigiaan (Literature, language, and linguistics). Patiala: Punjabi University. 2003b surkhiiaan di bhaasha (The language of headlines). Punjabi Media 4: 29–30. 2004 mohn singh di kaav bhaasha (The language of Mohan Singh’s poetry). Khoj Patrika 60: 181–189. 2006 sahit ate bhaasha vigiaanak adhian maaDal (Literature and the linguistic models of study). In Sahit Adhian Vidhiaan: Vartmaan Parpekh (The study of literature: contemporary context), H. S. Bhatia (ed.), 213–224. Amritsar: Guru Nanak Dev University. 2007 vishavikaran ate panjaabi bhaasha (Globalisation and the Punjabi language). Trishanku 40: 76–79. Sangha, Sukhwinder S., and Palwinder Kaur 2001 gurmukhi lipi da adhiaapan (Teaching of Gurmukhi script). Khoj Darpan 47: 195–212. 2003 vigiaapan bhaasha (Language of advertising). Alochana 200: 57– 69. Sara, Amanpal 2000 panjaabi bhaasha de miaarikaran da masla (The issue of standardization of Punjabi language). Chirag 32: 37–43. Sekhon, Manmander S. 2007 Malwai Kosh (Malwai dictionary). Patiala: Punjabi University. Shan, Harnam S. 2001 guru granth sahib di koshkaari: ik sankhep adhian (Lexicography of Guru Granth Sahib: A brief study). Punjabi Dunian 51 (10): 145–156. Sharma, T. R. 2003 praaimri skuulaan vich angrezi kion nahin (Why English not in primary schools). Alochana 200: 14–21. Shastri, Rajinder S. 2001 panjaabi vich dhuni parivartan (Sound change in Panjabi). Punjabi Dunian 51 (10): 39–46.

258 Joga Singh Sidhu, Gurmel 2005 gaihn sankaTgrast panjaabi bhaasha (The Punjabi language in serious crisis). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 39–61. Ludhiana: Chetna Parkashan. Sidhu, Parmjit S. 2001a bhagat naamdev di baaNi di bhaasha de adhian diaan sambhaavnaavaan (The potentials of the study of the language of Bhgat Namdev’s Bani). Khoj Darpan 47: 77–94. 2001b panjaabi lokdhaara da bhaasha vigiaanak adhian (Linguistic study of Punjabi folklore). Khoj Darpan 48: 79–93. 2002a saadhu bining de sachch di simiotiks (The semiotics of Sadhu Bining’s Sachch (Truth)). Khoj Darpan 50: 1–16. 2002b tukhaari chant mahalla paihla, baaraamaaha di bhaasha: shabad ruup te shabad parbandh (The language of Tukhari Chhand Mahala Paihla, Baramah: Word forms and word structure). Khoj Darpan 40: 167–182. 2003a panjaabi koshkaari vich kampiuuTar di varton (Use of computer in Panjabi lexicography). Alochana 200: 125–134. 2003b globalikaran te panjaabi bhaasha di samaaj bhaasha vigiaanak sthiti (Globalisation and the sociolinguistic situation of the Punjabi language). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 83–92. Patiala: Punjabi University. 2006a paaThaatmak vidhi – sidhaantak adhaar (Textual method: theoretical foundations). In Sahit Adhian diaan Vidhiaan (Methodologies of the study of literature), H. S. Bhatia (ed.), 230–239. Amritsar: Guru Nanak Dev University. 2006b sidh goshT: arth-vigiaanak jugat (The semantic structure of Sidh Gosht). In Sidh GoshT: Bahupakkhi Adhian (Sidh Gosht: Multifaceted studies), Manjinder Singh (ed.), 252–262. Amritsar: Guru Nanak Dev University. Sidhu, Parmjit S., and Ramanprit Kaur 2003a bhagat baaNi di simiotiks (The semiotics of Bhagat Bani). Nanak Parkash Patrika 2: 144–155. 2003b mahalla paihla diaan vaaraan: arthvigiaanak jugat (The ballads of the First Guru: The semantic structure). Khoj Patrika 58: 97–107. 2004a aasa di vaar da chihn vigiaanak adhian (Semiotic study of Asa di Var). In Aasa di Vaar: Bahupakkhi Adhian (Asa di Var: Multifaceted studies), Rai Jasbir Singh (ed.), 47–58. Amritsar: Guru Nanak Dev University.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 259 2004b

bhaTTaan de svaiie: paaTh-vigiaanak vishleshan (Text linguistic analysis of Bhats’ Svaiye). Nanak Parkash Patrika 1: 55–67. 2004c shabad guru di simiotiks (The semiotics of Shabad Guru). In Guru Granth Sahib de Vividh Parpekh (The various dimensions of Guru Granth Sahib), J. S. Dhillon (ed.), 171–186. Amritsar: Guru Nanak Dev University. 2005a janeu sanskaar: chihn vigiaanak vishleshaN (Semiotic study of the ritual of Janeu). Khoj Darpan 56: 65–74. 2005b siri guru granth sahib vich bhashaai saanjh (Linguistic a‰nity in Sri Guru Granth Sahib). In Sundar Parteek – Siri Guru Granth Sahib (The beautiful symbol – Guru Granth Sahib), Bhai Kirpal Singh (ed.), 258–272. New Delhi: Gobind Sadan. 2005c siri guru granth sahib di bhashaai vilakkhaNta (The linguistic uniqueness of Sri Guru Granth Sahib’s language). In Siri Guru Granth Sahib: Vibhin Pasaar (Guru Granth Sahib: Various dimensions), Ravinder Bhathal (ed.), 116–25. Ludhiana: Punjabi Sahit Academy. Singh, Agyajit 2003 maat bhaasha de manovigiaanak sidhaant (Psychological theories of mother tongue). Panjabi Dunian 53 (4): 25–30. Singh, Atamjit 2000 panjaabi koshkaari (Punjabi lexicography). Aks 20 (2): 3–6. Singh, Bakhshish 2002 Panjaabi Angrezi Muhaavara Kosh (Panjabi-English dictionary of idioms). Amritsar: Guru Nanak Dev University. 2006 bhaashaa de vikaas lai koshkaari di mahatata (The importance of lexicography for the development of language. SirjNa 126: 41–49.

Singh, Davinder 2002

upbhashaavaan ate koshkaari (Lexicography and the dialects). Khoj Patrika 56: 229–230. Singh, Dharam 2001 pakistaan vich panjaabi bhaasha te sahit di sathiti (State of Punjabi language and literature in Pakistan). Punjabi Dunian 51 (10): 6–11. Singh, Gulwant 2001 panjaabi te phaarsi (Persian and Punjabi). Punjabi Dunian 51 (10): 157–180. Singh, Gurmukh 2001 gurmukhi lipi (The Gurmukhi script). Panjabi Dunian 51 (10): 28– 40. Singh, Gurbax, and Joga Singh 2004 panjaabi bhaasha di raajniti (The politics of Punjabi language). Sarokar 21: 53–57.

260 Joga Singh Singh, Gurnam 2000 panjaabi kinni sanskrit bolde hann (How much Sanskrit the Punjabis speak). PreetlaRi 67 (2): 56–60. Singh, Harkirat 2000 maajhi da sthaan (The place of Majhi). Punjabi Dunian 51 (10): 89– 101. 2001 panjaabi koshkaari da vikaas te sambhaavnaavaan (The development and the prospects of Panjabi lexicography). Panjabi Dunian 51 (10): 22–29. 2003a bhaasha te viaakraN (Language and grammar). Panjabi Dunian 53 (1): 21–24. 2003b panjaabi di uppbhaasha multaani (The Multani dialect of Panjabi language). Alochana 200: 43–56. 2004 SaaDi Bhaasha (Our language). Patiala: Punjabi University. Singh, Joga 2002 rupaantri viaakraN (Transformational Grammar). In Sidhaantak Bhaasha Vigiaan (Theoretical linguistics), Prem Prakash S. Dhaliwal (ed.), 121–139. Patiala: Madan Publications. 2003a bhashaayi sidhaant ate globali parsang vich bhashaai niti (The linguistic theory and the language policy in the global context). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 93–119. Patiala: Punjabi University. 2003b maat bhaasha da mahattav (The importance of mother tongue), modified version of Joga Singh (2001). Samdarshi 73: 48–60. 2003c panjaabi da padkram (Word order in Punjabi). Alochana 200: 87–94. 2004 Daaktar prem parkaash singh ate ohnaan di rachnaan sidhaantak bhaasha vigiaan (Dr. Prem Prakash Singh and his work ‘theoretical linguistics’). In Daaktar Prem Parkaash Samriti Granth (Festschrift for Dr. Prem Prakash Singh), K. S. Thind (ed.), 180–184. Prince George: South Asian Review. Singh, Joga, and Gurbax Singh 2004 panjaabi bhaasha di raajniti (The politics of Punjabi language). Sarokar 21: 53–57. Singh, Joga, and Poonam Dhillon 2007 paash ate paatar di paRnaaviin varton (The pronominal usage of Pash and Patar). Khoj Patrika 56: 245–250. Singh, Kulwinder 2001 paakistaani panjaab vich panjaabi bhaasha di sathiti (State of Punjabi language in Pakistan Punjab). Khoj Darpan 47: 165–176. Singh, Major Gurmukh 2002 Punjabi–English Dictionary. Patiala: Punjabi University.

Linguistic Publications in Punjabi (2000–2008): A Brief Survey 261 Singh, Onkar 2001a panjaabi bhaasha da kampiuTrikarn (Computerisation of Punjabi language). Khoj Darpan 47: 131–136. 2001b panjaabi bhaasha te panjaabi koshkaari (Panjabi language and Panjabi lexicography). Panjabi Dunian 51 (10): 5–12. 2003 panjaabi da vikaas (Development of Punjabi). Punjabi Dunian 53: 56–61. Singh, Parmjit 2005 Bhaashavigiaan: Bhashaai Padhar te Sanchaar MaaDal (Linguistics: Communication models at language level). Jallandhar: Sundar Book Depot. Singh, Prem Prakash 2002 Sanskrit-Panjaabi Kosh (Sanskrit-Punjabi dictionary). Patiala: Punjabi University. 2005 guruubaaNi da bhaasha saruup (The form of the language of Gurbani). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 10–15. Ludhiana: Chetna Parkashan. Singh, Pritam 2003 maa boli panjaabi de ajj baare (The present of mother tongue Panjabi). PreetlaRi 70 (4): 41–43. 2006 asTrelia dian aad-vaasi bhashaavaan de panjaabi hindi naal sanjhe shabad (The common words between Australian aboriginal languages and Punjabi and Hindi). PreetlaRi 73 (5): 24–26. Singh, Sewak 2002 gurbaaNi viaakraN: kujh nukte (Few points about Gurbani grammar). Nanak Parkash Patrika 2: 95–106. 2004 gurbaaNi arthaan de sambandh vich (About meanings in Gurbani). Nanak Parkash Patrika 2: 42–46. Singh, Surjeet 2003 vigiaan sikkhia ate panjaabi bhaasha (Punjabi language and the teaching of science). Alochana 200: 31–42. Sital, Jit S. 2001 kosh da lacchaN: puraaNe koshaan di loR te baNtar (Nature of dictionary: The need and form of old dictionaries). Punjabi Dunian 51: 95–144. Suman, Preet 2005 panjabi angrezi kosh (Punjabi – English dictionary), review of AngloPunjabi Dictionary by K. S. Kang. Khoj Patrika 61: 604–612. Taangh, Ishar Singh 2000 panjaabi bhaasha di niiti (The policy of Punjabi language). Punjabi Dunian 50 (8–9): 105–108.

262 Joga Singh Tanvi, Arvinder K. 2005 Do-Bhaashi Koshaan da Kosh-Vigiaanak Adhian (Lexicographic study of bilingual dictionaries). Chandigarh: Lokgeet Parkashan. Thind, Karnail S. 2003 paakistaan vich panjaabi bhaasha di sthiti te sambhaavnaavan (The state and prospects of Panjabi language in Pakistan). In Proceedings of the Nineteenth International Conference on Punjabi Language Development (Panjaabi Bhaasha, Sahit te Sabhiaachaar: Global Parpekh (Punjabi language, literature and culture: Global context)), Dhanwant Kaur (ed.), 161–175. Patiala: Punjabi University. Thind, Karnail S. (ed.) 2004 Daaktar Prem Parkaash Samriti Granth (Festschrift for Dr. Prem Prakash Singh). Prince George: South Asian Review. Zaman, Fakhar 2005 Pkistaan vich panjaabi zubaan (The Punjabi language in Pakistan). In Panjaabi Bhaasha: Kaumaantri Parpekh (The Punjabi language: Global context), Darshan Gill (ed.), 62–69. Ludhiana: Chetna Parkashan.

Reviews

Gregory D. S. Anderson The Munda Verb: Typological Perspectives Mouton de Gruyter: Berlin. 2007. xvi, 306 pp. US $ 145. ISBN 978-3-11-018965-0

Reviewed by Ayesha Kidwai

This is an engaging and scholarly book that will serve as an important resource for linguists working in the field of Munda and Austro-Asiatic synchronic and diachronic linguistics for many years to come. Certainly the most wide-ranging study of any aspect of the Munda languages in the last five decades, it is also the most ‘modern’ in its tenor and orientation – not only does it systematically attempt to integrate its descriptive findings with similar phenomena in the world’s languages, the tone of such description is consistently scientific, such explanation as it o¤ers is always structural, and without recourse to the anthropological specificities of its speakers. Although the book’s title suggests that it is an exercise in qualitative typology – i.e. the investigation of Munda verb morphosyntax in terms of a framework of cross-linguistically viable types – the inquiry it contains is guided by another question as well: How is the bewildering variation between North (Korku, Kherwarian) and South Munda (Juang, Kharia, Gutob, Remo, Gtaffl, Sora, Gorum) and in particular, the Kherwarian (Santali, Mundari, Bhumij) languages, to be reconciled with respect to the reconstruction of a common Proto-Munda ancestor? This question is a constant theme in the eight chapters of the book; indeed, it becomes increasingly the dominant question. This preoccupation, however – as I shall argue in the discussion of individual chapters below – yields mixed results. While the quest certainly acts as a unifying factor for the discussion of what are ultimately quite disparate areas of Munda grammar, it also has the e¤ect of making the task of synchronic description somewhat secondary to the goal of linguistic reconstruction. Chapter 1 provides an overview of the Munda language family, its classification, the sociolinguistic profile of its speakers, and a (very welcome) description of the history of linguistic inquiry into the language family. The chapter concludes with a concise survey of its phonology and nominal

266 Reviews morphology; the discussion of syntax being quite brief, limited to clausal constituent order and subordination as it is. In fact, there is quite generally a morphology-based focus to the book, as even though most of the phenomena investigated in the book are syntactic in nature – from voice and noun incorporation to negation, referent indexing, and switch reference – Anderson’s description of them remains grounded in the morphological exponence associated with these syntactic processes. This prevents him from making the kinds of connections between these processes that a syntactic approach would have a¤orded, with the result that each chapter largely stands on its own, and no systematic syntactic typology of the Munda languages can emerge. In this review of Anderson’s work, I would like to focus on the possible advantages that a more synchronic perspective to the morphosyntax of the Munda languages will bring not only to a description of each language, but also to the reconstruction problem itself. In the remarks that follow, I will concentrate on the chapters on voice, referent indexing, noun incorporation and tense, aspect, mood morphology, and have nothing to say about the chapters on negation and auxiliary verb forms. Chapter 2 discusses argument-structure changing morphology in Munda languages, and demonstrates it to be significantly di¤erent in North vs. South Munda. With regards to the morphological causative, whereas South Munda languages uses variants of the original Proto-Munda causative
0 b as a prefix or infix, North Munda employs historically secondary su‰xes. Reciprocal marking is realized in North Munda only as an infix, but in South Munda as obligatorily prefixal. However, both groups mark applicatives with su‰xal morphology, and are also similar in allowing a single morpheme to have a passive or reflexive function. The chapter concludes with a discussion of ‘‘version’’ – a grammaticalised discourse notion encoding ‘‘salience’’, ‘‘a¤ectedness’’, and the like – in the South Munda language Gorum. Anderson suggests that although North Munda languages have the means of indicating some of the functional contrasts encoded by other morphology, nothing comparable to a distinct version morphology is to be found in North Munda. The facts that Anderson presents are very interesting, but, in terms of description, a number of questions arise: How do these various a‰xes distribute vis-a`-vis each other in a single verbal complex? Is it possible to derive a passive reciprocal causative in both South and North Munda? Furthermore, does the variation in the prefixal/infixal nature of North and

G. D. S. Anderson. The Munda Verb. Reviewed by Ayesha Kidwai

267

South Munda causative and reciprocal marking have distinct consequences in terms of case-marking of arguments? If not, then the reconstruction problem is really one of morphological (re-)analysis (in a syntactic environment); but if yes, then the problem relates to syntax, suggesting that morphological change triggers a di¤erent syntax altogether. Chapter 3 discusses referent-indexing morphology on the Munda verb. Once again, the North vs. South divide is on show, with greater variation besides. Taking subject marking first, the main di¤erence between the Kherwarian (Korku having lost subject clitics altogether) and South Munda is that, in the former, subject marking may be enclitic on the penultimate phrase, but in the latter, it is indexed through a series of enclitics/proclitics closely associated with the verb/verbal complex. A closer examination, such as the one that Anderson undertakes, shows that aspects of the North Munda distribution is instantiated in South Munda as well, and vice-versa. For example, in Kharia (p. 66), while enclisis on the verb is expected in all contexts, we find that (preverbal) negation attracts the pronominal, yielding the Kherwarian pattern. Similar facts hold for Gutob (p. 72). At the same time, instantiations of the South Munda type of enclisis to the verbal complex abound in Kherwarian as well, particularly in intransitive contexts with pronominal arguments (p. 77). In the case of object-marking, the North-South divide disappears. Both groups of languages integrate object markers (if they have them) as enclitics into the verbal complex, peripheral to tense/aspect markers, but preceding moods and the finiteness marker. This leads Anderson, in his focus on the reconstruction problem, to suggest that a set of ‘‘su‰xal’’ object markers must be reconstructed for Proto-Munda. For subject markers, he speculates that ‘‘it seems likely that the system of subject pro-clitics (or unbound prosodically weak resumptive pronominal elements) reconstructed for South Munda was also characteristic of Proto-Munda as well, preserved as such in Proto-South Munda and its daughter languages . . . , but reinterpreted as enclitic markers on the word preceding the verb in Proto-North-Munda, a system preserved in Proto-Kherwarian and its modern attested daughter languages . . . (p. 82).’’ Once again, the almost exclusive focus on forbears creates descriptive problems. The simultaneous existence of two patterns of enclisis – to the penultimate phrase or to the verbal complex – in a language (e.g., Santali [North Munda] and Kharia [South Munda]), suggests that the problem is as much synchronic as diachronic. In order to solve this descriptive problem

268 Reviews (at least as it appears to me), it is necessary to make a conceptual move from morphology to syntax, as signaled by a disambiguation of the terms ‘‘subject/object marker/agreement’’, vs. ‘‘subject/object clitic’’. This is a move that Anderson does not make (more appropriately, make enough – as the quote above indicates that this would be an analysis he would favor), as the distribution of clitics is necessarily governed by the syntax, rather than morphology. In my own work, I have explored this option for Santali (‘Santali Backernagel Clitics’, in The Yearbook of South Asian Languages, edited by Rajendra Singh and Tanmoy Bhattacharya. 2006. Mouton: The Hague). In general terms, I have sought to understand the two main patterns of Santali subject enclisis – penultimate phrase or verbal complex – as resulting from a complex of three main syntactic factors: (1) obligatory clitic doubling for non-pronominal animate subject arguments; (2) the need for clitics to find a host, and (3) the imperative for pronominal arguments or clitic doubles to incorporate into their selecting predicates. In the analysis I develop there, clitics are number/person heads in noun phrases in Santali, that may occur either on their own (i.e. as a pronominal), or associated with referentially independent nouns. I then show that when a noun phrase is a (subject or object) pronominal, it directly incorporates into the verbal complex; however, if it is a person/number marker associated with a referentially independent noun, (i.e. a clitic double), it is ‘too far’ from the verb and must look for an alternate host, which (mostly) turns out to be the penultimate phrase of the sentence. Anderson’s work now suggests that these imperatives hold of the entire family, so the reconstruction problem could really be a question about syntactic change – while in South Munda, verbs by and large are ‘close’ enough to subject clitics in their base positions so as to serve as their hosts (barring the case when negation is ‘closer’), language change somehow imposed a greater distance between verbs and subject clitics in North Munda (barring contexts of pronominal incorporation and bare intransitives). In such a scenario, either of two scenarios could result: (a) the dropping of clitic doubling and subject enclisis altogether, as in Korku, or (b) a relaxation of the word requirement on hosts to a phrase requirement, as in Kherwarian. One prediction of the approach that the above approach would make is that the inability of Kherwarian verbs to be close enough to subject clitics should be signaled somehow, perhaps by verb morphology. Anderson’s outstanding discussion of tense, aspect, and mood on Munda verbs in Chapter

G. D. S. Anderson. The Munda Verb. Reviewed by Ayesha Kidwai

269

4 indicates a North-South divide here as well. In very simplified terms, the divide essentially reduces to whether an original distinction in tense/aspect morphology in terms of transitivity is maintained (South Munda), or not (North Munda). Here the South Munda languages di¤er amongst themselves as well – while modern Kharia, Juang, Gutob, and Remo preserve a distinction between transitive and intransitive past attributable to ProtoMunda, modern Gtaffl, Sora, and Gorum have neutralized the distinction between transitive and intransitive past altogether. All the South Munda languages encode aspectual categories outside this basic tense distinction marking. In the Kherwarian system, on the other hand, transitivity is not the sole consideration, as this along with tense and aspect information fuse to yield two series of morphology – Series A morphology signals non-past/ imperfective/ indefinite temporal indexation fused with a low degree of transitivity, while Series B morphology is associated with a high degree of transitivity and past/ perfective/aorist meanings. As Anderson demonstrates, this conflation results in several di¤erent means for the realization of a single tense/aspect category. Although the data/analysis is too complex to reproduce, the chief e¤ect of this is a coalescence of tense and aspect (and transitivity) in these languages, whereas in South Munda, they remain largely separate. For example, in Santali, progressive is realized as –kan in the present with transitives, as zero with intransitives/passives and reflexives, and in the past as kan tah kana (both transitives and intransitives). It appears to me that one way of looking at the divide may be in terms of tense (South Munda) vs. aspect (North Munda), rather than tense/ aspect, as Anderson attempts. In fact, Anderson does not discuss this option at all, even though ever since the Hopi scandal, it is known that aspect can do quite a good job of e¤ecting temporal reference. In support of this claim, one could garner another distinction between North Munda and South Munda that Anderson does not dwell upon – in the former, but not in the latter, sentences must be marked with a ‘‘finitizer’’ final vowel (-a in Kherwarian, and –ba in Korku). This requirement for the overt expression of finiteness suggests that there may be no Tense morphology on North Munda verbs at all. If these speculations are on the right track, we might have the beginnings of a real structural account of the North-South divide with respect to pronominal argument enclisis, incorporation, doubling in terms of verb raising: in Tensed South Munda, verbs are attracted to Tense, with such raising

270 Reviews bringing them close(r) to the external argument. Aspectual North Munda, however, lacks the imperative for such verb raising, and hence the promiscuity of subject clitics in Kherwarian and their loss in Korku. Conceived of thus, the reconstruction problem becomes both a question of morphosyntactic change – the loss of verb raising conditioned by the loss of Tense morphology. Finally, Chapter 6, on noun incorporation, also relates directly to the questions addressed here in the review. Once again, North Munda di¤ers from South Munda, in seeming ‘‘generally to have lost the ancestral noun incorporation construction of the type found in South Munda’’ (p. 201). The most elaborate of the noun incorporation paradigm is to be found in Sora, with other South Munda allowing it in more restricted contexts. Anderson begins his description of Sora noun incorporation by noting that the free forms of nouns in Sora are derived forms, as the original roots (‘combining forms’, in his terms) are monosyllabic. The free forms of these words invariably are the product of a‰xation (chiefly prefixation and infixation), or reduplication or compounding. It is also these forms that surface in noun incorporation constructions. He then goes on to make two claims about Sora noun incorporation that, if correct, would require a radical reappreciation of our understanding of the phenomenon: (a) noun incorporation does not block pronominal object marking in Sora, and (b) agent nouns may also undergo noun incorporation. As support for his claims, Anderson notes that not all cases of noun incorporation are accompanied by the appearance of a detransitivisation marker, –n. While object incorporation usually triggers such marking, contexts in which an incorporated object is doubled by a pronominal argument cannot surface with this marker. Similarly, agent incorporation is also not accompanied by detransitivisation – in fact it can be witnessed in imperatives as well – suggesting that incorporation in Sora is not a valencyreducing operation at all. In my opinion, however, the Sora noun incorporation facts suggest that Sora is a polysynthetic language, in terms of Mark Baker’s formulation (Mark Baker 1996. The Polysynthesis Parameter. New York: Oxford University Press). Baker suggests that languages may be typologised along a polysynthesis parameter, which states that ‘‘every argument of a head element must be related to a morpheme in the word containing that head (an agreement morpheme, or an incorporated root). Polysynthetic languages would be those that were positively specified for this parameter.

G. D. S. Anderson. The Munda Verb. Reviewed by Ayesha Kidwai

271

The argument for Sora as a polysynthetic language could go something like this, provided we make the (crucial) assumption that pronominal clitic doubling and pronominal incorporation is a pan-Munda property. A close inspection of the Sora data (pp. 187–88) reveals that the cases of apparent pronominal doubling of an incorporated object involve ditransitive predicates like give and bring. In these cases, the pronominal does not double the incorporated bare noun bearing the semantic role of theme, but rather is the incorporated pronominal of the goal argument. Thus, what we have here is two kinds of incorporation – pronominal incorporation and noun incorporation, and the absence of the –n marker could then be traced to this fact. In other words, –n fills the slot of a clitic double, when an argument has been absorbed by incorporation or, as Anderson notes (p. 187), by reflexivisation. In the case of ditransitives that instantiate clitic doubling as well as noun incorporation, that slot is ‘taken’, as it were, by the clitic double, and hence no –n is possible. –n then does not mark detransitivisation per se, but the unavailability of a pronominal clitic to fill a slot. Turning now to agent incorporation, note first that the chief evidence that this is an instance of noun incorporation comes from the fact that as a free form a nominal subject must be marked with an absolutive/nominal su‰x – interestingly, again an –n. By the logic of the argument I have just developed, the absence of –n is then a signifier that there is no slot for the pronominal clitic in the verb; however, as the verb is transitive, it can only be the agent argument that is taking its place; in other words, the agent argument must be incorporated. This analysis now reframes the typological and historical questions that we must answer. It seems to me that the divide between the languages of the Munda family does not play out along expected North-South lines, given that many of the South Munda languages, just like North Munda, do not instantiate any productive rule of noun incorporation at all. It appears that all the languages of the family can be seen as arranged along a cline beginning with noun incorporation and approaching polysynthesis. Gutob, Remo, Kharia, Juang, Santali, Bhumij and Korku do not appear (at least from Anderson’s description) to have the distinction between free and combining forms, and allow noun incorporation in very limited (mostly lexical) contexts, and must therefore be placed very low on the cline. Gorum, however, must be placed higher than these languages, as it allows object incorporation and goal argument doubling of the sort we witnessed in Sora; however, it does not seem to make a synchronic distinction between free

272 Reviews and combining forms. Gtaffl is perhaps the closest to Sora, both in terms of productive object noun incorporation (with stranded modifiers), and a requirement that all free forms be bimoraic/bisyllabic; however, Anderson does not report it to have agent noun incorporation. This sub-group insensitive typology now poses uncomfortable historical questions, either way one looks at the question. If polysynthesis was the original system and di¤erentiation of free and combining forms was the original system (as Anderson suggests), how do we explain the fact that a majority of Munda languages have lost this system? On the other hand, if the Sora polysynthesis is a language-specific ‘innovation’, what are the possible factors that ultimately created the conditions for such change? Furthermore, what could be the reasons why such a direction of change was not available for Gorum, Sora’s closest relative, with which it shares many other properties not found in the rest of South Munda? Finally, if the proposals that I make in this review for the syntax of Munda languages, what are the theoretical reasons why a syntax of pronominal incorporation does not automatically entail a syntax of, at the very least, nominal incorporation, and at most, polysynthesis? Eloise Jelinek’s ‘‘pronominal argument hypothesis’’ – wherein she argued that in pronominal argument languages, all nominal arguments are necessarily adjuncts – may well be one possible avenue for explanation if a ban on adjunct incorporation is tenable. However, this would nevertheless leaves the reconstruction problem una¤ected – what change enabled pronominal argument languages like Sora and Gorum to escape the ban on adjunct incorporation? Whatever the answers to these questions may turn out to be, the direction that such explanation must take into account as much of the syntax of individual Munda languages, as it does their morphology. Although one can only hope that such research is in the o‰ng, one thing is certain: at the foundation of all such future work will lie Anderson’s seminal contribution to Munda and Austro-Asiatic linguistics.

Josef Bayer, Tanmoy Bhattacharya, and M. T. Hany Babu (eds.) – Linguistic Theory and South Asian Languages: Essays in Honour of K. A. Jayaseelan Linguistik Aktuell/Linguistics Today 102, Amsterdam/Philadelphia: John Benjamins, 2007. 282 pages, Hardbound. Price: Not given. ISBN 978 90 272 33 66

Reviewed by Anjum P. Saleemi

This is a volume in honour of the Indian linguist K. A. Jayaseelan, which, in the editors’ words, ‘‘is more than a Festschrift’’ (p. 1). It contains articles on (South Asian) syntax, written by people who have been associated with Jayaseelan in di¤erent capacities over the decades. Most, though not all the articles, employ data from South Asian languages. Those that focus on other languages include the articles by Boeckx, Kayne, Gue´ron, and Reuland. As somewhat indirectly acknowledged in the Introduction by the editors (p. 5), the paper by Kayne is a reprint (possibly from Kayne 2005), and thus the Introduction does not provide complete information on the volume in question. The areas of linguistics and linguistic theory which are heavily covered in the present volume are syntax and syntactic theory, though a whole section, namely the very last one, is devoted to phonology. Predictably, a number of aspects of syntax, and a limited range of phonological themes, come under the scrutiny of the contributors. This is evident from the independent sections on Clause Structure, Modification in DP, Binding, and Complementizers and Complementation, each of which contains individual contributions related to the broad topic the section in which it appears is devoted to. The section on Clause Structure contains four articles, the one on Modification in DP only one paper, and the sections on Binding, Complementizers and Complementation, and Phonology contain three papers each. The general viewpoint of the volume is slanted towards theoretical issues rather than merely descriptive ones, reflecting the nature of J. K. Jayaseelan’s own range of interests, a bibliography of whose works appears towards the end of the volume. In addition, as is expected from any well-

274 Reviews edited collection of this kind, indices of names, languages, and topics are part of its concluding material. The first syntax paper (by Beermann, Sahoo, and Hellan) is about the phenomenon of argument sharing, with almost exclusive attention to VPserialization in the language Oriya. A good deal of the discussion centres around the patterns of object sharing and the corresponding syntactic configurations, and a comparison of the mechanisms of token sharing and function sharing, maintaining a relatively neutral stance towards the predominantly configurational theories of the Government-and-Binding type, and more function-oriented theories represented by Lexical Functional Grammar. Boeckx, on the other hand, is an attempt to investigate specificational pseudoclefts within a purely configurational framework favouring strictly-leftward derivational anlyses. A crucial component of the account o¤ered is the split-CP approach popularized by Rizzi (1997) and Cinque (1999). Non-copular pseudoclefts, such as (1), are mentioned but not discussed because they invariably have a predicational reading, and are thus not specificational. It is duly noted, though, that some copular pseudoclefts also have only a predicational reading, much as some others can be understood only as being specificational; so (2a) can only mean something like (2b), and cannot have the interpretation shown in (3c); likewise, (3a) cannot be understood as meaning (3b), and only has the interpretation (3c). (1)

What John didn’t buy was any pictures of Fred.

(2)

a.

What John is is worthwhile.

b.

‘John is x & being x is worthwhile.’

c. *‘John is worthwhile.’ (3)

a.

What John is is proud.

b. *‘John is x & being x is proud.’ c.

‘John is proud.’

Following Boeckx’ paper is another one by Madhavan on cleft constructions in Malyalam, the analysis proposed being based upon the split-CP framework already mentioned, in this case also incorporating Jayaseelan’s (2000) take on Focus Phrase. The final paper in this part of the book is by Srikumar, also dealing with Malyalam data, but demonstrating the role of

J. Bayer et al. Linguistic Theory. Reviewed by Anjum P. Saleemi

275

clausal pied-piping in providing an escape hatch in long-distance questionword dependencies, thus bypassing subjacency constraints. The proposal outlined appeals to the Minimal Link Condition and is therefore formulated within the minimalist viewpoint. The section on Modification in DP consists only of Kayne’s paper on the syntax of quantity in English, in particular considering the distribution of adjectival modifiers of nouns implicitly expressing NUMBER and QUANTITY. Some example of such modifiers are: many, few, much, little, large, small, a few, quite a few, and numerous, which, it is proposed, is subject to the Principle of Decompositionality, supposedly a general constraint of U(niversal) G(rammar) requiring that no more than one interpretable syntactic feature be associated with a single lexical item. Evidence from polarity and scope is a part of the argumentation in this heavily data-oriented account, which, as is usual in Kayne’s work, is far from reticent on issues of theoretical import. The section on Binding begins by discussing co-reference violations of Principle B in a paper by Gue´ron, initially inspired by Jayaseelan (1992); English as well as Malyalam data are exploited in the development of Gue´ron’s own proposal, which relies on the minimalist mechanism of the checking of formal features, and also considers Jayaseelan’s idea of using LF movement as a reconstruction device at least insofar as ‘inverse copular sentences’ (see the contrast between 4a and 4b below) are concerned. (4)

a. John is my best friend. b. My best friend is John.

Reuland’s Perspectives on binding, the next paper, builds on Jayaseelan’s (1997) work, too, and grapples with the binding issues by resurrecting some of their fundamental problems. The approach to binding adopted is by and large the one developed by Reinhart and Reuland himself (e.g., Reinhart and Reuland 1993). The concluding paper in this section (by Yadava) is on subject-to-subject raising from within a tensed clause, a phenomenon which is attested in Maithili but is obviously not found in the English-type languages. A possible explanation for this special type of ‘raising’ is o¤ered that is largely couched in terms of the Government-Binding theory, and which strives for a reformulation of the notion of a governing category. Among the papers that appear in the section on Complementizers and Complementation, Dasgupta is a study of the clause-internal complementizers in Bangla. The distribution of these particle-like items in finite

276 Reviews complement and adjunct clauses, it is suggested, can be accounted for by assuming that they are base-generated in situ, and that their features undergo covert movement at LF. Davison’s paper, on the other hand, is a cross-linguistic investigation of parameters bearing on the position of C in a number of South Asian languages, such as Hindi, Punjabi, Kashmiri, Sinhala, Bangla, Gujrati, Oriya, Nepali, and Marathi. These otherwise head-final Indic languages exhibit a complex and unexpected pattern of variability in regard to the placement of the complementizer elements, which appear head-finally only in Sinhala, but are head-initial in Hindi, Punjabi and Kashmiri, and both head-initial and head-final in many others in the class of languages under discussion, e.g., Gujrati, Marathi, Sindhi, Bangla, Assamese, and Oriya. Like some of the other contributors to the volume, Davison describes this variation in the projection of C in terms of Rizzi’s (1997) well-known proposal about the extended structure of CP. Faced with the range of variety of this sort, it is not surprising that Davison remains ‘‘agnostic’’ (p. 191) insofar as the plausibility of the universal Specifier-Head-Complement order posited by Kayne’s (1994) is concerned. Davison’s interesting wide-ranging observations and insights are followed by Barbora’s more narrowly focused examination of the status of the particle ne, a disjunctive marker with inherent negative characteristics, in the Yes-No interrogatives in Assamese. As in many other accounts relating to the CP projection, Rizzi’s (1997) basic ideas are adopted as the backdrop of the analysis sketched out here, which, in addition, makes a crucial appeal to the category often called Disjunctive Phrase. A few general remarks are in order at this point about the syntax papers briefly reviewed so far, and about the approach taken therein. These papers evince the usual mix of theoretical insights and empirical observations, and the readers are exposed to the richness and diversity of both. With the exception of Reuland and Kayne, though, the semantic factors involved in, or underlying, the various syntactic constructs are not given the attention that they deserve, which is not to say that syntax cannot be done without semantic considerations but just to emphasize the fact that several syntactic phenomena are more sensitive to semantic and logical considerations than is typically acknowledged, with the result that some extremely common linguistic patterns tend to recede into the background of the syntactic theory (or simply of the notation thereof ) prevalent at any give point in time within a particular community of researchers. Take, for instance, Barbora’s paper: Yes-No questions, as their very name indicates, are disjunctive in

J. Bayer et al. Linguistic Theory. Reviewed by Anjum P. Saleemi

277

nature, with the disjunction being between a positive proposition (P) and a negative one (sP). So their general logical form could be assumed to be something like: (5)

Q(P4sP).

Described in these terms, it might appear a lot easier to notice how widely attested this pattern is, as the following examples from English (9), HindiUrdu (10) – and countless comparable A-not-A type constructions from languages like Mandarin Chinese will likewise – clearly demonstrate. (6)

Is this your book (or not)?

(7)

(tum) aao gee ya nahı˜? (you) come-2SF/M will-2SM or not ‘Will you come or not?’

The issue at stake is essentially a simple one: wherever appropriate, any syntactic analysis accompanied by some supportive semantic generalization will be reinforced by it, and should as a matter of principle be preferred over one that lacks such support. We believe that this is a methodological advantage that should be self-evident, and therefore must not be undervalued. After all, autonomous though (major) parts of it may be, what good is syntax if it cannot be shown to interface with meaning in a robust manner? Let us now turn to the phonology papers, namely: Das’s paper on Malyalam phonology, Vijayakrishnan’s study of syllable structure, and Sailaja’s article on the relationship between writing systems and phonological awareness. Das attempts to specify the ways in which Malyalam manages to avoid the cross-linguistically rare nasal-plus-voiceless-obstruent cluster. It is shown that the two strategies adopted for the purpose in the language in question are nasal gemination and post-nasal voicing, of which the latter is the unmarked one. Further, Das outlines an explanation for this order of preference by exploiting constraint interaction in optimality theory and the theory of underspecification. Vijaykrsihnan’s paper considers the preponderance of disyllabic words in Bangla, Punjabi and Tamil, and investigates the constraints on the length and prosodic shape of the minimal word in these languages. The conclusion that this paper reaches is that the favoured

278 Reviews strategies deployed to conform to these constraints are vowel lengthening in Bangla, consonant gemination in Punjabi, and both consonant gemination and vowel epenthesis in Tamil. This reviewer feels that the claim about Punjabi is based on facts which are incomplete: it appears that optionally Punjabi also operates like Tamil to some extent, and, in fact, uses a combination of consonant gemination and vowel lengthening. Further, at least insofar as Punjabi is concerned, the vowel lengthening can be considerable and tonal (see some relevant data below, where, informally, the V symbol followed by three ellipsis dots indicates word-final vowel lengthening of this sort, VOC ¼ vocative, and tone is not indicated), such that the added syllable can be used for various purposes, the vocative nature of a nominal (8), being one of a variety of functions so performed (including addition of emphasis, irony, etc.). (8)

S li:mmV . . . Saleem-VOC ‘Saleem’ [A proper name.]

(9)

t§ llV . . . go-VOC ‘Go!’

e

e

An omission in terms of citations is Malik’s (1995) fairly comprehensive study of Punjabi phonology and morphology. The final paper in the volume, i.e., the one by Sailaja, is an experimental study of the interaction between the two distinct writing systems, and the corresponding phonological systems, of Telugu and English. The study involves speakers biliterate in Telugu and English, and strives to examine the way they are processed, specifically in order to determine whether, and to what extent, the respective semi-syllabic and alphabetic writing systems of these two languages contribute towards an awareness of phonological units like syllables and phonemes. The findings appear to suggest that each language is processed in terms of its own script, but that the native language (Telugu) script is slightly more dominant and consequently may influence the processing of the second language (English). To sum up, the volume under review is an interesting collection of papers covering a variety of syntactic, semantic, and phonological issues, and an even wider variety of languages. As previously stated, the linguistic data

J. Bayer et al. Linguistic Theory. Reviewed by Anjum P. Saleemi

279

examined focus on, but are not limited to, South Asian languages. By way of conclusion, it should be pointed out that, in spite of the range of empirical coverage in this volume, its outlook remains predominantly theoretical; in short, it manages to maintain a reasonable balance between breadth and depth.

References Cinque, Guglielmo 1999 Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Jayaseelan, K. A. 1992 Coreference restrictions on pronouns in copular complements. Conference paper, Universite´ de Paris VIII. Jayaseelan, K. A. 1997 Anaphors as Pronouns. Studia Linguistica 51 (2): 186–234. 2000 A Focus Phrase above vP. Proceedings of the Nanzan GLOW Conference: 195–212. Kayne, Richard 1994 The Antisymmetry of Syntax. Cambridge, MA: MIT Press. 2005 A note on the syntax of quantity. In Movement and Silence, Richard Kayne (ed.), 176–214. New York: Oxford University Press. Malik, Amar Nath 1995 The Phonology and Morphology of Punjabi. Delhi: Munshiram Manoharlal. Reinhart, Tanya, and Eric Reuland 1993 Reflexivity. Linguistic Inquiry 24 (4): 657–720. Rizzi, Luigi 1997 The fine structure of the Left Periphery. In Elements of Grammar: Hand book in Generative Syntax, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer.

Colin P. Masica (ed.) – Old and New Perspectives on South Asian Languages: Grammar and Semantics 2007. Delhi: Motilal Banarsidass Publishers, 357 pp. Price: Rs. 1190. ISBN 978-81-208-3208-4

Reviewed by Ashwini Deo

This volume brings together papers from The Fifth International Conference on South Asian Linguistics (ICOSAL-5), held in Moscow in 2003. The papers span the range of language families found in the subcontinent and cover topics ranging from case marking and event structure to expression of information structure, strategies for disambiguation, and areal connections beyond the South Asian area. Reviewing a proceedings volume, which necessarily contains a heterogenous set of perspectives and datasets, is essentially a task of providing synoptic information and pointers to connections with the wider literature, wherever relevant. And that is what I undertake here. The book is divided thematically into three parts; Historical and etymological perspectives (Part I), Crosslinguistic areal or typological perspectives (Part 2), and Old and new descriptive perspectives (Part III). The organization of the review mirrors this division only imperfectly. I first deal with the papers on case, which are scattered through the volume. Then, I move on to the sections as organized in the volume.

Case Bubenik’s paper deals with the transition of the impoverished Middle Indo Aryan case system into the adpositional system for marking semantic case in the New Indo-Aryan languages, building up on his own previous work on the collapse of Old Indo-Aryan case categories (Bubenik 1996, 1998). Besides providing precise information on the semantics of adpositions in late MIA, and the cognates of the NIA postpositional clitics, the paper

C. P. Masica. South Asian languages. Reviewed by Ashwini Deo

281

also discusses the basic syntactic properties of the adpositional system of the understudied Romani (East Slovak variety) branch of NIA. Sigorsky’s paper within the same thematic part, is an exploration into issues of structural case within Hindi diachrony and the Hindi dialect continuum. Filled with rich historical data gleaned from a variety of sources, the paper proposes that loss of the original OIA inflectional marking resulted in a system for Hindi dialects that did not overtly morphologize the ergative-nominative contrast. The paths for the resolution of this pattern are reflected in the innovation of postpositional clitics in Western Hindi and the loss of the syntactic contrast in Eastern Hindi. While the claim itself is not entirely new (see Deo & Sharma 2006 for a formal analysis), the contribution of the paper lies in the detailed examples presenting the variety of structural case patterns in Broader Hindi ergative clauses. Liperorvsky’s paper on Braj, in comparison with standard Hindi in Part II, is comparable to Sigorsky’s in the issues it discusses, though narrower in scope. The paper brings out empirical data from Braj that shows that Braj (a) has both the dative/accusative postpositions attested in the languages of the Western/Central subgroup -ko/nai – one of which is homophonous with the ergative; (b) ergative case marking is optional; (c) case-marked objects in perfective clauses trigger agreement. Taken together, these three papers provide a range of interesting data about diachrony and variation on the semantic and structural case patterns of Indo-Aryan languages that demands a theoretical account. To this section in the review also belongs the paper by Kashi Wali and Omkar N. Koul (located in Part III of the book), a comparison of alternative analyses of double possessive case marking in Kashmiri. The relevant a‰xes are un/uk and hund, which attach to an oblique form of the possessor noun and inflect in agreement in gender, number, and case with the possessee noun. They argue that this phenomenon cannot be considered to be an instance of double case (as suggested by Payne 1995) since it does not cohere with typological patterns observed for double case. They establish the superiority of their hypothesis within the minimalist framework that the un/uk and hund a‰xes are referential and that the gender, number, case features on the possessor are concordial and do not carry case value. In the first section of the book, but of a slightly di¤erent flavor, is the lexical-typological study by Oranskaiya that undertakes a survey of the etymology of the psych verbs meaning fear in Indo-Aryan languages. She shows that this semantic notion is closely connected with the atelic verb of

282 Reviews involuntary motion tremble across languages. tremble is, in turn, etymologically related to the meanings encoded in beat, strike, and split.

Areal perspectives Hans Hock examines features shared by both Dravidian and Indo-Aryan (Vedic) on the one hand, and Turkic (Old and modern Turkish) and the intermediate Iranian and Tibetan languages on the other. The relevant features mainly cluster around non-finite verbal structures, but include properties such as whole word reduplication (a¯mredita) and echo-reduplication. Following up on Masica’s (1976) hypothesis that featural similarities indicate a longer period of contact between Turkic and (particularly South) Indian languages than just the past few centuries of Turkic rule, Hock o¤ers a concrete historical proposal for such ancient contact. The areal similarities, put together with evidence from the discovery of Niya documents and the Ga¯ndha¯ra inscriptions in the area, and similarities between Tibetan and Tocharian A, according to him, point to a Central Asian zone of intense multilingual contact between the three families. While this proposal is not without its problems in the details, Hock considers this proposal a serious alternative to the typological proposal (Hock 2001) which holds that the apparent areal similarities follow from independent diachronic developments having to do with a finiteness constraint in structures with non-finite clauses in verb-final languages. The paper by Hsin-Hsin Liang and Peter Hook aims to demonstrate that verb sequences that look like compound verbs in Mandarin Chinese, are not functionally comparable to the compound verbs attested as an areal feature in Asia. The broader setting of the paper is within the literature on the cross-linguistic similarities between complex predicate constructions (Masica 2001; Butt 2003, a.o), which, according to the authors, fails to give a precise understanding of the di¤erences. While Hindi and Chinese compound verbs pattern alike with respect to orientational minimal pairs (e.g. lena¯-dena¯ give-take in Hindi), completive entailments, incompatibility under the progressive/imperfective or aspectual verbs (this is debatable data as far as Hindi is concerned) etc. there are also significant di¤erences. Chinese compound verb sequences are ambiguous between sequential and aspectual readings, in contrast to Hindi, where the ambiguity holds at very low frequencies. Hindi vector verbs (the non-lexical, aspectual element of the complex predicate) are predominantly

C. P. Masica. South Asian languages. Reviewed by Ashwini Deo

283

transitive while those in Chinese are mainly intransitive. The conclusion from these di¤erences is that the suggestion in Masica (2001) that compound verbs be taken as part of evidence for a single linguistic area incorporating China is untenable. To my view, whether these di¤erences also show conclusively that the structure and semantic contribution of complex predicates in the two languages is di¤erent, remains to be seen. Subbarao’s paper describes a set of common devices used to disambiguate scopally and anaphorically ambiguous clauses across Indo-Aryan, Dravidian, and Tibeto-Burman. It brings out the role of focus particles in restricting the scope of negation, and the role of light verbs with benefactive readings in disambiguating the antecedent of reflexives. The data is rich and the crosslinguistic (within the South Asian area) parallels point to a promising area for more detailed research. Khokhlova and Charanjit Singh demonstrate that Hindi/Urdu and Punjabi resultative participles di¤er with respect to the participant of which the state denoted may be predicated. The resultative may be predicated of the agent, and the paper shows that this type of predication occurs with resultatives from both telic and atelic verbs. A construction, particular only to Punjabi within Indo-Aryan, and similar to Russian is identified, where a possessive marked argument refers to an agent or some other beneficiary/possessor of the event denoted by the verb in the resultative aspect.

Descriptive perspectives The final part, Old and new descriptive perspectives contains papers whose frameworks range from the classical grammatical traditions of Tolkappiyam and Asó ta¯ó dhya¯yi to the modern syntactic and functional-typological ones. The papers with traditional viewpoint are described first. Chevillard is concerned with the relative importance of complex compounding (which he calls unmarked syntax) and non-finite participial structures (marked syntax) in the corpus of Classical Tamil poetry. The relevant compounding structures are those compunding nouns with uninflected verb roots (vinaiyintokai) and compounds licensing case relations (ve¯rrumait tokai). The participial forms are associated with vernacular speech with a reflex in Modern Tamil. Chevillard contrasts the massive presence of uninflected verb forms in compound structures with their near-absence in other varieties of later Tamil, suggesting that these reflect a diglossic

284 Reviews situation holding across the periods of poetic composition. The close examination of and statistical data on the Classical Tamil corpus makes this paper especially interesting from the perspective of research on evidence of diglossia within textual corpora. Boris Zakaryn discusses Pa¯n¸ini’s technique for distinguishing between the mismatch between semantic/morphological function and overt morphological form, showing how Hindi, like every other language, displays such covert operations or formal syncretism in some sub-domains of its morphology. His position that derivational zeromorphemes must be posited does not take into consideration advances in morphological theory that employ more sophisticated methods to resolve mismatches between form and meaning. Davison builds up on the arguments of Hock (1989) concerning the right analysis of relative/correlative structures in Old Indic. Hock had proposed that for Old Indic, these clauses are best analyzed as loosely related paratactic rather than subordinating structures. Davison, concurring with this view, o¤ers an analysis of relative clauses in Hindi-Urdu, which show unambiguous evidence for subordinating structure. Ungrammaticality of whwords inside relative clauses and the simplexity constraint (Dasgupta 1980) are invoked to contrast Hindi with Sanskrit. Davison suggests that relative structures in Hindi involve adjunction, which satisfies the simplexity constraint that the referent of the relative must be within the nearest major consituent to which the relative is adjoined. Sanskrit lacks such a constraint. The paper also discusses discourse structural properties of left and right adjunction of relative clauses. Masica’s paper follows the discourse structural theme of the previous paper, being concerned with the discourse function of post-verbal subjects in Telugu and more generally. The paper shows that like Turkish, post-verbal material often constitutes backgrounded information that is presupposed and part of the common ground. Similar to Tamil (Herring 1994), Telugu also employs the post-verbal position to e¤ect a Topic switch, the reactivation of previously salient discourse referents. He contrasts this function of post-verbal placement with Hindi-Urdu, where it is entirely absent. The paper discusses the question of the placement of new discourse referents (indefinites) in the post-verbal position, attested in both Tamil and Telugu (as in many Indo-Aryan languages) and draws on Herring’s notion of presentational focus to account for these facts. Peterson identifies a dialect of Kharia which freely allows any non-eventdenoting predicate (of individuals) in the language to denote the corresponding inchoative event. The data on attributive and predicative marking

C. P. Masica. South Asian languages. Reviewed by Ashwini Deo

285

and participial/nominalized forms of verbs is adduced to argue for a precategorial open class of elements which may serve predicative or argumentlike functions in the clause as determined by the closed class of grammatical morphology in the language. The Kharia facts connect to the wider typological discussion on word class fluidity (claimed for languages from the Austronesian family and the languages of the Pacific Northwest as well), which challenges the basic typological universal that languages distinguish between nouns and verbs (See Evans and Osada 2005 for a detailed review of the issues involved). In light of Ewans and Osada’s findings for Mundari, the alternative analysis that languages with fluidity between nouns and verbs, in fact, have a significant amount of (often idiosyncratic) zero conversion, the Kharia facts present a rich area for re-examination. The second of the two papers on Munda languages in this volume is by Stichinava and it deals with the temporal/aspectual/modal senses of the past tense markers in Santali. The typologically interesting contribution of this paper is the observation that some of these markers have modal uses. The connection between aspectual and modal meaning has been observed at least since Dowty (1977) and more recently explicated in the domain of the perfect aspect by Portner (2003) and this paper provides a dataset for future explorations in this area.

Evaluation The topics covered in the book are mostly, except for the exception of the discussions on case, too far apart for them to be of overlapping interest or to answer related questions. But this is hardly to be expected of a Proceedings volume. The papers are written by scholars from diverse research traditions, not all of which are equally accessible to the reader. But for the most part, the terminology is transparent and the analyses are not too complicated. For the patient reader, and one well-acquainted with South Asian language and linguistic research, the volume provides rich material for thought. The con seems to be that much of this material is presented in a pretheoretical descriptive way, which often makes it di‰cult to see connections with questions of a more general linguistic interest. The typological connections and parallels drawn between the languages examined and the ones referred to often remain at an anecdotal level, since an explanation for the phenomena are not always o¤ered. However, the volume is

286 Reviews definitely useful as an addition to our empirical knowledge of South Asian languages, through the detailed information it provides on diverse sets of phenomena.

References Bubenik, Vit 1996 The Structure and Development of Middle Indo-Aryan dialects. Delhi: Motilal Banarasidass. 1998 A Historical Syntax of late Middle Indo-Aryan Apabhramsa. Amsterdam: John Benjamins. Butt, Miriam 2003 The light verb jungle. In G. Aygen, C. Bowern, and C. Quinn (eds.) Harvard Working Papers in Linguistics 9: 1–49. Dasgupta, Probal 1980 Question and Relative and Complement clauses in a Bangla grammar. Ph.D. diss., Department of linguistics, New York University. Deo, Ashwini and Devyani Sharma 2006 Typological variation in the ergative morphology of Indo-Aryan languages. Linguistic Typology 10:3, 369–480. Dowty, David 1977 Toward a semantic analysis of verb aspect and the English imperfective-progressive. Linguistics and Philosophy 1.1: 45–77. Evans, Nick and Toshiki Osada 2005 Mundari: The myth of a language without word classes. Linguistic Typology 9: 351–390. Herring, Susan 1994 Postverbal position in Tamil. In Theoretical Perspectives on Word Order in South Asian languages, Butt Miriam and Tracy Kind (eds.) 163–198. Stanford: CSLI. Hock, Hans H. 1989 Conjoined we stand: theoretical implications of Sanskrit relative structures. Studies in the Linguistic Sciences 19.1, 93–126. 2001 Typology vs. convergence: The issue of Dravidian/Indo-Aryan similarities revisited. Yearbook of South Asian Languages and Linguistics 2001: 63–99. Masica, Colin 1976 Defining a Linguistic Area: South Asia. Chicago: University of Chicago Press. 2001 The definition and significance of linguistic areas. Yearbook of South Asian Languages and Linguistics 2001: 205–67.

C. P. Masica. South Asian languages. Reviewed by Ashwini Deo

287

Payne, John 1995 Inflecting postpositions in Indic and Kashmiri. In Double Case: Agreement by Su‰xaufnahme, Frans Plank (ed.). Oxford: Oxford University Press. Portner, Paul 2007 The (Temporal) Semantics and Modal Pragmatics of the Perfect. Linguistics and Philosophy 26.4, 459–510.

Ghanshyam Sharma – Il dizionario di Hindi: Dizionario Hindi-Italiano, Italiano-Hindi 2004. Bologna: Zanichelli.1120 pages

Reviewed by Alice Davison

As the title indicates, this is a bilingual dictionary of Hindi and Italian, intended primarily for Italian learners of Hindi, but also designed to be equally useful for Hindi speakers learning Italian. It is therefore a pedogogical dictionary, focussed on a useful working vocabulary of common and grammatical important words in Hindi. Like all dictionaries, it is based on predecessors, which are listed in the reference section. These sources include important comprehensive dictionaries of Hindi and Urdu such as Platts 1884, Sya¯msurda¯s 1965–1975, McGregor 1995; English-Hindi dictionaries like Bulcke 1986, Kapoor 2000, students’ Hindi-English dictionaries such as Chaturvedi and Tiwari (1980); and comprehensive dictionaries of Italian and Sanskrit. The use of these sources shows that the author has given much care and thought in making the contents of the lexical entries as accurate as possible, and a reliable guide for speakers of two rather di¤erent languages. As a pedagogical dictionary, it is relatively short. The Hindi to Italian section is about 500 pages, the Italian to Hindi section a little longer, for reasons to be commented on below (1)–(2). The particular character of this dictionary originates, it seems, in the many years in which the author has taught Hindi to speakers of Italian at the University of Venice ‘Ca’ Foscari’. The central concern of the author is to convey to the speakers of Italian the particular features of Hindi, by showing corresponences between the two languages. This focus comes from his experience in teaching the grammatical structure of Hindi using the most direct methods, and avoiding when possible technical terms and abstract explanation. There is, to be sure, a common vocabulary of grammatical terms which are familiar to educated speakers of a language, such as lexical categories noun and verb, intransitive and intransitive, modal and

G. Sharma. Il dizionario di Hindi. Reviewed by Alice Davison

289

auxiliary verb, etc. There is a clear and complete list of the abbrieviations used for these terms, in Italian. The grammar of Hindi does present some challenges to Italian speakers, in phonetics, morphology, syntax and semantics. The introduction to the dictionary addresses the first of these. The Hindi-Italian section is arranged in the Devanagari order, in Devanagari with a phonetic transliteration of the standard Indological trandition, rather than the IPA. The Italian pronunciation, with word stress, is given in a Devanagari transliteration. The details of the Devanagari order are clearly explained. The Devanagari symbols are defined according to the phonetic features of place of articulation, manner of articulation, voicing, and aspiration. Corresponding Italian vowels and consonants are used as examples; retroflexion is defined in non-technical articulatory terms. The symbols anusvar and chandrabindu are defined, along with their particular order, but curiously nasalization itself is not defined. This would be a feature of pronunciation which Italian speakers should find di‰cult, as they do in learning French. The syntax of Hindi has several important characteristics for which Italian speakers have nothing which corresponds in their language. One is the ergative case (mostly) of transitive subjects. The author opts for equating ergative subjects with transitivity, so that even verbs like bool-naa ‘say’ which take an object complement are labelled intransitive, because the subject must be nominative, and technically transitive verbs like Dar-naa ‘fear’ are intransitive because the subject is nominative and the direct object has a postposition -see. Nuances of optionally ergative subject verbs such as jiitnaa ‘win’ are omitted in favor of transitivity. The use of postpositions on subjects and objects is expressed in a very clear way, by ‘meeting points’. These are schematic diagrams in the Italian to Hindi section which match the Italian subject and object with the Hindi subject and object, with the appropriate Hindi postposition. These are defined in the introduction and in Sharma (no date). Some of these meeting points are inherent in the entries. Many of the dative subject verbs in Hindi have dative subject counterparts in Italian, such pasand aa-naa ‘like’ and piacere ‘like’; others are categorized as ‘antitranstive’, or dative subject verbs (1). In other cases, the meeting points are spelled out. For example, the entry for incontrare ‘meet’ gives the correspondences for di¤erent uses of mil-naa ‘meet, get’ and a compound verb (2)–(3). Some entries also give sentence examples which illustrate case uses.

290 Reviews 1) miln-aa ‘ottenere’ (get, receive) Antitransitive [dative subject verb] uskoo noobeel puruskaar milaa ‘He received the Nobel Prize’. 2) Incontrare: ‘to meet on purpose’ mil-naa: Italian object takes -see in Hindi. 3) ‘to meet casually’ mulaakaat/bheeNT hoo-naa Italian subject takes -kii in Hindi, Italian object takes -see in Hindi. These meeting points go a long way toward informing the learner about the cases selected by the verb, but if I have any major criticism of this (or most other) student dictionaries, it is that not enough information is consistently given about which postposition to use for the subject and object of specific verbs. Even comprehensive dictionaries such as McGregor 1995 are not always complete in this regard. Verb classes present some di‰culties. There are many compound verbs made up of a noun or adjective (the principal lexical entry) and a verb such as kar-naa ‘do, make’ and hoo-naa ‘be, become’, which, as the author notes, directly indicate the transitivity of the combination. Other possibilities are listed in the N or A entries. One of the most useful features of the entries concerns the Verb-Verb combination, a main verb followed by a ‘vector’ verb which adds adverbial and aspectual information. The author distinguishes these from verbs used only as auxiliaries, such as sak-naa ‘be able’. The most common and idiomatic vector verbs are listed with the main verb in the Hindi section. I assume these choices represent the author’s own intuitions, but a comprehensive study such as Nespital 1997 might also have been a useful source. The vocabulary of Hindi is distinguished by the many sources of the words, particularly in the fact that there are corresponding words from Persian, Arabic and Sanskrit along with the historically derived Indic vocabulary. With the permission of R. S. McGregor, Dr. Sharma has included the source language or etymon of each word in McGregor 1995, where this is information is known. This information is useful to students in giving them a sense of the nuances of Hindi, Urdu and the common language, as well as conveying indirectly some idea of the history of the language. Dr. Sharma is to be commended for undertaking the complex project which this work represents. It may not go quite far enough in some respects, but within the limits established by the author, the results are clearly organized and well executed. The book itself is printed in extremely legible

G. Sharma. Il dizionario di Hindi. Reviewed by Alice Davison

291

fonts, and for a work of such detail seems to be free of typographical errors. While this work may be of limited interest to an English speaking audience, there are nevertheless some benefits. For those who read Hindi, the organization and choices in the Hindi section could give some ideas for a useful vocabulary for students of Hindi whose native language is English or another language. The ‘meeting points’ idea may inspire some useful methods of teaching grammar. The Italian text is very clear and not di‰cult to read if one has some familairity with Italian through French or Spanish.

References Bulcke, C. 1968

Angrezı¯-hindı¯ kos´. English-Hindi Dictionary. New Delhi: S. Chand and Co. Chaturvedi, M. and B. N. Tiwari 1980 A Practical Hindi-English Dictionary. New Delhi: National Publishing House. Kapoor, B. N. 2000 Prabha¯t Practical English-Hindi Dictionary. New Delhi: Prabhat Prakashan. McGregor, R. S. 1997 The Oxford Hindi-English Dictionary. Oxford: Oxford University Press. Nespital, H. 1997 Hindı¯ kriya¯ kos´. Allahabad: Lokbharati Prakashan. Platts, J. T. 1884/ A Dictionary of Urdu¯, Classical Hindi and English. New Delhi: Mun2000 shiram Manohar Lal. Sharma, Ghanyshyam (No date) On how to incorporate linguistic research findings into a bilingual dictionary. MS, University of Venice. Sya¯msundarda¯s (ed.) 1965– Hindı¯ s´abd-sa¯gar. Banaras: Na¯garı¯ praca¯rin¸¯ı sabha¯. 1975

Dialogue

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh

Shishir Bhattacharja

Dasgupta and Ghosh (henceforth DG) published in this journal in 2007, is, to some extent, the continuation of the classical generative study of the structure of B(engali) D(eterminer) P(hrase), inaugurated in the pioneering work by Dasgupta (1983) and then followed by works like Bhattacharya (1999) among others. In this intervention we intend to focus, rather selectively, on those claims of theirs which attempt to answer the following two questions which are, in our view, the most pertinent ones in the study of BDP structure: 1. Why does (3a) have an indefinite reading whereas (3b), (3c) and (3d) have a definite one? 2. Why does the NP move in (3c) despite the presence of an overt Dem whereas in (3d) it does not? 3a. /tin-Ta boi/ (three-Cl-book) ‘Three books’ (Indefinite) 3b. /boi tin-Ta/ (book-three-Cl) ‘The three books’ (Definite/Specific) 3c. /ei boi tin-Ta/ (Dem-book-three-Cl) ‘These three books’ (Definite) 3d. /ei tin-Ta boi/ (Dem-three-Cl-book) ‘These three books’ (Definite) Bhattacharya (1999) claims, along the line of Chomsky (1995), that in (3b) a NP carrying a feature [þspecific] moves, in order to check the feature, to Spec Q(uantifier) P(hrase), a fused head combining both Q(uantifier)/ Num(eral) (henceforth Q) and Cl(assifier). 4.

DP [Spec

D 0 DQP [NPi Q 0 Q ti ]]

However, DG point out the following facts which are not compatible with Bhattacharya’s account:

296 Shishir Bhattacharja 5. With some classifiers some NPs are not allowed to move, as illustrated in (6) and (7) below: 6. */chelejon/ (boy-Cl) ‘The boy’ or 7. */chele-du-jon/ (boy-two-Cl) ‘The two boys’ (but note that /upacarjo-du-jon/ (Vice-chancellor-two-Cl) is acceptable) (see Dasgupta in the present volume) 8. The NP movement beyond Q depends on the content of the quantifier, as exemplified in (9) below: 9. */boi egaroTa/ (book-eleven-Cl) ‘The eleven books’ In order to analyze BDP structure DG adopt what they refer to as the Substantivist theory (elaborated in Dasgupta et al. (2000) and Dasgupta (2005)).1 Unlike the majority of works on DP structure this approach does not consider the classifiers and demonstratives as distinct morphosyntactic entities. It involves W(ord) E(xtension) S(trategies) like (11) which are extentions of W(ord) F(ormation) S(trategies) like (10) of the Whole Word Morphological model of word formation elaborated in Ford et al. (1997). According to Dasgupta (2005: 61) a WES ‘‘maps words onto extended words by attaching a clitic.’’ DG do not formally define either the term ‘clitic’ or the term ‘extended word’. However, in the context of Bengali, we can see the former as an entity concatenated to the input of a WES and the latter as phrases or syntactic sub-trees that merge at particular sites in a syntactic structure. This said, we will avoid the term ‘clitic’ in this text and will designate the cliticised entities like /tin-Ta/ in (11) as ‘constants’ of WES.2 10. /X/Unclassified Q $ /XTa/Classified Q /tin/ ‘three’ $ /tinTa/ ‘three-Cl’ /koyek/ ‘a few’ $ /koyekTa/ ‘a few-Cl’ 11. /X/NðounÞ; Indefinite; Unquantified ! /XtinTa/NP; Definite; Quantified; Plural 3 /boi/ ‘book’ ! /boi-tin-Ta/ (book-three-Cl) ‘the three books’ According to DG we can handle (3b) with (10) and (11). The right-hand outputs of (10) appear as constant in (11) and if /boi/ ‘book’ is mapped onto this WES, it outputs (3b). In absence of an overt Dem, the outputs of (11) merge at Q and then Q moves to SpecDemP. Subsequently, the null Dem moves to D in order to check the [þdefinite] feature, because, as (13)

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh 297

shows, Dem alone cannot yield definiteness in a good number of cases. DG assume that head-movement is allowed for functional heads like Dem and Q. 12.

DP [Spec

D 0 D DemP [Spec Dem 0 Dem

QP [Spec

Q 0 Q]]]

13. */ei bhalo nOy/ (this-good-not) ‘this is not good’ DG claim that (3d) can be handled with two WES: (14) and (15). The output of (14), a DemQ sequence, appears as constant in (15) which outputs a quantified, definite and plural NP. This NP merges at Q and then, Q moves to Spec DemP in order to check the [þdefinite] feature, and consequently, its left most element, Dem /ei/ ‘this’ is niched in D. 14. /XTa/Q; Classified ! /eiXTa/Q; Classified; Definite; Plural /tin-Ta/ (three-Cl) ‘the three’ ! /ei-tin-Ta/ (Dem-three-Cl) ‘these three’ 15. /X/N; Indefinite; Unquantified ! /ei-tin-TaX/NP; Quantified; Definite; Plural /boi/ ‘book’ ! /ei-tin-Ta-boi/ (Dem-three-Cl-book) ‘these three books’ DG attempt to answer (1) and (2) and try to account for (5) and (8). However, some of the examples which are problematic for Bhattacharya (1999) are also problematic for DG. The present intervention aims to draw our attention to those examples. Subsequently, in order to account for (3c) on the one hand, and to handle some di¤erent data on the other, we propose some modifications in the WES structure and also in the syntactic structure of BDP a` la DG. According to DG (5) and (8) can be handled (13) ‘‘by constraining the Strategies.’’ The formation of (6) can indeed be avoided by claiming that there exists no strategy like (16). We can form (7) with (17) although some speakers may find it less acceptable than other outputs of this WES. Sequences like (9) can be formed with (18) or (19) but as we can see below, speakers are far from being unanimous about the acceptability of their outputs. The general rule in this respect seems to be the following: Ns denoting objects of comparatively smaller size are allowed to precede a bigger Num. However, the acceptability of the sequences in which the N precedes Q depends, as (18) and (19) show, both on the size of the Num and the nature of the N. Therefore, if we want to constrain these WES, we must make a list of the Ns that are not allowed to be mapped onto them, but this is

298 Shishir Bhattacharja indeed a very di‰cult task. In our view, idiosyncratic examples like (7) and (9) are di‰cult to handle both in the classical generative approach and in the substantivist one.This said, as idiosyncrasies are rather unusual in syntax these examples should in principle be better handled in approaches like DG. 16. /X/N; Human; Indefinite; Singular $ /Xjon/N; Human; Definite; Plural /chele/ ‘boy’ $ */chelejon/ (boy-Cl) ‘the boy’ /shikkhok/ ‘teacher’ $ */shikkhokjon/ (teacher-CL) ‘the teacher’ 17. /X/N; Unquantified; Indefinite; Singular ! /Xdujon/NP; Quantified; Definite; Plural /chele/ ‘boy’ ! ?/chele-dujon/ (boy-two-Cl) ‘the two boys’ /Dakat/ ‘bandit’ ! /Dakat-dujon/ (bandit-two-Cl) ‘the two bandits’ 18. /X/N; Indefinite; Unquantified; Singular ! /XegaroTa/NP; Definite; Quantified; Plural /boi/ ‘book’ ! ?/boi-egaro-Ta/ (book-eleven-Cl) ‘the eleven books’ /Taka/ ‘money’ ! /Taka-egaro-Ta/ (money-eleven-Cl) ‘the eleven rupees’ 19. /X/N; Indefinite; Unquantified; Singular ! /XpOncashTa/NP; Definite; Quantified; Plural /licu/ ‘lichi’ ! /licu-pOncash-Ta/ (lichi-fifty-Cl) ‘the fifty lichis’ /desh/ ‘country’ ! ?/desh-pOncash-Ta/ (country-fifty-Cl) ‘the fifty countries’ It seems that (3c) cannot be formed with a WES a` la DG because the input of a WES must be a word. We cannot input Dem because DG prefer to treat categories like Dem as constants in WES rather than as inputs because there are fewer Dems than members of other categories like Q and N. If we decide to input N, then, (3c) will involve two ‘circumfixal’ constants: Dem and Q. We can however overcome this problem if we are allowed to map ‘extended words’ like the outputs of (11) onto a WES like (21), or activate P(hrase) F(ormation) S(trategies) like (22) or (23).4 A PFS involves several variables and can perhaps be better formulated as: 20. /X/a ! /X 0 /b where a is a lexical or phrasal category and b is a phrasal category.

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh 299

21. /XTa/NP; Non-deictic definite ! /ei/XTa/NP; Deictic definite; Plural 5 /boi-tin-Ta/ (book-three-Cl) ‘the three books’ ! /ei-boi-tin-Ta/ (this-book-three-Cl) ‘these three books’ 22. //X/N /Y/Q /Z/Cl /NP; Non-deictic definite ! //W/Dem /X/N /Y/Q /Z/Cl /NP; Deictic definite; Plural /boi-tin-Ta/ (book-three-Cl) ‘the three books’ ! /ei-boi-tin-Ta/ (this-book-three-Cl) ‘these three books’ /ciThi-shat-khana/ (letter-seven-Cl) ‘the seven letters’ ! /sheiciThi-shat-khana/ (that-letter-seven-Cl) ‘those seven letters’ 23. /X/N ! //W/Dem /X/N /Y/Q /Z/Cl /NP; Deictic definite; Plural /boi/ ‘book’ ! /ei-boi-tin-Ta/ (this-book-three-Cl) ‘these three books’ /ciThi/ ‘letter’ ! /shei-ciThi-shat-khana/ (that-letter-seven-Cl) ‘those seven letters’ Following Bhattacharya (1999) DG consider Q as a fused head. However, examples like (24–25) make us suspect such a decision. If we compare (24) with (3a), it becomes clear that when Q follows Cl the NP lacks number specification. We can note in (26–27) that M(easure) W(ord)s also behave like Cl in this respect.6 24. /goTa tin boi/ (Cl-three-book) ‘more of less three books’ 25. /goTa koyek boi/ (Cl-a few-book) ‘a few books’7 26. /bOsta dui cal/ (MW-two-rice) ‘about two bags of rice’ 27. /dui bOsta cal/ (two-MW-rice) ‘two bags of rice’ DG seem to consent to the claim by Bhattacharya (1999) that (3b) has a specific meaning. In our view, (3b) cannot have a specific reading because, in this sequence, N precedes Q and such an order, as we can see below (28– 31), is unacceptable in a context of specificity but acceptable in a context of definiteness. We consider (29–30) as contexts of specificity and (31) as a context of definiteness because in (29–30), the BDP substitutes its antecedent (28) partially whereas in (31), it substitutes it completely. We assume with Enc¸ 1991 and Campbell 1996 that a specific NP must substitute its antecedent partially, and the definite one, completely. 28. /Tebile [shatTa boi] ache/ (on the table-seven-Cl-book-are) ‘there are seven books on the table’

300 Shishir Bhattacharja 29. */er moddhe [boi tinTa] ami cai/ (of this-among-book-three-Cl-Iwant) ‘I want three books among them’ (Definite Specific) (see endnote5) 30. /er moddhe [tinTa boi] ami cai/ (of this-among-three-Cl-book-Iwant) ‘I want three books among them’ (Indefinite Specific) 31. /[boi shatTa] ami cai/ (book-seven-Cl-I-want) ‘I want those seven books’ (Definite) Bhattacharya (1999) considers Dem as an XP adjoined to QP because for him examples like (32–33) are unacceptable. As he (77) argues, ‘‘If the Dem is a head, then it is di‰cult to see how it can act as a barrier to XP movement.’’ In our view, (32–33) are acceptable and therefore, Dem is a barrier to no head or XP other than Q and Cl.8 32. /boi ei tin-Ta/ (book-this-three-Cl) ‘These three books’9 33. /nutan ei tin-Ta boi/ (new-Dem-three-Cl-book) ‘These three new books’ If the objection against Dem heading a projection is overruled, then, we can propose (34) as the structure for BDP which has, as we can see, a D(emonstrative) P(hrase) between DP and ClP. 34.

D 0 D DemP [Spec Dem 0 Dem 0 NP [Spec N N]]]]] DP [Spec

ClP [Spec

Cl 0 Cl

QP [Spec

Q0 Q

DG propose to merge the outputs of WES at Q. In our view, they can also merge as the complement of Q and then move to particular sites in order to check their features. In (3a) the output of (35) moves to Spec ClP in order to check number specification and then /tinTa/ (three-Cl) is niched in Cl and /boi/ ‘book’ in N. In (3b) an output of (11) moves to Spec DP (via Spec DemP) to check the non-deictic definite feature and then /tinTa/ (three-Cl) is niched in Cl. In (3c) and (3d) the outputs of (22) and (15) respectively move to Spec DemP to check the deictic definite feature and then /ei/ ‘this’ is niched in D and /tinTa/ in Cl. The di¤erence between (3c) and (3d) is that in the former N is niched in Spec DemP whereas in the latter, it is niched in N (via Dem, Cl and Q). We note that in our approach i) no extra movement is required for checking

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh 301

[þdefinite] feature and ii) the head movement constraint does not need to be violated. 35. /X/N; Indefinite; Unquantified ! /tinTaX/NP; Quantified; Definite; Plural /boi/ ‘book’ ! /tin-Ta boi/ (three-Cl-book) ‘the three books’ (3c) and (32) are problematic for the classical generative approach because in these examples N moves overtly when it does not have to. In presence of an overt Dem [þ definite] feature should be checked lexically and/or covertly through Probe-Goal (cf. Epstein and Seely 2006). Now, it is not impossible that some speakers find these examples ‘more definite’ than (3d) and we can account for this fact by claiming that it is so because unlike in (3d), in (3c) and (32) the N is niched in Spec DemP and Spec DP respectively – the two sites that are responsible for checking definiteness. (33) is also problematic for the classical generative approach because the motivation behind AP movement is not clear. If we compare (36) with (37), then it become clear that if AP precedes Q the NP has a definite reading. Now, the question we may ask is whether it is possible for the NP in (33) and (37) to check its [þdefinite] feature by sending one of its adjuncts to Spec DP. (33) and (37) are problematic for the Substantive approach as well because it is not yet clear whether we can handle instances of adjunction and complementization with PFSs like (38) and (39) respectively.10 36. tin-Ta nutan boi (three-Cl-new-book) ‘Three new books’ 37. /nutan tin-Ta boi/ (new-three-Cl-book) ‘The three new books’ 38. //X/N /Y/Q /Z/Cl /NP; Non-deictic definite ! //W/A /X/N /Y/Q /Z/Cl /NP; Non-deictic definite; Plural /boi-tin-Ta/ (book-three-Cl) ‘the three books’ ! /nutan-boi-tinTa/ (new-book-three-Cl) ‘the three new books’ /ciThi-shat-khana/ (letter-seven-Cl) ‘the seven letters’ ! /puronociThi-shat-khana/ (old-letter-seven-Cl) ‘the seven old letters’ 39. /X/NP ! //Y/Pronoun /X/NP /Z/Inflected verb /IP=S 11 /boi-tin-Ta/ (book-three-Cl) ‘the three books’ ! /ami-boi-tin-Taporechi/ (I-book-three-Cl-have read) ‘I have read the three books’ /boi-Ta/ (book-Cl) ‘the book’ ! /tumi-boi-Ta-porecho/ (youbook-Cl-have read) ‘You have read the book’

302 Shishir Bhattacharja Unlike the week lexicalist approaches of the Generative school which put Syntax at the service of Morphology, the Substantive approach puts, to some extent, Morphology at the service of Syntax. In our view, DG o¤er a better account of the syntax of BDP as compared to previous analysis done in light of the classical generative theory. However, some problems like (8) persist and they should be taken care of in the future research.

Abbreviations BDP Cl ClP Dem DP DemP IP

Bengali Determiner Phrase; Classifier; Classifier Phrase; Demonstrative; Determiner Phrase; Demonstrative Phrase; Inflectional Phrase;

MW Num QP PFS Q WES WFS

Measure Word; Numeral; Quantifier Phrase; Phrase Formation Strategy; Quantifier; Word Extension Strategy; Word Formation Strategy.

Notes 1. According to Dasgupta (2005: 60) ‘‘The substantivist approach operates on the assumption that levels of linguistic characterization must converge on formal objects. Morphology and Syntax co-specify a word. Syntax and pragmatics co-characterize a sentence.’’ 2. According to Ford et al. (1997) some words can be analyzed into two subcomponents: i) variable and ii) constant by mapping it onto a relevant WFS. For example, an English word like friendly can be analyzed into the variable ( friend ) and the constant (ly) or an Arabic word like /kitab/ ‘book’ can be analyzed into the constant /i/-/a/ and the variable /k/-/t/-/b/. Constants can be represented by any phonic element: stress, phonemic change, single phoneme, meaningless sound cluster, simple or complex word, discontinuous and inseparable segmental as well as supra-segmental means. Quite coincidentally, when a constant is represented by a continuous sequence of phonemes, it can have phonic resemblance to word parts which some grammatical traditions would label as a‰xes (see Singh and Agnihotri 1997 or Bhattacharja 2007b for details). 3. The lexical relatedness between the inputs and the outputs of a WFS is shown by a bidirectional arrow $ in order to symbolize its bidirectional implication. In a WES on the other hand, the rather ‘syntactic’ relatedness between the input and the output is shown with a unidirectional arrow !. Although the

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh 303 pattern underlying a WFS and a WES must be repeated, we give only one example for each WES to save space, unless required otherwise. 4. In Dasgupta et al. (2000: 171), one of the expositions of the Substantive approach (DG 2007 being a di¤erent exposition), a PFS has the following form: 40. [X]properties ! [Xþf ]properties 0 According to the authors (171) PFSs are expected to underwrite syntagms which are ‘‘permitted by the syntax, to which they are accountable – to be tighter than normal syntactic constructions and to exhibit other opacities. The unidirectional arrow of (40) ‘‘derives a functor-headed phrase Xþf (linear order of f, X immaterial) from a word X and simultaneously maps the set Properties into the set of Properties’. The authors claim (171) that ‘‘this mapping and the substantive (phonological and semantic) relation between f and X are continuous with and accountable to morphology. Xþf as a formal structure is continuous with and accountable to syntax.’’ 5. Following Bhattacharja (2007a) we claim that both definiteness and specificity are based on the familiarity which can be: i) Deictic, or ii) Non-deictic (anaphoric, relational, pragmatic, etc.). We also claim that specificity can be i) Definite and ii) Indefinite. 6. One of the di¤erences between the syntactic behavior of Cls and MWs is that unlike the former, the latter are not generally concatenated to N: /chele-Ta/ (boy-Cl) ‘the boy’, /boi-khana/ (book-Cl) ‘the book’, but not */gom-bOsta/ (wheat-bag) ‘a bag of wheat’ or */ca-kap/ (tea-cup) ‘a cup of tea’. This said, as we have seen in example (6), not all Cls are allowed to be concatenated to N either. Ns are allowed to move beyond Q-MW sequence (41–42), and this movement yields, as with the QCl sequence (3b), definiteness to the NP: 41. cal dui-bOsta (rice-two-bag) ‘the two bags of rice’ 42. ca tin-kap (tea-three-MW) ‘the three cups of tea’ 7. The semantic di¤erence between (25) in which Q follows Cl and (43) in which Q precedes Cl is not reflected in their common English gloss. 43. /koyekTa boi/ (Q-Cl-book) ‘a few books’ 8. We can propose the following syntactical constraints for the structure of BDP: i. Q-Cl (or Cl-Q) sequence must not be interrupted; ii. Q must not precede Dem; iii. Cl must not precede Dem. 9. We are aware of the fact that (3b) and (32) can be pronounced with a pause (represented with ‘–’ in (44–45)) after the N and also probably with some variation in the stress pattern. We agree with DG that in examples like (44– 45) /boi/ is not a constituent but a DP which (22) ‘‘begins its career as a DP

304 Shishir Bhattacharja adjoined to a DP and then moves from that site to a non-argument position in the clause . . .’’ We assume that the semantic reading of (3b) and (32) is di¤erent from the semantic reading of (44) and (45) respectively. 44. /boi – ami ei tin-Ta porechi/ (book – I-Dem-three-Cl-have read) ‘As for books, I have read these three’ 45. /boi – ami tin-Ta porechi/ (book – I-three-Cl-have read) ‘As for books, I have read three of them’ 10. The PFSs we propose here are compatible with the minimalist approaches like Epstein and Seely (2006) in the sense that they can be used as tools for derivational operations untill spellout. Epstein and Seely claim, consonant with the Minimalist program, that the operations merge and move each takes two objects, join them together (as a set) and then project one or the other, hereby creating a label for the resulting object. Hence, if X and Y are merged creating C, then C is necessarily the input to both LF and PF, which interpret as much of C as possible, while C may serve as input to subsequent derivational operations. 11. It is conceivable that an output of (39) can be mapped onto the syntactic tree of IP (46) and then, the subject pronoun (/ami/ ‘I’) is niched in Spec IP, and the inflected verb (/porechi/ ‘I have read’) in I. 46. IP [Spec (ami) I 0 VP [Spec V 0 VDP [Spec (boi) D 0 D DemP [Spec Dem 0 Dem 0 0 ClP [Spec Cl Cl (tin-Ta) QP [Spec Q Q NP [Spec N N]]]]]] I (porechi)]

References Bhattacharya, Tanmoy 1999 Specificity in Bangla DP. The Yearbook of South Asian Languages and Linguistics 1999: 71–99. Bhattacharja, Shishir 2007a Familiarity, specificity, (in)definiteness and substitution. Journal of the Institute of Modern Language, University of Dhaka 2006–2007: 21–26. 2007b Word Formation in Bengali: A Whole Word Morphological Description and its Theoretical Implications. Munich: Lincom Europa. Campbell, Richard 1996 Specificity operators in spec DP. Studia Linguistica 50(2): 161–188. Chomsky, Noam 1995 The Minimalist Program. Cambridge: The MIT press. Dasgupta, Probal 1983 The Bangla Classifier /Ta/, its penumbra and definiteness. Indian Linguistics 44: 11–26.

Bengali Determiner Phrase Revisited: A Response to Dasgupta and Ghosh 305 2005

Q-baa and Bangla clause structure. The Yearbook of South Asian Language and Linguistics 2005: 45–81, Berlin: Mouton de Gruyter. Dasgupta, Probal, Alan Ford, and Rajendra Singh 2000 After Etymology: Towards a Substantive Linguistics. Muenchen: Lincom Europa. Dasgupta, Probal, and Rajat Ghosh 2007 The nominal left periphery in Bangla and Asamiya. Annual Review of South Asian Languages and Linguistics 2007, Rajendra Singh (ed.), 1–27. Berlin/New York: Mouton de Gruyter. Enc¸, Mu¨rvet 1991 The semantics of specificity. Linguistic Inquiry 22: 1–25. Epstein, Samuel David, and Daniel T. Seely 2006 Derivations in Minimalism. Cambridge: Cambridge University Press. Ford, Alan, Rajendra Singh, and Gita Martohardjono 1997 Pace Panini, towards a Word-based Theory of Morphology. New York: Peter Lang. Singh, Rajendra and Rama Kant Agnihotri 1997 Hindi Morphology, a Word-based Description. Delhi: Motilal Banarsidass publishers.

The Study of ‘Indian English’: What is It Good for?

Lionel Wee

Perhaps a medical analogy will be useful in helping readers anticipate my line of argument. In a nutshell: I largely agree with Singh (2007) that the Indian English patient is su¤ering from the ill e¤ects of various unfounded theses. However, while Singh diagnoses localized infections for which he recommends the surgical separation of ‘form-related linguistics’ and a political economy of English as a cure, I see the need for something more akin to therapy, where assumptions regarding the ontology of language need to be revisited.

Singh (2007): A brief summary Singh (p31) argues that the study of Indian English is plagued by the following three theses: (1) that IE is both a non-native and deviant variety of English, (2) that IE is not a ‘genuinely Indian language’, and (3) that any understanding of IE has no choice but to also attend to ‘the sociolinguistics of English in India and the political economy of the contemporary world and India’s place in it.’ He points out that, as a consequence of these theses, analyses of the grammatical properties of IE have been overly concerned with highlighting the ways in which it di¤ers from native varieties, and attributing such di¤erences to substratal influences as though native varieties were not themselves the products of similar influences. This attention to grammatical ‘exotica’ comes at the price of overlooking the commonalities that IE in fact shares with other varieties of English, such as broader morphological, syntactic and phonological patterns or rules (p36–7). Such unwarranted obsession with documenting ‘exotica’ to the point where commonalities are ignored arises, Singh suggests, from the failure to separate questions about ‘the place of English in India’ from questions about the grammatical structure of IE (p40). The former is a sociolinguistic venture, which he feels is better reconstituted as an investigation into the political economy of English with

308 Lionel Wee all its attendant focus on socio-historical contingencies; the latter is a properly linguistic enterprise committed to the unearthing of psycholinguistic universals. Singh traces this unhappy state of a¤airs to two major sources. The first is the traditional influence of language pedagogy, which has not only characterized non-native productions as errors, but has compounded this by comparing what are in fact performance data of ‘non-natives’ with the presumed competence of ‘natives’. The second is the field of sociolinguistics, in particular, that branch concerned with the study of World Englishes, which, by introducing the notion of ‘non-native varieties’, has e¤ectively undermined the idea that all linguistic systems are essentially autonomous. The labeling of IE as a ‘non-native variety of English’ has e¤ectively doubly condemned it to an outsider status, both vis-a`-vis more established varieties of English, but also in relation to other Indian languages (p38). Singh’s proposal is that progress in the study of IE can only come about if analysts put aside the belief that ‘a sociolinguistics of English in India or elsewhere’ is a viable scholarly enterprise (p40). Instead (op. cit, upper case in original): . . . there can only be a form-related linguistics and a political economy of English, and the latter can determine only when the linguistic argument would be heard and NOT what it would look like.

Scholarly activities that have been wrongly conflated therefore need to be separated. Pursuing a form-related linguistics would entail recognizing that ‘we are all native speakers of the steady state grammar we develop’ (p34) and any correlative investigations would have to be ‘grounded squarely in the reality and psycholinguistics of multilingualism’ (p36). Pursuing a political economy of English, on the other hand, might involve studying how colonial and postcolonial policies, in concert with the ever-increasing economic and political interdependence that often goes under the rubric ‘globalization’, are influencing the cultivation and movement of speakers and language institutions (p40).

Degrees of regularity Clearly, a form-related linguistics of IE must be capable of dealing with both the exotic and the mundane, while aiming to capture the relationship

The Study of ‘Indian English’: What is It Good for?

309

between them. In this regard, some examples of the exotic may be easily absorbed as rather uninteresting lexical instantiations of more general patterns. For example, ‘the rules that can and do generate room-mate and directorate will also generate batch-mate and collectorate’ (p37). More to the point, such rules are not unique to IE – they are also found in other varieties – and any analysis that celebrates the uniqueness of items like batch-mate while failing to take note of the available broader generalizations is simply guilty of bad faith. There are, however, other cases that are less blatantly or clearly exotic, and so deserve attention without necessarily being reduced to a set of crossvarietal commonalities. Thus, it’s been noted (Schneider 2007: 169 and references within) that /t/ and /d/ in IE tend to be retroflexed, that syllable-initial voiceless stops are generally unaspirated, that some mass nouns are pluralized (alphabets, furnitures), and that adjectives and verbs reduplicate. While no single feature on its own serves to distinguish IE from other varieties, it seems fair to say that it is existence of these congregations of grammatical features that (taken together) contributes to the relative distinctiveness of IE, and therefore help to give a sense of coherence to the intuition that IE can exist as an object of study. One could possibly posit a rule or pattern of maximal generality so that reduplication, say, becomes seen as just another variant of a morphological operation that at some higher level is shared with other varieties of English. The same, with su‰cient energy, could be achieved for observations regarding the phonology and syntax of IE. But it would appear that more is lost than gained by insisting on such an enterprise, since the analytical challenge is to be able to say something useful about why there is this variety that (at least some) people refer to as ‘Indian’ and ‘English’. It is therefore important to be able to focus on both the exotic and the mundane, and to recognize that no sharp line separates one from the other. Ignoring one at the expense of the other either loses us any sense that we are investigating a distinct variety (thus the ‘Indian’ part of IE is lost), or it prevents us from connecting this variety to others (thus the ‘English’ part of IE is nowhere to be found). Appreciating that the distinction between the exotic and the mundane is a matter of degree also means recognizing that patterns or rules are not always fully productive, and some of the patterns that do license more specific instantiations are themselves fairly idiosyncratic – that is, not easily reducible to larger patterns. This is the main insight that drives the enterprise of construction grammar (Croft 2001; Fillmore, Kay and O’Connor 1988;

310 Lionel Wee Goldberg 1995). Constructions are patterns of form and meaning, and these can be of varying degrees of specificity. For example, Kay and Fillmore (1999) argue that sentences such as What’s this fly doing in my soup? and What is this scratch doing on the table? are examples of a relatively idiomatic What’s X doing Y? construction. ‘What’s’ and ‘doing’ are lexically invariant parts of the construction while ‘X’ and ‘Y’ represent empty slots that can be satisfied by various noun phrases. As they point out, the construction has a very specific pragmatics associated with it. It does not simply ask about an activity; rather, it indicates the speaker’s judgment that there is an incongruity (for example, the presence of the fly in the soup). In contrast, a less idiomatic expression such as The man painted the house is an instance of a Transitive Verb Construction, which is highly schematic in that it does not demand the presence of any specific lexical item. This construction in turn instantiates an even more general Subject Predicate Construction. In this way, construction grammar aims to account for the entire continuum of linguistic expressions, ranging from those that are highly irregular to those that are more prosaic, by looking at how constructions are related to each other, and how they combine to form newer ones. Thus, Goldberg (2007: 10) suggests that an expression such as What did Liza buy Zach? involves a combination of the following constructions: Liza, buy, Zach, what, do constructions Ditransitive construction Question construction Subject-Auxiliary inversion construction VP construction NP construction Likewise, the expression What’s Zach doing buying Liza’s soup?, too, involves a combination of constructions, including Liza, buy, Zach, what and do. But crucially, to get at the right kind of relationship between the morphosyntax and the pragmatics, this expression also makes use of the What’s X doing Y? construction, which plays no part in What did Liza buy Zach?. The idea of a construction is a response to the empirical observation that even within what we would consider a single variety of English, there are unpredictable pairings of form and meaning. Extrapolating from this observation across di¤erent varieties of English, it is not only possible to agree with Singh (p37) that ‘there are no structural features at any level

The Study of ‘Indian English’: What is It Good for?

311

of grammatical description, that characterize all ‘‘non-native’’ varieties of English to the exclusion of all ‘‘native’’ varieties, it in fact becomes necessary to go further and ask whether the homogeneity presumed in phrases like ‘all ‘‘non-native’’ varieties’ and ‘all ‘‘native’’ varieties’ is justified. Some constructions, it is true, are more commonly distributed across di¤erent varieties of English (e.g. the Transitive Verb Construction); others, on the other hand, may be more specific to particular cultures or sub-domains of use (e.g. the What’s X Doing Y Construction). But more significantly, the distribution of lexicogrammatical constructions crosscuts the boundaries of varieties of English, regardless of whether these are ‘native’ or ‘nonnative’. At this point, the following questions present themselves: (1) How would pragmatics fit into a form-related linguistics of IE? (2) What do we gain by continuing to speak of varieties as though their labels (e.g. ‘Indian English’, ‘Singapore English’, ‘American English’, etc) reflect meaningful grammatical boundaries?

A place for pragmatics . . . and, ultimately, sociolinguistics If we are interested in understanding what constitutes the linguistic knowledge of a competent speaker of IE (I’ll continue to speak of ‘IE’ as a variety for ease of exposition, although it would actually be more precise to think of ‘IE’ here as referring to an individual speaker’s repertoire of constructions), then there is no reason to exclude pragmatic knowledge, given the observations regarding collocations of form and meaning discussed in the previous section. In other words, a ‘form-related linguistics’ does not, and probably should not, be taken to mean ‘form only’. Rather, insofar as particular forms of language are connected with pragmatic knowledge, then ‘form-related’ should be understood as ‘pragmatics-related’ as well. In fact, the distinction between pragmatics, semantics and discourse functions is best considered a matter of degree, depending on just how conventionalized the relationship between a particular meaning and a given form happens to be. But at this point, it becomes necessary to acknowledge that it is not all that easy to separate sociolinguistics from a form-related linguistics, especially the kind of interactional sociolinguistics that treats linguistic forms as resources for the localized negotiation of meanings (Coupland 2007;

312 Lionel Wee Eckert 2000). This kind of work recognizes that linguistic forms index various kinds of meanings, and even primarily ‘referential’ meanings convey socially evaluative stances (Agha 2007). For example, pronominal choices do not merely refer to (groups of ) individuals, but can be deployed as solidarity or distance markers (Brown and Gilman 1960). Studies of French tu and vous are perhaps paradigmatic, but the general point extends to the study of other grammatical devices, and – in the context of the present paper – to the study of IE as well as other varieties of English. Similarly, given the observations mentioned above that /t/ and /d/ in IE tend to be retroflexed, that syllable-initial voiceless stops are generally unaspirated, etc., it is not inconceivable that such properties can function as resources for the marking of social group identity, along the lines spelt out by Bell’s (1984) audience design framework or accommodation theory (Giles, Coupland and Coupland 1991). One might suggest that interactional sociolinguistics is a rather di¤erent animal from the study of the political economy of English. But while it is true that each enterprise can be pursued separately, it is also the case that how individual speakers decide to marshal their linguistic resources to manage local interactions is often influenced by their perceptions of how said resources are likely to be received. Here, political economic considerations enter the picture in the sense that broader societal understandings of language do play a role in a¤ecting the social distribution of grammatical forms and potentially impacting on how such forms are reproduced or reanalyzed. And conversely, the reproduction or re-analysis of grammatical forms in localized interactions can also contribute to the perpetuation and transformation of the political economic status of (varieties of ) Engish. For example, attempts by language activists to improve the social acceptability of otherwise stigmatized varieties (e.g. Hawai’ian Creole English) so that it can be used in a wider range of social domains oftentimes involve processes of functional elaboration, codification, and graphization (Blommaert 2001: 137), all of which potentially lead to changes in the forms and their associated meanings. This interaction between broader societal understandings of language and the properties of grammatical forms also suggests we need to tweak the description of the native speaker as someone in possession of a ‘steady state grammar’. A ‘steady state grammar’ does not preclude change, though perhaps someone like Lightfoot (1991) might want to insist on a sharp distinction between ‘core’ and ‘peripheral’ realms of the grammar. Be that as it

The Study of ‘Indian English’: What is It Good for?

313

may, it seems clear that there is no principled reason for excluding the relatively less stable or less entrenched aspects of grammar from any study of what speakers of IE know and/or do. This is especially because it is becoming increasingly clear that an important part of what it means to be a native speaker is the knowledge of conventionalized sequences, stock phrases and idiomatic collocations that exist alongside more general rules of grammar (Pawley and Snyder 1983; Widdowson 1989). Thus, knowing what the conventional ways of saying things are is an important part of knowing how a particular discourse community works and how interactions with members of such a community are appropriately conducted. But it is crucial to bear in mind that while all those individuals that we would characterize as ‘speakers of IE’ may have share (largely) the same grammar, there are also going to be unpredictable di¤erences such that things that we might want include in one speaker’s steady state might turn out to be better located in another speaker’s newly acquired list of grammatical innovations. We therefore need to more consistently appreciate the dynamics of language, and not forget that the continued employment of the label ‘Indian English’ in fact disguises a whole series of ongoing changes and variations, both at the intra-individual and inter-individual levels. The study of IE is not the study of a completed linguistic system with hard boundaries separating it from other varieties. We in fact have to rethink the ontological status of a language: it is ‘not a circumscribed object but a confederation of available and overlapping social experiences’ (Hopper 1998: 171). As language conventions change – and they inevitably do – what were once regularities may now be perceived as irregularities, and vice versa. Hopper (1998) explains this by describing grammar as emergent since ‘structure, or regularity, comes out of discourse and is shaped by discourse in an ongoing process. Grammar is, in this view, simply the name for certain categories of observed repetitions in discourse’ (1998: 156). This leads us to the second question mentioned at the end of the preceding section, namely, the extent to which it is useful to continue to speak of IE as a variety.

‘Indian English’ as a heuristic We need to be careful when speaking of distinct varieties, whether these are regarded as separate languages or dialects of a language, for the labels that

314 Lionel Wee we give to each variety can easily mislead us into assuming that it is unproblematic to be treat language as a bounded entity. And, in particular, to speak of specific names as inextricably tied to specific identities is to buy into cultural essentialism. In fact, understandings of how particular constructions belong to specific varieties are usually the historical products of broader socio-political acts of labeling: particular constructional repertoires become categorized as ‘languages’ while others are classified as ‘dialects’ or ‘jargon’. These metadiscursive regimes or representations about language reflect invented understandings of language, in particular, conceptualizations of languages as ‘separate and enumerable categories’ (Makoni and Pennycook 2007: 2). The invention of languages has been discussed in a number of works (Harris 1980; Mu¨hlha¨usler 2000), and a brief example will su‰ce here. As Makoni and Pennycook (2007: 9) point out, Sir George Abraham Grierson’s linguistic Survey of India, which was completed in 1928, had to face the problem of deciding on the boundaries between languages and dialects. To do this, Grierson openly admitted the need to invent language-names while ignoring the complexity of actual language use (1907: 350), quoted in Makoni and Pennycook 2007: 10: . . . nearly all the language-names have had to be invented by Europeans. Some of them, such as Bengali, Assamese, and the like, are founded on words which have received English citizenship, and are not real Indian words at all, while others, like ‘Hindostani’, ‘Bihari’, and so forth, are based on already existing Indian names of countries and nationalities.

The significance of this, as Makoni and Pennycook (2007: 10) observe, is that ‘these were not just new names for existing objects . . . but rather the invention and naming of new objects. The naming performatively called the languages into being.’ So we have to be careful that we not get too carried away with the label ‘Indian English’, and blithely proceed as though this describes an object that is sharply distinct from other labeled objects such as ‘Singapore English’ or ‘British English’. The boundaries that separate one variety from another are porous at best. To better appreciate this simple fact, we need only observe that even the label ‘Indian English’ is often a convenient cover term for a range of regional and social varieties, that sometimes go by names like ‘Marathi English’, ‘Hindustani English’, ‘Boxwallah English’, ‘Babu English’ and ‘Kitchen English’ (Schneider 2007: 168). And in fact,

The Study of ‘Indian English’: What is It Good for?

315

Baldauf (2004, in Schneider 2007: 171) has suggested that the ‘fastestgrowing’ variety in the country is a mixture of English and Hindi or ‘Hinglish’. It is therefore not inconceivable that what started out as two distinct varieties will, over time, converge to form a single variety as recurrent interactions lead speakers to routinize linguistic patterns. Auer (1999), for example, suggests that there is a unidirectional continuum from codeswitching to language mixing to fused lects. With codeswitching and language mixing, speakers still have the option of deciding whether or not to juxtapose elements from di¤erent varieties. With fused lects, speakers no longer have such an option; the drawing together of elements from erstwhile di¤erent varieties has stabilized to the point where it constitutes a grammatical obligation or constraint. Therefore, what is otherwise construed as a particular variety of language, that is, what is assumed to be a stable, highly systemic entity with clearly defined boundaries, is really better construed as a network of constructions that are more or less loosely connected. The interconnections between the constructions will inevitably change over time, in the light of different speakers’ ongoing social experiences. Even though speakers possess sets of constructions that are generally organized into language groupings, and there may be broad agreement across speakers about what constructions belong to which language groupings, it is obvious that no two speakers possess exactly the same constructional set. Furthermore, there will be constructions whose association with particular varieties of language may be open to dispute. Or put di¤erently, there may be constructions that speakers use regularly, but which speakers do not necessarily associate with a named variety until they are actually confronted with the need to make such an association. Here, the desire to associate particular constructions with a label such as ‘Indian English’ (or some other label) is inevitably informed by broader social understandings of what the label entails – which leads us once again to issues of political economy.

In closing The fact that there exists a group of speakers for whom the label ‘Indian English’ reflects a linguistic reality is a phenomenon that can be approached either via a form-related linguistics or in political economic terms. But unless the label is one that has been conjured up by an analyst with no

316 Lionel Wee regard for the group’s own beliefs, the name of a variety is also a reflection of speakers’ (growing, emerging) metalinguistic awareness that there are shared commonalities in their linguistic practices, or at the very least, expectations that such commonalities do/should exist. As an ingroup metalinguistic label, it very possibly orients the linguistic practices of these speakers towards each other, and towards those whom they might consider nonmembers of the group. E¤ects on grammar are plausible as how linguistic forms are manifested and how their meanings are interpreted gradually become adjusted or calibrated in the course of recurrent interactions. It therefore seems to me a wholly legitimate enterprise if one were to also attempt to elucidate on the inter-connections between a form-related linguistics and the surrounding political economic factors. But such an attempt needs to be undertaken with careful attention to the dynamic interplay between what the name of a variety ostensibly refers to and the language practices of the speakers themselves.

References Agha, Asif 2007 Auer, Peter 1999

Language and social relations. Cambridge: Cambridge University Press.

From codeswitching via language mixing to fused lects: Toward a dynamic typology of bilingual speech. International Journal of Bilingualism 3: 309–332. Baldauf, Scott 2004 A Hindi-English jumble, spoken by 350 million. The Christian Science Monitor, Nov. 23. Csmonitor.com/2004/1123/p01s 03-wosc.html. Bell, Alan 1984 Language style as audience design. Language in Society 13: 145–204. Blommaert, Jan 2001 The Asmara Declaration as a sociolinguistic problem: Reflections on scholarship and linguistic rights. Journal of Sociolinguistics 5/1: 131– 142. Brown, Roger and Gilman, A. 1960 The pronouns of power and solidarity. In T. A. Sebeok (ed) Style in language: 253–77. Cambridge, Mass: MIT Press. Coupland, Nikolas 2007 Style: Language variation and identity. Cambridge: Cambridge University Press.

The Study of ‘Indian English’: What is It Good for?

317

Croft, William 2001 Radical construction grammar. Oxford: Oxford University Press. Eckert, Penelope 2000 Linguistic Variation as Social Practice. Oxford: Blackwell. Fillmore, Charles J., Kay, Paul, and O’Connor, Mary Catherine 1988 Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64: 501–38. Giles, Howard, Coupland, Justine, and Coupland, Nikolas (eds) 1991 Contexts of accommodation: Developments in applied sociolinguistics. Cambridge: Cambridge University Press. Goldberg, Adele E. 1995 Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Goldberg, Adele E. 2007 Constructions at work: The nature of generalization in language. Oxford: Oxford University Press. Harris, Roy 1980 The language-makers. Ithaca, NY: Cornell University Press. Hopper, Paul J. 1998 Emergent grammar. In The new psychology of language, M. Tomasello (ed.), 155–75. Mahwah, NJ: Lawrence Erlbaum. Kay, Paul, and Fillmore, Charles J. 1999 Grammatical constructions and linguistic generalizations: The What’s X Doing Y? construction. Language 75: 1–33. Lightfoot, David 1991 How to set parameters:Arguments from language change. Cambridge, Mass: MIT Press. Makoni, Sinfree, & Pennycook, Alastair 2007 Disinventing and reconstituting languages. In Disinventing and reconstituting languages, Sinfree Makoni and Alastair Pennycook (eds.), 1–41. Clevedon: Multilingual Matters. Mu¨hlha¨usler, Peter 2000 Language planning and language ecology. Current Issues in Language Planning 1/3: 306–67. Pawley, A. and Snyder, Francis H. 1983 Two puzzles for linguistic theory. In Language and communication, Jack Richards and Richard Smith (eds), 192–226. New York: Longman. Schneider, Edgar W. 2007 Postcolonial English: Varieties of English around the world. Cambridge: Cambridge University Press.

318 Lionel Wee Singh, Rajendra 2007 The nature, structure, and status of Indian English. Annual Review of South Asian Languages and Linguistics 2007, Rajendra Singh (ed), 31–44. Berlin/New York: Mouton de Gruyter. Widdowson, Henry G. 1989 Knowledge of language and ability for use. Applied Linguistics 10: 128–37.

Appendices

Announcements

The Gyandeep Prize We are happy to re-announce the continued availability of this annual prize. It will be awarded to the most outstanding student contribution to ARSALL.

Housekeeping As it is still our intention to bring out future issues earlier than November/December of each year, potential contributors to ARSALL should get in touch with the editor as soon as possible. Our new deadlines are: November 1:

initial submission

March 1:

final versions of accepted papers (A4 and PDF).

Papers submitted after these deadlines will be processed, but only for a later issue. A paper initially submitted after November 1, 2009, for example, will be considered only for the 2010 issue of ARSALL. Potential contributors are encouraged to send their initial submissions on the A4 template for contributions to edited volumes, furnished by Mouton at their web-site: http://www.degruyter.com/cont/imp/mouton/moutonAuthors.cfm Authors of accepted contributions must, however, follow the Mouton style-sheet and furnish their articles as word files (on the appropriate template) and their PDF versions. We reserve the right not to process papers requiring unnecessary editorial work. We would also like potential contributors whose primary language is not English to have their initial submissions looked at by a competent writer of English.

Notes on Contributors

Tej K. Bhatia, Ph.D. (Illinois) is Professor of Linguistics at Syracuse University, Syracuse, U.S.A. He has published extensively in the domains of descriptive and applied linguistics (including sociolinguistics). [email protected] Shishir Bhattacharja, Ph.D. (Montre´al) is Associate Professor of French at the University of Dhaka. He was trained in Linguistics and Indology at the University of Sorbonne, Paris. His most recent books are Sanjanani Byakaran (1998), a collection of articles, written in Bengali, on generative syntax, and Word formation in Bengali: a Whole Word Morphological description and its theoretical implication (2007), an exhaustive morphological description of Bengali. [email protected] Probal Dasgupta, Ph.D. (New York University) has taught in Kolkata, Pune, Hyderabad and is now professor of linguistics at the Indian Statistical Institute, Kolkata. His best known books include Kathaar kriyaakarmo (1987), The otherness of English: India’s auntie tongue syndrome (1993), Primico (1977), and Projective syntax: theory and applications (1989). [email protected] Niladri Sekhar Dash, Ph.D. (Calcutta University) works at Linguistic Research Unit, Indian Statistical Institute, Kolkata. He has published several books and papers in various international and national journals. He has acted as visiting faculty in some Indian universities and research institutes, as a Co-investigator in the TDIL Project of the Govt. of India, as an Expert in ASI@IT&C project of European Commission, and as a Member in the Indo-African project of the International scientific research network. He is also a Consultant for Mobile Labs, Sweden. [email protected] Alice Davison, Ph.D. (Chicago) is Professor of Linguistics at the University of Iowa. She is interested in the interrelation of Chomskyan linguistic

324 Notes on Contributors theory and the particular structures and features of Hindi/Urdu and other South Asian languages. She has also taught at SUNY Stony Brook, University of Illinois, Champaign-Urbana and University of Wisconsin. [email protected] Ashwini Deo, Ph.D. (Stanford) is Assistant Professor of Linguistics, Yale University. Her main area of research is Historical linguistics with a focus on Indo-Aryan languages. [email protected] Hans Henrich Hock, Ph.D. (Yale) is Professor Emeritus of Linguistics and Sanskrit, The University of Illinois, Champagne. He is well-known for his work in Historical linguistics. He is the author of the Mouton text book on Historical linguistics and one of the editors of the Mouton series Trends in linguistics. [email protected] Ayesha Kidwai, Ph.D. (J.N.U.) teaches syntax and morphology at Jawaharlal Nehru University, New Delhi, and is the author of XP-adjunction in Universal Grammar: Scrambling and Binding in Hindi-Urdu (2000, Oxford University Press, New York). [email protected] Rajend Mesthrie, Ph.D. (Texas) is Professor of Linguistics at the University of Cape Town. His areas of interest include new varieties of English, pidgins and creoles as well as Indic dialects and Anglo-Saxon. He has published extensively on pidgins, Creoles, koines and language shift. His recent books include Introducing Sociolinguistics (Edinburgh University Press, 2000) and Language in South Africa (Cambridge University Press, 2002). [email protected] Annie Montaut is Professor of Hindi at INALCO, Paris, and member of the Center of studies on India and South Asia at EHESS/CNRS. She is the author of Hindi grammar (Lincom 2005) and editor of Les mots du discours (Cahier de Linguistique de l’INALCO, 2002) and Litte´rature and poe´tique en Asie du Sud (Purushartha 24, 2004). [email protected].

Notes on Contributors 325

Prashant Pardeshi, Ph.D. (Kobe University) is a lecturer in the Graduate School of Humanities, Kobe University. The focus of his research is on issues related to transitivity, voice and compound verbs. He is also interested in and has written about: linguistic typology, areal linguistics, Japanese linguistics, and pedagogy. [email protected] Anjum P. Saleemi (Ph.D. Essex) is Professor of Linguistics at Government College, Lahore. He has been engaged in work on linguistic theory, the acquisition and lernability of syntax, and the grammar of Hindi-Urdu, Punjabi and other South Asian languages. His publications include the book Universal Grammar and language learnability (Cambridge University Press, 1992), and several papers. Another major area of his interest is the philosophy of language and mind. [email protected] Ghanshyam Sharma, Ph.D. (Agra and Bologna) is on the faculty of Indian Studies at the University of Venice and has worked as adjunct at the University of Bologna. His research interests are in the fields of pragmatic theories of meaning, Hindi grammar, and forms of literary ennunciation in Hindi. [email protected] Joga Singh, Ph.D. (York, U.K.) is Professor of Linguistics and Punjabi Lexicography, Punjabi University, Patiala. He has published extensively on Punjabi and Hindi Syntax and presented several papers at national and international seminars/conferences. He frequently writes on language issues in the Punjabi media. He has also been associated as researcher/teacher with Osmania University, Hyderabad, Delhi University, and South Gujarat University, Surat. [email protected] Sanford B. Steever, Ph.D. (University of Chicago) is an independent scholar. He has studied Tamil and holds a Diploma in it. He has written extensively on Dravidian linguistics, including Comparative Dravidian, historical syntax and auxiliary verb systems. His recent books include The Dravidian languages (Routledge, editor) and The Tamil auxiliary verb system (Routledge). [email protected]

326 Notes on Contributors Lionel Wee, Ph.D. (Berkeley) is Associate Professor in the Department of English Language and Literature at the National University of Singapore. His research interests include metaphor, new varieties of English, Malay morphology, and language policy and planning. His publications include articles in Discourse and society, Journal of pragmatics, World Englishes, and English world-wide. [email protected]