Annual Review of South Asian Languages and Linguistics: 2011 9783110270655, 9783110270570

Citation preview

Annual Review of South Asian Languages and Linguistics 2011

Trends in Linguistics Studies and Monographs 241

Editor

Volker Gast Founding Editor

Werner Winter Editorial Board

Walter Bisang Hans Henrich Hock Heiko Narrog Matthias Schlesewsky Niina Ning Zhang Editor responsible for this volume

Hans Henrich Hock

De Gruyter Mouton

Annual Review of South Asian Languages and Linguistics 2011

edited by

Rajendra Singh Ghanshyam Sharma

De Gruyter Mouton

ISBN 978-3-11-027057-0 e-ISBN 978-3-11-027065-5 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. ” 2011 Walter de Gruyter GmbH & Co. KG, Berlin/Boston Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany. www.degruyter.com

Contents

Editorial Preface ....................................................................................... vii General Contributions Metatypy in Sri Lanka Malay ..................................................................... 3 Umberto Ansaldo Benglish verbs: A case of code-mixing in Bengali .................................. 17 Shishir Bhattacharja Agreement and non-finite verbs in Bangla: A biaxial approach .................................................................................... 35 Probal Dasgupta On the role of protases in conditional statements: Some evidence from Hindi ........................................................................ 49 Ghanshyam Sharma Special Contributions: The Indigenous Languages of South Asia Aspects of Kharia grammar: A Role and Reference Grammar approach ............................................... 81 John Peterson Grammatical voice in Gorum .................................................................. 125 Félix Rau Regional Reports India ......................................................................................................... 161 Pingali Sailaja

vi Contents Reviews M. Kaul and A. Aklujkar (Eds.) – Linguistic Traditions of Kashmir ...... 183 by Shishir Bhattacharja I. Hasnain and S. Chaudhary (Eds.) – Problematizing Language Studies. Cultural, Theoretical and Applied Perspectives. Essays in Honor of Rama Kant Agnihotri .............................................................................. 188 by Ghanshyam Sharma Appendices Announcements: The Gyandeep Prize/Housekeeping ............................ 207 Notes on Contributors ............................................................................. 209

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 fifth issue of ARSALL as part of the series Trends in Linguistics: Studies and Monographs. Each volume of this annual generally 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 generally invited single or multiple contributions on important issues, ranging from the narrowly grammatical to the wide-scope socio-linguistic/socio-political. When many contributions are included, the section will in effect 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 fifth, issue of ARSALL contains four contributions: Ansaldo’s Metatypy in Sri Lanka Malay, Bhattacharja’s Benglish Verbs: A Case of Code-mixing in Bengali, Dasgupta’s Agreement and Nonfinite Verbs in Bangla, and Sharma’s On the Role of Protases in Conditional Statements: Some Evidence from Hindi. The Special Contributions section is dedicated to the indigenous languages of South Asia. In their contributions to this section, Peterson and Rau carefully examine aspects of Kharia and Gorum. As India still 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 recent Indian research on South Asian languages in this issue of ARSALL. The Review section of this issue contains reviews of Hasnain and Chaudhary`s recent collection on language and the social and cultural matrices in which it is embedded and of Kaul and Aklujkar’s compendium on the linguistic traditions of Kashmir. We regret that this issue of ARSALL has no Dialogue section. Hopefully, we will have a substantial one next year. We are grateful to Prof. Hans Henrich Hock and Mr. Wolfgang Konwitschny for help and support in the preparation of this issue. Rajendra Singh and Ghanshyam Sharma

General Contributions

Metatypy in Sri Lanka Malay Umberto Ansaldo

This paper reviews the notion of metatypy in relation to the genesis of Sri Lanka Malay. Metatypy is a process of typological congruence known to occur in prolonged and intense situations of contact due to wide-spread multilingualism. In this process, contact-induced transfer of predominantly semantic categories leads to the evolution of grammatical patterns that emerge as syntactic compromises between the actual grammars in contact (Ross 2006). In this paper I show that metatypy efficiently explains the evolution of a new language – Sri Lanka Malay – as the outcome of the contact between Sinhala, Lankan Tamil and Trade Malay. The grammar of Sri Lanka Malay shows an extremely high degree of syntactic compromise, due to contact-induced transfer of semantic categories from Sinhala and Lankan Tamil in the everyday usage of Trade Malay. This is first and foremost illustrated in the nominal domain, where Sri Lanka Malay exhibits a typical South Asian case system, though the verbal domain also shows interesting metatypic effects in its Tense and Aspect categories. In explaining a majority of the features of Sri Lanka Malay grammar, metatypy emerges as a more convincing explanation than previous accounts of its genesis (Ansaldo 2009). In addition, it sheds light on the reasons for the evolution of a new language, relying on the principles of second/third language transfer and contact-induced cognitive compromise known to occur in metatypic settings. 1. Introduction This paper explores the notion of metatypy in relation to the restructuring that occurs in the case system of Sri Lanka Malay (SLM). The possibility of metatypic effects in SLM grammar was first suggested in Ansaldo (2005) and Bakker (2006) who assume typological convergence in the evolution of SLM. Here I first show that a narrow definition of metatypy as developed in Ross (2006) can indeed be applied to capture what happens in the evolution of SLM. I then proceed to spell out the implications this has

4

Umberto Ansaldo

for our understanding of why this happens in the history of a language. I conclude with some observations on the relationship between metatypic processes and language genesis. 2. The narrow definition of Metatypy Metatypy is a process of typological congruence known to occur in prolonged and intense situations of contact due to widespread bi- or multilingualism. In this process, contact-induced transfer of syntactic and semantic categories leads to the evolution of syntactic patterns that emerge as compromises between the actual grammars in contact. In recent work, Ross (2006, 2007, 2008) narrows down his original definition of metatypy (Ross 1996) to exclude lexical and grammatical calquing. These may be preconditions for metatypy, but not necessarily part of the process. Simply put, metatypy is a change in typological profile that occurs through language contact. In a bilingual – or for that matter multilingual context – speakers may restructure (aspects of) one of the languages in their repertoire using another language of that repertoire as a model. While it is most often the case that the emblematic language in a multilingual context undergoes metatypy under the effect of a dominant language, the opposite can occur, as illustrated by the case of Singlish (Ross 2006: 97, 2007: 131). Clear examples of metatypic process have been documented in a number of contexts. Ross (2007, 2008) illustrated this process in Takia, an Austronesian language that has undergone metatypy under influence of Papuan languages of Trans-New Guinea type. Other cases of metatypy have been found in in Tariana, an Arawak language restructured on the model of East Tucanoan languages (Aikhenvald 2002); Semitic languages of Ethiopia remodeled on Cushitic grammar (Harris and Campbell 1995); and Arvanitic, the Albanian spoken in Greece, which has shifted towards a Greek type (Sasse 1985). We no longer need to uphold a distinction between metatypy and typical processes involved in language genesis (Ross 2008). Singlish, Mixe Basque (Ross 2007: 124) and Maluku Malay (Bowden 2005) are all examples of recently formed contact languages with clear metatypic processes which show that metatypy is indeed involved in contact language formation. In what follows we will see that we can comfortably add the Malay of Sri Lanka to this list, which has shifted from Austronesian to Lankan type.

Metatypy in Sri Lanka Malay

5

Crucial to this paper is an understanding of exactly why metatypy should happen. Metatypy has been described as the result of a heavy psychological load that results from intense bilingualism requiring communication in distinct systems (Nadkarni 1975). The work of Sasse (1985, 1990) explains a number of typical outcomes of contact as the result of bilinguals’ needs to express the same thought in different languages. I am not subscribing to a general claim that all multilingual situations are inherently unstable and must be resolved through some process of shift, be it convergence, simplification or metatypy. But, as clearly evidenced by the existence of well-documented linguistic areas, speakers of different languages in prolonged and intense situations of contact can affect each other to the point that their linguistic system will become more similar to each other over time. This makes sense in the light of the overwhelming accommodating nature of human communication which results from a wider behavioral coordination characteristic of humans (Trudgill 2010). 3. Metatypy in SLM The claim that metatypic processes might be at play in the formation of SLM is not new. This was first suggested in Bakker (2006), followed up in Ansaldo (2005, 2008) and Ross (2008). I therefore here briefly review the basis for the claim and focus on explaining the implications of the facts. The most obvious illustration of metatypy in SLM is to be found in the nominal domain, in particular the case system (see Table 1). The emblematic language, a Malay variety variously referred to as Trade or Vehicular, more correctly a Pidgin-derived Malay variety following the typology of contact Malay varieties put forward in Adelaar (1996, 2005) has, in line with other Austronesian languages of the Malayic subgroup, no morphological marking of case on the Noun. The dominant languages in the multilingual feature pool of SLM, Sinhala and Lankan Tamil, both show typical case systems of the South Asian type. Due to the known congruence between these two languages, there is a certain degree of functional overlap in the two systems (Ansaldo 2009, 2010). What is crucial here is that SLM shows systematic restructuring of its NP to map the case systems of Lankan type, as shown in Table 1 below (synthesized from Ansaldo 2009: 129– 131)

6

Umberto Ansaldo

Table 1. Case in SLM (Sri Lankan Malay), Sinhala and Lankan Tamil.1 Abbr.: Experiential (Exp), Goal (G), Benefactive (Ben), Possessive (Poss), Tempo-rary possession (Temp. poss), Locative (Loc), Instrumental (Instr). CASE DATIVE NOMINATIVE ACCUSATIVE GENITIVE INSTRUMENTAL COMITATIVE

FUNCTIONS SLM Exp, G, Ben, Poss Agent Patient Possession Instr. Source Association

SINHALA Exp, G, Ben, Poss Agent Patient Temp.poss, Loc Instr. Source Association

TAMIL Exp, G, Ben, Poss Agent Patient Temp.poss, Loc Source Association

Worth noting for our purposes here are the following points: (1) The overwhelming typological restructuring that leads an original Malay variety without morphological case marking (typical for Malayic in general) to develop a case system of the South Asian type, with a dominant, multifunctional and obligatory Dative case and a weak Accusative case mostly used to mark definite objects (Ansaldo 2005, 2008, 2009). (2) The complete structural overlap in the first three core cases. (3) The particular structural parallel in the SLM and Sinhala InstrumentalAblative syncretism. (4) SLM case markers are reanalyzed PDM prepositions or verbs (Ansaldo 2009: 129). This overwhelming typological restructuring, I have argued (Ansaldo 2005, 2008, 2009, 2011), can be explained through the typological pressure that Sinhala and Tamil combined exercise over the emblematic original Malay variety, in a typical metatypic scenario. Furthermore, note that there are three noteworthy syntactic-semantic alignments happening here in the emblematic language: (a) General VO > OV shift; from Austronesian to Lankan word-order (b) Following (a) the grammaticalization of [Prep + N] into [N-Case Marker] (c) In parallel with (b) the semantic obligation to express (core) cases.

Metatypy in Sri Lanka Malay

7

It is not easy to claim that these changes must occur in the order above. It is highly plausible that (a) is the first step; word-order is extremely susceptible to change in contact situation and in the trilingual pool in which SLM evolves OV is dominant as it is present in both Sinhala and Lankan Tamil. Thus Malayic VO is overruled. It is also quite safe to say that (b) follows (a), since we know that [OV – N-PP] is a robust typological correlation. But this does not necessarily entail that (c) happens only after (b) is completed. In a sense, (c) has to co-occur with (b) if we want to explain how prepositions end up being reanalyzed as case markers. Stage (c) is, crucially, a semantic shift, one in which speakers reorganize cognitively in order to align themselves with a Lankan way of doing things, i.e. by obligatorily expressing a number of thematic roles on the Noun. In that sense, there must be an overlap of (b) and (c) at some point. It has been pointed out that this is a rather radical restructuring for a number of reasons (Ansaldo 2009): (i) (ii)

What we see is a case of development of morphology, i.e. an increase in morphological complexity not usually associated with contact-induced change (Ansaldo and Nordhoff 2008) The development of case systems is normally a complex gradual process that requires many generations of speakers.

Such radical restructuring leads us to believe that an overwhelming typological pressure must lie behind the changes, which suggests that trilingualism in Malay, Sinhala and Tamil, rather than bilingualism in Malay and one of the dominant languages, is indeed the correct scenario for the evolution of SLM. Another note-worthy issue is the fact that we do not see the same degree of systematic restructuring in the VP. There are various ways to explain this. 1. Ansaldo (2009) analyzes this situation in terms of type frequency. He notes that there is overwhelming overlap in the two dominant languages in the nominal domain; this creates a ‘gang-up’ effect which results in a nearly complete adherence of the Malay NP to the Lankan type. On the other hand, in the VP, the two dominant languages are more divergent, causing competition in the multilingual pool. This allows for a higher degree of retention of emblematic features. 2. Ross (2006) puts forward a strong hypothesis of metatypy, which implies that, when metatypy happens, it tends to run to completion throughout

8

Umberto Ansaldo

the grammar under restructuring. In his view, then, the fact that the SLM VP does not show full metatypic effect is simply a matter of time. Either it is happening, or it has happened but has not been detected yet. In other words, the onus is on us to find metatypy in the verbal domain. Indeed, as shown in Ansaldo (2009: 135), a change that can be seen in the SLM domain is that from a predominantly aspectual system to a mixed TenseAspect system, with a past-non-past distinction of the South Asian type. Be that as it may, what really matters here is the significance of metatypy for the genesis of SLM to which I now turn. 4. Theoretical implications Let us review the facts: – – –

SLM is undoubtedly an instance of metatypy, even if the process is not (yet) complete (see also Ross 2008: 153); Metatypy is a gradual, ongoing change; Metatypy occurs because of bi- or multilingual-induced pressure on the cognitive system.

In relation to the last point, the general idea is that bilinguals lead the way in this restructuring, modelling the grammar of the emblematic language on the more dominant L2 models. As argued in Ansaldo (2005, 2009), in the multilingual context of SLM the socially dominant language is Sinhala, which must be considered a major model for the metatypic change. In addition, the presence of Lankan Tamil in the context further strengthens, at least in the nominal domain, the Lankan type, and it is possibly this combined force that accounts for what appears to be a rather fast pace of restructuring.2 If it is the case that the strong version of metatypy is correct, we should expect the VP to soon be fully metatypified. If on the other hand issues of frequency play a role, the situation may not change. Other factors enter into this consideration, of course, for example the relative language loss and shift to English observed in a number of Malay communities of Sri Lanka (Ansaldo and Lim 2006; Lim and Ansaldo 2006, 2007). The study of SLM within a metatypic framework also confirms a number of recent observations on the nature of metatypy. It seems to be safer to say that metaypy is not an outcome of contact distinct from shift, pidginization and mixing (Ross 2007: 132), but rather that it cannot be clearly dis-

Metatypy in Sri Lanka Malay

9

tinguished from other processes of contact-induced change, and that it must be considered part of them (Ross 2008: 163). More importantly, this paper clearly shows that metatypy is involved in the genesis of new grammar, and can thus in principle not be separated from other processes of contact language formation. At this point it is worth considering what this approach implies for the nature of the bilingual process that underlies grammatical restructuring of this kind. It seems safe to suggest that metatypy falls, as a contact-induced process of change, under the general, and perhaps often too generic, notion of convergence, in particular radical convergence involving typological shift. The standard view on the evolution of SLM has so far mostly subscribed to an exceptionalist view of restructuring (Ansaldo and Matthews 2007), whereby individuals in multilingual ecologies fail to acquire a linguistic system in its entirety and end up with some kind of approximation. In particular, as discussed in Ansaldo (2009: 141), the tradition of the ‘Tamil bias’ constructs a scenario in which Malay-Tamil bilingual households create the conditions for imperfect transmission leading to the evolution of SLM (see Smith et al. 2004; Slomanson 2006). This tradition appears to fail on two counts: (i) the scanty historical evidence for widespread MalayTamil intermarriage, and (ii) the lack of structural evidence to corroborate significant Tamil influence vs. the abundant evidence of Sinhala influence in SLM (Nordhoff 2009). Having already thoroughly illustrated the weakness of the Tamil bias (but see section 5), I here want to focus on the positive lessons to be drawn from a metatypic approach. In many multilingual communities around the world, where multilingualism is not institutionally supported through schools, education etc., multilingual individuals may experience shifts within their multilingual competence For example, if one of the codes used becomes limited to the home domain, it typically grows weak, and interference from the more frequently used languages is expected. This essentially is language attrition and also shows the following characteristics (Thomason and Kaufman 1988): – – – –

There is gradual abandonment of ancestral language (AL). It is typical for minority groups under (a) colonization and (b) nation-expansion processes. It is typical of stigmatized linguistic codes. It happens in the transition from monolingualism in AL to multilingualisms in L2/L3.

10

Umberto Ansaldo

– –

–

L2/L3 take over domains of usage. The number of AL speakers gradually decreases and so does competence in AL. If AL has vitality, some features survive; if not, death occurs.

In the transition process described here, L2/3 dominant individuals lead the change through intense code-mixing, structural and lexical transfer in which features from the dominant languages interfere with features from the weaker code. As noted in Haugen (1978: 37), the speaker “dismantles and reorders the language he already knows”. One common motivation is the AL speakers’ increasing accommodation to L2/ 3 under strong social and structural pressure through increased multilingual competence (Ansaldo 2006, 2007; Ansaldo and Lim 2006). In rare contexts where AL is vital, a partial maintenance is indeed possible (e.g. Anglo-Romani, Thomason and Kaufman 1988), typically in cases where the change is gradual (Winford 2003). This is a most plausible scenario for the genesis of SLM, which does not require exceptionalist explanations but relies on well-established sociolinguistic observations. The SLM communities were and are characterized by a solid integration within the larger communities of Sinhala and Tamil speakers, as their high degree of trilingualism to this date shows. At the same time, however, they have a very strong sense of their own SLM identity, as testified by their vital cultural, grass-root organizations as well as the active engagement in their own linguistic practices, seeking to improve the status of their language (Ansaldo and Lim 2006; Lim and Ansaldo 2006, 2007). As pointed out in Matras (2010: 66), bilingual individuals should be seen as possessing an enriched linguistic system (rather tan two or more impoverished ones), which they are capable of appropriately adapting to the context in which they function. In this sense, language contact phenomena can be seen as function-drive choices that speakers make (not necessarily consciously) in goal-oriented communicative interaction. Convergence thus emerges as a process that “offers speakers the opportunity to accommodate and generalize and yet still hold on to a mental demarcation between subsets of word forms within their repertoire.” (Matras 2010: 76).

Metatypy in Sri Lanka Malay

11

5. Final remarks Nordhoff (2009: 49) offers a thorough and clear overview of the different theories of genesis put forward in the scholarship on SLM so far. I summarize his observations below: (a) Bakker’s (1995) rapid convergence hypothesis. The argument here is that SLM would have been created within two generations through a process of rapid convergence. There seems to be no clear historical or sociolinguistic evidence in support of this claim (Nordhoff 2009: 53– 54). There is, moreover, some counter-evidence in Ansaldo and Matthews (2007) who argue that case system is known to emerge gradually in language evolution. Whether one agrees with these objections or not, the mechanism involved in this view is one of convergence and thus not incompatible with the metatypic view proposed here. Note however that, as Ross makes it clear, metatypy occurs gradually, not abruptly. (b) Smith et al. (2004) Tamil substrate hypothesis. The claim here is that SLM is the product of relexification processes in bilingual TamilMalay households. Some putative historical evidence for this is presented in the work of Hussainmiya (1990), but it is disproved in Ansaldo (2008, 2009), based on historical records. Nordhoff (2009: 59– 60) shows that a number of features of SLM used in this hypothesis to claim stronger Tamil influence are actually incorrectly analyzed and can be derived from Sinhala, rather than Tamil influence. Finally, it is difficult to reconcile bilingualism as required by relexification with the obvious trilingual ecology in which SLM functions (Ansaldo 2009). (c) In my own work (Ansaldo 2005, 2008, 2009, 2011; Aboh and Ansaldo 2007; Ansaldo and Nordhoff 2009) I have argued that SLM evolves in a trilingual ecology and undergoes heavy typological restructuring due to the ganging-up of typologically similar Sinhala and Lankan Tamil. The effects of both languages are evident at the structural level in all domains of grammar; the nature of the ecology is confirmed by sociohistorical evidence. For these reasons it seems safe to suggest that metatypy in SLM makes historical and theoretical sense, being supported within typological, sociolinguistic and acquisitional frameworks. Though there is no serious counterevidence, as noted in Nordhoff (2009: 63–64), two aspects of SLM stand out: (i) there is clear Tamil evidence vs. very little Sinhala in the lexicon; (ii) there is clear

12

Umberto Ansaldo

Sinhala evidence vs. no Tamil influence in phonology. One possibility here is that the two adstrate languages may have played different roles at different times. According to the Founder Principle, influences that occur in the early days of contact are particular robust and survive in later stage of evolution (Mufwene 2001). In this view, we could postulate that initially Malay-Tamil contacts may have been more intense, and led to the lexical influence still noticeable today. Clearly though, over time Sinhala became the more dominant language in the trilingual ecology, and ended up dominating in phonology. Be that as it may, metatypy is a framework designed to capture grammatical restructuring due to multilingual competition at the cognitive level, and is not concerned with either lexical or phonological restructuring. These areas thus do not undermine the strong claim of metatypic effects presented here to account for the genesis of SLM. The hypothesis that Tamil might have played a significant – even dominant – role in the formation of SLM has been critized in Ansaldo (2008, 2009); the claim that SLM may be the product of bilingual Tamil-Malay interaction has been shown to be historically and grammatically incorrect, and is explained as the result of certain ideological bias I refer to as the ‘Tamil bias’ (Ansaldo 2008). Slomanson (2011: 385 fn.4) appears to take issue with the claim of a Tamil bias in SLM studies, but he falls short on a number of counts. Firstly, while criticizing archival data in support of a Tamil bias, he fails to provide archival evidence in favor of his claim that “Muslim evidence for highly frequent Malay – ‘Mloaish’ (Shonamspeaking) intermarriage is completely clear and unambiguous”. The claim is further weakened by the fact that he does not qualify the nature of obscure varieties such as Shonam or Mloaish which come across as somewhat ad hoc linguistic constructs and are thus impossible to evaluate. Thirdly, he incorrectly attributes claims relating to the role of Shonam to Ansaldo (2008). To the best of my knowledge, this variety had not been ‘uncovered’ at the time and could have not been referred to in any possible way. More important however is the fact that the upholders of the biased Tamil-Malay bilingual genesis scenario have so far not offered any convincing explanation for the abundant presence of Sinhala features in SLM shown in Ansaldo (2008, 2009, 2011) and even more robustly in Nordhoff (2009). A metatypic account of SLM genesis, on the other hand, has no problems with this aspect of SLM grammar. Finally, any claim of bilingual genesis seems to ignore the fact that, contemporary contact linguistics and creolsitic

Metatypy in Sri Lanka Malay

13

scholarship have welcomed the insight that language genesis is by and large a product of multilingualism, and not mere bilingualism, apart from very few marked cases (see Croft 2000; Mufwene 2011, 2008; Ansaldo 2009; Matras 2010). The bilingual genesis view assumed in the Tamil bias thus remains historically, sociolinguistically, structurally and theoretically unwarranted. Notes 1. 2.

Based on the variety of SLM spoken in Kirinda, South. For a more detailed account of case in the Upcountry variety of SLM (Kandy) see Nordhoff (2009: 583). Though this could be also due to the size and nature of the community, small and tight-knit (see Trudgill 2001).

References Adelaar, Alexander 1996 Contact languages in Indonesia and Malaysia other than Malay. In Atlas of languages of intercultural communication in the Pacific, Asia and the Americas, S. A. Wurm, P. Muhlhausler, and D. T. Tryon (eds.), 695–712. Berlin/ New York: Mouton de Gruyter. 2005 Structural diversity in the Malayic subgroup. In The Austronesian languages of South East Asia and Madagascar, Alexander K. Adelaar and Nicholaus P. Himmelmann (eds.), 202–26. (Routledge Language Family Series.) London: Routledge. Aikhenvald, Alexandra Y. 2002 Language contact in Amazonia. Oxford: Oxford University Press. Ansaldo, Umberto 2005 Typological admixture in Sri Lanka Malay: The case of Kirinda Java. Ms. University of Amsterdam. 2008 Revisiting Sri Lanka Malay: Genesis and classification. In A World of Many Voices: Lessons from Documenting Endangered Languages; A. Dwyer, D. Harrison and D. Rood (eds.), 13–42. (Studies in Language.) Amsterdam/Philadelphia: John Benjamins. 2009 Contact languages. Ecology and evolution in Asia. Cambridge: Cambridge University Press. 2011 Sri Lanka Malay and its Lankan adstrates. In Creoles, their substrates and language typology, C. Lefevbre (ed.), 367–82. Amsterdam/ Philadelphia: John Benjamins

14

Umberto Ansaldo

Ansaldo, Umberto, and Lisa Lim 2006 Globalisation, empowerment and the periphery: The Malays of Sri Lanka. Paper presented at Vital Voices: Endangered Languages and Multilingualism. Proceedings of the FEL X Conference. 39–46. Bath: Foundation for Endangered Languages & Mysore: Central Institute of Indian Languages. Ansaldo, Umberto, and Sebastian Nordhoff 2008 Complexity and the age of languages. In Complex processes in new languages. Creole Language Library, E. O. Aboh and N. Smith (eds.). 345–363. Amsterdam/ Philadelphia: John Benjamins. Bakker, Peter 2006 The Sri Lanka Sprachbund: The newcomers Portuguese and Malay. In Language Areas, Y. Matras, A. McMahon, and N. Vincent (eds.). 135–159. Houndsmill Basingstoke: Palgrave McMillan. Bowden, John 2005 Language contact and metatypic restructuring in the directional system of North Maluku Malay. Concentric: Studies in Linguistics 31: 133–158. Harris, Alice C., and Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Hussainmiya, B. A. 1990 Orang Rejimen: The Malays of the Ceylon Rifle Regiment. Kuala Lumpur: Universiti Kebangsan Malaysia. Lim, Lisa, and Umberto Ansaldo 2006 Keeping Kirinda vital: The endangerment-empowerment dilemma in the documentation of Sri Lanka Malay. Amsterdam Center for Language and Communication Working Papers 1, 51–66. 2007 Identity alignment in the multilingual space: The Malays of Sri Lanka. In Linguistic identity in multilingual postcolonial spaces, E. Anchimbe (ed.), 218–243. New Castle upon Tyne: Cambridge Scholars Publishing. Matras, Yaron 2010 Contact, convergence and tytpology. In Handbook of Language Contact, R. Hickey (ed.), 66–85. Malden, MA/ Oxford: WileyBlackwell. Mufwene, Salikoko S. 2001 The Ecology of Language Evolution. Cambridge: Cambridge University Press. Nadkarni, Mangesh V. 1975 Bilingualism and syntactic change in Konkani. Language 51: 672– 683.

Metatypy in Sri Lanka Malay

15

Nordhoff, Sebastian 2009 A grammar of Upcountry Sri Lanka Malay. University of Amsterdam: Landelijke Onderzoekschool Taalwetenschap. Ross, Malcolm 1996 Contact-induced change and the comparative method. Cases from Papua New Guinea. In The comparative method revisited. Regularity and irregularity in language change, Durie Mark and Ross Malcolm (eds.), 180–217. New York: Oxford University Press. 2006 Metatypy. In Encyclopedia of Language and Linguistics; Keith Brown (ed.), vol. 8: 95–99. Oxford: Elsevier. 2007 Calquing and Metatypy. Journal of Language Contact Thema 1: 116–143. 2008 A history of metatypy in the Bell languages. Journal of Language Contact Thema 2: 149–164 Sasse, Hans-Jurgen 1985 Sprachkontakt und Sprachwandel: die Grazisierung der albanischen Mundarten Griechenlands. Papiere zur Linguistik 32: 37–95. 1990 Language decay and contact-induced change: similarities and differences. Arbeitspapier (Institut für Sprachwissenschaft, Universität zu Köln) 12 (Neue Folge): 30–56. Slomanson, Peter 2006 Sri Lankan Malay morphosyntax. In Structure and Variation in Language Contact, Ana Deumert and Stephanie Durrleman (eds.), 135–158. (Creole Language Library 29.) Amsterdam: John Benjamins. 2011 Dravidian features in the Sri Lanka Malay verb. In Creoles, their substrates, and language typology, C. Lefebvre (ed.), 383–411. Amsterdam/ Philadelphia: John Benjamins. Smith, Ian, Scott Paauw, and B. A. Hussainmiya 2004 Sri Lanka Malay: The state of the art. Annual Review of South Asian Languages and Linguistics, 197–215. Trudgill, Peter 2001 Contact and simplification: Historical baggage and directionality in linguistic change. Linguistic Typology 5: 371–374. 2010 Investigations in Sociohistorical linguistics: Stories of colonization and contact. Cambridge: Cambridge University Press.

Benglish verbs: A case of code-mixing in Bengali* Shishir Bhattacharja

In this article, we show how grammar can account for Benglish verbs, a particular type of complex predicate, which are constituted of an English word and a Bengali verb (e.g. /EksiDenT kOra/ ‘to have an accident’, /in kOra/ ‘to get/come/put in’ or /kOnfuz kOra/ ‘to confuse’). We analyze these verbs in the light of a couple of models (Kageyama 1991, Lieber 1992, Matsumoto 1996) which claim that complex predicates are necessarily formed in syntax. However, Benglish verbs like /in kOra/ or /kOnfuz kOra/ are problematic for these approaches because it is unclear how a preposition (e.g. in or off) or a verb (e.g. confuse or justify) can appear as the argument of the verb /kOra/ ‘to do’ in an underlying syntactic structure. We claim that all Benglish verbs can be satisfactorily handled in Morphology in the light of Whole Word Morphology (elaborated in Ford et al. 1997 and Singh 2006).

1. Priliminaries As in many other languages (see Moravcsik 1975, 1978 and Wohlgemuth 2009 among others) there exists a particular type of complex predicates in Bengali constituted of two items, one chosen from among various categories of words: noun, verbal forms, adjective, preposition, adverb, onomatopoeic word, etc., and the other, a duly inflected verb. The first item is usually called a pole and the second one a vector. Complex predicates are generally put into two different groups on the basis of the syntactic category of their pole: (i) Compound verbs (1) that categorically involve a verb (usually a non-inflectional verbal form such as infinitive, participle, absolutive or past gerund), and (ii) Conjunct verbs (2) that involve categories other than the verb. The main characteristic of compound and conjunct verbs is that they must denote one single action.1 (1)

Rik eSe Rik having come ‘Rik has just come.’

poreche has fallen

18 (2)

Shishir Bhattacharja

Rik bajar kore Rik market does ‘Rik does shopping.’

As in other South Asian languages (see Butt 1995, 2010, Dasgupta 1977, 2003, Hook 1974, Masica [1976] 2005, Mohanan 1993, 1994, among others) there is a particular type of conjunct (3–9) and compound (10-11) verbs in Bengali in which the pole is an English word chosen from among various types of nouns (3–6), adverbs (7), adjectives (8), prepositions (9) and verbs (10-11) while the vector is chosen from among a closed set of Bengali verbs consisting mainly of /kOra/ ‘to do’, /hOwa/ ‘to be/to happen/to become’, /dewa/ ‘to give’, /newa/ ‘to take’ and a few others. In this article, we will call these particular instances of code-mixing Benglish verbs, and we will try to show how grammar can account for them. We presume that most of these verbs are used by a particular group of more or less urbanized Bengali speakers who are at different stages of bilingualism (in the sense of Singh and Backus 2000: 83).2 (3)

a. EksiDenT kOra accident do ‘to have an accident’ b. EksiDenT hOwa accident be ‘to have an accident’

(4)

ribhEnj newa revenge take ‘to take revenge’

(5)

grup kOra group do ‘to put (things/persons) in a group’

(6)

OfiS (or Ofis) kOra office do ‘to work in an office’

(7)

slow kOra slow do ‘to make slow’

Bengalish verbs

(8)

kOmpaTibOl hOwa compatible be ‘to be compatible’

(9)

in kOra in do ‘to get/come/put in’

19

(10) kOnfuz kOra confuse do ‘to confuse’ (11) jasTify kOra justify do ‘to justify’ We note that not all words can appear as pole with all vectors.3 For example, /newa/ and /dewa/ cannot appear as vector in Benglish compound verbs. Hence we can have /pripEar kora/ (prepare-do) and /priparEshOn newa/ (preparation-take) ‘to prepare’, but ?/priparEshOn kora/ (preparation-do) and */pripEar newa/ (prepare-take) are unacceptable (however, there is no such ban on Bengali compound verbs: kore newa (having donetake) ‘to have something done’ or kore dewa (having done-give) ‘to do something for somebody as a service’). As with any other simple or complex predicates in Bengali each Benglish verb has its own subcategorical features. For example, /EksiDEnT kOra/ (3a) and /EksiDEnT hOwa/ (3b), both denoting ‘to have an accident’ require their agent nouns to be casemarked differently, the former with nominative (marked with zero affix) (12), and the latter with genitive (13).4 (12) Kawsar EksiDEnT koreche Kawsar accident has done ‘Kawsar had an accident.’ (13) Kawsarer EksiDEnT Kawsar-GEN accident ‘Kawsar had an accident.’

hoyeche has been

In some cases, speakers can alternate between a Bengali conjunct verb (e.g. /obhijog kOra/ (14)) and its Benglish counterpart (e.g. /kOmplEin kOra/

20

Shishir Bhattacharja

(15)). However, many Benglish verbs (e.g. /ofish kOra/ (6) and /kOmpaTibOl hOwa/ (7)) do not have a verbatim Bengali conjunct verb counterpart. In some cases the English pole cannot alternate with its Bengali counterpart, for instance, /durghOTona/ cannot replace the pole accident in (3a) (*/durghOTona kora/, but /durghOTona hOwa/ is acceptable) although /accident/ is frequently used as one of the synonyms of /durghOTona/. These examples show that not all Benglish verbs can be obtained by simply replacing the Bengali pole (of a conjunct verb) with its English counterpart. (14) Rik Fahimer kache Gargir biruddhe obhijog Rik Fahim-GEN near Gargi-GEN against complain ‘Rik has complained to Fahim against Gargi.’

koreche did

(15) Rik Fahimer kache Gargir biruddhe kOmplEin Rik Fahim-GEN near Gargi-GEN against complain ‘Rik has complained to Fahim against Gargi.’

koreche did

This article is organized as follows. In section 2 we demonstrate how Benglish verbs can be handled in the light of different models of morphology of our time which claim that complex predicates must be handled in syntax, and then we point out some examples that would be problematic for theses approaches. In section 3, after a brief description of W(hole) W(ord) M(orphology), we try to demonstrate how this model can account for all types of Benglish verbs in morphology. In section 4 we discuss whether Benglish verbs are words or not, and finally, we draw conclusions.

2. Benglish verbs in the light of different models of morphology In this section we will analyze Benglish verbs in the light of a couple of models that have been used to account for word formation in other languages. In Lieber (1983, 1992 and 2004), compound formation is constrained by the A(rgument) L(inking) P(rinciple) according to which a verb or a preposition must be able to link its internal arguments which she (1983: 257) defines as follows: “all obligatory (i.e. lexically specified) arguments with the exception of the subject are internal.” For instance, the

Bengalish verbs

21

verbs /kOra/ ‘to do’ in (15) and /newa/ ‘to take’ in (16) link their theme /kOmplEin/ and /ribhEnj/ respectively. Equally, the predicates in (5) and (6) link their respective arguments. (16) Rik Gargir upOr ribhEnj niyeche revenge has taken Rik Gargi-GEN on ‘Rik has taken revenge on Gargi.’ It is unclear how ALP can be satisfied in (7) through (11) because in these examples the pole is represented by categories that cannot usually function as arguments of some predicate  adverb (7), adjective (8), preposition (9) and verb (10) and (11). If the nominal forms of the verbs in (10) and (11) appeared as poles, for instance, */kOnfuSOn/ ( Indonesian downloadin (25) English realize > Hungarian realiz-ál Whether Wohlgemuth’s LVA strategies are really different from each other is debatable. In the following section we will try to demonstrate that the whole phenomenon of LVA can be handled with one single strategy in the light of W(hole) W(ord) M(orphology). 3. Benglish verbs in the light of WWM In what follows, after a brief description of WWM (elaborated in Ford et al. 1997 and Singh 2006) we will try to demonstrate how Benglish verbs can be handled in this model. According to Singh (2006: 578):

24

Shishir Bhattacharja All that needs to be said about word structure in any language (of any type whatsoever) can and must be said by instantiations of the schema in (S1). These instantiations are referred to as Word Formation Strategies (WFSs) because, as generalizations drawn from known particular facts, they can be activated in the production and understanding of new words. WFSs must be formulated as generally as possible, but  and this is crucial  only as generally as the facts of the matter permit. S1. /X/a l /Xƍ/b where 1. /X/a and /X'/b are words and X and X' are abbreviations of the forms of classes of words belonging to categories a and b (with which specific words belonging to the right category can be unified or on to which they can be mapped). 2. ' represents (all the) form-related differences between /X/ and /X'/ that fall outside of automatic phonology. 3. a and b are categories that may be represented as feature bundles. 4. The l represents a bidirectional implication (if /X/ then /X'/, and if /X'/, then /X/). 5. The interpretation of /X/a is a semantic function of /X'/b and vice versa. 6. ' can be null iff D z E.

Singh (2006: 578) adds that S1 offers a unified account of what have sometimes been seen as different types of morphologies and encapsulates the rejection of multipartite analysis of words into ‘roots’, ‘affixes’, ‘stems’, and so on, entries that are hard to define and harder to tell apart.”

For WWM, words have no internal (non-phonological) hierarchical structure. According to this model (cf. Singh 2006: 578): Morphological complexity is a matter of analyzability (# segmentability) of a word into a variable and a constant component with respect to a WFS.

According to Singh (2006: 578) WWM sees ‘morphology’ “... not as a combinatorics of morphs or morphemes but as a system of generalized and abstract bidirectional correspondence among patterns instantiated by sets of whole words that exploit the same contrast.”

For example, on the basis of morphologically related sets of words like (26) and (27), one can obviously set up a WFS like (28). We note that (28) is licensed by a set of semantically related word-pairs that manifest the same (i) formal contrast: X/Xli on the one hand, and (ii) categorical affiliation: Noun/Adjective on the other. According to (28) bad, kind, famous etc. pro-

Bengalish verbs

25

vide the differing values for the variable X in (26) and (27) while the phonemic representation /li/, which remains constant throughout (27), provides the particular value of the prime (') in the schema (S1). (26) kind famous bad etc.

l l l

(27) kindly famously badly etc.

(28) /X/N l /Xli/ADJ We will now move onto demonstrate how WWM can account for Benglish verbs. We assume that the lexicon of a bilingual Bengali speaker contains sets of word-pairs constituted of English verbs like confuse or complain and Benglish verbs like /kOnfuzkOra/ or /kOmplEinkOra/ (29a-29b). Such pairs license (29) which can be used to form, analyze and retrieve other Benglish verbs like /EnalaizkOra/ ‘to analyze’ or /jasTifykOra/ ‘to justify’. Benglish verbs such as (9) can be obtained through (30). (29) a. b. c. d. (30)

/X/V, INF l /XkOra/V, INF /confuse/ l /kOnfuzkOra/ ‘to confuse’ /complain/ l /kOmplEinkOra/ ‘to complain’ /insult/ l /insalTkOra/ ‘to insult’ /group/ l /grupkOra/ ‘to put (things/persons) in a group’

/X/P l /XkOra/V, INF a. /in/ l /inkOra/ ‘to put (something) in’ b. /off/ l /OfkOra/ ‘to put (something) off’

We note that pairs manifesting the same formal difference but different categorical affiliations (e.g. (29–31)) must license different WFSs. WFSs also differ from each other on the question of semantic relatedness manifested in the pairs licensing them. For example, although the formal difference and categorial affiliations are the same in (31) and (32) they are different WFSs because they are licensed by pairs that manifest different semantic relatednesses.

26

Shishir Bhattacharja

(31) a. b. c. d. e. (32)

/X/N l /XkOra/V, INF /EksiDEnT/ ‘accident’ l /EksiDEnTkOra/ ‘to have an accident’ /insalT/ ‘insult’ l /insalTkOra/ ‘to insult’ /obhijog/ ‘complaint’ l /obhijogkOra/ ‘to complain’ /kOmplEin/ ‘complaint’ l /kOmplEinkOra/ ‘to complain’ /group/ l /grupkOra/ ‘to put (things/persons) in a group’ /X/N l /XkOra/V, INF /OfiS/ ‘office’ l /OfiSkOra/ ‘to work in some office’ /kOlEj/ ‘college’ l /kOlEjkOra/ ‘to attend some college’

The same Benglish verbs can be outputted with different WFSs. For instance, /kOmplEinkOra/ ‘to complain’ can be obtained with (29) and/or (31). Some Benglish verbs and their Bengali conjunct verb counterparts can be formed with the same WFS. Hence, Benglish /kOmplEinkOra/ and Bengali /obhijogkOra/ can be obtained by mapping the nouns /kOmplEin/ (< English Complaint) and /obhijog/ respectively onto (31).8 We also note that some of the left-hand inputs of (29) (e.g. insult, group) have double categorical affiliation they can be used either as a noun or as a verb in English and hence, they can be mapped onto (31) as well.9 We mentioned in section 2 that according to Wohlgemuth (2009) there are three different strategies of LVA. We claim that most of the cases of LVA mentioned in his voluminous work (about hundred languages belonging to about seventy language families) can be handled with relevant WFSs instantiating (S1). These ‘code-mixed’ words can be analyzed into the variable instantiated by loan words and the constant by sequences like kOra (kOmplEinkOra in Bengali) ((29) or (31)), en (downloaden in German) (23) in (downloadin in Indonesian) (24) or ál (realizal in Hungarian) (25) if we map them onto relevant WFSs. Finally, we claim that there is no need to treat Benglish verbs in syntax, or to list each one of them in the lexicon of individual speakers because they can be handled in morphology in the light of WWM. In this approach, the so-called vector verbs are supplied by relevant WFSs, and therefore, they do not need to be listed separately either. However, some Benglish verbs (but, not necessarily the same ones for all speakers) must be listed in different individual lexicons, which together with relevant English words form adequate pairs that license different WFSs which can be activated, in case of need, to form, retrieve or understand other Benglish verbs.

Bengalish verbs

27

4. Benglish verbs: are they words? As with many compound and conjunct verbs in Bengali, one may argue that Benglish verbs are not words but phrasal constructions because some of them (e.g. 34-35) lack formal cohesion in the sense that their pole and the vector part can be interrupted with some other words. (34) EksiDEnT Kawsar kOkhon koreche? accident Kawsar when did ‘When did Kawsar have an accident?’ (35) kEarful Rikke OboSSoi careful Rik-ACC certainly ‘Rik has to be careful.’

thakte remain

hObe will be

Linguists generally tend to agree that i) a large number of compounds derive from phrases, and ii) many affixes derive from words that used to appear in compounds (see Dressler 2006). We may divide this whole process of grammaticalization into four consecutive stages: (i) Loose compounds (see Dasgupta 2003 and Dressler 2006) > (ii) Tight or normal compounds > (iii) Affixoidal words > (iv) Affixal words. In the first stage, the two components of the compound in question may be interrupted by some other words, which is no more possible in the second stage. In the third stage, one of the components of the compounds undergoes phonological modification and becomes affixoid (see Booij 2004, Bauer 2005) (such as Africa>Afro, India>Indo, etc.) before finally becoming an affix (such as like>ly in English). In some dialects of Bengali (e.g. Chittagonian (36) and Kishoreganj dialect (37)) there are examples of compound verbs that have been fused into simple verbs. We claim, on the basis of examples like (36) and (37), that verbs that appear as vector in Benglish verbs (and also in other compound and conjunct verbs) have already stepped into the process of grammaticalization and are heading towards becoming something that are generally described as affixes (such as dom in kingdom and hood in boyhood which were regular words at some point of diachrony), which however does not mean that all vector verbs must cease to be used as regular verbs in the long run. (36) khai phalai (having eaten-I have thrown off) > khaialai “I have finished eating’

28

Shishir Bhattacharja

(37) khaia phalaichi (having eaten-I have thrown off) > khaialchi “I have finished eating’ We are aware of the fact that (36) and (37) are neither Benglish verbs nor words of the standard dialect of Bengali, with which we are concerned here. However, (38) and (39) point to fact that not all Benglish verbs can be interrupted, which means that some of them have already acquired some sort of formal integrity. It is possible that the Benglish verbs that lack cohesiveness are in their first stage of grammaticalisation and those which cannot be interrupted are in the second stage. Hence, although some Benglish verbs lack formal integrity it is likely that most of them will acquire it in the course of time. (38) *Rik kOnfuz bondhuder Rik confuse friend-ACC/PLU ‘Rik confuses his friends.’

kore does

(39) ?plEin ranoEte in kOkhon korbe? aeroplane runway-Loc in when will do ‘When will the aeroplane enter the runway? As with other compound and conjunct verbs, Benglish verbs are semantically opaque in the sense that they denote one single action. For example, /EksiDEnT kOra/ does not mean two simultaneous or consecutive actions such as ‘to have an accident’ and ‘to do something’. Equally /kOnfuz kOra/ either means i) ‘to mix up things (or persons) in the speaker’s mind which are otherwise distinct’, or ii) ‘to make someone else to mix up things (or persons)’, the agent performing in both cases nothing but one single ‘action of confusing’. In our view, semantic opacity is a more reliable and (probably) more universal criterion for word-hood as compared to formal integrity because some languages have been reported (Sadock 1998) to have words that allow insertion of lexical and grammatical elements. As Benglish verbs categorically lack semantic compositionality, they are more likely to be words and can be handled in morphology with WFSs à la WWM. Save a few exceptions (e.g. Grimshaw and Mester 1988), linguists generally agree that complex predicates are not phrases.10 If one takes this stand as valid, then Benglish verbs cannot be but words. However, one can also propose some intermediate category between phrase and word such as Word+ (word plus) (cf. Kageyama 2001), or, following Mohanan (1994), put words into different subcategories such as ‘morphological word’ (e.g.

Bengalish verbs

29

complex predicates) and ‘phonological word’. But, as long as the wordhood of the Benglish verbs is not seriously challenged we do not see why they should be a challenge for WWM. 5. Conclusions In this article we have studied Benglish verbs, a particular type of complex predicates in Bengali which are constituted of an English word and a Bengali verb, in the light of different models of morphology of our time. We have tried to demonstrate that some of these verbs are problematic for models that usually account for word formation in syntax and that all of them can be satisfactorily handled in morphology using the anlytical tools of WWM. Finally, it needs to be said that although we have used traditional terms like compound, compound verb, conjunct verb, complex predicate, pole, vector, light verbs, etc. throughout this article to facilitate discussion, there will, in fact, be no need of this if we adopt the WWM framework. Benglish verbs are verbs and like any other verb (or any other word) they can be formed, analyzed or retrieved with relevant WFSs that instantiate the schema: /X/a l /Xƍ/b, and therefore, neither they nor their subcomponents require different names.

Abbreviations = Accusative; ADJ = Adjective; AI = Abstract incorporation; ALP = Argument linking principle; CL = Classifier; GEN = Genitive; INF = Infinitive; LVA = Loan verb accommodation; N = Noun; P = Preposition; PLU = Plural; PRO = covert subject pronoun; TOP = Topicalizer; NOM = Nominative; V = Verb; VN = Verbal noun; WFS = Word Formation Strategy; WWM = Whole Word Morphology. ACC

Notes *

An earlier version of this article was presented at the 27th Pacific Asia conference on language, information and computation (PACLIC), 4-7 November, 2010, Tohoku University, Sendai, Japan, and at the 5th Bangladesh English Language Teachers’ Association (BELTA) International Conference, 20-22 May, 2011, Dhaka, Bangladesh. I am grateful to Professor

30

1.

2.

3.

Shishir Bhattacharja Peter Hook and anonymous reviewers of this paper for their comments and suggestions. I thank Professor Taro Kageyama, my post-doctorate research supervisor in Japan for the thought-provoking discussions I had with him concerning complex predicates and word formation. I thank my senior colleague Dr. Dil Afroze Quader who read several versions of this paper, gave comments, and tightened up my prose. Usual disclaimers apply. For financial support, I gratefully acknowledge the beneficence of the Japan Society for the Promotion of Science (JSPS) and University of Dhaka, Bangladesh, while for logistic support, I am indebted to the National Institute for Japanese Language and Linguistics, Tokyo, Japan. One of our reviewers mentions a major finding by Dasgupta (no year is mentioned, supposedly Dasgupta 1977) – ‘complex predicates may carry a serial verb interpretation goes unnoticed’ in our paper. S/he also states that our conclusion ‘CPs denote one single meaning’ is not true. Neither Dasgupta (1977) nor this writer has made any such claims or come to such conclusions. Dasgupta (1977) has shown that some complex predicates are serial verbs, we have tried to show that all compound and conjunct verbs must denote one single meaning. In fact, compound and conjunct verbs should not be confused with two other types of predicates which are also known to be complex: (i) Phrasal verbs (e.g. put out or lie down in English) and (ii) Serial verbs (in Bengali: potrika kine, phon kore, bajar niye hente chole asho (newspaperhaving bought-phone-having done, market-having taken-having walkedhaving moved-come) ‘buy the newspaper, make a phone call, do your shopping and then come on foot’). Unlike Compound and Conjunct verbs Serial verbs denote more than one action. Serial verbs and phrasal verbs are not the concerns of the present article. Singh and Backus (2000: 83) put bilinguals into two different groups: (i) Perfect or true bilinguals and (ii) less than perfect bilinguals. The latter are further divided into four subgroups: a. Very competent, b. Competent, c. Weak and d. Apparent. How these weak and apparent bilinguals become aware of the subcategorical features of English verbs is a question left unanswered in the pesent paper. One of the reviewers states that we must “look at the class of the verb that vector belongs to, e.g., unacccustives will have different argument taking properties than transitives. Within intransitives, internally caused verbs, nonvolitional changes of state verbs and existence, apperance and disappearance will show different behavior, further within transitives, one would expect considering change of state, manner of motion, inherently directed motion, externally caused emission, etc.” Dasgupta (1977: 78–79) also states that “there might be transitive stems which, when vectors, occur exclusively with tansitive poles. This phenomenon, ‘transitiviy harmony’, does indeed appear over a surprisingly wide range of data.” However, intransitive verbs like /aSa/ ‘come’, /boSa/ ‘sit’, /mOra/ ‘die’, etc. randomly appear as vector with transi-

Bengalish verbs

31

tive poles, such as /Suna/ ‘to listen/ hear’, /likha/ ‘to write’, /bOla/ ‘to say/speak/tell’: /Sune aSchi/ ‘we have been hearing’, /likhe boSbe/ ‘all of a sudden he will write (something that he should not have)’, /bole mOrchi/ ‘I continue to say in vain’. In our view, it is hard to propose ‘transitivity harmony’ (or any other type of harmony between the pole and the vector) as a general tendency in the formation of Bengali compound and conjunct verbs, and therefore, we do not see what can be gained by looking at the class of verb that the vectors and poles belong to. 4. According to one of our reviewers Benglish verbs with transitive verbs as vectors must function intransitively because, as s/he mentions, their pole which is in fact the internal argument of their vector is linked inside the compound. However, for another reviewer “It is extremely unlikely that they are all intransitive.” Be that as it may, in examples like /Daktar rogi OparESonkOreche/ (Doctor-patient-operation-did) ‘the doctor did an operation on the patient’ /Ami khata karekSon kOrechi/ (I-answer script-correction-did) ‘I have corrected the answer scripts’ Benglish verbs are used transitively. Similar examples abound in the language. 5. In examples like /tumi khabeTa ki/ (you-will eat-Ta-what) ‘What (the hell) will you eat?’ or /she geloTa kothay/ (she/he-went-Ta-where) ‘Where (the hell) has she gone!’ [Ta] is concatenated to a flexional verb. 6. Words like /kOnfuz/ ‘confuse’ cannot be considered as adjectives because sequences like */ek kOnfuz lok/ (a-confused-person) ‘a confused person’ or */uni kOnfuz/ (he-confused) ‘he is confused’ are not acceptable. One has to say instead /ek kOnfuzd lok/ and /uni kOnfuzd/ respectively. 7. According to Wohlgemuth (2009) light verb strategy is the second most frequently used strategy in world’s languages, and it can be found in languages spoken in all regions of the world and in most language families. 8. One of our reviewers has complained that our article “cannot take into account the extreme productivity of /kora/ compounds in Bangla, an indication that it is likely to be dealt satisfactorily in the syntax rather than morphology.” In our view, a WFS like (29) is productive because any verb can be mapped onto it. Equally, any noun can be mapped onto (31). The fact that the righthand outputs of these WFSs end in /kOra/ can be seen to be a coincidence. 9. English nouns like 'insult or 'import differ from the verbs in'sult and im'port as regards the placement of stress. However, /insalT/ is usually pronounced unstressed by bilingual Bengalis irrespective of whether it is a noun or a verb. 10. Grimshaw and Mester (1988: 213) claim that the formation of Japanese conjunct verbs can be “assimilated to a more general theory of phrasal constructions, which governs the behavior of idioms (like kick the bucket) and other lexical expressions that do not constitute single words.”

32

Shishir Bhattacharja

References Baker, Mark 1988 Incorporation: a Theory of Grammatical Function Changing. Chicago: The University of Chicago Press. Bauer, Laurie 2005 The Borderline between Derivation and Compounding. In Morphology and its Demarcations; W. U. Dressler, D. Kastovsky, Oslar E. Pfeiffer, and Franz Rainer (eds.), 97–107. Amsterdam: Benjamins. Booij, Geert 2004 Compounding and derivation, evidence from construction morphology. In Morphology and its Demarcations; Wolfgang U. Dressler, Dieter Kastovsky, Oslar E. Pfeiffer, and Franz Rainer (eds.), 109– 131. Amsterdam: John Benjamins Publishing Company. Butt, Miriam 1995 The Structure of Complex Predicate in Urdu. Stanford: CSLI. 2010 The Light Verb Jungle: Still Hacking Away. In Complex Predicates: Cross Linguistic Perspectives on Event Structure; Mengitsu Amberber, Brett Baker and Mark Harvey (eds.), 49–78. Cambridge: Cambridge University Press. Dasgupta, Probal 1977 The Internal Grammar of Compound Verbs in Bangla. Indian linguistics, 38, 68–77. 2003 Bangla. In The Indo-Aryan Languages; G. Cardona and D. Jain, (eds.), 351–390. London: Routeledge. Dressler, Wolfgang. U. 2006 Compound Types. In The Representation and Processing of Compound Words; Gary Libben, and Gonia Jarema (eds.), 23–44. New York: Oxford University Press. Ford, Alan, Rajendra Singh, and Gita Martohardjono 1997 Pace Panini, towards a Word-based Theory of Morphology. New York: Peter Lang. Grimshaw, Jane, and Armin Mester 1988 Light verbs and ș-marking. Linguistic Inquiry, 19 (2): 205–232. Hook, Peter Edwin 1974 The Compound Verb in Hindi. Michigan: Center for South and Southeast Asian Studies, The University of Michigan. Kageyama, Taro 1991 Light Verb Constructions and the Syntax-Morphology Interface. In Current English Linguistics in Japan; H. Nakazima (ed.), 169–203. Berlin: Mouton de Gruyter.

Bengalish verbs 2001

33

Word Plus: The Interaction of Words and Phrase. In Issues in Japanese Phonology and Morphology; J. V. D. Weijer and T. Nishihara (eds.), 245–270. Berlin: Mouton de Gruyter. Lieber, Rochelle 1983 Argument Linking and Compounds in English. Linguistic Inquiry 14 (2): 251–285. 1992 Deconstructing Morphology: Word Formation in Syntactic Theory. Chicago: The University of Chicago Press. 2004 Morphology and Lexical Semantics. Cambidge: Cambridge University Press. Masica, Colin P. 2005 Reprint. Defining a Linguistic Area: South Asia. Chicago: The University of Chicago Press. Original edition, 1976. Matsumoto, Yo 1996 Complex Predicates in Japanese, a Syntactic and Semantic Study of the Notion ‘Word’. California: CSLI Publications and Tokyo: Kurosio Publishers. Mohanan, Tara 1993 Verb Agreement in Complex Predicates in Hindi. In Complex Predicates in South Asian languages; M. K.Verma (ed.), 163–175. New Delhi: Manohar publishers. 1994 Argument Structure in Hindi. Stanford: CSLI. Moravcsik, Edith A. 1975 Verb Borrowing. Wiener Linguistiche Gazette, 8: 3–30. 1978 Language Contact. In Method and Theory, Universals of Human Languages, Vol. 1; J. Greenberg (ed.), 93–122. Stanford: Stanford University Press. Sadock, Jerald M. 1998 On the Autonomy of Compounding Morphology. In Morphology and its Relation to Phonology and Syntax; Steven G. Lapointe, Diane K. Brentari, and Patrick M. Farrel (eds.), 161–187. California: CSLI Publications. Singh, Rajendra 2006 Whole Word Morphology. Elsevier Encyclopedia of Linguistics, 2nd edition, 578–579. Singh Rajendra, and A. Backus 2000 Code-Mixing and Bilingual Proficiency. The Yearbook of South Asian Languages and Linguistics 2000, 61–92. Wohlgemuth, Jan 2009 A Typology of Verbal Borrowings. Berlin: Mouton de Gruyter.

Agreement and non-finite verbs in Bangla: A biaxial approach* Probal Dasgupta

The class of acceptable sequences of nonfinite verbs in Bangla becomes easier to characterize if we employ a notion of agreement that generalizes from its prototypical morphological exponence. The present study of these and related issues proposes a framework of syntactic analysis that develops substantivist assumptions in a biaxial direction – in the sense of combining syntagmatic and paradigmatic devices in its descriptive apparatus. 1. Methodological remarks This study brings some recent theoretical elaborations (Dasgupta 2011) to bear on earlier descriptions of syntactic linkages across clause boundaries (Dasgupta 2006). The context of the substantivist realignment of linguistic theory (Dasgupta, Ford, and Singh 2000; Dasgupta 2008a, 2008b, 2009b) is reshaped in Dasgupta (2011) in ways that lead us to reexamine such linkages and to retool syntax in a substantivist direction. One point of entry into the line of thought pursued here is the proposal, explored in parametric syntax and related frameworks, that ‘control clauses’ like to stay in John wanted to stay or like finding my comments offensive in Finding my comments offensive, John exploded have a phonologically null ‘control Pro’ subject. ‘Pro’ names a bundle of phi-features (Person, Number, Gender) for which the grammar provides either an arbitrary reading, ‘arbitrary Pro’, or coreference with a ‘controller’, here John. That mode of inquiry encodes the linkage of the finite verb (wanted or exploded) with the infinitive to stay or the participle finding by giving the non-finite verb a separate, atomic ‘Pro’ subject of its own. The shadow that John casts on the non-finite construction takes the form of an atomic element Pro ‘controlled’ by John. Within the assumptions of this atomizing style, consider the parallel between the syntax of ‘control’ and that of subject-verb agreement in a finite

36

Probal Dasgupta

clause like John exploded, formalized as John Agr Tense explode, specifying Tense as Past and Agr as a phi-feature bundle agreeing with – technically, controlled by – John. Why do these moves not lead atomistic work to describe the verb-object construction welcome them as welcome v they, with a little v feature-bundle controlled by the real verb welcome and incarnated as accusative case? Why does atomism not describe against them as against p they, with a little p embodied as oblique case and controlled by the real preposition against? Even atomists treat case differently from agreement, then, as a phenomenon, and from ‘control’, as a formal device. These remarks may not serve as a sufficient introduction to our mode of inquiry, which regards agreement as a fundamental feature of sentences and other freely assembled structures. For further motivation, consider some empirical material first presented in Dasgupta (2006: 148) and not yet annexed to the regime of any known formalist treatment of agreement: “The common feature is that relative phrases in Bangla may trigger first person agreement, as in sequential [(1)] or headed [(3)], or second person agreement, as in sequential [(2)] or headed [(4)]. This is surprising because in [(1)] and [(3)] we find first and second person agreement without a triggering first or second person pronoun anywhere; such a pronoun is in fact excluded there. [1]

jaaraa kambol aani ni taaraa sabaai kurxi who blankets brought haven’t they all twenty txaakaa debo rupees will-pay ‘Those of us who have not brought blankets will all pay Rs 20.’

[2]

jaaraa kambol aano ni taaraa sabaai kurxi who blankets brought haven’t they all twenty txaakaa debe rupees will-pay ‘Those of you who have not brought blankets will all pay Rs 20.’

[3]

aamraa sabaai kurxi txaakaa debo jaaraa kambol we all twenty rupees will pay who blankets aani ni brought haven’t ‘All of us will pay Rs 20 who haven’t brought blankets.’

Agreement and non-finite verbs in Bangla

[4]

37

tomraa sabaai kurxi txaakaa debe jaaraa kambol you all twenty rupees will pay who blankets aano ni brought haven’t ‘All of you will pay Rs 20 who haven’t brought blankets.’

It seems plausible that the feature traffic underpinning such ‘heteropersonal’ agreement crucially depends on the ‘sequent’ subsystem (instantiated by pronouns like taaraa ‘they’) as a cross-clausal mediator. Be that as it may, what we need is the result that even agreement with overt morphological exponence can operate across a finite clause boundary. The overall argument of Dasgupta (2006) is that the relative pronoun has a ‘Pro’-like character; if we construe ‘control’ as an extension of agreement, it follows that a relative clause formally agrees with matrix material. This line of thinking converges with theoretical considerations as follows. The grammar modules that rigorously operate with morphological, syntactic and discursive entities – call these modules Mu, Sigma, Delta, to distinguish them from their phenomenal counterparts whose boundaries are less well-defined – are so configured that agreement plays a role only in Sigma. It stands to reason that Mu should have no room for agreement, if agreement is merely a cementing device. The facts bear this out. Mu, even in its dealings with ‘compound words’ (see Singh and Dasgupta 1999 on why substantivist work uses scare-quotes here), encounters very few clear cases of ‘internal agreement marking’; French petite-fille [vs *petit-fille] for ‘granddaughter’ is far less typical, cross-linguistically, than French grandmère [vs *grande-mère] for ‘grandmother’. In striking contrast, examples of ‘internal case marking’ are far easier to find; Sanskrit gives us pituhxsxvasaa ‘father.GEN.sister’, maatuhxsxvasaa ‘mother.GEN.sister’, khecarahx ‘sky.LOC.dweller’, priyamxvadaa ‘pleasant.ACC.speaker’, etc. This makes sense if distinct case markers signal distinct roles. Word integrity, though it surely does render mere cementing devices pointless, need not mean that the signalling of distinct roles loses all pertinence. On the other side of Sigma, we would expect Delta to allow a whole range of cementing devices. Neighbouring sentences are indeed connected to each other by pronouns, anaphoric epithets, and various other deictic expressions, which lend ‘cohesion’ to a body of discourse. However, agreement has never been known to cross sentence boundaries in a discourse – as distinct from clause boundaries within a sentence.

38

Probal Dasgupta

In other words, agreement counts as one of Sigma’s signature-level traits. The present study explores the two roles – intraclausal and transclausal – that agreement plays in Sigma. To this end, we propose new, substantivist tools for grammatical inquiry; we then apply these to specific tasks in the description of Bangla, focusing on non-finite constructions. Earlier substantivist work – especially Dasgupta (2008b; 2011) – on the interplay between Saussurean arbitrariness and the free assembly property of propositions leads us to launch here a biaxial approach to syntactic inquiry that gives equal but distinct space to lexical arbitrariness and free assembly. We develop here some biaxial tools for the interfaces where syntax meets morphology and discourse; these prolegomena narrow the space from which second approximation studies will make device choices for the syntax module. Suppose we base Mu on ‘arbitrariness’ and Sigma on ‘free assembly’. Agreement becomes a strategic point, where arbitrary gender (invoking Mu) meets person/number values chosen by speakers freely assembling their sentences (invoking Sigma), and where the syntagmatic link between concord partners meets coreference, a paradigmatic relation. Thus the biaxial approach we propose here deploys devices co-invoking the syntagmatic and paradigmatic axes to explicate – with special reference to agreement – what the arbitrary and freely assembled aspects of linguistic structuration do and do not share. Taking Whole Word Morphology (WWM) as a point of departure, the biaxial approach populates the morphology-syntax interface with Binodal Dependency Declarations. The categorized variables at the two nodes of every Binodal must formally connect through a connective functor or feature binding. For concreteness, consider two Hindi-Urdu sentences with overt agreement and some relevant node pairs: (5)

lambaa laRkaa dauRegaa boy run-FUT.M.3SG tall-M.SG ‘The tall boy will run.’

(6)

lambii laRkii kitaabeM books tall-F.SG girl ‘The tall girl will buy books.’

(7)

lamb@ larxk@ ‘the tall boy/girl’ (@: a notation for aa/ii/e and their variants)

khariidegii buy-FUT.F.3SG

Agreement and non-finite verbs in Bangla

(8)

laRkaa dauRegaa boy run-FUT.M.3SG ‘the boy will run’

(9)

kitaabeM khariidegii books buy-FUT.F.3SG ‘will buy books’

39

Binodals are asymmetric. A head word as an independent x may lexically specify a slot for the syntax to fill; the head-initiated Binodal (11) formalizes such a dependency illustrated in (9). Or a dependant y can ask a head node for broad spectrum, lexically unspecified, structural acceptance; such a dependant-initiated Binodal (10) formalizes (7). In our no doubt insufficiently economical conventions, dependants precede heads. In statement (10), y-dash states that A depends on N, as in (7). The dash-x notation in (11) says that (9)’s V governs N. Angled brackets indicate bound features, either intrinsic phi-features like gender or extrinsic kappa-features like kasus (case). Square brackets [ ] enclose or follow the head’s label to show whether the Binodal participant is the head [H] qua lexical entry or the syntactic category H[ ]. Similar fine-tuning of the dependant’s label, not attempted here, may prove necessary. (10) A y— N[ ] (11) N —x [V] Such Binodals arrange syntagmatic head-dependant links and address paradigmatic questions of non/coreference. V in Binodal (11) binds N’s kappa-feature accusative, whose paradigmatic identification serves to link the V lexeme syntagmatically to its object. But (10) has the adjective’s phifeatures bound by the nominal node, not the particular nominal lexeme (to cut a few expository corners). A third option emerges in Binodal (12), which deals with predication dependencies like (8): (12) N —x V.T[ ] Binodal (12) contrasts not just with (10), where the lexical head [V] invites dependant N, but with (11), where dependant y moves in on N[ __ ] without reference to the lexeme occupying [ __ ]. V specified for Tense heads (12); the dot in V.T says that no matrix-internal boundary separates

40

Probal Dasgupta

verbhood from tense features. [ ] is blank as neither ‘run’ nor ‘Future’ nor the word ‘will.run’ does the inviting. The abstract tense-bearer initiates the Binodal and heads the dependency. Comparison with familiar frameworks may help readers to follow biaxial proposals. Call A in (10) an adjunct, then, N in (11) a complement, and N in (12) a specifier, but keeping the limitations of cross-framework translations in view. At the level of analysis, the biaxial approach can distinguish specifiers of T from specifiers of V, if desired, at (13), where V feature-binds the N’s agent theta role feature (or a version where a little v binds the feature). One may relate (13) to (12) representationally or derivationally; both choices are biaxiality-compatible. We must choose whether to have V.T’s phi-features bound separately, (14), or to co-specify nominative binding and phi-agreement, (15): (13) N —x V[ ] (14) N y–– V.T[ ] (15) N ––x V.T[ ] Actual choices among formally admissible options will require argumentation, as in any framework. Our interest here is in properties of the modules, not in mechanism choice. These Binodals at the Mu-Sigma interface are tools enabling the study of how Mu and Sigma’s respective signatures – arbitrariness and free assembly – play out in the grammar. Free assembly can be either syntagmatic or biaxial. Sigma’s Mucontiguous phrasal constructions are assembled syntagmatically, leaving paradigmatic linkage to Sigma-Mu interface devices like Binodals. At the other end Sigma meets Delta, which in its own realm involves free juxtaposition. Distinct sentences in a discourse are juxtaposed, not syntagmatically composed; paradigmatic properties like non/coreference associate them with each other, implementing cohesion. We propose that Sigma assembles Delta-contiguous predicational constructions on the syntagmatic and paradigmatic axes simultaneously, keeping the Sigma-Delta interface in view. On this view of the interplay of morphological and discursive factors with the syntax, agreement becomes formally coextensive with predicational constructions.1 We propose the following division of labour in Sigma, where ‘juxtaposition’ refers to Juxtapose Alpha (Dasgupta 2011):

Agreement and non-finite verbs in Bangla

(16)

41

(a) Phrasal constructions per se are assembled only syntagmatically, but open paradigmatic windows at binodal dependency points. (b) The basic predicational construction, the simple finite sentence, is assembled biaxially: the axes meet at subject-predicate agreement. (c) Adjacent sentences are in parataxis; what Sigma particularizes as coordination functors is paratactic juxtaposition, Delta’s composition baseline. (d) Subordination, a form of predication, joins two clauses through agreement procedures managed in the complementizer zone of the lower clause.

The discussion of poetry in Dasgupta (2011) could of course be forced into a straitjacket, subsuming the anchor-kite relation under some ‘generalized agreement’. But we take it that agreement and predication are in fact specific to Sigma; poetry’s use of discursive resources follows a different route. The Delta-Sigma interface takes Sigma’s agreement-linked headdependant pairs and juxtaposes them with various paradigmatic neighbours. Delta’s interface with the poetry module Pi involves treating anchor-kite pairs in a similar fashion, and further mechanisms reflecting the specificity of versification. These, we must add, stem from major subsystems of Delta itself. Delta is about hypertext information and invokes a whole range of paradigmatic coaxialities. Coreference is not even half the story. Quotation, allusion, mention, or such tropes as irony cannot be reduced to effects of indexing in index theories or of chain formation in chain theories, and abound in ordinary conversation.2 Delta requires more than ‘purely linguistic’ tools of inquiry if by this one means a slightly augmented form of ‘sentence grammar’. As cognitive science advances, the coarticulation of pragmatics and semiotics with grammar in this domain will become clearer, at which point workers in these fields can decide whether to place Delta within or beyond the borders of ‘linguistics’. 2. Descriptive proposals We turn to some empirical matters that we are hoping these tools will help illuminate. Consider the Bangla gerundials shown in (1) and (3) and infinitivals shown in (2) and (4):

42

Probal Dasgupta

(1)

dilip pharaasi sekhaano suru koreche Dilip French teaching begin AUX ‘Dilip has begun teaching French.’

(2)

dilip pharaasi sekhaate suru koreche Dilip French to.teach begin AUX ‘Dilip has begun to teach French.’

(3) * dilip kaauke pharaasi sekhaano suru karaa Dilip anybody French teaching begin AUX.ING pichiye deoaa pachondo kare naa NEG off putting like AUX *‘Dilip doesn’t like postponing beginning teaching anybody French.’ (4)

?dilip kaauke pharaasi sekhaate suru korte deri korte Dilip anybody French to teach begin to AUX delay to AUX bhalobaase naa NEG likes ‘Dilip doesn’t like to take for ever to begin to teach anybody French.’

Self-embedding the infinitival construction is relatively straightforward, as (4) illustrates, while a self-embedded gerund crashes,3 as we see at (3), for reasons that await clarification. At first sight, the gerundial/infinitival problem looks unrelated to agreement. But recall that we have decided to assimilate control to agreement. Suppose we revisit, with reference to Bangla data, the cross-linguistically familiar contrast between gerunds and infinitivals seen in the control examples (5)-(8) adapted from Dasgupta (2006: 163).

(5)

(6)

projitaa bitxkel posaak paraa pachondo karen Projita bizarre clothes wearing like AUX ‘Projitai doesn’t like wearingi/arb bizarre clothes.’

NEG

naa

projitaa bitxkel posaak porte pachondo karen Projita bizarre clothes to wear like AUX ‘Projitai doesn’t like to weari/*arb bizarre clothes.’

NEG

naa

Agreement and non-finite verbs in Bangla

(7)

(8)

bitxkel posaak paraa projitaar bhaalo laage bizarre clothes wearing Projita-GEN good feels ‘Projitai doesn’t like wearingi/arb bizarre clothes.’

NEG

bitxkel posaak porte projitaar bhaalo laage bizarre clothes to wear Projita-GEN good feels ‘Projitai doesn’t like to.wear i/*arb bizarre clothes.’

NEG

43

naa

naa

Cutting across the duality of ordinary sentences like (5) and (6) and experiencer subject sentences like (7) and (8), we find that the gerundials in (5) and (7) permit either subject sharing with the matrix verb (this is notated as subscript i) or what we shall call an implicit arbitrary subject (here arb).4 But the infinitivals shown in (6) and (8) permit only subject sharing, never the arbitrary option. Do the ideas of section 1 underwrite an enlightening account of (5)–(8) that helps us to understand (1)–(4)? We propose that a gerundial construction as in (5) and (7) is a syntagmatically assembled phrase while the infinitivals of (6) and (8) are predications, ergo biaxial. Agreement is then a core phenomenon for infinitivals and costlessly iterable, as in (4). In contrast, getting a phrase to masquerade as a predication is hard work; iterating the operation is even harder; too many iterations lead to unintelligibility, as in (3). To unpack this claim formally would involve fashioning substantivist tools for Sigma proper right here – not a task we have space for in this study. To give readers a sense of what is intended in these remarks about ‘strain’, we shall briefly comment on another case of (non-)iterability in the grammar, located in the Mu module. Dasgupta (2009a: 24–27) argued that the general non-iterability of Word Formation Strategies – observed in the contrast between well-formed Asianist ‘a scholar in Asian studies’ and illformed *Buddhistist ‘a scholar in Buddhist studies’, for instance – was a consequence of the formal structure of strategies. While the existence of some specific iterabilities, costly and exceptional, was informally noted (2009a: 37), that study did not present the escape hatch exempting them from the ‘strategy shadow theorem’. The answer, as some readers may have discovered, is that a word like great-grandfather underwrites the formation of great-great-grandfather and beyond by means of an iteration-enabling supplementary strategy [greitX]N ÅÆ [greitgreitX]N. Since each strategy that exceptionally permits iteration requires the formulation of a separate supplementary strategy (Whole Word Morphology prohibits two-variable schemata), the formalism adequately reflects the extra cost of permitting

44

Probal Dasgupta

such exceptions. Our remarks about costly iterations in Sigma are to be construed along the lines of this far more concrete formulation with respect to Mu. To return to our empirical account, it is possible that skeptics will find our evidence insufficient to motivate our conclusions even if other frameworks happen not to have dealt with the material we adduce. We shall therefore take our argument one step further. Consider the following data:

(9)

chobitxaa dekhle aami khusi habo picture-CLA see-COND I happy will be ‘I will be happy if I/others see the picture.’

(10) chobitxaa

dekhle khusi hole aami I picture-CLA see-COND happy be-COND majaa pele comke jeo naa NEG surprised AUX.COND startled AUX ‘Don’t be startled if I’m amused if others are (/*I am) happy to see the picture.’

In Bangla the conditional participle, unlike the infinitival, licenses nominative agent subjects, and correspondingly5 permits both subject sharing and arbitrary subject interpretations. Hence the two readings at (9). But on our terms this must be a case of predication, exhibiting syntactic agreement without overt morphological exponence. Why then is the iterated agreement reading ill-formed in (10), where repeated cross-clausal agreement (subject sharing) is required in order to yield the ‘I am’ reading – which is, in fact, crashingly bad? Our stand is that conditional participials, though predicational, should behave precisely like gerundials on this score. Having to choose at each interclausal juncture between agreement and an arbitrary6 implicit subject, repeatedly, causes overload both at (3) and at (10). Readers who find puzzling our disregard for what they take to be the boundary separating grammar from psycholinguistics may find Ford, Singh, and Martohardjono (1997: 109n1) to be a useful reference. These relatively straightforward problems were useful at the stage of introducing our tools and their uses. We turn now to a difficult problem. Some background facts need to be provided first. As van der Wurff (1989) notes, conditional participles like khele ‘if (I/we, you, they) eat’ license nominative agent subjects, while conjunctive participles like kheye

Agreement and non-finite verbs in Bangla

45

‘having eaten’ do not; the former need not share a subject, but the latter type goes in for a subject sharing arrangement that van der Wurff describes in detail. Briefly, if the superordinate verb takes a nominative agent subject, then a conjunctive participle dependent on it must have no independent nominative subject of its own. When the superordinate verb has an overt or implicit experiencer subject, a dependent conjunctive can license an unaccusative subject, as in (11), where ‘fall-PcP’ shares no subject with ‘was’:

(11) gaach

theke eatogulo aam porxe (tomaar) kono tree from so many mangoes fall-CJV (you-GEN) any laabh holo naa profit was not ‘So many mangoes having fallen from the tree, it was no use (to you).’

In this context, we inquire whether conjunctive or conditional participle status is compatible with the properties of complex constructions like the passive or the obligational. It turns out that a passive or obligational verbal complex can appear in the conditional participial form, (12) and (13), but not in the conjunctive, (14) and (15):

(12) dilipke

dillite paatxhaano hole silpaa kintu Dilip-ACC Delhi-LOC send AUX.COND Shilpaa however muskile porxbe problem-LOC AUX ‘If Dilip is sent to Delhi, however, Shilpaa will have problems.’

(13) pritike

ceaxcaate hole silpaa kintu Priti-ACC to shout AUX.COND Shilpaa however muskile porxbe problem-LOC AUX ‘If Priti has to shout, however, Shilpaa will have problems.’

(14) *dilipke

dillite paatxhaano hoye (tomaar) kono (you.GEN) any Dilip.ACC Delhi.LOC send AUX.CJV laabh habe naa profit will be NEG ‘Dilip having been sent to Delhi, it will be of no use (to you).’

46

Probal Dasgupta

(15) *pritike

ceaxcaate hoye (tomaar) kono laabh habe Priti-ACC to shout AUX.CJV (you-GEN) any profit will be ‘Priti having had to shout, it will be of no use (to you).’

naa NEG

Standard syntactic frameworks provide no means for stating a generalization covering these facts. Holding the conditional participle’s ability to license a nominative agent subject responsible does not work in the first round: passives and obligationals cannot take such subjects. Let us make a second round effort and propose that the ability to license such a subject for simplex verbs correlates with eligibility to appear in the conjunctive form. That proposal fails when we consider another non-finite form, the circumstantial participle, which can license a nominative agent subject for simplex verbs, (16), but whose inflectional features are incompatible with a passive or obligational complex, (17) and (18):

(16) oraa dilipke

eato paesaa so much money

deoaae give-CRC

silpaa Shilpaa

they Dilip-ACC abaak hoyeche surprised is ‘Shilpaa is surprised at their giving Dilip so much money.’

(17) *dilipke

dillite paatxhaano haoaae (tomaar) kono (you-GEN) any Dilip-ACC Delhi-LOC send AUX.CRC laabh habe naa profit will be NEG ‘On Dilip being sent to Delhi, it will be of no use (to you).’

(18) *pritike

ceaxcaate haoaae (tomaar) kono laabh Priti-ACC to shout AUX.CRC (you-GEN) any profit habe naa will be NEG ‘On Priti having to shout, it will be of no use (to you).’

At this point, formalist syntaxes run out of resources. The tools of section 1, however, enable a move or two. Circumstantial participles are locative-inflected gerunds, and license nominatives known to be quirky and to require stipulations (Dasgupta 1994: 62–5). Gerunds (see above) are syntagmatically assembled constructions on which the grammar can, with dif-

Agreement and non-finite verbs in Bangla

47

ficulty, impose some predicational properties. The generalization we seek, then, is:

(19) The passive and the obligational feature bundles are compatible only with a participle bundle capable of choosing whether to agree (in the subject sharing mode) with a superordinate verb. It is possible to conclude that a syntax accountable to cognitive science can be built – and is consistent with our current understanding of the interface at which poetry meets other modules of linguistic structure. Notes *

1.

2.

3. 4.

5.

6.

Support from the Esperantic Studies Foundation for research on substantivist linguistics and from the Department of Science and Technology (Government of India) for the ‘Language and Brain Organization in Normative Multilingualism’ project is gratefully acknowledged. In that project we intend to study common sense as encoded in proverbs; Dasgupta (2011) and the present study prepare for this goal by building and testing crucial tools not otherwise available. Although the limited distribution of morphologically overt agreement denies it such a fundamental role, its independent strategic importance in the realtime study of production and comprehension (Chandra 2011) corresponds to that of syntactic agreement, which it betokens. That many linguists are able to believe that poetry is about special effects invented by artists shows that the ordinary properties of discourse have barely begun to register, conceivably because the overexamined syntagmatic axis has eclipsed paradigmatic inquiry for decades. Speakers vary as to how many iterations it takes to bring about a crash. We are cutting corners by confining ourselves to arb alone, for the implicit subject can also be a distinct individual salient in the discourse; adopting a pickwickian sense of arb that includes that option leaves our argument unaffected and saves space. For a thorough description that profiles adverbial participles and other nonfinite constructions in Bangla, focusing on case marking as well as subject coreference options, see Bagchi (1993). See note 4.

48

Probal Dasgupta

References Bagchi, Tista 1993 Clausal Subordination in Bangla: A Cross-modular Approach. Ph. D. diss., Department of Linguistics, University of Chicago. Chandra, Pritha 2011 Redundant agreement, phases, and memory. In Language-Cognition Interface: State of the Art, Ramesh Kumar Mishra and Narayanan Srinivasan (eds.), 256–71. Muenchen: Lincom Europa. Dasgupta, Probal 1994 Verbless and nonfinite clauses in Bangla. PILC Journal of Dravidic Studies 4 (1): 57–67. 2006 Unifying relativization and control in Bangla. In Nyaya-Vasistha: Felicitation Volume of Professor V.N. Jha, Manabendu Banerjee et al. (eds.), 138–70. Kolkata: Sanskrit Pustak Bhandar. 2008a Knowledge and language in classical Indian linguistics: some observations. In Forms of Knowledge in India: Critical Revaluations, Suresh Rawal et al. (eds.), 89–104. New Delhi: Pencraft. 2008b Transparency and arbitrariness in natural language. In Annual Review of South Asian Languages and Linguistics 2008, Rajendra Singh (ed.), 3–19. Berlin/ New York: Mouton de Gruyter. 2009a Strategies and their shadows. In Annual Review of South Asian Languages and Linguistics 2009, Rajendra Singh (ed.), 3–40. Berlin/ New York: Mouton de Gruyter. 2009b Corepresentation of linguistic structures. Unpublished manuscript, Indian Statistical Institute, Kolkata. 2011 Compositionality, the prose default and poetry: A cognitive approach. International Journal of Mind, Brain and Cognition 1 (2). Dasgupta, Probal, Alan Ford, and Rajendra Singh 2000 After Etymology: Towards a Substantivist Linguistics. Muenchen: Lincom Europa. Ford, Alan, Rajendra Singh, and Gita Martohardjono 1997 Pace Panini: Towards a Word-Based Theory of Morphology. New York: Peter Lang. Singh, Rajendra, and Probal Dasgupta 1999 On so-called compounds. In The Yearbook of South Asian Languages and Linguistics 1999, Rajendra Singh et al. (eds.), 265–75. New Delhi: Sage Publications. Van der Wurff, Wim 1989 The syntax of participial adjuncts in Eastern Bengali. Journal of Linguistics 25: 373–416.

On the role of protases in conditional statements: Some evidence from Hindi* Ghanshyam Sharma

Providing evidence from Hindi, the present study determines the crucial role played by the protasis (or antecedent or P-clause) in the actualization of conditional constructions. It maintains that tense-aspect-mood elements of the protasis establish different degrees of hypotheticality in a conditional statement. The paper questions some widespread misunderstandings about the clause ordering in conditional constructions, namely that conditionals can exhibit both the protasis-apodosis (P-Q) and apodosis-protasis (Q-P) orderings. By examining data from Hindi and making reference to data from other languages, the paper reiterates our earlier claim that protasisapodosis ordering in conditional constructions cannot be reversed and advances a new idea that a ‘conditional string’ is the defining characteristic of a conditional construction. In other words, it is either the proposition contained in the apodosis – not the entire apodosis – which is preposed by the speaker to accomplish different communicative tasks, or the entire ‘conditional string’ which is added to a non-conditional statement to give it conditional dimensions. 1. Overview Despite being a most extensively studied subject, conditionals still remain a hotly debated topic – both in philosophy and linguistics – largely due to the peculiarity of these constructions. The present paper looks into the crucial role played by the protasis (i.e. P-clause, antecedent, subordinate clause or dependent clause)1 in the overall structure of conditional constructions. It discusses certain evidence from Hindi to determine the role played by the tense-aspect-mood (TAM hereafter) elements in the classification of conditional statements. In addition, it makes yet again an attempt to reiterate that protasis-apodosis (i.e. P-Q) ordering is the only possible clause order in natural languages. Thus, the paper rejects the idea that conditionality may

50

Ghanshyam Sharma

be expressed through apodosis-protasis ordering. As an additional point, the paper puts forward a new theory which – however bizarre it might seem at first glance – may have some implications for different brands of syntactic analyses. The basic idea is that in a conditional statement the conditionality is introduced exclusively by a conditional string which stretches from the protasis marker, if in English, up through what is generally considered to be an apodosis marker, i.e. then or [Ø], in English. This conditional string, in my view, is the sole factor in the actualization of a conditional statement, to which can be attached any kind of proposition in a language, thus generating wide varieties of conditional statements. In this manner, the paper argues that the attachment of a conditional string (i.e. protasis plus the so-called apodosis marker) to the apodosis proposition should not be viewed as analogous with the attachment of an adverbial clause to the main clause (as is common in most brands of syntactic analyses) nor should there be a need to propose two different kinds of attachments, one for the sentence-initial protasis (namely as IP/CP adjunction) and the other for the sentence-final protasis (namely as VP adjunction).2 2. What is a conditional statement? It is a widely accepted belief that conditional statements are complex sentences built up from two constituent clauses: the first being the if-clause and the second the then-clause. However, given the diverse syntactic varieties of conditional statements attested in different languages, it proves difficult, if not impossible, to give a clear-cut definition which might encompass all the characteristics of conditional statements. To begin with, it is generally assumed that conditional statements may have either an overt or a covert marking of two clauses. Besides, it is also commonly believed that they may overtly mark just one of the two clauses or may exhibit the same marker in both clauses. On the other hand, conditional statements are reported to be employed in situations where they convey quasi-conditional meanings rather than real hypothetical meanings.3 Furthermore, there are documented cases of reduced and pseudo-conditionals4 across the languages in which conditional statements may have either a covert protasis or a covert apodosis in the common ground. Realizing this difficulty, Declerck and Reed (2001: 9) have devised a general definition which tries to cover the basic characteristics of English conditionals, namely, “a conditional is a two-clause structure in which one of the clauses is introduced by if (possi-

On the role of protases in conditional statements

51

bly preceded by only, even or except) or except if (viz. unless).” Needless to say, this definition too is inadequate and says almost nothing about the ‘conditionalhood’ of English conditionals, let alone the characteristics of conditionals in general. Moreover, the mere presence or absence of the protasis marker in a statement cannot be the sole basis for considering a statement conditional since, as mentioned above, there are languages which do not obligatorily mark the protasis and others which do not mark either of the two clauses. Thus, I argue that it is the conditional string (viz. ‘if—— then’…, or, to put it in non-English terms, ‘Į——ȕ’…) – with or without overt markers – which is the defining characteristic of a conditional statement. According to this view, a conditional string is to be regarded as equipped with what others have taken to be the markers of two clauses – either overtly or covertly – and can be attached to any clause (viz. an assertion, an order, a question, etc.) of a language to fashion it into an apodosis clause in a conditional statement, as in (1)–(5) (1) (2) (3) (4) (5)

[If you heat water to 100 degrees celsius, Ø] it boils. [If it rains this afternoon, Ø] everybody will stay home. [If I become President, Ø] I’ll lower taxes. [If it’s raining now, then] your laundry is getting wet. [If it’s raining now, then] don’t go outside.

where the conditional string – which contains all the elements between square brackets – consists of the protasis and either a null marker Ø or the overt marker then. As is clear from examples (1), (2) and (3), the conditional string in English may contain a null marker [Ø] as well. As a consequence, only the remaining part of the statement – which lies outside the square brackets – is to be considered an apodosis proposition. Without entering into the syntactic details of the argument, we claim that for an adequate analysis of conditionals, any syntactic theory has to take into account the concept of the conditional string rather than dividing a conditional statement into two traditionally established parts (i.e. the if-clause and the then-clause), basically following conventional wisdom. The conditional string, in my view, is the sole factor which renders a statement a conditional statement. The protasis, thus, is the founding clause rather than a dependent or subordinate clause in a conditional statement, as has wrongly been assumed in most of the literature on conditionals.5 We reject the orthodox division of a conditional statement into two separate entities: (6)

[[If-PROPOSITION] [then-PROPOSITION]]

52

Ghanshyam Sharma

in which then proform is heedlessly considered to be a marker of the apodosis and thus is considered to be an integral part of it. We rather propose the division of a conditional statement along the following lines: (7)

[[If-PROPOSITION then] [-PROPOSITION]]

In this way, it can be said that a conditional statement is a statement which obligatorily contains the conditional string. The conditional string induces the hearer to suppose a case which is different than the case present in the common ground and thus introduces an alternative situation to the discourse. In semantic terms, the conditional string is the sole means of introducing conditionalhood to any sentence type in a language. Borrowing truth-conditional semantics terminology, it can be stated that a conditional string creates a suppositional world – a possible world – which is different from the actual world and in which the proposition contained in the apodosis has to be true. Using Fauconnier’s (1994) terminology, it can be said that the conditional string sets up a mental space wherein the content of an apodosis (an assertion, an imperative, an interrogative, etc.) is understood to exist. Moreover, in terms of mental model theory, it can be said that the conditional string serves as a basis for constructing a mental model of the state of affairs from which the conclusion in the apodosis has to be drawn.6 As stated above, the conditional string may introduce various meanings into a conditional statement. For example, it may serve as a premise in inferential or hypothetical conditionals or it can create a situation in which the speaker has to deliver his or her illocutions in the apodosis, producing the so-called speech-act conditional. To sum up, the conditional string hypothetically may be equipped either with both of the markers or just one of them. 3. The position of the protasis in a conditional statement In this section we make an attempt to determine the exact position of the protasis in a conditional statement, thus providing some evidence in support of our earlier claim (Sharma 2010), namely that in a conditional statement there may be the protasis-apodosis ordering only. As is widely reported, Greenberg (1963: 66) in his Universal of Word Order 14 states that: “In conditional statements, the conditional clause precedes the conclusion as the normal order in all languages.” Indeed, numerous studies have suggested that the protasis-apodosis clause order is in fact the universal order

On the role of protases in conditional statements

53

as it has a resemblance to the order of human reasoning and furthermore shows “parallels between order of elements in language and order of elements in experience”.7 In spite of this universally accepted fact, it is also acknowledged that many languages seem to exhibit the apodosis-protasis ordering as well. In fact, the abundance of data displaying the apodosisprotasis ordering in world languages has led linguists to believe that the protasis in final position indicates an after-thought on the part of the speaker. Another speculative argument goes as follows: “Given that it seems to be commoner cross-linguistically for the protasis to be marked overtly as non-factual than for the apodosis to be so marked …, placing the overtly marked protasis in front of the unmarked apodosis avoids the apodosis being interpreted as a factual statement” (Comrie 1986: 84). As I maintain elsewhere (Sharma 2010), contrary to the widely-held belief, I show that no language seems to exhibit a marked apodosis in the sentenceinitial position. In other words, there is no evidence of any sort to demonstrate that a language may indeed have the apodosis-protasis ordering in conditionals. Broadly speaking, the apodosis marking is considered to be done through a proform – generally a resumptive pronoun or a time adverbial – such as then in English. However, as will become clear from our subsequent discussion, we do not consider then to be an apodosis marker. Furthermore, there are no data from any language to suggest that the socalled sentence-initial apodosis can ever be marked, either overtly or covertly. According to our line of reasoning, the protasis is the basis of a conditional string which includes the so-called apodosis marker as well. Thus, the idea that the protasis may follow the apodosis is simply untenable and misleading. Let us now develop some further arguments in support of our hypothesis. In different typological studies it has been extensively reported that most languages overtly mark either the protasis or the apodosis, or both (Comrie 1986; Xrakovskij 2005). Following this line of thought, we can postulate the following four hypothetical types of clause combinations under which all languages can be classified. (8)

a. [Overtly marked P b. [Overtly marked P c. [Not overtly marked P d. [Not overtly marked P

+ + + +

Overtly marked Q] Non overtly marked Q] Overtly marked Q] Not overtly marked Q]

No language seems to exclusively belong to just one of the above four categories. As mentioned in the previous section, some overtly mark both the

54

Ghanshyam Sharma

protasis and the apodosis (8a), others mark either the protasis (8b) or the apodosis (8c) and there may also be languages without any means to overtly mark either of the two. Furthermore, even if a language has at its disposal the means to overtly mark both the protasis and the apodosis, it may or may not obligatorily employ them. For example, English and Hindi apparently have means to overtly mark both the protasis and the apodosis, but English obligatorily marks the protasis only whereas Hindi obligatorily requires the use of the particle to (i.e. then) to express conditionality, which is generally considered to be the marker of the apodosis. Likewise, Mandarin is reputed to have the means to mark both clauses, but unlike English and Hindi, it is reported to not mark them obligatorily. In fact Mandarin may allow both the protasis and the apodosis to remain unmarked. Thus, bearing in mind the obligation or non-obligation for overt marking in the protasis and the apodosis, we can revise the list of different types of combinations expressed in (8), and hypothetically have four different categories, as in (9) (9) a. [Obligatorily marked P b. [Obligatorily marked P c. [Non-obligatorily marked P d. [Non-obligatorily marked P

+ + + +

Obligatorily marked Q] Non-obligatorily marked Q] Obligatorily marked Q] Non-obligatorily marked Q]

As mentioned above, languages such as English, French and Italian are reported to obligatorily mark the protasis and thus can be said to belong to the category in (9b), whereas Hindi – and presumably many other IndoAryan languages as well – can be considered to belong to the category in (9c) since it does not mark the protasis obligatorily and requires the apodosis to be marked obligatorily. Mandarin, on the other hand, does not require any clauses to be obligatorily marked and therefore should be included in the category in (9d). Now, let us examine the above four classes, one by one, to check if indeed our argument can be proved valid. 3.1. Class 1: Obligatorily marked P + Obligatorily marked Q Only reliable data from this group of languages could provide any conclusive proof to demonstrate if a language may in fact have apodosis-protasis ordering in a conditional statement. However, to our knowledge, there is no record of any language in which the obligatory marking of two clauses is done through two overt and distinct markers.8 Nevertheless, closely related

On the role of protases in conditional statements

55

to this class, there is the case of the Ngiyambaa language in which the same marker is reported to be used in both clauses. In fact Comrie (1986: 84) cites the unique case of Ngiyambaa, observing that “in Ngiyambaa, with past tense counterfactuals, both clauses have the same overt marking (with the clitic -ma), and the first must be interpreted as protasis, …” (10) Nginuu-ma-ni burray giyi, ngindu-ma-ni yada gurawiyi Lit. ‘your-counterfactual-this child was, you-counterfactual-this well looked-after’ ‘If this child had been yours, you would have looked after it well.’9 It is not quite clear whether the so-called past tense counterfactual morpheme attested in the protasis and the apodosis in (10), viz. ma, does in fact mark both the protasis and the apodosis or if it performs different functions in different contexts, nor whether this phenomenon is limited to this type of conditional only or is attested in all other types of conditionals as well. However, the simple fact is that in Ngiyambaa, in the case of marking of two clauses through the same morpheme – no matter the purpose – the first clause is always considered the protasis, which clearly supports our hypothesis that there is indeed only protasis-apodosis ordering in natural languages. 3.2. Class 2: Obligatorily marked P + Not-obligatorily marked Q Now let us consider the second class, a class of languages in which according to traditional analyses only the protasis is believed to be marked obligatorily. The data from English, French and Italian, for example, seem to suggest an obligatory marking of the protasis (obtained through if, si and se, respectively) and an optional marking of the apodosis (obtained typically through some sort of proform, viz. then, alors and allora, respectively, which cannot appear sentence-initially). Although it is a topic of a separate study – thus outside the scope of this paper – whether the above mentioned proforms are in fact markers of the apodosis or of some other pragmatic elements, even a cursory examination of then reveals that English in reality has a null marker [Ø] and uses then to express bi-conditional (i.e. if and only if) meanings, e.g. in (10)10 (10) a. If you mow my lawn, I’ll pay you ten dollars. b. If you mow my lawn, then I’ll pay you ten dollars.

56

Ghanshyam Sharma

(10b) in fact means that ten dollars would be paid if, and only if, the lawn is mowed. Given that this purported English apodosis marker then carries a bi-conditionality meaning derived from the pragmatic scalarity in the protasis, a plausible explanation would be that it is associated with the protasis rather than with the apodosis. Then in fact is a pragmatic marker which induces implicatures, giving rise to bi-conditional readings, a topic that still requires much more serious investigation than it has received so far. Thus, to show the pragmatic affiliation of then with the protasis, we can roughly present (10) in the following manner: (11) a. [If you mow my lawn, Ø] I’ll pay you ten dollars. b. [[If you mow my lawn Ø] then] I’ll pay you ten dollars. where both (11a) and (11b) have a null marker and (11b) in addition has a pragmatic marker then which gives it if and only if reading. We can notice this meaning when we prepose the apodosis proposition. Thus, (11a) and (11b) can be rewritten as (12a) and (12b) respectively: (12) a. I’ll pay you ten dollars [if you mow my lawn Ø]. b. I’ll pay you ten dollars [if, and only if, you mow my lawn Ø]. Therefore, according to our line of thinking, then is a pragmatic marker rather than an apodosis marker and pragmatically belongs to the protasis. Thus, it goes without saying that we do not regard (12a) and (12b) as cases of apodosis-protasis ordering. We consider that in these examples, the apodosis proposition – which was a part of the apodosis in a normal protasisapodosis ordering – has simply been preposed. We argue that then is a part of the conditional string, rather than of an apodosis and makes an anaphoric reference to the protasis. In order to marshal our argument, let us examine some other characteristics of then. Firstly, as Bhatt and Pancheva (2006)11 have noted, then has to be adjacent to the protasis, as in (13a): (13) a. If it rains, then I think that we should stay at home. b.* If it rains, I think that then we should stay home. In syntactic terms, it has been argued that the surface location of then marks a predicate that combines with the if-clause and therefore then must be structurally adjacent to the if-clause. However, the fact that then in such circumstances has to be adjacent to the protasis also proves that it is pragmatically associated with the protasis rather than with the apodosis as it has

On the role of protases in conditional statements

57

to pick out the scalarity from the protasis. Furthermore, then has a pragmatic role to play which is derived from the protasis and has no semantic role to play as far as its presumed affinity with the apodosis is concerned. Secondly, there is a restriction on the use of then in those situations in which the protasis contains pragmatic elements expressing other pragmatic scales. In fact, being a pragmatic scalarity marker, then conflicts with other scalarity markers such as even if and only if, as is clear from examples (14) and (15): (14) Even if it rains, *(then) the football game will happen. (15) Only if it is sunny, *(then) I will visit you. If we look at (14) and (15) carefully, it becomes evident that Bhatt and Pancheva’s account of the above mentioned phenomenon does not seem to hold. It is not clear in the least, for example, what syntactic rules are presumed to preclude an appearance of then in (14) and (15) whose position is claimed to mark a predicate. Likewise, there is yet another restriction on the use of then when it is employed in the so-called generic conditionals. Iatridou (1994) and Dancygier and Sweetser (1997), for instance, have variably argued12 that there is a restriction on the use of then when the protasis contains reference to generic time or event, as in (16): (16) If Mary bakes a cake, *(then) she gives some slices of it to John. where then anaphorically picks up a generic time reference from the protasis rather than from the apodosis and requires the apodosis to have the similar generic reference. The above mentioned characteristics of then clearly reveal its pragmatic nature and structural association with the protasis rather than with the apodosis. We can thus conclude that data from this class of languages do not provide convincing evidence to consider apodosis-protasis ordering valid. 3.3. Class 3: Not obligatorily marked P + Obligatorily marked Q Now let us consider the class of languages in (9c) to which Hindi belongs. The strongest evidence against the purported apodosis-protasis ordering of clauses in conditional statements in fact comes from Hindi.13 The use of the so-called apodosis marker to (then) in Hindi is obligatory regardless of the

58

Ghanshyam Sharma

presence or absence of the protasis marker agar (if), as can be seen in (17). It is attached to the protasis even when the apodosis has to dislocate in the apodosis-protasis ordering, as is attested in (18): (17) (agar) RƗm ƗyƗ, to mѓ˾ us-se pnjchnj˾gƗ then I he-ABL ask-FUT.M.1SG (if) Ram come-PFV.M.SG ‘If Ram comes/came, I will/would ask him.’ (18) mѓ˾ RƗm-se pnjchnj˾gƗ agar vo ƗyƗ he come-PFV.M.SG I Ram-ABL ask-FUT.M.1SG if ‘I will/would ask Ram if he comes/came.’

to then

As mentioned above, in Hindi the apodosis marker to (then) has to remain attached to the protasis even when a conditional statement has to have the so-called apodosis-protasis clause ordering, as in (18). In fact an absence or a displacement of to (then) renders a Hindi conditional either ungrammatical or semantically odd, as can be noticed in (19a)–(19f): (19) a.* agar RƗm ƗyƗ ʊ maƭ us-se pnjchnj˾gƗ he-ABL ask-FUT.M.1SG if Ram come-PFV.M.SG ʊ I b.* ʊ RƗm ƗyƗ ʊ maƭ us-se pnjchnj˾gƗ ʊI he-ABL ask-FUT.M.1SG ʊ Ram come-PFV.M.SG c.? ʊ maƭ RƗm-se pnjchnj˾gƗ agar vo ƗyƗ he come-PFV.M.SG ʊI Ram-ABL ask-FUT.M.1SG if d.* to maƭ RƗm-se pnjchnj˾gƗ agar vo ƗyƗ he come-PFV.M.SG then I Ram-ABL ask-FUT.M.1SG if e.* to maƭ RƗm-se pnjchnj˾gƗ — vo ƗyƗ then I Ram-ABL ask-FUT.M.1SG — he come-PFV.M.SG f.* ʊ maƭ RƗm-se pnjchnj˾gƗ ʊ vo ƗyƗ ʊI Ram-ABL ask-FUT.M.1SG ʊ he come-PFV.M.SG ‘I will/would ask Ram if he comes/ came.’ On the grammaticality test, the situation in the Hindi examples in (19) can be summarized as in (20) where Į and ȕ are markers of the protasis and apodosis, respectively, P and Q the propositions contained in the protasis and the apodosis, and Ø, a null marker: (20) a. [ĮP ĺ ØQ] A grammatically unacceptable conditional string b. [ØP ĺ ØQ] A grammatically unacceptable conditional string c. [ØQ ĸ ĮP] A semantically unacceptable conditional string

On the role of protases in conditional statements

59

d. [ȕQ ĸ ĮP] A grammatically unacceptable conditional string e. [ȕQ ĸ ØP] A grammatically unacceptable conditional string f. [ØQ ĸ ØP] A grammatically unacceptable conditional string Hindi conditionals require the particle to (then) to follow the protasis irrespective of the type of conditional (i.e. relevance or ‘biscuit’14 conditionals (e.g. 21a and 21b), conditionals giving advice (e.g. 22a and 22b) or any other type of idiomatic conditional expressions (e.g. 23a and 23b). (21) a. (agar) Ɨp cƗhͅ to bisku‫ ܒ‬ƗlmƗrƯ=mͅ if you want-SUBJ.3PL then biscuits sideboard=in rakhe hѓ˾ placed aux-PRES.M.3PL ‘If you want, there are biscuits on the sideboard.’ b. bisku‫ ܒ‬ƗlmƗrƯ=mͅ rakhe hѓ˾ agar biscuits sideboard=in placed aux-PRES.M.3PL if Ɨp cƗhͅ to you want-SUBJ.3PL then ‘There are biscuits in the sideboard if you want them.’ (22) a. (agar) Ɨp merƯ bƗt mƗnͅ to if you my advice accept-SUBJ.3PL davƗ le lenƗ acchƗ hogƗ medicine take better be-FUT.3SG ‘If I were you, I would take some medicine.’

koƯ then some

b. koƯ davƗ le lenƗ acchƗ hogƗ agar some medicine take better be-FUT.3SG if Ɨp merƯ bƗt mƗnͅ to you my advice accept-SUBJ.3PL then ‘If I were you, I would take some medicine.’ (23) a. (agar) Ɨp burƗ na mƗnͅ to ek bƗt if you bad notconsider-SUBJ.3PL then one thing kahnƗ chƗhnj˾gƗ say want-FUT.3SG ‘If you don’t mind, I would like to say something.’

60

Ghanshyam Sharma

b. ek bƗt kahnƗ chƗhnj˾gƗ agar Ɨp burƗ you bad one thing tell want-FUT.3SG if na mƗnͅ to then notconsider-SUBJ.3PL ‘If you don’t mind, I would like to say something.’ Moreover, in Hindi only the presumed apodosis marker to (i.e. then) carries all sorts of pragmatic meanings derived from scalarity (i.e. even if, only if, if and only if, etc.) in the protasis, as is clear from the examples in (24): (24) a. (agar) unhõne mujhe bulƗyƗ to invite-PFV.M.SG then if they-ERG me mѓ˾ pƗr‫ܒ‬Ư=mͅ jƗnj˾gƗ I party=in go-FUT.M.1SG ‘If they invite/invited me, I will/would go to the party.’ b. (agar) unhõne mujhe bulƗyƗ to hƯ invite-PFV.M.SG then only if they-ERG me mѓ˾ pƗr‫ܒ‬Ư=mͅ jƗnj˾gƗ I party=in go-FUT.M.1SG ‘I will/would go to the party only if they invite/invited me.’ c. (agar)unhõne mujhe bulƗyƗ to bhƯ invite-PFV.M.SG then even if they-ERG me mѓ˾ pƗr‫ܒ‬Ư=mͅ nahƯ˾ jƗnj˾gƗ I party=in notgo-FUT.M.1SG ‘Even if they invite/invited me, I will/would not go to the party.’ All the examples in (24) unequivocally suggest a closer affinity of to (then) with the protasis rather than with the apodosis and provide proof that there is no apodosis-protasis ordering as such and that the conditionality of a sentence derives typically from the conditional string which includes the so-called apodosis marker as well. 3.4. Class 4: Not obligatorily marked P + Not obligatorily marked Q On the basis of different studies, Mandarin can be classified among those languages which seem to have characteristics described in category (9d), since in it “the protasis necessarily precedes the apodosis, whether the protasis alone is marked for non-factuality (by a conjunction such as rúgu΅

On the role of protases in conditional statements

61

‘if’), whether the apodosis alone is marked (for instance by nà and/or jìu ‘then, in that case’), whether both are marked, or whether neither is marked.” (Comrie 1986: 85), as in (25) (rúgu΅) ZhƗngsƗn hƝ j΃u, w΅ (jìu) (If) Zhangsan drink wine I (then) ‘If Zangsan drinks wine, (then) I will scold him.’

mà tƗ scold him

where, as Comerie points out, the protasis must precede the apodosis, irrespective of whether either protasis or apodosis is marked overtly. Needless to say, the case of Mandarin lends even stronger support to our hypothesis that there is only protasis-apodosis ordering in conditional statements regardless of the presence or absence of any overt clause markers. We can sum up our discussion of above four possible classes exhibiting different possible orderings of two clauses in a conditional statement in table 1. Table 1. Table sumarizing the four classes of conditional statement.15 1. Class 1 2. Class 2 3. Class 3 4. Class 4

PĺQ [ĮPĺȕQ] [ĮPĺØ(ȕ)Q] [(Į)PĺȕQ] [(Į)Pĺ(ȕ)Q]

Q, P ĺ — [Q[ĮPĺØ]] [Q[ĮPĺȕ]] —

As is clear from table 1, languages belonging to classes 2 and 3 seem to exhibit the so-called apodosis-protasis (QĸP) ordering. In fact, data from these languages have led linguists to believe that a conditional statement may exhibit both clause orderings (i.e. PĺQ and QĸP). However, as we have argued above, these cases in no way violate the universal of conditional clause ordering (namely PĺQ), since in such cases only the proposition contained in the apodosis is preposed. A null-marker at the end of the conditional string in class 2 effectively indicates that in these languages a preposing of the apodosis-proposition does regularly take place as a result of speaker’s different pragmatic strategies. Furthermore, data from the languages which belong to class 3 indubitably demonstrate that in these languages the so-called apodosis marker, say ‘ȕ’, cannot be omitted and, except for a few cases,16 has to remain at the end of the conditional string even in cases where the apodosis-proposition has to be preposed or fronted. We believe that the preposing of apodosis-proposition attested in languages that belong to classes 2 and 3 of table 1 requires further pragmatic research

62

Ghanshyam Sharma

in order to be better understood. One of the reasons of apodosis-proposition preposing seems to derive from the fact that conditional statements are always discourse-bound (see e.g. Akatsuta 1986). In fact, the phenomenon of apodosis-proposition preposing, reduced conditionals (i.e. delition of either P or Q), nonconditional conditionals (see Lycan 2001 for details) and pseudo-conditionals can only be understood in the light of discourse-bound nature of conditional statements. Thus, we believe that depending on the elements of knowledge shared by the speaker and hearer (available from the previous part of the discourse) the speaker may consider it necessary to prepose the apodosis in order to highlight the information contained in it thus violating the normal clause-ordering. Another reason for apodosisproposition preposing has to do with type of modality contained in it. In fact, when the speaker has to express deontic modality in making requests or orders he invaribly begins the conditional statement by preposing the apodosis-proposition, as in, (25) a. Sit down, if you want! b.? If you want sit down! We believe that similar observations can be made about other conditional statements where the apodosis carries non-assertive illocutionary forces – as is the case in uttering exclaimations, interrogatives etc. – and thus expresses a non-epistemic modality. Subject to further language specific research, we can expect apodoses carrying non-epistemic modal meanings to be always preposed – irrespective of syntactic differences in languages. In sum then, our analysis does not support the widely held belief according to which conditionals can have both orderings: PĺQ and QĸP. 4. Do protases contain speaker’s knowledge or beliefs? In order to determine an overall semantic contribution of the protasis to a conditional statement, it is necessary to establish the speaker’s attitude towards the nonfactuality in the protasis. In other words, it is important to see whether the protasis can contain speaker’s knowledge or beliefs. From the point of view of the speaker, conditional statements typically consist of three elements – the protasis, the apodosis and the relationship between the two – in which speaker’s epistemic stand can be assessed.17 Now, if we look at the four traditional types of English conditionals, we notice that there are four different situations. In the first case, there is conditional type

On the role of protases in conditional statements

63

0 such as If you heat water to 100 degrees celsius, it boils where the speaker knows the relation that holds between P and Q, but neither knows nor believes the stae of affairs in P and Q. In the case of second type of conditionals, such as If it rains this afternoon, everybody will stay home, the speaker neither knows nor believes the state of affairs in the protasis but believes in the relation that holds between P and Q, and believes also that either necessarily or possibly Q (i.e. everybody will stay home). In the third type of conditionals, such as If she came back early, he wouldn’t be able to go to the party the speaker neither knows nor believes that P, but believes that possibly Q. The speaker believes that the relation between P and Q necessarily holds. The only category in which the speaker knows that ‘necessarily not-P’ is the fourth category. In fact, only the fourth type of conditionals, such as If she had bought a lottery ticket, she would have become rich contains the speaker’s counter-to-fact knowledge, namely that ‘she did not buy a ticket’. We can thus summarize the epistemic state of the speaker in four types of conditionals in the following table. Table 2. Table showing the distribution of epistemic elements in the three parts of four types of epistemic conditional statements.18 1. Conditional type 0 2. Conditional type 1 3. Conditional type 2 4. Conditional type 3

P ¬ KP/¬BP ¬ KP/¬BP ¬ KP/¬BP K¬P

ĺ K B B K

Q ¬ KQ/¬BQ BQ BQ K¬Q/ B¬Q

Thus, as explained above in epistemic conditionals protases do not contain speaker’s knowledge or beliefs19 except in the fourth category of countefactuals where the speaker has a piece of counterfactual knowledge of the state of afairs in P and Q. In the first category, the speaker does not possess any piece of knowledge regarding specific occurances of P and Q, but rather of their generic occurances only. In other words, at all times, in all the cases of P it is necessarily Q. As Comrie (1986: 88) has rightly observed, it is difficult to accept a neat biparttite or tripartite division of conditional statements with a clear-cut boundary between the two or three types. In fact, different degrees of hypotheticality in conditional statements are very much subjective. However, if we look at the speaker’s epistemic state in table 2 we notice that there may be three types of conditionals only, as can be seen in table 3 where we have regrouped four types into three, namly A, B and C. The conditional types 2 (namely If she gives me ten dollar, I will do her

64

Ghanshyam Sharma

job) and 3 (namely If she gave me ten dollar, I would do her job) attested in languages such as English express exactly the same epistemic stand of the speaker, as is clear from table 3. These three types can still be reduced to two types as the conditionals belonging to type 0 contain the lowest degree of hypotheticality and therefore are not real conditionals expressing hypotheticality. The conditionals in type 4 on the other hand contain the highest degree of hypotheticality. Table 3. A regrouping of four types of conditional statements presented in table 2. A 1. Conditional type 0 B 2. Conditional type 1 3. Conditional type 2 C 4. Conditional type 3

P ¬ KP/¬BP ¬ KP/¬BP ¬ KP/¬BP K¬P

ĺ K B B K

Q ¬ KQ/¬BQ BQ BQ K¬Q/ B¬Q

Let us now take another issue which has a direct bearing on the role played by the protasis in a conditional statement. Syntactically, conditional clauses have generally been thought to belong to, thus considered to have similar qualities of, a class of adverbials or complementizers. For example, Bhatt and Pancheva (2006) synthesize a widespread common belief among linguists according to which the conditional clause belongs to a class of adverbial clauses that includes, among others, clausal adverbial of time, cause, and concessions, as can be seen in examples in (26). (26) a. b. c. d.

If Andrea arrived late, Clara must have got upset. When Andrea arrived late, Clara got upset. Because Andrea arrived late, Clara got upset. Although Andrea arrived on time, Clara got upset.

It is argued that in (26a) the if-clause is attached to the then-clause in the same way as it is in the rest of examples (i.e. 26b, 26c, 26d). Thus, according to this line of research, the conditional clause should get a treatment similar to the one in other cases. In pragmatic terms, though, this idea does not seem to stand up to close scrutiny. In our view, one of the striking characteristics of the protasis is that it, unlike other syntactic complementizers, does not carry and thus does not attach (or contribute) to conditional statements any epistemic elements derived from speaker’s knowledge or beliefs. To see the epistemic state of examples in (26) we can rephrase them in (27). We claim that, contrary to the widely held belief, protases do not

On the role of protases in conditional statements

65

carry speaker’s knowledge or belief, as can be seen in (27a) where the speaker neither knows nor believes that Andrea in fact came late at time t and thus the speaker’s necessary belief (B): (27) a. ?The speaker knows that ‘Andrea arrived late at time t’, and the speaker believes that ‘Clara got upset at time t’. b. The speaker knows that (i) ‘Andrea arrived late at time t’, and (ii) ‘Clara got upset at time t’. c. The speaker knows that (i) ‘Andrea arrived late at time t’, and (ii) ‘Clara got upset at time t’ and believes that ‘(i) caused (ii)’. d. The speaker knows that (i) ‘Andrea arrived on time at time t’, and (ii) ‘Clara got upset at time t’. The speaker believes that ‘(ii) took place notwithstanding (i)’ In fact, (26a) cannot be paraphrased as (27a) since the speaker in making an utterance of (26a) does not know that Andrea in fact arrived late. Had the speaker known that, he would have instead made an utterance using other constructions such as: Given that/ Since/ As/ For Andrea arrived late ... Consequently, we believe that it is a fundamental mistake to equate a conditional clause with an adverbial complementizer. There are no proofs to support the idea that (26a) is semantically similar to (26b), (26c) and (26d). Let us consider another similar argument which is related to the question of the presence or absence of speaker’s epistemic elements, namely knowledge and beliefs in the protasis. In various syntactic analyses, it has been observed that conditional clauses (i.e. protases) are incompatible with Main Clause Phenomenon (exemplified in English by argument fronting). For example, it has been argued that English does not permit argument fronting, as (28b) is ungrammatical: (28) a. If you don’t pass these exams, you won’t get the degree. b.*If these exams you don’t pass, you won’t get the degree. Similarily, the protasis cannot contain the so-called Speaker Oriented Adverbs. For example, an adverb such as frankly, renders (29b) ungrammatical. (29) a. If he’s unable to cope, we’ll have to replace him. b. ??*If frankly he’s unable to cope, we’ll have to replace him. We believe that these syntactic phenomena have to be considered from the point of view of the speaker’s epistemic stand only since it is difficult to

66

Ghanshyam Sharma

provide an elegant syntactic analysis of conditionals without taking into consideration modal meanings of the speaker. 5. The distribution of TAM elements in the Hindi protasis The TAM elements in the protasis are responsible for establishing different degrees of hypotheticality in a conditional statement as a whole, although they do not indicate the same meanings as they do when used in factual propositions. In order to show the distribution of Hindi TAM elements, we provide a list of possible combinations of the protasis and the apodosis in Hindi conditional statements in Appendix 3. Table 4. Distribution of possible TAM elements in the protasis and the apodosis of the Hindi conditional statements.20 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

Simple perfective (aorist) Past (aspect-less) Past habitual Past progressive Past perfective Present (aspect-less) Present habitual Present progressive Present perfective Future (aspect-less, presumptive) Future habitual (presumptive) Future progressive (presumptive) Future perfective (presumptive) Subjunctive (aspect-less) Subjunctive habitual Subjunctive progressive Subjunctive perfective Counterfactual (aspect-less) Counterfactual habitual Counterfactual progressive Counterfactual perfective Imperatives Interrogatives Exclamations

P

Q

YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES NO NO NO

YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES

On the role of protases in conditional statements

67

As can be seen in more detail in Appendix 3, in Hindi the conditional type 0 employs the present habitual in P and Q whereas the conditional type 3 necessarily employs TAM elements listed in 18, 19, 20, 21 in table 4. The conditional type 2 requires P to be in the aorist tense and Q in the future tense. The conditional type 1 makes use of the rest of Hindi TAM elements. 6. Conclusions Analyzing data mainly from Hindi, we have shown that a conditional statement obligatorily contains – either overtly or covertly – a conditional string, namely PʊQ, which is indivisible and that the order of the two elements in the conditional string can never be reversed. This conclusion will have important implications for any type of syntactic analysis which aims to provide a division of a conditional statement according to the traditional line of thinking, namely [[Pʊ], [Qʊ]]. Our proposal in this regard is that a syntactic analysis of conditional statements should have the following form: [[Pʊ, Q] ʊ]] and, with a possible alternation, [ʊ [Pʊ, Q]], due to the preposing of the proposition contained in the apodosis. Thus we reject the syntactic analyses which look at the phenomenon along these lines: [[Qʊ], [Pʊ]]. We have also demonstrated that the Hindi TAM elements in the protasis set up different degrees of hypotheticality of a conditional statement. In addition, we have shown that contrary to widely held belief, the protasis carries neither speaker knowledge nor speaker belief. This pragmatic fact also explains why the protasis can contain neither the fronted argument nor speaker-oriented adverbs.

Appendix 1 To see the distribution of the Hindi TAM elements in the protasis (i.e. P), we provide in this appendix a list of examples which contain 21 out of 24 categories we registered in Table 3. Needless to say, imperatives, interrogatives and exclamations cannot appear in the protasis. P-TAM-1 Simple perfective/ aorist agar/us-ne/ un dinõ/ vakƗlat/ kƯ/ to … if/ he-ERG/ those days/ advocacy-F/ do-PFV.F.3SG/ then … ‘If he practiced law those days, then …’

68

Ghanshyam Sharma

P-TAM-2 Past (aspect-less) agar/ vo/ un dinõ/ vakƯl/ thƗ/ to … if/ he/ those days/ lawyer/ be-PST.M.3SG/ then … ‘If he was a lawyer those days, then …’ P-TAM-3 Past habitual agar/ vo/ un dinõ/ vakƗlat/ kartƗ/ thƗ/ to … if/ he/ those days/ advocacy-F/ do-IMPFV.M.3SG/ AUX-PST.M.3SG/ then … ‘If he practiced/ would practice law those days then …’ P-TAM-4 Past progressive/ agar/ vo/ us samay/ vakƗlat/ kar rahƗ/ thƗ/ to … if/ he/ that period/ advocacy-F/ do-PROG.M.3SG/ AUX-PST.M.3SG/ then … ‘If he was practicing law during that period, then …’ P-TAM-5 Past perfective agar/ us-ne/ us samay/ vakƗlat/ kƯ/ thƯ/ to … if/ he-ERG/ at that time/ advocacy-F/ do-PFV.F.3SG/ AUX-PST.F.3SG/ then … ‘If he then had practiced law, then …’ P-TAM-6 Present (aspect-less) agar/ vo/ vakƯl/ hѓ/ to … if/ he/ lawyer/ be-PRES.3SG/ then … ‘If he is a lawyer, then …’ P-TAM-7 Present habitual agar/ vo/ ab/ vakƗlat/ kartƗ/ hѓ/ to … if/ he/ now/ advocacy-F/ do-IMPFV.M.3SG/ AUX-PRES.3SG/ then … ‘If he practices law now, then …’ P-TAM-8 Present progressive agar/ vo/ ab/ vakƗlat/ kar rahƗ/ hѓ/ to … if/ he/ now/ advocacy-F/ do-PROG.M.3SG/ AUX-PRES.3SG/ then … ‘If he is practicing law now, then …’ P-TAM-9 Present perfective agar/ us-ne/ vakƗlat/ kƯ/ hѓ/ to … if/ he-ERG/ advocacy-F/ do-PFV.F.3SG/ AUX-PRES.3SG/ then … ‘If has practiced law, then … ’ P-TAM-10 Future (aspect-less) agar/ vo/ vakƗlat/ karegƗ/ to … if/ he/ advocacy-F/ do-M.3SG/ then … ‘If he will practice law, then …’ P-TAM-11 Future habitual (presumptive habitual) agar/ vo/ vakƗlat/ kartƗ/ hogƗ/ to … if/ he/ advocacy-F/ do-IMPFV.M.3SG/ AUX-PRESM/ then … ‘If he must be practicing law, then …’

On the role of protases in conditional statements P-TAM-12 Future progressive (presumptive progressive) agar/ vo/ vakƗlat/ kar rahƗ/ hogƗ/ to … if/ he/ advocacy-F/ do-PROG.M.3SG/ AUX-FUT.M.3SG/ then … ‘If he must be practicing law, then …’ P-TAM-13 Future perfective (presumptive perfective) agar/us-ne/ vakƗlat/ kƯ/ hogƯ/ to … if/ he-ERG/ advocacy-F/ do-PFV.F.3SG/ AUX-FUT.F.3SG/ then … ‘If he must have practiced law, then …’ P-TAM-14 Subjunctive (aspect-less) agar/ vo/ vakƗlat/ kare/ to … if/ he/ advocacy/ do-SUBJ.3SG/ then … ‘If he should practice law, then …’ P-TAM-15 Subjunctive habitual agar/ vo/ vakƗlat/ kartƗ/ ho/ to … if/ he/ advocacy-F/ do-IMPFV.M.3SG/ AUX-SUBJ/ then … ‘If he may be practicing law, then …’ P-TAM-16 Subjunctive progressive agar/ vo/ vakƗlat/ kar rahƗ/ ho/ to … if/ he/ advocacy-F/ do-PROG.M.3SG/ AUX-SUBJ.3SG/ then … ‘If he may be practicing law, then …’ P-TAM-17 Subjunctive perfective agar/ us-ne/ vakƗlat/ kƯ/ ho/ to … if/ he-ERG/ advocacy-F/ do-PFV.F.3SG/ AUX-SUBJ.3SG/ then … ‘If he may have practiced law, then …’ P-TAM-18 Counterfactual (aspect-less) agar/ vo/ vakƗlat/ kartƗ/ to … if/ he/ advocacy-F/ do-IMPFV.M.3SG/ then … ‘If he had practiced law then …’ P-TAM-19 Counterfactual habitual agar/ vo/ un dinõ/ vakƗlat/ kartƗ/ hotƗ/ to … if/ he/ those days/ advocacy-F/ do-IMPFV.M.3SG/ AUX-IMPFV.M.3SG/ then … ‘If he had been practicing law those days, then …’ P-TAM-20 Counterfactual progressive agar/ vo/ un dinõ/ vakƗlat/ kar rahƗ/ hotƗ/ to … if/ he/ those days/ advocacy-F/ do-PROG.M.3SG/ AUX-IMPFV.M.3SG/ then … ‘If he had been practicing law those days, then …’ P-TAM-21 Counterfactual perfective agar/ us-ne/ vakƗlat/ kƯ/ hotƯ/ to … if/ he-ERG/ advocacy-F/ do-PFV.F.3SG/ AUX-IMPFV.F.3SG/ then … ‘If he had practiced law, then …’

69

70

Ghanshyam Sharma

Appendix 2 In this appendix, we provide examples of the apodosis (i.e. Q) containing all the 24 types of TAM elements listed in Table 3. Q-TAM-1 Simple perfective (aorist) (to)/ usne/ zarnjr/ ferrƗrƯ/ kharƯdƯ (then)/ he-ERG/ certainly/ Ferrari/ buy-PFV.F.3SG ‘(then) he certainly bought a Ferrari.’ Q-TAM-2 Past (aspect-less) (to)/ vo/ bahut pѓse vƗlƗ/ thƗ (then)/ he/ very rich/ be-PST.M.3SG ‘(then) he was very rich.’ Q-TAM-3 Past habitual (to)/ vo/ roz/ wiskƯ/ pƯtƗ/ thƗ (then)/ he/ everyday/ whisky/ drink-IMPFV.M.SG/ AUX-PST.M.3SG ‘(then) he drank whisky everyday.’ Q-TAM-4 Past progressive (to)/ vo/ us samay/ wiskƯ/ pƯ rahƗ/ thƗ (then)/ he/ at that moment/ whisky/ drink-PROG.M/ AUX-PST.M.3SG ‘(then) he was drinking whisky at that moment.’ Q-TAM-5 Past perfective (to)/ usne/ zarnjr/ ferrƗrƯ/ kharƯd lƯ/ thƯ (then)/ he-ERG/ certainly/ Ferrari/ buy-take-PFV.F.SG/ AUX-PRES.3SG ‘(then) he certainly had bought a Ferrari.’ Q-TAM-6 Present (aspect-less) (to)/ vo/ bahut pѓse vƗlƗ/ hѓ (then)/ he/ very rich/ be-PRES.3SG ‘(then) he is very rich.’ Q-TAM-7 Present habitual (to)/ vo/ roz/ wiskƯ/ pƯtƗ/ hѓ (then)/ he/ everyday/ whisky/ drink-IMPFV.M.SG/ AUX-PRES.3SG ‘(then) he drinks whisky everday.’ Q-TAM-8 Present progressive (to)/ vo/ is samay/ wiskƯ/ pƯ rahƗ/ hѓ (then)/ he/ this moment/ whisky/ drink-PROG.M.SG/ AUX-PRES.3SG ‘(then) he is drinking whisky at the moment.’ Q-TAM-9 Present perfective (to)/ usne/ zarnjr/ ferrƗrƯ/ kharƯd lƯ/ hѓ (then)/ he-ERG/ certainly/ Ferrari/ buy-take-PFV.F.SG/ AUX-PRES.3SG ‘(then) he certainly has bought a Ferrari.’

On the role of protases in conditional statements Q-TAM-10 Future (aspect-less, presumptive) (to)/ vo/ zarnjr/ bahut pѓse vƗlƗ/ hogƗ (then)/ he/ certainly/ very rich/ be-FUT.3SG ‘(then) he must certainly be very rich.’ Q-TAM-11 Future habitual (presumptive) (to)/ vo/ zarnjr/ roz/ wiskƯ/ pƯtƗ/ hogƗ (then)/ he/ certainly/ everyday/ whisky/ drink-IMPFV.M.SG/ AUX-PRES.3SG ‘(then) he must be drinking whisky everyday.’ Q-TAM-12 Future progressive (presumptive) (to)/ vo/ zarnjr/ is samay/ wiskƯ/ pƯ rahƗ/ hogƗ (then)/ he/ certainly/ this moment/ whisky/ drink-PROG.M.SG/ AUX-PRES.3SG ‘(then) he must be drinking whisky at the moment.’ Q-TAM-13 Future prefective (Presumptive) (to)/ usne/ zarnjr/ ferrƗrƯ/ kharƯd lƯ/ hogƯ (then)/ he-ERG/ certainly/ Ferrari/ buy-take-PFV.F.SG/ AUX-FUT.F.3SG ‘(then) he certainly must have bought a Ferrari.’ Q-TAM-14 Subjunctive (aspect-less) (to)/ ho saktƗ hѓ ki/ vo/ bahut pѓse vƗlƗ/ ho (then)/ it may be that/ he/ very rich/ be-SUBJ.3SG ‘(then) maybe he is very rich.’ Q-TAM-15 Subjunctive habitual (to)/ ho saktƗ hѓ ki/ vo/ roz/ wiskƯ/ pƯtƗ/ ho (then)/ may be that/ he/ everyday/ whisky/ drink-IMPFV.M.SG/ AUX-SUBJ.3SG ‘(then) it may be that he drinks whisky everyday.’ Q-TAM-16 Subjunctive progressive (to)/ ho saktƗ hѓ ki/ vo/ ab/ wiskƯ/ pƯ rahƗ/ ho (then)/ may be that/ he/ now/ whisky/ drink-PROG.M.SG/ AUX-SUBJ.3SG ‘(then) he may be drinking whisky now.’ Q-TAM-17 Subjunctive perfective (to)/ ho saktƗ hѓ ki/ usne/ ferrƗrƯ/ kharƯd lƯ/ ho (then)/ may be that/ he-ERG/ Ferrari/ buy-take-PFV.F.SG/ AUX-SUBJ.3SG ‘(then) he may have bought a Ferrari.’ Q-TAM-18 Counterfactual (aspect-less) (to)/ vo/ bahut pѓse vƗlƗ/ hotƗ (then)/ he/ very rich/ be-IMPFV.M.3SG ‘(then) he would have been very rich.’ Q-TAM-19 Counterfactual habitual (to)/ vo/ roz/ wiskƯ/ pƯtƗ/ hotƗ (then)/ he/ everyday/ whisky/ drink-IMPFV.M.SG/ AUX-IMPFV.M.3SG ‘(then) he would have been drinking whisky everyday.’

71

72

Ghanshyam Sharma

Q-TAM-20 Counterfactual progressive (to)/ vo/ is samay/ wiskƯ/ pƯ rahƗ/ hotƗ (then)/ he/ this moment/ whisky/ drink-PROG.M.SG/ AUX-IMPFV.M.3SG ‘(then) he would have been drinking whisky now.’ Q-TAM-21 Counterfactual perfective (to)/ usne/ zarnjr/ ferrƗrƯ/ kharƯd lƯ/ hotƯ (then)/ he-ERG/ certainly/ Ferrari/ buy-take-PFV.F.SG/ AUX-IMPFV.F.3SG ‘(then) he certainly would have bought a Ferrari.’ Q-TAM-22 Imperative (to)/ use/ zarnjr/ bulƗo (then)/ him/ certainly/ invite-IMP.2PL ‘(then) invite him!’ Q-TAM-23 Interrogative (to)/ uskƗ/ be‫ܒ‬Ɨ/ kyƗ/ kartƗ/ hѓ (then)/ his/ son/ what/ do-IMPFV.M.3SG/ AUX-PRES.3SG ‘(then) what does his son do?’ Q-TAM-24 Exclamations (to)/ mujhe/ kyƗ/ lenƗ-denƗ! (then)/ to me/ what/ take-give ‘(then) who cares!’

Appendix 3 In this appendix we provide a list of possible combinations of different types of protasis (P) and apodosis (Q) in conditional statements in Hindi. All the numbers refer to those in Table 3. Although the list is derived bearing in mind a possible combination of types in Appendix 1 and Appendix 2, not all the following combinations can be obtained by simply combining the examples we have supplied since their semantic content is not always coherent. P-TAM-1 + [Q-TAM-2]/ [Q-TAM-6]/ [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-2 + [Q-TAM-2]/ [Q-TAM-3]/ [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-3 + [Q-TAM-2]/ [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ [Q-TAM-13]/ TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ TAM-23]/ [Q-TAM-24]

[Q[Q[Q[Q[Q[Q-

On the role of protases in conditional statements

73

P-TAM-4 + [Q-TAM-2]/ [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ [Q-TAM-13]/ [QTAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-5 + [Q-TAM-2]/[Q-TAM-6]/ [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ [QTAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [QTAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-6 + [Q-TAM-6]/ [Q-TAM-7]/ [Q-TAM-8]/ [Q-TAM-9]/ [Q-TAM-10]/ [Q-TAM11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-7 + [Q-TAM-6]/ [Q-TAM-7]/ [Q-TAM-8]/ [Q-TAM-9]/ [Q-TAM-10]/ [Q-TAM11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-8 + [Q-TAM-6]/ [Q-TAM-7]/ [Q-TAM-8]/ [Q-TAM-9]/ [Q-TAM-10]/ [Q-TAM11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-9 + [Q-TAM-6]/ [Q-TAM-7]/ [Q-TAM-8]/ [Q-TAM-9]/ [Q-TAM-10]/ [Q-TAM11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-10 + [Q-TAM-10]/[Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-22]/ [Q-TAM-23]/ [QTAM-24] P-TAM-11 + [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-12 + [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-13 + [Q-TAM-10]/ [Q-TAM-11]/ [Q-TAM-12]/ [Q-TAM-13]/ [Q-TAM-14]/ [QTAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-14 + [Q-TAM-14]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-15 + [Q-TAM-10]/[Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [QTAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-16 + [Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [Q-TAM-17]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-17 + [Q-TAM-6]/[Q-TAM-10]/[Q-TAM-14]/ [Q-TAM-15]/ [Q-TAM-16]/ [QTAM-17]/ [Q-TAM-22]/ [Q-TAM-23]/ [Q-TAM-24] P-TAM-18 + [Q-TAM-18]/ [Q-TAM-19]/ [Q-TAM-20]/ [Q-TAM-21]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-19 + [Q-TAM-18]/ [Q-TAM-19]/ [Q-TAM-20]/ [Q-TAM-21]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-20 + [Q-TAM-18]/ [Q-TAM-19]/ [Q-TAM-20]/ [Q-TAM-21]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24] P-TAM-21 + [Q-TAM-18]/ [Q-TAM-19]/ [Q-TAM-20]/ [Q-TAM-21]/ [Q-TAM-22]/ [QTAM-23]/ [Q-TAM-24]

74

Ghanshyam Sharma

Abbreviations 1 = first person; 2 = second person; 3 = third person; ABL = ablative; ACC = accusative; AUX = auxiliary; CFV = contrafactive (counterfactual); CONT = continuous; ERG = ergative; F = feminine; FUT = future; HON = honorific; IMP = imperative; IMPFV = imperfective; M = masculine; OBL = oblique; PART = particle; PASS = passive; PFV = perfective; PL = plural; PRES = present; PRESM = presumptive; PROG = progressive; PST = past; SG = singular; SUBJ = subjunctive.

Notes * 1.

2.

3. 4. 5.

I am grateful to Professor Peter Edwin Hook (Michigan) for helpful suggestions. Needless to say, I am solely responsible for all errors and inaccuracies which may occur. Although familiar with the technicalities of different brands of theories of conditionals, both formal and informal, semantic as well as syntactic, we intend to avoid any direct discussion of them altogether since our aim in this brief study is to put forward certain new ideas and our conviction is that once one starts discussing the technicalities of any theory, one remains entrapped in it, in a no-go situation. This is especially when, in order to follow these theories, you have to begin by using the terms which have been inappropriately used for centuries. Most of the logical theories of conditionals, to cite just one case, have misused the terms ‘indicative’ and ‘subjunctive’ in their discussion, without ascertaining the basic meaning of the terms. Thus, we will be using the following terms and symbols: protasis = the first clause or P-clause or the antecedent, apodosis = the second clause or Q-clause or consequent, Ƒ = modal necessity operator, ¸ = modal possibility operator, ¬ = negative modality operator. After giving a concise account of different syntactic theories of conditional, Bhat and Pancheva (2006) advance an idea by which there is a need to provide two different solutions for two types of attachments, one for the sentence-initial protasis and another for the sentence-final protasis. We believe that syntacticians will rather have to find some other solutions for a dislocation of the proposition contained in the apodosis. A good discussion of such quasi-conditionals in English can be found in “Non-conditional Conditionals”, Michael L. Geis and William G. Lycan in Real Conditionals W. G. Lycan, Oxford: Clarendon Press, pp. 184-205. For an excellent survey of such pseudo-P or pseudo-Q conditional constructions, see Renaat Declerck and Susan Reed (2001). Without paying due attention to the crucial role played by the protasis (conditional clause) in the actualization of a conditional statement, most studies

On the role of protases in conditional statements

75

have by and large accepted the terms ‘subordinate clause’ or ‘dependent clause’ to designate the protasis. We argue that this is misleading since the protasis is the founding clause of a conditional statement and thus cannot be termed ‘dependent’. Citing the Cambridge International Dictionary of English, Declerck and Reed (2001) argue that a ‘subordinate’ clause is the one which cannot form a separate sentence but which can form a sentence when joined with a main clause. In the present study we do not follow this line of thought. 6. See Johnson-Laird 1986. 7. Elisabeth Traugott, Alice Ter Meulen, Juddy Snitzer Reilly, and Charles A. Ferguson, 1986, p. 9. 8. As mentioned in Sharma (2010), our claim is based exclusively on the data discussed in the following typological studies: (1) Typology of Conditional Constrctions, Victor S. Xrakovskij (ed.), (2005) for Bulgarian (Rousselina Nicolova), Armenian (Natalia A. Kozintseva), Dari (Boris Ya. Ostrovsky), Hindi (Tatayana I. Oranskaya), Homeric Greek (Ilja A. Perelmouter), Early Latin (Margarita K. Sabaneyeva), French (Elena E. Kordi), German (Svetlana M. Kibardina), English (Tatiana G. Akimova, Natalia A. Kozintseva), Finnish (Hannu Tommola), Estonian (Irina P. Külmoja), Hungarian (László Jaszai, Ethelka Tóth), Hausa (Myrrah A. Smirnova, Nikolaj A. Dobronravin), Klamath (Viktor A. Stegniy), Indonesian (Alexander K. Ogloblin), Cambodian (Natalia M. Spatari), Vietnamese (Igor S. Bystrov, Nonna V. Stankeviþ), Chinese (Tamara N. Nikitina), Even (Andrej L. Malchukov), Evenki (Igor V. Nedjalkov, Nina Ya. Bulatova), Eskimo (Nikolaj B. Vaxtin), Aleut (Evgeniy V. Golovko), Yukaghir languages (Elena S. Maslova) and Japanese (Vladimir M. Alpatov, Vera I. Podlesskaya); (2) The semantics of Clause Linking, R. M. W. Dixon, and Alexandra Y. Aikhenvald (eds.), (2009) for Akkadian, Galo, Khan, Manambu, Iquito, Aguaruna, Ojibwe, Fijian, Toqabaqita, Martuthunira, Korean, Goemai, Konso and Mali; (3) The Indo-Aryan Languages, Colin P. Masica, (1991) for various Indo-Aryan languages. None of the languages discussed in above mentioned works seems to have ‘apodosisprotasis’ ordering with overt markers. 9. Donaldson (1980: 251-2), cited in Comrie (1986: 84). 10. In logic, conditionals (i.e. material implications: PĺQ) are defined as a relation between P and Q which is said to be true in the following three cases: (1) P and Q are true (2) P is false and Q is true (3) both P and Q are false. The relation is false only when P is true and Q is false. In bi-conditionality (PļQ), the relation is false in the second case: P is false and Q is true. Geis and Zwicky (1971) were perhaps the first to notice this and to claim that the actual interpretation of many conditional sentences is a bi-conditional (‘if and only if’) interpretation. Thus, If it rains, they will cancel the game (P, Q) is taken to mean No rain, no cancellation (¬P, ¬Q). Likewise, Iatridou (1991, 1994) ar-

76

11. 12.

13. 14. 15.

16.

17.

18. 19.

20.

Ghanshyam Sharma gues that the use of then in a conditional statement suggests that an iff-reading is not false. Bhatt and Pancheva (2006) cite these examples from R. Izvorski, The Present Perfect as an Epistemic Modal. In A. Lawson (ed.) Proceedings of SALT VII, Cornell University, Ithaca, NY, Cornell Linguistic Club, 1997. According to Iatridou (1991, 1994) then in conditional statements is not vacuous and is associated with a particular presupposition. A conditional of the form ‘If P, then Q’ presupposes that some of the ¬P are ¬Q. We assume that further supportive evidence may come form other Indo-Aryan languages, although we have made no attempt in this direction. Named so after Austin’s famous example, “If you are hungry, there are biscuits on the sideboard.” (Austin 1961) In this table, P and Q stand for the propositions contained in the protasis and the apodosis respectively. Į and ȕ are overt markers of the protasis and the apodosis respectively. The element within brackets () is not-obligatory. Ø is a null marker. As mentioned above, Hindi conditionals require the so-called apodosis marker to ‘then’ to be used always, no matter what the cluase ordering is (i.e. [agarPĺ toQ] or [Q-[agar-Pĺ to]]). Nevetheless, it is possible to encounter examples such as the following one where the use of to in [Q-[agar-Pĺ to]] is not obligatory: mѓ˾/zarnjr hƯ/ pƗr‫ܒ‬Ư mͅ/ gayƗ hotƗ/ agar/ usne/ mujhe/ bulƗyƗ hotƗ I/ certainly EMP/ party=in/ go-PFV-M-SG aux-IMFV-M-SG/if/ he-ERG/ invitePFV-M-SG aux-IMFV-SG ‘I certainly would have gone to the party if he/she had invited me.’ In this discussion we are concerned exclusively with the protases which contain some epistemic elements. Thus, we do not attempt to analyze conditional statements such as “If you open the window, I will kiss/kill you” where the protasis carries the speaker’s indirect illocutions, invitation/prohibition to carry out the task rather than speaker’s epistemic stand. As in the table 1, P and Q stand for propositions of the protasis and the apodosis respectively. K and B stand for ‘knowledge’ and ‘belief’ and ¬ is a negative marker. The arrow indicates the relationship between P and Q. We do not consider example such as If you love Clara so much, why don’t you marry her a counter-example even if in such cases the speaker may know the facts reported in the protasis since it is the speaker’s epistemic comunicative stand that counts not what he may know or believe. Examples for P-clause and Q-clause are provided in the Appendices 1 and 2, respectively.

On the role of protases in conditional statements

77

References Akatsuka, Noriko 1986 Conditionals are discourse-bound. In On Conditionals, Elizabeth Closs Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson (eds.), 333–351. Cambridge: Cambridge University Press. Austin, John L. 1961 Ifs and cans. In Austin, J. L., Philosophical Papers (3rd edition, 1979, edited by J. O. Urmson and G. J. Warnock). Cambridge: Cambridge University Press, 153–180. Bhatt, Rajesh, and Roumyana Pancheva 2006 Conditionals. In The Blackwell Companion to Syntax, Vol. 1, Martin Everaert, and Henk van Riemsdijk (eds.), 638–687. Oxford: Blackwell. Comrie, Bernard 1986 Conditionals: A typology. In On Conditionals, Elizabeth Closs Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson (eds.), 77–99. Cambridge: Cambridge University Press. Dancygier, Barbara, and Eve Sweetser 1997 Then in conditional constructions. Cognitive Linguistics 8–2: 109– 136. Declerck, Renaat, and Susan Reed 2001 Conditionals, A Comprehensive Empirical Analysis. Berlin: Mouton de Gruyter. Fauconnier, Gilles 1994 Mental Spaces. Aspects of meaning construction in natural language. Cambridge: Cambridge University Press. [1st Edition, 1985, Cambridge: The MIT Press. Geis, Michael, and Arnold M. Zwicky 1971 On invited inferences. Linguistic Inquiry 2: 561–566. Greenberg, Joseph H. 1963 Some universals of grammar with particular reference to the order of meaningful elements. In Universals of language, Joseph H. Greenberg (ed.). Cambridge, Mass.: MIT Press. Haiman, John 1978 Conditionals are topics. Language 54: 565–589. 1986 Constraints on the form and meaning of the protasis. In On Conditionals, Elizabeth Closs Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson (eds.), 215–227. Cambridge: Cambridge University Press. Iatridou, Sabine 1994 On the contribution of conditional then. Natural Language Semantics 2: 171–199.

78

Ghanshyam Sharma

Johnson-Laird, P. N. 1986 Conditionals and mental models. In On Conditionals, Elizabeth Closs Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson (eds.), 55–75. Cambridge: Cambridge University Press. König, Ekkehard 1988 Concessive connectives and concessive sentences: cross-linguistic regularities and pragmatic principles. In Explaining Language Universals, John A. Hawkins (ed.), 145–166. Oxford: Basil Backwell. Levinson, Stephen C. 2000 Presumptive Meanings – The Theory of Generalized Conversational Implicature, Cambridge, MA : MIT Press. Lycan, William G. 2001 Real Conditionals. Oxford: Oxford University Press. Masica, Colin P. 1991 The Indo-Aryan Languages. (Cambridge Language Surveys) Cambridge: Cambridge University Press. McGregor, Ronald Stuart 1995 Outline of Hindi Grammar. Oxford: Oxford University Press. Monatut, Annie 2004 A Grammar of Hindi. (LINCOM Studies in Indo-European Linguistics) Munich: Lincom Europa. Oranskaya, Tatyana I. 2005 Conditional constructions in Hindi. In Typology of Conditional Constructions, Victor S. Xrakovskij (ed.), 218–245. (LINCOM Studies in Theoretical Linguistics 25.) Munich: Lincom Europa. Sharma, Ghanshyam 2008 A pragmatic account of the Hindi presumptive. In Annual Review of South Asian Languages and Linguistics, Rajendra Singh (ed.), 83– 113. Berlin: Mouton de Gruyter. 2010 On Hindi conditionals. In Annual Review of South Asian Languages and Linguistics, Rajendra Singh (ed.), 107–134. Berlin: Mouton de Gruyter. Traugott, Elizabeth Closs, Alice ter Meulen, Judy Snitzer Reilly and Charles A. Ferguson (eds.) 1986 On Conditionals. Cambridge: Cambridge University Press. Van der Auwera, Johan 1986 Conditionals and speech acts. In On Conditionals, Elizabeth Closs Traugott, Alice ter Meulen, Judy Snitzer Reilly, and Charles A. Ferguson (eds.), 197–214. Cambridge: Cambridge University Press. Xrakovskij, Victor S. (ed.) 2005 Typology of Conditional Constructions. (LINCOM Studies in Theoretical Linguistics 25.). Muenchen: Lincom Europa.

Special Contributions: The Indigenous Languages of South Asia

Aspects of Kharia grammar: A Role and Reference Grammar approach John Peterson

Role and Reference Grammar (RRG) is a grammatical theory which makes a minimal number of assumptions with respect to human language. For example, although the theory allows for the presence of lexical classes such as noun and verb, it does not assume that all languages necessarily possess these categories. Rather, they are only assumed to the extent that they are necessary for the description of a particular language. In addition, as RRG does not consider grammatical relations such as “subject”, “object”, etc., to be universal, it does away with these entirely, making use instead of presumably universal notions such as the macroroles “Actor” and “Undergoer” and language- and construction-specific “privileged syntactic arguments” which can serve to organize much of a particular language’s morphosyntax. The present study employs the basic tenets of RRG to describe the South Munda language Kharia, a language which has been argued not to possess lexical classes such as noun, adjective and verb and also not to have any clearly definable grammatical relations such as “subject” or “object”. As such, RRG seems to be the natural choice for a theoretical discussion of this language. 1. Introduction This study presents an introduction to the South Munda language Kharia within the framework of Role and Reference Grammar (RRG), a monostratal theory of language which makes relatively few theoretical assumptions with respect to universal features of human language. Due to the enormous variation found within languages, such an approach offers a number of advantages when, e.g., categories which we are familiar with from English are not found in other languages. One example are grammatical relations such as subject and object, which many theories consider universal but which have proven enormously difficult to

82

John Peterson

define cross-linguistically and which have no place in RRG. Similarly, RRG makes no assumptions with respect to the universality of parts of speech, particularly lexical classes such as noun, verb and adjective, although it does allow for them. Although at first glance notions such as subject and object appear relatively straight-forward in Kharia, upon closer examination a number of problems surface if these categories are assumed for the language. Furthermore, as I have argued elsewhere, assuming the presence of lexical classes such as noun, adjective and verb in Kharia actually causes more problems than assuming that they do not exist, hence an approach such as RRG which makes no direct reference to these is clearly preferable over one which forces these categories onto the language. The study is structured as follows. Section 2 presents a brief overview of Kharia grammar. Section 3 then gives a very basic introduction to the principles of RRG, which are then applied to Kharia in Section 4. Section 5 provides a summary of the main results. 2. Brief overview of Kharia Kharia is a member of the southern branch of the Munda family, which forms the western branch of the Austro-Asiatic phylum. It is spoken predominantly in southwestern Jharkhand, eastern Chhattisgarh and northwestern Orissa in India. The present study deals with the variety spoken in and around the city of Simdega in the district of the same name in southwestern Jharkhand. Kharia is predominantly predicate-final and almost all grammatical categories such as person/number/honorific status, TAM, case, etc., are marked by enclitics; only very few categories, such as the causative, are marked by affixes. Like other Munda languages, Kharia has basic voice, i.e., there is an active/middle opposition in most TAM categories. For our purposes it will suffice to note that most predicative morphemes which may appear in both voices are transitive in the active voice and intransitive in the middle voice.1 In the variety of Kharia dealt with here there is no direct evidence for assuming the presence of lexical classes such as noun, adjective and verb. Instead, the lexicon can more conveniently be divided into two classes: The first is the open class, consisting of contentive morphemes. These can all be used in predicative, referential and attributive function without any

Aspects of Kharia grammar

83

derivational morphology, “light verbs”, etc. Consider by way of example the use of lebu ‘man; human’ in examples (1)–(3).

(1) lebu яel=ki. man

come=MID.PST

‘The / a man came.’

(2) bhagwan lebu=ki God

ro

яel=ki.

man=MID.PST and come=MID.PST

‘God became man [= Jesus] and came [to earth].’ [adapted from Malhotra, 1982:136]

(3) lebu human

jati2 caste

‘humanity’ The second group is the closed class. This consists of two subclasses: 1. deictic forms such as iяaҌ ‘yesterday’ and forms translating as pronominals in English, such as iѪ ‘1SG’, am ‘2SG’, etc. These morphemes may all be used in predicative and referential function without any “class changing” derivational morphology, light verbs, etc. (cf. (4)), but require the genitive to appear in attributive function.

(4) (in a play about me and you, in which both of us will be taking part): “naҀak=te iѪ=ga

ro am=ga

ho=kaѵ=na=iѪ

play=OBL 1SG=FOC that=SG.HUM=MID.IRR=1SG

iѪ=na=m.” 1SG=MID.IRR=2SG

“umboҌ. no

and 2SG=FOC

am=na

um=iѪ

pal=e.

2SG=INF

NEG=1SG

be.able=ACT.IRR

яirekҀar

seƾ=gaҌ

iѪ=te

ho=kaѵ=oҌ.

director

early=FOC

1SG=OBL

that=SG.HUM=ACT.PST

am=ga

am=na=m.”

2SG=FOC

2SG=MID.IRR=2SG

‘“In the play I will be him and you will be me.” “No. I can’t be you. The director already made me him. You will be you.”’ 2. functional morphemes such as case markers, TAM, person, etc., which cannot appear in any of these three functions.

84

John Peterson

Figure 1 presents a schematic overview of the Kharia lexicon. Lexicon Open class (contentive morphemes, may be used in all functions)

Closed class

Proforms / Deictics (referential and predicative function, used attributively only if marked by the genitive)

Grammatical Morphemes (never used referentially predicatively, or attributively)

Figure 1. The Kharia lexicon (from Peterson, 2011: 78)

For reasons of space, the present study will not attempt to demonstrate yet again the “precategoriality” of Kharia contentive morphemes.3 Instead, (5), from Peterson (in press), merely presents a number of illustrative examples from various (intuitive) semantic categories demonstrating the general flexibility of the open class.

(5) Interrogatives: i ‘what; which; do what?’ Indefinites: jahã ‘something; some (attributive); do something’ Quantifiers: moѪ ‘one (referential/attributive); become one’ Properties: rusuƾ ‘red one; red (attributive); become red’, maha ‘big one; big; grow, become big’ Proper names: aҌghrom ‘Aghrom (name of a town) (referential/attributive); come to be called “Aghrom” (middle voice), name [some place] “Aghrom” (active voice)’ Status and Role: ayo ‘mother; become a mother (middle voice), accept someone as a mother (active voice)’ Physical objects and animate entities: kaяoƾ ‘fish; fishy; become a fish (middle), turn s.th./s.o. into a fish (active)’ Locative: tobluƾ ‘top; rise (middle), raise (active)’ Activities: siloҌ ‘plowing (n.); plowed; plow’ However, the “precategoriality” of contentive and deicic morphemes is only one aspect of this issue: Not only can individual contentive morphemes appear in all three functions, entire (non-endocentric)

Aspects of Kharia grammar

85

phrases can as well, due at least in part to the enclitic nature of grammatical markers, which attach to the last element of these “phrases” and not to “stems” or “roots”. Cf. (6)–(9).4

(6) [ho rochoҌb]=ki=Ѫ that

side=MID.PST=1SG

‘I went to that side’

(lit.: “I that-side-d”)

(7) [ubar rochoҌb]=ki=Ѫ 2

side=MID.PST=1SG

‘I moved to both sides’ (i.e., this way and then that)

(8) iѪ

ho=kaѵ=te

[iѪ=aҌ]=yoҌj.

1SG that=SG.HUM=OBL 1SG=GEN=ACT.PST.1SG

‘I adopted him/her (i.e., I made him/her mine).’

(9) [oҌ=yaҌ

teloƾ]=ki.

[oҌ=yaҌ

teloƾ]=oҌ=ki.

house=GEN roof=MID.PST

house=GEN roof=ACT.PST=PL

‘The house’s roof was thatched.’

‘They thatched the house’s roof.’

As I have argued elsewhere (e.g., Peterson, 2005; 2007; 2011; in press), although it may be possible to assume that Kharia has nouns and verbs, it is more economical to simply assume the presence of two structurally defined categories, the CASE- and the TAM/PERSON-syntagmas. These two units are defined by the grammatical marking which attaches to the unit as a whole, i.e., either case or TAM/person marking, whereas the unit they attach to has the same potential structure in both syntagmas. Put differently, these units have two heads, a semantic head, with the same potential structure in both units, and a functional head, which defines the status of the unit as a whole. This is illustrated schematically in Figure 2.5 Common structure of CASE- and TAM/PERSON-syntagmas

Semantic Head

Functional Head

Figure 2. Common structure of TAM/PERSON- and CASE-syntagmas in Kharia (Peterson, 2011: 71).

86

John Peterson

The semantic head has the (simplified) potential structure given in Figure 3. The Kleene star (“*”) indicates that any number of lexemes, including zero, may be present; non-final lexemes modify the lexeme to their right, similar to the use of lebu ‘human’ in (3). Note that possessive and number marking as well as the genitive are part of the semantic base, not the functional head of the CASE-syntagma, as these can appear in both TAM/PERSON- and CASE-syntagmas. 6 (LEX=GEN) (DEM) (QUANT (CLASS)) (LEX=GEN) (LEX*) (=POSS) (=NUM) or PRO Figure 3. The structure of the semantic base of CASE- and TAM/PERSONsyntagmas.

In Peterson (in press), this is formalized as in (10), where the unit portrayed in Figure 3 is referred to as the “content head” or “CONTHEAD”. “PRO” here refers to the fact that this entire unit may be replaced by a deictic element. Note also that this unit is recursive, i.e., there can be CONTHEADs contained within a CONTHEAD (e.g., as a genitive attribute), marked here for ease of presentation as CONTHEADGEN,8 or in conjoined CONTHEADs, similar to English those two old chairs and these three new tables.

(10) CONTHEADo

(CONTHEADGEN) (DEM) (QUANT (CLASS)) (CONTHEADGEN) (LEX*) (POSS) (NUM) (GEN) › PRO (GEN)

The CASE-syntagma consists of this unit plus case marking, where the Kleene plus sign “+” denotes that any number of CONTHEADs may be present, as long as at least one is present.

(11) CASE-SYNTAGMA o CONTHEAD+

CASE

“Case” refers to one of the following markers which denote the relation of the CASE-syntagma to the predicate. Note again that the genitive is not considered a case here but is part of the content head. Direct (unmarked)  Subjects (A / S), non-definite countable and noncountable direct objects (P). As we shall see below, the

Aspects of Kharia grammar

87

TAM/PERSON-syntagma is always overtly marked as such, hence an unmarked clause-level unit is automatically interpreted as a CASEsyntagma in the direct case Oblique (=te)  definite P, “indirect objects” (recipient, goal), temporal and locative adjuncts Postpositions – adjuncts, e.g., buƾ ‘INST’, tay ‘ABL’, sori ‘COM’. (12) and (13) provide two simple examples of CASE-syntagmas.

(12) bides=aҌ

lebu=ki=yaҌ

abroad=GEN person=PL=GEN

rupraƾ=te appearance=OBL

‘the appearance of foreigners’ (e.g., as the object of a clause)

(13) ho

pãc

that five

lebu=ki person=PL

‘those five persons’ (e.g., as subject)

[Biligiri, 1965: 77]

The structure of the TAM/PERSON-syntagma is somewhat more complex. In Peterson (in press), the phrase-structure rules which account for this structure are as follows: This unit consists of one or more “TPBASES” (denoted here by the Kleene plus sign) plus subject marking. This is given in (14).

(14) TAM/PERSON-SYNTAGMA o TP-BASE+

SUBJ

Table 1 provides an overview of the enclitic subject markers in Kharia.9 Table 1. Markers for Person/Number/Honorific status Singular 1 2 3

=(i)Ѫ =(e)m



Dual /HON Inclusive Exclusive =naƾ =jar =bar (=kiyar)

Plural Inclusive Exclusive =niƾ =le =pe (=ki / =may)

The TP-BASE consists of a perfect base (PERFBASE) and “TV”, a combination of TAM and basic voice markers (BV). Finally, the perfect base consists of the telic V2s and the perfect, both of which are optional. This is summarized in (15)–(19) and Table 2.

(15) TP-BASE o PERFBASE

TV

88

John Peterson

(16) PERFBASE o CONTHEAD (TEL) (PERF) (17) TV = (TAM š BV) › OPT › PSTII10 (18) TAM= (PAST › PRESENT › PROGRESSIVE › IRREALIS) (19) BV = ACTIVE › MIDDLE Table 2. : “TV PAST š ACTIVE

=oҌ =ki =te =ta =teҌj =taҌj =e =na guяuҌ =khoҌ

TAM š BASIC VOICE

PAST š MIDDLE PRESENT š ACTIVE PRESENT š MIDDLE PROGRESSIVE š ACTIVE PROGRESSIVE š MIDDLE IRREALIS š ACTIVE IRREALIS š MIDDLE OPT PSTII

These rules account for structures such as (20), with a simple content head, and also (21) and (22), with complex content heads with varying degrees of finite marking of the non-final unit. Structures such as that in (21) are possible since CONTHEAD is a potentially recursive category, to which TV is conjoined to form a TP-BASE, whereas examples such as (22) are possible as more than one TP-BASE – which is marked for TV but not for the subject – can be present.

(20) яoko=ki sit.down =MID.PST

‘[he or she] sat down’

(21) tar ol=e=pe kill bring=ACT.IRR=2PL

‘kill [the animal and] bring [it back]!’

(22) u=kiyar taƾ el=aҌ this=DU

now 1PL.EXCL=GEN

sori=ga Ѫog=e COM=FOC

uя=e=kiyar.

eat=ACT.IRR drink=ACT.IRR=DU

‘They two will now eat and drink with us.’ [adapted from Roy & Roy, 1937: 180f.]

Aspects of Kharia grammar

89

(23) and (24) present examples of the content head bidesaҌ lebukiyaҌ rupraƾ, 'the appearance of foreigners” in a CASE-syntagma and in a TAM/PERSON-syntagma, respectively (cf. Peterson (in press) for further examples).

(23)

CASE-SYNTAGMA (= example (12)) CONTHEAD CONTHEAD CONTHEAD LEX

bides

LEX

CASE LEX

NUM GEN

GEN

=aҌ lebu =ki =yaҌ

rupraƾ

(24)

=te

TAM/PERSON-SYNTAGMA TP-BASE PERFBASE

SUBJ TV

CONTHEAD CONTHEAD CONTHEAD LEX

LEX

LEX

NUM GEN

GEN

bides

=aҌ

abroad

=GEN person =PL =GEN

lebu =ki =yaҌ

rupraƾ

=ki

=may.

appearance =M.PST =3PL

‘They took on the appearance of foreigners (e.g. by living abroad so long).’ Having given a general, theory-neutral overview of Kharia, Section 3 provides an introduction to Role and Reference Grammar (RRG) which

90

John Peterson

will serve as the basis for our discussion of Kharia within the framework of this theory in Section 4. 3. Introduction to RRG Role and Reference Grammar (RRG) is a monostratal theory of language based on two general considerations of clause structure (Van Valin, 2005: 3):  A theory of clause structure should capture all of the universal features of clauses without imposing features on languages in which there is no evidence for them.  A theory should represent comparable structures in different languages in comparable ways. RRG has three main levels of representation: syntactic structure, corresponding closely to the actual form of the utterance, semantic representation and information structure (Van Valin, 2005: 1). All three levels are equally important and none is considered more basic than the other two. Nonetheless, for reasons of space, the present study concentrates on syntactic structure, referring to the semantic representation only as this is necessary to understand the syntactic structure, while a discussion of information structure in Kharia must be left for a future study.11 Syntactic structure As a monostratal theory, RRG does not derive actual sentences from an underlying level, hence it also does not postulate movement or traces. Syntactic structure in RRG is based on the notion of the “layered structure” of the clause and on two further fundamental contrasts: arguments/nonarguments and predicating/non-predicating elements (Van Valin, 2005: 4). These are viewed as universal features of human language and hence figure prominently in RRG. This is shown schematically in Figure 4.

Predicate

+ Arguments

Non-Arguments

Figure 4. Universal opposition underlying clause structure (Van Valin, 2005: 4).

Aspects of Kharia grammar

91

Structurally speaking, the basic unit of the clause is the nucleus, which contains the predicate. This unit, together with the arguments, is referred to as the core. Non-arguments are found in the periphery, which combines with the core to form the clause. This is illustrated in Figure 5. CLAUSE CORE PERIPHERY

NUCLEUS

CLAUSE CORE ARGUMENTS Dana

saw

Pat

PERIPHERY ADJUNCTS yesterday

in the library

NUCLEUS Figure 5. Components of the layered structure of the clause (Van Valin, 2005: 4).

Of central interest for the present study is that RRG makes no assumptions with respect to lexical classes: Arguments, adjuncts and predicates may all be nouns, adjectives, verbs, or none of the above. The status of these units is a language-specific feature and is of no interest for the general theory. Furthermore, no assumption is made that a language will even possess these lexical classes. All that is required is that all structures which are obligatory in a particular language be represented in the structural representation of that language. (25), from Van Valin (2005: 30, simplified and given here without the operator projection), provides a simple English example. The arrow from the periphery to the core denotes that the periphery modifies the core and is not an argument.

92

John Peterson

(25)

SENTENCE CLAUSE CORE NP

NUC

PERIPHERY NP

PRED V The man

PP

speared the wallaby in the mountains

Finally, in many so-called head-marking languages the predicate, with its morphological marking for units such as the subject, object, etc., can stand alone as the core (or even sentence). Consider (26), from Van Valin (2005: 16, there from a work by Judith Aissen), from Tzotzil, a Mayan language, which can also be realized as (27).

(26) Ҍi-Ø-s-pet ASP-3ABS-3ERG-carry

loۖel Ҍantz

ti

Ģule.

away

DEF

rabbit-DEF

woman

‘The rabbit carried away the woman.’

(27) Ҍi-Ø-s-pet. ASP-3ABS-3ERG-carry

‘He/she carried him/her/it.’ As RRG does not presuppose any version of the lexical integrity principle, in these languages the marking on the predicate for subject, object, etc., is considered the actual arguments. Hence, any units (usually NPs) referring to these arguments are core-external non-arguments which only serve to further identify the units marked on the predicate. The differences, but also the similarities, between head-marking languages and languages such as English are shown in (28) and (29), based on (26) and (27) and their English translations.12 The main difference is that in English, unlike Tzotzil, the arguments are phonological and morphosyntactic words whereas they are affixal in Tzotzil. Furthermore, the NPs referring to these two units in the left-hand structure in (28) are core-external in Tzotzil, whereas in the left-hand structure of (29) the

Aspects of Kharia grammar

93

rabbit and the woman are core-internal arguments, as are he and her in the right-hand structure of (29).

(28) SENTENCE

SENTENCE

CLAUSE

CLAUSE

CORE

CORE

PRO PRO NUC

‫ݦ‬i- Ø-

(29)

s-

NP

PRO PRO NUC

PRED

PRED

V

V

pet loরel ‫ݦ‬antz

SENTENCE

‫ݦ‬i- Ø-

s-

pet

CLAUSE

CORE NUC

ti Ģule.

SENTENCE

CLAUSE

NP

NP

CORE NP

NP NUC

PRED

PRED

V

V

‘The rabbit carried away the woman.’

NP

He carried her.

This distinction between core-internal and core-external referential phrases (§ NPs) will be of importance again in the discussion of Kharia within the framework of RRG.

94

John Peterson

Grammatical relations RRG assumes neither subjects nor objects, which have been used here purely intuitively until now. Instead, it acknowledges two macroroles  Actor and Undergoer – which can play a role in the morphosyntax. These are determined by their respective positions within the logical structure (LS) of the respective lexical entry, to which we now turn. Predicates (including predicative adpositions) may be described in terms of Aktionsart features as being stative or dynamic, change of state, activity, etc. These are determined by the various tests described in Van Valin (2005, chapter 2). For our purposes the following types will suffice: “States are represented as bare predicates: know' (x, y), dead' (x). Activity verb representations all contain the element do', e.g. do' (x, [cry' (x)]) ‘cry’, do' (x, [eat' (x, (y))]) ‘eat’.” (Van Valin, 2005: 42). Note that do' marks that an action is carried out by some entity, whether or not this occurs volitionally. If the event denoted by a certain lexeme is necessarily agentive, the operator DO appears at the beginning of the LS. For example, kill in English is potentially agentive, whereas murder is obligatorily so (Van Valin, 2005: 56):

(30) The man accidentally killed his neighbour. (31) *The man accidentally murdered his neighbour. This difference is expressed as follows in the lexical entries of these two morphemes:

(32) kill

:

[do' (x, ‡)] CAUSE [BECOME dead' (y)]

(33) murder : DO (x, [do' (x, ‡)] CAUSE [BECOME dead' (y)])13 x and y are determined for a specific predicate by their respective positions in the continuum shown in Figure 6.

Aspects of Kharia grammar

Arg. of DO AGENT

1st arg. of do' (x, … EFFECTOR MOVER ST[ATIC]-MOVER L[IGHT]-EMITTER S[OUND]-EMITTER PERFORMER CONSUMER CREATOR SPEAKER OBSERVER USER

1st arg. of pred' (x, y) LOCATION PERCEIVER COGNIZER WANTER JUDGER POSSESSOR EXPERIENCER EMOTER ATTRIBUTANT

2nd arg. of pred' (x, y)

95

Arg. of state pred' (x)

THEME STIMULUS CONTENT DESIRE JUDGMENT POSSESSED SENSATION TARGET ATTRIBUTE PERFORMANCE CONSUMED CREATION LOCUS IMPLEMENT

PATIENT ENTITY

Figure 6. Thematic relations continuum in terms of logical structure argument positions (Van Valin, 2005: 58).

Once the lexical entry of a particular morpheme has been determined, the two macroroles can also be determined. In general, the Actor is that argument which is furthest to the left, the Undergoer that which is furthest to the right in Figure 7.14 ACTOR Arg. of DO

UNDERGOER 1st arg. of do' (x, …

1st arg. of pred' (x, y)

2nd arg. of pred' (x, y)

Arg. of state pred' (x)

Figure 7. Actor-undergoer hierarchy (Van Valin, 2005: 61).

For example, English eat has the structure do' (x, [eat' (x, (y)]), where x is Actor and y is Undergoer. The parentheses “( )” signal that y is optional here. On the other hand, use has the structure do' (x, [use' (x, y)]), where x is Actor and y, which is not optional, is Undergoer (Van Valin, 2005: 55). For lexical entries with only one macrorole, its status is determined as follows (Van Valin, 2005: 63): If the morpheme denotes an activity, i.e., its lexical entry contains a do', the macrorole is an Actor, otherwise it is an Undergoer. For example, the single macrorole of walk, with the structure

96

John Peterson

do' (x, [walk' (x)]), is an Actor, since the lexical entry contains a do'. exist, on the other hand, has the LS exist' (x), hence x here is an Undergoer. The lexical entry of a predicate may also have two arguments but only one macrorole. For example, be at involves a location and a theme, but has only one macrorole. This is indicated by adding [MR 1] for “one macrorole” to the lexical entry to denote that it exceptionally has only one macrorole, despite having two arguments, i.e., be-at' (x, y) [MR 1]. There is no need to indicate whether this is an Actor or Undergoer: As this lexical entry does not contain a do', its single macrorole is Undergoer and, according to Figure 7, in a two-place predicate, the best candidate for Undergoer status is the argument on the right, i.e., y in be-at' (x, y) [MR 1]. x is then a non-macrorole argument marked by the preposition at. Finally, it has often been noted that grammatical relations – above all the “subject” – tend to play a central role in syntax, controlling operations such as coreferential deletion of the type in (34), where [Ø] can only refer to The woman, i.e., the subject of the first clause.

(34) The womani saw the manj and [Øi/*j] left. Such operations are often termed “subject properties” in older typological works. One problem with this term, however, is that many languages do not have such properties. Furthermore, it is quite common for a language to have such “subject properties” for a number of syntactic processes, but with the different processes pointing towards different “subjects”. As RRG does not make use of the notions “subject” or “object”, it avoids these problems entirely. Instead, case marking and predicate marking are defined independently of one another and without any reference to such notions. Furthermore, the “pivot” or controller of various syntactic operations is referred to as the “privileged syntactic argument” or “PSA” in RRG. This is a language-specific and construction-specific unit which may control operations such as coreferential deletion, reflexivization, or “subject identity” with converbs, etc. Consider, e.g., the S/A pivot for coreferential deletion in example (34) above, i.e., in English only the S or A may be controller or the omitted NP between clauses. This is the PSA for this construction in English. There are two major types of PSAs involving the neutralization of the semantics of an intransitive argument, i.e., cases in which the status of an intransitive argument (S) as Actor or Undergoer is irrelevant. This is shown in Figure 8. Depending on the language and the particular operation, the

Aspects of Kharia grammar

97

choice of PSA may begin on the left and proceed to the right until a suitable argument is found (“accusative syntax”) or it may begin on the right and proceed to the left (“ergative syntax”). Thus, in English, those operations which are often considered “subject properties” are in effect all PSAs of the same type, i.e., the leftmost argument along the hierarchy in Figure 8.15 Arg. of DO > 1. Arg. of do' > 1. Arg. of pred' (x, y) > 2. Arg. of pred' (x, y) > Arg. of pred' (x) Figure 8. Privileged syntactic argument (PSA) selection hierarchy (Van Valin, 2005: 100).

The choice of a PSA can be further restricted in a particular construction, e.g., by limiting the PSA determined by Figure 8 only to nominative-case marked arguments. Another possibility is that the argument determined by Figure 8 may only be a PSA if it is also a macrorole argument, i.e., Actor or Undergoer, whereas in other languages it may also be a dative-marked argument (e.g., an experiencer in a so-called “dative subject” construction). Thus, PSAs may be restricted in some languages only to certain cases (e.g., nominative), or only to macrorole arguments, and all different types of PSA may be found for particular operations even within the same language. 4. Kharia from an RRG perspective In the present section, the principles of RRG discussed in the preceding section will be applied to Kharia.

4.1. Case marking in Kharia We can use the basic tenets of RRG to formulate rules for case marking in Kharia:

(35) Direct case  the highest macrorole argument is unmarked for case (= direct case); i.e., an Actor, if present, is unmarked. Otherwise, the Undergoer, as the single macrorole, is unmarked.

98

John Peterson

(36) Oblique case – If there is a second macrorole, it is mandatorily marked by =te if it is [+DEF/+HUM] or optionally if it is [+DEF/HUM]. [-DEF/-HUM] Undergoers are generally unmarked (= direct case). =te is also the default marker of non-macrorole arguments which do not qualify for instrumental marking (cf. (37)).

(37) Instrumental/Dative – The rule given in Van Valin (2005: 110) for the instrumental and dative also applies to Kharia if we equate “dative” here with the oblique marker =te. Note that Rule b, the default rule, only applies when Rule a does not: “a. Assign instrumental case to [a non-macrorole] b argument if, given two arguments, a and b, in a logical structure, with (1) both as possible candidates for a particular macrorole and (2) a is equal or higher (to the left of b) [in Figure 8], b is not selected as that macrorole. b. Assign [=te] to non-macrorole arguments (default).” (cf. (35)). These rules are illustrated in (38)–(40).16

(38) dura ruҌ=ki door

open=MID.PST

‘The door opened.’

LS: BECOME open' (dura)

 Only one macrorole argument, by default the highest macrorole argument (Undergoer, due to the absence of do') ĺ direct case.

(39) rayem dura=te Rayem door=OBL

rukh=oҌ open=ACT.PST

‘Rayem opened the door.’ LS: [do' (rayem, Ø)] CAUSE [BECOME open' (dura)]  rayem is the highest macrorole argument (Actor) and therefore appears in the direct case; dura is the lowest argument (Undergoer). As it is definite, it may be marked for the oblique case, as in (39).

(40) rayem saroj=te pothi=te ter=oҌ. Rayem Saroj=OBL book=OBL give=ACT.PST

‘Rayem gave Saroj the book.’ LS: [do' (Rayem, Ø)] CAUSE [have' (saroj, pothi)]

Aspects of Kharia grammar

99

 Here again, rayem is the highest macrorole argument (Actor) and appears in the direct case; pothi is the other macrorole argument, the Undergoer, and as it is [+DEF/-HUM], it can appear in the oblique case, as in (40). saroj in (40) is not a macrorole argument nor does it compete with rayem for Actor status with the predicative morpheme ter.17 According to Rule (37)b, it is therefore marked by the oblique marker =te. Note also the alternation in the LSs 0 and (39) for ruҌ ‘open’, reflecting the fact that (38) is intransitive with the middle voice and (39) is transitive with active voice. This alternation is quite productive in Kharia and can be represented as in (41). We return to this issue in Section 4.4.

(41) ruҌ : BECOME open' (x) (Middle voice)

ļ [do' (x, Ø)] CAUSE [BECOME open' (y)] (Active voice)

4.2. CASE- and TAM/PERSON-syntagmas within the framework of RRG As noted in Section 3, RRG considers lexical categories to be languagespecific categories whose existence in a particular language is an empirical question, not a universal feature of language. In recent work on RRG (e.g., Van Valin, 2005: 28), it has therefore become customary to speak not only of “predicates” – which may or may not be verbal – but also of “referential phrases” or “RPs”, which are generally – although not necessarily – realized as NPs. I will adhere to this practice and speak only of predicates and RPs in the following. As the only elements which are required in a structural analysis in RRG are those structures which are obligatory in the respective language, this means that the TAM/PERSON- and CASE-syntagmas discussed in Section 2 will also be used in our RRG analysis. The only exception to this is the PERFBASE, which is not an obligatory part of the TAM/PERSON-syntagma and which can hence be omitted. Thus, in our format, TP-BASE consists directly of CONTHEAD and TV, both of which are obligatory (cf. once again (15) and (16)). Consider once again the CASE-syntagma given in (11). This consists of at least one element from the content head, given in (10), plus case marking. This is illustrated in an RRG structural representation in example

100

John Peterson

(42). “NUC” is always the right-most element of the content head up to and including LEX.18

(42)

CASESYNTAGMA (Structure of example (23)) CONTHEAD CASE NUC

bides=aҌ

lebu=ki=yaҌ

abroad=GEN person=PL=GEN

rupraƾ

=te

appearance

=OBL

Recall from (14) that the TAM/PERSON-syntagma consists of subject marking (Table 1) plus the TP-BASE. The latter then consists of the TAM/BASIC VOICE markers shown in Table 2 and the PERFBASE, which itself consists of the content head plus optional marking for telicity (“V2s”) and the perfect. However, as just noted, since neither of these two latter categories is obligatory, they are not included in the structure representation. Instead, they are considered (nuclear) operators and will not be discussed further. The distinction between the TAM/PERSON-syntagma and the TP-BASE is necessary as subject marking is not always directly attached to the right side of the TP-BASE but may also directly precede it, e.g., in negation: Unlike in non-negated TAM/PERSON-syntagmas, where the subject marker always follows the TP-BASE, the subject in negated TAM/PERSONsyntagmas is obligatorily marked on the left side of the TP-BASE in all persons other than the second person, singular. In the second person, singular, it may appear on either side, either as an enclitic to the negative marker preceding the TAM/PERSON-syntagma or on the right-hand side of the TAM/PERSON-syntagma, as in (43).

(43) ubhroƾ these.days

um=em яe=na NEG=2SG

OR

come=MID.IRR

ubhroƾ

um яe=na=m.

these.days

NEG

come=MID.IRR=2SG

‘These days you do not come.’ (adapted from Malhotra, 1982: 285)19 As I will argue in more detail in Section 4.4, the TAM/PERSONsyntagma is in fact the entire core of its respective clause and the enclitic marker is its subject, i.e., it is an RP.

Aspects of Kharia grammar

101

As in most cases the TAM/PERSON-syntagma is the predicate of that clause, the content head will generally contain the nucleus of the predicate, while TV will contain marking for the core, clause and sentence-level operators.20 (44) and (45) present two examples of TAM/PERSONsyntagmas.

(44)

TAM-SYNTAGMA (Structure of example (24)) TP-BASE CONTHEAD

RP

TV

NUC bides=aҌ

lebu=ki=yaҌ

abroad=GEN person=PL=GEN

(45)

rupraƾ

=ki

=may

appearance

=MID.PST

=3PL

TAM-SYNTAGMA (Structure of example (21)) TP-BASE CONTHEAD

RP TV

NUC

NUC

tar kill

ol bring

=e =pe =ACT.IRR =2PL

The following section discusses the few PSAs found in Kharia before Section 4.4 examines the status of overt RPs as core arguments. 4.3. “PSAs” in Kharia There are a small number of morphosyntactic operations in Kharia which make use of a PSA, although the PSA is not the same for all operations. These will now be discussed individually.

102

John Peterson

4.3.1. Person / Number / Honorific marking on the predicate “Subject” marking in Kharia is based on a nominative/accusative pattern: The predicate is marked for the person / number / honorific status of A/S, as opposed to P, in general typological terminology. In RRG terms, this means that the highest ranking macrorole argument is marked on the predicate – Actor, if there is one, otherwise the Undergoer. (46) provides a simple example; its LS is given in (47). As rayem is the highest macrorole argument, it is chosen as PSA for predicate marking.

(46) rayem saroj=te pothi=te ter=oҌ(=Ø). Rayem Saroj=OBL book=OBL give=ACT.PST(=3SG)

‘Rayem gave Saroj the book.’

(47) [do' (rayem, Ø)] CAUSE [BECOME have' (saroj, pothi)] This is the unmarked PSA in Kharia with respect to marking on the predicate. It is possible, however, to make the Undergoer of a transitive clause the PSA, i.e., pothi in (46). The Actor is then demoted to adjunct status or omitted entirely, and the Undergoer is now the PSA and therefore marked on the predicate.

(48) Passive: Saroj=te

pothi ter яom=ki(=Ø).

Saroj=OBL book

give

PASS=MID.PST(=3SG)

‘A book was given to Saroj.’ This corresponds to what Van Valin (2005: 116) refers to as both “PSA modulation voice”, as it permits an argument other than the default argument to be the PSA, and also “Argument modulation voice”, as it gives non-canonical realization to a macrorole argument, as the Actor of the unmarked construction is an adjunct in the passive.21 Only the Undergoer, i.e., the right-most semantic argument according to Figure 8, can be promoted to PSA in the passive, while other, nonmacrorole arguments may not. For example, the recipient of ter ‘give’, although an argument, may never be the PSA with respect to predicate marking as it is not a macrorole, hence (49) is ungrammatical, whereas (48) above is grammatical.

Aspects of Kharia grammar

(49) Passive *am 2SG

103

pothi=te ter яom=ki=m. book=OBL give

PASS=MID.PST=1SG

‘You were given the book.’ 4.3.2. Control constructions / “Equi-NP deletion” We find the same PSA in complex predication with “auxiliaries” such as try, be able, want, etc., which take an infinitive. In these predicates, the S/A of the infinitive may not be overtly expressed and is obligatorily interpreted as the enclitic subject (= S/A) of the modal auxiliary, i.e., the highest macrorole argument.

(50) co=na go=INF

lam=t=iѪ. want=ACT.PRS=1SG

‘I want [I / *you / *he / *she /…] to go.’ If the A/S of the infinitive is different than the PSA of the modal auxiliary, this must be expressed by another construction. In the case of lam, a finite complement clause is used.22

(51) lam=t=iѪ want=ACT.PRS=1SG

no

ho=kaѵ

co=na.

CMPL

that=SG.HUM

go=MID.IRR

‘I want him/her to go.’ 4.3.3. Reflexivization There are two types of reflexivization in Kharia, which PSAs. The first type is expressed directly on the predicate яom which can denote either a passive, as we saw above, yield a reflexive interpretation. Thus, (52) can have interpretations, depending on the context.

have different by the marker or which may two different

(52) yo яom=ki=kiyar. see

REFL=MID.PST=DU

‘They saw themselves (e.g., in the mirror) / they were seen.’ Although the data for this reflexive construction are rather meager, it appears that the PSA for this type of reflexivization is the same as that found in predicate marking and in control constructions, i.e., the highest macrorole according to Figure 8. This holds for (53), with an indirect

104

John Peterson

reflexive interpretation, and (54), with a direct reflexive interpretation, in both of which the controller of reflexivization is the “subject” marked on the predicate itself.

(53)… taҌj distribute

яom=ki=may

komaƾ=te.

REFL=MID.PST=3PL

meat=OBL

‘… they distributed the meat amongst themselves.’

(54) koяpuѵuҌ=ki

gam=na=Ҍ

hosiyar

man=PL

say=INF=GEN

intelligent say

gam

[AK, 1:56]

яom=ta=ki. REFL=MID.PRS=PL

‘Men call themselves intelligent in (= of) speech.’ [Kerke‫ݚݚ‬Ɨ,1990: 12] In the second construction, which does not appear to differ semantically from яom, reflexivization is expressed syntactically. Here, the reflexive proform apan is used, borrowed from Kharia’s Indo-Aryan neighbor Sadri. It generally refers to an agentive subject, as in (55).

(55) khaѵiya apan raƾ=te Kharia

REFL

bancay guяu[Ҍ].

culture=OBL save

OPT

‘The Kharia should save their own culture.’

[MS, 1:263]

However, unlike morphologically expressed reflexivity, the controller of syntactic reflexivity can also be an experiencer (56). It may not, however, refer to a recipient (57).

(56) ho=kaѵ=te that=SG.HUM=OBL

apan=aҌ

ma=яom

REFL=GEN

mother=3POSS memory come=MID.PST

iyed

яel=ki.

‘S/he remembered his/her own mother.’

(57) rayem saroj=te apan=aҌ Rayem Saroj=OBL

REFL=GEN

pothi=te o-y-eƾ=oҌ. book=OBL

CAUS-y-return=ACT.PST

‘Rayemi gave Sarojj back heri, *j book.’ All this suggests that apan in Kharia refers back to the highest argument, whether or not this is a macrorole argument: In (55), whose LS structure is given in (58), the khaѵiya are both the highest argument and a macrorole (Actor). In (56), whose LS is given in (59), hokaѵte is not a macrorole, although it is the highest argument in the LS and with that the PSA of reflexivization, i.e., what is often referred to as a “dative subject”.

Aspects of Kharia grammar

105

In (57), however, whose LS is given in (60), rayem is both macrorole (Actor) and highest argument in the LS, whereas saroj is neither. Hence, only rayem qualifies as PSA.23

(58) do' (khaѵiya, [save' (khaѵiya, raƾ)]) (59) remember' (3SG, maяom) [MR 1] (60) [do' (rayem, Ø)] CAUSE [BECOME have' (saroj, pothi)] Thus, despite the similarities between the two reflexive constructions, there is nevertheless a slight difference. Although the data are still somewhat meager for morphologically expressed reflexivization, it appears that this type of reflexivization is controlled by the highest macrorole argument, whereas syntactically expressed reflexivization is controlled by the highest argument, regardless of whether this element is a macrorole. 4.3.4. Other operations With the exception of the PSA for predicate marking, control constructions in complex predicates and reflexivization, no other constructions seem to possess a PSA in Kharia. Thus operations which are generally considered “subject properties”, such as coreferential deletion across clauses and “subject identity” with converbs (= “conjunctive participles”) do not have a PSA in Kharia. At first glance, coreferential deletion might seem to follow an S/Apivot, where the A/S of the one clause is identical with that of the next, as in (61).

(61) ro

beҌҀ=яom=ki=jo

and son=3POSS=PL=ADD

umay NEG.3PL

juƾ=oҌ ask=ACT.PST

no

i

CMPL

what animal

jhãut

heke

lekin kaҌ kom=ki dhoҌ=ke [‡] muҌ

goҌя=ki=may.

QUAL.PRS

but

C:TEL=MID.PST=3PL

bow arrow=PL grab=SEQ

emerge

‘And his sons also didn’t ask which animal it is, but, having taken their bows and arrows, [they] went off.’ [AK, :11] However, this is not necessarily so. In (62) the subject of the second clause refers to another group of people who carry the Kharia men around, not the Kharia men, who are the subject of the first clause.

106

John Peterson

(62) laҌ … khaѵiya lebu=ki pujapaҀh karay=na laҌ=ki=may, then

Kharia

man=PL

soub=aҌ thoƾ, ro all=GEN

for

sacrifice

ho=ki=te

do=INF

яoli=te

IPFV=MID.PST=3PL

яoko=ke

and that=PL=OBL palanquin=OBL sit.down-=SEQ

[Ø] яoҌя=na

goҌ=ke carry.on.shoulders =SEQ

take=INF

laҌ=ki=may. IPFV=MID.PST=3PL

‘The Kharia men used to perform sacrifices, for all, and [they z the Kharia men] used to seat them [= the Kharia men] on a palanquin and carry them on their shoulders.’ [AK, 2:6] Similarly, “subject identity” between the converbal clause and the matrix clause is not syntactically or semantically restricted. To be sure, the S/A of the converb is generally coreferential with that of the main predicate, as with dhoҌke in (61), whose A is coreferential with the A of umay juƾoҌ ‘they didn’t ask’ and the S of muҌ goҌяkimay ‘they went off’. However, as (63) shows, the S/A of the converb can refer to the nonmacrorole experiencer of the matrix clause (a so-called “dative subject”), but it can also refer to a unit which is not even an argument in the matrix clause, as in (64), where the S of awkon is jiƾray, i.e., the porcupine who the daughter (= [herk]) is being sent off with.

(63) etwa=te

u=ki=yaҌ

haleit

[Ø] yo=kon lebui laҌ=ki.

Etwa=OBL this=PL=GEN condition

see=SEQ love

‘Etwai, [Øi] seeing their condition, felt compassion.’

(64) thor toҌ [Ø] few

day

aw= kon jiƾray=aҌ stay=SEQ

EMOT=MID.PST

[RD, 2:108]

sori яaƾ goҀh=oҌ.

porcupine=GEN with send

C:TEL=ACT.PST

‘[‡i] having stayed a few days, [hej] sent [herk] off with the porcu[Pinnow, 1965: 40] pinei.’ 4.4. The status of overt RPs It was noted in Section 3 that with head-marking languages in RRG, subject and object marking on the predicate (using traditional terminology for the moment) is considered to be the actual subject and object of the core/clause and not merely agreement marking. As such, any overt RPs referring to

Aspects of Kharia grammar

107

these units are clause-internal but core-external and only serve to further specify the identity of these arguments, similar to apposition. However, many languages are not simply either head-marking or dependent-marking but are rather somewhere in-between the two, for example when the subject is marked on the predicate while other arguments are not. In general, it is assumed in RRG in cases such as these that the subject marker on the predicate is a core argument, whereas the object and other arguments are expressed via overt RPs. With this information in mind, let us now reexamine the Kharia data. 4.4.1.

“Subjects”

Recall that subject marking on the predicate in Kharia, which is obligatory, is also enclitic. As such, this marking is a unit of the syntax and must be considered the subject of the core/clause. Although the presence of the subject enclitic is compatible with the presence of an overt RP referring to the same entity, I argue that the overt RP is not the subject but simply refers to the same unit as the enclitic subject marking. A number of arguments can be made for this analysis. First, the overt RP, like all RPs in Kharia, is not obligatory and can always be omitted, whether its identity is unknown or is considered unimportant by the speaker. Second, if we consider the overt RP to be a subject, we would then have a clause with two subjects, as the enclitic subject marking is, as already noted, an obligatory unit of syntax and hence cannot be considered agreement marking. But perhaps most importantly, the person / number features of the overt RP and the enclitic subject often differ, as in (65).

(65) botoƾ=ta=pe

ho=ki

lutui

su=kon

peҌ cakhnaҌ

fear=MID.PRS=2PL that=PL clothes put.on =SEQ rice curry

iҌjthaƾ

kinbhar=na=pe,

iѪ=ko

laѵe=na=iѪ.

cow.dung courtyard=MID.IRR=2PL 1SG=CNTR fight=MID.IRR=1SG

‘Those of you who are afraid, you put on your [house] clothes and see to your house work like cooking and cleaning the courtyard with cowdung, but I will fight.’ [Kerke‫ݚݚ‬Ɨ, 1990: 7] While the overt RP in (65) is hoki, the head of a “relative clause” which consists of the TAM/PERSON-syntagma botoƾ=ta=pe ‘you (PL) are afraid’, the predicate is marked as having a second-person, plural subject, i.e., =pe. Nonetheless, the two units refer to the same referent: The utterance is directed toward a large group of people, i.e., ‘(those of) you who are

108

John Peterson

afraid’, but not ‘(all of) you’. This type of “semantic agreement” is typical of Kharia and is not any kind of production error of the type found in the so-called constructio ad sensum in classical languages, which is usually considered performance errors in lively conversation: Native speakers I consulted saw nothing strange or incorrect about such examples, hence they cannot be considered performance errors. (66)–(68) show that person and number of the overt RP and the enclitic subject often do not agree, as long as both can be understood as having the same reference.

(66) ber=jo

i=jo

who=ADD what=ADD

aҌ=pe

gam=e.

NEG.MOD=2PL

say=ACT.IRR

‘Don’t any of you say anything.’

(67) behar who

[Kerke‫ݚݚ‬Ɨ, 1990: 2]

bujhay=e=pe? explain=ACT.IRR=2PL

‘Which of you will explain?’ moѪ khõѵha kayebar ghaҀo tay eяoҌ=ga toҌbluƾ one

section

Khyber

valley

ABL

more=FOC high

яel=ki=may. come=MID.PST=3PL

‘One section [of the group] came (PL) up even higher than the Khyber Pass.’ [MT, 1:48] In (68) the (singular) addressee is being asked if she will meet with her fiancé, Dele, who is not present in this discussion.

(68) яele buƾ kol kui=na=bar? Dele

COM REC

bariya=ga goѵjhuƾ

find=MID.IRR=2DU both=FOC

path

yokuҌj=na=le. wait.for=MID.IRR=1PL.EXCL

‘Will [you (SG)] meet up (= DU) with Dele? (i.e., “[You] with Dele, will you two meet?”) We will be expecting [you] both.’ [Kerke‫ݚݚ‬Ɨ, 1990: 16] Also, as noted above (cf. note 5), an overt RP need not even be a CASEsyntagma but may also be a TAM-PERSON-syntagma. This further demonstrates that the overt RP is not the subject but merely identifies this unit further if the speaker considers this appropriate.

Aspects of Kharia grammar

(69) kunяab aw=ki behind

109

tomliƾ khaѵiya gam яom=na laҌ=ki=may

QUAL=MID.PST

milk

Kharia

tomliƾ uҌя=ga

ina no

u=ki

because

this=PL milk

say

PASS=INF

IPFV=MID.PST=3PL

яel=ki=may.

drink=FOC come=MID.PST=3PL

‘[Those who] were behind were called “Milk Kharia” because they came drinking milk.’ [MT, 1:180] With that, the apparent “overt subject” RP is core-external but clauseinternal. The following presents a few simple examples along with their respective syntactic representations.24

(70) og=aҌ

koleҌ

leƾ

col kan=ki.

house=GEN parrot

fly

go

CONT=MID.PST

‘The parrot of the house continued flying away.’

(71)

SENTENCE CLAUSE RP

CORE

CASE-SYNTAGMA

PRED

RP

TAM/PERSON-SYNTAGMA CORER CASE

NUC

NUCR

og=a‫ ݦ‬kole‫ݦ‬

TP-BASE

‡

TV

NUC

NUC

leƾ

col

kan =ki

‡.

RP

(DIR)

this=DU

RP

CLAUSE

now

taƾ

NUCR

CORER

(DIR)

Ø

CASE

1PL.EXCL=GEN

el=aҌ25

NUCR

CORER

COM=FOC

sori=ga

CASE

uя =ACT.IRR drink

=e

NUC

=ACT.IRR

=e

TV

TP-BASE

=DU

=kiyar.

RP

[adapted from Roy & Roy, 1937: 180f.]

eat

Ѫog

NUC TV

TP-BASE

TAM/PERSON-SYNTAGMA

CORE

SENTENCE (= structure of (22), simplified here somewhat)

CASESYNTAGMA CASESYNTAGMA

RP

‘They two will now eat and drink with us.’

‡

u=kiyar

NUCR

CORER CASE

PERIPHERY

John Peterson

CASESYNTAGMA

(72)

110

Aspects of Kharia grammar

4.4.2.

111

Non-subjects

There are three types of units in Kharia which are not subjects but which can be considered arguments. As with all other arguments, these elements need never be expressed overtly: Undergoer – This unit was discussed in Section 4.1. It is the lowest macrorole argument and is always the lowest argument in the LS. It is obligatorily marked by the oblique marker =te if definite and human, otherwise it may be unmarked. Non-macrorole arguments. These include both recipients and locatives in bitransitive clauses, e.g., the recipient with ter ‘give’ or the locative with maѵay ‘put, place’. Both groups are obligatorily marked by the oblique marker =te, regardless of animacy/definiteness. In addition, the presence of a benefactive argument can be indicated on the predicate by the benefactive “V2” kay. As the predicate does not indicate the identity of these units, it would seem reasonable to assume that the overt RPs referring to these units are arguments. However, this is problematic for two reasons: 1. RPs referring to these units are never obligatory and can always be omitted even if their identity is unknown or if the speaker feels this information to be unimportant, 2. The presence of these units is unambiguous in Kharia, even when they are not overtly expressed. With respect to 1: As we have repeatedly noted, no overt RP is ever obligatory in Kharia, regardless of the identifiability of its referent. This is shown in (73) and (75), from Peterson (2011: 357). Note that in both examples, no information is considered known, indicated in the respective LSs by “Ø” to denote an unknown entity. (Upon entering a room in a state of disorder:) (73) Q: ute i hoy=ki? A: surum=oҌ=ki. here what become=MID.PST

‘Q: What happened here?

steal=ACT.PST=PL

A: [Someone, PL] stole [something].’

(74) surum=oҌ=ki do' (3PL, ‡) CAUSE [BECOME not-have' (‡, ‡)]

112

John Peterson

(75) Q: laҌ then

i

karay=oҌ ?

A: mãѵay=oҌ.

what do=ACT.PST

‘Q: Then what did s/he do?

put=ACT.PST

A: S/he put [something somewhere].’

(76) mãѵay=oҌ [do' (3SG, ‡)] CAUSE [BECOME be-at' (‡, ‡)] With respect to 2: Most contentive morphemes in Kharia in predicative function have a fixed number of arguments, semantically speaking, such as ter ‘give’ and maѵay ‘put, place’, which are both bitransitive, yar ‘flee’, яel ‘come’ and col ‘go’, which are intransitive, amo࣯я ‘wash (the face)’, gujuƾ ‘wash (the feet)’, which are monotransitive. However, a considerable number of contentive morphemes can have either one or two arguments. In such cases this is unambiguously indicated through the use of the active (transitive) or middle (intransitive) voice. For example, a morpheme such as ruҌ denotes in the middle voice ‘open (ITR)’, whereas in the active voice it means ‘open (TR)’. This is summarized in (77).

(77) Class A: Fixed valency Class B: Varying valency: Active– transitive Middle – intransitive Similarly, the presence of benefactives can be indicated on the predicate itself, although again not its identity. Consider examples (78)–(80): In (78) the predicate is overtly marked by the “V2”kay ‘BEN’ as a kind of “applicative” which indicates the presence of a benefactive participant. Its identity is indicated by the overt RP hokaѵaҌ thoƾ ‘for him/her’. Again, like all RPs in Kharia, this RP is not obligatory and can be omitted if its identity is unknown or considered unimportant, as in (79).

(78) ho=kaѵ=aҌ

thoƾ

that=SG.HUM=GEN for

caha

bay

kay=oҌj.

tea

make

BEN=ACT.PST.1SG

‘I made tea for him/her.’

(79) caha tea

bay

kay=oҌj.

make

BEN=ACT.PST.1SG

‘I made tea for [him, her, us, you, … someone].’ However, (80) clearly demonstrates the non-argument status of the overt beneficiary-RP: Here the same unit is explicitly mentioned and has the same form as in (78) but its presence is not signaled by the predicate.

Aspects of Kharia grammar

(80) ho=kaѵ=aҌ

thoƾ

that=SG.HUM=GEN for

caha

bay=oҌj.

tea

make=ACT.PST.1SG

113

‘I made tea for him/her.’ Thus, I argue that it is the marking on the predicate alone which denotes the number of arguments in the LS, whether or not the identity of these units is specified (or even recoverable from context), whereas hokaѵaҌ thoƾ in (80) is an adjunct, as it is not signaled on the predicate. In fact, the data, taken together, strongly suggest that all clauses in Kharia are syntactically intransitive as only one argument, the enclitic subject, is ever obligatory.26 Hence, I assume that all overt RPs, whether TAM/PERSON- or CASEsyntagmas, are core-external but clause-internal if they refer to semantic arguments. If not, they are then adjuncts in the periphery. Thus, in this interpretation, the core in Kharia consists only of the predicate and its subject marking. The remainder of the clause then consists of optional, core-external RPs which further specify the identity of the arguments when the speaker considers this appropriate. In RRG terms, the clause contains as many arguments as are found in the LS, although the identity of these units is not expressed by arguments but rather by coreexternal RPs. (81) presents an example of a clause with two semantic arguments and the syntactic structure I propose. The underlying assumption motivating this analysis is that three basic processes are involved which can be accomplished either independently or in combination with one another. These are the following: 1. Indicate the number of semantic arguments of the predicate; 2. Indicate their identity; 3. Indicate their functions (Actor / Undergoer / other).

‡

(DIR)

person=P L

lebu=ki

NUCR

CORER

(DIR)

‡

CASE

‘The people tell [this] story in the villages.’

story

kahani

NUCR

CASE

CASESYNTAGMA

CASESYNTAGMA

CORER

RP

RP

(81)

PRED

village=PL

khoѵi=ki

NUCR

CORER

=OBL

=te

CASE

=MID.PRS =PL

=ta

TV

[AK, 1:4, Peterson, 2011: 439]

tell

kayom

NUC

RP

=ki

CORE

CASESYNTAGMA TAMSYNTAGMA

RP

PERIPHERY

CLAUSE

SENTENCE

114 John Peterson

Aspects of Khariya Grammar

115

– With respect to 1.: As noted above, the number of semantic arguments is unambiguously indicated in Kharia by a combination of the identity of the predicating morpheme and the basic voice marking of the predicate and in no way depends on the number of overt RPs in the clause. This combination serves to open the appropriate number of slots in the LS (proceeding from syntax to semantics for the moment), but it does not fill these slots. This is thus accomplished independently of the other two processes in Kharia. – With respect to 2: The indication of the identity of a particular argument is left entirely to the discretion of the speaker – it may be made explicit, but it can also remain unspecified (denoted by “Ø” in the LS). – With respect to 3: As was noted in Section 2, the function of an argument in Kharia is expressed by the case marking of the RP, to the extent that this unit is overtly mentioned, as well as the “subject marker” on the predicate, which is always the highest macrorole argument. Thus, processes 2 and 3 are combined to some extent in Kharia, but are independent of process 1. This is a very different strategy than, e.g., in English, where all three processes are dependent on the overt presence of all arguments.27 As the view taken here with respect to non-subject arguments in Kharia is not mainstream RRG, this topic requires some discussion. In the next section I will argue that Kharia is in fact typologically quite common, and that this warrants a review of our current thinking on the status of RPs in general. 4.4.3. A typological interlude28 Although the classical type of “pro-drop”, i.e., the omission of a subject RP, is often viewed as a marked feature in human language, the opposite in fact seems to be the case. For example, the data given in Dryer (2008) strongly suggest that, far from being unusual, “pro-drop” languages are in fact that norm and languages such as English, which require the overt mention of all arguments, are rather infrequent. Dryer (2008) examined altogether 674 languages with respect to the expression of “pronominal subjects”, i.e., clauses which do not contain a full NP subject (or RP in our terminology) but “where the subject is expressed at most by a morpheme or morphemes coding semantic or grammatical features of the subject, such as person, number, or gender.” He divides these languages into the following categories, where:

116

John Peterson

 Pronominal subjects are expressed by pronouns in subject position that are normally if not obligatorily present, 77  Pronominal subjects are expressed by affixes on verbs, 409  Pronominal subjects are expressed by clitics with variable host, 33  Pronominal subjects are expressed by subject pronouns that occur in a different syntactic position from full noun phrase subjects, 64  Pronominal subjects are expressed only by pronouns in subject position, but these pronouns are often left out, 61  More than one of the above types with none dominant, 30 Of these, we may consider the second type, where the subject is expressed exclusively by affixes on the predicate (which alone accounts for more than half of the languages in the survey), and the fifth type, where pronominal subjects are optional, to be “pro-drop languages” (both here in bold print). Together, these account for 470 of the 674 languages or roughly 70%. Hence, “pro-drop” of the type found in Kharia can be classified as “run-of-the-mill” from a typological perspective. However, “subject dropping” is by no means dependent on the presence of subject marking on the predicate, as is quite often assumed. For example, among the 61 languages given in Dryer (2008) in which the expression of a pronominal subject is optional, we also find languages which do not show any person marking on the predicate. To demonstrate this, we can combine the data given in Dryer (2008) with that given in Sierwierska (2008), which deals with person marking on the predicate, although only in transitive clauses. Hence “subject” here is restricted to the A argument, but not S. Sierwierska (2008) examined 378 languages with respect to the person marking on the predicate and came to the following results. Of interest to us here are the first and third types (given here in bold), comprising altogether 106 languages or ca. 28% of the total, which never show any marking on the predicate for A.  No person marking of any argument, 82  Person marking of only the A argument, 73  Person marking of only the P argument, 24  Person marking of the A or P argument, 6  Person marking of both the A and P arguments, 193 Combing the data from Dryer (2008) and Sierwierska (2008) with the combining function provided on the internet version of the World Atlas of Language Structures (Haspelmath et al., 2008), we get the following

Aspects of Khariya Grammar

117

results: Of the 61 languages described by Dryer (2008) as having optional pronouns, i.e., languages in which the subject need not be expressed pronominally even if no NP/RP is present, 26 do not show any person marking on the predicate while one marks only for P. In other words, 27 out of 61 of these “pro-drop” languages, or roughly 44%, may omit any overt reference to the subject and with that can provide even less information on the identity of the subject than Kharia.29 In other words, the presence of subject marking is not a prerequisite for the omission of the subject-denoting RP and the assumption that a semantic argument must either be expressed as an overt RP or marked on the predicate is highly questionable: The presence of a semantic argument (although perhaps not its identity) is a matter of the logical structure. The syntax may or may not contain this unit in its structure as an argument – it can be an argument (as in English, cf. (29)), it may not be present at all, or it may be outside of the core. In fact, this principle is not at all new to RRG – as noted above in head-marking languages in general the overt RPs are regularly not syntactic arguments but are core-external.30 Data such as these suggest that a considerable number of languages do in fact treat at least two of the three criteria mentioned above independently, i.e., the number of semantic arguments and their identity, while other languages, such as English, combine all three in the overt expression of an NP/RP. As such, I suggest that it would be wisest to assume, per default, that overt RPs are not found in the core unless it can be shown that they do indeed belong to the core, e.g., because an overt RP is obligatory, or because the presence of person marking on the predicate is not compatible with the presence of an overt RP, etc. Although such an analysis in effect goes against conventional wisdom by forcing us to justify assuming that overt RPs are syntactic arguments instead of having to justify “exceptions” where they are not, there are good reasons against requiring that a semantic argument must be overtly contained in the syntactic representation: Its presence, its identity, and its function may but need not all be overtly signaled, and of those that are overtly signaled, there is no general requirement that all be signaled in the core. 5. Summary and outlook The preceding pages have shown that a language such as Kharia, for which neither nouns nor verbs need be assumed, can easily be accounted for

118

John Peterson

within the framework of Role and Reference Grammar (RRG). In addition, we have also seen that grammatical relations, especially the notoriously difficult concept of “subject”, are not needed to describe such a language. RRG provides an alternative to grammatical relations in the form of a language-specific and construction-specific “privileged syntactic argument” or PSA: While some operations, such as person marking on the predicate, control constructions and one type of reflexivization all make use of a one type of PSA in Kharia, another type of reflexivization makes use of a similar – but different – PSA, while yet other operations, such as coreferential deletion across clauses and “subject identity” with converbs (“conjunctive participles”) have no PSA whatsoever. Assuming a blanket term such as “subject”, no matter how this is defined, will not be able to account for these facts. Finally, it was argued for Kharia that overt referential expressions (RPs) which appear to be syntactic arguments are in fact core-external units which merely serve to specify the identity of the semantic arguments, if the speaker considers this appropriate. As no RP in Kharia is obligatory – whether or not anything is known about its identity – and since the number of semantic arguments is unambiguously signaled by the predicate, I argue that we cannot simply assume that all semantic arguments will necessarily be expressed as syntactic arguments in languages. Rather, we must distinguish between means of indicating the presence of these arguments, of indicating their identities, and of indicating their functions. While languages such as English require that all semantic arguments be expressed overtly – thereby simultaneously indicating their presence, identity and function – there is much typological evidence that this is not a universal feature of human language, and that languages exist, such as Kharia, which distinguish these three criteria to a greater degree. Abbreviations 1, 2, 3 = first/second/third persons; ABL = ablative; ACT = active voice; BEN = benefactive; CNT = continuous; CNTR = contrastive focus; COM – comitative; CONT = continuous; CORE(R) = core (of an RP); C:TEL = culminatory telic; DEM = demonstrative; DU = dual; EXCL = exclusive; FOC = restrictive focus; GEN = genitive; HUM = human; INF = infinitive; IPFV = imperfective; IRR = irrealis; LEX = lexeme; LS = logical structure; MID = middle voice; NPT = nonpast; NUC(R) = nucleus (of an RP); NUM = number; OBL = oblique; PASS = passive; PL = plural; PERF = perfect; POSS = inalienable possession; PRO = proform;

Aspects of Khariya Grammar

119

PSA = privileged syntactic argument; PST = past; QUAL = marker of qualitative predication (§ copula); QUANT = quantifier; REC = reciprocal; REFL = reflexive; REP = repetition; RP = referential phrase; RRG = Role and Reference Grammar; SEQ = sequential converb; SG = singular.

Notes 1. 2.

3. 4. 5. 6.

7.

8.

9.

For a detailed overview of the functions of the active and middle in Kharia, cf. Peterson (2011: 258–293). In Peterson (2011: 118–121) it is shown that a structure such as that in (3) is a simple juxtaposition of two lexemes, both of which are phonological and morphosyntactic words, thus (3) is not an example of a compound but merely one lexeme (lebu) modifying another (jati). For a detailed discussion, see Peterson (2011, Chapter 4). For a detailed discussion of these phonological and morphosyntactic enclitics in Kharia, see chapters 2 and 3 in Peterson (2011). Furthermore, CASE- and TAM/PERSON-syntagmas may fulfill any of the three basic functions of reference, attribution and predication. For further discussion, cf. Peterson (2011, chapter 4). Cf. examples (8) and (9) for the use of the genitive in TAM/PERSONsyntagmas. For examples of the use of possessive and number marking in the semantic base of a TAM/PERSON-syntagma, see Peterson (2011: 110, examples (71) and (72)). For the sake of simplicity, all “vectors” or “V2s”, i.e., markers for Aktionsart and other categories such as durativity, the passive, benefactive and a number of other categories, will be treated in this study as part of the functional head of the TAM/PERSON-syntagma. Cf. however Peterson (in press), where it is argued that these in fact do not form a single homogeneous category, as telic V2s are part of the functional head of the TAM/PERSON-syntagma whereas non-telic V2s are part of the semantic head. As the present study contains no examples with V2s, this does not affect our discussion and is merely intended to simplify the discussion. Although it is unnecessary to note that these content heads appear in the genitive, as the content head itself may also end in the genitive (cf. the righthand side of example (10)), I use the term CONTHEADGEN here as an aid to the reader. This has no further theoretical implications. Number marking in the 3rd person is not obligatory, only in the 1st and 2nd persons. Hence, while the presence of =kiyar overtly marks an entity as dual and =ki / =may as plural, the lack of number marking in the 3rd person is semantically unmarked and can refer to any number of entities, although a singular interpretation usually results. Note that =kiyar and =ki are also found

120

10.

11.

12. 13.

14.

15. 16. 17.

18.

John Peterson in the CASE-syntagma and are, properly speaking, number markers, not person markers. Hence, they will be glossed as ‘DU’ and ‘PL’, respectively. In the 3rd person, plural, there are no semantic distinctions between =may, which is restricted to the TAM/PERSON-syntagma, and =ki and the two are freely interchangeable in this syntagma. The “Past II” does not differ semantically from the (simple) past in any way. It is a relatively recent development in the language and represents a further development of the past perfect. It differs from the (simple) past only in that it does not show a distinction for basic voice, i.e., the Past II does not mark for active or middle voice. However, its use is frowned upon by many older speakers, further suggesting its status as a new category. Again for reasons of space, no discussion of operators and their representation will be given here, and non-universal structures such as pre- and postcore slots (PrCS, PoCS) as well as Left and Right Dislocated Positions (LDP, RDP) which are not found in Kharia will be omitted from the following discussion. Both structures in (28) as well as the right-hand structure in (29) are from Van Valin (2005: 17), while the left-hand diagram in (29) is my own. “‡“ refers to an unspecified activity, i.e., what x does which causes the death of y. CAUSE and BECOME are operators referring to causativity and change of state, respectively. As their presence or absence is irrelevant for the following discussion, they will not be discussed further but should be intuitively clear enough to cause the reader no difficulties. Figure 7 actually differs somewhat from the final version of this hierarchy (Van Valin, 2005: 126): However, it will suffice for our purposes as the Undergoer in Kharia is always the lowest-ranked argument, i.e., there is no alternation in the selection of the Undergoer in Kharia. For reasons of space, we will deal only with “syntactic” PSAs, as these are the only ones which appear to play a role in Kharia. For a discussion of other PSA types, cf. Van Valin (2005, Section 4.3). Again, for reasons of space, I will refrain from providing an example illustrating the use of the instrumental marker buƾ, as this postposition does not figure prominently in the following discussion. In fact, there is no contentive morpheme denoting possession in Kharia, and the logical entry have' (x, y) given below example (40) for Kharia, would have the form have' (x, y) [MR 1], with only one macrorole (Undergoer). Here, x appears in a locative case while y is an Undergoer. However, what is at issue here is whether rayem and saroj compete for Actor status for the same predicate, i.e., ter, which is not the case. Hence, (37)a does not apply here. For reasons of space, we cannot go into the internal structure of the content head here. The “non-NUC” elements, including genitive attributes (as in (42)) and demonstratives but also possessive and number marking, are operators of

Aspects of Khariya Grammar

19. 20. 21.

22. 23.

24. 25. 26.

121

the RP and are depicted on an independent level, that of the operator projection (cf. Van Valin, 2005: 21–30). However, as these elements are not obligatory, their omission here will not affect the following discussion. They will be treated in a later study. For a further example showing that TP-BASE and SUBJ are independent levels, here in a complex predicate, cf. Ѫog=e uя=e=kiyar in (22) and (72). As stated for the CASE-syntagma in note 18, for reasons of space these operators cannot be dealt with in the present study. For an introduction to the operators of the core, clause and sentence levels, cf. Van Valin (2005: 8–16). The passive in Kharia is also compatible with intransitive predicates. Here, the intransitive Actor / Undergoer of the unmarked construction is omitted and the predicate has default marking for person, i.e., 3rd person, singular (= unmarked). However, as this construction does not have a PSA, it will not be dealt with here. Each auxiliary employs a different strategy for dealing with cases in which there is no “subject agreement”. For reasons of space, only one example is given here. There is a slight complication here in that the reflexive may also refer to certain other “oblique” functions, such as the (non-specified) possessor of the affected body in (i) . (i) apan=aҌ Ѫeri=te kosu dhokh=oҌ. REFL=GEN body=OBL pain grab=ACT.PST ‘Someone felt pain / sick in their own body.’ (lit. ‘Pain grabbed [his/her/my/…] own body.’) (i) appears to involve “possessor raising” involving an experiencer, i.e., the non-macrorole argument. As (ii) shows, what in (i) is the “possessor” of the body can also be expressed as a simple oblique-marked experiencer: (ii) musniƾ raҀa=yaҌ ap=яom=te kosu dhokh=oҌ. [BB, 2: 20] one.day Rata=GEN father=3POSS=OBL pain catch=A.PST ‘One day, Rata’s father got sick (= pain grabbed Rata's father).’ Hence, I (tentatively) assume that (i) is not an exception to the general rule but rather that a non-macrorole argument can be the PSA even when it is morpho-syntactically expressed as the possessor of the affected body, as it is still an experiencer. Further research is required here. For the sake of presentation, the direct case and the 3rd person, singular, both of which are unmarked, are given in the following diagrams as “Ø”. Note that the genitive is found here as it is required by the postposition sori when the RP is definite. Indeed, as Hengeveld & Rijkhoff (2005: 426ff.) note, languages which do not possess nouns and verbs as lexical categories tend to have only intransitive contentive morphemes, which fits in well with the Kharia data if “intransitive” is equated with SYNTACTIC intransitivity.

122

John Peterson

27. In addition to marginal subject marking on the verb in the 3rd person, singular, non-past in English. Note furthermore that criteria 2 and 3 above need not combine, e.g., in Santali (North Munda), where no RP is obligatory and where the functions of Actor vs. Undergoer (among others) are indicated directly on the predicate itself via the relative positions of these markers. 28. The following discussion derives from suggestions originally made in a talk by Balthasar Bickel, although I alone am responsible for the discussion of the data given here. 29. Note that Dryer (2008, §1) writes that in languages in which the pronominal subject is optional, “the pronoun will typically only be left out if the reference is clear in the context.” However, recent work in at least some of these languages, such as Tao (1996) on Mandarin, suggest that this may not be as widespread as is generally assumed. Further research in this area is required. We should also note here the apparent discrepancy in the figures given above: As the 61 languages cited by Dryer (2008) as only signaling the subject by means of an optional pronoun would seem to suggest that none of these languages should mark for A, the figure of 44% actually seems quite low, as one might expect this figure to be 100%. In fact, although there are undoubtedly also other factors involved, Dryer (2008) and Sierwierska (2008) may be using somewhat different classificatory criteria. E.g., Sierwierska (2008, §1) writes that a number of factors may combine to determine whether person marking is found on the verb. E.g., in Paluan, person marking on the predicate is found in the presence of nouns but not free pronouns. Further factors include the location of the NP argument corresponding to the person marker, animacy, definiteness / referential status, person, tense, aspect, mood and polarity. In these cases, Sierwierska (2008) includes such languages among those showing person marking “on the basis of the maximal person marking that can occur in a single clause.” Hence, I assume that this will account for at least much of the discrepancy noted. Further research is required. 30. In fact, in languages such as English, which possess a precore slot, we also quite often find a semantic argument contained in this position, which is then clause-internal but core-external.

Aspects of Khariya Grammar

123

References Biligiri, Hemmige Shriniwasarangachar 1965 Kharia: Phonology, Grammar and Vocabulary. (Building Centenary and Silver Jubilee Series, 3.) Poona: Deccan College, Postgraduate and Research Institute. Dryer, Matthew S. 2008 Expression of Pronominal Subjects. In The World Atlas of Language Structures Online, Martin Haspelmath, Matthew S. Dryer, David Gil and Bernard Comrie (eds.). Munich: Max Planck Digital Library, chapter 101. Available online at http://wals.info/feature/101. Accessed on . Haspelmath, Martin, Matthew S. Dryer, David Gil, and Bernard Comrie (eds.) 2008 The World Atlas of Language Structures Online. Munich: Max Planck Digital Library. Available online at http://www.wals.info/. Accessed on . Hengeveld, Kees, and Jan Rijkhoff 2005 Mundari as a flexible language. Linguistic Typology 9 (3): 406–431. Kerke‫ݚݚ‬Ɨ, Khrist PyƗrƯ 1990 jujhair яƗ˾ ѵ (khaѵiyƗ nƗҀak). Ranchi: JanjƗtƯya BhƗৢƗ AkƗdamƯ, BihƗr SarkƗr. [“The Battle-Field (Kharia Drama)”, in Kharia.] Malhotra, Veena 1982 The Structure of Kharia: A Study of Linguistic Typology and Language Change. Ph.D. diss., New Delhi: Jawaharlal Nehru University. Peterson, John 2005 There’s a grain of truth in every “myth”, or, Why the discussion of lexical classes in Mundari isn’t quite over yet. Linguistic Typology 9 (3): 391–441. 2007 Languages without nouns and verbs? An alternative to lexical classes in Kharia. In Old and New Perspectives on South Asian Languages. Grammar and Semantics. Papers growing out of the Fifth International Conference on South Asian Linguistics (ICOSAL5), held at Moscow, Russia in July 2003, Colin P. Masica (ed.) (MLBD Series in Linguistics, XVI.) Delhi: Motilal Banarsidass. 274–303. 2011 A grammar of Kharia. A South Munda language. (Brill’s Studies in South and Southwest Asian Languages, 1.) Leiden / Boston: Brill. in press Languages without nouns and verbs. A formal account. In Flexible word classes: a typological study of under-specified parts-of-speech, Jan Rijkhoff and Eva van Lier (eds.). Pinnow, Heinz-Jürgen 1965 Kharia-Texte (Prosa und Poesie). Wiesbaden: Otto Harrassowitz.

124

John Peterson

Siewierska, Anna 2008 Verbal Person Marking. In The World Atlas of Language Structures Online, Martin Haspelmath, Matthew S. Dryer, David Gil and Bernard Comrie (eds.). Munich: Max Planck Digital Library, chapter 100. Available online at http://wals.info/feature/description/102. Accessed on . Roy, Sarat Chandra, and Ramesh Chandra Roy 1937 The KhƗ‫܀‬iƗs. Ranchi: Man in India. Tao, Hongyin 1996 Units in Mandarin Conversation. Prosody, Discourse, and Grammar. (Studies in Discourse and Grammar, 5). Benjamins: Amsterdam / Philadelphia. Van Valin, Robert D., Jr. 2005 Exploring the syntax-semantics interface. Cambridge: Cambridge University Press.

Grammatical voice in Gorum* Felix Rau

A new analysis for the Gorum verb system is proposed and it is argued that contrary to the established view Gorum has a grammatical voice system similar to the systems found in most other Munda languages. The analysis presented here provides a more regular picture of the verb morphology of Gorum and can be supported by comparative data from other Munda languages. Finally, the new analysis allows a reconstruction of the voice morphology for Proto-Sora-Gorum and partially for Proto-Munda.

1. Introduction The present article proposes a reanalysis of the verb system of Gorum, a Munda language of India. Languages of the Munda family generally distinguish between active and middle voice forms in their verbal morphology. The voice category is a central part of the verb system of all Munda languages and is relatively similar in the whole family, although it is sometimes phrased in terms of transitive and intransitive instead of voice. Gorum – sometimes with Sora or Gta‫ – ݦ‬has always been considered exceptional, because according to the prevalent view, it has “neutralized ... the transitive/intransitive inflection contrast altogether” (Anderson 2007: 107). Contrary to this established view, I argue that Gorum features exactly such a voice distinction and that it pervades its whole verb system. Furthermore, the voice system is regular with the active voice suffix -u and the middle voice suffix -nuҌ as its base. The main focus of this article is on the morphological and historical aspects of the Gorum voice system, and especially on the middle voice morpheme -nuҌ. Its exceptional morphophonological behavior hides the regularity and pervasiveness of the voice system and has been the main reason why the system has eluded researchers until now – including the present writer (cf. Anderson and Rau 2008). The new analysis presented here paints a more regular picture of the verbal morphology and shows that the verb

126

Felix Rau

system of Gorum is much less exceptional within its language family than previously thought. Even though the analysis presented here is fundamentally different from previous discussions of the verb system of Gorum, certain aspects of it can be found in one or more of the accounts of Zide (1972, 1990), Aze (1973), Anderson (2007) and Anderson and Rau (2008). However, none of these has recognized the pervasive active/middle voice distinction in all verb forms and its consequences for the verb system as a whole. Instead, these previous accounts have focussed on parts of what is analyzed here as middle voice and phrased the analysis in terms of intransitivity and affectedness (Zide 1972, 1990; Anderson 2007; Anderson and Rau 2008) or undergoer function in the case of Aze (1973). The present analysis replaces the marginal transitive/intransitive distinction found in the forms called infinitive in all earlier research as well as the affectedness category and shows that intransitive as well as affectedness-marked verb forms are middle voice forms. Aze’s (1973) undergoer function resembles the middle voice category more closely, but his analysis and the resulting verb system also differ significantly from the present proposal. 2. Verbal morphology in Gorum In the analysis presented here, voice marking has a central role in the verbal morphology of Gorum. Besides voice affixes, verbs can carry tense, aspect, mood and negation morphology, as well as a ventive marker. Furthermore, Gorum possesses an elaborate person marking system in which two roles can be marked on one verb form. Generally, speech act participants are marked following an active/stative-like pattern. The third person is mostly unmarked, except for third person plural participants in certain constellations, if they are actors or subjects. The subtleties of this system go beyond the scope of this article and are of no further concern here. Verb forms in Gorum display a considerable degree of complexity. The four verb forms below give a good overview of the range of possible forms. What follows is a short sketch of the morphological structure of the verb as well as a description of the morphological system that forms the basis for these forms. (1)

ne-r-ab-so’ї-om 1sA-NEG-CAUS-learn-ACT:2sU ‘I will not teach you. / I have not taught you.’

127

Grammatical voice in Gorum

(2)

baҌ-t-aj=gi come-ACT:NPST-VEN=3pA ‘They will come.’

(3)

mo-laҌ-r-iƾ-aj 2sA-hit-ACT:PST-1sU-VEN ‘You have hit me.’

(4)

imbur-r-aj=ni wind-ACT:PST-VEN=PROG ‘The wind is blowing.’

Apart from some minor morphosyntactic irregularities, the morphological structure of the verb is regular. The following template gives the structure of the verb and the morphological slots it consists of. Prefix 2

1

A

MOOD

Prefix Domain

Bare Stem

CAUS

RDL

ROOT

Suffix

LV

Stem

Clitics

1

2

3

1

2

VOICE:TNS

U

VEN

PL

PROG

Suffix Domain

Enclitics

Verb

Figure 1. The morphological structure of the Gorum verb

In this morphological structure, the bare stem takes the central position. It minimally consists of a root or, in case of loanwords, of the root and the loan verb suffix -ej (LV). Additionally, it can contain the reduplicant of the verb root (RDL) and the causative prefix ab-/a֓ b- (CAUS). The bare stem as the central morphological unit is also the main reference point for voice marking. It is prosodically very stable and generally not altered by morphophonological processes. This prosodic stability is crucial for its relation to voice marking. The rest of the verbal morphology can be understood in relation to the bare stem. Two prefix slots precede the stem. Prefix slot 1, labelled MOOD

128

Felix Rau

in the template, can contain the irrealis mood prefix aj- or a negative prefix – for non-past negation the affix or- and for past negation ar-. The second prefix slot is where the actor prefixes are positioned. The prefix domain is not relevant for voice marking in Gorum and is of no further interest for the present purpose. The suffix domain, on the other hand, features prominently in the following discussion. It is the location of voice marking, and the other suffixes interact heavily with the voice morphemes. The suffix domain contains three distinct suffix slots. The first position, following the bare stem, is the prime locus of voice marking. This position is shared by the voice suffixes -u (ACT) and -nuҌ (MID), which are the main subjects of this article, and the tense suffixes -ru (PST) and -tu (NPST). The interaction between voice and tense morphology is the most complex aspect of voice marking and will be explained in detail below. The next morphological slot is the position of the undergoer suffixes and the third person subject suffix -ej1 – both are labelled U in the template above. The following slot is occupied by the ventive suffix -aj (VEN). The discussion of voice in Gorum is centered on these three suffix positions. Two clitic positions follow the suffix domain. These formatives have morphosyntactic properties that differ from those of the suffixes. Two different types of plural enclitics occur in the first clitic slot, viz. the third person plural actor clitic =gi and the enclitic =bu, which marks the plural of addressees in the imperative paradigm. The last morphological slot of the verb is the locus for the progressive enclitic =ni. This is the complete morphological structure of the Gorum verb. Of the four structural domains – prefix domain, bare stem, suffix domain and enclitics – only the bare stem and the suffix domain are of relevance for the understanding of voice in Gorum. 3. The voice system in Gorum The morphology of virtually all languages of the Munda family features a grammatical voice system and verb forms are marked for either active or middle voice. These two voice categories have been called active and middle in Sora (Ramamurti 1931, Starosta 1967), Santali (Neukom 2001), Gutob (Griffiths 2008), Mundari (Cook 1965) and Kharia (Peterson 2008, 20), while they are called transitive and intransitive in Remo (Fernandez 1968) and in one account of Mundari (Osada 2008). The concrete forma-

Grammatical voice in Gorum

129

tives involved in these voice systems differ considerably, but the morphological structure is reasonably homogeneous across the family. In fact, in all of these languages, with the exception of Sora and Gorum, voice marking is combined with tense marking in the form of tense-voice portmanteau suffixes. The syntax and semantics of the two voice categories are also comparatively similar over the whole family. Their distribution and their syntactic and semantic effects are determined by the verb class of the lexeme. Many verbs occur only in one of the two categories, while others occur in both. As a rough tendency, active voice is associated with transitive verbs and middle voice with intransitive ones, especially those of posture and motion, but also verbs of grooming. Furthermore, middle voice has a detransitivizing effect on transitive verbs and denotes passive, middle passive, reflexivity, and indirect or self-benefactive middle. Beyond its central function, it can also convey non-volitional semantics or suddenness. Generally, the syntactic and semantic properties of the Munda middle voice characterize it as a typical middle voice system as described by Kemmer (1993). The short sketch given here barely scratches the surface of this complex aspect of grammar in Munda languages and for more information I must refer to the analyses in the description of the individual languages, such as Starosta (1968), Fernandez (1968), Neukom (2001) and especially Peterson (2008, to appear) as the most comprehensive discussion of voice in a Munda language.2 For all intents and purposes of cross-linguistic comparison, the situation in Gorum is similar to the one in other Munda languages. Voice is an obligatory category on verb forms in this language. The system distinguishes two voice categories, which will be called active and middle voice. Depending on their class, some verbs occur only in either the active or middle voice, in which case this obligatory category is a purely formal property of this verb, while other verbs can be marked for either active or middle voice. In these cases, voice has significant influence on the syntactic and semantic properties of the verb in question. Generally, active voice is associated with transitive verbs, but also occurs with some intransitive activities and states. Middle voice, on the other hand, is associated with intransitive verbs, especially those of posture, motion and grooming. The following list gives an impression of the lexemes which are associated with a given voice category.

130

Felix Rau

Table 1. Voice categories and corresponding verbs Active

jer ‘to run’ b‫ܚ‬tu ‘to be hungry’ aяaҌ ‘to be thirsty’ imbur ‘to blow’ (of wind)

Middle

яuku ‘to be, to remain’ uj ‘to go’ koko ‘to sit down, to sit’ a֓ ziƾ ‘to wash one’s own feet’

Active/Middle

ta֓ j ‘to give’ laҌ ‘to hit’ gi'ї ‘to see’ ruҌ ‘to pour’

The first two parts of this list show verbs that occur with only one voice category.3 The third part lists verbs that occur in both active and middle voice. These are mostly transitive verbs. Middle voice has a clear detransitivizing effect with these lexemes. Predominantly, it denotes reflexivity, middle passive, indirect (self-benefactive) middle or passive. The following gives some examples for the effect of voice on the syntax and semantics of transitive verbs. Phonologically, the difference between the two voice forms in these sentences is minimal: active voice is unmarked, while a glottal stop or creaky voice phonation of a vowel in the suffix domain indicates middle voice. This often makes it difficult to distinguish these forms for non-native speakers. These examples give a first impression of why the morphology of middle voice marking is the most complex part of the voice system. The issue will be discussed in more detail later. Henceforth I will gloss voice in the first position following the stem irrespective of where it is realized. In any case, for now the crucial aspect in these examples is the voice category of a verb form and its syntactic and semantic effects. In the examples above, the transitive verb ruҌ ‘to pour’ is in its active usage in (5) a transitive verb with at least an agent and a theme. Additionally, an optional benefactive participant role may be expressed, as can be seen from the object pronoun eniƾ in (6). In the middle voice, the verb’s argument frame is significantly different. In the first middle voice example (7), the agent and theme role are unaffected by the change of voice. The difference lies with the optional benefactive role which has to be inter-

Grammatical voice in Gorum

131

preted as coreferential with the agent; the sentence must thus be read as a self-benefactive. In (8), a further participant is removed: Here, besides the absence of the optional benefactive, the agent role is missing. This results in an impersonal or middle passive reading of this sentence.4 (5)

aяi

penяom=яi etur ruҌ-t-ej millet.beer=DEF OBJ pour-ACT:NPST-3pS ‘They poured out the millet beer’ DET.DIST

(6)

no'd eniƾ яaҌ ruҌ-r-iƾ 3sDIR.PRO 1sOBL.PRO water pour-ACT:PST-1sU ‘He poured water for me.’

(7)

яaҌ ne-ruҌ-ruҌ water 1sA-pour-MID:PST ‘I took a shower’ literally: ‘I poured (myself) water.’

(8)

ruҌ-ruҌ u‫ے‬j enu go:MID:PST DEM.PROX pour-MID:PST ‘This one was spilled.’

In other sentences the middle voice marking has an even clearer middle passive meaning: In (9), balbal ‘to warm, to heat’ in the active voice has both an agent and a theme. In the middle voice in (10), on the other hand, the agent is syntactically and semantically absent. (9)

buboƾ=яigin яa'd яaҌ ne-balbal-tu for water 1sA-warm-ACT:NPST child=DEF:PL ‘I will warm water for the children.’

(10) balbal-luҌ sunnen eno'dgi ne-ruҌ-tu COMP 3pOBL.PRO 1sA-pour-ACT:NPST warm-MID:PST ‘When it is warm, I will pour it out for them.’ The middle voice can also bring about passive-like changes to the argument structure, as is exemplified by the verb gi'ї ‘to see’. In (11), the third person plural actor clitic =gi denotes the agent of the event of seeing, while the object is left unspecified. In the middle voice form in (12), the same actor clitic denotes the theme participant. Due to the habitual interpre-

132

Felix Rau

tation of the non-past, the middle voice form of the verb gi'ї ‘to see’ is read in this case as ‘to look like’. (11) gi'ї-t-aj=gi see-ACT:NPST-VEN=3pA ‘They will look here’ (12) rumaƾ lukun gi'ї-t-a֓ j=gi cat like see-MID:NPST-VEN=3pA ‘They look like cats.’ The same lexeme in middle voice with a different combination of tenseaspect morphology can give the semantics ‘to be visible’, as in (14). (13), for comparison shows the equivalent active voice form. Differences in the contextual frame also contribute to this change in meaning of middle voice. (13) gi'ї-j-aj=ni see-ACT:PST-VEN=PROG ‘He/she/it is looking.’ (14) gi'ї-j-a֓ j=ni see-MID:PST-VEN=PROG ‘He/she/it is visible.’ Although the middle forms of gi'ї ‘to see’ in (12) and (14) differ in their meaning, the changes to the argument structure as opposed to the one associated with the active voice form are relatively uniform. In both cases, the most agent-like participant is removed, and only the patient-like participant remains. Additionally in (12), the undergoer from the active voice argument structure of the verb becomes the actor argument and is marked by the actor clitic =gi. These substantial argument structure alterations are similar to passive constructions. There are, however, no grammatical means to reintroduce the agent as an adjunct. This cursory treatment of the syntax and semantics of voice in Gorum suffices for the present purpose. The most important point is that its basic properties are as expected from an active/middle voice system (Kemmer 1993). Some lexemes only occur in one voice form, while others occur with both voice categories. In such a case, voice manipulates the argument structure associated with a verb to a substantial degree. Generally, middle voice

Grammatical voice in Gorum

133

has a detransitivizing effect and is associated with passive and reflexive semantics. Voice marking morphology As stated earlier, all verb forms in Gorum – except for the bare stem – are marked for either active voice or middle voice. The bare stem occurs as a free form in the complement position of the verb яa ‘to do, to become’ in light verb constructions. Furthermore, the bare stem is the basis for all verb forms and the central reference point for voice marking. This morphological unit in its minimal form consists of a root, or, in the case of a loan word, of a root and the loan verb suffix -ej. In addition, the bare stem can contain a reduplicant of the root (RDL) and the causative prefix ab-/a֓ b- (CAUS). The schematic representation of the resulting structure is repeated here in Figure 2 from the complete verb template in Figure 1 above. BARE STEM CAUS-RDL-[ROOT-LV]

Figure 2. The bare stem

The bare stem always consists of one or more syllables, and no morphophonological process can alter its prosodic structure:5 While in the domains of the prefixes and suffixes a hiatus is avoided by vowel deletion, this does not occur at the boundary between these affixes and the stem. This prosodic stability of the bare stem accounts for the behavior of the negative past tense prefix ar- as opposed to that of the causative prefix ab-/a֓ b- in examples (15)–(18). While the negative is reduced to r- to avoid a hiatus in the prefix domain, the causative remains a֓ b- and a hiatus occurs between the actor prefix ne- and the stem in (18). (15) ar-koko-nuҌ NEG:PST-sit-MID ‘he/she/it did not sit’ (16) ne-r-koko-nuҌ 1sA-NEG-sit-MID ‘I did not sit.’

134

Felix Rau

(17) a֓ b-koko-ruҌ CAUS-sit-MID:PST ‘he/she/it made him/her/it/them sit’ (18) ne-a֓ b-koko-ruҌ 1sA-CAUS-sit-MID:PST ‘I made him/her/it/them sit’ This prosodic stability of the bare stem also explains why there is no resyllabification if a CVC stem is followed by a vowel. In the suffix domain, the same structure triggers resyllabification. Example (19) shows such a CVC syllable in the suffix domain. In (20), the ventive suffix -aj causes resyllabification that distributes the undergoer suffix -iƾ over two syllables. Parentheses indicate syllable boundaries. (19) (laҌ)(-t-iƾ) hit-ACT:NPST-1sU ‘he/she/it will hit me’ (20) (laҌ)(-t-i)(ƾ-aj) hit-ACT:NPST-1sU-VEN ‘he/she/it will hit me’ There is no indication that such a resyllabification happens with the bare stem ga'd ‘to cut’ and the active voice suffix -u in (21) (21) (ga'd)(-u) cut-ACT ‘to cut, cutting’ This phonological structure violates the maximal onset principle, which Gorum otherwise adheres to. However, in this case, speakers articulate these two syllables very clearly in slow speech. Further evidence for the syllable structure comes from the presence of the preglottalized stop in this word. In Munda languages, the preglottalized stop /'d/ is generally considered an allophone of the phoneme /‫ܩ‬/ that occurs in the coda of a syllable, but cannot occur in the onset. This distribution is also found in the native Gorum vocabulary. However, in most other Munda languages the combination of morphemes phonologically similar to ga'd and -u would result in

Grammatical voice in Gorum

135

[ga.‫ܩ‬u] and not [ga'd.u]. Gorum, however, does not show any signs of such a resyllabification. In summary, the bare stem in Gorum is a self-contained, highly stable prosodic unit. Its morphological boundaries always coincide with syllable boundaries. This holds true for all regular verbs. Only a handful of high frequency items such as яuku ‘to be’ and uj ‘to go’ have verb forms in which the prosodic structure of the stem is exceptional. The prosodic stability of the bare stem is central for the understanding of voice marking in Gorum and its history. The bare stem can occur as an independent word in some constructions such as the light verb construction and is the only freestanding form of the verb that does not have voice marking. Apart from this, all other verb forms are marked for voice. 4. The extended stem The extended stem is formed on the basis of the bare stem by placing one of the voice suffixes -u or -nuҌ in the morphological slot directly following the bare stem. As stated above, the bare stem is a self-contained, stable prosodic structure, so that there is no significant morphophonological interaction between it and the voice suffixes. EXTENDED STEM BARE.STEM-VOICE

Figure 3. The extended stem

The extended stem is best considered a non-finite form and can function as either a nominal form or as the predicate in a purposive clause. It also is the basis of negative, irrealis and other finite verb forms. These verb forms are illustrated in the table below on the example of verbs which are fixedly marked as active or middle, respectively, as well as one verb compatible with either voice suffix.

136

Felix Rau

Table 2. Verb forms directly based on the extended stem Extended Stem

Negative Past

Negative Non-Past

Negative (underspecified for tense)

Irrealis

Active gaҌ-u eat-ACT ‘to eat; eating’

Middle koko-nuҌ sit-MID ‘to sit; sitting’

laҌ-u hit-ACT ‘to hit s.o.’ ar-gaҌ-u NEG:PST-eat-ACT ‘he/she/it did not eat’

laҌ-nuҌ hit-MID ‘to hit oneself’ ar-koko-nuҌ NEG:PST-sit-MID ‘he/she/it did not sit’

ar-laҌ-u NEG:PST-hit-ACT ‘he/she/it did not hit him/her/it’ or-gaҌ-u NEG:NPST-eat-ACT ‘he/she/it will not eat’

ar-laҌ-nuҌ NEG:PST-hit-MID

‘he/she/it did not hit himself/herself/itself’ or-koko-nuҌ NEG:NPST-sit-MID ‘he/she/it will not sit’

or-laҌ-u

or-laҌ-nuҌ

NEG:NPST-hit-ACT

NEG:NPST-hit-MID

‘he/she/it will not hit him/her/it’ ne-r-gaҌ-u 1sA-NEG-eat-ACT ‘I will not/ did not eat’

‘he/she/it will not hit himself, herself, itself’ ne-r-koko-nuҌ 1sA-NEG-sit-MID ‘I will not/ did not sit’

ne-r-laҌ-u 1sA-NEG-hit-ACT ‘I will not/ did not hit him/her/it’ aj-gaҌ-u IRR-eat-ACT ‘he/she/it would eat’

ne-r-laҌ-nuҌ 1sA-NEG-hit-MID ‘I will not/ did not hit myself’ aj-koko-nuҌ IRR-sit-MID ‘he/she/it would sit’

aj-laҌ-u IRR-hit-ACT ‘he/she/it would hit him/her/it’

aj-laҌ-nuҌ IRR-hit-MID

‘he/she/it would hit himself, herself, itself’

Grammatical voice in Gorum

137

The combination of the voice affixes and the bare stem is traditionally called the infinitive (e.g. in Aze 1973 or Anderson and Rau 2008), based on its function in purposive clauses. However, since it is also the basis for finite verb forms as those exemplified in Table 2, the term is avoided here, and the unit is called extended stem instead. On its own, however, the extended stem is a non-finite verb form. All finite forms of Gorum can be understood to be based on the extended stem. However, the two voice suffixes – and, consequently, the right boundary of the extended stem – are morphophonologically very unstable. This makes the identification of the voice marking system difficult, and its regularity hard to perceive. 5. The morphophonology of the extended stem The two voice suffixes -u and -nuҌ are morphophonologically unstable. In contrast to the verb forms presented above, they display considerable morphophonological interaction with the suffixes following them. The active voice suffix -u is particularly affected by the interaction with other suffixes. The phoneme /u/ is the least specific vowel of Gorum and in unaccented position, the suffix is often realized as [‫ ]ݜ‬or [ԥ], as in (22). It can also be assimilated to the preceding consonant, as in (23). Furthermore, where -u is followed by another suffix with an initial vowel, such as the ventive -aj in (24), the /u/, and consequently all phonological substance of the active voice suffix, is deleted. (22) don-u take-ACT ‘to take’

[d‫ܧ‬n.‫ ]ݜ‬or [d‫ܧ‬n.ԥ]

(23) ta֓ j-u give-ACT ‘to give’

[ta࡭ j.ܼ]

(24) don-aj (< don+u+aj) take-ACT:VEN ‘to bring’ As a consequence, active voice marking is in most verb forms phonologically zero. Additionally, word-final /u/ is frequently lost in casual

138

Felix Rau

speech. Hence, the /u/ of the active voice suffix -u may become zero even if no suffix follows. The middle voice suffix -nuҌ has more phonological substance than its active counterpart. The basic morphological principles for middle voice marking are similar to those described for the active suffix. Morphologically, the middle voice marking occurs in the same position as the active voice suffix. It is positioned in suffix slot 1, following the bare stem, as is demonstrated in examples (25)–(27). (25) яiҌ-nuҌ finish-MID ‘to become finished’ (26) koko-nuҌ sit-MID ‘to sit down, to sit’ (27) bas-ej-nuҌ smell-LV-MID ‘to smell (intr.)’ Similar to its active counterpart, the extended stem, consisting of the bare stem and the suffix -nu‫ݦ‬, can be seen as the basis for all other middle voice forms. However, its morphophonological properties are considerably different. To account for these differences, it is crucial to understand the behavior of the glottal stop as one instantiation of the more general suprasegmental feature of glottalization. In Gorum, glottalization is a property of the rhyme and has three phonemic realizations: the glottal stop /‫ݦ‬/, creaky voice /V࡭ / and preglottalization of voiced obstruents. The last realizations is not relevant for middle voice marking and will not be discussed further. The distribution of the remaining two, glottal stop and creaky voice, in middle voice marking is determined by syllable structure. In open syllables, the glottalization is realized as a glottal stop, resulting in a syllable of the structure CV‫ݦ‬. On syllables in which the coda position is occupied by another phoneme, glottalization is realized as creaky voice on the nucleus, yielding a CV࡭ C syllable. This is illustrated in Figure 4, in which glottalization of the rhyme is represented as the privative feature constricted glottis [CG].

Grammatical voice in Gorum

139

Figure 4. Realization of glottalization in different syllable types

Given the unspecific and phonetically unstable nature of the /u/ and the two possible realization of glottalization, middle marking could – on a very abstract level – be represented as /nVG/. This representation is intended to describe a syllable with the phoneme /n/ in the onset, followed by a glottalized rhyme, where /G/ represents the presence of the glottalization feature [CG] in the rhyme. /u/, the least specific vowel phoneme of Gorum, is the default nucleus. Similar to the active voice suffix -u, it is often realized as [ԥ]. The glottalization of the rhyme /G/ is realized as a glottal stop if the coda position is not otherwise occupied. If the coda position is occupied by a consonant, the glottal feature is realized as a creaky voice on the nucleus. In cases where the vowel /u/ is deleted due to morphophonological interaction with other suffixes, the phonological realization of the glottalization feature occurs on structures that are morphologically part of other suffixes, which follow the position of the voice markers. As such, the glottalization is the exceptional aspect of middle voice marking. The /u/ of the middle voice suffix -nuҌ, on the other hand, displays the same behavior as the vowel of the active voice suffix. The combination of the voice morphology in suffix slot 1 with the the speech-act participant undergoer suffixes and the third person plural suffix in suffix slot 2 is regular, given the morphophonological behavior of the vowel and the glottalization of the rhyme:

140

Felix Rau

Table 3. Interaction between the voice morphology and person markers Active

Middle

1sU

-iƾ

< /u+iƾ/

-n-‫ے‬ƾ

< /nVG+iƾ/

2sU

-om

< /u+om/

-n-o֓ m

< /nVG+om/

1pU

-ileƾ

< /u+ileƾ/

-n-iҌleƾ

< /nVG+ileƾ/

2pU

-ibeƾ

< /u+ibeƾ/

-n-iҌbeƾ

< /nVG+ibeƾ/

3pS

-ej

< /u+ej/

-n-‫ۀ‬j

< /nVG+ej/

The ventive suffix -aj, which occupies suffix slot 3, is placed in the position following the undergoer suffixes in suffix slot 2. However, if no suffix occupies suffix slot 2, the ventive directly follows the voice suffixes. In this case, the combination of the middle voice suffix -nuҌ and the ventive aj yields /na࡭ j/, as in (29). (28)

koko-nuҌ sit-MID ‘to sit down’

(29)

koko-n-a֓ j sit-MID-VEN ‘to sit down (here)’

The form -n-a֓ j demonstrates that the prosodic behavior of the glottalization is completely independent of the underlying morphological structure. Since the ventive suffix -aj is positioned in slot 3 following the slot of the undergoer suffixes, the combination of the middle voice suffix -nuҌ and the ventive suffix is different in its morphological structure from the combination of -nuҌ with the person marking suffixes. The first combination involves slot 1 and slot 3, while the second involves the directly adjacent pair slot 1 and slot 2. The prosodic principles that govern the form and placement of the glottalization, however, are identical in both structures. The respective combinations of the voice markers with the ventive marker demonstrate the similarities and differences between the morphophonology of the active and middle suffixes very well.

Grammatical voice in Gorum

141

Table 4. Interaction between the voice morphology and the ventive suffix Active

Middle

without VEN

-u

-nuҌ

with VEN

-aj < /u+aj/

-n-a֓ j < /u+nVG/

In accordance with the representation of the middle marker as /nVG/, the realization of the glottalization is solely determined by the prosodic structure of the involved suffixes, so that the glottalization can occur on any morpheme that forms the rhyme of the syllable following the bare stem. The form that glottalization takes depends on the phonological structure of that rhyme, with /‫ݦ‬/ added directly to open syllables and creaky voice to closed syllables. The only morphological structure that is relevant to the placement of the glottalization is the right boundary of the bare stem. This boundary always coincides with a syllable boundary, so that morphological and prosodic structures are necessarily aligned at this point. For the glottal aspect of voice marking, this is the one crucial reference point in the morphological structure of the verb. Beyond that, the glottalization of the middle voice marker is independent of the morphological structure of the verb’s suffix domain. The following template of the suffix domain shows the placement of the glottalization in respect to the morpheme positions, demonstrated with a few morphemes. Table 5. Possible morpheme combinations in the suffix domain Suffix slot 1

Suffix slot 2

Suffix slot 3

-MID

-nuҌ

-MID-VEN

-n

-MID-1sU-VEN

-n

-‫ے‬ƾ

-aj

-MID-1pU-VEN

-n

-iҌleƾ

-aj

-a֓ j

The previous discussion only covers situations in which voice is the only category marked in the first suffix slot. As indicated in the verb template on page 3, there is another category besides voice that appears in the same morphological position, namely tense: The first slot of the suffix domain is

142

Felix Rau

also the locus of the past tense suffix -ru and the non-past suffix -tu. While the interaction between the active voice marker -u and the tense markers is straightforward, a complication arises in the middle voice marking patterns. In fact, I believe this interaction is at the core of the failure to recognize the pervasiveness of the active/middle voice distinction in previous accounts. The active voice suffix -u is completely lost in the presence of a tense suffix. The combination of the middle voice suffix -nuҌ with the tense suffixes, on the other hand, is characterized by the disparity between the /nu/ component of the former and its glottalization component. The relevant morphological aspect in this constellation is that the /n(u)/ component of the middle voice suffix is clearly located in the same slot as the past and non-past suffixes, -ru and -tu, in their respective active forms. However, if a verb is marked by both, a tense suffix and the middle voice, the component /nu/ of the middle voice suffix is replaced by the tense suffixes. This behavior of the /nu/ component contrasts with the glottalization component, which is, as we have seen, independent of the morphological structure and is placed relative to the stem boundary following purely prosodic principles. Since /nu/ cannot co-occur with the tense suffixes, the medium forms of past and non-past can be given as -ruҌ and -tuҌ, respectively, since the underlying forms are open syllables. These forms differ from their active counterparts only through the presence of the glottal stop coda. Table 6. Interaction between voice morphology and tense suffixes Active

Middle

Extended Stem

-u

-nuҌ

Past

-ru

-ruҌ

Non-Past

-tu

-tuҌ

Except for the glottalization, the active and middle voice forms are thus identical. This glottalization has previously been analyzed as an affectedness morpheme in the accounts of Zide, Anderson and myself. Since the /nu/ of the middle voice marking is missing in these forms, the past and non-past middle voice forms were not recognized as forms representing the same category as the corresponding extended stem with -nuҌ. However, the following paradigms of the active verb jer ‘to run’ and the middle verb koko ‘to sit’ show that these forms are indeed all part of a pervasive voice system rather than representing distinct categories.

Grammatical voice in Gorum

143

Table 7. Voice and tense forms Active

Middle

Extended Stem

jer-u run-ACT ‘to run’; ‘running’

koko-nuҌ sit-MID ‘to sit’; ‘sitting’

Past

jer-ru run-ACT:PST ‘He/she/it ran.’

koko-ruҌ sit-MID:PST ‘He/she/it sat.’

Non-Past

jer-tu run-ACT:NPST ‘He/she/it will run.’

koko-tuҌ sit-MID:NPST ‘He/she/it will sit.’

The two tense and middle suffixes -ruҌ and -tuҌ interact with subsequent suffixes in the same way as -nuҌ and can therefore be given the same abstract description for their rhyme. Table 8. The phonological representation of middle voice tense affixes MID

-nuҌ

/nVG/

MID:PST

-ruҌ

/rVG/

MID:NPST

-tuҌ

/tVG/

The morphophonological behavior of these three suffixes is identical. The following examples show the position of the affixes and the prosodically motivated placement of the glottalization in combination with other suffixes. (30)

ne-koko-ruҌ 1sA-sit-MID:PST ‘I sat.’

(31)

koko-r-a֓ j sit-MID:PST-VEN ‘He/she/it sat.’

144

Felix Rau

(32)

koko-t-a֓ j sit-MID:NPST-VEN ‘He/she/it will sit.’

(33)

яuku-r-‫ے‬ƾ be-MID:PST-1sU ‘I had/possessed ...’

(34)

яuku-r-iҌleƾ be-MID:PST-1pU ‘We had/possessed ...’

6. The irregular verb duku ‘to be’ Examples (33) and (34) are instances of the irregular verb яuku ‘to be, to remain, to have’, which possesses an additional “tense-neutral” form. This form is an exception to the system described above. The general morphophonological behavior of яuku is irregular, as it is one of the very few verbs whose stem is altered through affixation, as in (35). (35)

ne-ku-ruҌ 1sA-be-MID:PST ‘I was.’

The formation of the extended stem as well as the affixation of the past and non-past suffixes for this lexeme are regular. However, the tenseneutral form does not carry a tense suffix, nor does it display the complete -nuҌ suffix: The tense-neutral form blocks the /n/ component of the middle suffix without filling the relevant position. As such, the tense-neutral form differs from the extended stem as shown Table 9. Table 9. Forms of ‫ܩ‬uku ‘to be’ Extended Stem

яuku-nuҌ

Past

яuku-ruҌ

Non-Past

яuku-tuҌ

Neutral

яukuҌ

(< яuku+Ø+nuҌ)

Grammatical voice in Gorum

145

Interestingly, if no further suffix is present, the glottalization occurs on the final /u/, which appears to be part of the stem. On the other hand, this /u/ interacts with following suffixes in a way similar to the /u/ of the tense and voice affixes. The neutral tense form of ‫ܩ‬uku could thus be represented as /‫ܩ‬ukVG/, analogous to the other middle voice forms. Thus, if a vowelinitial suffix follows the tense-neutral form, the glottalization component of the middle voice occurs with this suffix, as in (36). (36)

яuk-‫ے‬ƾ be:NEUT:MID-1sU ‘I have/possess ...’

7. Voice as a grammatical category in Gorum – Summary Voice marking in Gorum is a pervasive and mostly regular morphological process. As I have argued, all verb forms are based on the extended stem, which consists of the bare stem and the active voice suffix -u or the middle voice suffix -nuҌ. Both suffixes interact with the tense morphology that is located in the same morpheme slot. However, the glottalization component of the middle voice marker behaves in a way fundamentally different from the rest of the suffix. This voice system of Gorum has never been recognized in its full extent. This is remarkable, since virtually all other Munda languages make a distinction between active and middle voice in their verb system. The suffixes -u and -nuҌ (sometimes represented as -nu) have generally been analyzed as transitive and intransitive infinitive suffixes (e.g. Zide 1972, Anderson and Rau 2008). Astonishingly, the categories have not been connected to the voice distinction in other Munda languages, even though it has been phrased in terms of a transitive/intransitive distinction in several of those languages as well. In my view, the source for this problem lies in the misinterpretation of the morphophonological behavior of the middle-marking suffix -nuҌ. As we have seen, there is an apparent difference between the behavior of the /nu/ component of the middle voice marker with its fixed morphological position and its interaction with tense suffixes, and the prosodic nature of the glottalization, which ignores morphological slots. This difference has lead previous accounts to analyze the two components as separate morphemes representing completely different categories: an intransitive

146

Felix Rau

infinitive in the case of -nu(Ҍ) and affectedness (or undergoer focus) in the case of the glottalization. One problem with this analysis, however, is that the intransitive category is obligatorily connected to affectedness marking. In any case, since the assumed intransitive infinitive and the affectedness morpheme have a similar function, all previous approaches have had to stipulate a connection between the two categories while focussing on the formal, lexical character of the intransitive and the semantico-pragmatically motivated character of the affectedness marker. This led to a situation where the middle voice in a sentence such as (37) was interpreted as an instance of the affectedness category, while the middle voice in (38) was interpreted as an instance of the category intransitive. This obscures the regularity of the system and the fact that both forms are based on the middle voice form gi’ї-nuҌ ‘to be visible, to look like s.th., to be seen’ that corresponds to the active gi’ї-u ‘to see’. (37)

gi'ї-j-a֓ j=ni see-MID:PST-VEN=PROG ‘It is visible’

(38)

aяuka gi'ї-n-a֓ j ar-luҌn shelter see-MID-VEN NEG:PST-lift ‘The shelter is not visible’

Under the new analysis, (37) and (38) do not involve different categories such as affectedness and intransitive. They do, however, feature competing progressive constructions in Gorum. The first construction consists of a combination of the past form of a verb, here with the -ju allomorph of the past suffix -ru, with a clitic =ni. Middle voice marking is in this construction reduced to the glottalization /VG/, as the /nu/ part is replaced by an allomorph of the paste tense suffix -ru. The second construction combines the extended stem with the auxiliary luҌn ‘to lift’ in a periphrastic construction. In this case, no tense morphology is present, so that the middle voice suffix occurs in its /nVG/ form. Therefore, what is at issue here is not a matter of category, but rather of the form the middle voice marking takes with the extended stem and past verb forms. Both examples thus contain the verb gi'ї ‘to see’, the middle voice morpheme -nuҌ as well as the ventive -aj and are in progressive form, although the latter is expressed through two different constructions. Apart from that, the examples only differ in grammatical polarity.

Grammatical voice in Gorum

147

The received view for the last forty years thus adhered to an analysis that put these two forms in (37) and (38) in completely different categories. Furthermore, it assumed that Gorum lacked a voice system, even though practically every other Munda language was analyzed as possessing precisely such a system. The analysis of the verb system presented here thus differs radically from previous analyses by Zide (1990), Aze (1973), Anderson (2007) as well as Anderson and Rau (2008). All these analyses assumed that Gorum has a transitive/intransitive distinction which is only represented in some word forms, such as the infinitive, negative and irrealis forms, as well as in imperatives. This was distinct from the category that is called affectedness by Zide (1990), Anderson (2007) as well as Anderson and Rau (2008), and undergoer focus by Aze (1973). These two categories were viewed as independent parts of the verb system. While the intransitive category seemed to be a formal property of lexemes, affectedness was mostly viewed as an optional category with a more or less specific semantic function. While my analysis is a clear deviation from this traditional view on the verb system of Gorum, it brings Gorum much closer to the other Munda languages. Historical evidence from Sora and the Kherwarian languages supports this analysis. The comparative evidence – especially the connection of -nuҌ to the middle voice markers of other Munda languages – is discussed in the following section. 8. The history of voice marking The active/middle voice distinction in Gorum is not only similar in function and structure to the voice systems of other Munda languages, it can also be connected to them etymologically. As in other Munda languages, the voice system found in Gorum pervades the verb system of finite and non-finite verb forms. The active voice suffix -u is difficult to relate to the active marking morphology in other Munda languages, as it is the unmarked vowel and phonologically very unstable – the current understanding of the Munda sound laws does not allow for a reliable reconstruction. However, comparative evidence on middle marking is more telling and strongly suggests that, although Gorum has undergone some morphological changes from the original state, the middle voice marker -nuҌ is a reflex of an old middle voice marker. The following discussion lays out the evidence for this conclusion.

148

Felix Rau

In present-day Gorum, voice (VOICE) in the form of the middle voice marker -nuҌ and the active voice -u occurs in the first slot of the suffix domain, following the bare stem of the verb. This is the first of three suffix positions, which the voice morphology shares with the tense (TNS) suffixes. The template below summarizes the constellation.

BARE STEM

1

2

3

Clitics

-TNS/VOICE

-UNDERGOER

-VENTIVE

(=PL)

Figure 5. Suffix domain of the Gorum verb

Evidence from other Munda languages suggests that the middle voice marking suffix -nuҌ is not an innovation of Gorum. Sora, Gorum’s closest relative, has a suffix -n which is clearly cognate to -nuҌ in Gorum. This suffix is the distinctive feature of one conjugational paradigm in Sora, which is called middle voice (MID) here, as it is the functional equivalent of middle voice in Gorum and other Munda languages. It is identical to paradigm IV in Biligiri (1965). The verbal system of Sora is considerably different from that of Gorum, even though most of the suffixes involved are cognate with Gorum suffixes. Sora has paradigmatic inflectional patterns that differ with regard to intransitive/middle voice marking, ventive marking and object marking (as in the impersonal paradigm). The interpretation of a given suffix may vary in different inflectional paradigms. This paradigmatic conjugational organization has no direct correspondence in Gorum, where all affix classes combine freely and their interpretation is constant. Four paradigms are given in the Table 10, following Biligiri (1965). These paradigms are a selection of the paradigms presented by Biligiri (1965) and cover only parts of the complexity of the verbal morphology of Sora. They do however contain all the morphological structure relevant for a comparison of the indicative verb forms of Sora and Gorum. For a description and analysis of the Sora verb system, the reader may consult Stump (2005), who also discusses the multiple functions of the morpheme called middle voice here. As with Gorum, the suffix domain is the relevant part of the verb for the morphology of voice marking. All forms presented here are based on the stem (STEM) with a directly following tense suffix (TNS).

Grammatical voice in Gorum

149

Table 10. Verb paradigms of Sora ACT

MID

MID/VEN

IMERSONAL

1 Sg

STEM-TNS-ay

STEM-TNS-n-ay

STEM-TNS-n-ay

STEM-TNS-iѪ

2 Sg

STEM-TNS-ѓ

STEM-TNS-n

STEM-TNS-n-ay

STEM-TNS-Ωm

3 Sg

STEM-TNS-ѓ

STEM-TNS-n-ay

STEM-TNS-n-ay

STEM-TNS-e

1 Pl incl.

STEM-TNS-be

STEM-TNS-n-be

STEM-TNS-n-ay-be STEM-TNS-ay

1 Pl Ω-STEM-TNS-ay excl.

Ω-STEM-TNS-n

Ω-STEM-TNS-n-ay

STEM-TNS-lѓn

2 Pl

Ω-STEM-TNS-ѓ

Ω-STEM-TNS-n

Ω-STEM-TNS-n-ay

STEM-TNS-ben

3 Pl

STEM-TNS-ѓ-ji

STEM-TNS-n-ji

STEM-TNS-n-a-ji

STEM-TNS-Ω-ji

The paradigm called active voice (ACT) here features the suffix -ѓ as the relevant voice morpheme. This suffix is probably cognate to the active voice suffix -u in Gorum. Based on Zide 1982, the active voice suffix of Proto-Sora-Gorum has to be reconstructed as *-Ω (see Zide 1982: 332-333 for the relevant correspondences). However, the vocalism of Sora-Gorum is not well understood and the suffix is difficult to relate to other formatives outside of Sora-Gorum. Consequently, this brief discussion of the history of the Gorum active voice must suffice for the present. The simple middle voice paradigm (MID) in Sora features the middle voice suffix -n. The suffix -n follows the tense suffixes and is thus positioned in the same slot as the active voice suffix. The Sora middle voice suffix is clearly cognate to the Gorum suffix -nuҌ; the Proto-Sora-Gorum suffix can unequivocally be reconstructed as *-n (cf. Zide 1982: 22). The middle voice ventive paradigm (MID/VEN) adds the ventive suffix -ay [-aj] after the voice suffix in all verb forms.6 This suffix is identical in Sora and Gorum and seems to be an innovation of this branch of Munda languages. The so-called impersonal paradigm has object markers following the tense suffixes. The first person inclusive suffix -ay in the impersonal paradigm is probably identical with the ventive suffix -ay, which also functions as a first person marker in other paradigms. This suggests that the object markers are in the same morphological slot as the ventive suffix. Following this morphological position, the suffix -be marks plural in most of the first person inclusive forms, while the suffix -ji [-ܱi] marks third person plural. Based these paradigms, the suffix domain of Sora can be analyzed with four slots following the stem. The first position after the stem is occupied

150

Felix Rau

by the tense suffixes. The active voice suffix -ѓ and the middle voice suffix -n follow in the second slot. The third slot can be occupied by either the ventive or the undergoer suffixes and, finally, the fourth position is the place of the plural person suffixes. The structure of the suffix domain in Sora is given in the template below.

STEM

1

2

3

4

-TNS

-VOICE

-UND/VEN

-PL

Figure 6. Suffix domain of the Sora verb

The main differences between the Gorum and the Sora verbal system pertain to structural aspects and not so much to the form of the markers itself. For instance, tense suffixes and the -n middle marker can co-occur in Sora, but are mutually exclusive in Gorum, where only the glottal element of the middle voice suffix -nuҌ survives in combination with the tense suffixes. On the other hand, undergoer suffixes and ventive can be marked at the same time in Gorum, while this seems to be impossible in Sora. Furthermore, there is a significant difference between the two Sora formatives -be and -ji [-ܱi] indicating first and third person plural, respectively, and their Gorum cognates =bu and =gi. In Sora, these formatives have been analyzed as suffixes by Biligiri (1965). Their Gorum counterparts, on the other hand, are clearly clitics and have distinct morphosyntactic properties setting them apart from the suffixes. Nonetheless, a direct comparison of the two systems clearly shows a common morphological structure from the stem onwards. 1 Gorum

BARE STEM

-TNS/VOICE

Sora

STEM

-TNS

2 -VOICE

3

4

-U

-VEN

-U/VEN

5

Clitics (=PL)

-PL

Figure 7. Comparison of the Gorum and Sora suffix domains

This comparative evidence suggests that the ancestor of Sora and Gorum had four morphological positions following the verb stem and another position for plural marking, which had a different status than the four preceding positions. Since it is more likely that a clitic becomes bound to the verb more closely and turns into a suffix than vice-versa, it is assumed here that Gorum preserves the original status of the plural formatives. Proto-Sora-Gorum thus possessed four suffix slots and one position for

Grammatical voice in Gorum

151

enclitics. The original place of the voice suffixes was in the slot following the tense suffixes and preceding the undergoer and ventive suffixes. The morphological structure of the suffix domain of the verb in Proto-SoraGorum thus has to be reconstructed as follows.

STEM

1

2

3

4

Clitics

-TNS

-VOICE

-U

-VEN

(=PL)

*-n (MID) *-Ω (ACT) Figure 8. Reconstructed Proto-Sora-Gorum suffix domain

The structure found in present-day Gorum, with the voice morphology directly following the stem, is a later development that is the result of the merger of slots 1 and 2. This gives the following historical picture for Gorum:

STEM

1+2

3

4

Clitics

-[TNS/VOICE]

-U

-VEN

(=PL)

Figure 9. Historical development in Gorum

Supporting evidence for this historical scenario comes from the middle voice morphology of Kherwarian languages such as Santali, Mundari and Ho. Santali, for instance, does not have independent voice markers, but portmanteau markers for tense and voice, like most other Munda languages. Four pairs of these suffixes show a regular correspondence between the active and middle voice form. The below table follows Neukom (2001: 62) and shows a component /n/ in the middle voice suffixes that is missing from their active voice counterparts. Similar suffixes can be found in all Kherwarian languages.

152

Felix Rau

Table 11. Santali tense affixes ACTIVE

MIDDLE

PAST

-ke't

-en

PLUPERFECT

-le't

-len

PERFECT

-aka't

-akan

IRREALIS

-le

-len

The analogy to the situation in Sora-Gorum is striking. Not only do the middle forms contain a reflex of the formative *-n, but its position also fits the reconstructed morphological structure perfectly. The /n/ follows the material common to both tense forms, suggesting an earlier sequence of a tense morpheme followed by the voice morphology. Thus, even though Santali does not possess an independent middle voice suffix -n, the system retains reflexes of precisely such a marker that followed the tense morphemes. The degree of similarity between the verb systems of Kherwarian and Sora-Gorum is very nicely exemplified by the following two word forms: (39) de-le-n (Sora) get.up-PST-MID ‘he got up’ (Biligiri 1965: 235) (40) dal-len-a=e (Santali) strike-PLUP:MID-IND=3sS ‘he had been struck’ (Neukom 2001: 79) The Sora example shows the past tense suffix -le, which is cognate to Gorum -ru and presumably goes back to a Proto-Sora-Gorum form *-lV, most likely *-lΩ, followed by the middle voice suffix -n. The Santali example features the pluperfect middle voice suffix -len. The first part of this suffix, -le, is common to both the active and middle voice variants of the suffix. Assuming that the Sora past tense suffix -le is cognate to this common part of the pluperfect morphemes in Santali, -le-n in Sora and -len in Santali differ only in that Sora preserves a morpheme boundary that has been lost in Santali. Moreover, other Kherwarian languages such as Mundari (Osada 2008: 119) as well as Ho and Bhumij (Anderson, Osada and Harrison 2008: 221) seem to have preserved this morpheme boundary as well.

Grammatical voice in Gorum

153

Further evidence supporting the reconstruction of the middle voice marker comes from Korku, Juang and Kharia. The Korku intransitive/middle past suffix -en (Nagaraja 1999: 71; Zide 2008: 274) and the Juang intransitive/middle simple past suffix -an (Patnaik 2008: 531) are parallel to the middle voice past -en in Santali. The Juang suffix is the only unequivocal remnant of the middle marker -n in South Munda outside of Sora-Gorum. There are other combined tense/voice forms – shared by Juang and Kharia – that could be related to the middle voice formatives containing *-n, namely the Juang intransitive/middle simple future -na (Patnaik 2008: 531) and the similar irrealis middle clitic =na in Kharia (Peterson 2008: 462). However, in this case the /n/ is morpheme-initial, setting the suffixes in question apart from all other tense-voice portmanteau suffixes. This makes the connection somewhat dubious, though still possible. In summary, the comparative evidence suggests that Munda had a middle voice marking suffix *-n that is well preserved in Sora-Gorum as well as in Santali and other Kherwarian languages and seems to have left traces in Korku, Kharia and Juang. The suffix has been entirely lost in Gutob, Remo and Gta‫ݦ‬, which are, geographically speaking, the closest neighbors of Gorum. The data also suggest a morphological reconstruction in which the voice morphology – or at least the middle marker *-n – followed the tense-aspect suffixes. This system is well preserved in Sora and partially preserved in fossilized form in Santali and other Kherwarian languages. The reconstruction of the morphological structure of the Proto-Sora-Gorum verb is thus confirmed by comparative evidence from all major branches of the family except Gutob, Remo and Gta‫ݦ‬. All reconstructions of the Munda family assume a primary branching into North Munda (Kherwarian languages and Korku) and South Munda (including, Gorum, Sora, Kharia, Gutob). The data from Proto-Sora-Gorum and Kherwarian allow the first two slots of the reconstructed suffix domain of the Proto-Sora-Gorum verb to be posited for Proto-Munda. The restricted reconstruction is given in the template below. It is similar to the relevant part of the reconstruction of the Munda verb by Pinnow (1966: 179–181).

STEM

1

2

-TENSE/ASPECT

-VOICE *-n (MID)

Figure 10. The partially reconstructed Proto-Munda verb

154

Felix Rau

9. The historical development of voice in Gorum Tense and voice morphology is positioned in the same slot in Gorum as a result of the merger of two originally distinct slots. Despite of this, tense and voice morphology has remained distinct in Gorum. The same merger occurred in other Munda languages such as Santali, Kharia or Korku, but unlike in Gorum, the voice and tense morphemes in these languages formed portmanteau suffixes as a result. In Gorum, on the other hand, the middle voice suffix moved into the same morphological position as the tense markers, but retained its independent status. So far, I have ignored the history of the other component of the Gorum middle voice marker -nuҌ, namely the glottalized rhyme /V‫ݦ‬/ or /V࡭ C/. Nothing in the comparative data suggests that the glottalization is an original part of the middle voice marking, although it is synchronically an integral part of middle voice in Gorum. Consequently, the question arises as to the source of this component. There is currently no known cognate for the glottalization component of the Gorum middle voice suffix in other Munda languages. To understand its historical development in Gorum and to find related phenomena in other Munda languages, it is hence crucial to determine whether the glottalization was originally connected with the middle voice marking slot in which the suffix -n is still positioned in present-day Sora, or whether it was always connected to the position following the stem – the morphological slot into which the middle voice marking moved in Gorum. The synchronic behavior displayed by the glottalization /G/, with its placement in relation to the right stem boundary, follows prosodic principles, rather than morphological ones. Its sole morphological reference point is the right boundary of the stem; the morphological structure of the suffix domain is irrelevant for its occurrence. As shown above, the right stem boundary of a regular verb in Gorum always coincides with a syllable boundary, so that prosodic and morphological structures are necessarily aligned in this position. This is the crucial structural configuration that determines the positioning of the glottalization /G/. This alignment of prosodic and morphological structure is missing in other positions in the suffix domain, so that a placement according to prosodic principles in relation to the original slot of the middle voice suffix *-n is virtually impossible. This clearly favors a historical situation in which the /VG/ component of the middle voice suffix was connected to the position immediately following the stem boundary, rather than to the original position of the middle

Grammatical voice in Gorum

155

voice marker -n, i.e. suffix slot 2 in Figures 8 and 10. Thus, internal evidence from Gorum suggests that the association of the middle voice marker with glottalization occurred after the tense suffix slot and the middle voice suffix slot coalesced in Gorum. Hence, two distinct morphemes have to be reconstructed for Gorum: a middle voice suffix *-n, that changed its morphological position but is well supported by comparative evidence, and a morpheme of unknown function that is represented by the glottalization /G/ and was positioned relative to the stem. Interestingly, if not ironically, the picture that arises from this historical analysis – with its two distinct morphemes, the middle voice suffix -n and a second morpheme of uncertain function represented by glottalization – is virtually identical to the traditional analysis of present-day Gorum as found in Aze (1973), Zide (1990) as well as Anderson (2007) and Anderson and Rau (2008). While I have shown here that the old analysis does not satisfactorily capture its morphology and function, the historical evidence explains the peculiar behavior of the middle voice suffix -nuҌ, which was the focus of the preceding discussion. Further research is necessary to understand the history of the glottalization, which is now part of the middle voice marking in Gorum. 10. Conclusions Voice is a general verbal category in Gorum. The voice system makes a distinction between active and middle voice, and every verb form – except for the bare stem – is marked for either of these two voices. All finite verb forms are based on the extended stem, which consists of the bare stem and either the active voice suffix -u or the middle voice suffix -nuҌ. The voice system of Gorum is thus similar to the system found in other Munda languages, and comparative evidence allows for a reconstruction of voice morphology for Proto-Sora-Gorum and partially for older stages. A part of the middle voice marker -nuҌ can be shown to have cognates in Sora, Santali and some other Munda languages. The distribution of the cognates allows for further statements about the history of middle voice marking in the Munda family. All versions of the Munda family tree assume a fundamental division between North Munda languages, such as Santali, and South Munda languages, such as Sora and Gorum. On this basis, the middle voice suffix *-n and its morphological position in the

156

Felix Rau

second suffix slot following the stem and the tense-aspect suffixes can be reconstructed for Proto-Munda. Abbreviations 1s = first person singular; 1p = first person plural; 2s = second person singular; 2p = second person plural; 3s = third person singular; 3p = third person plural; A = actor; ACT = active voice; caus = causative; comp = complementizer; DEF = definite; dem.prox = proximal demonstrative; det.dist = distal determiner; DIR = direct case; IND = indicative; IRR = irrealis; LV = loan verb; MID = middle voice; neg = negative; NEUT = neutral tense; NPST = non-past tense; OBJ = object; OBL = oblique case; PL = plural; PLUP = pluperfect; PROG = progressive; PST = past tense; S = subject; U = undergoer; VEN = ventive.

Notes *

1.

2. 3. 4. 5. 6.

I would like to thank John Peterson for his comments on the paper and the encouragement to write it as well as Juliette Huber for patiently reading and improving earlier versions of this text. Furthermore, I have to thank David Stampe for his extremely helpful remarks on the Sora data. The data used in this article have been collected during two field trips which were partially funded by the Leiden University Fund. The suffix -ej marks actors on verbs with two or more arguments and on verbs with a single actor argument, but also the undergoer on verbs with a single undergoer argument, such as aяaҌ ‘to be thirsty’. Morphologically, it behaves identically to the undergoer suffixes and both types consequently subsumed into a single category here. See Anderson (2007) for a different approach. There are secondary derivational processes that can alter the association of these verbs to a given voice category, but these will not be discussed here. The second verb uj ‘to go’ is an auxiliary-like explicator verb that adds a component of non-volitionality or undesirability and has many properties that fit the affectedness category as described in Anderson (2007). The term prosodic structure as used here refers to phonological structures larger than the segment, i.e. syllables and higher units such as the phonological (or prosodic) word. The form -a in the third person plural is a phonological variant of -ay which occurs in the combination with the third person plural marker -ji.

Grammatical voice in Gorum

157

References Anderson, Gregory D. S. 2007 The Munda verb: typological perspectives. Berlin: Mouton de Gruyter. Anderson, Gregory D. S., Toshiki Osada and K. David Harrison 2008 Ho and other Kherwarian languages. In The Munda Languages, Gregory D. S. Anderson (ed.), 195–255. London: Routledge. Anderson, Gregory D. S. and Felix Rau 2008 Gorum. In The Munda Languages, Gregory D. S. Anderson (ed.) 381–433. London: Routledge. Aze, Richard 1973 Clause patterns in Parengi-Gorum. In Patterns in clause, sentence, and discourse in selected languages of India and Nepal Part I, Richard L. Trail (ed.), 235–312. Norman: Summer Institute of Linguistics. Biligiri, Hemmige Shriniwasarangachar 1965 The Sora verb: a restricted study. In Indo-Pacific Studies, Part II: descriptive Linguistics (Lingua 15), G. B. Milner and Eugénie J. A. Henderson (eds.), 231–250. Amsterdam: North-Holland. Cook, Walter Anthony 1966 A descriptive analysis of Mundari: A study of the structure of the Mundari language according to the methods of linguistic science, with particular attention to the units of sound, the units of meaning, the units of grammar, and their mutually contrastive arrangement patterns. Ph. D. diss., Georgetown University. Fernandez, Frank 1969 A grammatical sketch of Remo: a Munda language, Ph. D. diss., University of North Carolina at Chapel Hill. Griffiths, Arlo 2008 Gutob. In The Munda Languages, Gregory D. S. Anderson (ed.) 633–681. London: Routledge. Kemmer, Suzanne E. 1993 The middle voice. (Typological studies in language 23.) Amsterdam: Benjamins. Nagaraja, Keralapura Shreenivasaiah 1999 Korku language: grammar, texts, and vocabulary. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies. Neukom, Lukas 2001 Santali. München: Lincom.

158

Felix Rau

Osada, Toshiki 2008 Mundari. In The Munda Languages, Gregory D. S. Anderson (ed.) 99–164. London: Routledge. Patnaik, Manideepa 2008 Juang. In The Munda Languages, Gregory D. S. Anderson (ed.) 508–556. London: Routledge. Peterson, John 2008 Kharia. In The Munda Languages, Gregory D. S. Anderson (ed.) 434–507. London: Routledge. 2011 A Grammar of Kharia: A South Munda Language. (Brill’s studies in South and Southwest Asian Languages 1.) Leiden: Brill. Pinnow, Hans-Jürgen 1966 A Comparative Study of the Verb in the Munda Languages. In Studies in Comparative Austroasiatic Linguistics, Norman H. Zide (ed.), 96–193. The Hague: Mouton. Ramamurti, Giঌugu Veৄkata 1931 A Manual of the So:ra: (or Savara) Language. Madras: Government Press. Starosta, Stanley Herman 1967 Sora syntax: a generative approach to a Munda language, Ph. D. diss., University of Wisconsin-Madison. Stump, Gregory T. 2005 Referrals and Morphemes in Sora Verb Inflection. In: Yearbook of Morphology 2005, Geert Booij and Jaap van Marle (eds.), 227–251. Dordrecht: Springer. Zide, Arlene R. K. 1972 Transitive and Causative in Gorum. Journal of Linguistics 8: 201– 216. 1982 A reconstruction of Sora-Juray-Gorum phonology, Ph. D. diss., University of Chicago. 1990 Cislocative, affectiveness, and the experiencer subject in Gorum. In: Experiencer Subjects in South Asian Languages, Manindra K. Verma and Karuvannur Puthanveettil Mohanan (eds.), 343–354. Stanford: Center for the study of language and information. Zide, Norman H. 2008 Korku. In The Munda Languages, Gregory D. S. Anderson (ed.) 256–298. London: Routledge.

Regional Reports

India Pingali Sailaja

1. Introduction This report aims to present an overview of the published research on linguistics in India. It is delimited to works written in English, from 2005 to 2010, and is also somewhat constrained by accessibility of works. Linguistic research, apart from being published by the well-known publishing houses, is also available in the many journals of different Universities and departments. All of these journals are not easily accessible. Two major linguistic journals are Indian Linguistics and International Journal of Dravidian Lingusitics. Several articles are published in these journals regularly. Some select articles from them are referred to in the sections that follow. Indian linguistics is more oriented towards applied areas than theroretical areas. A very broad view of linguistics is adopted here. It includes theoretical and applied areas and also grammars and course books for languages. Additionally, there is acknowledgement of the considerable amount of work done on English—focussing on the teaching of the language and its linguistic, political and cultural ramifications. This report is divided into what seem to be the major areas dealt with, in the research done in this region. It is divided into sections according to the language families, and additionally some of the emerging areas that have got attention recently. We begin with research on Dravidian languages, then move on to Indo-Aryan languages, and the North-East. Finally, areas that have always been discussed in linguistics—English (education) in India and more recently, endangered languages, clinical linguistics etc. Indian grammatical tradition has always received considerable attention and the last few years have seen a spurt in research. The final section on miscellaneous items contains those works that are not easily classifiable into any of the areas described in the previous sections. Many works there belong in more than one category. Revised versions of books reflect the trends in current thinking and are thus included in this survey, as are South Asian/Indian editions of popular books, especially in language education.

162

Pingali Sailaja

This report is not a critical assessment but merely a brief annotated review of the research done in India on language and linguistics. 2. Dravidian languages Trautmann (2006) documents the evidence regarding the interconnectedness of the Dravidian language family constituents. His contention is that linguistic analysis cannot be carried out in isolation; it has to take into account ethnographic features also. He argues that Indian linguistic analyses were incorporated into European accounts of historical linguistics. Kavithaasaran (2008) is a reprinted version of the second edition (which, according to the editor, is the authentic version rather than the commonly prevalent third edition) of Caldwell’s comprehensive description of the grammatical aspects of the Dravidian languages. Sjoberg (2009), in her collection of essays written over several years, explores aspects of the Dravidian people and their culture; some essays discuss specific linguistic features. P. S. Subrahmanyam’s (2008) work on Dravidian languages, meant to be a textbook, deals with the classification into groups of the Dravidian languages from the proto forms, and comparative phonology. It gives details about each of the Dravidian languages. Pilot-Raichoor (2008) attempts an explanation of the so-called zero negative in Dravidian. Krishnamurti (2009) is a volume consisting of fourteen reprinted articles on Telugu written over a number of years. The theroetical orientation in this work changes according to the original time of writing. The articles are primarily on Telugu inflections, Sandhi, compound verbs, complex predicates and other aspects of Telugu. A couple of articles are on Indian languages: causatives and stative expressions. Sastry & Murthy (2009) trace the evolution of the Telugu script from pre-3rd century BCE. Balusu (2010), although claiming to be on Dravidian, presents earlier research on Kannada and own research on Telugu. It is an experimental investigation of how children and adults interpet ambiguous sentences. Emeneau (2006a) is a recently discovered article (published along with another article on Toda in the same volume, 2006b) on Dravidian noun compounds and states some of the problems in their treatment. Vasanta (2006) presents an experimental study of the role of semantic transparency in processing compounds by adult Telugu speakers. Nagamma Reddy (2009) discusses Telugu fricatives.

India

163

Ramakrishna Reddy (2009) demonstrates that the number system of Manda, one of the South-Central Dravidian tribal languages, is unique and that several linguistic levels interact. Gamliel (2009) discusses Jewish Malayalam. 3. Indo-Aryan languages Ananthanarayana (2005) deals with the Prakrits, tracing the origins and describing the dialects, the grammatical literature, and the linguistic features. The paper also relates the Prakrits to other languages. Sar (2007) discusses the passive marker in Kashmiri. Sarma (2007) discusses variation in non-phonemic vowel length for Assamese. Dutta Baruah (2007) presents a contrastive analysis of the morphology (derivation, reduplication, compounding and inflections) of Assamese and Oriya, with the particular aim of serving a pedagogical purpose. Mahanta (2009) compares prominence in Bengali, Oriya and Assamese. Das (2006) discusses the structure of the Oriya syllable according to the sonority sequencing principle. Dalai (2008) discusses some sociolingusitic characteristics of Oriya personal pronouns. Mohanty & Malik (2008) is a collection of essays on aspects of literature, culture and language, primarily of Orissa. Two of these essays deal with language movements in colonial Orissa. Bardhan (2010) examines nominal derivational affixation in Chakma, an Indo-Aryan tribal language spoken in the Chittagong Hill-tract of Bangladesh, and considers the morpho-phonemic changes that take place. The topics studied are vowel harmony, vowel copying and obstruent voicing. Dan (2010) studies the linguistic features, particularly, the semantic features of number names of Bangla. Bandyopadhyay (2007) traces the history of grammar writing in Bengali. Bandyopadhay (2008) argues that Magadhi Prakrit, as it evolved, influenced Modern Bengali. Dakshi (2008) discusses the semantics of Bengali collocations. Wali (2006) is a collection of essays, some co-authored by Wali with others, on several syntactic features of Marathi. The book contains sections on Subjects and Agreement, Anaphors and Pronouns, Facts and Propositions, Structure of Causatives, and finally, Correlatives and Modals. Since these were written over a period of time, the theoretical framework changes across the chapters.

164

Pingali Sailaja

4. The North-East Morey & Post (2008) deal with North-Eastern languages. Many of the papers deal with more than one language at a time. The phonology, morphology, syntax and semantics, and the lexicon are discussed. Two papers deal with doing field-work and the quest for a script. Joseph & Burling (2006) compare the phonological structure of four Boro-Garo languages belonging to the Tibeto-Burman language family. The authors give phonological inventories for the languages and then move on to the phonological processes, including those apparent with compounding, affixation, simplification, etc. They also discuss the correspondences among the four languages. Nagaraja (2010) is a description of the phonology, morphology and syntax of Konyak a Tibeto-Burman language. Kapfo (2005) describes the linguistic aspects of Kheza—the phonology, morphology and syntax. The first part describes the ethnology of the Khezas. S. H. Sharma (2007) discusses the issue of the Manipuri script. S.H Sharma (2006) discusses negation in Manipuri. Singh & Sharma (2008) discuss nominalization in Manipuri. S. I. Singh (2010) studies the pronominal structures of Kom and Aimol, Tibeto-Burman languages. Walling (2010) examines the allative subject of Ao within a Minimalist framework. Anderson and Murmu (2010) present a hitherto unknown tribal language of Arunachal Pradesh called Koro, belonging to the Tibeto-Burman language family. The language is spoken by a small group within the Aka tribals. The structural features of this language, which seem to be different from other languages of the region, are described. Baishya (2008) describes the word-formation processes of Dimasa, a Tibeto-Burman language belonging to the Bodo-Garo group. A. K. Mishra (2009) examines the phonology of Indo-Aryan and English borrowings in Khasi. 5. Grammars and learning materials An interesting aspect is that several grammars have been written on different languages and a Routledge series is available. But here we mention some of those that have been produced in India. Ramanarasimham (2006) is a course on Telugu for those who wish to learn the language. Using a cyclic syllabus, the book introduces the script and the language to the reader.

India

165

Mahapatra (2007) provides a grammar of Oriya dealing with the phonology, the features of nouns and verbs, and indeclinables and sentences. The early chapters give an introduction to Orissa/Oriya and the orthography. Ramaswami (2007) is a grammar of Ho, a Munda language, spoken in Orissa, Bengal and Bihar. The grammar describes phonetics, phonology, morphology and syntax. The author’s aim is to establish that Ho is not a dialect of Mundari but an independent language, a claim made by the speakers themselves. 6. Endangered languages, neuro-lingusitics, language disorders and clinical lingusitics Manoharan & Gnanasundaran (2007) after presenting the background, document the vocabulary of Great Andamanese, a dying language. Ramakrishna Reddy (2010) highlights the details of endangered tribal languages of Central India, belonging to Munda, Dravidian and Indo-Aryan language families. He categorises them according to the degree of endangerment. R. K. Mishra (2006) reviews research on the prevalent view that specific brain areas are involved for specific language faculties, bringing in insights from recent research which finds that a wider neuronal network is involved. Kartikeyan and Priyadarshi (2010) is the result of a study on the coarticulatory problems of groups of Tamil-speaking children and adults with Down’s syndrome. The authors compare the results with those of ‘normal’ people. They also study the maturational effects. Dattamajumdar (2006) studies the learning process of Bangla consonant clusters with r, by hearing-impaired children with different degrees of impairment. Narang et al (2010) study the numerical and arithmetic abilities of twenty seven stroke affected patients, and compare the results with five normal adults. Indian Journal of Applied Linguistics (2010), vol 36/1-2 is a special issue on clinical linguistics edited by Duggirala Vasanta. 7. English (education) in India A great amount of work seems to have been done on researching a variety of aspects of English in India. Krishnaswamy & Krishnaswamy’s (2006) monograph deals with the history of establishing English as an integral part of the nation. They classify the historical process into five phases, from the

166

Pingali Sailaja

earliest times to the the current period of globalization. The work deals with various official documents. Advani (2009) looks at the issue of education and textbooks in a postcolonial society in relation to the ideology, culture, society and politics that inform them, and the role of language in all this. Dua (2008) focusses on language and language education in a multilingual society, the politics and ideology associated with these, and argues for a pluralistic approach. Ramanathan (2007) argues that there are ideological underpinnings in language policy and the English versus vernacular divide, focusing on Gujarat. S. Mishra (2006) covers the contribution of the University Grants Commission (UGC), a nodal agency that governs higher education, in terms of its funding and policy. The role of the UGC in the Indian education scenario is brought out exhaustively by the author. The work deals with aspects like organization, governance, etc. One chapter is devoted to the language problem in higher education. This chapter discusses the history of the issue, and the problems of the language of instruction in a multi-lingual country like India, and issues concerning language instruction, and the consequences of all of these for language policy. Dheram (2008) is an edited volume divided into two sections. Section I titled “Towards Critical English Language Learning and Teaching” is more theoretical in nature, while Section II titled “Critical Classroom Practices” examines classroom issues from different perspectives. Rassool (2007) deals with language education as it intersects with politics and cultural studies, across a few post-colonial countries. Nihalani et al’s (2005) work compares British and Indian English. The first part deals with usage and the second with pronunciation. Both parts have separate introductions; Part II is prescriptive and the prescribed variety is called IRP—Indian Recommended Pronunciation. Aslam & Kak (2007) is a textbook on phonetics and phonology. Nagaraj (2008) is an introduction to the concepts relevant in English language teaching. The second edition incorporates the changes and innovations that have taken place since the previous edition, such as the lexical approach and the Council of Europe’s A Common Framework of Reference for Languages: Learning, Teaching, Assessment. The work is meant to “educate teacher-participants/teachers about ELT, not train them in it” (Nagaraj 2008: viii). It is divided into two sections that deal with the approaches and methods in teaching English (beginning with the grammartranslation method), and a section (Part II) on techniques of teaching specific skills, teaching aids, and testing. Kaur (2006) is a book on the teaching

India

167

of English across a range of theoretical issues relevant in language teaching. It also consists of practical examples for teaching. V. D. Singh’s (2008) work is organised like a dictionary, defining and describing key terms in the areas of language teaching, learning and testing. P. C. Sharma (2005) is a guide to the teaching of English, and acts as an aid to English teachers. It deals with aspects like grammar, composition, reading and also literary genres. It shows how newspapers can be used in the English classroom. Banerjee’s (2006) work is on reading and writing in English, and constructing tests. One chapter is on the use of technologies in language teaching and a final chapter on some theoretical issues on language learning. One chapter deals with cultural aspects in language teaching. Khan (2005) focusses on issues concerning the teaching and testing of English, with particular reference to the bilingual context of India and Saudi Arabia. Sinha’s (2005) work gives a general background to English and the relevance of its teaching, and a history of English in India. The main thrust of the book is to present an empirical study of a comparison of the abilities of Oriya medium students with English medium students, by administering a series of tests. This is followed by an examination of the teaching methodologies and materials used in the different types of schools. Shah and Sinroja’s (2006) Seminar proceedings of the Department of English, Shri Jasani Arts and Commerce College, Rajkot deal with the “status and future of English in a global and Indian context” (Shah and Sinroja 2006: xi). Sheorey (2006) is an empirical study of Indian college students’ English language learning beliefs, how they relate to other groups of students (American, Korean and Chinese), the learning strategies of Indian students and whether there are any differences due to gender and proficiency levels. Raja & Nair (2007) is a developmental study of the English plural concord among Tamil children from 8 to 10 years. A number of English-teaching books continue to be published in India. Books that have done well over the years are often reprinted as Indian editions. While some of the works are meant for teaching general English, others claim to teach communication skills (for example, Sureshkumar et al 2010, Murphy 2009); some others are oriented towards specific disciplines, such as Business and Management (Ober 2005). Parrott (2010) is an English grammar book for English language teachers. Baker (2008) is meant to teach pronunciation at a basic level. Jain et al (2008) deal with phonetics, grammar, punctuation, vocabulary, letter writing, verbal communication, oral presentation, technical written communica-

168

Pingali Sailaja

tion, report writing etc. This work is prescribed as a textbook in various universities. Dutt et al (2008) have lessons to impart various skills. Shemesh & Waller’s (2010) work is a systematic guide to teaching English spelling, with focus on the sound-spelling relationships. The chapters include lesson plans and provide guidance based on the level of students being taught: beginners, intermediate and advanced. These are just a few of the several such books brought out in India. 8. Indian grammatical tradition Kapoor (2005) is an explication of Panini’s grammar primarily, beginning with an introduction to Indian lingusitic tradition and moving on to other works related to a discussion of Ashtadhyayi. Also included is a chapter on computer applications. This is a book form of the courses and lectures given by the author over a period of time. Grimal et al (2006) and Grimal et al (2007) are monumental volumes that present examples to the sutras in Panini’s grammar. Grimal et al (2006) which is the first volume, contains two parts. Part I gives examples and reference to the corresponding sutras as given by commentators. This is followed by a reference to the commentaries in which they appear. Part II gives the sutras as given in Ashtadhyayi and then refers to the examples. Grimal et al (2007) focuses on select compounds. Joshi & Roodbergen (2006, 2007) are part of the series of volumes brought out on Ashtadhayi by the Sahitya Akademi. The volumes contain translations and explanatory notes. The 2006 volume deals with sutras 7.3.1-7.3.120, and the 2007 volume with 7.4.1-7.4.97. Bharati & Kulkarni (2010) argue, by examining three specific sutras, that Panini had information coding in mind. They argue that these sutras point out precisely the location, quantity and manner in which information is coded in Sanskrit. Ganeri (2006) discusses different Indian theories of meaning, focussing primarily on Navya-Nyaya school of thought. Tatacharya (2005), the first volume of an intended series, deals with the notion of the sentence in Indian theories of language. A substantial introduction to the topic is in English and the rest of the work is written in Sanskrit. Ananthanarayana (2007) presents an account of the phonetic sciences as described in Indian grammatical tradition.

India

169

9. Language technology A considerable amount of research is being conducted in the country on language technology, both in universities and in the corporate world. Much of this research is available in the form of seminar proceedings. Bendre et al (2005, 2007), Darbari et al (2008) and Sharma et al (2009, 2010) are all proceedings of the annual International Conference on Natural Language Processing in the respective years. The papers in these volumes cover natural language processing, including interpretation and synthesis of human languages, machine translation, speech processing, and speech technology etc., of a range of languages, both Indian and foreign. Bhattacharya et al (2010) is a summary of the proceedings of a WordNet conference. Basu & Singh (2005) present the proceedings of another conference, with focus on morphology and phonology of Indian languages in the areas of computational linguistics and NLP. Sinha & Thakur (2007) analyse the multiple uses of kyaa in Hindi with the aim of capturing the multiple mapping patterns of translation into English. 10. Miscellaneous works Agnihotri & Dewan (2010) is a collection of papers read at a seminar on “the construction of knowledge”. Some essays are theoretical in nature, and deal with the relation between mind, cognition and knowledge. Others focus on practical work related to teaching different subjects. Hasnain & Chaudhary (2010) is a large volume of essays on a wide range of topics concerning language, with primary focus on India. The essays are divided into three sections: cultural, theoretical and applied areas. They are on aspects such as language deates in India, code-switching, linguistic aspects of specific languages, and issues concerning teaching languages. Mohanty (2008) discusses convergence across language families in India. Kapfo (2009) discusses mother tongue education within neurological frameworks. Mohanty et al (2009) is a collection of essays arguing for multilingual education and dealing with the theory and practice of it across several countries. Chaudhary (2009) is an account of the several foreign languages that came into India. The story of these languages and their assimilation is told in simple terms that are accessible to the lay reader. Apart from the better

170

Pingali Sailaja

known languages such as Sanskrit, Arabic, Persian, Turkish, English, French, Dutch and Portuguese, the work discusses the less known languages such as Greek, Hebrew and Armenian. The role of the East India Company and its language policy are also discussed. Patel et al’s (2007) collection of essays deals with the Indic scripts that are derived from Brahmi. The evolution of the scripts is discussed, followed by the relevance of the notion of ‘akshara’ to modern phonological theory. S. K. Sharma (2006a) and (2006b) are two volumes of essays on the language issue in India. The articles are on a range of topics such as a national/official language for India, the origin of Sanskrit, Hindi, Sindhi, the reports of the language commissions and so on. Khuong et al (2009) present the conference proceedings of the fifth Asian GLOW conference. Most of the articles are on specific syntactic aspects of different languages. Huddar et al (2006) is a manual, introducing concepts of language and communication for learners with disabilities. Consequently, many concepts that are not dealt with in mainstream textbooks (for eg. sign language) are discussed in detail in this work. Masica (2007) is a collection of papers from the Fifth International Conference on South Asian Linguistics (ICOSAL-5), Moscow, 2003. Divided into three Parts, it contains papers dealing with historical and typological aspects, etymology and descriptions of particular languages. Dash (2005) describes Corpus Linguistics and its use with reference primarily to Bengali. Mohanty & Kohler (2008) is a collection of essays dealing with various aspects of quantitative and statisitical measures in conducting linguistic research. K. Subrahmanyam (2008) is an explication of Indian grammatical theories (such as Purva-Mimamsa, Nyaya, Vaiseshika, etc.), and also Western linguistic theories (such as Humboldtian, structuralism, government and binding theory, etc.). Apart from describing the systems, the author also presents a comparative study of the oriental and the occidental. Tikkanen & Hettrich (2006) is a collection of select papers in linguistics, presented at the Twelfth World Sanskrit Conference, Helsinki, 2003. The papers range from common Indo-Iranian to Modern Indo-Aryan. “The general focus is, nevertheless, on Old and (Early) Middle Indo-Iranian”. (v) The linguistic aspects dealt with are: phonology, morphosyntax, etymology, loan words, substrata/convergence, translation, lexicography and computer processing. Levitt (2010) argues that Dravidian sound laws can

India

171

be seen in Indo-European as well. Fortunatov’s Law (l + dental becomes a single retroflex consonant), he argues, reflects a Dravidian combination. Satapathy et al (2007) brings together the discussions of a symposium on the relevance of Panini in analyzing English grammar. Sharada (2006) presents in book form a complete index with a citation index of the works by renowned linguist M. B. Emeneau. Tickoo (2008) is an unusual work on the contribution of Harold E. Palmer, linguist, grammarian and ELT practitioner. The book presents chronologically the wide range of Palmer’s thinking and contribution to his chosen disciplines. Viswanatham (2009) is a collection of essays written by the author over a period of several years. The essays are grouped under the following sections: language education, theoretical and comparative linguistics, language teaching and testing, and Telugu linguistics. Sahu (2005) contains articles on a wide range of topics ranging from communication, effects of modern science on English language and literature to Hinglish, and generative phonology. It is mostly on English as communication and its teaching. Kostic (2007) draws parallels between reduplication in Indo-European languages and Slavic languages. Tarai & Devaki Reddy (2007) provide an acoustic analysis of vowel length in Oriya English and compare it with British English. Rathna Kumar et al (2010) study the effects of auditory feedback on phonatory control among 40 young adults. Volumes 67 (2006), 69 (2008) and 70 (2009) of Indian Linguistics are felicitation volumes for D. P. Pattanayak, V. I. Subramoniam and Bh. Krishnamuti respectively. Ten articles in International Journal of Dravidian Linguistics volume XXXIV/2 (2006) are in honour of B. Emeneau, some of which discuss his contribution to different areas. International Journal of Dravidian Linguistics XXXIX/2 is a V. I. Subramonian memorial volume. These volumes cover a wide range of articles by many authors, such as Asher (2009), Bhaskarrao & Ladefoged (2009), Subbarao & Arora (2009), Krishnamurti (2010), Prabodhachandran Nayar (2010), P. S. Subrahmanyam (2010), and several others. It may be concluded that there is a wide range of research in terms of the number of areas of study covered, and that a considerable amount of research is being carried out in each area. New, redefined frames of reference also seem to be evolving.

172

Pingali Sailaja

Acknowledgements I am grateful to Amba Kulkarni, Sridhar Maisa, Rajarshi Mitra and S. Ganesh for their help in procuring relevant works, and to Rama Kant Agnihotri for inputs.

References Advani, Shalini 2009 Schooling the National Imagination: Education, English and the Indian Modern. New Delhi: Oxford University Press. Agnihotri, R. K & H. K. Dewan (eds.) 2010 Knowledge, Language and Learning. New Delhi: Macmillan. Ananthanarayana, H. S. 2005 Prakrits. International Journal of Dravidian Linguistics. XXXIV/1: 1–16. 2007 SikSaas as source book on Indian phonetics. International Journal of Dravidian Linguistics XXXVI/1: 79–86. Anderson, Gregory D. S. & Ganesh Murmu 2010 Preliminary notes on Koro, a ‘hidden’ language of Arunachal Pradesh. Indian Linguistics 71/1–4: 1–32. Asher, R. E. 2009 Literature as a possible source of data on spoken dialects of Dravidian languages, with particular reference to Malayalam. Indian Linguistics 70/1–4: 39–46. Aslam, Mohammad & Adil Kak 2007 Introduction to English Phonetics and Phonology. New Delhi: Cambridge University Press/Foundation. Baishya, A. K. 2008 Word formation in Dimasa. International Journal of Dravidian Linguistics XXXVII/1: 125–136. Baker, Ann 2008 Reprint. Tree or Three: An Elementary Pronunciation Course. New Delhi: Cambridge University Press, 1982. Balusu, Rahul 2010 Children’s command of quantification: Some investigations in Dravidian. The EFL Journal 1/2: 61–81. Bandyopadhyay, Anita 2007 Evolution of Bengali grammar. International Journal of Dravidian Linguistics XXXVI/1: 63–78. 2008 Magadhi Prakrit and Bengali. International Journal of Dravidian Linguistics XXXVII/2: 57–85.

India

173

Banerjee, A. K. 2006 Teaching English as a Foreign Language. Jaipur: Pointer Publishers. Bardhan, Susanta Kumar 2010 Vowel harmony and nominal derivational suffixes of Chakma. Indian Linguistics 71/1–4: 33–48. Basu, Anupam & U. N. Singh 2005 SIMPLE 05: Proceedings of the Second Symposium on Indian Morphology, Phonology and Language Engineering. Mysore: Central Institute of Indian Languages. Bendre, S. M., Amitabha Mukherjee & Rajeev Sangal (eds.) 2005 Proceedings of the International Conference on Natural Language Processing (ICON-2005). New Delhi: Allied Publishers. Bendre, S. M., Inderjeet Mani & Rajeev Sangal (eds.) 2007 Proceedings of the International Conference on Natural Language Processing (ICON-2007). Delhi: Macmillan India Ltd. Bharati, Akshar & Amba Kulkarni 2010 Information coding in a language: Some insights from Paninian Grammar. Dhimahi 1/1: 77–91. Bhaskarrao, Peri & Peter Ladefoged 2009 Timing constraints within gestures: A re-examination of Toda sibilants. Indian Lingusitics 70/1–4: 73–78. Bhattacharya, Pushpak, Christiane Fellbaum & Piek Vossen (eds.) 2010 Principles, Construction and Application of Multilingual Wordnets: Proceedings of the 5th Global Wordnet Conference. New Delhi: Narosa Publishing House. Chaudhary, Shreesh 2009 Foreigners and Foreign Languages in India: A Sociolinguistic History. New Delhi: Cambridge University Press/Foundation. Dakshi, Alibha 2008 The role of collocative meaning in semantic relations in context bound Bengali utterances. International Journal of Dravidian Linguistics XXXVII/2: 87–106 Dalai, U. P. 2008 Socio-linguistic study of Oriya personal pronouns. International Journal of Dravidian Linguistics XXXVII/1: 95–106. Dan, Mina 2010 A linguistic account of number names in Bangla, Assamese and Oriya. Indian Linguistics 71/1–4: 57–71 Darbari, Hemant, Rajeev Sangal & Dipti M. Sharma (eds.) 2008 Proceedings of the International Conference on Natural Language Processing (ICON-2008). Delhi: Macmillan Publishers India Ltd.

174

Pingali Sailaja

Das, R. K. 2006 Dash, N. S. 2005

Role of SSP in syllabification of words in Oriya. International Journal of Dravidian Linguistics XXXV/1: 97–110.

Corpus Linguistics and Language Technology. New Delhi: Mittal Publications. Dattamajumdar, Satarupa 2006 Study of Bangla consonant clusters. International Journal of Dravidian Linguistics XXXV/2: 89–105. Dheram, Premakumari (ed.) 2008 Negotiating Empowerment: Studies in English Language Education. Hyderabad: Orient Longman. Dua, Hans 2008 Language Education: The Mind of Society. Mysore: Yashoda Publications. Dutt, Kiranmai P., Geetha Rajeevan & C. L. N. Prakash 2008 A Course in Communication Skills. Hyderabad: Foundation Books. Dutta Baruah, P. N. 2007 A Contrastive Analysis of the Morphological Aspects of Assamese and Oriya. Mysore: Central Institute of Indian Languages. Emeneau, M. B. 2006a Some Dravidian noun compounds. International Journal of Dravidian Linguistics XXXV/2: 1–8. 2006b Toda studies. International Journal of Dravidian Linguistics XXXV/2: 9–38. Gamliel, Ophira 2009 Jewish Malayalam. International Journal of Dravidian Linguistics XXXVIII/1: 147–175. Ganeri, Jonardon 2006 Artha: Meaning. New Delhi: Oxford University Press. Grimal, F, V. Venkataraja Sarma, & S. Lakshminarasimham 2007 Paninian Grammar through its Examples volume II: The Book of Compound Words. Tirupati: Rashtriya Sanskrit Vidyapeetha. Grimal, F, V. Venkataraja Sarma, V. Srivatsankacharya & S. Lakshminarasimham 2006 Paninian Grammar through its Examples, volume I The Book of Examples: 40,000 Entries for a Text. Tirupati: Rashtriya Sanskrit Vidyapeetha. Hasnain, Imtiaz & Shreesh Chaudhary (eds.) 2010 Problematizing Language Studies: Cultural, Theoretical and Applied Perspectives: Essays in Honour of Rama Kant Agnihotri. New Delhi: Aakar Books.

India

175

Huddar, Asmita, Rekha More, Prabha Ghate & Varsha Gathu 2006 Language and Communication. New Delhi: Kanishka Publishers & Rehabilitation Council of India. Jain, A.K, Pravin S.R. Bhatia & A. M. Sheikh 2008 Reprint. Professional Communication Skills. New Delhi: S. Chand & Company, 2001. Joseph, U. V. & Robbins Burling 2006 The Comparative Phonology of the Boro-Garo Languages. Mysore: Central Institute of Indian Languages. Joshi, S. D. & J. A. F. Roodbergen 2006 The Astadhyayi of Panini. Vol XII. New Delhi: Sahitya Akademi. 2007 The Astadhyayi of Panini. Vol XIII. New Delhi: Sahitya Akademi. Kapfo, Kedutso 2005 The Ethnology of the Khezhas and the Kheza Grammar. Mysore: Central Institute of Indian Languages. 2009 A Neurocognitive Approach to Mother Tongue Education. Mysore: Central Institute of Indian Languages. Kapoor, Kapil 2005 Dimensions of Panini Grammar: The Indian Grammatical System. New Delhi: D. K. Printworld. Kartikeyan, B. M. & Brajesh Priyadarshi 2010 Comparison of co-articulatory effects in speech of Tamil-speaking children and adults with Down’s Syndrome: An acoustic explanatory study. Indian Linguistics 71/1–4: 105–112. Kaur, Rajpal 2006 Teaching of English: New Trends and Innovations. New Delhi: Deep & Deep Publications. Kavithaasaran (ed.) 2008 Reprint. A Comparative Grammar of the Dravidian or South Indian Family of Languages by Rt. Rev. Robert Caldwell. Chennai: Kavithaasaran Pathippagam, 1875. Khan, Intakhab Alam 2005 Teaching of English: The Bilingual Context. New Delhi: Academic Excellence. Khuong, Nguyen Chi Duy, Richa & Samar Sinha 2009 The Fifth Asian GLOW: Conference Proceedings. Mysore: Central Institute of Indian Languages, and New Delhi: FOSSSIL. Kostic, Svetislav 2007 Grammatical reduplication in Indic and Slavic. Indian Linguistics 68/1-2: 1–10. Krishnamurti, Bhadriraju 2009 Studies in Telugu Linguistics. Hyderabad: C. P. Brown Academy.

176

Pingali Sailaja

2010

Language typology, regular sound change. International Journal of Dravidian Linguistics XXXIX /2: 17–29. Krishnaswamy, N. & Lalitha Krishnaswamy 2006 The Story of English in India. Delhi: Foundation Books. Levitt, Stephan Hillyer 2010 Fortunatov’s Law and Dravidian. International Journal of Dravidian Linguistics XXXIX/1: 19–90. Mahanta, Shakuntala 2009 Prominence in Oriya, Bangla and Assamese – A phonetic and phonological investigation. International Journal of Dravidian Linguistics XXXVIII/1: 101–127. Mahapatra, Bijay Prasad 2007 A Synchronic Grammar of Oriya (Standard, Spoken and Written). Mysore: Central Institute of Indian Languages. Manoharan, S. & V. Gnanasundaram 2007 Linguistic Identity of an Endangered Tribe: Present Great Andamanese. Mysore: Central Institute of Indian Languages. Masica, Colin P. (ed.) 2007 Old and New Perspectives on South Asian Languages: Grammar and Semantics. Delhi: Motilal Banarsidass. Mishra, Awadesh K. 2009 Phonology of borrowed words in Khasi. International Journal of Dravidian Linguistics XXXVIII/1: 87–99. Mishra, R. K. 2006 Grammar and speech in the brain. International Journal of Dravidian Linguistics XXXV/2: 39–66. Mishra, Sharda 2006 UGC and Higher Education System in India. Jaipur: Book Enclave. Mohanty, Ajit, Minati Panda, Robert Phillipson & Tove Skutnabb-Kangas (eds.) 2009 Multilingual Education for Social Justice: Globalising the Local. New Delhi: Orient Blackswan. Mohanty, P. 2008 Dravidian substratum and Indo-Aryan languages. International Journal of Dravidian Linguistics XXXVII/1: 1–20. Mohanty, Panchanan & Reinhard Kohler (eds.) 2008 Readings in Quantitative Linguistics. Delhi: Indian Institute of Language Studies. Mohanty, Panchanan & Ramesh C. Malik (eds.) 2008 Language, Culture and Society: Studies in Honour of Acharya Bhabananda. New Delhi: Indian Institute of Language Studies. Morey, Stephen & Mark Post (eds.) 2008 North East Indian Linguistics. New Delhi: Cambridge University Press/Foundation.

India

177

Murphy, Raymond 2009 Reprint. English Grammar in Use. New Delhi: Cambridge University Press, 2000. Nagamma Reddy, K. 2009 Fricatives in Telugu. International Journal of Dravidian Linguistics XXXVIII/1: 37–56. Nagaraj, Geetha 2008 Reprint. English Language Teaching: Approaches, Methods, Techniques. Hyderabad: Orient Longman, 1996. Nagaraja, K. S. 2010 Konyak Grammar. Mysore: Central Institute of Indian Languages. Narang, Vaishnav, Deepa Monibarua & Ritu Yadav 2010 Number recognition deficit and arithmetical disorder in cases of stroke. Indian Linguistics 71/1–4: 123–146. Nihalani, Paroo, R. K. Tongue, Priya Hosali & Jonathan Crowther 2005 Reprint. Indian and British English: A Handbook of Usage and Pronunciation. New Delhi: Oxford University Press, 1979. Ober, Scot 2005 Contemporary Business English. New Delhi: Biztantra. Parrott, Martin 2010 Reprint. Grammar for English Language Teachers. New Delhi: Cambridge University Press, 2000. Patel, P. G., Pramod Pandey & Dilip Rajgor (eds.) 2007 The Indic Scripts: Palaeographic and Lingusitic Perspectives. New Delhi: D. K. Printworld. Pilot-Raichoor, Christiane 2008 The Dravidian zero negative: A ‘myth’ or a challenging structural reality? International Journal of Dravidian Linguistics XXXVII/2: 1–32. Prabodhachandran Nayar, V.R. 2010 Plurisegmental features and intonation in Malayalam. International Journal of Dravidian Linguistics XXXIX /2: 185–196. Raja, Lalitha R & N Rajasekharan Nair 2007 Development of plural concord in English among school children. Indian Linguistics 68/1-2: 123–134. Ramakrishna Reddy, B. 2009 Number system in Manda. International Journal of Dravidian Linguistics XXXVIII/1: 19–35. 2010 Endangered tribal languages of Central India International Journal of Dravidian Linguistics XXXIX/1: 105–115. Ramanarasimham, P. 2006 Reprint. An Intensive Course in Telugu. Mysore: Central Institute of Indian Languages, 1985.

178

Pingali Sailaja

Ramanathan, Vaidehi 2007 Reprint. The English-Vernacular Divide: Postcolonial Language Politics and Practice. New Delhi: Orient Longman, 2005. Ramaswami, N. 2007 Ho Grammar. Mysore: Central Institute of Indian Languages. Rassool, Naz 2007 Global Issues in Languages, Education and Development: Perspectives from Postcolonial Countries. Hyderabad: Orient Longman. Rathna Kumar, S. B., Panchanan Mohanty, S. G. R. Prakash & K. Anusha 2010 The effect of auditory masking on phonatory stability in normal speaking young subjects. Indian Linguistics 71/1–4: 153–172. Sahu, Nandini (ed.) 2005 Post modernist delegation to English language teaching: The Quixotic deluge. New Delhi: Authors Press. Sar, Mohan Lal 2007 Yun ‘To Come’: The passive marker in Kashmiri. Indian Linguistics 68/1-2: 27–32. Sarma, Mukul Kumar 2007 Vowel length variation in Assamese conforming to the bimoraic word minimality. Indian Linguistics 68/1-2: 33–48. Sastry, P. V. Parabrhma & N. S. Ramachandra Murthy 2009 Telugu Script: Origin and Evolution. Hyderabad: Andhra Pradesh Government Oriental Manuscripts Library & Research Institute. Satapathy, H. K., S. Satyanarayana Murthy, R. J. Rama Sree & Srinivasa Varakhedi (eds.) 2007 English Grammar Paninian Perspective: Proceedings of National Symposium held from 7th to 9th Nov 2005. Tirupati: Rashtriya Sanskrit Vidyapeetha. Shah, Nila & Amee Sinroja (eds.) 2006 English in India: Issues and Approaches. New Delhi: Creative Books. Sharada, B. A. 2006 M. B. Emeneau: Bibliography with Citation Index. Mysore: Central Institute of Indian Languages. Sharma, Dipti Misra, Rajeev Sangal & Sudeshna Sarkar (eds.) 2010 Proceedings of the International Conference on Natural Language Processing (ICON-2010). Delhi: Macmillan Publishers India Ltd. Sharma, Dipti Misra, Vasudeva Varma & Rajeev Sangal (eds.) 2009 Proceedings of the International Conference on Natural Language Processing (ICON-2009). Delhi: Macmillan Publishers India Ltd. Sharma, P. C. 2005 Effective Methods of Teaching English. Jaipur: Sublime Publications.

India

179

Sharma, Suresh K. 2006a Language in Contemporary India. vol. 1. Delhi: Vista International Publishing House. 2006b Language in Contemporary India. vol. 2. Delhi: Vista International Publishing House. Sharma, Surmangol H. 2006 Negation in Manipuri (Meiteilon). International Journal of Dravidian Linguistics XXXV/2: 155–162. 2007 Elements of Meitei Mayek (Meitei/Manipuri script). Indian Linguistics 68/1-2: 49–58. Shemesh, Ruth & Sheila Waller 2010 Reprint. Teaching English Spelling: A Practical Guide. New Delhi: Cambridge University Press, 2000. Sheorey, Ravi 2006 Learning and Teaching English in India. New Delhi: Sage Publications. Singh, Ch.Yashawanta & H. S. Sharma 2008 Nominalization in Manipuri. International Journal of Dravidian Linguistics XXXVII/1: 107–123. Singh, Soibam Imoba 2010 Pronominal structures in Kom and Aimol. Indian Linguistics 71/1– 4: 173–182. Singh, V. D. 2008 Language Learning, Teaching and Testing: A Companion. New Delhi: Cambridge University Press/Foundation. Sinha, R. M. K. & Anil Thakur 2007 Kyaa in Hindi: Identification and categorization of its multiple uses for Hindi-English MT. Indian Linguistics 68/1-2: 59–70. Sinha, Smita 2005 English Language Teaching: Prospects, Problems & Suggestions. Jaipur: Mangal Deep Publications. Sjoberg, Andrée F. 2009 Dravidian Language and Culture. Kuppam: Dravidian University. Subbarao, K. V. & Harbir K. Arora 2009 The conjunctive participle in Dakhini Hindi-Urdu: Making the best of both worlds. Indian Linguistics 70/1–4: 359–386. Subrahmanyam, Korada 2008 Theories of Language: Oriental and Occidental. New Delhi: D. K. Printworld. Subrahmanyam, P. S. 2008 Dravidian Comparative Grammar—I. Mysore: Central Institute of Indian Languages.

180

Pingali Sailaja

2010

The Evolution of finite verb in Dravidian. International Journal of Dravidian Linguistics XXXIX /2: 31–58. Sureshkumar, E., P. Sreehari, & J. Savithri 2010 English for Success. Bangalore: Foundation Books. Tarai, Shashikant & S. Devaki Reddy 2007 Variation in Indian English: A case study of vowel length in Oriya English. Indian Linguistics 68/1-2: 11–26. Tatacharya, Ramanuja N. S. 2005 Sabdabodhamimamsa, An Enquiry into Indian Theories of Verbal Cognition: Part I—The Sentence and its Significance. New Delhi: Rashtriya Sanskrit Sansthan. Tickoo, Makhan L. 2008 Harold E. Palmer: From Teacher-Learner to Legend. Hyderabad: Orient Longman. Tikkanen, Bertil & Heinrich Hettrich (eds.) 2006 Themes and Tasks in Old and Middle Indo-Aryan Linguistics. New Delhi: Motilal Banarsidass. Trautmann, Thomas R. 2006 Languages and Nations: The Dravidian Proof in Colonial Madras. New Delhi: Yoda Press. Vasanta, D. 2006 Processing of Telugu Compounds. International Journal of Dravidian Linguistics XXXV/2: 107–115 Viswanatham, Kasturi 2009 New Horizons in Language and Linguistics. Mysore: Central Institute of Indian Languages. Wali, Kashi 2006 Marathi: A Study in Comparative South Asian Languages. New Delhi: Indian Institute of Language Studies. Walling, Pangersenla 2010 The Allative subject construction in Ao. Indian Linguistics 71/1–4: 219–227.

Reviews

Mrinal Kaul and Ashok Aklujkar (Eds.) Linguistic traditions of Kashmir, Essays in Memory of Pandit Dinanath Yaksh 2008. Jommu: D.K. Printworld (P) Private Ltd. 609 pp., ISBN 8124604401

Reviewed by Shishir Bhattacharja This book is a collection of twenty one articles on the work and life of a number of scholars of Kashmir starting from the medieval period up to the beginning of the 20th century. The main objective of this volume is to present a compendium of the contribution of Kashmir to the study of Sanskrit language and linguistics as well as other subjects such as philosophy, aesthetics and theology to the extent that they concern language and grammar. As most of the articles in this volume do not, unfortunately, deal with grammar or lexicon directly (had it been the other way round, its title ‘Linguistic traditions in Kashmir’ would have been more justified!), I shall comment only on those they do. In the first article Mukulabhatta and Vyanjana (28–40) M. M. Agrawal describes the view of Mukulabhatta, a Kashmiri grammarian, regarding laksana (metaphoric use) of words and the counter criticism of his views by Mammata, another grammarian from Kashmir. The second, the third and the fourth are three long articles by Ashok Aklujkar. In Patanjali’s Mahabhasya as a key to Happy Kashmir (41–87) Aklujkar tries to show with citations from different grammarians including Kalhana that (i) Mahabhasya was studied in Kashmir and (ii) Patanjali was a Kashmirian by birth. In Gonardiya, Gonika-putra, Patanjali and Gonandiya (88–172) he claims that Gonandiya and Patanjali are two different names of the same person whereas Gonardiya and Gonika-putra may not be the names of Patanjali. In Patanjali: A Kashmirian (173–205) Aklujkar gives arguments in support of his claim that Kashmir was the homeland of Patanjali. Bettina Baümer, the author of the fifth article, The three grammatical persons and Trika (206–222) shows, following Abhinavagupta, how the three persons – I, you and he/it – instantiate the god Shiva, the goddess Shakti and Nara, the human beings respectively. In the sixth article, The treatment of the Present tense in the Kasmirasabdamrta of Isvara Kaul: A Paninian grammar of Kashmiri (223–270) Estella Del Bon and Vincenzo Vergiani describe the contents of Isvara Kaul’s grammar Kasmirasabdam-

184

Shishir Bhattacharja

rta. The seventh and the eighth articles are written by Johannes Bronkhorst. In A note on Kashmir and Orthodox Paninian Grammar (271–280), Bronkhorst tries to show how Kashmir had played a key role in the preservation of the grammatical tradition associated with Patanjali’s Mahabhasya. In Udbhata, a grammarian and a Carvaka (281–299), he points to the existence of different non-orthodox Paninian grammatical traditions, which, since at least the time of Bhartrhari, (i) used Panini’s Astadhyayi as their main reference but (ii) gave new interpretations to a number of its rules, and (iii) did not recognize Patanjali, one of the trimuni (three sages or three great grammarians of Ancient India) (Panini and Katyayana being the two others), as an authority. In Theoretical precedents of the Katantra (300–367), the 9th article, George Cardonna explains why Katantra is not just a brief textbook but a different way of looking at Sanskrit grammar. Katantra is believed to be written by Sarvavarman to teach Sanskrit grammar to the king Satavahana within a very short period of six months. According to Kathasaritsagara, a collection of stories, Satavahna wanted to excel in grammar as soon as possible because one of his queens, during a common bath, laughed at him for his poor knowledge of Sandhi ‫ ޤ‬he confused modakai (ma-udak-ai) paritaraya (don’t-water-with-throw) ‘don’t throw water at me’ for modak-ai paritaraya (sweet-balls-with-throw) ‘throw sweet-balls at me’, had some sweet-balls brought by some servant and thrown them towards the queen. There was a debate among Indian grammarians regarding i) whether one needs rules to derive words with taddhita affixes, or ii) the words that are supposed to be outputs of those rules are in fact listed in the lexicon. For example, one can derive a fruit name like amalaka from its tree name amalaki, or consider both of the words as distinct lexical entries. According to Cardonna many grammarians including the composer of Katantra believed that both amalaka and amalaki (and many other similar words) are listed in the lexicon, and hence, there is no need to include rules to derive fruit names from tree names or vice-versa in grammar (morphology). In Ksirakhyata Catuspadi, notes on Ksirasvamin's comments on the four basic grammatical categories (368–376), the 10th article by M.G. Dhadphale, the author talks about different works by Ksirasvamin on Sanskrit nouns, verbs, upasargas (prefixes) and nipatas. In the 11th article Three Kashmirian texts on Sanskrit syntax: Kudaka’s Samanvayadis, Devasarmana’s Samanvayapradipa and Samanvayapradipa sanketa (377–398), Oliver Hahn talks about three different texts on Sanskrit syntax which deal with the whole range of possible syntactic relations between words and

Reviews

185

gives definition of a sentence and its parts. These manuscripts which now belong to the National Library of Austria in Vienna had been collected by the archeological explorer Marcus Aurel Stein in 1894. In the 12th article Jayanta’s Interpretation of Panini 1.4.42 (399–409), V.N. Jha talks about the grammarian Jayantabhatta’s theory of Pramana, his opponents’ objections against this theory and Jayantabhatta’s reply to those objections. S.D. Joshi explains in the 13th article On Nagesbhatta’s misunderstanding of Kaiyata, the Kashmirian commentator of Patanjali’s Mahabhasya (410–418) how Nageshbhatta (1670–1750) misunderstood some rules of Pradipa a commentary on Panini’s grammar by the 11th century grammarian Kaiyata. In the 14th article The Sarada manuscripts of the Kasikavritti – Part II (419–428) Malhar Kulkarni compares different manuscripts of Kasikavritti, one of the major works in the tradition of Panini’s grammar, in order to identify different stages of development of this text. In Uvata, the Kashmirian Pratisakhya commentator (429–445), the 15th article, Nirmala Kulkarni discusses mainly the contribution of Uvata, a grammarian-cum-phonetician, to Indian grammatical tradition. Kulkarni also revisits the ‘Uvata controversy’ – different issues concerning his personal history, his time, birthplace and genealogy. In The mythico-ritual syntax of omnipotence, on Utpaladeva and Abhinavagupta’s use of kriyakaraka theory to explain Siva’s action (446–488), the 16th article, David Peter Lawrence examines the interpretation of Sanskrit syntax from the point of view of Shiva-Shakti mythology by two Kashmiri philosopher, Utpaladeva and Abhinavagupta who lived in the 10th and 11th century respectively. In his short note (the 17th article), Some Peculiar vocables in the Paippalada-samhita (489–494) Hukam Chand Patyal brings to light some rare words that appear in the Paippalada-samhita of the Atharva veda. In the 18th article The first among the learned: Kashmiri poeticians on grammarians (495–507) C. Rajendran describes how poeticians used to regard grammarians in Kashmir. According to Rajendran there were two different periods in the history of Kashmiri poetics: (i) The Pre-dhani period and (ii) the Post-dhani one. For the poeticians of the Pre-dhani period (Bhamaha, Vamana and Anandavardhana among others) the grammarians are prathame hi vidvamso vaiyakaranah ‘the first among the learned’. In this period the knowledge of grammar was considered a prerequisite for writing poetry because it was thought that poets should know how to use (correct) words that are allowed by the grammar and only those.

186

Shishir Bhattacharja

Mahimabhatta, a poetician-cum-grammarian of the Post-dhani period, does not agree with the abovementioned criteria of absolute grammaticality of poetic works. Following Panini’s famous dictum sarve sarvarthavacakah (all-all sense-denote) Mahimabhatta states that all words can theoretically denote all senses (this reminds me Frege (1848–1925) who said that no word is used in the same sense twice in the same text). He averred that terms like shabda ‘word’, apashabda ‘slang’ and asadhushabda ‘improper word’(?) used by grammarians are only relative concepts and not absolute ones. Mahimabhatta draws our attention to the indispensability of pragmatics when he says in his book Vyaktiviveka that contextual and exralinguistic features must be taken into account for some act of communication to take place. Hence, a proper word may be unable to convey the intended meaning if it lacks contextual/pragmatic reinforcements while even a wrong word can denote the intended sense if it is accompanied with contextual and other accessories. Mahimabhatta counters the arguments of Patanjali regarding the use of correct word with the explanations given by Patanjali himself ‫ ޤ‬the demerit produced by the use of a grammatically incorrect ‘bits and pieces’ here and there is counterbalanced by the ‘good final results’ of the whole poem, or as Rajendran states (P. 504), “by the profuse merit generated by the hearing, retaining, understanding and practicing of the ideas contained in the scientific discourse.” Notwithstanding that Mahimabhatta is reverential to trimuni and gives reference from their work in support of his own views, he makes an interesting distinction between Panini and his blind followers whom he calls sarcastically Khandikopadhyayas (piece-expert) ‘experts of bits and pieces’. Mahimabhatta also refers to Bhartrhari abundantly but, unlike Abhinavagupta among others, he is skeptical about Bhartrhari’s key concept of Sphota. In the 19th article From an adversary to the main ally, the place of Bhartrhari in the Kashmirian Saiva Advaita (508–524), the author Raffaele Torella describes how some adversaries of Bhartrhari, namely Somananda and Utpaladeva who belong to the Pratyabhijna school of Kashmir later became his followers to a certain extent and used some of Bhartrhari’s doctrines to construct their own views about the world and the language. In the 20th article Helaraja’s defence of the Padavadhika method of grammatical explanation (525–549), Vincenzo Vergiani presents a particular view of Sanskrit grammar by Helaraja, a grammarian and a commentator of Bhartrhari, who lived in Kashmir in the 10th century.

Reviews

187

In the 21st article The impact of Candra Vyakarana on the Kasika (550– 561), the author P. Visalakshy explains the extent to which the author of Kasika(vritti), an explanatory text on Panini’s grammar as well its commentaries (varttikas) by Katyayan and Patanjali, is influenced by Candra byakarana, one of the eight ancient schools of grammar (Indra, Kasakrtsna, Apisali, Sakatayana, Panini, Amara, and Jainendra being the seven others). Kasika is believed to have been composed in the 5th century by Jayaditya (identified as Jayapida by Belvelkar), a king of Kashmir and his minister Vamana. Some of the authors cited above describe the theoretical stand of particular scholars whereas some others try to identify those scholars both historically and geographically. Although there are scholars such as Patanjali whose direct relation with Kashmir is less than certain, most of the scholars cited in this volume do belong to Kashmir. This collection shows that the grammatical tradition in Kashmir has remained mainly Paninian, notwithstanding that other grammatical schools namely Katantra and Candra (which were not necessarily composed in Kashmir) were very popular in this area (as in other parts of South Asia, specially Bengal) until very recently. Although these grammars were used as pedagogical manuals and/or abridged versions of Astaddhayi, they were rich in insights that are different from those of Panini’s. It is said that Katantra and Candra were used by the general populace who tried to access Sanskrit language whereas Astaddhayi was reserved exclusively for higher studies in the Sanskrit language and grammar. This volume has a foreword by Kapila Vatsyayan (vii–ix) and a preface by Ashok Aklujkar (xi–xiv), one of the two editors. Mrinal Kaul, the other editor, gives a life sketch of Pandit Dinanath Yaksh (1921–2004) who has been a doyen in the field of Sanskrit grammatical exegesis in Kashmir and to whose memory the present volume is dedicated (xxvi–xxxiii) and a detailed introduction (1–27). The twenty one articles mentioned above are followed by five appendices (565–584) in which Mrinal Kaul presents a list of the manuscripts found in the area of Kashmir, a note on each one of the contributors (585–592) and a general index (593–609). To conclude, this well-edited book of a considerable size (about 650 pages) helps the reader guess the extent to which Kashmir was an important site for linguistic research in ancient and medieval India.

S. Imtiaz Hasnain & Shreesh Chaudhary (Eds.) Problematizing Language Studies. Cultural, Theoretical and Applied Perspectives. Essays in Honor of Rama Kant Agnihotri 2010, Aakar Books, Delhi. 601 pages, ISBN 978-93-5002-084-5

Reviewed by Ghanshyam Sharma In the face of stiff competition from highly specific publications, bringing out a Festschrift to honor an eminent scholar is increasingly becoming a most challenging task. Scientific publications in almost all branches of human knowledge have become so technical and specialized (thus sectorized) that any non-specialized publication risks being easily ignored or ridiculed. Yet in spite of being fully aware of all the risks, S. Imtiaz Hasnain & Shreesh Chaudhary (Editors, hereafter) have admirably undertaken the task to honor the highly esteemed Indian linguist Professor Rama Kant Agnihotri for his services to applied linguistics in general and sociolinguistics in particular. This 601-page Festschrift is entitled Problematizing Language Studies and is divided into three broadly-defined sections, namely Cultural Perspective, Theoretical Perspective and Applied Perspective. Beginning with a scholarly Foreword by Professor Rajesh Sachdeva, it is a collection of 36 contributions from 38 scholars spanning not only all the major branches of theoretical and applied linguistics but also covering Indian History, Urdu literature, and Indian culture. A caveat regarding the organization of our following discussion thus is in order: given the widely diverse nature of the papers, I prefer to organize my discussion by author rather than attempt to rearrange the articles by topic, assuming that the Editors must have had their reasons for arranging the volume into broad categories rather than well-defined topics. I also feel it necessary to specify that, again because of the vastness of the volume’s content, I do not possess a specialist’s knowledge of all of the topics and thus in my discussion of several articles I can do no more than rely on my general knowledge alone. I therefore offer my sincere apologies to any authors who find my comments on their work inappropriate.

Reviews

189

Section A: Cultural Perspective Full of detailed and highly informative notes yet lacking in accurate references (it is not at all clear, for example, which book by Bangha the author refers to in note 43 on page 32, or what the name of the publisher is in note 29 on page 31, to mention just a few inaccuracies), Sanskritic vs Syncretic: Sir William Jones and Indian Pluralism and Plurality by Michael J. Franklin (13–34) cleverly opens up the often ignored question of the contribution made by the Orientalists working in the East India Company in the eighteenth century. The Orientalists, Franklin reminds us, are unfairly accused of having a presumed bias towards the Hindu past of India. The author makes a concerted effort to stress the need to reconsider works by the Orientalists like Sir William Jones. He calls our attention to their firm commitment not only to discover works relating to the Hindu past of India but also to that devoted to the prophets of the so-called ‘syncretic’ culture and communal harmony in India, including Amir Khusro, Akbar the Great, Dara Shikoh and many others. Needless to say, no other contribution devoted to cultural syncreticism could have done better than this one does. In spite of some superfluous notes (e.g. note 1) and inconsistencies in references (“see McGregor 2003, 1984, 1968”), the essay entitled Braj Reinvented: Some Colonial Approaches to Hindi Dialects by Arthur Dudney (35–47) concisely describes the divergence between the colonial approach to different terms viz. Hindi-Hindustani-Urdu-Braj and the indigenous speakers’ perception of these languages or dialects. The paper thus very insightfully describes “how colonial research into vernacular languages turned Braj Bhasa into a synecdoche for all non-Islamicate vernaculars in northern India.” (p. 42). In The Language of Love: A Study of the Amorous and Erotic Associations of Urdu, Tariq Rahman (48–75) guides us through the erotic and amorous garden of Urdu literature, providing us with a detailed, accurate and vivid description. While describing the historical development of Urdu literature, Rahman rightly observes that its amorous and erotic associations were “suppressed in the wake of the political urgency and identity politics of the Pakistan movement and then the perpetual struggle the Pakistani ruling elite waged against the ethnic elites of Pakistan on the one hand and the Western and Indian ‘Other’ on the other.” (p. 71). However, to understand the development of the Urdu literary tradition of amorous and erotic poetry described by Rahman, perhaps it would also be have been useful to compare it with two other traditions of the amorous and the erotic in the

190

Ghanshyam Sharma

Indian literature: the amorous in classical Sanskrit literature and the amorous-erotic in so-called riti poetry in Hindi where one can easily notice poet’s above mentioned proclivity to the themes of romantic love, separation, fidelity on the part of lover, and indifference and fickleness on the part of the beloved. The theme of boy love, however, is unique to Urdu literature and can exclusively be put down to the influence of Persian and Turkish Muslim cultures. Analyzing statistical data on language change and shift in Pahari, Mandiali and Garhwali speech communities – the three prominent speech communities in the Himalayan region of India – Himalayan Languages, Hindi and English: Contentment or Containment? (76–82) by Mahendra K. Varma poses some real sociolinguistic questions regarding the position of the minority languages of the Himalayan region vis-à-vis major languages such as Hindi and English: should the language shift – i.e. from native tongue to Hindi and English – in these speech communities be considered a case of speakers’ contentment or language containment? Varma leaves it to further sociolinguistic and/or ethnographic research to find out what language policy to adopt, but invites the State to take some action lest these languages should disappear for ever. Although I accept the author’s careful analysis of the phenomenon, according to which “The smaller languages, with no literacy tradition and no state support, are beginning to appear more as symbols of past heritage encapsulated in folk-lore” (p. 80), and share his concerns as a linguist, I do not think, however, that any State should intervene to revert the process. Furthermore, would not such an intervention be counter to the natural sociolinguistic phenomenon of language-shift? There are countless cases of language shift and containment the world over, certainly not limited to the three speech communities in question. Well supplied with important and detailed notes (56 in total) and an almost complete list of references, and extensive citations, English in India and the Role of the Elite in the National Project by Annie Montaut (83– 116) provides an accurate in-depth analysis of the role of the Indian Elite and their weapon-language –English – in the making of modern India. Montaut correctly observes that although it is spoken by a minority of eight to eleven percent of all Indians (I would put this figure even lower since not all those who claim to speak it can be said to do so) as a second language, it is an official language of India and is accredited with much more power than it should be. English consequently has created what she calls a ‘schism’ within Indian society. Montaut rightly quotes Krishna Baldev

Reviews

191

Vaid in saying that English has created an ‘enchanted circle’, imposing a mode of communication on Indians which keeps the majority of them out. Needless to say, this process will inevitably produce what she calls some ‘dramatic consequences’ not only for the vernacular languages of India but also for the nation as a whole since “without acknowledging (and first knowing) what is going on intellectually and culturally outside the enchanted circle, there will be no scope for a true dialogue (in equal terms) with the West. The prerequisite for the international dialogue is of course the inner dialogue, which means, in the matter, gapping the schism created by English”. (p. 104). As far as the reciprocal alienation of Hindi and Urdu is concerned, I fully agree with Montaut’s endorsement of Agnihotri’s analysis that due to standardization of Hindi and politicization of these two varieties “a new generation had grown up: unfamiliar not only with Ghalib and Faiz but also with Kabir and Premchand; nor could they understand Prasad or Nirala (...); the staple diet was Bombay film Hindi. The damage that inevitably accompanies the loss of literary sensitivity in a community is there for everyone to see.” In his well-written article entitled Has Globalization Given a New Legitimacy to Diversity? (117–133), Paroo Nihalani investigates the question of pedagogical standards in the teaching of English at a time of globalization throughout the world. Following his first-hand experience in the teaching of English as a second language, Nihalani endorses the observations made by other researchers in the field of world Englishes, namely: “In recorded history there has never been a language to match the present global spread and use of English.... There has never before been a language that has been spoken by more people as a second language than a first”. (p. 118). Nihlani rightly argues that it is not necessary to stick to teaching ‘received pronunciation’ since, as he puts it, “the so-called received pronunciation (RP) was indeed adopted fifty years ago by the BBC for use by its newsreaders and remains still valid as the foundation of a model for imitation abroad. R.P., to my mind, is a dead horse in the 21st century, and I hate to beat the dead horse.” (p. 126). However, in his lengthy conclusion, he recommends “the ‘top-down’ approach in which emphasis is shifted from the teaching of segments to the teaching of supra-segmental features” (p. 129) and suggests that it may be rewarding for the learner to be exposed to many varieties of pronunciation, including native and, more importantly, non-native as well, in order to enrich his repertoire. Although Nihlani’s recommendation that “we must abandon the teaching of Phonetics of English per se”, is acceptable to certain extant, it is difficult to agree with him

192

Ghanshyam Sharma

when he says that “a sound grounding in General Phonetics will create more space and help learners with broadening their perceptual skills, which I may call ‘multi-dimensional view of intelligibility.” (p. 131). I believe that the teaching of General Phonetics without providing any direct reference to some sort of standard pronunciation of English – be it RP or any other accepted ‘standard’ pronunciation – will never ever achieve the desired result. Like it or not, the teaching of the distinctive features of the English sound system will have to remain there in the field of TESL. Hindi, Urdu, Hindustani: Evolution of the Language of a People (134– 159) by Shreesh Chaudhary is a well-documented and informative research paper. After analyzing carefully the abundant historical facts and sociolinguistic data on Hindi, Urdu and Hindustani, Chaudhary amazingly comes to the following ‘emotional’ conclusion: “Here [in the case of Hindi-Urdu, interpretation is mine] it is one and the same language made two by the choice of script. Different scripts are keeping this language from occupying its rightful place, with one script it may become the mass language of unprecedented reach and can truly be called ‘the language of India’. It may truly become the most preferred second language for most Indians if it adopts a script of wider and greater reach. That will obviously make Roman script the first choice, followed by Devanagari and Arabic in that order.” (p. 154). Needless to say, S. K. Chatterji (1960) too made a similar proposal and advocated the use of Roman script for Hindi so that Hindi could be read by all Indians easily. However, given the present state of affairs, I do not think Chaudhary’s dream can ever be realized unless India compels its citizens to accept the Roman script. Ayesha Kidwai in her research paper entitled The Case for Hindustani Revisited (160–168) rightly questions the Hindi-Urdu-Hindustani continuum and provides some data in support of her claim that Hindi and Urdu are two separate sociolinguistic identities – two different registers used by their speakers for different purposes. Unlike Chaudhary, Kidwai believes though that “... even as there can be no doubt of the validity and the necessity of the task of cultural recovery, to place its entire burden on the reconstruction of a quasi-mythical common core is mistaken. (p. 166). Similar to M. K. Varma’s contribution, The Politics of Multiculturalism by Tanmoy Bhattacharya and Haobam Basantarani (169–184) discusses the case of minor languages spoken in the North-Eastern state of Tripura. The authors’ in-depth knowledge of political theories in general and the political situation in the state in particular, together with a careful analysis of the data on the minor languages, depict a very pessimistic and unfortunate pic-

Reviews

193

ture of the linguistic situation in the state of Tripura. Nevertheless, I believe that their plea for State intervention to support minority rights in a democratic system may leave these minorities in an even more dramatic political situation. In his paper entitled Contextuality, Critical Discourse Analysis and Structuration Theory: Sociological Basis for Analyzing ‘Heavenly Ornaments’ (Bihishti Zewar) (185–200) S. Imtiaz Hasnain makes a strong case for analyzing a text in its social context. The author carefully examines Bihishti Zewar (i.e. Heavenly Ornaments) – a text written in the early nineties by Maulana Ashraf ‘Al-Thanawi – who sought to improve the quality of Islamic education, increase personal piety, and to spread the observance of Islamic law more widely among Muslims in India and feels the need to do that in the light of the usefulness of Systemic Functional Linguistics (SFL), as developed by Halliday, and Critical Discourse Analysis (CDA) since both provide, as he puts it, ‘contextualization of analysis’. After analyzing the above-mentioned text, Hasnain comes up with the following suggestion for both SFL and CDA: “...CDA and SFL must not only incorporate the notion of chains of practices but should also explore the elements of intercontextuality for looking at the way every social practice interrelates with other social practices.” (p. 198). In his short essay entitled On “Puri Boli” (201–211), B.N. Patnaik considers the meaning of some taboo words in a number of expressions of a dialect spoken in the city of Puri in Orissa. These expressions, so the author informs us, can deliver different meanings depending on the context. He discusses various expressions used during the annual Chariot procession when taboo words carry different connotations. Without providing any linguistic arguments to justify an urgent need for an analysis of Puri Boli texts, he nonetheless seems to persuade researchers to undertake an analysis of the dialect. Well equipped with a long list of the author’s unpublished materials, the article entitled On Unearthing a Civilization by Ranjit Singh Rangila (212– 233) discusses sophistication and efficiency in the discourse of the Jenu Kuruba tribe which inhabits the jungles of the Karnataka-Tamil Nadu border. The Editors inform us that according to Rangila’s investigation “Once a verb is realized as an act, then all the extensions of the verb in this language start showing up as extensions of the act. Jenu Kuruba people can match act for act with sophisticated dramatic performances like Yaksha Gana.” (p. 5).

194

Ghanshyam Sharma

In spite of incomplete references (one cannot know the identity of the persons when she refers to Neil McCarthy and Steven Raymond on page 255, for example), the article entitled From Margin to Mainstream the Growing Domain of Language Switching and Mixing in Contemporary South African Television Drama by Kay McCormick (234–259) well describes a situation of code switching and mixing between Afrikaaner, Zulu and Xoso in the South African television drama. The author tells us how the popular television programs in South Africa try to play with tribal loyalty and national loyalty using the tools of code-mixing and code switching. After analyzing different cases of code-mixing and code-switching in present-day South Africa, the author comes to the following conclusion: “Taking together all that has emerged from documents, interviews and observations, I would say that, although it clearly occurs for a variety of reasons – some idealistic and some mundane – the flexible weaving of languages in South African tv drama is an index of a new ‘imagined community’.” (p. 256). Full of interesting excerpts from Indian English newspapers, Code Switching and the Politics of Nostalgia by Rakesh M. Bhatt (260–275) deals with an increasingly popular aspect of code-mixing (not codeswitching as the title inadvertently suggests!) in English and comes to the following conclusion “...the use of Hindi in English is arguably strategic: it activates historical-cultural memory in contemporary modern linguistic practices animating a politics of nostalgia, closely intertwined with Hindilanguage/ Hindu-religion as its cultural expression.” (p. 274). While one can easily agree with Bhatt’s analysis of the evident influence of Hindu fanaticism attested in the terms borrowed from the dominant Hindu culture and the Hindi language, it seems quite difficult to sustain his claim that the tendency for code-mixing can exclusively be put down to Hindu fanaticism. There are, I believe, many additional reasons as well (e.g. inappropriateness of English idiomatic expressions with local culture and an urgency to express subtle nuances, hard to express otherwise) which equally contribute to the trend of code-mixing in Indian English newspapers. Section B: Theoretical Perspective Confessions of A Minor Morphologist: Some Remarks on ‘Morphological Complexity’ by Rajendra Singh (279–285) succinctly raises several key points in the theory of morphology. The foremost question in morphology

Reviews

195

is whether the notion of ‘morphological complexity’ has any important role to play. Following the new line of research developed under the title Whole Word Morphology (cf. Singh 2006), Singh answers it categorically saying that: “What is needed is NOT ‘morphological complexity’, inheritance, right-headedness, or percolation but an account of the interplay between variables and constants, which play their game of word formation-strategies in which they figure.” (p. 284). He proposes that the notion of ‘length’ (or ‘weight’) is sufficient to account for what is normally called morphological complexity. Not that Singh’s task is an easy one, particularly when we see all sorts of so-called sophisticated morphological theories around which aim to provide a consistent one-to-one mapping between all types of forms and functions (and consequently remain trapped in their own technical jargon!), he puts forward a radically different theory which intuitively seems to fare much better than other theories which intend to deal with the question of ‘morphological complexity’. I believe that his WWM theory, which is supported by his own research on Hindi morphology (Singh and Agnihotri 1987, Singh and Neuvel 2003, among others) and draws on ideas developed by Patanjali and Bhatrihari, could have a much bigger role to play in morphology than might be appreciated at first blush. I firmly believe that further research in cognitive and brain sciences could provide tangible scientific proof to determine whether the complexities of many so-called sophisticated morphological theories have any cognitive basis or are just the product of linguists’ analyses. In his short research note (comprising 19 concise paragraphs yet claiming to be based on “detailed statistical data”) entitled The Hindi Long Vowel Problem: A Substantive Approach, (286–291) Probal Dasgupta critically examines the case of the Hindi long vowel in morphological alternation. He does so taking into consideration Singh & Agnihotri’s (1997) proposal to consider the long vowel problem from the morphological point of view. Without going into the details of Dasgupta’s proposal, however, I can certainly raise some doubts about his claim (paragraph 19, page 290), namely, “My informants have absolutely never heard anybody say /dvitiiyaa, tritiiyaa/ with a long /ii/.” (p. 290). I believe that he may have to withdraw such a strong claim. It has long been established that lexical stresses and accents are not as functional in Indian languages as they are in European languages such as English, French, Italian and others. In her very well-documented yet poorly formatted (see (13) for example on page 304) research paper entitled Indian Languages as Intonational ‘Phrase Languages’, Caroline Féry (292–316)

196

Ghanshyam Sharma

convincingly argues that a new intonational category should be introduced which could account for phrase level phonetic properties of languages such as Hindi, Bengali, Tamil and Malayalam which, Fréy authoritatively claims, “show common intonational properties.... belong to a group of languages called phrase languages, which have no lexical stress and also no pitch accent.” (p. 313). The paper gives me abundant fresh food for thought since it clearly establishes that the above mentioned languages: “... organize the tonal pattern of their declarative sentences on the basis of phrase tones, which are anchored at the level of the prosodic phrase. Lexical items do not project any tones. Very few pragmatic meanings are conveyed by changes of tones, though manipulation of pitch range may be a more common device, requiring in-depth studies.” (p. 314). Exploring the line of research followed by Singh & Agnihotri (1997), Gisbert Fanselow in his well-written paper entitled More Freedom on the Left (317–331) proposes a reconsideration of the formation of nominal compounds in German which, in his view, is a productive morphological process. However there are, he warns, some syntactic constraints which can be accounted for if dealt with at the syntactic rather than morphological level. After a very careful analysis, Fanselow comes to the following conclusion: “The strong lexical restrictions we observe here do not disappear in non-head context. This suggests that these constraints are due to lexiconinternal regularities. We have speculated that AN compounds do not arise from the standard merge operation, but rather due to word formation strategies in the sense of Agnihotri and Singh. If correct, this implies that recursive structures can indeed arise by two different mechanisms in natural language.” (p. 329). A list of abbreviations (e.g. AN, PN, QN, etc.) would have been helpful for the reader to follow the discussion without much difficulty. In his paper entitled Morphosyntax of Hindi Infinitives, Rajesh Kumar (332–346) proposes a two-tier treatment of Hindi infinitives. A Hindi infinitival phrase, Kumar observes, has both VP an NP structure inside it and thus has to be treated differently. According to the author, the infinitives have a NP structure externally, but a VP structure internally. However, I find it difficult to identify what theoretical and empirical questions the ‘exocentric structure’ of infinite phrase he is referring to in his conclusion (p. 345). A careful proofreading and style-check would have saved the paper from many typos and wrong alignments (e.g., see example 35) that are present throughout.

Reviews

197

Compositionality in Complex Predicates (347–371) by Achla M. Raina deals with complex predicates in Hindi/Urdu and Kashmiri. In her careful syntactic analysis, she very ably takes into consideration different aspects and forms of complexities attested in verbal predicates of the languages in question. In spite of considering some marginal data ((27a) and (32a), for example) and not checking the draft correctly (31a is not followed by any 31b, just to mention only a few), Raina thoroughly analyses the phenomenon and proposes a compositional account of complex predicates. She discusses the constraints on complex predication and offers a semantic typebased explanation for them. Despite poor proof-reading (Lakoff is repeated in References, for example) Limits to Compositionality of Meaning: Some Notes with Reference to English and Hindi (372–386) by Rajneesh Arora is a fine attempt to show the limits of the theories which aim to provide a compositional account of meaning. Arora seems to be a fervent proponent of D. A. Cruse’s account (1986, 2000) of the phenomenon in question. He analyzes data from Hindi to prove that it is not enough to look merely at the constituent parts to arrive at the real meaning of an utterance as a whole. Context, speech acts, conversational implicatures and metaphors play the most important role in establishing the ultimate meaning of an utterance. On the Nature of Reason in the Present-day Academic Research (387– 397) by Pramod Pandey is a short yet highly thought-provoking article. Its overly ambitious plan to provide an account of the entirety of present-day ‘academic research’ (I fail to understand what precisely he means by this) by undertaking cases simply from linguistics, and exclusively from Chomsky’s famous book Knowledge of Language: Its Nature, Origin, and Use (1986) – which is missing from the References – is not a good starting point, I fear. Dichotomizing all scientific endeavors along an ‘A versus -A’ line is nothing but mystifying the current state of linguistic research. It is customary among scholars in India to ridicule any theory by simply referring to some special kind of ‘reasoning’ from the ‘East’. Pandey says “Considering the negligible role of developing countries form the East, such as India, where knowledge has not been based on such a mode of reasoning, it appears plausible to characterize -AR as Western, albeit purely in the socio-historical sense of having to do with the modern times.” (p. 394). He does not consider it necessary though to explain to the reader which ‘mode of reasoning’ in India he is talking about. I assume that he has the classical Indian philosophical mode of reasoning in mind, but to my knowledge there has never been only one single way of thinking.

198

Ghanshyam Sharma

The essay entitled Towards Branding Indian Linguistics, A Linguistic Destination Next by Shailendra Kumar Singh (398–418) contains the author’s reflections on the current state of linguistic research in India. After considering various aspects of linguistic research, the author concludes by saying that “... neither the linguistic canon of the West has to be domesticated nor foreignized, but interventionist options have to be reinvented to energize the capacity of recognizing the [...betweens], which is occupying the space of [...third].” (p. 418). Singh ends his reflections by mystifying them even further: “Whether we welcome or denounce it, it is undeniable that linguistic temptation interrogates drastically many time honored linguistic myths and, in return, may therefore also offer us alternative ways of negotiating the place of Indian Linguistics in the transitory and often distorting competency of the West.” (p. 418). As the title itself suggests, Rough Notes by Udaya Narayana Singh (419–429) roughly describes some empirical facts and expresses some worries about ‘monolingualism’ and ‘linguistic purity’. Yet in spite of its inclusion in the theoretical section of the present volume, I have failed to find any theoretical points made. ‘How Do I Know that You are not a CBI Agent?’ Examining the Identity of Researcher in Sociolinguistic Fieldwork by Rizwan Ahmad (430–439) describes the author’s own field work experience in the city of Delhi. The author informs us that the notion of ‘insider’ and ‘outsider’ in sociolinguistic research is complex in that, as he puts it, “Different social, political, and other contextual factors go into the construction of an identity of the researcher.” (p. 437). As a conclusion to his account of sociolinguistic fieldwork in Delhi, Ahmad has the following very obvious advice to give to any scholar interested in field work: “... the best way to win the trust and confidence of the community you want to study is through establishing contact with key personalities in the community”. (p. 437). Section C: Applied Perspective Using the Mother Tongues for Learning and Teaching Other Tongues: Theory and Practice by A. L. Khanna (443–456) discusses different aspects of the role of the mother tongue (L1) in learning a second or foreign language (L2). Some second language teaching theorists advance the idea that L1 should be completely banned from L2 teaching. Khanna presents a good summary of the discussion on swings and roundabouts of using the L1

Reviews

199

while teaching the L2. After an evaluation of different language teaching and learning theories, Khanna comes to support the idea that the mother tongue should be perceived as a rich resource in the initial stages of learning the L2. “A judicious use of the mother tongue both by the teacher and learners instead of causing an obstacle may help learners to achieve their goals faster in the target language and also make them cognitively richer and more mature.” (p. 452). Khanna contends that “Simply increasing exposure to the target language may not yield the desired result particularly with the lower-level student. The may benefit more by explanation in L1 than immersion in the TL.” (p. 453). On the Teaching of Indian Language and Literatures Abroad by Omkar N. Koul (457–465) describes the current situation of the teaching of Indian languages and literatures abroad, focusing mainly on the teaching of Hindi and Urdu abroad. He provides some guidelines regarding “framing clearcut objectives, devising of learner-centered need-based courses, preparation and selection of instructional materials including print and audio-visual materials, developing of online courses, use of different suitable language teaching methods, administering of language tests in measuring the achievement levels.” (p. 465). In her article entitled Teaching English to Young Learners: How Far Have We Come? How Far Can We Go? (466–478), Rama Mathew deals with the question of the teaching of English to adult learners. She suggests that both languages (i.e. L1 and English) can go hand in hand because “The additive bilinguals enrichment principle and the interdependence or common underlying proficiency principle (Cummins 1981) suggest that when students add a second language to their intellectual toolkit while continuing to develop conceptually and academically in their first language, there is substantial linguistic, cognitive, or academic growth.” (p. 475). S. R. Prahlad in his article entitled Restructuring General English Language Curricula at the Tertiary Level: Issues and Possibilities (479–504) expresses anxiety over not taking into account learners’ wants and needs in the preparation of language teaching materials. After discussing at length different cases of language teaching materials and demonstrating the need for a renewal of the English curriculum, he establishes some principles for designing sourcebooks. In his paper entitled English at the Primary Level: A Study of Teaching Vocabulary Panchanan Mohanty (505–513) undertakes the case of an English Reader published by the Andhra Pradesh State Government to be introduced at class IV all over the state. To judge the suitability of the book

200

Ghanshyam Sharma

in question, Mohanty uses three criteria, namely language and cultural distance, pronounceability and use of basic vocabulary, and finds it unsatisfactory. Unfortunately, the paper itself has some linguistic flaws (the first sentence in the Conclusion, for example). Well in line with anti-Chomskian ideas, Mohammad Aslam, in his very informative article entitled Communicative Language Teaching and Large Classes Conflicts and Convergence, (514–525), begins his discussion of ‘Communicative Language Teaching’ (CLT) method by referring to Dell Hymes’s objection to Noam Chomsky’s famous distinction between language competence (knowledge of grammatical rules necessary to understanding and producing language) and language performance (production of actual utterances). Following Hymes, Aslam contends that Chomsky’s competence restricts itself to the grammatical well-formedness or correctness of the native languages, whereas communicative competence, as he puts it, ‘moves beyond grammar and defines the use of language as a vehicle for negotiating meaning and meaningful communication.’ This is so because communicative competence includes grammatical, lexical, discourse, strategic, sociolinguistic and pragmatic competencies. However, I do not remember Chomsky ever advocating his concept of linguistic competence as the basis of any language teaching methodology. Attacks on Chomskyan ideas apart, CLT method has both merits and demerits which have long been discussed in the literature. I have personally witnessed CLT’s effectiveness in my classes over a period of 20 years. Nevertheless, I believe that different elements of language require variegated methodologies to be dealt with properly. Just one methodology – be it CLT or any other – does not deal properly with all the topics to be taught in classrooms. Thus, in teaching certain elements of language, I have found going back-tobasics much more helpful than any other deemed to be superior methodologies. When it comes to the language teaching in large classrooms in India, though, I entirely agree with the guidelines and suggestions provided by Aslam. Process of Learning English EFL vs. ESL: A Study of Errors in Acquisition vs. Learning (526–537) by Vaishna Narang & Mi Yang Cha analyses errors of Koreans learning English as a foreign language (EFL) and second language (ESL). Through analysis of errors made by students learning English in Korea (i.e. EFL) and students learning English in Delhi (i.e. ESL), the authors show that in the case of EFL it is ‘learning’ whereas in ESL it is ‘acquisition’.

Reviews

201

In his paper entitled Kashmiri-English Contact: Some Phonological Issues (538–555), Aadil A. Kak analyzes abundant data to highlight the phonological difficulties Kashmiri speakers face while learning English. Needless to say his main concern is to find phonological differences between Kashmiri and English. He notices that due to the ‘English influence’, the English /f/ has entered Kashmiri phonology. He concludes by saying that “The increase in status of English, its global value and greater dependence on it in many spheres indicates that the influence of English on Kashmiri and the level of assimilation is going to increase, rather than decrease, with the passage of time.” (p. 555). Lexico-Semantic Features of Indian Pidgin English by Priya Hosali (556–573) clearly depicts the nature of Indian Pidgin English. After her careful analysis of the Indian variety of pidgin English, she concludes by saying that “... pidgins / creoles as repositories of man’s innate ‘communicative competence,’ have proved themselves indispensable in sociohistorical contexts. The study of these linguistically poor relations of great language families, bred in harsh and limiting conditions, may yet have much to tell us about the nature of human interaction – from the first halting attempts at communication to the ultimate recognition of a shared identity.” (p. 572). The article entitled Globalization and English Language in India: An Overview by Raj Nath Bhat (574–580) views the importance of the English language to India in the present globalizing world. He recommends that ways and means should be found to bring English within easy reach of all as it is and will remain the main official language of the union government and of many states. He concludes by saying “As of date, the future of English in India and the world stands unchallenged and the Indian higher education has the potential to make best from it.” (p. 580). In the copy of Festschrift I have received, The study of Lexical Phrases Among Proficient Users of English by Anju Sahgal Gupta & Madhu Gurtu (581–592) does not read well at all and requires a thorough editing. However, the title suggests that the paper intends to highlight the important role played by ‘lexical phrases’ in language acquisition. An addendum in the end of the volume entitled Rama Kant Agnihotri: Publications, Awards, Current Projects and Activism (593–601) lists the honoree’s publications, current projects as well the awards bestowed on him by different scientific organizations. To sum up then, as mentioned in the beginning of this review, the Editors have accomplished an arduous task. They have attempted to summarize

202

Ghanshyam Sharma

the contributions in their Introduction, which is a useful guide for anyone who wants to have a glimpse of what the different contributions are all about. However, when it comes to the organization of the volume they seem to have set neither a rigorous editorial plan nor any scientific requirement for inclusion in the Festschrift; probably taking extra-scientific factors into consideration. They have failed, I believe, in accommodating the contributions in their rightful places, thus unjustifiably devising three arbitrary sections in the volume as if to promote a three-in-one mixed-bag. Needless to say, if one publishes a paper by Albert Einstein on the theory of Relativity and a poem by Charles Baudelaire together, one does justice to neither (both get applause at their proper places only!), nor even to the reader who gets completely lost in this amalgam. Similarly, their choice of the title Problematizing Language Studies is, in my view, itself problematic since it does not accurately relate to the actual meaning of the term ‘problematizing’. The Editors take no pains to explain how these pieces cohere together and can be seen as constitutive of the problem they see as uniting them. They express their customary thankfulness to the ‘meticulous copy editing’ (p. 8), but there is hardly any page free of typos and other mistakes, grammatical and otherwise. A simple word processing search-andreplace tool would have saved many typos. Some contributions require a close and critical reading while others need editing by a competent English speaker. The last two contributions (i.e. 34 and 35) do not read well because of misplacement of pages. The paper by Anju Sahgal Gupta & Madhu Gurtu is missing entirely. Although they fail to discharge their technical and intellectual responsibility, South-Asianists in India must be grateful to S. Imtiaz Hasnain & Shreesh Chaudhary for keeping the tradition of Festschrift alive. I hope to see other initiatives of this type to honor other eminent Indian linguists. I have found in the volume useful discussions of many topics I must brush up my knowledge on, and I have no doubt that South-Asianists will find the volume equally useful, no matter which fields of linguistic or literary enquiry they are engaged in.

References Chatterji, S. K. 1960 Indo-Aryan Hindi. Calcutta: Firma K. L. Mukhopadhyay. Chomsky, Noam 1986 Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger Publishers.

Reviews

203

Cruse, D. A. 1986 Lexical semantics. Cambridge: Cambridge University Press. 2000 Meaning in language. Oxford: Oxford University Press. Singh, Rajendra 2006 Whole Word Morphology. In Encyclopedia of Language & Linguistics, Keith Brown (ed.), Vol. 13, 578–579. Oxford: Elsevier. Singh, Rajendra, and Rama Kant Agnihotri 1997 Hindi morphology – A word-based description. Delhi: Motilal Banarsidass. Singh, Rajendra, and S. Neuvel 2003 When the Whole is smaller than the Sum of its Parts. Chicago Linguistics Society Papers 38, 299–307.

Appendices

Announcements The Gyandeep Prize

We are happy to re-announce the continued availability of this annual prize of US$ 400. 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. Papers submitted after these deadlines will be processed, but only for a later issue. A paper initially submitted after November 1, 2011, for example, will be considered only for the 2013 issue of ARSALL. Potential contributors are encouraged to send their initial submissions as word and pdf files. The submissions must be prepared following the Mouton style-sheet which can be obtained from [email protected] Authors of accepted contributions must, however, furnish their Word files on the appropriate template and their PDF versions. They must also send a 50 word bio-blurb as a word file at the time of final submission. The template can be obtained by writing to the Editor or directly from: http://www.degruyter.com/cont/imp/mouton/moutonAuthors.cfm. 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

Umberto Ansaldo, Ph.D. (Stockholm) is Associate Professor of Linguistics in the School of Humanities at The University of Hong Kong. His interests include contact linguistics, evolutionary theories of language, and description and documentation of South and Southeast Asian languages. He is the author of Contact Languages: Ecology and Evolution in Asia, Cambridge University Press 2009. Shishir Bhattacharja, Ph.D. (Montréal) is Associate Professor of French at the University of Dhaka. He was trained in Indology and Linguistics at the University of Sorbonne, Paris and the University of Montreal. 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]) John Peterson, Ph.D. (Kiel), currently visiting professor of Linguistics at the University of Leipzig, has published extensively on the Indo-Aryan and Munda languages of South Asia. He is founder and co-manager (with Tej Bhatia, Syracuse) of the email-discussion group Vyâkaran and is co-editor (with Anju Saxena, Uppsala) of the series Brill’s Studies in South and Southwest Asian Languages. He also manages the “Bibliography for Seldom Studied and Endangered South Asian Languages” (http://www.southasiabibliography.de)

210

Notes on contributors

Felix Rau is research assistant in the Department of Linguistics at the University of Cologne and doctoral candidate at Leiden University. He has studied General Linguistics and Indology and has been working in the Koraput Area of Orissa (India) since 2002. The focus of his research has been Gorum as well as the Indo-Aryan Desia Oriya. He has also worked on other Munda and Dravidian languages of the Eastern Ghats. ([email protected]) Pingali Sailaja, Ph.D. (CIEFL, Hyderabad) is Professor of English at the University of Hyderabad. She is interested in the historical, educational and linguistic aspects of English in India, and in morphology. Her previous publications include English Words: Structure, Formation and Literature (2004, Pertinent Publishers, Mumbai) and Indian English (2009, Edinburgh University Press). ([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 enunciation in Hindi.