280 12 1MB
English Pages 289 [273] Year 2004
Rhythm and the Synthetic Drift of Munda* PATRICIA DONEGAN and DAVID STAMPE
im«l«mlEn StAnley H. StArostA-n «s«nensenAs«n, 1939±20021
1. The Structural Divergence of Munda and Mon-Khmer The South Asian (Munda) and South-East Asian (Mon-Khmer) branches of the Austroasiatic language family2 are perhaps the most divergent in the world. They are opposite in structure at every level (Table 1): Table 1. Polarizations in Munda vs Mon-Khmer discussed in this paper.3
Grammar: Word Order: Phrases: Words: A½xation: Timing: Fusion: Syllables: Consonants: Tonality: Vowels:
Munda
Mon-Khmer
§§
Synthetic Head-last: OV, Postpos. Falling (initial) Falling (trochaic) Pre/in®xing, Su½xing Isosyllabic/Isomoric Agglutinative (C)V(C) Stable/Assimilative Level (rare) Harmonizing/Stable
Analytic Head-®rst: VO, Prepos. Rising (®nal) Rising (iambic/monosyll.) Pre/in®xing or Isolating Isoaccentual Fusional  (ù/V9 )(C) (C(«()) (C)V Shifting/Dissimilative Contour (common) Reducing/Diphthongizing
3, 8 3, 8 4 5, 8 5 5±8 7.1 7.1 7.2 7.3 7.4
In Sora,4 a Koraput Munda language of Orissa, the sentence `He doesn't want to give me the rice' is head-last and synthetic, as in (1), but in Khmer5 (Cambodian), it is head-®rst and analytic, as in (2):
Anin dN- øEn dAr«j -«n «- tiy -ben idsöm he/she obj- me rice -art inf- give -inf want -tE ted -3pr not
(1)
Sora:
(2)
Khmer: ko(«t /«t cAN /aoy baay køom he/she not want give rice me
4
Patricia Donegan and David Stampe
The polarity of Munda vs Mon-Khmer recalls that of ancient vs modern Indo-European: synthetic head-last vs analytic head-®rst (Lehmann 1974). But Munda and Mon-Khmer are far more divergent. Indo-European was never polysynthetic, but many Munda languages are. For example, in Sora an even more synthetic (and more idiomatic) rendering of sentence (1) crystallizes it into a single complex word «dm«ltiydAriødAe: (3)
Sora: «d- m«l- tiy -dAr -iø -dA -e not- want- give -rice -me -aux -3pr
And while all modern Indo-European languages, even English, retain some in¯ection, most Mon-Khmer languages lack it entirely, and VietnameseM°¡ng lacks a½xation entirely. The Munda and Mon-Khmer branches of Austroasiatic, rarely studied by typologists, provide a nearly exhaustive inventory of the extremes of di¨erence in human language structure.
2. The Structural Oppositions of South Asia and South-East Asia The main reason for the neglect of the Munda and Mon-Khmer divergence is that each is spoken in a linguistic area (Sprachbund ) where its structure is su½ciently typical as to seem unremarkable.6 Most of the Munda traits in Table 1 are found also in the other language families of the South Asian areaÐDravidian, Indo-Aryan, and Tibeto-Burman, as well as isolates like Nihali and Burushaski. And most of the Mon-Khmer traits in Table 1 are found also in the other language families of the mainland South-East Asian areaÐTai-Kadai, Hmong-Mien, Chamic (Austronesian), and Chinese. In some respects Munda and Mon-Khmer are not typical of their areas. For example, South Asian languages are predominantly su½xing (and Dravidian exclusively so), but Munda languages are also pre®xing and in®xing: (4)
g«{b}rj-l -Ay Sora: «- «dn- «lpl- not- recip- {caus} feel-ashamed -pa -1st `We (exclusive) didn't shame each other'
And South-East Asian languages are pre®xing or isolating, but Nicobarese is also su½xing. However, in most respects Munda and Mon-Khmer are more like the unrelated languages of their areas than they are like each other. Most of the boundary lines between major typological di¨erences in Asia in the maps in Masica (1976) run precisely between South and SouthEast Asia. Except in groups that have crossed over that boundary in the past two millenia, there has been deep and divergent coalescence in the respective areas.
Rhythm and the Synthetic Drift of Munda
5
Areal contact certainly might explain the similarities within each area. But it cannot explain the di¨erences between them. And Munda and MonKhmer, and the South and South-East Asia areas, are not just di¨erent from each other, they are systematically opposite at every level. To explain the holistic polarization of structures in Munda vs Mon-Khmer, and in South vs South-East Asia, we seek a linguistic opposition which might pervade and organize every level from syntax to phonetics. The only plausible candidate is initial vs ®nal accent in phrases and in words. We will speak of these as falling vs rising rhythms. Munda and other South Asian languages have falling phrase rhythms (as in noÂun poÁstposition) and, excepting some Indo-Aryan languages, also falling word rhythms (as in baÂse-suÁ½x). Mon-Khmer and other South-East Asian languages have rising phrase rhythms (as in preÁposition noÂun) and rising word rhythms (as in preÁ®x-baÂse). We will argue that this opposition of falling vs rising rhythm is what maintains the opposition of South Asian vs South-East Asian structure, and that it was a change from a rising rhythm in proto-Austroasiatic (which we will show originally had a rising rhythm and analytic typology) to a falling rhythm in proto-Munda that channeled the drift of the individual Munda languages as their highly synthetic structures evolved. 3. Historical Issues Grierson, in the introduction to the Linguistic Survey of India, at ®rst doubted whether languages with such an opposite ``order of ideas'' as Munda and Mon-Khmer could be related at all (1904: 2). Schmidt (1906) established their genetic relationship, and Pinnow (1959 et passim) has removed all reasonable doubts. But there remain disagreements about what proto-Austroasiatic was like, and therefore about how the polar opposition of Munda and the eastern Austroasiatic languages came about. 3.1. The analytic basis of proto-Austroasiatic Pinnow argued that proto-Austroasiatic had SVO order, based on the order of elements in the Munda verb (1960, 1966). Repeating our own examples (sentences 1, 3, 2, respectively), the Sora verb phrase is head-last: (5)
Sora: Anin dN- øEn dAr«j -«n «- tiy -ben idsöm -tE he/she obj- me rice -art inf- give -inf want -3pr me rice give want ted not not
6
Patricia Donegan and David Stampe
The head-last order of non-in¯ectional elements (bold-faced) in the Sora verb phrase (5) is the exact opposite of their head-®rst order in the Sora polysynthetic verb in (6) and their head-®rst order in the Khmer verb phrase in (7). The glosses apply to both (6) and (7): (Anin) «d- m«l- tiy -dAr -iø he not want give rice me
(6)
Sora:
(7)
Khmer: ko(«t
/«tcAN
-dA -e -aux -3pr
/aoy baay køom
Not surprisingly for languages separated for many millenia, there is but one cognate form in (6) and (7): the Sora pre®x «d- and the Khmer verb /«t (lit. `lack'), from proto-Austroasiatic *«t. But the patterns of (6) and (7) are cognate: the analytic Mon-Khmer pattern of the verb phrase in Khmer (7), /«t cAN /aoy baay køom, has crystallized into the polysynthetic Munda pattern of the verb stem in Sora (6), «d-m«l-tiy-dAr-iø-. From cognate patterns like these in Munda and Mon-Khmer, Pinnow concluded that proto-Austroasiatic must originally have had the analytic subject±verb±object (SVO) pattern that persists in Mon-Khmer syntax and in Munda morphology. Lehmann, in an article (1973) that extended the notion ``VO/OV'' to mean head-®rst/head-last order in other phrases, also extended Pinnow's conclusion about proto-Austroasiatic to other phrases: If we examine further evidence provided by Pinnow, we note that Munda contains VO characteristics. It has VO order in compounds (Pinnow [1960], 97); it also provides examples of NG [noun±genitive] order and of pre®xes. Since the Khmer-Nicobar languages are consistently VO, I assume that it was the Munda languages which were modi®ed syntactically. . . . We may conclude that ProtoAustro-Asiatic was VO and non-agglutinative in morphological structure. (Lehmann, 1973: 57)
Sora is particularly rich in examples of such ``VO'' (head-®rst) survivals: (8)
Noun±genitive compounds: head-®rst with ®rst and second person pronouns: si/iN1 -øEn2 `my2 house1 '; m/d1 -n«m2 `your2 eye1 '; grjAN1 -lEn2 `our2 village1 '; oA1 -y«N2 -ben3 `your (pl.)3 parents (lit. father1 -mother2 )'.
(9)
Noun±noun compounds: head-®rst is recessive but is still regular for many nouns: kyky1 -im2 `chicken2 tail-plume1 '; «söN1 -d«d2 `snail2 shell (lit. house1 )'; «sN1 -tAN2 `cow2 dung1 '; «loN1 -söN2 `interior1 of a house2 '; «bAy1 -m«d2 `pupil (lit. seed1 ) of the eye2 '.
(10)
Verb±object (object incorporated) compounds, all head-®rst: «1 -si2 `wash1 the hand2 s'; «1 -jeN2 `wash1 the feet2 '; gu111 -«r2 `erect1 a
Rhythm and the Synthetic Drift of Munda
7
funeral stone2 '; gu111 -söN2 `erect1 a house2 '; g«d1 -bör2 `clear (lit. cut1 ) a hill2 for shifting cultivation'; g«d1 -bN2 `sacri®ce (lit. cut1 ) a bu¨alo2 '; gAy1 -sAN2 `dig1 turmeric-root2 '; gAy1 -loN2 `dig1 a pit2 '; jom1 -dAr2 `feast (lit. eat1 rice2 )'. Similarly the re¯exive is head-®rst: g«d1 -d«m2 `cut1 oneself2 '; «1 -si2 -d«m3 `wash1 one's own3 hand2 s'. (11)
Other verb±dependent compounds, also all head-®rst: (agent) ø«m1 -köd2 `be caught1 by a tiger2 '; m1 -kol2 - `be swallowed1 by a ghost2 '; duN1 -yoN2 `for the sun2 to rise1 '; (instrument) r«j1 -kon2 `chop1 with a knife2 ', töd1 -dAN2 `beat1 with a stick2 '; (locative) d«ko1 -söN2 `stay1 at home2 '; (completive) jom1 -Aj2 `eat1 up2 '.
(12)
Auxiliary±verb compounds, also all head-®rst: y«r1 -mEN2 `revive, lit. return1 -live2 '; m«l1 -gij2 `want1 to see2 '; r«bti1 -«md«N2 `can1 hear2 '. (But like all phrases in Sora, auxiliary phrases are headlast, e.g. rA/A1 -n2 «3 -gij4 -ben3 «5 -r«bti6 -l7 -E8 p9 ? `Were7 you8 (pl5 ) able6 to3 see4 the2 elephant1 ?9 '.)
The shapes of the elements in the Sora compounds in (8) through (12) echo the bare and often monosyllabic shapes of independent words in Mon-Khmer, and their head-®rst internal order echoes the order of phrases and compounds in Mon-Khmer. All this, together with the striking rarity of head-last order in Mon-Khmer, supports Lehmann's conclusion that protoAustroasiatic syntax was head-®rst not only in the predicate (VO) but in phrases generally. Both Pinnow's and Lehmann's conclusions rest on an implicit but welltested hypothesis that syntactic patterns become morphological patterns, but morphological patterns do not become syntactic patterns.7 An original proto-Austroasiatic head-®rst syntactic pattern, maintained in Mon-Khmer, has become a head-®rst morphological pattern in the Munda polysynthetic verb. The converse changeÐdissolving a polysynthetic word like Sora «dm«l-tiy-dAr-iø- (6) into an analytic phrase like Khmer /«t cAN /aoy baay køom (7)Ðseems quite impossible. Clearly, it is Munda that innovated, ®rst by joining proto-Austroasiatic head-®rst analytic phrases (7) into head®rst polysynthetic words (6), a process paralleled also in a few other Austroasiatic languages like Nicobarese, and then, along with a reversal from rising to falling rhythm, by reversing head-®rst analytic phrases into head-last in¯ected phrases (5). Despite his own evidence for SVO structure in proto-Austroasiatic, Pinnow (1960) hypothesized that proto-Austroasiatic might have been synthetic, on the basis of variation in Khmer word-®nal consonants that he took as evidence for former su½xes. But Jacob (1992), citing extensive Khmer data, showed that this consonant variation is not grammatical but
8
Patricia Donegan and David Stampe
a¨ective, and that it involves not only ®nal but also initial consonants and even vowels. Such a¨ective variation is widespread in Mon-Khmer (see Jacob's list of references, 1992: 71) and in Munda (Kuiper 1965). More recently, Zide and Anderson (2001) have assembled cognates in various eastern Austroasiatic languages for a number of verbal formatives in the Munda languages. They take many of these not only to have been part of proto-Munda but also of proto-Austroasiatic. Some are derivational pre®xes or in®xes of nominalization, causativization, and so on, that have long been accepted as proto-Austroasiatic. Others are in¯ectional elements like person and number a½xes, tense/aspect a½xes, and so on, which were also reconstructed by Pinnow in his work on the Munda pronouns (1965) and the verb (1966). Zide and Anderson's work on these incorporates South Munda data that provide a much clearer view of early Munda than was available to Pinnow. They criticize Donegan and Stampe (1983) for the view that the Munda morphology must be seen as in large part due to the independent synthetic drift of the daughter languages rather than due to the breakdown of a fully formed verbal system in proto-Munda. Our views are based not on some a priori scepticism about reconstruction, but on the di½culty of explaining the variety of combinations of elements of the verb in the daughter languages if those elements were already a½xes rather than free forms in proto-Munda. At the level of proto-Austroasiatic the di½culty is far greater, because in the vast majority of eastern languages the cognate elements are free forms. To reconstruct synthetic morphology for protoAustroasiatic implies that the vast majority of eastern languages lost verbal in¯ection and morphology and became analytic.8 But that would entail that in most Mon-Khmer languages that former a½xesÐelements characteristically faded in pronunciation, grammatical autonomy, and meaningÐhave been restored to full lexical speci®city and function. It is far more likely that Munda and the few eastern languages that show some signs of synthesis have innovated it. The reason is that most of the elements that are functionally identical, whether cognate or not, are placed at one end of the verb in Munda and at the other end the eastern languages. The only obvious explanation for that is that those elements were still free forms that could trade places with the verb after Munda changed from head-®rst and pre®xing to head-last and su½xing. 3.2. Syntax: Progession vs regression9 There is an old view that languages naturally change from synthetic to analytic, but that the opposite change, if it were to occur at all, would have to be due to external causes. The same scepticism has extended to changes from head-®rst to head-last order.
Rhythm and the Synthetic Drift of Munda
9
Friedrich von Schlegel (1808) contrasted the ``ancient and artful'' forms of Sanskrit, Greek, and Latin, which expressed structure morphologically, by a½xing or modifying a root, with the younger Romance or English, which expressed structure by adding separate words like auxiliaries and prepositions. The general tendency for languages to change from synthetic to analytic ``shows up everywhere the same'', Schlegel said; ``no external cause is necessary'': The ingenious structure is readily lost through wearing away by common usage, . . . and the grammar with auxiliaries and prepositions is actually the shortest and most convenient, like an abbreviation for simple, general usage; in fact one could almost establish the general rule that a language is the easier to learn, the more its structure has been simpli®ed and approximated to this abbreviation (translation by Lehmann 1967: 26).
Jespersen (1922: chapters 18±20) concurred, arguing that the change from synthetic to analytic is not ``decay'' but ``progress'': (1) The forms are generally shorter. . . . (2) There are not so many of them to burden the memory. (3) Their formation is much more regular. (4) Their syntactic use also presents far fewer irregularities. (5) Their more analytic and abstract character facilitates expression by rendering possible a great many combinations and constructions which were formerly impossible or unidiomatic. (6) The clumsy repetitions known under the name of concord have become super¯uous. (7) A clear and unambiguous understanding is secured through a regular word order. . . . This development may truly . . . be termed a progressive evolution (1922: 364).
The implication is clear: that an opposite change from analytic to synthetic would be regressive, and would occur only due to some external cause. The typological classi®cation of languages as head-last vs head-®rst has come to be treated as an even more basic ``parameter'' of grammar than analytic vs synthetic. Roughly half the world's languages are of each type. From a purely logical view of grammar, there is no reason to regard one as more natural than the other. But Yngve (1960) argued that left-branching (head-last) structure puts a burden on short-term memory and that many transformations function to reduce left-branching. Chomsky (1965: 197± 198) quarreled with this, but neither he nor Yngve observed that head-last structure seems to require synthesis. As Greenberg (1963: 96) put it, ``if in a language the verb follows both the nominal subject and nominal object as the dominant order, the language almost always has a case system''Ðas in Dravidian and Indo-Aryan. And if not a case system, we could add, then subject and object marking on the verb, as in Munda. There is rarely such marking either of nouns or verbs in Mon-Khmer and other languages of South-East Asia, where the verb precedes the object.
10
Patricia Donegan and David Stampe
Also, head-last sentences seem uncomfortable with more than one ®nite verb. Clauses that head-®rst sentences comfortably embed often must be nominalized in head-last sentences. Head-last clauses are ¯attened down into phrases. In Sora and many head-last languages, a conditional clause must be nominalized and treated as the object of a postposition equivalent to the preposed subordinating conjunction if of head-®rst languages. An explanation of these remarkable asymmetries was proposed by one Mark Twain in The Awful German Language (1880). Based on research ``upwards of nine full weeks'' (616), Twain found that a German sentence treats of fourteen or ®fteen di¨erent subjects, each enclosed in a parenthesis of its own, with here and there extra parentheses which re-enclose three or four of the minor parentheses, making pens within pens; . . . after which comes the VERB, and you ®nd out for the ®rst time what the man has been talking about. . . . German books are easy enough to read when you hold them before the looking-glass or stand on your headÐso as to reverse the construction (Twain 1880: 603).
He provided an example of a verb-last subordinate clause, with a literal translation, and parentheses and hyphens to help the English reader: Wenn er aber auf der Strasse der in Sammt und Seide gehuÈllten jetz sehr ungenirt nach der neusten Mode gekleideten Regierungsrathin begegnet . . . But when he, upon the street, the (in-satin-and-silk-covered-now-veryunconstrainably-after-the-newest-fashion-dressed) government counselor's wife met'', etc., etc. (Twain 1880: 604).
``You will observe'', Twain said, ``how far that verb is from the reader's base of operations''. Putting this in terms linguists can understand, the verb is the head of its sentence, and the head of a construction can stand for the whole. In Twain's example, the verb, begegnet `met', singlehandedly gives us the gist of the sentence: a meeter met a meetee. If the verb came early, everything else would just elaborate that gist. But the verb comes last and its elaborations come ®rst! So to make the best of a bad thing, we are given case marking to help us sort out which elaborations are whichÐ er is the masculine doer, and der the feminine doee, etc.Ðuntil at last we get to the verb and ®nd out what he did, upon the street, to her, dressed after the latest fashion, the government counselor's wife: he met her. Case marking or verb agreement occur in most languages, but as MonKhmer shows, both can be dispensed with entirely in head-®rst languages. Head-last order may be logically equal to head-®rst order, but if it requires synthetic structure, whatever the reason for that may be, then it is not psychologically equal. Elements that are compounded or a½xed are altered
Rhythm and the Synthetic Drift of Munda
11
in form and meaningÐgrammaticized or lexicalized. To say our example sentence in Khmer, (2), we only need to know the Khmer words for he not want give rice I. To say it in Sora, (1), we need to know not only the Sora words but also how to mark want for tense, person, number, and other things that we won't even mention, what ambi®x to wrap around give to show that it isn't the head verb, whether rice needs an article, and how to show that I isn't the subject. It is hardly surprising that a change from head-last to head-®rst order and synthetic to analytic structure could be viewed as progress, but that a change from head-®rst to head-last order, and from analytic to synthetic structure, could be viewed as regressionÐ something that would happen only under outside in¯uence.
3.3. Outside in¯uence? Scepticism about whether a `natural' or `internal' drift could be toward head-last and synthetic structure accords with the persistent view that this structure in common Munda arose due to areal in¯uenceÐDravidian rather than Indo-Aryan in¯uence, given the relatively late date of Indo-Aryan settlement in the subcontinent. It is still customary to assume that areal similarities that do not have a genetic basis must be due to areal contact. This assumption resulted in classi®cations of languages as ``mixed'' in the South-East Asian as well as the South Asian area. In particular, Vietnamese was thought to be mixed until Haudricourt (1954) showed that its tones were not derived from Thai or Chinese but arose by rephonologization of the phonation types of its inherited Austroasiatic consonants. Sora was labeled mixed until the appearance of Ramamurti's grammar (1931), which brought him a letter from Edward Sapir, saying, in part: I note from the references to Savara in recent general linguistic surveys by Kieckers, by Meillet and Cohen and by Father Schmidt that Savara is classi®ed as a mixed MundaÅ language, owing to supposed serious in¯uence exerted by Aryan and Dravidian. I gather from what you say that the language is quite de®nitely of the MundaÅ type and is to be classi®ed without reservation with such typical MundaÅ languages as SantaÅli. (Letter quoted in an advertisement in Ramamurti 1933: 259±260)
The classi®cations of Sora to which Sapir referred10 all simply echoed the statement by Konow in his Munda and Dravidian volume of the Linguistic Survey of India that ``Savara has been largely in¯uenced by Telugu and is no longer an unmixed speech'' (1906: 218). Konow says that all he knows of Sora was gleaned from the texts submitted to the Survey, but there is nothing Telugu in those texts. Konow had some knowledge of Santali and
12
Patricia Donegan and David Stampe
other North Munda languages, and apparently, wherever Sora seemed to di¨er from them, he simply assumed it was due to non-Munda in¯uence. But while Santali is ``typical'' of Munda languages in having assimilated every phoneme of a neighboring Indo-Aryan language, Sora alone has no foreign phonemes. Santali and most Munda languages have adopted much vocabulary from neighboring non-Munda languages (Pinnow estimates that Kharia has 40 per cent Indo-Aryan words). And they have adopted Ç processes and constructions along with them. But Sora has grammatical adopted only one, the Indo-Aryan ¯exional -/-i of Oriya for naming males/females, as in tAb«n/tAb«ni, a man/woman calledÇ `bamboo shoot' (tAb«N), or g«d-s«r-gAn/-gAni, a man/woman born in g«d1 -s«r2 -gAj3 `cut1 -rice2 -moon3 '. Finally, speakers of most Munda languages also speak a local nonMunda language, and some speakers of the less populous languages use their mother tongues only in private. For example, Das Gupta (1978: 4) could ®nd no monolingual Juangs at all. But among Hill Soras, inquiries during our 1980's ®eld work about how many know any language besides Sora were answered not with numbers but with names. Hill Soras expect to be spoken to in Sora, and do their trading via agents bilingual in Sora and Oriya (Vitebsky 1993). But Çnot only Sora but all Munda languages, despite the recent in¯uence of non-Munda languages, remain solidly Munda in their basic structures. Despite foreign phonemes, they have kept native phonemes and processes intact. Though most languages have lost the central vowel series of protoMunda, these vowels are retained in Sora, and they are reconstructible at every branch in the Munda family tree (see section 7.4). As for consonants, Munda languages retain the treatment of ®nal stops as checked and voiceless but morphophonemically voiced (see section 7.3)Ðeven Kharia and the North Munda languages, which have added the full complement Çof released ®nal stops (e.g. p ph b bh) from Indo-Aryan. As to an ancient in¯uence of Indo-Aryan or Dravidian on Munda, no convincing evidence has been presented to support the theory that Munda synthesis and head-last order were borrowed. Even in vocabulary, few Indo-Aryan or Dravidian words appear in a form in Munda languages that would indicate ancient borrowing.11 And the converse is true as well: Few of Kuiper's Proto-Munda Words in Sanskrit (1948), for example, have turned out to be proto-Munda, and only a handful of words in the large etymological dictionaries of Indo-Aryan by Turner (1966) or of Dravidian by Burrow and Emeneau (1984) seem likely to turn out to be Munda.12 As for grammatical form, though Munda is head-last and synthetic like Indo-Aryan and Dravidian, it di¨ers from them in a way which Nichols (1986, 1992) has argued is more resistant to change than is word order.
Rhythm and the Synthetic Drift of Munda
13
Indo-Aryan and Dravidian languages are mainly dependent-marking, marking noun phrases for their case relation to their verbal head. The Munda languages are mainly head-marking, marking verbal heads for their relation to their noun phrase arguments. Most Mon-Khmer languages mark neither. The hypothesis of a Dravidian substratum would imply something that seems unimaginable: that dependent-marking speakers adopted a nonin¯ecting language and made it head-marking.13 Sapir said, ``Language is probably the most self-contained, the most massively resistant of all social phenomena; it is easier to kill it o¨ than to disintegrate its individual form'' (1921: 220). Indeed, the impact of Indo-Aryan and Dravidian languages on Munda languages has been less to change them than to replace them. Munda languages were once spoken over much of IndiaÐsee Elwin's summary (1955: chapter 1) of ancient Indic and Greek references to SÂabara or Savara or SuariÐbut most of that vast area is now Indo-Aryan or Dravidian-speaking. But death is not di¨usion. There is little solid evidence of assimilation of early Munda to Indo-Aryan or Dravidian, or the reverse. Languages of India share traits like those in the Munda column of Table 1, but so do head-last languages everywhere. These and many other traits Indologists have regarded as areal were shown by Masica (1976) to extend over central and northern Eurasia, as well as to occur in geographically remote areas. Munda languages have been seen as genetically related not only to MonKhmer, but also Dravidian, Tibeto-Burman, Burushaski, Nihali, Vedda, and geographically remote head-last languages like Finno-Ugric, Turkic, Australian, Basque, and Japanese, by linguists who found deep similarities in Munda.14 Such comparisons have been ridiculed when the similarities proved not to be inherited nor di¨used, but they deserve some explanation. The structures of languages are not just inherited or borrowed, they are also shaped to the needs of their speakers and hearers, and of those needs, two are inseparable: ®rst, a consistent grammatical form, and second, a consistent rhythmic sca¨olding for realizing that form in utterances that must be constructed, communicated, and comprehended in real time.
4. Falling and Rising Rhythm ``Finding a way into a conversation'', said Tannen (1994: 18), ``is like joining a line of dancers''. Speaking and listening, and all regular voluntary action, in ensemble or solo, outwardly or in imagination, is performed and perceived to a tacit real time rhythmic score. Knowing a language is not just a knowledge of words and constructions and propositions, it is also the ability to hear and speak them as beats and phrases and melodies.15
14
Patricia Donegan and David Stampe
The characteristic rhythm of a language depends on how it ®ts words and phrases to the beats and measures of the rhythmic score. If the fronts of words and phrases go on the main beats, that is falling rhythm; if their ends go on the main beats, that is rising rhythm. The distinction between falling and rising rhythm is particularly evident in oral verse and song. Old English had a falling rhythm, and an alliterative verse that foregrounded the fronts of words at the fronts of phrases (13a), but Modern English has a rising rhythm, and a rhyming verse that foregrounds the ends of words at the ends of phrases (13b): (13)
a.
Falling: stoÂrmaÁs Q^ r staÂnclõÁfu beÃotaÁn storms there stone-cli¨s beat `Storms beat the cli¨s of stone there' (Seafarer, l. 23a)
b.
Rising: aloÃng caÁme a spõÂder and saÃt doÁwn besõÂde her (Little Miss Mu¨et, l. 3)
Old English with its falling rhythm had a mainly head-last syntax, while Modern English with its rising rhythm has a mainly head-®rst syntax. The concomitant historical drifts in English that reversed the Germanic rhythm, word order, and verse structure also occurred independently in Latin and Romance, Celtic, and other western Indo-European families. Falling phrase rhythms are typical of head-last languages, such as the Munda group and other languages of the South Asian area, Australia, northern Eurasia, Korea, and Japan. Rising phrase rhythms are typical of head-®rst languages, such as the Mon-Khmer group and other languages of the South-East Asian area, Oceania, western Europe, and much of subSaharan Africa. 4.1. Heads and dependents A phrase consists of a head word or phrase plus zero or more dependent words or phrases which specify (qualify, quantify, modify) the head. The head can stand for the whole phrase, but the dependent cannot (to readH a bookD is to read, not a book). Even when the head is pronominalized or deleted (The brownD bookH or the blueD oneH ?ÐThe brownD oneH , please), its presence is always implicit. The dependent must be explicit. This is well illustrated by questions, which assume some information and ask for more information about it, and by answers, which present the assumed information as head (or just omit it) and the new information as dependent. This relationship of known and unknown information has universally been grammaticalized and lexicalized by the provision of interrogative pronouns
Rhythm and the Synthetic Drift of Munda
15
that request a dependent for a known head, e.g. WhichD bookH ?ÐThe brownD bookH . But there are no interrogative pronouns that request a head for a known dependent, e.g. *WhatH brownD ?ÐThe brownD bookH . As is shown in Table 2, in every phrase type, the head corresponds to old information and the dependents to new information, but not the reverse, in a potential question and answer.16 Table 2. Question words correspond to dependents (in bold), not heads. [V O]
She [got a book].
What did she get? *WH-Verb she a book?
A book. *Got it.
[V Adv]
He [speaks clearly].
How does he speak? *WH-Verb he clearly?
Clearly. *Speaks.
[Mod V]
He [can type].
What can he do? *WH-Modal he type?
Type. *Can.
[Prep N]
[from there]
Where from? *WH-Prep there?
There. *From.
[N Rel]
The [book that we lost].
Which books? *What that we lost?
*The books.
[N Gen]
A [book of John's].
Whose book? *What of John's?
John's. *A book.
The ones we lost.
The dependent also gets the main beat (accent) relative to the head. This is true not only if it is newer information than the head, as in answering the questions of Table 2, but also if the head and the dependent are both old or both new, as in We didn't readH a ÈbookD , we bakedH a ÈcakeD . The usual situation has been conventionalized as the usual rhythm: modern English has a rising rhythm, and most dependents have moved to the phrase-®nal beat.17 Where the ancient dependent-®rst word order has not shifted to dependent-last, as in English [AdjD NH ] redD booksH (cf. French [NH AdjD ] livresH rougesD ), there is a discrepancy: if the dependent is newer than the head, it steals the main beat from the end of the phrase: ÈredD booksH . But if there is no contrast in newness, the default English phraselast rhythm is dominant: redD ÈbooksH . 4.2. Munda vs Mon-Khmer heads and dependents In section 3.1 we showed that Munda morphology has the same head-®rst order of elements in Mon-Khmer syntax, that proto-Austroasiatic must have had head-®rst syntax, and that Munda must have shifted its syntax
16
Patricia Donegan and David Stampe
from head-®rst to head-last. In Table 3 we contrast Munda and MonKhmer order in various phrase types, showing how their opposite word orders re¯ect their opposite phrase rhythms. In either rhythm, dependents go on the main beat, so that in Munda with its falling phrase rhythm there is a dependent±head phrase order, while in Mon-Khmer with its rising phrase rhythm, there is a head±dependent phrase order. Table 3. Opposite order in Munda vs Mon-Khmer: [È] marks dependents.18 Munda (Sora)
Mon-Khmer (Khmer)
reads1 the book2
[ÈO V] k«mbl2 -«n ken-l1 -t-E
[V ÈO] m««l1 si«wphöw2
gives1 the book2 to me3
[ÈDat [ÈO V]] dN-øEn3 k«mbl2 -«n tiy1 -t-E
[[V ÈO] ÈDat] qaoy1 si«wphöw2 khøom3
goes1 quickly2
[ÈAdv V] omeN2 yer1 -t-E
[V ÈAdv] töw1 rhah2
stays1 in2 the house3
[È[ÈN Adpos] V] si/iN3 -leN2 -«n d«ko1 -t-en
[V È[Adpos ÈN]] nöw1 knoN2 pte«h3
very1 small2 house3
[È[ÈAdv Adj] N] byby1 s«nnA2 si/iN3 -«n
[N È[Adj ÈAdv]] pte«h3 tooc2 nah1
grandfather1 's2 house3
[ÈGen N] jojo1 -n «2 -si/iN3 -«n
[N ÈGen] pte«h3 taa1
the book1 which5 is2 in3 the house4
[ÈRel N] «5 -si/iN4 -leN3 -«n d«ko1 t-en-«n5 k«mbl1 -«n
[N ÈRel] si«wphöw1 nöw2 knoN333 pte«h4
Reversals of rhythm and word order, as in Indo-European and in Munda, are not very common in the world's languages. Niger-Congo (GivoÂn 1975) had a progressive shift like Indo-EuropeanÐfalling to rising and head-last to head-®rst. Tibeto-Burman, given the pre®xing character of proto-Sino-Tibetan (Benedict 1972), may have had a regressive shift like MundaÐrising to falling and head-®rst to head-last. A regressive shift entails the construction of an in¯ectional system (section 3.3), and surely takes far longer than a progressive shift. Judging from the time depth of the far less complete reversal of type in Indo-European, Munda must have a time depth of several millenia. The recorded histories of Indo-European languages show that reversals of grammatical structure proceed gradually, construction by construction,
Rhythm and the Synthetic Drift of Munda
17
hingeing on local analogies and ambiguities. Even where we can ®nd the grammatical function of a given change, it is often clear that alternative changes were available. To understand why local changes taking many generations can result in a consistent global reversal of word order requires a factor both persistent and pervasive, namely a reversal of phrase rhythm. 5. Phrase Rhythm and Word Rhythm Munda and most South Asian languages, with falling phrase rhythm, also have falling word rhythm, while Mon-Khmer and most South-East Asian languages, with rising phrase rhythm, also have rising word rhythm. Some languages with falling phrase rhythm (and head-last order) are described as having word-®nal accent, for example the Turkic languages, but those we have heard put the beginning of the word on the beat, and the ``accent'' that is described is merely a terminal rise in pitch. Korean has a rather similar rhythmic system. ``Accent'' usually coincides with the beat, but in some languages it has no rhythmic relevance whatever. For example, the pitch accents of Vedic or Homeric Greek or of Japanese seem to play no role in the rhythm even of their verse. By falling and rising word rhythm we refer only to whether the beginnings or ends of words come on the rhythmic beats, even if this does not coincide with other ``accents'' in the language. Just as falling and rising phrase rhythms do not necessarily put the very ®rst or last word of a phrase on the beat, word rhythms do not necessarily put the very ®rst or last syllable of a word on the beat. Some languages skip an initial or ®nal syllable, particularly pre®xes or su½xes. And even if only a root or stem is eligible, its very ®rst or last syllable may not be. Beats are not pulses but divisions of time, su½cient for a long syllable (like English Èstead ±) or two or perhaps three short syllables (Èsteady ±µ or Èsteadier ±µµ, the latter spoken as a triplet) of which the ®rst is the most prominent. We use English examples here because our readers will know the rhythm of English, and because these examples were carefully measured by Lehiste, who found that they are spoken isochronously by ``temporal compensation'' of their syllables (1971, 1977). A stress accent, as in English, foregrounds a stressed syllable by lengthening it at the expense of unstressed syllables. To use more precise notation, Èstead, Èsteady, Èsteadier take a dotted rhythm, ±, Ö.», Ö.= » respectively. The duration of any syllable varies inversely with the durations of the other syllables in the beat. This was a structural fact of early English, where long vowels shortened phonemically before two syllables in the same beat (ÈsaÅne : ÈsaÆnity), and both vowels shortened in two-beat compounds that were reduced to one beat (ÈwaistÇcoat : Èweskit). Beats and syllables are not only the domains
18
Patricia Donegan and David Stampe
of timing but also they and their natural parts (beginning, rise, peak, fall, end) are the domains of phonological processing (Donegan and Stampe 1978). For example, sonorant nasalization is limited in English to one beat (deÈlõrÄ.õÄ.uÄm) or even just one syllable (deÈlir.i.uÄm). In many languages a short anacrustic syllable may be pre®xed to the beat, like a musical grace note, as in inÈstead _ ±. Like a grace note, such an anacrustic syllable is temporally inert: it is not part of the following or the preceding beat, and its presence does not perceptibly shorten nor does its absence lengthen the following or preceding syllables. It is the ``rhyme'' of a word that determines its timing, from the most prominent syllabic to the end of the word. Mappings of words and phrases onto beats or measures may be iambic, like the word inÈstead, or the verse line |and Èmiles| to Ègo| beÈfore| I Èsleep|, but as in western music notation, the beats and measures of the rhythmic score itself are always front-prominent.19 To ®t into a beat, which is universally bimoric, the word should begin either with two light syllables like Sora Èur.A ±µ `mango' or one heavy like Èy«N ± `mother' or ÈsinÇdi ± ± `date palm'. In a word consisting of a light plus a heavy syllable, the light syllable is ignored: Sora E È/El _ ± `ironwood', t« Èröb _ ± `cloud', which is rhythmically equivalent to eliminating the short syllable, tröb ±.20 Mon-Khmer and other mainland South-East Asian languages put the word-®nal syllable on the beat. To ®t the bimoric duration of the beat, short-voweled open ®nal syllables may be extended either by lengthening the vowel or by closing it with a glottal stop.21
5.1. Word rhythm and compound structure The accentuation of compounds also exempli®es the principle that heads are rhythmically subordinate to modi®ers. Mon-Khmer compound nouns are head-®rst, with a rising rhythm as in Mon-Khmer phrases and words: (14)
Çlaan - Ècnu«l Khmer: ÇsAc - Èmo«n Çbaay - Èprök ¯esh - chicken food - morning car - hire `chicken meat' `breakfast' `rental car'
In Munda, the picture is more complex. Sora, for example, has three patterns of compounds. There is a productive pattern exactly opposite the Mon-Khmer pattern: (15)
Sora: Èk«nsim - ÇjEl È«b«b - Çsu Èk«nsim - ÇsN chicken - ¯esh head - pain chicken - feces `chicken meat' `headache' `chicken manure'
Rhythm and the Synthetic Drift of Munda
19
But head-last compounds are fairly recent in Munda. Older compounds, such as Sora «drE1 -im2 `chicken egg (lit. egg1 -chicken2 )' or k«mbol1 -si2 `biceps muscle (lit. rat1 -arm2 )' have head-®rst structure like those of MonKhmer. The Munda falling rhythmic pattern has been imposed even on these older compoundsÐÈ«drE-im, Èk«mbol-si Ðso that the rhythm no longer ®ts the old head-modi®er order. In the formation of Sora verb stems, however, the rising rhythm is still intact, and the head-®rst structure still remains productive, e.g. g«d1 -Èim2 `sacri®ce a/the chicken' (lit. cut1 -chicken2 ). This is in contrast with the less idiomatic expression of this in separate words: Èk«nsim2 -«n3 g«d1 - (lit. `chicken2 art3 cut1 '), which must be head-last. 5.2. Word rhythm and a½xation Most Mon-Khmer languages, like other South-East Asian languages, lack su½xes entirely; only pre®xes and in®xes occur. Munda languages have some pre®xes and in®xes, but many more su½xes. The su½xes are not reconstructable as su½xes to proto-Austroasiatic, and only a minority of them reconstruct as su½xes even to proto-Munda. So su½xation must have been an ongoing tendency in Munda languages. This has often been attributed to contact with the su½xing Dravidian languages, but the Munda su½xes are not borrowings or even calques of Dravidian su½xes. Instead, pre®xation vs su½xation re¯ects rising vs falling word rhythm. Exclusively pre®xing languages, like the languages of South-East Asia, have rising word rhythms. Exclusively su½xing languages, like Dravidian, Finnic, and Turkic, have falling word rhythms. The reason must be to background a½xes by putting them at the far end of the word from the beat. Rising rhythm backgrounds pre®xes; falling rhythm backgrounds suf®xes. In Mon-Khmer, new a½xes are pre®xed; in Munda, they are su½xed. 6. Timing It should be clear from Table 1 that it is not only grammatical traits that are opposite in Munda and Mon-Khmer and their areas; phonological traits are opposite, as well. We are not referring here to phoneme inventories and morphophonological alternations, but to the living prosodic and featural processes that distinguish and polarize the two language areas, and to the word forms that result. The rhythmic type of a language is a pervasive in¯uence in its living phonology, and since phonological processes apply to rhythmic domains, they are strongly linked. In our 1983 paper, we attributed the divergent phonological typologies of Munda and Mon-Khmer to Munda's syllable rhythm (isosyllabic or
20
Patricia Donegan and David Stampe
isomoric, depending on whether short and long syllables are distinguished) and falling accent, versus Mon-Khmer's word rhythm (isoaccentual, in bimoric beats) and rising accent. But just as all speakers have the same phonetically motivated processes, but must inhibit them di¨erently to speak different languages, so too all speakers are motivated to give moras equal time, and syllables, and words, but they are forced by the structure of their languages to yield on one or more of the principles. In monosyllabic languages like those of the Vietnamese-M°¡ng group, the rhythmic principles do not con¯ict: each simple word is one syllable, and each syllable is heavy, so that any sequence of words is a sequence of bimoric beats. But there are two conditions on this kind of word-perfect rhythm: First, there should be hiatus at word boundaries: languages with liaison do not keep words rhythmically discrete and thus tend to time by moras (Greek, Latin) or syllables (Italian, French), not by words. Second, most words must be su½ciently short to ®t into a beat: languages with front accent and multiple su½xes (Dravidian, Uralic, Altaic) have words of highly variable duration, and ®nd it easier to time by moras or syllables. Mon-Khmer languages, with hind accent, hiatus, short words, and no su½xes, can have word timing. Munda languages, with front accent and a great accumulation of su½xes, must live with syllable or mora timing. These di¨erences, as is evident to the ear of any traveler, are true also of the distinct linguistic areas where Munda and Mon-Khmer are spoken. 7. Rhythm and Phonology Real-time rhythm is as central to phonology as it is to syntax. Beats and measures are the domains into which speech material is ®tted, and they are the domains to which the e¨ects of phonological processes are restricted. When too little or too much material is ®tted into a beat or measure, the phonological processes that the language has not inhibited apply to ®ll out the timing or to trim the excess. These processes create the syllable types, consonant formations, tonal patterns, and vowel inventories of languages. In this section, we will consider how the dominant rhythmic principles of each Austroasiatic branch have created opposite phonological types. 7.1. Rhythm and syllable canon Proto-Austroasiatic had isochronous words of one or two syllables. The disyllable had a rising rhythm, like *b«Èlu _ ± `thigh': its ®nal syllable got a full beat, like a monosyllabic word, and could contain a full long or short vowel plus consonant, while the initial syllable, called ``minor'' (Shorto 1960), had only a «-like vowel (Pinnow 1959; Shorto 1976).
Rhythm and the Synthetic Drift of Munda
21
In Mon-Khmer, a distinctively short vowel in the ®nal syllable was kept short by inserting a glottal stop b«Èlu/, but otherwise could be merged with the corresponding long vowels b«Èluù. The short initial syllable invites vowel reduction or deletion, and the long ®nal syllable invites diphthongization, as in Khmer Èplöw ± `thigh'. Vowel deletion creates the complex initial clusters typical of Mon-Khmer, but such consonant clusters may be reduced (Khmer slaap-Èpri« `spoon' @ s«Èpi«). Since minor syllables are non-moric, the initial consonant clusters that result from vowel deletion do not include geminates; and any consonant that is completely assimilated simply disappears, since it lacks any moric value, e.g. Khmer pram1 -b«y2 `eight' (lit., ®ve1 -three2 ) @ mb«y. Fusions of adjacent morphemes like this favor a fusional morphology, and in extreme cases, like Vietnamese, they have led to a complete loss of a½xation. In Munda, the disyllable was given a falling rhythm, *Èb«lu ±µ, ®tting the ®nal syllable into the beat by shortening its vowel (proto-Munda seems not to have had vowel length distinctions), and giving the initial syllable a full though short vowelÐoften, by harmony, Èbulu. The shift to falling accent encouraged encliticization and su½xation in Munda, e.g. Sora Èbulu1 -lEn2 ji3 `our2 thigh1 s3 '. Word rhythm became impracticable; the languages shifted to syllabic rhythm, which because of the inherited distinction between open and closed syllables is of the isomoric variety. This rhythm supports geminate consonants in the Munda languages: a syllable-®nal consonant completely assimilated to an adjacent consonant retains its moric value, as in Sora b«ttN `frighten' < b«{b1 }tN2 `cause1 to fear2 '. Where consonants retain their presence, even if not their full identity, the morphemes of which they are part retain their separate identities, as is typical of agglutinative morphology.
7.2. Rhythm and consonants Austroasiatic onsets had a stop voicing distinction, which was universally preserved in Munda; this conservation of phonation type in consonants seems typical for India, with only sporadic exceptions. But in Mon-Khmer and South-East Asia, consonant shifts are commonplace (Haudricourt 1965). One of the main causes of this is stress (word) timing, as also in Germanic. Also it may also be encouraged by the dissimilation of initial consonant clusters, which for reasons already explained are commonplace in Mon-Khmer (Khmer pdou `to exchange', tm«y `new'). Such clusters are susceptible to assimilation and if the assimilation is complete, to loss. Since assimilation favors more-similar sounds, dissimilation can block it, and such strengthenings may extend to initial consonants in general.
22
Patricia Donegan and David Stampe
A prosodic characteristic of languages that is often overlooked because it is not ``distinctive'' is whether or not ®nal consonants are pronounced with a vocalic release or are unreleased. This is a very important feature in the phonological typology of languages. A ®nal release is typical of Indo-Aryan and Dravidian languages and helps preserve voicing, aspiration, and a¨rication in ®nal consonants. In Oriya (Indo-Aryan), consonant-®nal words Ç a release that became identi®ed with a apparently were pronounced with vowel, so that Oriya words all end in vowels (Masica 1991: 197). In DraviÇ dian, the ®nal release also has created an ``enunciative'' vowel (Bright 1975). One result of this tendency is that released ®nal consonants may become continuants since they are between continuants (the preceding vowel and the following release), as in Tamil, or in Europe in the lenitions of Celtic, Spanish, Danish, etc. Austroasiatic, Sino-Tibetan, and the South-East Asian language groups pronounce ®nal consonants without release (rather like the /t/ in English right now). This prevents ®nal consonants from becoming continuant, but it eventually limits ®nal obstruents to voiceless stops with simultaneous glottalization, and in many languages of this type it has resulted in the loss of oral articulation in some ®nal stops, leaving only the glottal stop behind. In Mandarin this has happened to all ®nal stops. In Munda it happens in several languages to velars, e.g. South Munda *lAg `leaf', Juang olAg, but Kharia olA/, Gorum olA/A, Gutob olA/ (in Sora lA, even Ç the glottal stop was lost), etc., and in some to other positions of articulation as well, e.g. to coronals in Remo pine/, Gutob pine/ `¯ute', Gta/ pini `horn' (cf. Sora p{«n}Ed, nominalization of pEd `to blow a ¯ute'), and to labials in Gta/ gtA/ `ethnonym' (Gutob gutob), slA/ `tree' (Gutob sulob). Consonant release/nonrelease is a remarkably stable feature historically. If it is related to the distinction of falling/rising rhythm, the relationship is not clear to us. None of the languages of either area seem to have switched from releasing to nonreleasing, or vice versa. We have argued elsewhere, in fact, that one very odd characteristic of MundaÐthat it has only voiced stops morpheme-®nally even though word-®nally they are pronounced as checked and voiceless, as in Sora p«tod [p«to't] `a hole' beside the form with the article [p«tod-«n]Ðis due to the voicing of word-®nal checked stops at the moment that proto-Munda ®rst began to su½x vowel-initial su½xes. There were voiceless as well as voiced stops non®nally, as in Sora b«tN `fear' vs p«dAb `mushroom', so all morpheme-®nal stops had to be interpreted as phonemically voiced (b d j g). This innovation is not found elsewhere in Austroasiatic, or apparently in the world, and it strengthens the widely held view that Munda (North and South) is indeed a single genetic family in Austroasiatic (Donegan and Stampe 2002).
Rhythm and the Synthetic Drift of Munda
23
7.3. Rhythm, tone, and register Consonant phonations which are neutralized may have re¯exes in tonal distinctions, as Haudricourt (1954) demonstrated for Vietnamese, and as others (e.g. Matiso¨ 1973) have observed elsewhere in South-East Asia. The tones of Mon-Khmer languages, like those of other South-East Asian languages, include many contour tones, which re¯ect the typical bimoric structure of the stressed syllables on which they occur. Distinctions of voice-register (Henderson 1952) in Khmer and several other Mon-Khmer languages have similar origins (Hu¨man 1976, Gregerson 1976). So tone and register are the re¯exes of consonantal phonation distinctions that have been lost in the consonant shifts typical of South-East Asian languages. Phonation types are more stable in isomoric and isosyllabic languages, so rephonologization of consonant phonation as tone is rare in India. Where it has occurred, as in the Munda language Korku, the resulting tones are level rather than contour (Zide 1966b). 7.4. Rhythm and vowels Mon-Khmer languages show vowel reduction, but Munda languages often show vowel harmony. The di¨erence results from their opposite rhythms. In the Mon-Khmer rising word rhythm, the unaccented initial syllable is anacrustic, and there is pressure to minimize it. Since it is not in the beat with the rest of the word, its vowel can su¨er a fairly context-free reductionÐshortening, narrowing, and loss of color (labial, palatal or velar). This is a perpetual tendency, producing synchronic variation as in Khmer prÈlöm @ pr«Èlöm `dawn', prAÈhael @ pr«Èhael @ p«Èhael `similar', bANÈri«n @ b«NÈri«n `to teach', etc., even in borrowed words like French cravat > Khmer kraaÈwat @ kraÈwat @ kr«Èwat `necktie'. Indeed the vowel can be lost entirely, as in *b«Èluù > Khmer Èplöw `thigh', cited earlier, or baÈzaar > Khmer Èpsaa. In the Viet-M°¡ng group, this resulted in monosyllabism. Other language families of South-East Asia, since they share the same iambic word rhythm as Mon-Khmer, show a similar treatment of their unaccented vowels. Chamic languages, for example, shifted the penultaccented words in their inherited Austronesian lexicon to ®nal accent, and then reduced the rhythmically demoted vowels (Thurgood 1999). In the Munda falling word rhythm, on the other hand, the whole word is usually part of a single beat or measure, and in its syllable- or moratimed rhythm even unaccented syllables get at least one mora of time. They are less apt to be reduced (narrowed, bleached) than harmonized to features of other vowels in the word. Harmony can involve color, as in the *b«Èluù > Munda *Èbulu `thigh' example, or height, as in the o/u alternation
24
Patricia Donegan and David Stampe
of Korku kor `person' with the plural Èkur-ku, or the A/« and o/u alternations in Santali ÈA)ondA Ç«)undi `anxiously'. Every Munda subgroup shows evidence of synchronic or diachronic vowel harmony. See Donegan (1993) for examples, and Bodding (1930: 18±34) for a discussion of the association of harmony in Santali with its two-syllable stress-unit (our beat). Harmony also occurs in Dravidian (Bright 1966) and Indo-Aryan (e.g. Majumdar 1970: 118±119 on Oriya, Chatterji 1926: 387±402 on Bengali), as well as in remote falling-rhythm language families like Altaic and Uralic. The typical South-East Asian vowel system has a back or central unrounded series. These vowels can be reconstructed in every subgroup of Munda (Munda 1969, Norman Zide 1965, 1966b, Stampe 1978, Arlene Zide 1982), but they have been eliminated separately in each group by fronting, rounding, or lowering, so that most Munda languages have the ®ve-vowel systems typical of Indian languages. Sora is the only language that keeps these un-Indian central vowels, but even here they only occur in closed syllables, which suggests that their intrinsic shortness is not so compatible with a syllable rhythm. As in most Indian languagesÐand most languages with a syllable rhythm, Munda vowel qualities have apparently remained quite stable for centuries, except for this loss of central nonlow vowels. In marked contrast to the stable vowels of Munda, the stressed vowels of Mon-Khmer, like those of other languages with stress-timing, undergo repeated diphthongizations and vowel shifts. Many Mon-Khmer languages retain a vowel-length distinction; this was lost in Munda, probably because moving the accent o¨ the ®nal syllable in the change to falling rhythm caused the vowel of that syllable to shorten. And many languages reinforce the short vs long distinction with lax vs tense, and then also with in-gliding (centering) like i«, u«, µ«, vs outgliding diphthongs like Ai, Au, Aµ. Register di¨erences, as in Khmer, also can a¨ect vowel qualities (Gregerson 1976). This multiplication of vowel qualities often leads to further diphthongization and vowel shifts (Donegan 1993 gives examples and comparisons with European vowel shifts). As a result, Mon-Khmer languages often have large vowel inventories. While the typical Munda language has ®ve vowels (Sora is extraordinary in having nine), among Mon-Khmer languages, nine counts as a small vowel inventory. Khmer, in Hu¨man's analysis (1970a and b), has thirty-one vowels (including long and short monophthongs and diphthongs); other Mon-Khmer languages have even more. 8. Rhythmic and Grammatical Convergence in Head-last Structure We have proposed that the holistic oppositeness of Munda and MonKhmer linguistic structure could have been the result, after many millenia,
Rhythm and the Synthetic Drift of Munda
25
of a simple change of Munda from a rising to a falling rhythm. But falling rhythm imposes head-last order and synthetic structure, which are both so complicating (section 3.2) that one must ask why they persist in South Asia and nearly half the world's languages, and indeed why they exist at all. The answer must be that falling rhythm itself has intrinsic value. It has been noted that child speech shows a ``trochaic bias'' (Allen and Hawkins 1978, 1980). This bias is clearly a re¯ection of the fact that the division of real time into beats and measures is universally front-prominent: temporal compensation and phonetic processes operate within groupings of strong± weak, never weak±strong.22 Only words and phrases spoken in a falling rhythm ®t neatly into these universal divisions of time. In `A new knife is hard to sharpen', as spoken in Sora (16), dependents go on the main beats, so that phrases ®t into measures, and words into half-measures. (The pre®x in « -ÈtAji-Çben is squeezed in by anacrusis.) (16)
Falling (Sora): ± ± ¦¼ ± Ä à é onÇdi-n k [ Èt«bÇmE knife H new D
| ][
_ ±µ ± «-ÈtAji-Çben to.sharpen D
± ± ék«lÇk«l hard H
| ]
But words and phrases spoken in rising rhythm, as in the Khmer translation (17), never ®t into the rhythmic divisions. Putting the ®nal syllables of dependents on the main beats, the result is that phrases straddle measures, two-syllable words straddle half-measures, and many rests are needed: (17)
Rising (Khmer): ± | ± £ ä | ¦¼ £ ± ± £ | ± £ ä | £ ± ± Çk«m éb«t Ètm«y ] [ Çpi ébaa/ ÇsAmÈli«N ] [ knife H new D hard H sharpen D
Reducing minor syllables to anacruses (± ± ! _ ±) and extending major syllables over following rests ( ± £ ! ±_ ± ) ®ts the words into halfmeasures in a typically Mon-Khmer fashion, (17 0 )
Khmer with maximal anacrusis and legato: ¦¼ ä _ ±_ ± | ±_ ± ä | ä _ ±_ ± | _ ±_ ± ä | [ ][ ] km éb«ùt Ètm«ùy p« ébaaù/ smÈliù«N
but it does not ®t the phrases into the measures. To do that the words and phrases of (17) would have to be reversed from rising to falling rhythm: (18)
Pseudo-Khmer with falling rhythm: ± £ | ± ± ± ± ¦¼ ± ± étm«y ] [ épiÇbaa/ Ès«mÇli«N [ Èk«mÇb«t knife H new D hard H sharpen D
| ]
26
Patricia Donegan and David Stampe
But the accentuation is backward for head-®rst order: A knifeH that's new D is hard H to sharpen D . Falling phrase rhythm requires a head-last order: (19)
Pseudo-Khmer with falling rhythm and head-last word order: ¦¼ ± £ ± ± | ± ± ± ± | [ Ètm«y ék«mÇb«t ] [ ÈsAmÇli«N épiÇba/ ] new D knife H sharpen D hard H
This is exactly how Munda diverges from Mon-Khmer, with a drift from head-®rst to head-last order (19) accompanying a regularizing shift from rising to falling rhythm (18). The opposite drift to head-®rst in languages like English may be driven by grammatical simplicity and regularity (section 3.2), but this is achieved only with a shift from a tight, regular ®t of words and phrases into beats and measures (20a) to a loose, syncopated ®t (20b): (20)
(a) ( 13a) Old English: ¦¼ ± ± ± £ | ± ±µ [ stoÂrmaÁs Q^ r ] [ staÂnclõÁfu storms there stone-cli¨s
± ± beÃotaÁn beat
| ]
(b) ( 13b) Modern English: ¦¼ ä _ ± ± | _ Ö.» £ _ ± ± | _ Ö.» £ ä | [ aloÃng caÁme a spõÂder ] [ and saÃt doÁwn besõÂde her ] It does not seem unreasonable, then, to suppose that a head-last, synthetic drift, as in Munda, might be driven by rhythmic simplicity and regularity. 9. Conclusion The divergent typologies of the Austroasiatic languages of South vs mainland South-East Asia, and of the several other language families in either area, are not limited to synthetic and head-last vs analytic and head-®rst grammatical structure, but pervade every level of structure down to phonetics and prosody (section 1). The two areas are not only di¨erent, but opposite at every level, even in falling vs rising rhythms (2). Since the Munda languages of South Asia and the Mon-Khmer languages of South-East Asia are genetically related, one of them must have changed. Munda shows rich evidence of an earlier head-®rst analytic structure, and there is little evidence of earlier synthetic structure in Austroasiatic (3.1). Munda therefore must have changed, becoming synthetic and head-last in spite of the problems of that structural type (3.2). But the opinion that such a shift was due to areal in¯uence does not stand up to close scrutiny (3.3).
Rhythm and the Synthetic Drift of Munda
27
We propose that Munda had a genuine independent drift to synthetic and head-last structure due to a shift from rising to a falling rhythm. Rising vs falling rhythm go with head-®rst vs head-last syntax because of the backgrounding of heads relative to their dependents (section 4). And they go with pre®xing vs su½xing morphology due to the backgrounding of a½xes relative to their stems (5). Their e¨ects extend to timing as well: word (stress) isochrony is optimal, but languages must settle for mora or syllable isochrony if su½xing makes their words too variable in length (6). These timing di¨erences deeply a¨ect the phonology of syllables (7.1), consonants (7.2), tone and register (7.3), and vowels (7.4). While the shift from rising to falling rhythm in Munda might have been due to contact, it might instead be due to the fact that grammatical and rhythmic structure are in phase only with falling rhythm (8). The better-known drift of Indo-European from head-last to head-®rst structure is well attested in Celtic, Romance, and other western languages. That drift was reversed in India as Indo-Aryan was adopted by Mundas and Dravidians with their falling rhythms, and by the time of the Prakrits, Indo-Aryan was Indo-European in little but etymology. Munda structures are far more various and cognates far fewer than in Dravidian, and likewise than in eastern Austroasiatic. This suggests that the Austroasiatic people may have dispersed from South Asia rather than South-East Asia, and that the shift of Munda from rising to falling rhythm, after the eastern languages had moved eastward, may have been the cause rather than the e¨ect of the profound polarization of South and South-East Asian language structures. Notes *
Research on this paper began in the 1970's and was reported in papers cited under Donegan and/or Stampe in the bibliography. For help and/or argument, we thank Paul Benedict, Chris Court, GeÂrard Di¿oth, Ulli Dressler, Talmy GivoÂn, Joseph Greenberg, Ken Hale, EugeÂnie Henderson, Paul Hopper, Bernhard Hurch, Philip Jenner, Bhadriraju Krishnamurti, Bill Labov, Greg Lee, Winfred Lehmann, Ilse Lehiste, Christopher Longuet-Higgins, Jim McCawley, Miren Lourdes OnÄederra, Frans Plank, A. K. Ramanujan, Haj Ross, Chhany Sak-Humphry, Eric Schiller, Ron Scollon, Harry Shorto, Raj Singh, Frank Southworth, Richard M. Stallman, Stan Starosta, Larry Thompson, Graham Thurgood, Ben Ts'ou, Michael Witzel, Carl and Flo Voegelin, Theo Vennemann, Norman Zide, and Arnold Zwicky. For help in our work on the languages, we thank E. Annamalai, Sudhibhushan Bhattacharya, H. S. Biligiri, Beryl Girard, Allison MacDonald, Bijoy and Ranganayaki Mahapatra, Khageswar Mahapatra, Veena Malhotra, Ram-Dayal Munda, Arjunguru J. Patel, Heinz-JuÈrgen Pinnow, Dobek Pujari, Mnsi Raika, Taban Sara, Norman Zide and members of his Indian and American Munda Languages
28
1.
2.
3.
4.
5.
6.
Patricia Donegan and David Stampe Project in the 1960s (see Stampe 1965±1966), and grants from the University of Chicago, Ohio State University, the American Institute for Indian Studies, the Fulbright Program, the National Endowment for Humanities, and the National Science Foundation. And ®nally, we thank Yo Tomita for his TrueType Bach Musicological Font, available as shareware at http://www.music.qub.ac.uk/~tomita/bach-mf.html. `In honor of our old friend Stanley H. Starosta', who joked with us in Sora, the language used to represent Munda in this paper, and one of the many languages in which he was wittier, and about which he was wiser, than anyone else we know. A memorial for Stan is at http://www.ling.hawaii.edu/faculty/stanley/. On the divergent structures of Munda and Mon-Khmer, see Pinnow 1960; for a brief overview and maps, see Di¿oth (1978), Parkin (1991, ethnography only), and http://ling.lll.hawaii.edu/austroasiatic/. The only etymological dictionary and comparative phonology of Austroasiatic is Pinnow (1959), despite its modest title, Versuch einer historischen Lautlehre der Kharia-Sprache. Articles, reviews, and bibliography on Munda are now included in Mon-Khmer Studies. Table 1 is simply a table of contents, not a tabulation of invariable associations in these languages nor in languages of the world. Any theories here are in the text, not in the table. The column headings of the table, Munda and Mon-Khmer, re¯ect the custom in South-East Asian forums of calling all non-Munda Austroasiatic languages Mon-Khmer. This is an areal rather than a genetic grouping. In fact the genetic subgroupings of Austroasiatic are mostly undemonstrated. Our Sora phonemic transcriptions di¨er from IPA usage mainly in that y, j, .r, d are used for IPA [ j, ï, }, ê ]. The vowel transcribed u is a high unrounded (compressed) labial, as in Japanese and Finnish Swedish, for which there is no IPA symbol. The IPA-based transcription that christianized Soras use, adapted from one devised by Ramamurti (1931, 1933, 1938) after he heard Daniel Jones lecture on the new IPA in Madras in 1911, is unreliable for vowels, as is Ramamurti's, who inconsistently transcribed Sora vowel qualities in terms of accent and length, which led Pinnow (1959) wrongly to reconstruct vowel length for proto-Munda. Norman Zide's Munda project brie¯y revived linguistic work on Sora: in 1962 by Stampe and the late H. S. Biligiri with Monsi Raika and Daman Buy, then in 1963 by Starosta, Bijoy P. Mahapatra, and K. Ranganayaki [later K. Mahapatra] with Raika and Taban Sara. Later work has been done by Arlene R. K. Zide, Khageswar Mahapatra, Piers Vitebsky, Stampe and Donegan with various guides. Besides their works in the reference list, all contributed to a new dictionary of Sora which continues to grow (a link is at http://ling.lll.hawaii.edu/austroasiatic). Our Khmer transcriptions, except where we quote others' citations, are in the system of Hu¨man (1970a, 1970b), except that we use / instead of q, to parallel our Munda forms. On Khmer see also Maspero (1915), Henderson (1952), Jacob (1960 through 1993), Pinnow (1979a, b), Jenner and Pou (1980±81), Pou (1992), and Sak-Humphry (1996). The term linguistic area, Harry V. Velten's translation of N. S. Trubetzkoy's term Sprachbund, has geographic as well as typological reference. But in fact the Indian linguistic area, as Masica (1976) pointed out, actually extends far northwest of India and Pakistan. And it also includes the Tibetan-Burman languages, which
Rhythm and the Synthetic Drift of Munda
7. 8.
9.
10. 11. 12.
13.
14.
29
extend north through Tibet and eastÐin the case of Burmese and KarenÐto overlap the typologically South-East Asian languages (Mon, etc.) of Burma. On the other hand, the language area called (mainland) South-East AsiaÐjust in the case of Khasi (Mon-Khmer) and Khamti (Daic)Ðoverlaps the typologically South Asian languages of north-eastern India, and it extends north into China to include Chinese (at least in its southern varieties), and south to include the MonKhmer languages of the Malay Peninsula (Aslian) and of the Nicobar Islands. On the traits of the South Asian linguistic area, see Emeneau (1956), Kuiper (1966), and particularly Ramanujan and Masica (1969), Masica (1976). On IndoAryan see the surveys of Bloch (1965) and Masica (1991) and the etymological dictionary of Turner (1966). On Dravidian see the surveys of Steever (1998) and Krishnamurti (2003) and the etymological dictionary of Burrow and Emeneau (1984). On Tibeto-Burman, there is http://stedt.berkeley.edu/, the web site of James Matiso¨'s Sino-Tibetan Etymological Dictionary and Thesaurus project, with maps and bibliography, but no etymologies so far. On the traits of the South-East Asian area, see Henderson (1965), Hu¨man (1973), Matiso¨ (1973), and Gregerson (1976). The STEDT web site lists as forthcoming a volume Languages of Mainland Southeast Asia, edited by Matiso¨, in the Cambridge [University Press] Language Surveys series. On the reasons for the irreversibility of ``grammaticalization'', see, for example, Hopper and Traugott (1993). Zide and Anderson (2001) describe Donegan and Stampe (1983) as connecting polysynthesis to falling rhythm in Munda, but if we had done that we couldn't have explained why Munda polysynthetic constructions are head-®rst! The explicit links to falling/rising in our article did not include polysynthesis, which begins as compounding, and as we showed, Sora has both older rising and newer falling patterns in compounds. (On compounds in this paper see section 5.1.) Head-®rst vs head-last order, or right-branching vs left-branching structure, have been called progressive vs regressive (e.g. Yngve 1960). These are just technical terms (cf. their use in labeling perseveratory vs anticipatory assimilation), but in the present section, quoting Jespersen's evaluative use of the term progressive, their ambiguity seemed irresistibly apt. We argue in defense of regressive structure in section 8. Kieckers (1931), Schmidt (1926), and Meillet and Cohen (1924) (the ®rst edition, where the Munda section (385±403) was written by J. Przyluski). There are Dravidian loanwords in languages in contact with Oraon (Kurukh) in Chota Nagpur or with Telugu in Andhra Pradesh, but these look recent. Ironically, there is abundant evidence of early lexical in¯uence of Indo-Aryan (Sanskrit and Pali) on Mon-Khmer, as can be seen in the dictionaries of inscriptions in Mon (Shorto 1971) and in Khmer (Pou 1992). There is one group of Dravidian languages with some head-marking, incorporating pronominal objects, namely the South Central group, including Telugu and tribal languages spoken around the Koraput Munda area. But Steever's excellent study of analysis-to-synthesis in those languages (1993) shows little that is like Munda. A bibliography is in Pinnow (1959: 480±486).
30
Patricia Donegan and David Stampe
15. This applies of course to gestural language as well, and also to the coordination of speech and gesture. Jim McCawley once remarked in conversation that not only must both pronouns in He ate his lunch be deictic for both to be accented but also for both to be accompanied by pointing. The word, the gesture, and the accent are simultaneous. To a singer or dancer or comedian, there is nothing mysterious about this: the accent is a beat in the tacit real-time rhythmic stream to which we put the words and the movements. We linguists get this the wrong way around and speak of putting the accent on a certain syllable or word. That is like saying that Fridays come on payday! A rhythmic accent is a not an ``accentuation'': every complete utterance, even a one-syllable word, is ®tted into real time on a main beat (and thus takes a ``primary accent''), even if it is swallowed as in the very British example [Èk1kju] XÖ.» `Thank you' of Daniel Jones (Abercrombie [1964a] 1965: 20). The view that syllables are put on accents, just as words are put to music, and not vice versa, brings together the insights of Lashley on serial behavior (1951), Lehiste on speech rhythm (1970 etc.), Longuet-Higgins and Lee on musical rhythm (1984), and a long tradition in verse metrics. 16. The correspondence of dependents and interrogatives is an ancient observation. Starosta argued in a Tuesday linguistics seminar at the University of Hawai`i that the modi signi®candi of medieval grammar correspond to dependents in modern grammar. The modes of signifying (predicating) were based on the categories of Aristotle, which corresponded to the Greek interrogative pronouns. 17. We are grateful to Frans Plank for making us clarify that we do not of course mean that the dependent is always the newer information, but only that it is the default locus of newer information, and therefore that the grammatical association of the main beat with the dependent is a conventionalization of the pragmatic association of the main beat with new information. 18. The unglossed Sora in¯ections in Table 3 are the verbal a½xes -t `present tense', -E `third person singular subject' and the nominal a½xes dN- `dative/accusative' (which is su½xed to noun objects), and -(«)n `article'. In the examples in the ®nal row of the table, Sora and Khmer di¨er in a way that is typologically characteristic. Head-®rst Khmer embeds a relative clause with a ®nite verb, but head-last Sora avoids the ®nite verb and reduces the relative clause to a postpositional phrase. 19. The incompatible use of the term foot in verse metrics and in musicology (Cooper and Meyer 1960) makes us reluctant to follow Abercrombie (1964a) and many others in using foot to refer to beats. In the past we have sometimes used measure, but that traditional term is best reserved for larger structures built up of beats, e.g. ¦¼ Ö.» Ö.» Ö.» Ö.» | or ¦¼ _ Ö.= » ± ± Èel.eÇvat.or ßop.eÇrat.or inÈtell.i.gence Çtest ßcram
Ö.» | Çsession
We view pauses ( junctures) as phrase or word edges mapped onto rest beats, as in the contrast of meaning and phrasing in this traditional minimal pair: ¦¼ £ ± £ ± | _ Ö.» £ [ Çold [ Çmen and Èwom.en ] vs
ä ]
|
Rhythm and the Synthetic Drift of Munda ¦¼ £ ± ± £ | _ Ö.» £ [[ Çold ßmen ] and Èwom.en
ä ]
31
|
Often rest beats are realized in legato fashion, ®lled by speech material prolonged from preceding stronger beats, such as men and women in the preceding example, resulting in what linguists call ``®nal lengthening'': ¦¼ £ ± ± _ ± | _ Ö.»_± Çold ßme:::n and Èwom.en:::
ä
|
Viewing accents and junctures as levels of rhythmic duration rather than merely as prominences vs silences accounts for the four levels of accent and juncture that have traditionally been recognized as distinguishable in English words (NeÂwton Ö.»), vs compounds (NeÂwtoÁwn ± ±), vs phrases (neÃw toÂwn ± £ | ± £ or legato ±_± | ±_±), vs coordinates (eÂggs, oÂil, and leÂmons | ± £ ä | ± £ ä | _Ö.» £ ä | or legato | ±_± ä | ±_± ä | _Ö.»_± ä |). It explains the mutual shortening vs lengthening (temporal compensation) of the parts of tighter vs looser constructions, as in and their mutual levels of susceptibility to assimilation vs dissimilation. And ®nally, while it distinguishes rhythm from tempo, it explains why the phonological and phonetic e¨ects of rhythmic brevity parallel those of quick tempo. 20. Ignoring a short syllable before a long is not unusual in falling word rhythm, e.g. in Oriya (Majumdar 1970: 213) and Malayalam (Mohanan 1986: 111±115). Ç Because of the high frequency of ``iambic'' words in Munda, impressionistic descriptions have characterized some languages as having word-®nal accent, even languages like Sora and Remo, whose very names have initial accent. 21. This is paralleled in Japanese: one-mora words like ke Ä `hair' are extended utterance-®nally as ke/ ± in the Tokyo standard and as keù ± in other dialects. 22. The universal falling character of beats and measures is re¯ected in temporal compensation not only in speech but also in music. The non-anacrustic divisions of musical time begin with one prominent element and end before the next equally prominent element, whether it is beats being divided (± Ö.» Ö.= ») or measures (¦¼ | ¦¼ ° ° | ¦¼ ± ± ± ± |). The harmonic structure is mapped onto these front-prominent divisions, so that, in ¦¼ time, a harmonic change on the second half of a division entails one on the ®rst half. (That is true even in syncopated styles as in rock or jazz that put an ``accent'' on the even or ``back'' beat.) And exactly parallel to rising speech rhythms are rising musical rhythms, in which the most prominent note in a melodic phrase is the ®nal one: the phrase does not end in the measure where it began, but at the beginning of the next measure. This is why the ®nal note is usually lengthened: to ®ll that measure.
References Abbi, Anvita (ed.) (1997). Languages of Tribal and Indigenous Peoples of India: The Ethnic Space. (MLBD Series in Linguistics 10.) Delhi: Motilal Banarsidass Publishers. Abercrombie, David (1964a). A phonetician's view of verse structure. Reprinted in Abercrombie 1965: 16±25. Ð (1964b). Syllable quantity and enclitics in English. Reprinted in Abercrombie 1965: 26±34.
32
Patricia Donegan and David Stampe
Ð (1965). Studies in Phonetics and Linguistics. (Language and Language Learning.) Oxford: Oxford University Press. Allen, George D. and Sarah Hawkins (1978). The development of phonological rhythm. In Bell and Hooper, 175±185. Ð (1980). Phonological rhythm: de®nition and development. In Child Phonology, I: Production, Grace H. Yeni-Komshian, James F. Kavanaugh, and Charles A. Ferguson (eds.), 227±256. New York: Academic Press. Bell, Alan, and Joan Bybee Hooper (eds.) (1978). Syllables and Segments. Amsterdam: NorthHolland Publishing Company. Benedict, Paul K. (1972). Sino-Tibetan: a Conspectus. James A. Matiso¨, contributing editor. Cambridge: Cambridge University Press. Biligiri, H. S. (1965). The Sora verb, a restricted study. In Milner and Henderson, 1.231± 250. Bloch, Jules (1965). Indo-Aryan: From the Vedas to Modern Times. Translated by Alfred Master. Paris: Librairie d'Amerique et d'Orient Adrien Maisonneuve. Bodding, Paul Olaf (1930). Materials for a Santali Grammar I: Mostly Phonetic. 2d ed. Dumka: Santal Mission of the Northern Churches. Bright, William (1966). Dravidian metaphony. Language 43: 311±322 (1960). Reprinted in Bright 1990: 19±36. Ð (1975). The Dravidian enunciative vowel. In Dravidian Phonological Systems, Harold F. Schi¨man and Carol M. Eastman (eds.), 11±46. Seattle: South Asian Studies Program, Institute for Comparative and Foreign Area Studies, University of Washington. Reprinted in Bright 1990: 86±117. Ð (1990). Language Variation in South Asia. New York: Oxford University Press. Burrow, Thomas, and Murray B. Emeneau (1984). A Dravidian Etymological Dictionary. 2d ed. Oxford: Oxford University Press. [An announced reprint is actually of the 1st ed.] Chatterji, Suniti Kumar (1926). The Origin and Development of the Bengali Language. Part 1: Introduction, Phonology. Calcutta: University Press. (Reprinted 1975, Calcutta: Rupa & Co.) Chomsky, Noam (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Chomsky, Noam, and Morris Halle (1968). The Sound Pattern of English. New York: Harper and Row. Cooper, Grosvenor W., and Leonard B. Meyer (1960). The Rhythmic Structure of Music. Chicago: The University of Chicago Press. Das Gupta, Dipankar (1978). Linguistic Studies in Juang, Kharia Thar, Lodha, Mal-Pahariya, Ghatoali, Pahariya. Calcutta: Anthropological Survey of India. Di¿oth, GeÂrard (1978). Austro-Asiatic languages. Encyclopaedia Britannica 15th ed. 2.480± 484. Chicago: Encyclopaedia Britannica Inc. Donegan, Patricia Jane (1993). Rhythm and vocalic drift in Munda and Mon-Khmer. Linguistics of the Tibeto-Burman Area 116: 1±43. Donegan, Patricia Jane and David Stampe (1978). The syllable in phonological and prosodic structure. In Bell and Hooper, 25±34. Ð (1983). Rhythm and the holistic organization of language structure. In Papers from the Parasession on the Interplay of Phonology, Morphology, and Syntax, John H. Richardson, Mitchell Marks, and Amy Chukerman (eds.), 373±353. Chicago: Chicago Linguistic Society. Ð (2002). South-East Asian features in the Munda languages: Evidence for the analytic-tosynthetic drift of Munda. To appear in Papers from the 28 th Meeting of the Berkeley Linguistics Society. Berkeley, CA: Berkeley Linguistic Society. Elwin, Verrier (1955). Religion of an Indian Tribe. Bombay: Oxford University Press. Emeneau, Murray B. (1956). India as a linguistic area. Language 32.3±16.
Rhythm and the Synthetic Drift of Munda
33
GivoÂn, Talmy (1975). Serial verbs and syntactic change: Niger-Congo. In Li, 47±112. 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.), 73± 113. Cambridge, MA: MIT Press. Gregerson, Kenneth (1976). Tongue-root and register in Mon-Khmer. In Jenner, Thompson, and Starosta, 1.323±369. Grierson, George A. (1904). Mon-Khmer and Siamese-Chinese Families (including Khassi and Tai). (Linguistic Survey of India 2.) Calcutta: O½ce of the Superintendent of Government Printing. (Reprinted 1973, Delhi: Motilal Banarsidass.) Haudricourt, Andre G. (1954). De l'origine des tons en vietnamien. Journal Asiatique 242.69± 82. Ð (1965). Les mutations consonantiques des occlusives initiales en mon-khmer. Bulletin de la SocieÂte Linguistique de Paris 56.163±180. Henderson, EugeÂnie J. A. (1952). The main features of Cambodian pronunciation. Bulletin of the School for Oriental and African Studies 14.149±174. Ð (1965). The topography of certain phonetic and morphological characteristics of South East Asian languages. In Milner and Henderson, 2.400±434. Hopper, Paul J., and Elizabeth Closs Traugott (1993). Grammaticalization. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. Hu¨man, Franklin E. (1970a). Cambodian System of Writing and Beginning Reader. New Haven: Yale University Press. Ð (1970b). Modern Spoken Cambodian. New Haven: Yale University Press. Ð (1973). Thai and Cambodian: a case of syntactic borrowing? Journal of the American Oriental Society 93.488±509. Ð (1976). The register problem in ®fteen Mon-Khmer languages. In Jenner, Thompson, and Starosta, 1.575±589. Ð (1985). Bibliography and Index of Mainland Southeast Asian Languages and Linguistics. New Haven: Yale University Press. Hyman, Larry M. (ed.) (1993). Consonant Types and Tone. (Southern California Occasional Papers in Linguistics 1.) Los Angeles: University of Southern California. Jacob, Judith (1960). The structure of the word in Old Khmer. Bulletin of the School of Oriental and African Studies 23: 351±368. (Reprinted in Jacob 1993: 1±17.) Ð (1968). Introduction to Cambodian. London: Oxford University Press. Ð (1992). Some comments on the relationship between Khmer words having identical vowel nuclei and ®nal consonants. Mon-Khmer Studies 18±19: 67±76. (Reprinted in Jacob 1993: 167±178.) Ð (1993). Cambodian Linguistics, Literature, and History: Collected Articles. David A. Smith (ed.). London: School of Oriental and African Studies, University of London. Jenner, Philip M., Lawrence C. Thompson, and Stanley Starosta (eds.) (1976). Austroasiatic Studies. (Proceedings of the First International Conference on Austroasiatic Linguistics.) 2 vols. Honolulu: University of Hawai`i Press. Jenner, Philip M., and Saveros Pou (1980±81). A Lexicon on Khmer Morphology. (MonKhmer Studies 9±10.) Honolulu: University of Hawai`i Press. Jespersen, Otto (1922). Language: its Nature, Development and Origin. London: George Allen & Unwin, Ltd. Kieckers, Ernst (1931). Die SprachstaÈmme der Erde, mit einer anzahl grammatischer Skizzen. Heidelberg: Carl Winter. Konow, Sten (1906). MundaÅ & Dravidian Languages. (Linguistic Survey of India IV.) Calcutta: ÇÇ O½ce of the Superintendent of Government Printing. (Reprinted 1973, Delhi: Motilal Banarsidass.) Krishnamurti, Bhadriraju (2003). The Dravidian Languages. (Cambridge Language Surveys.) Cambridge: Cambridge University Press.
34
Patricia Donegan and David Stampe
Kuiper, F. B. J. (1948). Proto-Munda Words in Sanskrit. (Verhandeling der Koninklijke Nederlandsche Akademie van Wetenschappen, Afd. Letterkunde, N. R. 51.3.) Amsterdam. Ð (1965). Consonant variation in Munda. In Milner and Henderson, 1.54±87. Ð (1966). The genesis of a linguistic area. Indo-Iranian Journal 10.81±102. Lashley, Karl S. (1951). The problem of serial order in behavior. In Cerebral Mechanisms in Behavior, L. A. Je¨ress (ed.), 112±136. Reprinted in The Neuropsychology of Lashley: Selected Papers, F. A. Beach, D. O. Hebb, C. T. Morgan, and H. W. Morgan (eds.), 506± 528. New York: McGraw-Hill, 1960. Lehiste, Ilse (1970). Suprasegmentals. Cambridge, MA: MIT Press. Ð (1971). Temporal organization of spoken language. In Form and Substance: Phonetic and Linguistic Papers Presented to Eli Fischer-Jùrgensen, L. L. Hammerich, Roman Jakobson, and Eberhard Zwirner (eds.), 159±169. Copenhagen: Akademisk Forlag. Ð (1977). The timing of utterances and linguistic boundaries. Journal of the Acoustical Society of America 51.2018±2024. Lehmann, Winfred P. (ed.) (1967). A Reader in Nineteenth Century Historical Indo-European Linguistics. (Indiana University Studies in the History and Theory of Linguistics.) Bloomington: Indiana University. Ð (1973). A structural principle of language and its implications. Language 49.47±66. Ð (1974). Proto-Indo-European Syntax. Austin: University of Texas Press. Li, Charles N. (ed.) (1975). Word Order and Word Order Change. Austin: University of Texas Press. Longuet-Higgins, H. Christopher, and C. S. Lee (1984). The rhythmic interpretation of monophonic music. Music Perception 1.4.424±441. Reprinted in Longuet-Higgins, Mental Processes: Studies in Cognitive Science. (Explorations in Cognitive Science 1.) Cambridge, MA: MIT Press, 1987, 150±187. McCawley, James D. (1978). Where you can shove in®xes. In Bell and Hooper, 213±221. Majumdar, Paresh Chandra (1970). A Historical Phonology of OriyaÅ. Calcutta: Sanskrit Ç College. Masica, Colin (1976). De®ning a Linguistic Area. Chicago: University of Chicago Press. Ð (1991). The Indo-Aryan Languages. (Cambridge Language Surveys.) Cambridge: Cambridge University Press. Maspero, Georges (1915). Grammaire de la langue KhmeÁre (Cambodgien). Paris: EÂcole francËais d'extreÁme-Orient. Matiso¨, James A. (1973). Tonogenesis in Southeast Asia. In Hyman, pp. 71±95. Meillet, Antoine, and Marcel Samuel RaphaeÈl Cohen (1924). Les langues du monde. Paris: Centre national de la recherche scienti®que. Milner, George B., and EugeÂnie J. A. Henderson (1965). Indo-Paci®c Linguistic Studies. 2 vols. Amsterdam: North-Holland Publishing Company. Mohanan, Karuvannur P. (1986). The Theory of Lexical Phonology. Dordrecht: Reidel. Munda, Ram Dayal (1969). Proto-Kherwarian phonology. M.A. thesis, Linguistics, University of Chicago. Nichols, Johanna (1986). Head-marking and dependent-marking grammar. Language 62: 56± 119. Ð (1992). Linguistic Diversity in Space and Time. Chicago: University of Chicago Press. Parkin, Robert (1991). A Guide to Austroasiatic Speakers and their Languages. (Oceanic Linguistics Special Publications 23.) Honolulu: University of Hawai`i Press. Pinnow, Heinz-JuÈrgen (1959). Versuch einer historischen Lautlehre der Kharia-Sprache. Wiesbaden: Harrassowitz. È ber den Ursprung der voneinander abweichenden Strukturen der Munda- und Ð (1960). U Khmer-Nikobar-Sprachen. Indo-Iranian Journal 4.81±103. Ð (1963). The position of the Munda languages within the Austroasiatic language family. In Linguistic Comparison in South East Asia and the Paci®c, Harry L. Shorto (ed.), 140±52.
Rhythm and the Synthetic Drift of Munda
35
(Collected Papers in Oriental and African Studies.) London: School of Oriental and African Studies, University of London. Ð (1965). Personal pronouns in the Austroasiatic languages: a historical study. In Milner and Henderson, 2.3±42. Ð (1966). A comparative study of the verb in the Munda languages. In Zide 1966a, 96±193. Ð (1979a). Re¯ections on the history of the Khmer phonemic system. Mon-Khmer Studies 8.103±130. Ð (1979b). Remarks on the structure of the Khmer syllable and word. Mon-Khmer Studies 8.131±138. Plank, Frans (1998). The co-variation of phonology with morphology and syntax: a hopeful history. Linguistic Typology 2.195±230. Pou, Saveros (1992). Dictionnaire vieux khmer±francËais±anglais (Old Khmer±French±English dictionary). Paris: Centre de Documentation et de Recherche sur la Civilisation KhmeÁre. Ramamurti, G. V. (1931). A Manual of the So:ra: (or Savara) Language. Madras: Government Press. Reprinted 1986, Delhi: Asian Educational Services. Ð (1933). English±Sora Dictionary. Madras: Government Press. Ð (1938). Sora±English Dictionary. Madras: Government Press. Reprinted 1986, Delhi: Mittal Publications. Ramanujan, A. K., and Colin Masica (1969). Toward a phonological typology of the Indian linguistic area. In Current Trends in Linguistics, Thomas A. Sebeok (ed.), 5.543±577. The Hague: Mouton. Romaine, Suzanne (1988). Pidgin and Creole Languages. London & New York: Longman. Sak-Humphry, Chhany (1996). Khmer nouns and noun phrases: a dependency grammar analysis. Ph.D. dissertation, University of Hawai`i at MaÅnoa. Sapir, Edward (1921). Language. New York: Harcourt, Brace, and Company. È ber die Sprache und Weisheit der Indier: Ein Beitrag zur BeSchlegel, Friedrich von (1808). U gruÈndung der Alterthumskunde. Heidelberg: Mohr und Zimmer. (Selections in English translation in Lehmann 1967, 21±28.) Schmidt, Wilhelm (1906). Die Mon-Khmer-VoÈlker: ein Bindegleid zwischen VoÈlkern Zentralasiens und Austronesiens. (Archiv fuÈr Anthropologie, Neue Folge 5: 59±233.) Braunschweig: Vieweg und Sohn. Ð (1926). Die Sprachfamilien und Sprachenkreise der Erde. (Kulturgeschichtliche Bibliothek, 1 Reihe, Etymologische Bibliothek 5), 2 vols. Heidelberg: Carl Winter. Shorto, Harry L. (1960). Word and syllable patterns in Palaung. Bulletin of the School for Oriental and African Studies 23.544±577. Ð (1971). A Dictionary of the Mon Inscriptions, from the Sixth to the Sixteenth Centuries. (London Oriental Series 2.) London, Oxford University Press. Ð (1976). The vocalism of Proto-Mon-Khmer. In Jenner, Thompson, and Starosta, 2: 1041± 1067. Stampe, David (1965±66). Recent work in Munda linguistics I±IV. International Journal of American Linguistics 31.332±41 (1965), 32.74±80, 164±8, 370±87 (1966). Ð (1973). Speech as music, music as speech. Paper read to the Annual Meeting of the Linguistic Society of America, San Diego. Ð (1978). Proto-Central Munda phonology. Second International Conference on Austroasiatic Linguistics, Mysore. Starosta, Stanley (1967). Sora syntax: a generative approach to a Munda language. Ph.D. thesis, University of Wisconsin, Madison. Ð (1997). Sora noun in¯ection. In Abbi, 263±305. Steever, Sanford H. (1993). Analysis to Synthesis: The Development of Complex Verb Morphology in the Dravidian Languages. New York: Oxford University Press. Ð (ed.) (1998). The Dravidian Languages. (Routledge Language Family Descriptions.) London: Routledge.
36
Patricia Donegan and David Stampe
Tannen, Deborah (1994). Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse. Cambridge: Cambridge University Press. Thurgood, Graham (1999). From Ancient Cham to Modern Dialects: Two Thousand Years of Language Contact and Change. (Oceanic Linguistics Special Publication No. 28.) Honolulu: University of Hawai`i Press. Turner, Ralph L. (1966). A Comparative Dictionary of the Indo-Aryan Languages. Index, volume 2, 1969, compiled by Dorothy Rivers Turner. Oxford: Oxford University Press. Twain, Mark (1880). The awful German language. Appendix D of A Tramp Abroad. Hartford, Connecticut: American Publishing Company. Vennemann. Theo (1975). An explanation of drift. In Li, 269±305. Vitebsky, Piers (1993). Dialogues with the Dead: The Discussion of Mortality among the Sora of Eastern India. (Cambridge Studies in Social and Cultural Anthropology.) Cambridge, UK: Cambridge University Press. Yngve, Victor (1960). A model and an hypothesis for language structure. Proceedings of the American Philosophical Society 104: 444±466. Zide, Arlene R. K. (1982). A reconstruction of proto-Sora-Juray-Gorum phonology. Ph.D. thesis, University of Chicago. Zide, Norman H. (1965). Gutob-Remo vocalism and glottalised vowels in Proto-Munda. In Milner and Henderson, 1.43±53. Ð (ed.) (1966a). Studies in Comparative Austroasiatic Linguistics. The Hague: Mouton. Ð (1966b). Korku low tone and the Proto-Korku-Kherwarian vowel system. In Zide 1966a, 1.214±229. Ð (1969). Munda and non-Munda Austroasiatic languages. In Current Trends in Linguistics, Thomas A. Sebeok (ed.), 5.411±430. The Hague: Mouton. Zide, Norman H., and Gregory D. S. Anderson (2001). The Proto-Munda verb: some connections with Mon-Khmer. In Yearbook of South Asian Linguistics, B. Rao and K. V. Subbarao (eds.), 517±540. Delhi: Sage Press.
The Possible and the Impossible in Bengali Word Formation: Some Problems in Nominalization UDAYA NARAYANA SINGH and SUCHITA SINGH
This study makes a modest attempt to consider certain productive word formation rules in Bengali and tries to determine why certain potential but nonexisting words can gain acceptability among native speakers whereas certain others are not at all acceptable. An attempt will be made to give a psychosemantic interpretation to this problem which alone can explain what is possible and what is impossible in Bengali. In particular, this study is an attempt to understand the properties of noun formation rules by considering the entire verbal stock of Bengali in relation to certain formative a½xes. 1. Introduction This paper is concerned with the internal structure of Bengali words. In particular, it deals with certain possible nominals and adjectivals in Bengali derivable from the basic verbs. The productivity of certain derivational suf®xes in Bengali (such as On and naa changing verbs to nouns and that of (u)ni deriving adjectives from verbs) will be tested here. 2. Methodology In order to exhaust the nominal and other constructions under consideration, a large lexical database was ®rst created using dBase III Plus by inputting 6,251 verbal entries obtained from one of the standard dictionaries in BengaliÐthe Sahitya Samsad Dictionary. This included a large number of conjunct verbs, causative forms with some semantic di¨erences in shades of meaning, and the non-causative -no verbs, out of which an exhaustive set of 219 verbs was considered for this exercise in word generation to see if there was a pattern. The 250-odd non-causative roots with -no, which are listed in the appendix, were kept out of our list, because they would not participate in a construction involving -on. The data described here also includes possible but unavailable forms, which have been shown within parentheses with a question mark preceding
38
Udaya Narayana Singh and Suchita Singh
them. These are, indeed, forms which may or may not have actual realization but which must have been in conformity with the phonological, syntactic and semantic pattern of the language, which is evident from the fact that they pass the test of informant elicitation as possible (but unavailable) entries. 3. Theoretical Background Generative grammarians have only recently reached some kind of consensus regarding the properties of word formation processes and rules. But the issue as to where exactly this process of word formation takes place still remains unresolved. Whereas Kiparsky (1982) and Mohanan (1982) advocate that it occurs at the phonological level, Fabb (1984) and many others claim that it takes place in syntax. The argument seems to be that syntax not only has a role in the in¯ection of a given word but that it also plays a crucial role in word formation, particularly in the case of compoundings. Fabb (1984) postulates a set of diagnostics to determine whether word formation is a lexical or a syntactic phenomenon. According to him, for instance, a word formation process is syntactic if it (i) is productive, (ii) does not violate the Projection Principle, and (iii) is semantically transparent. Arono¨ (1976), in contrast to Fabb's claim for the syntactic role in the productive process, argues that several factors combine to decide the productivity of a process. In this study, however, we consider a few very productive processes, which operate at the level of lexis and as such are not syntactic. Even now, i.e. even after the introduction of many new forms in the last few decades, they show many gaps, although the forms generated and further generable are all semantically transparent. Although The Origin and Development of the Bengali Language (ODBL), the voluminous work of Chatterji (1926), did devote a few pages to the exempli®cation of use of these as well as a few other verb-to-noun formations (ODBL, Vol II: 651± 659), we are unaware of any literature on this particular problem where such word formation processes are to be checked. The productivity of a word formation rule can be judged on the basis of the frequency and acceptability of the words which are produced by a particular word formation rule, and are listed in the lexicon. The use of a new word depends upon the kind of productivity or the potentiality of the word formation rule, as is evidenced by the native speakers attitude in using new words (cf. Dowty 1979 and Dressler 1982). The general trend is the ready acceptance of the words produced by a productive word formation rule as against the reluctance to accept the words produced by the less productive ones. Arono¨ (1978) advocates testing the synchronic reality of the productivity experimentallyÐby using the technique of cognitive psychology (but
The Possible and the Impossible in Bengali Word Formation
39
by designing the experiment linguistically). Singh and Basu (1987) report on an experiment of a similar kind testing the social bases of productivity in Bengali. Arono¨ (1978) hopes to discover what he calls the `morphological competence' of the native speakers of a given language, which is also what is tested here. 4. Aims and Objectives This study makes a modest attempt to consider certain productive word formation rules in Bengali and tries to determine which potential but nonexisting words could gain acceptability among native speakers whereas certain other constructions are not at all acceptable. An attempt is made to give semantic and structural classi®cation of the basic verbs of Bengali, which show di¨erent patternsÐa classi®cation which could, at a later date, be tested with the help of syntactic and semantic features. It is hoped that the tentative interpretation of this problem explains what is possible as opposed to what is impossible in Bengali constructions in the given context. In particular, this study is an attempt to understand the properties of noun formation rules by considering the entire verbal stock of Bengali in relation to certain formative a½xes already mentioned. 5. Structure of Data Let us consider the following data (on marks a nominalization with the interpretation: `of the state of', naa has di¨erent types of meaning load, uni is usually used for an individual possessing the quality of the activity denoted by the verb, and the -i at the end indicates femininity in many cases) Data Structure Head
Nominal I On
Nominal II naa
Adjectival (u)ni
1.
aaMkaa
aaMkon
*aaMknaa
2. 3.
aaMcaa aaMTaa
Ð aaMTon
4. 5.
aakulaa *aachaa @ thaakaa
Ð thaakon
Ð *aaMTnaa/ aaMTkaa Ð *thaaknaa
*aaMk(u)ni aaMkiye Ð aaMTuni Ð thaakuni
40
Udaya Narayana Singh and Suchita Singh
6. 7. 8. 9. 10. 11. 12.
16. 17. 18. 19. 20. 21.
aanaa aaSaa ucchraa ubaa upjaa oThaa opRaa @ upRaa oRaa olaa kOhaa @ kOwaa kaaMdaa kaaMpaa kaacaa kaaTaa kaaRaa kuMthaa
22. 23. 24.
kuTaa kuhOraa kenaa
25. 26. 27.
koMdaa khaaoaa khaaTaa
28. 29. 30.
khElaa khoMjaa khodaa
31. 32. 33.
kholaa gOlaa *gOhaa
koMdon khaaon *khaaTon (dial) khElon *khoMjon khodon/ khodaai kholon gOlon gOhon
34. 35. 36. 37. 38. 39. 40.
gaaRaa gaadaa gaalaa gaaoaa gelaa golaa ghOSaa
gaaRon Ð gaalon gaahon gelon Ð ghOSon
13. 14. 15.
Ð Ð Ð Ð Ð oThon -(upTOn)
Ð Ð Ð Ð Ð Ð Ð
Ð Ð Ð Ð Ð (uThti) Ð
oRon olon kOhon/kaahon
oRnaa *olnaa *kOhnaa
uRni @ oRoni uluni *kouni
kaaMdon kaaMpon Ð *kaaTon kaaRaan kuMtaan/ koMkoTon kuhoron ?*kenon
kaannaa *kaaMpnaa Ð kaaTnaa *kaaRnaa *kuMthnaa
kaaMduni kaampuni Ð [?kaaTuni] *kaaR(u)ni *kuMthni
kuTnaa *kuhornaa ?kinnaa/ kenaa *kuMdnaa khaanaa *khaaTnaa
KuTni/kuRuni *kuhorni *kinuni
khElnaa *khoMjnaa *khodnaa
[?kheluni] [?khoMjuni] *khoduni
kholnaa *gOlnaa gOhnaa/ gOynaa *gaaRnaa Ð *gaalnaa gaaonaa *gelnaa Ð *ghOSnaa
khuluni [*goluni] *gOhni
[kudaali] *khaauni khaaTuni
*gaaRoni/-uni Ð [*gaaluni] gaauni *geluni Ð ghiSni
The Possible and the Impossible in Bengali Word Formation 41. 42.
gheraa ghoraa
*gherOn ghoron
*ghirnaa *ghornaa
43. 44. 45. 46.
cOTaa cORaa cOlaa caaoaa
Ð cORon cOlon caahon
Ð *cORnaa Ð *caahnaa
47. 48.
caaMchaa caakhaa
*caaMchon *caakhon
49. 50.
caaTaa caapaa
caaTon caapon
51. 52. 53. 54. 55. 56. 57. 58. 59. 60.
ciraa cukaa cumaa cenaa cepaa coSaa cetaa chORaa chOlaa chaaoaa
61. 62. 63.
chaaMkaa @ chaanaa chaaMTaa chaaMdaa
ceron Ð Ð Ð Ð coSon cetOn Ð chOlon chaaon (chaadon) *chaaMkon
*caaMchnaa *chaakhnaa (Taaknaa) *caaTnaa *caapnaa/ (caabnaa) *cirnaa Ð Ð Ð Ð *coSnaa cetnaa Ð chOl(o)naa *chaaonaa
64.
?ghirni ghurni/ ghuruni Ð *coR(u)ni [?coluni] caahoni/ caauni [?caaMchni] [?chaakhoni] caaTni [?caapuni] ciruni Ð Ð Ð Ð cuSni/cuSi *cet(u)ni Ð *chOluni chaauni
chaaMknaa
chaaMkni
chaaMTon chaaMdon
*chaaMTnaa chaaMdnaa
chaaRaa
chaaRon
65. 66. 67.
chaapaa chenRaa chuTaa
Ð Ð choTon
*chaaRnaa (chaaRaan) Ð Ð *chuTnaa
chaaMTuni cheni/ *chaaMdni *chaaRuni
68.
cholaa
cholon
69.
chEMcaa
70.
jOmaa
*cEMcaaon (cEMcaano) Ð
*cholnaa (cf Maithili cholni) *cEMcnaa Ð
41
Ð Ð chuTni/ chuTki [?chulni] cEMcaani Ð
42
Udaya Narayana Singh and Suchita Singh
71. 72.
jOlaa jaagaa
73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96.
jOlon jaagon/ (jaagorOn) jaaon (dial) jaacon jojhon Ð jhOron jhaaRon jhoMTon jholon/jhulOn TOlon Tepon Thokon (dial) *DOron *DhOlon Dhaakon Dhaalon *Dhokon (dial) tOron Ð tolon dOmon dOlon dOhon Ð dohon
102. 103. 104. 105. 106. 107. 108.
den dEkhon dolon dhOron *dhOSon/ dhOS dhaabon Ð Ð Ð nORon Ð naacon
jaaoaa jaacaa jojhaa joTaa jhOra jhaaRaa jhoMTaa jholaa TOla Tepa ThOka Doraa DhOlaa Dhaakaa Dhaalaa Dhokaa tOraa titaa tolaa dOmaa dOlaa dOhaa daagaa duha/ do(h) aa 97. deoaa 98. dEkha 99. dolaa 100. dhOraa 101. dhOSa dhaaoaa dhoMkaa dhunaa dhoaa nORaa noaa naacaa
*jOlnaa *jaagnaa
joluni [?jaagni]
*jaaonaa jaacnaa *jojhanaa Ð jhOrnaa *jhaaRnaa *jhoMTnaa *jholnaa *TOlnaa *Tepnaa Thoknaa *DOrnaa *DhOlnaa Dhaaknaa [?Dhaalnaa] *Dhoknaa *tOrnaa Ð *tolnaa *dOmnaa *dOlnaa *dOhnaa Ð *duhnaa
*jaauni [?jaacune] *jujhuni Ð *jhor(u)ni jhaaRuni ?jhuMT(n)i jhuluni ?TOluni Tipuni Thukuni [?Doruni] Dhuluni Dhaakni [?Dhaaluni] [?Dhuk(u)ni] tOroni Ð tuluni *domni *doluni ?dohuni Ð ?du(h)ini
denaa ?dEkhnaa dolnaa dhOrnaa *dhOSnaa
*deni (daani) *dekhni duluni [?dhoruni] *dhoSni
*dhaaonaa Ð Ð Ð *nORnaa Ð ?naacnaa
?dhaauni Ð Ð Ð ?noRune ?naacuni/ [naacune]
The Possible and the Impossible in Bengali Word Formation 109. 110. 111. 112. 113.
naaRaa naamaa niRaano nebhaa newaa/ (*lewaa) 114. pOcaa 115. pORaa 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128.
pOrokhaa paaoaa paakaa paaTaa paataa paadaa paaRaa paaraa piMja @ peMjaa piTaa @ peTaa puchaa puraa peTaa
*naaRon (dial) Ð Ð Ð len (nEon (dial)) pOcon pORon/ pOThon *pOrokhOn paaon *paakon (dial) paaTon paaton *paadon paaRon *paaron Ð peTon (gORon-) Ð purOn peTon
*naaRnaa Ð Ð Ð lenaa
?naaRune Ð Ð Ð *leni
*pOcnaa *pORnaa (pORtaa) *pOrokhnaa paaonaa *paaknaa paaTnaa *paatnaa *paadnaa *paaRnaa *paarnaa Ð
*pocuni poRoni (ODBL658) *pOrokhni *paauni *paakuni paaToni [?paatuni] [?paaduni] *paaRoni *paaraani Ð
*piTnaa
piTuni
Ð *purnaa *piTnaa/ *peTnaa *pi-/eSnaa *puRnaa *pu-/poSnaa
Ð *purni piTuni
129. peSaa 130. poRaa 131. poSaa
peSon *poRon poSon
132. 133. 134. 135. 136. 137. 138.
*phOlnaa Ð Ð *phaaTnaa Ð Ð Ð
139. phoMSaa 140. phOTaa
phOlon -(phaaMd) -(phaaMS) phaaTon/-ol Ð -(pherat) phoRon/ phoMRon Ð phoTon
141. pholaa 142. phElaa
*pholon phElon
*pholnaa phElnaa
phOlaa phaaMdaa phaaMSaa phaaTaa phaaRaa pheraa pho(M)Raa
43
Ð *phoTnaa
[?piSni] *poRoni [?puSni/ puSSi] *pholuni Ð Ð *phaaTuni Ð Ð *phuRni Ð *phuTuni/ -aani [?phuluni] [?pheluni]
44
Udaya Narayana Singh and Suchita Singh
143. 144. 145. 146. 147. 148.
160. 161. 162.
bOkaa bOkha bOma bOlaa bOSaa bOwaa/ bOhaa baaMkaa baaMcaa baaTaa baaMTaa baaMdhaa baachaa baajaa baaoaa/ baahaa baaRaa binaa biMdhaa/ beMdhaa boMjaa bojhaa bonaa
163. 164. 165. 166. 167. 168. 169. 170.
bEcaa bEla bhOraa bhaaMjaa bhaagaa bhaangaa bhaabaa bhaaSaa
149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159.
171. bhiRaa/ bheRaa 172. bhogaa 173. bholaa 174. bhejaa 175. mOraa 176. mOlaa 177. mOwaa
bOkon Ð Ð bOlon *bOSon (dial) bOhon
bOknaa Ð Ð *bOlnaa *bOSnaa *bOhnaa
bokuni Ð Ð *bolni *boSuni bouni
Ð baaMcon *baaTon Ð baaMdhon Ð baajon baahon
Ð *baaMcnaa baaTnaa Ð baaMdhnaa Ð baajnaa ?baahnaa (baaynaa) *baaRnaa *binna Ð
Ð *baaMcuni baaTni Ð baaMdhuni [?baachuni] *baajuni baauni (e.g. aauni-baauni) ?baaRni binuni Ð
*boMjon *bojhon bunon ( bunoT ) *bEcon bElon bhOron *bhaaMjon Ð bhaangon bhaabon *bhaaSon (bhaaSiye) bhiRon
*boMjnaa *bujhnaa *bunnaa
?buMjuni ?bujhni ?bununi
*bEcnaa bElnaa [?bhOrnaa] bhaaMjnaa Ð bhaangnaa bhaabnaa *bhaaSnaa
becuni beluni [?bhoruni] [?bhaaMjuni] Ð bhaangni *bhaabuni *bhaaSwe
*bhiRnaa
*bhiRuni
Ð [?bhulOn] *bhejon mOron mOlon *mOwon
Ð *bhulnaa *bhijnaa *mOrnaa *mOlnaa *mOwnaa
Ð [?bhuluni] [?bhijuni] *moruni [?moluni] mount (ODBL658)
baaRon *binon Ð
The Possible and the Impossible in Bengali Word Formation 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209.
maakhaa maagaa maajaa maaRaa maataa maanaa maapaa maaraa miTaa/ meTaa milaalmelaa meSaa muRaa/ moRaa mutaa/motaa raaMdhaa raakhaa rocaa rudhaa/ rodhaa roaa lOwaa newaa (See earlier) lORaa laagaa lekhaa luTaa @ loTaa lepaa lekhaa SOraa
SOhaa SaaMTaa Saajaa Saadhaa Saanaa Saaraa SiMcaa SeMcaa 210. Sekhaa 211. Secaa
45
maakhon maagon maajon *maaRon maaton *maanon *(maap) maaron Ð
maakhnaa maagnaa *maajnaa *maaRnaa *maatnaa *maannaa Ð *maarnaa Ð
*maakh(o)ni maagni *maajuni *maaRuni ?maatuni *manuni Ð *maar(u)ni Ð
milOn Ð *moRon
*milnaa Ð *muRnaa
miloni Ð muRaani
moton raaMdhon raakhon Ð rodhon
*mutnaa raannaa [?raakhnaa] Ð *rodhnaa
[?mutini] raaMdhuni [?raakhoni] Ð ?rodhuni
Ð
Ð
Ð
*lORon laagon lekhOn Ð
*lORnaa laagnaa *lekhnaa Ð
[?loRuni] *laagni/logni lekhoni Ð
lepon lekhOn *SOron (sOraa) SOhon Ð Saajon Saadhon Ð Ð SincOn
*lepnaa [?likhnaa] Ð
*lepuni lekhoni Ð
*SOhnaa Ð *Saajnaa *Saadhnaa Ð Ð *SiMcnaa
[?SOhoni] Ð Saajuni [?Saadhuni] Ð Ð *SiMcuni
-(SikkhOn) Ð
Ð Ð
Ð Ð
46
Udaya Narayana Singh and Suchita Singh
212. 213. 214. 215. 216. 217.
SEMkaa SoMkaa Sonaa hOwaa hOThaa haaMkaa
218. haaMcaa 219. haaSaa
Ð Ð Shunon *hOon (dial) Ð *haaMkan (haaMk/ haaMkaa) Ð Ð
Ð Ð *Sunnaa *hOonaa Ð ?haaMkna
Ð Ð Sunaani *houni Ð [?haaMkni]
Ð Ð
Ð [?haaSni, esp. in slang]
6. Observations If we take a closer look at the table presented above, we notice that there are indeed a large number of basic verbs that do not participate in this particular word form generation in Standard Chalit Bengali (SCB). These include about 99 out of the 219, i.e. 45.21% of the basic verbs already mentioned above. These stems include the following (where we have deleted the in®nitive marking, i.e. -aa): Level O 2. 6. 9. 18. 35. 43. 52. 55. 66. 76. 90. 104 109. 112. 121. 126. 134. 138. 141.
aaMc aan ub kaac gaad cOT cuk cep cheMR joT tit dhunaa naaR nebh paad puch phaaMS pho(M)R phol
4. 7. 10. 24. 39. 47. 53. 58. 70. 84. 95. 105. 110. 116. 123. 130. 136. 139. 144.
aakul aaS up(a)j ken gol caaMch cum chOR jOm DOr daag dhoaa naam pOrokh paar poR phaaR phoMS bOkh
5. 8. 12. 29. 41. 48. 54. 65. 73. 88. 103. 107. 111. 118. 124. 133. 137. 140. 145.
*aach ucch(a)r op(a)R/up(a)R khoMj gher caakh cen chaap jaa Dhok dhoMkaa no niR paak piMj/peMj phaaMd pher phOT bOn
The Possible and the Impossible in Bengali Word Formation 147. 154. 161. 166. 172. 177. 182. 186. 194. 199. 204. 211. 214. 217.
bOS baach bojh bhaaMj bhog mOw maat miT/meT rudh/rodh luT/loT SaaMT Sec Son haaMk
149. 159. 162. 167. 173. 178. 183. 188. 195. 201. 208. 212. 215. 218.
baaMk biMdh/beMdh bon bhaag bhol maakh maan meS ro lekh Saar SEMk hOw haaMc
152. 160. 165. 170. 174. 181. 184. 193. 196. 202. 210. 213. 216. 219.
47
baaMT boMj bhOr bhaaS bhej maaR maap roc lOR SOr Sekh SoMk hOTh haaS
As these stem-endings or their meaning/interpretation would show, there is no possibility of grouping them either in terms of semantic or morphophonological categories, which could tell us why they behave like the -no in®nitives (See appendix) that do not take -On, etc. This leads us to believe that in the matter of spread of these forms, it may be bene®cial to categorize or place these verbs on a percolation scale of 7-points to see the extent of their spread in these formations. Whereas 53% of the verbs above stand at the bottom of the percolation scale, only 9% or 19 verbs (out of 219) in all accept all these su½xes. These 19 verbs are: Level 7 RooT oR kaaMd kuT khol gaa ghOS chaaMd ThOk Dhaak dol paaT bOk baaMdh baa/baah
NOM I oRon kaaMdon koTon kholon gaahon ghOSon chaaMdon Thokon (dial) Dhaakon dolon paaTon bOkon baaMdhon baahon
NOM II oRnaa kaannaa kuTnaa kholnaa gaaonaa *ghOSnaa chaaMdnaa Thoknaa Dhaaknaa dolnaa paaTnaa bOknaa baaMdhnaa ?baahnaa
ADJ uRni @ oRoni kaaMduni kuTni/kuRuni khuluni gaauni ghiSni cheni/*chaaMdni Thukuni Dhaakni duluni paaToni bokuni baaMdhuni baauni
48
Udaya Narayana Singh and Suchita Singh
bEl bhaang maag raaMdh laag
bElon bhaangon maagon raaMdhon laagon
bElnaa bhaangnaa maagnaa raannaa laagnaa
beluni bhaangni maagni raaMdhuni *laagni/logni
Once again, we can apparently tell nothing by looking at the endings or the syllable structure, etc. Interestingly, however, there are a large number of verbs, 50 out of 219, i.e. 23%, which at least accept the -On marking, but have no corresponding forms under the other two sets. These are the following: Level 1 1 20 25 34 44 64 75 92 96 102 114 127 132 150 175 185 200 209
aaMkaa kaaRaa KoMdaa gaaRaa cORaa chaaRaa jojhaa dOmaa duha/do(h)aa dhaaoaa pOcaa puraa phOlaa baaMcaa mOraa maaraa/ lepaa siMcaa
11 21 30 36 45 68 81 93 98 106 120 129 135 157 176 190 203
oThaa kuMthaa khodaa gaal cOlaa cholaa TOla dOlaa dEkha nORaa paataa peSaa phaaTaa baaRaa mOlaa mutaa/motaa Sohaa
15 23 32 38 50 72 87 94 101 108 122 131 146 171 180 192 206
kOhaa/kOwaa kuhOraa gOlaa gelaa caapaa jaagaa Dhaalaa dOhaa dhOSa naacaa paaRaa poSaa bOlaa bhiRaa maajaa raakhaa Saadhaa
Although a number of these are tadbhava forms with instances of classical borrowing of their corresponding Sanskritic nominal or -On-forms, such as onkon, kuhorOn, gOlon, jaagorOn, duhon, nOrton, pOcon or purOn, this does not explain many other verbs which are not historically traceable to any Sanskritic root with similar nominal forms. While looking for the reasons, and also for the explanation behind recent emergence of many of these forms, we found that -On is a very productive su½x for nominalization in most dialects spoken in the erstwhile East Bengal, or the present Bangladesh. If we were to write a grammar of word formation of the stan-
The Possible and the Impossible in Bengali Word Formation
49
dard eastern spoken dialect of Dhaka-Mymensingh, a large number of the 53% non-participating verbs will also generate corresponding nominal forms. There will be very few verb forms left in the ®rst block of 116 verbs then. We could now probably make a guess as to why these forms have affected so many verbs in the above group of 49 verbs. There has been a large exodus of people from the east to the locus in and around Calcutta in the late forties, and many features of Bengali phonology show changes after this period of time, which is clear if we compare Chatterji's work on phonetics or his ODBL with Ferguson and Chowdhury's (1960) stand on Bengali phonology in the early ®fties, and then bring them into contrast with the recent studies by Sarkar (1976) and Paul (1985) or Dan (1998). There has to be a fall out in the area of morphology, and we guess this has been one such area. This brings us to the third point, namely that although the number 50 out of 219, or 23%, does not look very big, it has to be viewed in the context of there being more verbsÐat least more which need to be placed still higher on the percolation scale. These include the following groups which have to be placed higher than the level 1 verbs (although between levels 2 and 3, ordering is not so important): Level 2 (acceptance of the su½xes -On and -naa) 26 28 33 57 59 74 77 97 100 113 117 142 155 169 178
khaaoaa khElaa *gOhaa cetaa chOlaa jaacaa jhOra deoaa dhOraa newaa/(*lewaa) paaoaa phElaa baajaa bhaabaa maakhaa
khaaon khElon gOhon cetOn chOlon jaacon jhOron den dhOron len (nEon (dial)) paaon phElon baajon bhaabon maakhon
khaanaa khElnaa gOhnaa/gOynaa cetnaa chOl(o)naa jaacnaa jhOrnaa denaa dhOrnaa lenaa paaonaa phElnaa baajnaa bhaabnaa maakhnaa
*khaauni [?kheluni] *gOhni *cet(u)ni *chOluni [?jaacune] *jhor(u)ni *deni (daani) [?dhoruni] *leni *paauni [?pheluni] *baajuni *bhaabuni *maakh(o)ni
Level 3 (acceptance of -On and -uni forms only) 3 14
aaMTaa olaa
aaMTon olon
*aaMTnaa *olnaa
aaMTuni uluni
50
Udaya Narayana Singh and Suchita Singh
17 42 46 49 51 56 60 62 67 71 78 79 80 82 89 91 115
kaaMpaa ghoraa caaoaa caaToa ciraa coSaa chaaoaa chaaMTaa chuTaa jOlaa jhaaRaa jhoMTaa jholaa Tepa tOraa tolaa pORaa
kaaMpon ghoron caahon caaTon ceron coSon chaaon chaaMTon choTon jOlon jhaaRon jhoMTon jholon/jhulOn Tepon tOron tolon pORon/pOThon
*kaaMpnaa *ghornaa *caahnaa *caaTwa *cirnaa *coSnaa *chaaonaa *chaaMTnaa *chuTnaa *jOlnaa *jhaaRnaa *jhoMTnaa *jholnaa *Tepnaa *tOrnaa *tolnaa *pORnaa
125 148 187 198 205 207
piTaa/peTaa bOwaa/bOhaa milaa/melaa lekhaa Saajaa Saanaa
peTon (gORon-) bOhon milOn lekhOn saajon Ð
*piTnaa *bOhnaa *milnaa *lekhnaa *saajnaa Ð
kaaMpuni ghurni/ghumni caahoni/caauni caaTni ciruni cuSni/cuSi chaauni chaaMTuni chuTni/chuTki joluni jhaaRuni ?jhuMT(n)i jhuluni Tipuni tOroni tuluni poRoni (ODBL 658) piTuni bouni miloni lekhoni saajuni Ð
If we compare these 15 level 2 verbs above and the 25 level 3 verbs (a total of 4) to the initial list of 50, we see that gradually these forms are making inroads into all kinds of verbs, and consequently the Bengali lexicon is enriched more and more. Thus approximately 18% of the basic verbs have been drifting towards the highest level over the years. We could think of two more layers of forms for the other 3.5% verbs, assigned to levels 4 (7 verbs) and 5 (1 verb only) as shown below: Level 4 (verbs with only -(u)ni forms): 27 69 85 158 163 176 189
khaaTaa chEncaa DhOlaa binaa bEcaa mOlaa muRaa/moRaa
*khaaTon (dial) *cEMcaaon *Dholon *binon *bEcon mOlon *moRon
*khaaTnaa *cEMcnaa *Dholnaa *binna *bEcnaa *mOlnaa *muRnaa
khaaTuni cEMcaani Dhuluni binuni becuni [?moluni] muRaani
The Possible and the Impossible in Bengali Word Formation
51
Level 5 (verbs where only -naa is productive) 19 kaaTaa
*kaaTon
kaaTnaa
[?kaaTuni]
We would expect that all forms generated by adding the other two suf®xes would also have a corresponding -On form in each case. But this is further seen in the problem of the level 6 forms given below, which have both -naa and -(u)ni forms, but no trace of -On anywhere in SCB: Level 6 (verbs with both -naa and -(u)ni forms) 61 chaaMkaa @ chaanaa 151 baaTaa
*chaaMkon *baaTon
chaaMknaa baaTnaa
chaaMkni baaTni
The verbs in levels 4 through 6, although only 10 in number, call for an explanation which is not readily available. It is likely that either the forms do exist in the speech of some SCB speakers although not in the ones who have been our informants, or that there is a counter force blocking the spread of -On to SCB verb forms because of some sociological constraints. These are to be taken only as hints rather than conclusive remarks, because what we demonstrated here was only an interesting way of seeing a morphological change in progress. References Arono¨, Mark (1976). Word formation in generative grammar. Cambridge: Massachusetts Institute of Technology Press. Ð (1978). The relevance of productivity in a synchronic description of word formation. J. Fisiok (ed.), Historical morphology. The Hague: Mouton. Chatterji, Sunit Kumar (1926). The Origin and Development of the Bengali Language, vol. II. Calcutta: Calcutta University Press. Dan, Mina (1998). ``Bangla verb morphology: The actual derivation.'' Indian Linguistics, 59(1± 4): 43±79. Dowty, David R. (1979). Word meaning and Montague grammar. Dordrecht: Reidel. Dressler, Wolfgang (1982). On a polycentristic theory of word formation. Proceedings of 12th International Congress of Linguistics. Fabb, Nigel A. (1984). Syntactic a½xation. Doctoral dissertation, Massachusetts Institute of Technology. Ferguson, C.A. and Munier Choudhury (1960). ``The Phonemes of Bengali.'' Language 36: 22±59. Kiparsky, Paul (1982). Lexical morphology and phonology. I. S. Yang, ed. Linguistics in the morning calm. Seoul: Hanshin. Mohanan, Karuvannur P. (1982). Lexical phonology. Doctoral dissertation, Massachusetts Institute of Technology. Paul, Jayashree (1985). A Concrete Approach to Bengali Phonology. Unpublished Ph.D. Dissertation, Kanpur: Indian Institute of Technology.
52
Udaya Narayana Singh and Suchita Singh
Samsad Bengali Dictionary (1991 edition). Calcutta: Sahitya Samsad. Sarkar, Pabitra (1976). ``The Bengali Verb.'' International Journal of Dravidian Linguistics, 5(2): 274±297. Singh, Udaya N. and S. Basu (1987). Study of word formation processes in Indian languages: some directions in the analysis of creativity. Paper presented at the Seminar cum Workshop on Common Morphological Features in Indian Languages (March 9±13, 1987), Mysore. To appear in International Journal of Dravidian Linguistics.
Appendix The excluded -no in®nitives aaMTaano aagaano uKhRaano uSkhuSaano kOclaano kaamRaano kaaMcaano kodlaano olaano khaapaano khoMcaano khEpaano gonaano guTaano ghOngaano ghaamaano ghumaano cokcOkaano cOTcOTaano cOnmOnaano cOlaano caapaano caalaano cunaano cumRaano cEpTaano connano chORaano chaaMTaano jhaalaano Topkaano
aaTkaano aachRaano uchlaano utraano/utrono kOckOcaano koMkRaano kElaano khOtaano Eraano haablaano khuMTaano gOjraano guMRono gumraano ghaaMTaano ghucaano ghuraano cOTkaano cORaano cOraano caagaano caabkaano cibaano/cibono cuTaano culkaano cElaano copaano chaadaano choMkchoMkaano jhimaano/jhimono TaMpaano
aaglaano uMcaano uSkaano oRaano kaatraano kucono kuono khaaMkhraano khaaRaano khiMcono khoaano gORaano guchaano/guchono gongonaano ghaabRaano ghEMgaano gholaano cOTaano cORbORaano cOlkaano caapRaano caaraano cillaano cupSaano cetaano coTaano chOTphOTaano cORaano jhOlkaano TOnTOnaano TaaTaano
The Possible and the Impossible in Bengali Word Formation Tikaano/Tikono Thokraano taakaano tuRaano daapaano douRaano ghumRaano pichlaano pEMcaano pouMchaano/pouMchono phOSkaano phuSlaano phuraano baakhaano baahuRaano bichaano biiaano bERaano bhaaMRaano bhEngaano mOTkaano mucRaano moTaano rOTaa raangaano lOpTaano/lEpTaano lEngcaano Sekhaano Sudhraano SaaMtraano SaabRaano haaMkRaano haatRaano hiNcRaano/hENcaRano
Topaano ThOkaano taapaano toaano daabRaano dhOmkaano paanaano pichono pEMdaano pEnpEnaano phaaMpaano phukraano phEnaano baanaano bikaano/bikono binaano biSaano betaano bhEpSaano bhEbRaano miaano muSRaano rORaano rOSaano lOTkaano laaphaano lElaano Sijhaano SOmjhaano SaaMtlaano hoRkaano haaTkaano haataano
53
Thikraano DhuSaano taaSaano daaMRaano daabaano dhEstano paalaano piTpiTaano pohaano phOrphOraano phoMpaano phuTaano phoMSphoMsaano balSaano bigRaano biaano buRaano/buRono bhORkaano bhijaano mOckaano mukhaano melaano raangaano raaMdhaa lOtaano lukaano/lukono SaaSaano SiMTkaano SOraano SaapTaano hORbORaano haaMpaano haaraano
Case and the Argument Structure of Tamil* E. ANNAMALAI
Any grammar must deal with the issues of relationship amongst the semantic roles, grammatical functions and case relations of nouns with regard to the predicate, which are respectively Arguments (Agent etc.), Function Units (Subject etc.) and Markers (Case or Pre-/Post-positions). This paper proposes a schema as hypotheses regarding relationships between possible semantic roles of a predicate and its Arguments and between Arguments required by the predicate and Grammatical Functions required by syntax. It then goes on to discuss the issues in the marking of case in nouns in Tamil in relation to their occurrence and to their dependence on speci®c Arguments and Grammatical Functions. It shows that each case represents a bunch of distinguishable semantic features and that not all semantic distinctions mark cases separately. It also takes up some related issues.
1. The Morphological Status of Case Markers Tamil marks case by in¯ection of nouns. The case markers, which are added after nouns, may be bound or free and are called su½xes and postpositions respectively. The case markers may be morphologically simple with zero or one morpheme or complex with more than one morpheme. They are added to a special form of the noun called oblique form. All nouns, however, do not have an oblique form di¨erent from the bare form. When the case marker is zero, there is no di¨erence between the oblique and bare forms except in the case of genitive. With those nouns where there is no di¨erence, the bare form can be said to be notionally the oblique form. That is, oblique and bare forms of nouns are homophonous. The oblique forms have di¨erent morphological forms depending on the phonological shape of the noun, and one morphological form of the oblique can be said to be an empty morph, which is added to the bare noun. The nominative case marker is zero and so the nominative noun is homophonous with the bare noun. The case marker may be absent in the
58
E. Annamalai
accusative and dative (in its allative sense) cases in certain contexts, which are lexically, semantically or syntactically de®ned. The morphological scope of the case marker is lexical and not phrasal, though the it is attached to the head of NP and the head is repeated along with each of its modi®ers when the noun phrases are conjoined. That the scope is lexical is evidenced by the fact that in a conjunction of nouns making a conjoined NP, the case marker is added to each of the nouns conjoined. It may, however, be attached to the summative word at the end of conjunction, if the conjunction is not generated with the conjunctive su½x -um. The summative word at the end is ellaam `all' and aagiyoor `these people'. In doublets of nouns with similar meaning and syllabic structure, the case marker is attached only to the second word in its plural form. The doublets, however, are a lexical category. (1)
aaDu maaDukaLukku sheep cow-pl-to `To sheep and cows, i.e., to the cattle'
The lexical nature of case marking is strongly evidenced by the facts of modi®er ¯oating (Annamalai 2003). Though the case marker is added only to the post nominal modi®er when the head and the nominal modi®er are adjacent (2a), it is added obligatorily to both the head and the post nominal modi®er when there is an intervening article (2c). It can in fact be added to both optionally even when they are adjacent (2b). (2)
a.
naan payyanaga muuNupeerukku peenaa kuDutteen I boy-pl three people-to pen gave
b.
naan payyanagaLukku muuNupeerukku peenaa kuDutteen I boy-pl-to three people-to pen gave
c.
naan payyanagaLukku peenaa muuNupeerukku kuDutteen I boy-pl-to pen three-people-to gave
d.
*naan payyanaga peenaa muuNupeerukku kuDutten I boy-pl three people-to gave
e.
*naan payyanagaLukku peenaa muuNupeeru kuDutteen I boy-pl-to pen three-people gave `I gave pens to three boys.'
2. Semantic Roles and Grammatical Functions Morphological marking is believed to permit free word order because it permits the case of a noun to not be determined by its place in the linear order
Case and the Argument Structure of Tamil
59
of words in a sentence. The assumption is that the information of grammatical function (GF) like subject (SU), object (O), indirect object (IO) etc. is obtainable from the case, as also the information of thematic or semantic role (SR) like Agent, Goal, Object / Theme / Patient etc. The relationship between case, GF and SR is, however, not straightforward. To ®gure out their complex relationship is the concern of all linguistic theories. The predicate of a sentence carries with it a string of nouns which are linked to it syntactically and semantically. These nouns carry case, whether the case is morphologically expressed or not. They also carry GF and SR. SRs are pre-theoretical primitives. The number of SRs postulated and the kind of relation between SRs, irrespective of whether they form a hierarchy or have relative prominence, are theory-internal matters (Wechsler 1995). GFs may be primitive or derived depending on the theory. They are primitive in Lexical Functional Grammar and are derived from the con®gurational position of the noun in Government and Binding Theory (GB). The status of GF may depend on the nature of the language also. GFs cannot be derived from the constituent structure of the sentence in Tamil, as there is no VP and there is no con®gurational di¨erence between SU and O in Tamil (Annamalai 2003). They must therefore be primitive or derived from some other source like case. 3. Semantic Roles and Arguments This paper, however, does not examine the relationship between case, GF and SR. It is limited to examining how the nouns are assigned their case given the argument structure of the predicate. The nouns linked to the predicate are categorized into arguments (ARG) and adjuncts (ADJ). This distinction is based on the relation of the nouns to the SRs. The predicate by its meaning obligatorily licenses certain SRs. For example, for the predicate kuDu `give' to be meaningful requires an actor or Agent, an object or Theme and a receiver or Goal. These three SRs are licensed by this predicate. Another SR, say Location, is not required for this predicate to be meaningful and it may not be licensed obligatorily in a sentence. The obligatorily licensed SRs are ARGs from the semantic point of view of the predicate and those that are optionally added are ADJs. GFs are categorized into primary GF and oblique GF, or simply as being grammatical function and oblique function. This distinction is based on whether they play a role or not in some grammatical phenomenon like word order, agreement, coreference control, movement etc. GFs that play a role are Primary Grammatical Functions (henceforth simply called GFs) and which do not play a role are Oblique Grammatical Functions (henceforth called OFs). The nouns that carry Primary Grammatical Functions
60
E. Annamalai
are Arguments from the grammatical point of view. The nouns that carry Oblique Grammatical Functions are Adjuncts. Thus an ARG is a noun which carries an obligatorily licensed SR (henceforth SR refers only to this) and Primary Grammatical Function (GF). There could be a discrepancy between these two functions of a noun allowing an ARG to carry a GF but not an SR, and, conversely, to carry an SR but not a GF. The latter is more common than the former. The possibility of a discrepancy suggests that an ARG may carry both SR and GF or only SR or only GF. First we shall examine the relation between SR and ARG. Mohanan (1994) shows that out of ®ve possible relations between them, four (a,b,c,d below) are permissible in natural languages and one (e below marked by an asterisk) is not. In the diagram below, X, Y on the top line stand for SRs and A, B on the bottom line for ARGs. (3)
a.
X
Y
A
B
b.
X Y B
c.
Y A
B
d.
X Y
e.
A
*
X
A
B
(a) makes the claim that for each SR there is an ARG; (b) claims that an SR may not have a corresponding ARG; (c) claims that an ARG may not have a corresponding SR; (d) claims that two SRs may have one corresponding ARG; and (e) claims that one SR may have two corresponding ARGs. The ideal state is (a); but languages do have (b,c,d). (b) is the consequence of the fact that a GF need not be an SR. In other words, it is the consequence of the fact that GF does not have semantic motivation. For example, an SR like Source may play no role in any grammatical phenomenon like the ones mentioned above and so may not carry GF at all, causing it not to be an ARG from a grammatical point of view. A universal principle may be stated as (4)
The GFs of a predicate are equal to or less than its SRs.
(c) will result when a non-SRÐi.e. a semantically empty nounÐis made to carry a GF. This can be seen in a language like English which has semantically empty subjects like `it, there'. (c) will also result when a noun of non-SR is made to carry a GR by a movement rule. When the nominative noun of an embedded clause becomes the SU of the matrix clause (MS) by raising as in 5(b), this noun carries no SR with respect to the predicate of MS. (5)
a.
payyan nallavaanaa teriyidu boy good-he-adv appear-pr-png (neuter sg)
Case and the Argument Structure of Tamil b.
61
payyan nalllavaanaa teriyraan boy good-he-adv appear-pr-png (human masculine sg) `The boy appears to be good.'
(c) also results when the predicate allows one of the SRs not to manifest in the sentence formally, though it is obligatorily licensed conceptually. For example, the predicate villu `sell' requires the SRs Agent, Theme and Goal, but it allows Agent and Goal not to be manifested as in (6) (6)
onga pustagam nallaa vikkidu your book good-adv sells `Your book sells well.'
(d) results when a predicate allows an additional semantic feature like volition or causation and allows one of the two coindexed nouns not to manifest formally. For example, the predicate uruLu `roll' is or volitional. It licenses the SR Theme and additionally the SR Agent when it is volitional. When both these SRs are licensedÐi.e., when the predicate is volitionalÐand the Agent SR is coreferential with the Theme SR, which does not formally manifest, both SRs inhere in the same ARG. In 7(b) kumaar `Kumar' is both Agent and Theme. (7)
a.
malelerundu kallu uruNDuccu hill-from stone rolled `The stone rolled from the hill.'
b.
kumaar kooyilesutti uruNDaan Kumar temple-acc-around rolled `Kumar rolled around the temple.'
A common feature that makes (c) and (d) possible is the property of the SR to not be formally manifest in a noun. This suggests a putative universal principle (8). (8)
ARGs of a predicate are equal to or lesser than its SRs.
(8) is a corollary of (4) given the universal principle (10) below. This principle, however, is violated in languages that allow the syntactic rules mentioned above that give rise to (c)Ða semantically empty noun, that is, may ®ll in the position of SU and a noun not carrying SR of a predicate may be moved to become the SU of that predicate. This putative universal principle is overridden by language speci®c rules of syntax with reference to what can be SU in that language. If the SU is treated as a di¨erent kind of ARG (GB treats it as an external argument for a di¨erent reason), then the principle (8) can be maintained universally.
62
E. Annamalai
4. Arguments and Grammatical Functions Let us now examine the relation between ARG and GF. Mohanan (1994), following the Stratal Uniqueness Law of Relational Grammar (Perlmutter and Postal 1983) and Function-Argument Biuniqueness Principle of Lexical Functional Grammar (Bresnan 1982) states that ARGs and GFs have one to one relation. In the diagrammatic representation of this relation given below, where X, Y stand for GF and A, B for ARG, only (a) is permissible. (9)
a.
X
Y
A
B
b.
*
X Y A
c.
*
X X A
B
d. *
X X A
B
(b) claims that an ARG cannot carry more than one GF. (c) claims that two ARGs cannot carry one and the same GF. (d) is not exactly about the relation between GF and ARG like (a,b,c) and it represents the fact that a predicate will not have more than one occurrence of the same GF and consecutively two ARGs carrying same GF is not permissible. For example, for a ditransitive predicate, O and IO are not the same GF, even if a language has one case marker for both. What is true of GF is true of SR also. That is, a predicate will not have more than one occurrence of the same SR. For example, a causative predicate does not have two Agents, and the two actors of the predicate have different SRs like Instigator and Agent. A universal principle from the above is the following. (10)
ARGs of a predicate are equal to its GFs.
The claim that a predicate has only one instance of GF follows logically from this principle. Let us now examine the relation between ARGs and case. Their relation may be same as the relation between ARG and GF as represented in (9). That is, ARG and case will have one to one relation. This may be extended to ADJs also to give the universal principle (11). (11)
A predicate has only one instance of a case linked to it.
This is, however, an empirical matter and needs to be empirically validated. It is taken up later in this paper. Given the validity of the above universal principles, it follows that a predicate will have one instance each of an SR, of a GF and of a case. It does not follow that SR, GF and case are isomorphic and that their relation is constant with all predicates. Any particular manifestation of the relation between themÐfor example, which SR becomes the SU or which ARG
Case and the Argument Structure of Tamil
63
becomes nominativeÐwill depend on the pragmatic notion of prominence (Mohanan 1994) and on the syntactic constraints on rules to e¨ect the prominence. SRs are a set of language independent semantic primes. Each SR represents an abstraction of the senses codi®ed in a case that carries it. The case, in other words, may have many speci®c senses, which are obtained in conjunction with the meanings of other elements, particularly the predicate and the noun that carries the case. The speci®city of the senses may be of di¨erent degrees. The dative case, for example, that carries the SR Goal has the senses of recipient and destination; the latter is obtained with predicates of motion.
5. Multiple Senses of a Case The locative case has spatial and temporal senses on the noun that carries this case. There are more speci®c senses of the locative case depending on the meanings of the predicate, the noun and their combination as in the sentences in (12). (12)
a.
pustagam tarele keDakku book ¯oor-loc lie `The book is lying on the ¯oor.'
b.
pustagam payle irukku book bag-loc is `The book is in the bag'
c.
kuurele oru palli ooDudu roof-loc a lizard runs `A lizard is running on the roof'
d.
kaNNule taNNi varudu eye-loc water comes `Water is coming from the eyes'
e.
nii kaNNaaDile teriyre you mirror-loc appear `You are seen in the mirror'
f.
nii en kanavule vande you my dream-loc came `You appeared in my dream'
g.
naan maZele nanejiTTeen I rain-loc got wet `I got wet in the rain'
64
E. Annamalai h.
naan taNNile kayye kaZuvuneen I water-loc hand-acc washed `I washed my hands in water'
i.
ivanukku nenjile valikkidu he-to heart-loc pains `He has pain in the chest'
j.
ivan paNattule kuriyaa irukkaan he money-loc target-adv is `He is singularly concerned about money'
k.
roojaavule pala vage irukku rose-loc many kind is `There are many kinds of rose'
l.
pattu naaLLe varreen ten day-loc come `I will come in ten days' etc.
When a case has separate morphological markers for related senses, all of which are subsumed under the abstract or generic sense of that case, each marker may be taken to indicate a sub-case. They are sub-cases because they cannot have corresponding separate SRs and GFs. The locative case is again an example of this. It has di¨erent markers to refer to di¨erent orientations of location. A particular case marker is assigned to a noun that stands for that particular orientation of location or time. Assignment of sub-case is based purely on semantic speci®city and it is subordinated to the generic locative case. The sub-cases for speci®c sense orientations may be an open ended set. Their morphological form may vary from a su½x to a word. The forms of sub-case markers of the locative case in Tamil include the following: -le `in', meele `on, above', pakkam `by the side of', munnaale `in front of', pinnaale `behind', uLLe `inside', veLiye `outside', edure `opposite to', aDuttu `next to', pakkattule `next to', aDile `under', uccile `on the top of' etc.
6. Discovering Grammatical Functions and Arguments With regard to identi®cation of GF, the discovery procedure is to ®nd out the grammatical phenomena that operate on a noun and to designate the GF to the noun according to the phenomena it manifests. For example, there are grammatical properties identi®able with a GF like SU (Keenen 1975, Verma 1976).
Case and the Argument Structure of Tamil
65
There is no discovery procedure for the identi®cation of ARG. ARGs are obligatorily licensed by the predicate's meaning. If it is obligatory for the ARGs to be present syntactically in the sentence, this would help their identi®cation, but this is not the case. Tamil allows zero ARGs as in (13) and the sentences are well-formed without the licensed ARGs being present formally in them. (13)
a.
naan eDutteen I took `I took'
b.
naan kuDutteen I gave `I gave'
c.
naan paDutteen I lay `I lay down'
eDu `take' must have an object (SR Object / Theme) to take, which is absent in 13(a). kuDu `give' must have an object and a recipient (SRs Object / Theme and Goal), which are missing in 13(b). paDu `lie down' must have a place (SR Location) to perform the action, which is not there in 13(c). In spite of their absence, the sentences in (13) are well-formed in isolation. They are not truncated sentences found in a discourse. A heuristic test, however, has been suggested by Panevora (1978). When a speaker makes a statement without the licensed ARG and the hearer asks a question to inform himself about it, the speaker cannot answer felicitously `I don't know'. He can if the missing noun is an ADJ. In 14(a,b), the missing nouns of recipient and location are ARG and in 14(c,d) the missing nouns of source and destination respectively are ADJ. (14)
a.
A. Everyone gave. B. What did everyone give? A. *I don't know.
b.
A. Everyone slept. B. Where did everyone sleep? A. *I don't know.
c.
A. B. A.
Everyone bought a book. Where did everyone buy a book? I don't know.
d.
A. B. A.
Everyone disappeared. Where did everyone disappear? I don't know.
66
E. Annamalai
This heuristic test is, however, not foolproof, as (e) shows. The predicate `go' licenses Goal (i.e., the destination of going), and so Goal must be an ARG, but the answer `I don't know' is not infelicitous. e.
A. Everyone has gone. B. Where has everyone gone? A. I don't know.
This is so because `go' in the given syntactic context has the sense `leave' and the predicate with this meaning does not license Goal. Such instances bring in circularity to the test when there is no independent evidence to increase the number of senses of a predicate. 7. Case Like Forms Case is assigned to a noun by a combination of its semantic and grammatical functions in relation to the predicate. In Tamil, as mentioned in the beginning, cases are marked morphologically by distinct markers including zero. Two questions arise with regard to identifying a morphological form added to a noun: (i) which of these morphological forms are case markers and (ii) which forms should be treated as two di¨erent case markers. The list of morphological forms which are added to a noun and have a function similar to case marking includes the following. The list is open ended. (15)
muulam `through', vaZiyaa `through, by the way of', kurukke `across, through', padil/padilaa `instead of', eduraa `in opposition to, against', saadagamaa `in favour of', aadaravaa `in support of', patti `about', poola/maadiri `like', viDa `than' etc.
The last two forms are sentence complements occurring with nominalized S or relative clause. They also occur with simple nouns, which are in the accusative with accusative case marker. The simple nouns are extendible into sentences; in other words the simple nouns stand for reduced S. The noun may have cases other than the accusative, which are cases retained from the underlying sentence. (16)
a.
enakku piDikrade poolavve kumaarukkum raajaave I-to like-pr-it-acc like-emp Kumar-to-also Raja-acc piDikkum like `Kumar also likes Raja as I do.'
Case and the Argument Structure of Tamil b.
67
enne/enakku poolavee kumaarukkum raajaave piDikkum I-acc/I-to like-emp Kumar-ti-s-also Raja-acc like `Kumar also likes Raja like me'
Since poola/maadiri `like' and viDa `than' occur with another case marker (which is not used to make the bare noun into oblique for declension), they cannot be taken as case markers on the principle that a noun does not carry two cases. It may be noted that some forms in (15) also have the adverbial marker -aa. The adverbial in Tamil is derived by adding this su½x to a noun. This raises the question whether the forms in (15) which end in -aa are adverbs or case markers. They are not independent adverbs, in that they occur attached to another noun like case markers. The base nouns to which these forms are attached may be optionally absent, but are recoverable. Functionally, they do not modify the predicate directly but relate the base noun to the predicate like case markers. For these reasons, the forms with -aa in (15) may be taken as case markers. The formal similarity of some of the forms in (15) with adverbs also raises the question about the relation between case and adverb. Formally and semantically the adverb may vary with the comparable noun and case marker (CM). (17)
a.
kumaar mariyaadeyooDa peesunaan mariyaadeyaa Kumar respect-with spoke respect-adv `Kumar spoke with respect/respectfully'
b.
kaaru veegattule vandudu veegamaa car speed-in came speed-adv `The car came in speed/speedily'
c.
kumaar koobattooDa peesunaan koobattule koobamaa Kumar anger-with anger-in anger-adv `Kumar spoke with anger/in anger/angrily'
There is, however, a syntactic di¨erence between noun aa and noun CM. The former does not take noun modi®ers.
68 (18)
E. Annamalai a.
kaaru arubadu mayil
veegattule vandudu *veegamaa car sixty mile speed-in came speed-adv `The car came at a speed of sixty miles/*sixty mile speedily'
b.
kumaar anbu kalanda
mariyaadeyooDa peesunaan *mariyaadeyaa Kumar love mix-pa-rel respect-with spoke respect-adv `Kumar spoke with respect mixed with love/respectfully that was mixed with love'
This shows that the base of an adverb does not behave like a noun syntactically though it is a noun lexically. Since case is a relation between a noun (ARG or ADJ) and the predicate in a sentence, adverbs cannot be treated as an expression of case. The alternation between noun aa and noun CM must be treated as generated di¨erently but converging semantically. There is no simple criterion to decide which of those forms in the open list of (15) should be taken as case markers and which among the case markers are distinct cases and not sub-cases.
8. Two Markers and One Case The second question mentioned earlier is about one CM coding more than one case. The crucial criterion to decide this question is (11), the principle mentioned earlier, viz., a predicate has only one occurrence of a case. Cases of double nominative, double accusative etc. must have special explanations, if this principle is valid. If the same CM occurs twice with a predicate and the meanings of the two occurrences are contrastive i.e., one is not derivable from the other with the help of distributive or semantic factors, then the two nouns with the same CM may be taken to represent two distinct cases. The CM -kku in Tamil associated with the dative case can be examined with reference to this question. It was already mentioned that the two senses of this CM viz., recipient and destination do not code two cases because of the correlation of the latter sense with a semantic class of predicate viz., predicates of motion. The complex CM -kkaaga `for the sake of' derived from the dative CM by adding -aaga to it expresses the sense of benefactive, and this form must be treated as a distinct CM, constituting a distinct case, not as a variant and sub-sense of the dative, because of their distributive or semantic contrast.
Case and the Argument Structure of Tamil (19)
69
naan kumaarukku oru viiDu vaanguneen kumaarukkaaga I Kumar-to a house bought Kumar-for `I bought a house to Kumar/for Kumar'
Both CMs occur with nouns of the same predicate indicating separate semantic roles. (20)
naan kumaarukkaaga avan tambikki oru viiDu vaanguneen I Kumar-for his brother-to a house bought `I bought a house to his brother for the sake of Kumar'
The sense `for the sake of' in (20) is a sub-sense of the meaning `for' of the benefactive case. The benefactive case can be contrasted with the dative. Meaning can differentiate the actual recipient from intended bene®ciary as in (21). (21)
naan
kumaarukkaaga oru viiDu vaangi avan tambikki *kumaarukku I Kumar-for a house buy-pa his brother-to kuDuttuTTeen gave away `I bought a house for/*to Kumar and gave it away to his brother'
These facts argue for a separate case, benefactive. Its maker is morphologically complex (dative agga) like the CM of the ablative case, which is locative irundu. There is a CM -TTe, which is apparently in complementary distribution with the dative -kku (in its destination sense) and with the locative -le. Where -kku and -le occur with nouns with the semantic feature -animacy, -TTe occurs with nouns with the feature animacy. (22)
a.
kumaar
viiTTukku poonaan *viiTTuTTe Kumar house-to went house-prox `Kumar went to the house'
b.
kumaar
raajaaTTe poonaan *raajavukku Kumar Raja-prox went Raja-to `Kumar went to Raja'
70 (23)
E. Annamalai a.
en
paNam
kumaarTTe irukku *kumaarle my money kumar-prox is kumar-loc `My money is with Kumar'
b.
en
paNam
viiTTule irukku *viiTTuTTe my money house-loc is house-prox `My money is in the house'
There is a semantic contrast also between -TTe on one hand and -kku and -le on the other1. In 24(a) and 24(c,d), the animate noun takes both CMs, but their meanings are di¨erent, In 24(b) -TTe is not permitted with animate noun. (24)
a.
en
paNam kumaarTTe irukkaTTum kumaarukku my money Kumar-prox be-let Kumar-to `Let my money be kept with Kumar' for
b.
kumaarukku kaaccalu vandudu *kumaarTTe Kumar-to fever came Kumar-prox `Kumar got fever'
c.
maaTTule paalu ille cow-loc milk not `There is no milk in the cow'
d.
kaNNukuTTi maaTTuTTe ille calf cow-prox not `The calf is not with the cow'
The predicate can take a noun with -TTe along with a noun with -kku or -le as in (25). This should not be possible if both nouns carry the same case. (25)
naan kumaarTTe
avan tambikki paNam kuDutteen viTTule I Kumar-prox his brother-to money gave house-loc `I gave money to Kumar for his brother/in his house'
Case and the Argument Structure of Tamil
71
Their complementary distribution given in (22) and (23) follows a semantic distinction and therefore cannot be taken to support the view that what we have is simply two manifestations of the same case. One these grounds, -TTe may be considered a marker of a separate case, which is called proximate case (for want of a better term) to distinguish it semantically from being physically part of location or destination, as with the locative case.
9. Complex Case Markers The case marker may be simple or complex as already mentioned. The benefactive -kkaaga is a complex CM. Another complex CM is of the ablative case, which is formed by combining irundu with the locative -le or the proximate -TTe or with -e occurring with locative nouns like meele `above', kiiZe `below', or with locative pronouns like ange `there' etc. Another kind of complex CM is the verbal participle combined with the accusative CM -e as in the variant instrumental CM -evacci. (26)
kattiyaale narukku kattiyevacci knife-inst cut `Cut with the knife'
The oblique form of the noun to which CM is added may be in genitive or dative case. (27)
a.
en munnaale enakku my before I-to `Before me'
b.
meele viiTTu viTTukku house-gen above house-to `Above the house'
Di¨erent oblique forms may make a semantic di¨erence in some cases. For example, the genitive in 27(b) has the sense of `on, on top of' and the dative `over'. Such semantic di¨erence and the syntactic relation between object and verb when CM is verbal participle as in (26) make it di½cult to distinguish between oblique noun simple CM and bare noun complex CM.
72
E. Annamalai
The above description of the morphology of CM in Tamil shows that the questions of complex CM and of separateness or autonomy of cases remain far from satisfactorily answered.
10.
Zero Case Marker
Another morphological characteristic of CM in Tamil is that it can be zero, obligatorily with the nominative or optionally with the accusative and the dative. In the dative, CM can be zero when the sense is destination, the predicate is a motion verb like poo `go' and vaa `come' and the noun is a place or institution.2 The dative case noun should be close to the verb for the CM to be zero. (28)
a.
naan madurekki pooreen madure I Madurai-to go Madurai `I am going to Madurai'
b.
naan aa®isukku pooreen a½is I o½ce-to go o½ce `I am going to o½ce'
c.
*naan aa®is naDakreen I o½ce walk `I walk to o½ce'
d.
*naan madure rayille pooreen I Madurai train-in go `I am going to Madurai by train'
The distributive quanti®er is expressed by doubling the noun. In the doubled noun, the dative CM is always absent in the second noun3. (29)
a.
naanga viiTTukku viiDu poonoom naDandoom we house-to house went walked `We went / walked to every house'
b.
naanga uurukku uuru paNam kuDuttoom we village-to village money gave `We gave money to every village'
Case and the Argument Structure of Tamil
73
In this quanti®er sense, when -aa occurs with the second noun, dative CM is always absent from the ®rst noun as well. (30)
naanga viiDu viiDaa naDandoom we house house-adv walked `We walked to every house'
This is also true of the accusative. (31)
naanga viiDu viiDaa paattoom we house house-adv saw `We saw house after house'
The accusative CM is generally zero when the noun is de®nite and inanimate. In (32) the noun does not have these features, but the accusative CM is zero. The accusative noun is animate in 32(a,b) and is de®nite in 32(c). (32)
a.
avan naayi vaLakraan he dog grow-pr `He raises a dog'
b.
avan poNNu paakraan he girl see-pr `He is looking for a bride'
c.
naan inda puu paattadulle I this ¯ower see-it-not `I have not seen this ¯ower'
In 32(a,b), however, the animate noun is inde®nite and non-speci®c. In 32 (c), the inanimate noun is not de®nite as it does not refer to a de®nite and speci®c ¯ower despite inda `this'Ðit means `this kind of'. (32) seems to show that it is not animacy but de®niteness and speci®city are the features that determine the presence or absence of accusative CM. However, there are counter examples to this condition, as (32) shows. In 33(a), the noun is de®nite and speci®c with unique reference, but the accusative CM is zero. The di¨erence between 33(a) and (b) is that in the latter the action of the predicate does something to the object; i.e., the object does not remain the same after the act. This requires the presence of accusative CM. In 33(b) the action of the predicate has no such e¨ect on the object. This suggests that the function of the predicate plays a role in the presence or absence of the accusative CM. (33)
a.
ellaarum madure paatturukkaanga everyone Madurai seen-has `Everyone has seen Madurai'
74
E. Annamalai b.
yaaroo madureye ericciTTaanga someone Madurai-acc burnt-has `Someone has burnt Madurai'
When there is no CM on the noun which carries the SR Theme / Object it is not treated as the nominative case (as in passive construction) because there is another noun without CM which is nominative with syntactic properties like providing agreement on the predicate. Further, this noun can have the accusative CM added optionally. It cannot be conjoined with the nominative. It is not in the oblique (like the nominative) because the oblique form is a morphological property of in¯ection and not of case. The dative case with zero CM is also not in oblique for the same reason4. The minimum list of cases and their markers in Tamil is given below: nominative accusative dative benefactive locative proximate ablative instrumental associative
-e, -kku, -kkaaga -le -TTe -lerundu, -TTerundu, -eviTTu -aale, -TTe, -evaccu -ooDa, kuuDa
Genitive and vocative are not treated as case because the nouns in¯ected with them are neither ARGs nor ADJs of the predicate in the sentence.
11.
Assignment of Case to Nouns
We shall now examine how the nouns of a predicate are assigned case. The number of cases is decided by the meaning of the predicate. The noun carrying a case may be an ARG or ADJ. Since ARGs are obligatory subject to their being formally absent, cases with ARG nouns are obligatory. It is not the case that any particular case, including the nominative, is obligatory. The obligatory cases will di¨er according to the semantic property of the predicate. It is, however, necessary that there is a minimum of one case for any predicate, though it may be formally absent like the nominative with the predicate in the imperative or with agreement marker or as in impersonal sentences.5 Whereas 34(a) has only a nominative, other sentences in (34) have no nominatives:
Case and the Argument Structure of Tamil (34)
a.
kumaar siriccaan Kumar laughed `Kumar laughed'
b.
kumaarukku enne teriyum Kumar-to I-acc knows `Kumar knows me'
c.
enakku bayamaa irundudu I-to fear-adv was `I was afraid'
d.
enakku apDi tooNale I-to so appear-not `It does not appear to me so'
e.
enakku koDale poraTTikiTTuvandudu I-to intestine-acc turning up-was `My stomach was turning up'
f.
avane koDiyaale muuDiyirundudu he-acc ¯ag-inst cover-pa-be-pa-png `He was draped with the ¯ag'
g.
avane kaaNoom he-acc see-not `He is not seen'
h.
kumaaraale paDikka muDiyale6 Kumar-inst read-inf can-not `Kumar could not read'
75
The morpheme -aa occurs with the nominative, which is single or duplicated as in 35(a±c). This is not CM but a marking for intensity / plurality. It can be added to the noun of any case and with nouns having CM; it occurs after the CM as in 35(e). (35)
a.
inge maramaa irukku here tree-int is `There are plenty of trees here'
b.
enakku sirippaa vandudu I-to laughter-int came `I felt very much like laughing'
c.
avanukku koobangoobamaa vandudu he-to anger-anger-int came `He got very angry'
76
12.
E. Annamalai d.
naan pustagamaa vaanguneen I book-int bought `I bought a lot of books'
e.
kumaar puukaDekkaa poonaan Kumar ¯ower-shop-to-int went `Kumar went to a lot of ¯ower shops'
Case Assignment and Nominal Predicate
The predicate may be a noun in Tamil. It does not have any morphological marking. Being a predicate itself, it cannot carry any case and it cannot be said to be in the nominative. The sentences in (36) are some examples of a nominal predicate having case nouns with it. The meaning of the verbal predicate determines the case it takes. In the case of nominal predicates, it is not their meaning but the meaning of existence or change of state associated with the nominal predicate that appears to determine the cases. (36)
13.
a.
naanga viiTTule naNbargal we house-loc friends `We are friends at home'
b.
naanga pattuvayasulerundu naNbargal we ten-age-from friends `We are friends since the age of ten'
c.
naan iNNekki onnaale paNakkaaran I today you-inst rich man `I am a rich man today because of you'
d.
naan ippa onakkaaga oru DaakTar I now you-for a doctor `I am a doctor now for your sake'
Double Occurrence of a Case
As mentioned earlier, it is a universal principle that each case occurs only once with a predicate. This condition does not apply to conjoined nouns where more than one noun carries the same case. It also excludes a reduplicated noun with case for plurality or repetition of the act. 37(a) and (b) are respective examples of these. (37)
a.
kumaar kayyeyum kaaleyum aaTTunaan Kumar hand-acc-conj leg-acc-conj swayed `Kumar swayed his hand and leg'
Case and the Argument Structure of Tamil b.
77
kumaar kayye kayye aaTTunaan Kumar hand-acc hand-acc swayed `Kumar swayed his hand again and again'
Multiple locatives of a predicate are staggered locations embedded one into another. This is also excluded from this condition. (38)
en pustagam uurle viiTTule meejele oru muulele my book village-loc house-loc table-loc a corner-loc irukku is `My book is in a corner on the table in the house in the village'
There are many other instances of double cases of the predicate discussed by Lindholm (1976). We shall discuss whether they constitute a counter example to the principle of one occurrence per case with a predicate. (39)
a.
arasu tiruccile timukaavule seendaan Arasu Trichy-loc D.M.K in joined `Arasu joined the DMK party in Trichy'
b.
pongalukku piLLegaLukku pudu tuNi eDuttoom pongal-to children-to new cloth took `We bought new clothes for the children for the Pongal festival'
c.
naan piLLegaLukku pattu naaLekki paNam I children-to ten days-for money kuDutturukkeen given-have `I have given money to the children for ten days'
d.
DaakTar piLLegaLukku kaaccalukku nuuru doctor children-to fever-to 100 ruubaaykki marundu kuDuttaaru rupees-to medicine gave `Doctor gave children medicine for 100 rupees for fever'
e.
enakku sinnavayasulerundu amerikkaavulerundu pustagam I-to small-age-from America-from book varudu comes `I get books from America from my young days'
It may be noted that though the form of the CM is the same in all its occurrences in the sentences in (39), there is some di¨erence in their meaning.
78
E. Annamalai
This di¨erence is called sub-senses of a case. In 37(a), it is the lexical property of the verb seer `join' to take the locative case when `getting admitted into a party' is meant; the other locative is for the place where the event took place. In 39(b,c,d), the dative case in piLLegaLukku `to the children' indicates `goal' and the other dative case markers indicate senses of occasion, duration, purpose, and amount. In 39(e), the second ablative case indicates the source and the ®rst duration. These suggest that a distinction can be made between the primary sense of a case and its secondary senses, which do not relate to the SR of the predicate. It may also be noted that in each of the sentences above only one of the CMs of their multiple occurrences is with the ARG noun. This is the noun that carries the SR. On the basis of this evidence principle (11) should be modi®ed as below. (40)
A predicate has only one instance of a case among its Arguments
This is a natural corollary of the principle that each predicate licenses only one instance of an ARG. There are sentences in Tamil where there is apparent double occurrence of nominative and accusative. If (40) is to be maintained, it must be shown that at least one of the nouns of the duplicated case is not an ARG or some alternative explanation must be found for the apparent duplicated nominative and accusative cases.
14.
Case Status of the Noun in Compound Verbs
We shall ®rst explore the limits of the alternative explanation. Since the double case of this kind is typically without CM, we shall not prejudge whether it is nominative or accusative. There are di¨erent types of construction of double case where more than one noun is nominative or accusative. Lindholm (1976) discusses many construction types, but not all. His analysis is to treat the predicate and the noun adjacent to it as a verbal complex, which in turn functions as a predicate to assign case to other nouns in the sentence. He calls the double case constructions nested case constructions. The case that becomes part of the verb to create a verbal complex nests the same case. The main di½culty with this analysis is that it treats all double case constructions as the same, even those that seem to be structurally di¨erent. Let us ®rst take up N V construction type, which is called a compound verb, and is formed by compounding an auxiliary verb to a noun. The verbal element is an auxiliary because it is semantically empty except to mark presence or absence of transitivity. The (a) sentences below with the auxil-
Case and the Argument Structure of Tamil
79
iary verb paNNu `do', paDuttu `make someone experience' are transitive and (b) aagu `happen', paDu `experience' are intransitive. (41)
(42)
(43)
(44)
a.
kumaar eDatte kaalipaNNiTTaan Kumar place-acc empty-did `Kumar vacated the place'
b.
kumaar aaLu kaaliyaagiiTTaan Kumar person empty-happened `Kumar died'
a.
kumaar saappaTTe kaalipaNNiiTTaan Kumar food-acc empty-did `Kumar ®nished his food'
b.
saappaaDu kaaliaagiirucci food empty-happened `Food was ®nished'
a.
kumaar deeviye kalyaaNampaNNikiTTaan Kumar Devi-acc marriage-did `Kumar married Devi'
b.
kumaarukku kalyaaNamaagiirucci Kumar-to marriage-happened `Kumar is married'
a.
kumaar enne kaSTapaDuttunaan Kumar I-acc su¨ering-made experience `Kumar made me su¨er'
b.
naan kaSTapaTTeen I su¨ering-experience `I su¨ered'
It may be seen that in the above sentences (except in 43(b)) there are two nouns without CM. It is possible that in the transitive sentences the unin¯ected noun with the auxiliary verb (AV) is accusative and in the intransitive sentences it is nominative. This possibility, however, will arise only if the noun in N V is an independent ARG under S and is not part of the predicate. The noun in Tamil is characterized as a word that takes a CM and an adverbial su½x. Some of them take the adverbial su½x but not the CM. They may be said to possess less nominal properties. Nouns like kaali `emptiness' in (41) belong to this category of nouns. The question of case of such nouns does not arise. They are caseless nouns.
80
E. Annamalai
With regard to `full'' nouns, there is some evidence that in (42)±(44) they are not part of the predicate. The interrogative and emphatic clitics can occur between N and AV. (45)
a.
kumaar deeviye kalyaaNamaapaNNikiTTaan? Kumar Devi-acc marriage-int-did `Did Kumar marry Devi?'
b.
nii kaSTamaapaTTe? you su¨ering-intr-experienced `Did you su¨er?'
It should however be noted that even simple verbs allow clitics to occur inside them. The clitic is added to the in®nitive form of the simple verb and a dummy verb sey `do' is added to carry in¯ection. (46)
a.
nii paDikkavaaseyre? you read-inf-intr-do `Are you reading?'
Ns with some transitive AVs can optionally have the accusative CM present. (47)
a.
b.
kumaar kalyaaNattepaNNikiTTaan Kumar marriage-acc-made `Kumar married' *kumaar kaSTattepaDuttunaan Kumar su¨ering-acc-made experienced `Kumar made someone su¨er'
Some nouns with AV allow some modi®ers. (48)
a.
kumaar oru kalyaaNampaNNikiTTaan Kumar a marriage-made `Kumar married'
b.
kumaar aaDambarakalyaaNampaNNikiTTaan Kumar luxury-marriage-made `Kumar married luxuriously'
c.
*kumaar periya kaSTapaDuttunaan Kumar big su¨ering-made-experience `Kumar made someone su¨er'
It must be noted that Ns of N V which take accusative CM and modi®ers do not do so when there is an accusative case noun in the sentence.
Case and the Argument Structure of Tamil (49)
a.
*kumaar deeviye kalyaaNattepaNNikiTTaan Kumar Devi-acc marriage-acc-made `Kumar married Devi'
b.
*kumaar deeviye oru kalyaaNampaNNikiTTaan Kumar Devi-acc a marriage-made `Kumar married Devi'
81
The compound noun in 48(b), however, is possible with an accusative case noun in the sentence. This indicates that a compounded noun should be treated di¨erently from a modi®ed noun. The above shows that N AV is a lexical category with varying degrees of lexicality. This is further supported by the following facts. The N can be questioned. (50)
a.
kumaar deviye enna paNNunaan? Kumar Devi-acc what did `What did Kumar do of Devi?'
b.
kumaar deviye kalyaaNampaNNunaan Kumar Devi-acc marriage-did `Kumar married Devi'
It can be gapped. (51)
mohan deeviye kaadalum kumar kalyaaNamum Mohan Devi-acc love-and Kumar marriage-and paNNunaanga did `Mohan loved Devi and Kumar married her'
It can also be eclipsed. (52)
a.
kumaar deeviye kalyaaNampaNNapooraanaa? Kumar Devi-acc marriage-do-inf-going-int `Is Kumar going to marry Devi?'
b.
aamaa paNNapooraan yes do-inf-going `Yes, he is going to'
In addition, it can be relativized. (53)
a.
kumaar deeviye paNNuna kalyaaNam Kumar Devi-acc do-pa-rel marriage `The marriage which Kumar did with Devi'
82
E. Annamalai b.
kumaar enne paDuttuna kaSTam Kumar I-acc make experience-pa-rel su¨ering `The su¨ering which Kumar caused me'
And scrambled. (54)
a.
kumaar kalyaaNam deeviye paNNapooraan Kumar marriage Devi-acc do-inf going `Kumar is going to marry Devi'
b.
kumaarukku kalyaaNam naaLekki aagapoogudu Kumar-to marriage tomorrow happen-going `Kumar is going to be married tomorrow'
Scrambling does not make the N of N AV necessarily a daughter of S. A noun modi®er can be scrambled outside the NP in its nominal form (Annamalai 2003).
15.
Complex Predicate and Case Assignment
If an N with lexical category status is permitted to be a part of a Complex Predicate (CP) (Alsina et al (eds) 1996) then, it is possible that N AV is a CP. We shall explore this possibility. Unlike a simple predicate, components of a CP also license ARGs i.e., essential SRsÐbesides the verb. Compound verbs with di¨erent nouns and the same AV license di¨erent ARGs. If the verb alone licenses ARGs, then these compound verbs should license the same ARGs, which is not true. For example, kalyaaNampaNNu `marry' licenses two ARGs, viz. Agent and Theme / Patient. aDampaNNu `be obstinate' on the other hand, licenses only one ARG, viz. Agent. The speci®cation that both ARGs of kalyaaNampaNNu `marry' must be human also comes from the N, kalyaaNam `marriage'. Moreover, the AV paNNu `do' in kalyaaNampaNNu `marry' cannot license two Theme ARGs of a person and marriage (which can be in accusative as shown earlier), as it is prohibited by the principle that a predicate cannot license the same ARG more than once. Another example of compound verb whose N plays a role in licensing ARGs is paZivaangu `take revenge'. The verb vaangu `get, buy' licenses three ARGs, viz. Agent, Theme and Source. But paZivaangu `take revenge' as in (54) licenses only ARG and Theme. This is because of the noun paZi `revenge', which licenses only these two ARGs. The ARG structure in the sentence strongly suggests that N AV is a CP. The internal structure of CP in Tamil seems to vary among CPs, as demonstrated above. This requires a separate study.
Case and the Argument Structure of Tamil
83
Some more examples of the type in 41(a)±43(a) where the N AV predicate takes another accusative case noun are the following. The list is not exhaustive. (55)
16.
a.
kumaar bommeye muttamiTTaan Kumar doll-acc kiss-placed `Kumar kissed the doll'
b.
kumaar enne sattampooTTaan Kumar I-acc noise-put `Kumar shouted at me'
c.
parundu eliye vaTTamiTTudu kite mouse-acc circle-placed `The kite circled the mouse'
Complex Predicate among Compound Nouns
The following are di¨erent kinds of N V constructions in Tamil, which might possibly be treated as CP. In some of these constructions V is AV, which is semantically speci®c in di¨erent degrees and which plays a minimal role in licensing the ARG structure. Some of the N V constructions do not take any case marked accusative and some take two nouns marked for accusative case. We shall examine ®rst the former type. The list is not exhaustive. (56)
a.
kumaar enakku veeleseyraan Kumar I-to work-does `Kumar works for me'
b.
kumaar enakku vaZipaNNunaan Kumar I-to way-made `Kumar found me a way'
c.
kumaar enakku vaZiviTTaan Kumar I-to way-let `Kumar gave me way'
d.
kumaar enakku veelevaccaan Kumar I-to work-kept down `Kumar gave me (unnecessary) work'
e.
kumaar en veelekki olevaccaan Kumar my work-to stove-kept down `Kumar made me lose my job'
84
E. Annamalai f.
kumaar enakku aasekaaTTunaan Kumar I-to desire-showed `Kumar enticed me'
g.
kumaar enakku kaDanpaTTurukkaan Kumar I-to debt-experience-is `Kumar is indebted to me'
h.
kumaar enakku taDepooTTaan Kumar I-to ban-put `Kumar prohibited me'
i.
kumaar enakku naamampooTTaan Kumar I-to line on forehead-put `Kumar cheated me'
j.
kumaar bommekki saTTepoottaan Kumar doll-to shirt-put `Kumar put the shirt on to the doll'
k.
kumaar bommekki muttamkuDuttaan Kumar doll-to kiss-gave `Kumar gave a kiss to the doll'
It may be seen that some sentences (e,i) in (56) are idioms and are lexicalized and others ( j,k) are phrasal with object. It is clear that not all N V can be treated as CPs. Whether N V is a CP or not can be determined by the ARG structure in the sentence and the place of the N in that structure. The N cannot be an ARG of the V of N V because CP as a predicate cannot license an ARG within itself. One of the crucial ways to identify a CP then will be to ®nd out whether the N of N V is an obligatory SR and/or GF. If some Ns of N V in the sentences in (56) turn out to be ARGs, it may not be possible to answer the question whether a predicate, simple or complex, can have more than one occurrence of the same case. This is because the sentences in (56) do not have accusative case and if the ARG of N is Theme / Patient, it does not result in the predicate having two of the same case. 17.
Double Occurrence of Case with Simple Predicate
There are many instances in Tamil where there are two accusative nouns. To maintain (40), it needs to be demonstrated that one of the nouns with accusative CM is not an ARG or that each of them is a di¨erent ARG. In the latter case, there will be case syncretism. We shall ®rst describe the syntactic properties of double accusative constructions with or without CMs.
Case and the Argument Structure of Tamil (57)
kumaar onne
(58)
kumaar enne kaDan keeTTaan adee aLavu kaDan enna Kumar I-acc loan asked same amount of loan what `Kumar asked me for a loan' `Kumar asked me for a loan of same amount' `What did Kumar ask me?'
85
kore sonnan enna kore enna Kumar you-acc lack said what lack what `Kumar spoke ill of you' `What did Kumar speak ill of you?' `What did Kumar speak of you?'
The predicates in (57, 58) have semantic content and are not auxiliary verbs. They di¨er in this respect from their analogues in (56). There is one noun with accusative CM in (57, 58) and another noun without any CM. The modi®er occurs with the noun without CM. Interrogative noun and quanti®er case occur in the slot of this noun. The interrogative is not a rhetorical question but a real question. All these facts show that the N V in (57, 58) is not a CP. Then the N must be an ARG or an ADJ and must have a case. The choice of case is between nominative and accusative. The predicates in (57, 58) must have Theme / Object as one of their ARGs. The accusative case marked noun can occur with another case (or non case) as in (59). (59)
a.
kumaar onnepatti kore sonnaan Kumar you-acc-about lack said `Kumar spoke ill about you'
b.
kumaar engiTTe paNam keTTaan Kumar I-prox money asked `Kumar asked me for money'
On the basis of this alternation and the presence of another noun in the nominative, it may be concluded that the nouns before the verb in (57, 58) have accusative case. Otherwise, there will be no ARG of Theme / Object
86
E. Annamalai
in the sentence, which is licensed by the predicate, or this ARG must be said to be null. That the nouns adjacent to the verb without CM in (57, 58) are accusative is supported by the fact that when they are de®nite, as in (60), they take the accusative CM7. (60)
a.
kumaar ?enne on paNatte keeTTaan engiTTe Kumar I-acc your money-acc asked I-prox `Kumar asked me for your money'
There are other sentences like (57, 58), however, which do not alternate the ®rst noun in the accusative with another case. (61)
18.
a.
kumaar enne paDam verenjaan *inda paDatte Kumar I-acc picture drew this-picture-acc `Kumar drew a picture of me' *`this picture'
Genitive Case Alternative in Double Occurrence
There could be genitive case instead of the accusative case with the ®rst noun, but they do not seem to be synonymous semantically. The genitive forms a noun phrase and the genitive noun or the head noun cannot be scrambled or relativized like the accusative noun and the second noun. The genitive phrase can either refer to the picture already in existence which Kumar draws again or to the picture that comes into being by Kumar's drawing. The accusative noun only has the latter meaning8. 61(a) is closer semantically to 61(c)9, where the second noun is `adverbial', rather than to 61(b), where the ®rst noun is genitive. (61)
b.
kumaar en paDatte varenjaan Kumar my picture-acc drew `Kumar drew my picture'
c.
kumaar enne paDamaa varanjaan Kumar I-acc picture-adv draw `Kumar drew (made) me into a picture'
There are other constructions where the accusative and the genitive alternate as in (62).
Case and the Argument Structure of Tamil (62)
a.
kumaar enne kayye piDiccaan en Kumar I-acc hand-acc caught I-gen `Kumar caught hold of me with my hand' `my hand'
b.
kumaar maaTTe kombe piDiccaan maaTTu Kumar cow-acc horn-acc caught cow-gen `Kumar caught hold of the cow with its horns' cow's horns
c.
kumaar bommeye kaale oDeccuTTaan bomme Kumar doll leg-acc broke `Kumar broke the doll on its leg' `doll's leg'
87
In (62), both nouns carry accusative CM obligatorily. The obligatory presence of CM in the second noun suggests that the ®rst noun is genitive making the second de®nite. The two accusative nouns make a genitive noun phrase. Semantically, they have part-whole relationship. In a genitive phrase, the genitive can be relativized but not the head. (63)
a.
kumaar vaaZapaZa toole uriccaan Kumar banana skin-acc peeled `Kumar peeled the skin of the banana'
b.
kumaar toole uricca vaZapaZom Kumar skin-acc peel-pa-rel banana `The banana whose skin Kumar peeled'
c.
*kumaar vaaZapaZom uricca toolu Kumar banana peel-pa-rel skin `The skin which Kumar peeled of the banana'
The two accusatives in (62) behave in the same way, i.e., the ®rst noun can be relativized but not the second. (64)
a.
b.
kumaar kombe piDicca maaDu Kumar horn-acc catch-pa-rel cow `The cow whose horns Kumar caught hold of' *kumaar maaTTe piDicca kombu Kumar cow-acc catch-pa-rel horn `The horns which Kumar caught hold of'
88
E. Annamalai
This gives additional evidence that the ®rst accusative in (62) is genitive and can have an alternative CM of accusative when it is the whole and the second accusative is a part of it. This is not the case with 61(a) in spite of its genitive alternate 61(c). Both nouns in 61(a) can be relativized. (65)
a.
kumaar paDam varenja poNNu Kumar picture draw-pa-rel girl `The girl who Kumar drew a picture of'
b.
kumaar oru poNNe varenja paDam Kumar a girl-acc draw-pa-rel picture `The picture which Kumar drew of a girl'
Therefore, the double accusative in 61(a) cannot be explained as an instance of alternative case assignment from the genitive. In another set of double accusative constructions, the ®rst accusative alternates with the dative. (66)
a.
kumaar viiTTe veLLe aDiccaan viiTTukku Kumar house-acc white beat house-to `Kumar smeared the house white' `white paint to the house' i.e., `Kumar white washed the house'
b.
kumaar raajaave naamam pooTTaan raajaavukku Kumar Raja-acc mark on forehead put Raja-to `Kumar put a forehead mark on Raja' i.e., `Kumar cheated Raja'
It is possible to say that, as with the genitive in (62), the dative noun is alternatively assigned accusative case.
19.
Conditions for Alternating Case
It may be noted that the alternative case assignment of accusative is possible under the condition of the second noun losing its ARG status and becoming part of the predicate, making it a Complex Predicate. That this condition holds is supported by the fact that the second noun in a double accusative cannot be a phrase (i.e., cannot take a modi®er) (cf. 67a), a
Case and the Argument Structure of Tamil
89
compound (cf. 67b), or plural; that it does not take CM (cf. 67c); that it does not allow other nouns of the same semantic class (cf. 67d) and that it does not scramble or relativize. (67)
a.
kumaar raajaavukku periya naamam pooTTaan *raajaave Kumar Raja-to big forehead mark put Raja-acc `Kumar put a big forehead mark on Raja' i.e. `Kumar cheated Raja badly'
b.
kumaar viiTTukku veLLekalar aDiccaan *viiTTe Kumar house-to white colour beat `Kumar smeared white paint to the house'
c.
kumaar raajaavukku naamatte pooTTaan *raajaave Kumar Raja-to fore head mark put Raja-acc `Kumar put a forehead mark on Raja'
d.
kumaar viiTTukku kaavi aDiccaan *viiTTe Kumar house-to sa¨ron (colour) beat house-acc `Kumar smeared sa¨ron (colour) to the house'
Semantically, CP has lexical (cf. 66a) or idiomatic (cf. 66b) meaning. The verb cannot have speci®c semantic content; saattu `apply', which has semantic content, is less acceptable than pooTu `put' in 66(a) with double accusative. (68)
kumaar raajaavukku naamam saattunaan *raajaave fore head mark applied Kumar Raja-to Raja-acc `Kumar applied a fore head mark on Raja'
It was shown earlier that compound verbs like kalyaaNam paNNu `do marriage, i.e., marry' do not have the properties of CP and the noun in them can be modi®ed and compounded. But the noun cannot be case marked when another accusative noun is present, as shown earlier, nor can it be passivized with clear-cut acceptability. This suggests that these compound verbs may be treated as CP.
90 (69)
(70)
E. Annamalai a.
kumaarukku kalyaaNam paNNunaanga kalyaaNatte Kumar-to marriage did marriage-acc `They performed marriage to Kumar'
b.
kumaarukku deeviye
a.
b.
c.
kalyaaNam paNNunaanga *kalyaaNatte Kumar-to Devi-acc marriage did marriage-acc `They performed the marriage of Kumar with Devi' i.e., `They married Devi to Kumar' kumaarukku kalyaaNam seyyappaTTadu Kumar-to marriage do-pass-pa-png `Marriage was performed to Kumar' ??kumaarukku deeviye kalyaaNam seyyappaTTadu Kumar-to Devi-acc marriage do-pass-pa-png `Marriage was performed to Kumar with Devi' i.e., `Kumar was married to Devi' kumaarukku deevi kalyaaNam seyyappaTTaaL Kumar-to Devi marriage do-pass-pa-png `Devi was married to Kumar'
It is clear that CP does not share its properties universally. It is therefore not necessary that CP must possess all its features for the occurrence of double accusative. Further, CP is not a universal condition for the occurrence of double accusative as (57, 58) and (62), where the predicate is not CP, show. A double accusative arises through alternative case assignment of accusative under certain semantic conditions like part-whole relationship between the two accusative nouns or predicates of communication (sollu `say', keeLu `ask')10. The other apparent double accusative is not a double accusative and the second noun is part of the CP. The double accusative with picture nouns like paDam vare `draw picture' remains unexplained. With this exception and explanation, the principle that a predicate does not license more than one mark of the same case can be maintained. 20.
Double Occurrence with Special Predicates
The derived causative predicate is a CP and it licenses the ARGs Theme / Object and Agent in addition to Causer / Instigator. It licenses two accusations; Agent is assigned accusative case.
Case and the Argument Structure of Tamil (71)
a.
ammaa piLLekki nelaave kaaTTunaanga mother child-to moon-acc showed `The mother showed the moon to the child'
b.
ammaa piLLeye nelaave paakkavaccaanga mother child-acc moon-acc see-inf-keep-pa-png `The mother made the child see the moon'
91
Tamil has no separate CM for the ARG Causer and takes the CM of Agent, viz. nominative. The Agent in turn takes an accusative CM. This is an instance of case syncretism sensitive to the semantic context. There are sentences in Tamil with two nominatives. As previously mentioned, a noun used as a predicate, as in (72), does not carry case at all. The predicate only licenses case. In 72(c), mandiri aagu `become a minister' is the predicate and not the verb alone and therefore the question of the case of the noun in it does not arise (72)
a.
kumaar oru mandiri Kumar a minister `Kumar is a minister'
b.
kumaar mandiri ille Kumar minister not `Kumar is not a minister'
c.
kumaar mandiri aanaaru Kumar minister became `Kumar became a minister'
In sentences like (73), the second noun has nominative case relation with the nominal or verbal predicate. (73)
a.
tangam vele adigam gold price high `Gold is high in price'
b.
tangam graam nuuru ruubaa11 gold gram 100 rupees `Gold is 100 rupees per gram'
c.
tangam vele korenjirukku12 gold price reduce-pa-is `Gold has become less in price'
There are other sentences with double nominatives. Before we consider them, we should identify instances of two nouns with no CM of which the second is accusative. It is accusative because the accusative CM can be optionally present and the predicates are transitive13.
92 (74)
E. Annamalai a.
kumaar tale nimundaan taleye Kumar head raised head-acc `Kumar raised (his) head' (Idiomatically, `Kumar was proud')
b.
kumaar tale kuninjaan taleye Kumar head bent head-acc `Kumar bent (his) head' (Idiomatically, `Kumar was ashamed')
c.
kumaar mogam kooNunaan mogatte Kumar face twisted face-acc `Kumar twisted (his) face' (Idiomatically, `Kumar expressed dissatisfaction')
d.
kumaar kay viriccaan kayye Kumar hand spread head-acc `Kumar spread (his) hand' (Idiomatically, `Kumar let down')
e.
kumaar kaN muuDunaan kaNNe Kumar eye closed eye-acc `Kumar closed (his) eyes' (Idiomatically, `Kumar ignored')
f.
kumaar kaadu kutturaan *kaade Kumar ear pierces ear-acc `Kumar pierces (your) years' (Idiomatically, `Kumar fooled')
It may be noted that the absence of accusative CM is preferred in the idiomatic sense of N V above. The sentences in (74) have one nominative and one accusative and they are not instances of two nominatives14.
Case and the Argument Structure of Tamil
93
The following sentences with intransitive predicate have double nominatives15. (75)
a.
kumaar kaNNu kalanguraan Kumar eye moisten-pr-png `Kumar has his eyes ®lled with tears'
b.
kumaar manasu taLanduTTaan Kumar mind wear out-pa-png `Kumar lost heart'
c.
en kaNNu paarve mangudu my eye right lose-lustre-pr-png `My eye is losing sight'
d.
kumaar ale mooduraan Kumar wave hit-pr-png `Kumar is perplexed'
The N V in 75(d) is a CP as indicated by its idiomatic meaning. It remains to be examined whether the N V in other sentences is also a CP or whether the double nominative in them comes from another source. They have properties of a CP in that the N is not modi®ed, pluralized, relativized or scrambled, but the V has semantic content. 75(a±c) have alternative constructions with the genitive case as in 76(a±c). (76)
a.
kumaar kaNNu kalanguccu Kumar-gen eye moistened `Kumar's eyes ®lled with tears'
b.
kumaar manasu taLandurucci Kumar-gen mind wore out `Kumar's mind got dampened'
c.
en kaNNu paarve mangudu my eye-gen sight loses lustre `My eye sight is getting dim'
It may be noted that the second noun is a part of or an inherent property of the ®rst noun. Parallel to (62), 75(a±c) may be taken to have a genitive phrase as subject and the ®rst noun has the alternative case assignment of nominative. When the apparent double case constructions are analyzed as the ®rst noun carrying another case or the second noun being part of a complex predicate, the principle that a predicate licenses only one instance of a case is maintained. The instances of double case are then a result of alternative
94
E. Annamalai
case assignment to the same argument or of same case assignment to di¨erent arguments. 21.
Case Assignment to Nouns Revisited
We shall now take up the problem of case assignment to nouns in a sentence. The proto-typical correlation between ARG, GF and case as AgentSubject-Nominative, Theme / Patient-Object-accusative, though prevalent in languages, is neither universally valid for all languages nor valid in a speci®c language for all its predicates. The predicate licenses ARGs and it permits an ARG to take a certain GF. There is not a one-to-one relationship between ARG and GF as mentioned above. Certain predicates permit one or the other ARG like Agent, Theme / Object or Instrument to be the GF SU. The verb tera `open' is such a predicate and, given a hierarchy of ARGs, the ARG goes up in the hierarchy. It is again the predicate that permits one or the other case to be a GF like SU. The SU could be nominative, dative or instrumental case in Tamil depending on the predicate. Thus, case assignment is done by sets of predicates probably de®ned by shared semantic properties and modulated by the preferential hierarchy of GFs for cases, which in turn is modulated by the preferential hierarchy of SRs for GFs. Oblique GFs are determined by their direct semantic relation with the predicate. Consequently, case assignment to nouns carrying oblique GFs will represent this semantic relation. It may be hypothesized that ADJs (i.e., nouns which are not ARGs) necessarily have an oblique GF. Then the SR of an ADJ will be isomorphic with its oblique GF and the case will be assigned semantically in relation to the meaning of the SR. The subcases, like di¨erent locatives, will relate to speci®c manifestations of the SR. With regard to primary GFs, the speci®c semantic properties of the predicate determine the case. It is necessary to identify the primary GFs by their syntactic properties. To take, for example, SU in Tamil, the morphological features of agreement and zero CM are not useful in identifying SU because they arise after SU is identi®ed. The con®gurational position is not useful because in a ¯at structure there is no con®gurational di¨erence between SU and O. The position in linear order is not useful either because of freedom of word order. The only syntactic property that may uniquely identify a SU is being antecedent to anaphor. After SU is identi®ed, the meaning of the predicate will assign nominative, dative or instrumental case to it, subject to the condition that the ARG carrying the particular SR is permitted to become SU by the grammar. The principles of the Tamil grammar that allow SRs to be GFs and the semantic properties of predicates that assign cases to GFs need to be
Case and the Argument Structure of Tamil
95
studied in greater detail. This paper only outlines the relevance of issues in case marking for such a study. Notes *
The research for this paper was carried out while I was a Visiting Professor at the Institute for the Study of Language and Cultures of Asia and Africa, Tokyo in 1995±96. I am grateful to the Institute for the support and freedom it provided to pursue my research. 1. This CM -TTe is di¨erent from kiTTe `near', though sometimes the latter may be a free variant of the former. (1)
a.
naan DaakTarTTe pooneen I doctor-prox went `I went to the doctor'
b.
naan DaakTargiTTe pooneen I doctor-near went `I went near the doctor' `I went to the doctor'
It is also di¨erent from -TTe, which alternates with the instrumental CM -aale. (2)
naan kumaare
kayyaale aDicceen kayTTe I Kumar-acc hand-inst hit `I hit Kumar with my hand'
2. The dative CM may be absent in certain constructions, which are similar to constructions with complex predicate, even when the verb does not indicate physical motion. (3)
enakku on peeru nyabagattukku / nyaabagam vandudu I-to your name memory-to/ came `Your name came to my mind'
3. It is possible that the second noun in (29) is ablative which is never explicit as this sentence can be interpreted as `we went / walked from house to house'. This is supported by (4). (4)
nannga viiTTukku viiDu paNam vaangunoom we house-to house money got `We collected money from every house.'
4. In (5) mandiri `minister' does not take CMÐit is part of predicate. The noun with aagu `become' perhaps makes a complex predicate, in which case the question of it being an ARG does not arise. (5)
kumaar mandiri aanaan Kumar minister became `Kumar became a minister'
96
E. Annamalai
5. 6(a) is an impersonal sentence and is equivalent semantically to the passive. The nominative is generic third person plural and it is absent but is recoverable from the agreement marker. This contrasts with passive in 6(b), whose agreement marker is third person neuter singular to agree with the nominative. (6)
a.
paLLikkuuDam eppa terakraanga? school open open-pr-png `When do they open the school?'
b.
paLLikkuuDam eppa terakkudu? school when open-pr-png `When does the school open?'
6. If one takes the in®nitive to be the embedded clause, muDi `to be able to' is the predicate of the matrix sentence and it does have a nominative case, though the predicate of the embedded sentence paDi `read' has null nominative coindexed with the instrumental. 7. If the noun is accusative, passivization must be possible. In the corresponding written Tamil, which admits the passive, passivization of the second noun seems possible, though not as acceptable as the passivization of the ®rst noun with accusative noun. (7)
a.
naan kuRai collappaTTeen I lack say-inf-pass-pa-png `I was spoken ill of'
b.
?ennai kuRai collappaTTatu I-acc lack say-inf-pass-past-png `Ill was spoken of me'
c.
ennaipaRRi kuRai collappaTTatu I-acc about lack say-inf-pass-pa-png `Ill was spoken about me'
8. Note that the syntactically comparable construction, enne paDam eDu `take a picture of me' does not allow genitive alternation, *en paDatte eDu `take my picture'. paDam eDu `take picture i.e., photograph' being a compound verb does not allow genitive. 9. The accusative construction, however, is di¨erent syntactically from the adverbial construction. The second noun can be relativized in the former but not in the latter as (8) shows. (8)
a.
b.
kaDavuL agalyaave kallaa aakkunaaru god Akalya-acc stone-adv made `God made Akalya into a stone' *kaDavuL agalyaave aakkuna kallu god Akalya-acc make-pa-rel stone `The stone which God made Akalya into'
Case and the Argument Structure of Tamil (9)
a.
kumaar engiTTe pattu ruubaa kaDan keeTTaan kaDanaa Kumar I-prox ten rupee loan asked loan-as `Kumar asked me a loan of ten rupees'
b.
kumaar engiTTe kaDan keeTTa pattu ruubaa Kumar I-prox loan ask-pa-rel ten rupee `The ten rupees which Kumar asked me as a loan'
c.
*kumaar engiTTe pattu ruubaa keeTTa kaDan Kumar I-prox ten rupee ask-pa-rel loan *`The loan which Kumar asked me of ten rupees'
97
10. There are instances of alternations with other cases whose predicate is not a verb of communication. It is not clear whether all such instances can be treated as having CP. (10)
a.
naan meeselerundu tuusiye taTTuneen I table-from dust-acc knocked `I removed the dust from the table'
b.
naan meeseye tuusi taTTuneen I table-acc dust knocked `I cleaned the table of the dust'
11. Note that the second noun could be dative instead of nominative (11)
tangam graamukku nuuru ruubaa gold gram-to 100 rupee `Gold is 100 rupees per gram'
12. Such sentences are di¨erent from sentences where the second noun is the head of a genitive phrase and this phrase is the subject. (12)
a.
tanga vele adigam gold-gen price high `The price of gold is high'
b.
tanga vele korenjirukku gold-gen price reduce-pa-is `The price of gold has become less'
13. Some of the verbs like nimuru `raise (a body part)', kuni `bend (a body part)', kooNu `twist (a body part)' do not have transitivity when relativized even in their non-idiomatic sense. (13)
kumaar nimunda taleyooDa vandaan kuninja Kumar raise-pa-rel head-with came bend-pa-rel `Kumar came with raised head' `bent'
98
E. Annamalai
The head noun has subject relation with the relative participle as in tale nimundu / kuninji `raising / bending head'. This suggests that, in spite of the occurrence of accusative CM, they may not be transitive verbs. This will be a problem for case assignment. 14. Some N Vs in (74) take a dative case noun in their idiomatic sense, which suggests that N is accusative. (14)
kumaar enakku kay viricciTTaan kayye Kumar I-to hand spread hand-acc `Kumar ditched me'
Some take another accusative case noun. (15)
kumaar enne kay kaZuviTTaan kayye Kumar I-acc hand washed hand-acc `Kumar washed me o¨ his hands'
15. 16(a) looks like double nominative, but it may be a postposed modi®er of the noun as in 16(b) with contrastive stress on the numeral. (16)
a.
en veele paadi muDinjidu my work half was over `Half of my job was over'
b.
en veele oNNu muDinjidu my work one was over `One work of mine was over'
References Annamalai, E. (2003). The Constituent Structure of Tamil. In Singh, Rajendra (ed.). Yearbook of South Asian Languages and Linguistics. Berlin: Mouton de Gruyter. Alsina, Alex, Joan Bresnan and Peter Sells (eds.) (1996). Complex Predicates. Stanford: Center for Studies in Language and Information Publications. Bresnan, Joan (ed.) (1982). The Mental Representation of Grammatical Relations. Cambridge, MA.: Massachusetts Institute of Technology Press. Ð (1982). Polyadicity. In Bresnan, Joan (ed). The Mental Representation of Grammatical Relations. Cambridge, MA.: Massachusetts Institute of Technology Press. Keenen, E. (1975). Towards a Universal De®nition of Subject. In Li, Charles N. (ed). Subject and Topic. Chicago: Chicago University Press. Lindholm, James (1976). Nested Case Relations and the Subject in Tamil. In Verma, Mahindra K (ed). The Notion of Subject in South Asian Languages. Madison: University of Wisconsin, South Asian Studies. Mohanan, Tara (1994). Argument Structure in Hindi. Stanford: Center for Studies in Language and Information Publications.
Case and the Argument Structure of Tamil
99
Panevora, J. (1978). Inner Participants and Free Adverbials, In Prague Studies in Mathematical Linguistics. Perlmutter, D. and P. Postal (1983). Some Proposed Laws in Basic Clause Structure. In Perlmutter, D. (ed), Studies in Relational Grammar, Vol 1. Chicago: Chicago University Press. Verma, Mahindra K. (1976). The Notion of Subject in South Asian Languages. Madison: University of Wisconsin, South Asian Studies. Wechsler, Stephen (1995). The Semantic Basis of Argument Structure. Stanford: Center for Studies in Language and Information Publications.
The Semantics of Bangla Compound Verbs* SOMA PAUL
Compound Verb (CV) constructions in Indo-Aryan languages are generally made up of two verbal elements. In these constructions, the ®rst verbal element, V1, provides a semantic base or frame and the second, V2, which bears the in¯ection, pro®les or highlights a subpart of the event expressed by V1. This paper examines the details of the contribution of V2 to CV's in Bangla. 1. Introduction A Bangla1 compound verb (henceforth CV) sequence is a two-member construction in which the ®rst member (V1) chooses between the usual conjunctive participial form /±e/ and the (rare, for this construction) in®nitive form /±te/ while the second member (V2) bears the in¯ection. For example, Bangla uÿ h-lo (1) mee-ÿ a heS -e girl-CL2 laugh-CP rise-3PT `The girl burst into laughter' (2)
ge-le kQno? tumi mitu-ke ktha-ÿ a bol-te you Mitu-OBJ word-CL say-INF go-2PT why `Why did you have to say this to Mitu?'
Hindi-Urdu (3) l«} ki: ro p«} i: girl cry fall-3PT `The girl burst into tears' Assamese (4) suali-joni hah-i uÿ hi-la girl-CL laugh-CP rise-3PT `The girl burst into laughter'
102 Oriya (5)
Soma Paul jhi-ÿ i h s-i uÿ hi-la girl-CL laugh-CP rise-3PT `The girl burst into laughter'
The boldfaced sequences of verbs in (1)±(5) are the CVs. I label these expressions as periphrastic compounds. Periphrastic compounds are distinct, on the one hand, from synthetic expressions such as kor-echilam (``do-1PT PFT'') `(I) had done (it)' in which one word-form corresponds to one meaning and on the other hand, from those multi-word expressions in which the second verb either plays the role of an auxiliary or represents an independent event. The V2s within periphrastic compounds are not as completely grammaticalized3 as auxiliaries. When a V1 participates in periphrastic compounding, the semantics of the V1 is extended or modi®ed and the resultant CV acquires a meaning subtly distinct from that of the V1 itself. For example, let us consider the Bangla intransitive verbal root ghumono `sleep'. The verbal noun form of verbs is used here as the citation form. The verb ghumono `sleep' selects the V2s p}a `fall', neo» a `take' and jao» a `go'. Column 3 in the following table represents the CVs. The semantic overtone that the V2s add to the meaning of ghumono `sleep' is summarized in column 44: The following sentences illustrate the use of the CVs listed in column 3 of table 1. Henceforth all example sentences will be taken from Bangla: (6)
a.
mee-ÿ a ghumi-e po} -lo girl-CL sleep-CP fall-3PT `The girl fell asleep'
b.
mee-ÿ a khanikkhon ghumi-e ni-lo girl-CL for a while sleep-CP take-3PT `The girl took a nap for a while'
c.
mee-ÿ a Saradin Sudhu ghumi-e gQ Q-lo girl-CL entire day only sleep-CP go-3PT `The girl kept on sleeping all day long'
These CVs demonstrate that the core meaning of ghumono `sleep' is preserved in every case. However, the semantics of CVs becomes much more focused and restricted than that of its V1 associate. For example, the CV ghumie p}a `fall asleep' indicates a resultative viewpoint. When the speaker wants to focus on the immediate e¨ect of the event and not extend the focus over the span of the resultant state, he/she will most likely use the CV variant ghumie p}a `fall asleep'. The CV ghumie jao» a `continue to sleep', on the
The Semantics of Bangla Compound Verbs
103
Table 1. Semantics of a compound verb is the modi®cation or extension of its V1 associate V1 `sleep'
V2
CV
Semantic overtone
ghumono ghumono ghumono
p}a `fall' neo» a `take' jao» a `go'
ghumie p}a ghumie neo» a ghumie jao» a
A sense of immediateness Participant is self-bene®ciary Extendedness of the event
other hand, conveys the sense that the event of sleeping continues over a period of time. I presume that the V2 participant within CVs determines the focus. I explain the phenomenon in terms of pro®ling, a concept adopted from Langacker's pro®le-base distinction which I will be discussing in the next section. My contention is the following: Every V1 has an associated base or frame that represents speci®cation of an unmarked5 event type. The composition of a V1 with a V2 implies a focus on part of the V1 event. The speaker's choice and context in the discourse determine the focus. Compound verb construction is a lexical means to express the situation in focus linguistically. In this paper I identify two kinds of pro®ling: one is the lexical pro®ling of participants and the other is the supralexical pro®ling which includes the aspectual and temporal focusing on the V1 event6 by the V2s. I will illustrate them in section 3. In section 4, I will demonstrate that the pro®ling of V1 event by the V2s results in a new event type which is represented by the resultant CV. The syntactic behavior of the CVs might be di¨erent from that of their V1 associate.
2. Pro®ling I will mention here brie¯y how Langacker (1987) views the pro®le-base distinction in his cognitive model of perception. Langacker observes that human beings have a perceptual ability to recognize objects from various perspectives. The base is an underlying matrix, or set, of relevant cognitive domains that is required or evoked in comprehending a given expression. The profile, on the other hand, is the highlighted substructure within the base that the expression conceptually designates. Let us consider, for example, the English prepositions: `to' and `toward' vs. `from' and `away from'. These expressions pro®le di¨erent portions of the `source-path-goal' schema as shown in the following ®gure:
104
Soma Paul from
away–from
toward
to Figure 1. Pro®ling of di¨erent part of the source-path-goal schema by English prepositions
Langacker has commented that the meaning of an expression does not lie exclusively in either the pro®le or the base alone, but rather in the intimate interrelation of the former with the latter. It is important to realize that the base is just as much an integral part of semantic structure as the pro®le it contextualizes. Regarding the CV constructions under consideration, my assumption is that the event denoted by the V1 provides the base. When a V1 selects a V2, the semantics of the latter pro®les the base in the following two ways: 1. By highlighting the manner of involvement of the participant(s) engaged in the base event; and 2. By imposing temporal and aspectual focus on the event denoted by the resultant CV predicate The next section discusses the nature of these two kinds of profiling. 3. A V2 Pro®les the V1 Event The aforementioned expression `manner of involvement' is used as a cover term to include a range of abstract and complementary semantic concepts such as the following: a situation invoked versus a situation obtained, directedness of the result of the action towards the actor versus directedness away from the actor, directedness towards the deictic center and so on. Each of these abstract concepts de®nes a kind of manner in which the participant(s) is viewed as being involved in the event denoted by CVs. The theory of pro®ling as presented in this paper assumes that these restricted abstract senses are imported into the semantics of V1 by the V2s. When a V1 provides a V2 an appropriate base, the V2 in accordance to its semantic
The Semantics of Bangla Compound Verbs
105
requirement pro®les the relevant segment of the V1 event. As a result of this the semantics of a legitimate CV sequence is determined. The following table presents a summary of various manner of involvements of the participants in an event of banano `build': Table 2. A summary of various manner of involvements of the participants in an event of banano `build' V1 `build'
V2
CV
Semantic overtone
banano
deo» a `give'
banie deo» a
banano
neo» a `take'
banie neo» a
banano
rakha `keep'
banie rakha
E¨ect of the action directed towards a participant other than the actor E¨ect of the action directed towards the actor himself (self-bene®ciary) The result of the action is focused
The verb banano `build', like any other accomplishment verb, denotes a situation in which an actor performs some action and the action has a natural outcome. In case of ba}i banano `build a house', for example, the house is the entity to be built. Unlike verbs such as gchano `foist something on somebody', which distinctly entails that the actor acts upon a non-actor (patient) by persuading him/her to take the object transferred (theme), the situation denoted by the verb banano `build' does not indicate the person or entity (i.e., a bene®ciary) towards whom the e¨ect of the action is directed. That is, the builder can build a house for himself (as demonstrated in (7a)) or he can build a house for the bene®t of a receiver or bene®ciary as shown in (7b): (7)
a.
jonne ba}i-ÿ a bani-eche mheS nije-r Mahesh self-GEN for house-CL build-3PR PFT `Mahesh has built the house for himself'
b.
mheS ritur bana-len jonne ba}i-ÿ a house-CL build-3HON.PT Mahesh Ritu-GEN for `Mahesh built the house for Ritu'
The V2s deo» a `give' and neo» a `take' remove this vagueness by categorically focusing on one particular direction. The CV banie deo» a ``build-CP give'' speci®es that the e¨ect of the action directed towards a non-actor entity. However, verbs such as banano `build' does not license any a¨ected entity and the following sentence is ill-formed: (8)
*mheS mol-ke Qkÿ a ba}i bana-lo Mahesh Amal-OBJ one-CL house build-3PT `Mahesh built Amal a house'
106
Soma Paul
Therefore the linguistic consequence of this verb's selecting the V2 deo» a `give' is that an extra bene®ciary argument, the profiled entity, appears in the sentence as shown in (9). Amal is the bene®ciary in (9): (9)
mheS mol-ke Qk-ÿ a ba}i bani-e di-lo Mahesh Amal-OBJ one-CL house build-CP give-3PT `Mahesh built Amal a house'
Like the verb banano `build' the following dyadic verbs become ditransitive when they select the V2 deo» a `give': kena `buy' ÿ aNano `hang up' kaÿ a `cut (vegetables)' chka `plan' pa}a `bring down'
khoÄja `search' raÄdha `cook' aÄka `draw' ana `bring' ko}a `scrape' ko}ano `collect' ÿ oka `copy' joÿ ano `collect' kaÿ a `cut lekha `write' (vegetables)' kucono `cut into pieces' maja `wash (utensils), scrub'
bhaja `fry' kra `do' g}a `build' bacha `select'
The CV sequence banie neo» a ``build-CP take'', on the other hand, pro®les the self-directedness (or self-bene®ciary) reading inherent in the semantics of the verb banano `build' as shown in (7a). The following sentence suggests that Mahesh bought the book for his own bene®t. (10)
mheS dokan theke boi»-ÿ a kin-e ni-len Mahesh shop from book-CL buy-CP take-3HON.PT `Mahesh bought the book from the shop (implying that Mahesh is self-bene®ciary)'
The following sentence exempli®es the CV banie rakha ``build-CP keep'': (11)
nekguli ba}i bani-e mheS Shor-e Mahesh city-LOC quite a few house build-CP rekh-echen keep-3HON.PR PFT `Mahesh has built quite a few houses in the city (in anticipation of future need)'
The V2 rakha `keep' focuses upon the resultant situation. I assume that verbs such as banano `build' correspond to an activity event, which entails a logical end-state for the participant that occupies the object position. In an unmarked situation, the event presumes that `the house' has come into existence. The situation denoted by the CV sequence banie rakha
The Semantics of Bangla Compound Verbs
107
``build-CP keep'' in (11) focuses on the stative eventuality accompanying the event of banano `build'. It conveys a sense of building a house ahead of time in anticipation of future need. Thus table 2 presents cases in which participant roles are pro®led in various manners by V2s. I call this kind of pro®ling as lexical profiling of participants. Table 3 illustrates pro®ling of the second type mentioned above. Here the V2s impose aspectual and temporal restriction on the event type denoted by V1: Table 3. V2 imposes temporal and aspectual focus on V1 event V1 `laugh'
V2
CV
Semantic Overtone
haSa haSa haSa
oÿ ha `rise' phQla `drop' jao» a `go'
heSe oÿ ha heSe phQla heSe jao» a
Focus on inception Focus on result Focus on continuity
The verb haSa `laugh' is an activity verb. It expresses a durative atelic event of `laughing' in which the actor is volitionally involved. The V2 jao» a `go' pro®les the durative part of the action and the CV heSe jao» a `continue to laugh' conveys that the event of laughing occurs over an interval. Therefore this verbal expression is compatible with the durative adverbials. In (12), the expression Qknaga}e `at a stretch, continuously' conveys duration: (12)
mee-ÿ a Qknaga}e heS -e ja-cche girl-CL at a stretch laugh-CP go-3PR CON `The girl continues laughing at a stretch'
The V2 oÿ ha `rise', on the other hand, pro®les the initial segment of the situation denoted by the V1 as a result of which the V1oÿ ha `rise' construction indicates an inceptive aspectual viewpoint. The inceptive focuses on the beginning of an event. As Smith (1991) observes, the receiver could reasonably infer, with no information to the contrary, that the event continues. This is illustrated in the following sentence. The verbal sequence V-te `inf' thaka `stay' in Bangla implies that the event continues. (13)
oÿ h-e ar haS-te-i thak-e mee-ÿ a heS -e girl-CL laugh-CP rise-3PR and laugh-INF-EMP stay-3PR `The girl bursts into laughter and goes on laughing'
The V2 phQla `drop' focuses the end segment of the event haSa `laugh'. It also emphasizes the fact that the event occurred despite e¨orts to the contrary. Since the CV heSe phQla `laugh inadvertently' indicates a resultative
108
Soma Paul
viewpoint, this CV, unlike the other variant heSe oÿ ha `burst into laughter' (see (13)), does not occur in a context that conveys that the event continues. I refer to this kind of temporal and aspectual pro®ling of V1 event as supralexical profiling. 4. Pro®ling Results in a New Event Type Depending on which segment (or subpart) of the V1 event is being focused as it uni®es with a V2, the resultant CV represents an independent event type. For example, the following case demonstrates that the CVs might be semantically distinct from their V1 constituent in terms of the feature duration. Verbs popularly known as accomplishment verbs represent an event type that includes a process and an outcome or change of state. The process advances in successive stages to the point of culmination. The endpoint denotes a change of state. Like activity verbs these verbs can also occur in the context of adverb of duration. For example, the sentences in (14) exhibit that the accomplishment verbs such as aÄka `draw', lekha `write' and banano `build' are compatible with the durative adverbials. The durative adverbials are underlined: (14)
a.
theke Sondhe porjonto chobi aÄk-lo ritu Skal picture draw-3PT Ritu morning from evening till `Ritu drew the picture from morning till evening'
b.
chele-ÿ a Saradin dhore Qkÿ a ciÿ hi likh-che boy-CL whole day for one letter write-3PR CONT `The boy is writing a letter for the whole day'
c.
mheS babu tinbchor dhore ba}i-ÿ a Maheshbabu three years for house-CL bana-cchen build-3HON.PR CONT `Maheshbabu is building the house for three years'
The post-positional phrases (. . . theke . . . porjonto `from . . . till . . .' in (a) and (tinbchor) dhore `for (three years)' in (c)) express duration. When the accomplishment verbs (as illustrated in (14)) select the V2 phQla `drop', the end segment of the V1 event is only pro®led. Consequently the situation in focus acquires an intrinsic feature of instantaneousness. These CVs are not, therefore, compatible with durative adverbials as illustrated in (15): (15)
a.
*ritu Qkghnÿ a dhore chobi-ÿ a eÄk-e phel-lo Ritu one hour for picture-CL draw-CP drop-3PT `Ritu completed drawing the picture for one hour'
The Semantics of Bangla Compound Verbs
109
b.
*chele-ÿ a Saradin dhore Qkÿ a ciÿ hi likh-e boy-CL whole day for one letter write-CP phel-che drop-3PR CONT `The boy is completing writing a letter for the whole day'
c.
bani-e phel-lo *mheS tinbchor dhore ba}i-ÿ a house-CL build-CP drop-3PT Mahesh three years for `Maheshbabu completed building the house for three years'
However, when the accomplishment verbs select the V2 jao» a `go', the process segment of the V1 event is focused as a result of which the resultant CV event becomes atelic in nature. Smith (1991: 157) notes that completive adverbials are compatible with telic event and odd with atelics. In Bangla, the moddhe-PP `within-PP' phrase and the adverbial phrase such as Qkdine `within one day' represent the completive adverbials that locate an event at an interval, during which the event is completed. The CVs that have the verb jao»a `go' as the second constituent (illustrated in (16a) and (17a)) are not compatible with adverb of completion (the phrases underlined in (16) and (17)) even though their V1 associates are, as exempli®ed in ((16b) and (17b)): (16)
a.
b.
(17)
a.
b.
moddhe chobi-ÿ a eÄk-e *ritu Qkghnÿ a-r Ritu one hour-GEN within picture-CL draw-CP gQ Q -lo go-3PT `Ritu kept on drawing the picture within one hour' ritu Qkghnÿ a-r moddhe chobi-ÿ a aÄk-lo Ritu one hour-GEN within picture-CL draw-3PT `Ritu drew the picture within one hour' moddhe ba}i-ÿ a bani-e *mheS tinbchor-er Mahesh three years-GEN within house-CL build-CP gQ Q -lo go-3PT `Mahesh kept on building the house within three years' mheS babu tinbchor-er moddhe ba}i-ÿ a Maheshbabu three years-GEN within house-CL bana-len build-3HON.PT `Maheshbabu built the house within three years'
110
Soma Paul
5. Conclusion The speaker's choice and context in the discourse sometimes require an emphasis to be laid on part of the event denoted by a simple verb and some semantic nuances to be added to the original meaning of the verb. This paper proposes that compound verb constructions, or periphrastic compounds, are the lexical means to express the situation in focus. The semantics of V2s pro®le the base events denoted by V1s by highlighting the manner of involvement of the participants in the event and/or by imposing temporal and aspectual focus on the event. Thus the resultant compound verb represents a new event type. Notes * I am grateful to Professor Probal Dasgupta and Professor Gautam Sengupta for his invaluable guidance. 1. Bangla (also known as Bengali) is an Indo-Aryan language spoken in Bangladesh and the eastern region of India. 2. The following abbreviations are used in this article: CL: classi®er CP: conjunctive participle GEN: genitive INF in®nitive OBJ: objective case PR: present
CON: continuous EMP: emphatic HON: honori®c LOC: locative PFT: perfect PT: past
3. Sarkar (1975) elucidates Porizka's perception of grammaticalizationÐa stripping of the main dictionary meaning from the vector verb in order to reduce it to the role of `aspective'. 4. There is a great deal of discussion available in the literature regarding the semantics of V2. The following are some references related to the works on Indo-Aryan compound verb structure: Hook (1974), Sarkar (1975), Dasgupta (1989), Abbi (1991), Mohanty (1992), Bashir (1993) and Butt (1995). 5. Carlota Smith (1991) discusses markedness and aspectual choice in her book titled `The Parameter of Aspect'. A speaker uses conventional event types and aspect information in a predictable way; that is the way these events and aspects are meant to be used in the discourse. Smith calls it unmarked choice of the speaker. On the other hand, a speaker makes marked aspectual choices or event type choice to convey emphasis of some kind and also for various pragmatic reasons. 6. Here and elsewhere I use the phrase `V1 event' to mean `the event denoted by the V1' which is technically more accurate.
References Abbi, Anvita (1991). Semantics of explicator compound verbs. South Asian Languages, Language Sciences 13:2. 161±180.
The Semantics of Bangla Compound Verbs
111
Bashir, Elena (1993). Causal chains and compound verbs. Complex Predicates in South Asian Languages, ed. by M. K. Verma. New Delhi: Manohar Publishers and Distributors. Butt, Miriam (1995). The Structure of Complex Predicates in Urdu. Stanford University: Doctoral dissertation. Dasgupta, Probal (1989). Projective Syntax: Theory and Applications. Pune: Deccan College P. G. and Research Institute. Hook, Peter (1974). The Compound Verbs in Hindi. The Michigan Series in South and Southeast Asian Language and Linguistics. The University of Michigan. Langacker, Ronald (1987). Foundations of Cognitive Grammar, 1. Stanford: Stanford University Press. Mohanty, Gopabandhu (1992). The Compound Verbs in Oriya. Deccan College Post-Graduate and Research Institute, Pune: Doctoral dissertation. Sarkar, Pabitra (1975). Aspects of Compound Verbs in Bengali. Chicago University: Unpublished M.A. dissertation. Smith, Carlota (1991). The Parameter of Aspect. The Netherlands: Kluwer Academic Publishers.
Discourse Context and Word Order Preferences in Hindi* SHRAVAN VASISHTH
Discourse context has been argued to be the main factor responsible for increased processing di½culty in non-canonical order sentences: if appropriate discourse context is provided (the argument goes) both canonical and non-canonical order sentences are equally easy to process. This research suggests that this generalization may not be true across languages: the distance between arguments and verbs could a¨ect the ease with which the former can be integrated with the latter, and su½ciently increasing this distance makes processing di½cult, regardless of discourse context. 1
Introduction
Let us refer to any claim that increasing the distance between dependents and heads (e.g., arguments and verbs, respectively) results in increased processing di½culty at the verb as the distance hypothesis; for ease of exposition, we can de®ne distance as a linear function of the number of intervening words. Early Immediate Constituents (Hawkins, 1998) and Discourse Locality Theory (Gibson, 2000) propose variants of this distance hypothesis. In order to understand the rationale behind this claim, ®rst consider the Hindi single center embedding in (1). (1)
Siitaa-ne Hari-ko kitaab khariid-nekaa aadesh-diyaa Sita-erg Hari-dat book buy-inf ordered `Sita ordered Hari to buy a book.'
This is an instance of a control construction: as shown in Figure 1, the subject noun phrase (NP2) of the innermost verb khariid-nekaa, `to buy' is the indirect object of the main verb (here aadesh-diyaa, `ordered') which is one clause higher. As a result, this noun phrase is assigned the dative case marker -ko from the higher clause. Note also that this structure is a center embedding construction because the clause S2 is embedded within the main clause S1.
114
Shravan Vasishth S1
NP1
NP2
S2
V1
PRO
NP3
V2
Figure 1. Schematic representation of the single center embedding in (1).
In this structure, the distance between the verb V2 and its two arguments NP2 and NP3 can be varied in several di¨erent ways. Consider the situation where the indirect object NP2 (2a) or the direct object NP3 is fronted (2b). (2)
a.
Hari-ko Siitaa-ne kitaab khariid-nekaa aadesh-diyaa Hari-dat Sita-erg book buy-inf ordered `Sita ordered Hari to buy a book.'
b.
Kitaab Siitaa-ne Hari-ko khariid-nekaa aadesh-diyaa book Sita-erg Hari-dat buy-inf ordered `Sita ordered Hari to buy a book.'
Here, Early Immediate Constituents predicts that increasing distance (2b,c) between dependents and heads will result in increased processing dif®culty at the inner verb: more words need to be processed before the head of the embedded clause is recognized. That is, processing di½culty at the innermost verb is predicted to be greater in the direct-object and indirectobject fronting cases compared to the canonical order sentence in (1), but no di¨erence in processing di½culty is expected at the verb between the fronted indirect-object versus fronted direct-object cases. Discourse Locality Theory makes similar predictions, but the reason here is di¨erent: the number of new discourse referents between the head and dependents increases in the noncanonical orders and, since new discourse referents consume memory resources, integrating the fronted element with the verb is di½cult. Moreover, Discourse Locality Theory predicts that the processing cost at the innermost verb in the fronted direct-object sentence (2b) is higher than in the fronted indirect-object sentences (2a). In directobject fronting, there are two new discourse referents between the fronted NP and the verb (as opposed to zero new discourse referents in the canonical case), whereas in indirect-object fronting there are two new discourse referents intervening between the fronted indirect object and the verb (as opposed to one new discourse referent in the canonical case). That is, Discourse Locality Theory predicts that processing di½culty at the innermost
Discourse Context and Word Order Preferences in Hindi
115
verb will be greater in the fronted cases compared to the canonical case, and greater in the fronted direct-object sentence compared to the fronted indirect-object sentence. However, the discussion above ignores the well-known fact (Altmann & Steedman, 1988; Steedman, 2000; Kaiser & Trueswell, 2002; Weber & Neu, 2003) that appropriate discourse context can neutralize any increase in processing di½culty resulting from non-default word order, di¨erences in the number of presuppositions that must be computed or satis®ed, etc. Also, it is well-known that sentences with non-canonical order presented out of context are less acceptable in languages like English, German, Finnish, and Hungarian (see (HyoÈnaÈ & Hujanen, 1997) and references cited there). One should therefore not ®nd the claim surprising that non-canonical order sentences involve more processing cost when no preceding context is provided. What would be surprising is if an increase in processing di½culty occurs in spite of appropriate discourse context being present. This research demonstrates that non-default word orders can sometimes lead to greater processing di½culty irrespective of whether sentences are presented with appropriate context or not. One plausible explanation for this increased di½culty with non-canonical orders is increased argumenthead distance. 2
Previous Work
Vasishth (2003b) conducted two experiments in order to test the distance hypothesis. As these motivate the set of experiments that form the focus of this paper, I brie¯y discuss them in this section. The experiments used the non-cumulative self-paced moving window methodology (Just, Carpenter, & Woolley, 1982). A G3 Macintosh laptop running PsyScope (Cohen, MacWhinney, Flatt, & Provost, 1993) was used to present the materials to subjects. The task was to press the space bar in order to view each successive phrase; each time the space bar was pressed, the previous phrase would disappear and the next phrase would appear. Reading time (in milliseconds) was taken as a measure of relative momentary processing di½culty. A yes/no comprehension question was presented after each sentence; this was meant to ensure that subjects were attending to the sentences. Subjects with less than 70% correct responses were not included in the data analysis; typically this resulted in data from four or ®ve subjects being excluded in each experiment. The computer screen presentation was as described below with the aid of Figure 2. Each line in Figure 2 shows a particular stage during a trial. At the beginning of a trial (which could be a ®ller or stimulus sentence), the screen shows the uppermost line in Figure 2: a series of dotted lines separated by
116
Shravan Vasishth
Figure 2. Schematic view of moving window design.
spaces, with each line corresponding to a phrase in the sentence that will eventually be seen in that trial. Then the subject presses the space bar. The screen now displays the second line in the ®gure; at this point, only the ®rst phrase is visible. The subject then presses the space bar again, and the next phrase appears and the previous one disappears (see the third line in Figure 2). This procedure is repeated until the end of the sentence (marked by the period, a vertical line) is reached. After the entire sentence has been seen, a yes/no question is shown, to which the subject responds by pressing the frontslash (``/'') key for ``yes'', or the ``z'' key for ``no''. After the subject responds, the screen reverts to displaying the lines as in the uppermost line in Figure 2. This signals the start of a new sentence, and the procedure is repeated for the new sentence. The phrase segmentation of the stimuli is shown by the vertical bars between words/phrases in (3). The period was invariably presented as a separate segment. One experiment in (Vasishth, 2003b) investigated reading time di¨erences at the innermost verb resulting from increased distance between a verb (the innermost verb) and its dependents (its arguments) by fronting the indirect object; the stimuli were as in (3). (3)
a.
Riinaa-ne | Siitaa-ko | kitaab | khariid-neko | kahaa Rina-erg Sita-dat book buy-inf told `Rina told Sita to buy a book.'
b.
Siitaa-ko | Riinaa-ne | kitaab | khariid-neko | kahaa Sita-dat Rina-erg book buy-inf told `It was Sita who Rina told to buy a book.'
117
1000
1200
1400
1600
Indirect object fronted Canonical word order
800
Mean Reading Time (msec)
1800
Discourse Context and Word Order Preferences in Hindi
NP
NP
NP3
V2
V1
Position
Figure 3. Canonical versus IO-fronted order; raw RTs, with 95% con®dence intervals.
As shown in Figure 3, increasing distance resulted in a signi®cantly longer reading time (RT) at the innermost verb (V2) in the IO-fronted case compared to canonical order. This was as predicted by the distance hypothesis. In the second experiment the contrasting pairs were fronted direct object versus canonical order, as shown in Examples (4) below. (4)
a.
Riinaa-ne | Siitaa-ko | kitaab | khariid-neko | kahaa Rina-erg Sita-dat book buy-inf told `Rina told Sita to buy a book.'
b.
Kitaab | Riinaa-ne | Siitaa-ko | khariid-neko | kahii book Rina-erg Sita-dat buy-inf told `It was the book that Rina told Sita to buy.'
As illustrated in Figure 4, here too a longer reading time was observed at the innermost verb, V2, in the DO-fronted case. Recall that Discourse Locality Theory predicts that fronting direct objects should result in greater processing di½culty at the innermost verb compared to fronted indirect objects. Although the di¨erence was in the predicted direction, a t-test for reading times at V2 gave an inconclusive null result. These experiments ignored the fact that in Hindi non-canonical word order typically occurs due to an appropriate discourse context demanding it (Gambhir, 1981); even though Hindi is often described as a ``free word
Shravan Vasishth
700
800
900 1000
Direct object fronted Canonical word order
600
Mean Reading Time (msec)
1200
118
NP
NP
NP
V2
V1
Position
Figure 4. Canonical versus fronted DO; raw RTs with 95% con®dence intervals.
order'' language, there is always a discourse reason for word order variation, such as altered topic and/or focus, emphasis, discourse salience etc. For this reason, two self-paced reading studies were carried out to investigate the e¨ect of discourse context on processing di½culty of canonical and non-canonical order sentences. These are discussed next. The same stimuli were chosen as in the original experiments, although di¨erent ®ller sentences were used. 3 3.1
Discourse Context and Noncanonical Word Order Experiment 1
A self-paced reading experiment was carried out in order to test the e¨ect of discourse context on fronted-IO sentences. The stimuli were as shown in (5). Before each stimulus sentence, subjects saw a question whose focus was either (i) the innermost VP (e.g., What did Mary tell John?, where the appropriate answer would be Mary told John X, X a VP like to buy a book), or (ii) the entire sentence (What happened?). The questions were presented in their entirety (not word-by-word). In the VP-focus sentences, the word order of the ®rst and second NP was identical to that in the target sentence: canonical order sentences were preceded by questions with NP1 and NP2 in canonical order, and IO-fronted sentences (NP2 followed by NP1) had the word order NP2 NP1. I will refer to the VP-focus question type as salient context and the S-focus question as the non-salient context.
Discourse Context and Word Order Preferences in Hindi (5)
119
a.
Rita-ne Ravi-ko kyaa kahaa? Rita-ne Ravi-ko kitaab Rita-erg Ravi-dat what said? Rita-erg Ravi-dat book khariid-neko kahaa buy-inf told `What did Rita say to Ravi? Rita told Ravi to buy a book.'
b.
Ravi-ko Rita-ne kyaa kahaa? Ravi-ko Rita-ne kitaab Ravi-dat Rita-erg what said? Ravi-dat Rita-erg book khariid-neko kahaa buy-inf told `What did Rita say to Ravi? Rita told Ravi to buy a book.'
c.
Kyaa hua? Rita-ne Ravi-ko kitaab khariid-neko What happened? Rita-erg Ravi-dat book buy-inf kahaa told `What happened? Rita told Ravi to buy a book.'
d.
Kyaa hua? Ravi-ko Rita-ne kitaab khariid-neko What happened? Ravi-dat Rita-erg book buy-inf kahaa told `What happened? Rita told Ravi to buy a book.'
As shown in Figure 5, a contrast analysis showed that when a salient context is provided in canonical versus non-canonical order sentences, a IO fronted vs. canonical, context provided
Mean Reading Time (msec) 600 700 800 900
IO fronted Canonical
NP
NP
NP3 Position
V2
V1
Figure 5. Experiment 1, conditions (a) versus (b) in example (5), with 95% con®dence intervals.
120
Shravan Vasishth Table 1. Experiment 1, reading times (msecs) at V2 and V1, conditions (a) versus (b) in example (5) RTs at verb
(5a)
(5b)
F(1,47)
p-value
V2 V1
709 836
759 886
1.69 0.66
0.1998 0.4219
IO fronted, context varied
600
Mean Reading Time (msec) 800 1000 1200
What happened? NP2ko NP1ne Q V?
NP2
NP1
NP3 Position
V2
V1
Figure 6. Experiment 1, conditions (b) versus (d) in example (5), with 95% con®dence intervals.
signi®cant increase is not observed at V2 or in the spillover region V1 (although this is an inconclusive null result, it does contrast with Vasishth's ®rst experiment, where such an increase was observed). Table 1 shows the reading times in milliseconds at V2 and the spillover region V1 for conditions (a) and (b). Moreover, as shown in Figure 6, in IO-fronted sentences, reading time in the spillover region V1 is signi®cantly longer when no salient discourse context is present. This shows that discourse context indeed determines processing di½culty in fronted IO sentences, as predicted by existing research such as (Kaiser & Trueswell, 2002), (Weber & Neu, 2003). Table 2 shows the reading times at V2 and V1 for conditions (b) and (d). Here, the RT di¨erence approaches signi®cance in the region V2, but is signi®cantly di¨erent in the spillover region V1.
Discourse Context and Word Order Preferences in Hindi
121
Table 2. Experiment 1, reading times (msecs) at V2 and V1, conditions (b) versus (d) in example (5) RT at verb
(5b)
(5d)
F(1,47)
p-value
V2 V1
759 886
820 1036
3.95 5.39
0.0531 0.0250
In sum, the present results, taken together with Vasishth's ®rst experiment, suggest that providing appropriate discourse context can neutralize processing di½culty that occurs due to non-canonical order. 3.2
Experiment 2
In this self-paced reading study, the e¨ect of discourse context on canonical versus direct-object fronted sentences was compared. The preceding salient and non-salient contexts had essentially the same structure as in Experiment 1, with one important di¨erence: When the direct object was to be presented as a fronted element in the target sentence, it was fronted in the context question as well. This was done in order to topicalize the fronted direct object in the context question, so that its appearance in the target sentence as a fronted element would be felicitous. The stimulus examples are shown below. (6)
a.
Rita-ne Ravi-ko kyaa kahaa? Rita-ne Ravi-ko kitaab Rita-erg Ravi-dat what said? Rita-erg Ravi-dat book khariid-neko kahaa buy-inf told `What did Rita say to Ravi? Rita told Ravi to buy a book.'
b.
Kitaab ka Rita-ne kyaa kiyaa? kitaab Rita-ne book gen Rita-erg what did? book Rita-erg Ravi-ko khariid-neko kahii Ravi-dat buy-inf told `What did Rita do about the book? Rita told Ravi to buy the book.'
c.
Kyaa hua? Rita-ne Ravi-ko kitaab khariid-neko what happened? Rita-erg Ravi-dat book buy-inf kahaa told `What happened? Rita told Ravi to buy a book.'
122
Shravan Vasishth d.
Kyaa hua? kitaab Rita-ne Ravi-ko khariid-neko what happened? book Rita-erg Ravi-dat buy-inf kahii told `What happened? Rita told Ravi to buy a book.'
As Figure 7 shows, even when appropriate context is provided, the mean RT at the verb V2 is signi®cantly longer in the DO-fronted case than in the canonical order sentence. This result is interesting because it goes against the assumption in the literature that processing di½culty in non-canonical order sentences is due to the absence of discourse context (Kaiser & Trueswell, 2002). Table 3 shows the reading times at V2 and the spillover region V1 for conditions (a) and (b). Furthermore, in DO-fronted sentences, RTs are signi®cantly faster if appropriate discourse context is present (Figure 8), but only in the spillover DO fronted vs. canonical, context provided
Mean Reading Time (msec) 600 700 800 900 1000
DO fronted Canonical
NP
NP
NP Position
V2
V1
Figure 7. Experiment 2, conditions (a) versus (b) in example (6), with 95% con®dence intervals. Table 3. Experiment 2, reading times (msecs) at V2 and V1, conditions (a) versus (b) in example (6) RT at verb
(6a)
(6b)
F2(1,43)
p-value
V2 V1
709 836
759 886
10.19 1.05
0.0027 0.3120
Discourse Context and Word Order Preferences in Hindi
123
Mean Reading Time (msec) 800 1000 1200 1400
DO fronted, context varied
600
What happened? NP3"of" NP2ne Q V?
NP3
NP1
NP2 Position
V2
V1
Figure 8. Experiment 2, conditions (b) versus (d) in example (6), with 95% con®dence intervals. Table 4. Experiment 2, reading times (msecs) at V2 and V1, conditions (b) versus (d) in example (6) RT at verb
(6b)
(6d)
F2(1,43)
p-value
V2 V1
759 886
820 1036
1.88 4.51
0.178 0.03
region V1. Table 4 shows the reading times at V2 and V1 for conditions (b) and (d). 4
A Summing Up
The two experiments suggest that although discourse context does a¨ect the processing of non-canonical order sentences, the extent to which it does so depends on the distance between arguments and heads: if this distance is relatively small (as in Experiment 1) the absence/presence of appropriate discourse context is the deciding factor in determining processing di½culty; however, if the distance is increased (Experiment 2), discourse context cannot neutralize the e¨ect of this increased distance. However, there is a problem with this conclusion. Previous research (Hakes, 1972; Konieczny, 2000; Vasishth, 2003b, 2003a) has shown that increasing argument-head distance can in fact facilitate processing. This
124
Shravan Vasishth
appears to be true for a variety of constructions in English, German, Hindi, and, it appears (Gibson, personal communication), Japanese as well. To illustrate the phenomenon, consider the Hindi sentences in (7). (7)
a.
vo-kaagaz jisko us-larke-ne dekhaa bahut-puraanaa Ç that-paper which that-boy-erg saw very-old thaa was `That paper which that boy saw was very old.'
b.
vo-larkaa jisne us-kaagaz-ko dekhaa bahut-jigyaasu thaa Ç that-boy who that-paper-acc saw very-inquisitive was `That boy who saw that (piece of ) paper was very inquisitive.'
c.
vo-kaagaz jisko us-larke-ne mez ke-piiche gire-hue Ç that-paper which that-boy-erg table behind fallen dekhaa bahut-puraanaa thaa saw very-old was `That paper which that boy saw fallen behind a/the table was very old.'
d.
vo-larkaa jisne us-kaagaz-ko mez ke-piiche gire-hue Ç that-boy who that-paper-acc table behind fallen dekhaa bahut-jigyaasu thaa saw very-inquisitive was `That boy who saw that (piece of ) paper fallen behind a/the table was very inquisitive.'
Sentences (7a) and (7c) are object relative clause constructions with the contrast that the verb dekhaa, `saw', and its arguments larkaa, `boy', kaaÇ gaz, `paper', are adjacent in (7a) but in (7c) are separated by intervening material (mez ke piiche gire hue, `fallen behind the table'). Sentences (7b) and (7d) are subject relative clauses with the same contrast. These pairs of sentence types were investigated in a self-paced reading study; the research question was whether increasing argument-head distance renders processing harder. Distance was de®ned as in Gibson's theory, i.e., as the number of discourse referents intervening; note that the intervening material had a noun phrase and a ®nite verb, which also introduces a discourse referent in Gibson's theory. Thus the intervening material introduced two discourse referents. The reading time at the verb (dekhaa in the examples above) that immediately followed the intervening material was the region of interest: if increasing distance adversely a¨ects processing, reading time at this region would be longer. The results, however, show that increased distance facilitates
Discourse Context and Word Order Preferences in Hindi
125
Mean Reading Time (msec) 600 800 1000 1200
No extra DRs Two extra DRs
NP1
rel
NP2 NP3
P V3 Position
V2
Adj
V1
Figure 9. Object relatives (with 95% CIs).
500
Mean Reading Time (msec) 700 900 1100
No extra DRs Two extra DRs
NP1
rel
NP2 NP3
P V3 Position
V2
Adj
V1
Figure 10. Subject relatives (with 95% CIs).
processing; compare the reading times for the two conditions (zero versus two discourse referents intervening) at verb V2 in Figures 9 and 10. Given the above results, it is di½cult to argue that increased argumenthead distance is the reason for increased di½culty with DO-fronting as opposed to IO-fronting. A plausible interpretation for Experiments 1 and 2 is that the preceding discourse context actually primes the syntactic structure of the target sentence in the IO-fronting case but not in DO-fronting. An interesting di¨erence between the two experiments was that the IO-fronted
126
Shravan Vasishth
context question focused a constituent: to the context question What did A say to B, the answer was A told B to X, where X was a constituent. By contrast, the DO-fronted context question did not focus a constituent: to the context question What did A do with C? the answer formed the pattern A told B to buy C, where the new material told B to buy did not form a constituent (of course, the actual verbs etc. were varied in the stimuli sentences). The canonical order sentences were, of course, primed by the context question. Thus, in IO-fronted sentences the syntactic structure could have been primed by the context in both canonical and non-canonical orders, resulting in no di¨erence in processing di½culty in canonical versus IO-fronted orders. By contrast, in DO-fronted sentences there would have been no structural priming, so that these would be harder to process than canonical order sentences, which were primed by the context. Further experiments are being planned to explore the validity of the above explanation. If correct, this would suggest that in Experiment 2 it is not increased distance per se that is overriding any facilitation due to discourse context, but an asymmetry in syntactic priming. This would in turn imply that the present results are in fact consistent with the existing crosslinguistic results on the e¨ect of discourse context on the processing of non-canonical word order. In sum, an asymmetry in structural priming could in principle be responsible for the results of Experiment 2, and this would be consistent with the existing discourse processing literature and the other evidence against the distance hypothesis. On the other hand, there is extensive cross-linguistic evidence in favor of the distance hypothesis (Hawkins, 1994; Gibson, 2000) that cannot be ignored; therefore, until further experiments clarify the issue, a plausible conclusion from Experiment 2 is that increased distance can dominate over any processing facilitation due to discourse context in non-canonical order sentences. Notes * Earlier versions of this research were presented in the 2003 Architecture and Mechanisms of Language Processing conference (Glasgow), and the European Cognitive Science meeting at OsnabruÈck; I bene®ted greatly from the comments of the reviewers and the audience at these conferences. Professor Ayesha Kidwai and Mr. Samar Sinha provided invaluable assistance and logistical support for running the experiments at Jawaharlal Nehru University, New Delhi. Thanks also to Ivana Kruij¨-Korbayova for help with experiment design, and to PeÂter Dienes for comments and suggestions. This research was funded by SFB 378 (Resource-adaptive cognitive processes; EM 6 NEGRA), at Saarland University, Germany. The usual exculpations apply.
Discourse Context and Word Order Preferences in Hindi
127
References Altmann, Gerry & Mark Steedman (1988). Interaction with context during human sentence processing. Cognition, 30, 191±238. Cohen, Jonathan, Brian MacWhinney, Matthew Flatt & Je¨erson Provost (1993). PsyScope: An interactive graphic system for designing and controlling experiments in the psychology laboratory using Macintosh computers. Behavior Research Methods, Instruments and Computers, 25, 257±271. Gambhir, Vijay (1981). Syntactic restrictions and discourse functions of word order in standard Hindi. Unpublished doctoral dissertation, University of Pennsylvania, Philadelphia. Gibson, Edward (2000). Dependency locality theory: A distance-based theory of linguistic complexity. In A. Marantz, Y. Miyashita, & W. O'Neil (Eds.), Image, language, brain: Papers from the ®rst mind articulation project symposium. Cambridge, MA: MIT Press. Hakes, David T. (1972). On understanding sentences: In search of a theory of sentence comprehension. (Micro®lm, University of Texas, Austin) Hawkins, John A. (1994). A Performance theory of order and constituency. New York: Cambridge University Press. Ð (1998). Some issues in a performance theory of word order. In A. Siewierska (Eds.), Constituent order in the languages of Europe. Berlin: Mouton de Gruyter. HyoÈnaÈ, Jukka & Heli Hujanen (1997). E¨ects of Case Marking and Word Order on Sentence Parsing in Finnish: An Eye-Fixation Study. Quarterly Journal of Experimental Psychology, 50A(4), 841±858. Just, Marcel Adam, Patricia A. Carpenter & Jacqueline D. Woolley (1982). Paradigms and processes in reading comprehension. Journal of Experimental Psychology: General, 111(2), 228±238. Kaiser, Elsi & John Trueswell (2002). Anticipating upcoming discourse referents on the basis of discourse status. In Proceedings of the 26th Penn Linguistics Colloquium. Konieczny, Lars (2000). Locality and parsing complexity. Journal of Psycholinguistic Research, 29(6), 627±645. Steedman, Mark (2000). The Syntactic Process. Cambridge, MA: MIT Press. Vasishth, Shravan (2003). Quantifying processing di½culty in human sentence parsing: The role of decay, activation, and similarity-based interference. In Proceedings of the EuroCogSci conference. Osnabrueck, Germany. Ð (2003). Working memory in sentence comprehension: Processing Hindi center embeddings. New York: Garland Press. (In the series Outstanding Dissertations in Linguistics, edited by Laurence Horn) Weber, Andrea & Julia Neu (2003). Assignment of grammatical functions in discourse context and word-order ambiguity resolution. In Proceedings of the 16th Annual CUNY Sentence Processing Conference. MIT, Cambridge, MA.
Europe* JOHN PETERSON
1. Introduction Over the past few years, a trend which I noted in my report on Europe for the 2002 issue of this Yearbook has continued, or perhaps even gained momentum: Although Europe is undoubtedly best known in South Asian linguistics for its long tradition of historical research, especially with respect to the Indo-Aryan languages, there has been a growing movement ever since at least the late 1980's to document lesser known languages, especially those of non-Indo-European stock. As many of these languages have received little or no documentation since Grierson's massive survey, and especially since many of these languages are now on the verge of extinction, this is a development which can only be welcomed by anyone who is interested in preserving all we can of the rich linguistic heritage of the subcontinent. To date the main thrust of this development, at least in Europe, has been on the Tibeto-Burman languages, which make up one of the largest language families on earth, both in terms of the number of languages and speakers. However it is also pleasing to note that there has been a noticeable increase in work on the Austro-Asiatic languages, although these languages are still receiving relatively little attention. And of course, work on the Indo-Aryan languages is continuing to ¯ourish, as it has since the 19th century. Unfortunately, however, work on the Dravidian languages seems to be receiving very little attention in Europe: Whereas I was able to mention at least one title on a Dravidian language in the last report (Lehmann, 1998), I am not aware of any titles on these languages which have been published since 2001 (or are in press) by scholars residing in Europe. In the following, I will follow the same principle I followed in the last three reports, restricting my comments to a bare minimum and letting the titles speak for themselves as much as possible. Also, as in the last report, I will divide these works not by the section of linguistics which they deal with but rather by the respective language families.
132
John Peterson
2. Works on More Than One Language Family We begin our discussion here with works which deal with more than one language family before we proceed to discuss works on individual languages or language families. The ®rst work which should be mentioned here is a volume edited by Tej Ratna Kansakar (Kathmandu, Nepal) & Mark Turin (Cambridge, UK and Cornell, New York, USA), Kanasakar & Turin (2003). This is a collection of papers from a conference held in Kathmandu in September 1999, jointly organized by the Central Department of Linguistics, Tribhuvan University, the Royal Nepal Academy and the Linguistic Society of Nepal. It contains a preface, introduction, 18 articles on various Himalayan languages (four of which were written by European scholars and will be discussed below: van Driem, 2003; Ebert, 2003c; Huber, 2003b; Volkart, 2003), as well as ®ve ``research notes''. The book is jointly published by the South Asia Institute, Heidelberg (Germany) and the Tribhuvan University, Kathmandu (Nepal). Further works dealing with more than one language family include one by Balthasar Bickel (Leipzig, Germany) which deals with the syntax of experiencers in Himalayan languages, both Indo-Aryan and Tibeto-Burman (Bickel, in press, b). Finally, Annie Montaut (Paris, France) deals with the highly complex issue of hearsay, mirativity and other ``evidential'' categories in the IndoAryan languages Hindi and Nepali and also includes data from the Dravidian languages Telugu and Tamil (Montaut, in press, b). 3. Indo-Aryan As already noted, most work on Indo-Aryan languages in Europe over the past two years deals with the modern languages, although historical works are by no means lacking. I will begin here with those works which deal predominantly with the older languages or with ancient Indian linguistics before proceeding to the modern languages. We begin with two etymological articles. M.S. Andronov (Moscow, Russia) has recently published a work on the etymology of the word for ``rice'' in his Russian-language article (Andronov, 2003). In a similar vein, Enrica Garzilli (Milan, Italy) has just recently published a study of the etymology and semantics of the many words for ``lotus'' in the Rigveda, through which the author, among other things, hopes to clarify the interpretation of the relevant stanzas (Garzilli, in press). Leonid Kulikov (Leiden / Nijmegen, Netherlands) has published three articles dealing with Old Indo-Aryan from a typological perspective in the past two years: Kulikov (2002a) takes a look at reciprocal verbs in Vedic
Europe
133
Sanskrit, while Kulikov (2002b) deals with noun incorporation in Sanskrit (and Frisian!). Kulikov (in press) is a discussion of the Vedic re¯exive from a typological point of view, dealing with both the re¯exive function ``proper'' of Vedic svaÂ-/svayaÂm, tanuÅÂ-, aÅtmaÂn- and tmaÂn- (i.e., where there is coreference with the subject), but also with the so-called ``emphatic'' function of these forms throughout the Vedic period. Thomas Oberlies (GoÈttingen, Germany) has published a number of articles and books primarily on Old and Middle Indo-Aryan languages in recent years. Oberlies (2002) deals with ``short(ened)'' case endings in Indo-Aryan which have arisen through haplology, generally in complex noun phrases in which all components have the same case marking, so that non-®nal case marking within the NP is (or can be) shortened. Oberlies (2003a) is an overview of AsÂokan Prakrit and PaÅli, consisting of an introduction and sections dealing with phonology, and nominal and verbal morphology. Oberlies (2003b) is a detailed grammar of Epic Sanskrit (LVI 632pp.) in the series ``Indian Philology and South Asian Studies''. It deals with virtually all areas of the grammar, including phonology, morphology and syntax and also has an extensive appendix of roots and verb forms speci®c to Epic Sanskrit (168 pp.) and copious indices (48pp.). Finally, Oberlies (in press) is a historical grammar of Hindi. We now turn our attention to a number of recently published works by Boris Zakharyin (Moscow, Russia), which deal primarily with ancient Indian lingusitics. In Zakharyin (2002), the author takes a look at the di¨erentiation between phonetics and morphonology in works written by ancient Indian linguists, while Zakharyin (2003a) is an introduction to ancient Indian linguistics and linguophilosophy, both works written in Russian. Finally, Zakharyin (2003b), written in English, examines among other things the semantic category sambandha in the PaÅninian tradition and looks at Ç Ç and New Indo-Aryan. corresponding syntactic phenomena in both Old We come now to a large number of works dedicated to the modern languages, especially Hindi. Margot Gatzla¨-HaÈlsig (Leipzig, Germany) has published an extensive Hindi-German hand dictionary, in cooperation with a number of other scholars (Gatzla¨-HaÈlsig, ed., 2002, XXVIII 1448 pp.), while Gatzla¨-HaÈlsig (2003) is a revised edition of her Germanlanguage manual of Hindi grammar, both of which have appeared in the Buske Verlag in Hamburg. Similarly, Liperovsky (2003), by V.A. Liperovsky (Moscow, Russia), also presents an overview of Hindi grammar, written in Russian. Liudmilla Khokhlova (Moscow, Russia) has recently published two works dealing with various modern Indo-Aryan languages: Khokhlova (2003) is a discussion of the distribution of synthetic and analytic passives in the Indo-European languages of the western subcontinent while Khokhlova
134
John Peterson
(in press) deals with the syntax and morphology of second causative formations in Hindi-Urdu, Punjabi, Gujurati and Rajasthani. A very di¨erent approach to language research is taken by Stanislav Martynyuk (Kiev, Ukraine): Martynyuk (2002) is a statistical approach to the much discussed issue of the relationship between Hindi and Urdu. In his 13-page article, the author presents a list of the 100 most common Hindi and Urdu words based on text corpora (altogether well over 800,000 words from internet news services). The author notes that in his data Hindi and Urdu di¨er in about one-third of the most common words, although he cautions that much more data of this type is still necessary. Annie Montaut (Paris, France) has also recently published a number of articles on modern Hindi: Montaut (2002, 2003a, b; in press, a). These deal with the particle to (Montaut, 2002), the aorist and counterfactual (Montaut, 2003a) and the category ``subject'' (in press, a) in modern Hindi, while Montaut (2003b) discusses, among other things, the limited relevance of grammatical categories compared with the importance of semantic roles and information ¯ow. Works by Bhuvana Narasimhan (Nijmegen, Netherlands) in the past two years also deal with topics in Hindi: Narasimhan (2003) discusses manner of motion verbs in Hindi and their compatibility with path phrases, such as The bottle ¯oated into the cave, which are typically allowed in English but which are more restricted in languages such as Hindi. Similarly, Bowerman et al. (2002), which is co-authored by Narasimhan, discusses data from eight di¨erent languages (including Hindi) dealing with the acquisition by children of this type of information in di¨erent languages. Finally, Narasimhan et al. (in press) deals with schwa-deletion in Hindi. Neukom & Patnaik (2003), by Lukas Neukom (Zurich, Switzerland) and Manideepa Patnaik (Bhubhaneshwar, Orissa), appeared just as this report was being compiled. It presents a detailed overview of the eastern IndoAryan language Oriya (454 pp.) and is perhaps the ®rst description of this magnitude in a western language on Oriya. One of my own works (Peterson, OsnabruÈck, Germany) also pertains to the Indo-Aryan langauges: Peterson (2002b) is a holistic account of the Nepali converbs, in which I examine the form, meaning and scopal behavior of the various converbs of the language (in -õÅ, -e, -era, -daÅ and -dai) and propose a new classi®cation of the forms on the basis of these factors. Another work on Nepali is that by Mark Turin (Cambridge, UK and Cornell, New York, USA), Turin (2002b), which takes a look at Nepali kinship terms, with special emphasis on loan words such as anÇkal and aÅntÅõ, often used for ÇÇ addressing foreigners. Ruth Laila Schmidt (Oslo, Norway) has recently published a number of works on Shina and one article on Urdu. Schmidt (2003) is an overview of
Europe
135
Urdu in the massive volume edited by Cardona & Jain (2003), covering virtually all aspects of the language, including a general introduction, phonology, morphology, syntax and a section on the writing system. Schmidt (forthcoming, a, b), as well as Schmidt & Koistani (2001), deal with various aspects of the Shina language, ranging from nominal in¯ection to converbs and a comparison of the various Shina dialects. Another contribution in Cardona & Jain's (2003) overview of the IndoAryan languages which should be mentioned here is by Christopher Shackle (London, UK), in which the author presents a detailed overview of the Panjabi language, again covering virtually all major areas of the grammar (Shackle, 2003). Finally, Claus Peter Zoller (Heidelberg, Germany) has recently completed the ®rst volume of his description of the Dardic language Indus Kohistani. The dictionary (Zoller, in press) contains around 8,000 entries, along with etymological information. It also contains two indices, English ÐIndus Kohistani and Old Indo-AryanÐIndus Kohistani. Volume II, which is now in preparation, will be a detailed grammar of the language. 4. Tibeto-Burman We begin our discussion of studies on the Tibeto-Burman languages with a number of works by George van Driem (Leiden, Netherlands). van Driem (2003) is a brief review of the evidence for and against assuming a ``Mahakiranti'' branch of Tibeto-Burman. While the author comes to a rather negative conclusion on this issue, he notes that there is now more evidence for a Newaric or ``MahaÅnevaÅrõÅ '' branch, including Newari, BaraÅm and Thangmi. van Driem (2002a) is the ®rst in a series of survey articles in Lingua under the heading ``The decade in . . .'', whose aim is ``to publish overview articles that will give nonspecialist linguists an insigt into the way in which speci®c areas of expertise have developed in the last 10 years'' (editors' note, van Driem 2002a:79). In this work, van Driem gives a brief overview of the revival in the last decade of the past century of the old Tibeto-Burman theory. The article is also a valuable source of bibliographic information on many of the Tibeto-Burman languages. Finally, van Driem (2002b) deals with the prehistory and phylogeny of Tibeto-Burman. Other recent works on Tibeto-Burman languages include two by Balthasar Bickel (Leipzig, Germany). Bickel (2003) is a general overview of the Kiranti language Belhare, while in Bickel (in press, a) the author takes a look at the notion of ``subject'' in this language, paying special attention to pivots. DoÈrte Borchers (GoÈttingen / Berlin, Germany) has recently published two works dealing largely with the Kiranti language Sunwar (Koints): Borchers
136
John Peterson
(2003) deals with the topic of language variety and language policies in Nepal, while Borchers (in press) discusses the reasons for language death. Karen Ebert (Zurich, Switzerland), has also recently published a number of works on the Kiranti languages of eastern Nepal: Ebert (2003a) is an overview of the Kiranti languages in general, dealing with phonology, nominals, the verbal systems, syntax and genetic and areal groupings. Ebert (2003b), in the same volume, is a general overview of the Kiranti language Camling. Finally, Ebert (2003c), in Kansakar & Turin (2003), deals with equivalents of ``conjunctive participles'' or converbs in Kiranti languages. There is a group of scholars working in Berne, Switzerland whose works since 2002 deal predominantly with various aspects and dialects of Tibetan: Roland Bielmeier, Brigitte Huber and Marianne Volkart. Bielmeier (2003) is a research report on the Comparative Dictionary of Tibetan Dialects (CDTD) and Bielmeier (in press, a) is a review of van Driem (2001). Bielmeier's remaining two works deal with particular topics in Tibetan: Bielmeier (in press, b) takes a look at the formation of Tibetan verb paradigms in light of Shafer's proto-West Bodish hypothesis while Bielmeier (in press, c) deals with lexical variation and change in Tibetan as illustrated by designations for various body parts. Huber (2003a) is a study of relative clauses in Kyirong Tibetan, Huber (2003b) takes a closer look at tone in the same dialect, both from a synchronic and diachronic perspective, and Huber (in press) is a study of the Lende subdialect of Kyirong Tibetan, inlcuding historical notes. Finally, Volkart (2003) deals with types of nominal compounds in Written Tibetan. The author compares these with the types found in Sanskrit and discusses the extent to which the compound types described by indigenous Sanskrit grammarians can be carried over to Tibetan. Two authors from the CNRS in Paris, France, Martine Mazaudon and Boyd Michailovsky, have published a number of works in the last two years on Tibeto-Burman languages. Mazaudon (2003a, to appear) are overviews of the Tamang language of Nepal, written in English and French, respectively. Mazaudon (2003b) deals with the logic behind the counting systems in a number of Tibeto-Burman languages. Finally, Mazaudon (in press) deals with the gradual grammaticalization of various discourse particles into markers of subordinate clauses in Tamang. Michailovsky (2002) is a Limbu-English dictionary of the Mewa Khola dialect (with an EnglishLimbu index) while Michailovsky (2003, to appear, b) are overviews of Hayu and Limbu, respectively. Michailovsky (to appear, a) deals with time-ordinals in Kiranti languages. Following Matiso¨, the author divides these ordinals into ``runs'', i.e., that part which denotes whether the timeordinal is a unit of days or years (and whether in the past or future), and ``counters'', roughly the ``ordinal'' part of the time-ordinal (i.e., ®rst,
Europe
137
second, third, etc.). Finally, Michailovsky (to appear, c) is a brief overview of the Tibeto-Burman language family, written in French. Heleen Plaisier (Leiden, Netherlands) has recently published an overview in the extensive volume on Sino-Tibetan languages (Thurgood & LaPolla, 2003) of the Lepcha or RoÂng-rõÂng language, spoken in the Sikkim and Darjeeling district of West Bengal, the Ilam District of Nepal and in parts of Bhutan. The overview includes a discussion of phonology, nominal and verbal morphology as well as a brief annotated Lepcha text (Plaisier, 2003). Anju Saxena (Uppsala, Sweden) has published a considerable number of works in the past two years which deal primarily with the Kinnauri language. Two of her works (Saxena, to appear, b, c) will appear in a volume she herself has edited (Saxena, to appear, d), about which we will undoubtedly have more to say in the next report, after her book has appeared. Saxena (to appear, b) deals with the discourse functions of the ®nite verb in Kinnauri narratives while Saxena (to appear, c) deals with the study of Himalayan languages in general. The other works by Saxena which have appeared since 2002 (2002a,b; to appear a, e), deal with various aspects of Kinnauri, ranging from strategies for expressing requests and commands (2002a) and corpus linguistics (2002b) to strategies for reporting speech (to appear, a) and other aspects of Kinnauri narratives (to appear, e). Finally, Saxena (2002±2003) is a Swedish-language article dealing with multilingualism and language contact in South Asia. Turin (2002a) (Cambridge, UK and Cornell, New York, USA) deals with Thangmi (Nepal) ethnicity from a linguistic-anthropological perspective, providing information on the genetic position of the Thangmi language, the designation for Thangmi by its own speakers and also Thangmi designations for other ethnic groups such as the Tamang and Newars, among other topics. Finally, Turin (2003) is a geolinguistic analysis of historical writings on the Thangmi. 5. Austro-Asiatic As a researcher on Munda languages, I ®nd it especially pleasing to see a renewed interest in the half-forgotten Austro-Asiatic languages, both Munda and Mon-Khmer. Daladier (2002) (Paris, France) deals with the expression of de®niteness in the War-language Amwi (Mon-Khmer), closely related to the betterknown Khasi, primarily by means of what the author refers to as ``syntactic stress'', a phenomenon which closely resembles person-marking. Daladier (in press, a) is a brief, 13-page overview of the Mon-Khmer language Khasi of northeast India, in French, which also contains some comparative data on the relationship between Khasi and the War languages. Finally, Daladier
138
John Peterson
(in press, b) deals with kinship and spirit terms as classi®ers of animate nouns in Austro-Asiatic. Neukom (2001) (ZuÈrich, Switzerland) deals with the well-known but highly complex issue of parts-of-speech in the Munda languages, more precisely with the question of nouns and verbs in the north Munda language Santali. The author comes to the conclusion that there are nouns and verbs in Santali, adding that the distinctions between these classes are not based on formal criteria but rather on syntactic criteria which also take the semantics of the lexemes into account. In one of my own works (Peterson, 2002a; OsnabruÈck, Germany) I take a look at the question of ®niteness in the south Munda language Kharia, in which, in addition to fully ®nite and non-®nite predicates, there are also predicates which contain a large amount of morphological marking but lack certain markers, generally markers for person/number/honori®c status, and are hence ``semi-®nite'', a phenomenon which is typical of a large number of other Munda languages but also many Kiranti (Tibeto-Burman) languages (cf. the data in Ebert (1993; 1999) for an overview). Peterson (in press) is a 17-page article dealing with a rather uncommon morpheme in Kharia, -may, the ``totality'' marker, which expresses that an event is telic but also that one of the arguments or adjuncts involved is totally a¨ected by the event. I also provide a possible etymology of this marker from the homophonous marker of the 3rd person, plural, probably originally functioning as a marker of object agreement on the predicate. As a ®nal note, I would like to call the reader's attention to my own online bibliography of endangered and seldom studied South Asian languages (Peterson, no date), which has been online since January, 2001 and which is updated at regular intervals. Contributions are always welcome: http://www. SouthAsiaBibliography.de/. 6. Summary The long list of titles presented here, although undoubtedly not exhaustive, is certainly impressive for a two-year period and shows that the study of South Asian languages in Europe is currently ¯ourishing. Although I found no studies from Europe dealing speci®cally with Dravidian linguistics, the paucity of studies in this area is compensated for by an increasingly large number of predominantly descriptive studies on the modern Indo-Aryan, Tibeto-Burman and Austro-Asiatic languages, as well as works which have a largely diachronic perspective. Unfortunately, as encouraging as the present situation may seem, there is also some cause for concern with respect to the future. In recent years, pressure has been steadily increasing to cut costs at state-run institutions, which
Europe
139
in Europe generally includes the university system. In such a competitive situation, smaller departments, such as Indology and often linguistics, are easy targets for closure, based less on arguments of quality than on the number of students studying there. It is clear in such a situation that these departments will in the future ®nd it increasingly di½cult to compete for funding with the larger departments. Although perhaps not yet acute, a number of departments have already been singled out for closure, and these cost-cutting actions are only just beginning to take e¨ect. Nonetheless, whatever the future may bring, research on South Asian languages is ¯ourishing in Europe at the moment, albeit with a di¨erent emphasis than 100, 50 or even 25 years ago. If there is a decline in these studies in the coming years, it will certainly not be due to a lack of interest on the part of researchers. Rather, it will be the unfortunate result of shortsighted cost-cutting measures coupled with the present global economic slump. One can only hope that future reports on South Asian studies in Europe can continue to be at least as extensive as the present one. Notes * I would like to take this opportunity to thank the many people who provided me with information both on their own work as well as that of their colleagues: Balthasar Bickel, Roland Bielmeier, DoÈrte Borchers, Anne Daladier, George van Driem, Enrica Garzilli, Arlo Gri½ths, Brigitte Huber, Liudmila Khokhlova, Leonid Kulikov, Stanislav Martynyuk, Annie Montaut, Bhuvana Narasimhan, Lukas Neukom, Thomas Oberlies, Christina Oesterheld, Tatiana Oranskaia, Anju Saxena, Ruth Laila Schmidt, Mark Turin, Boris Zakharyin and Claus Peter Zoller. Thanks to all of you! Of course, I alone am responsible should any errors have found their way into this report, as well as for any accidental omissions.
References Andronov, M.S. (2003). ``K etimologii slova ris.'' V.M. Alpatov (ed.), VostocÏnoye Yazykoznaniye. Moskva: Izdatel'stvo VostocÏnaya Literatura. 49±53. [In Russian: ``On the etymology of the word rice.''] Bickel, Balthasar (2003). ``Belhare.'' Graham Thurgood and Randy LaPolla (eds.), 546±70. Ð (in press, a). ``Hidden syntax in Belhare.'' Anju Saxena (ed.), Himalayan Linguistics, Past and Present. Berlin: Mouton de Gruyter. (Trends in Linguistics, Studies and Monographs, 149) Ð (in press, b). ``The syntax of experiencers in the Himalayas.'' Peri Bhaskararao & Karumuri V. Subbarao (eds.), Non-nominative Subjects in South Asian Languages. Amsterdam: Benjamins. Bielmeier, Roland (2003). ``Comparative Dictionary of Tibetan Dialects (CDTD). A research report.'' European Bulletin of Himalayan Research 23:97±101. Ð (in press, a). ``On the Languages of the Himalayas and their Links (nearly) around the World. Review article of George van Driem, Languages of the Himalayas. An Ethnoloinguistic Handbook of the Greater Himalayas, containing an Introduction to the Symbiotic
140
John Peterson
Theory of Language. (Handbook of Oriental Studies, Section 2, India). European Bulletin of Himalayan Research 24. 27 pp. Ð (in press, b). ``Shafer's proto-West Bodish hypothesis and the formation of the Tibetan verb paradigms.'' Anju Saxena (ed.), Himalayan Languages, Past and Present. (Trends in Linguistics, Studies and Monographs, 149). Berlin / New York: Mouton de Gruyter. 395±412. Ð (in press, c). ``Lexikalische Variation und lexikalischer Wandel im Tibetischen am Beispiel einiger KoÈrperteilbezeichnungen.'' Wiltrud Mihatsch & Reinhild Steinberg (eds.), Lexical Data and Universals of Semantic Change. TuÈbingen: Stau¨enburg (Stau¨enburg Linguistik). [In German: ``Lexical variation and lexical change in Tibetan illustrated by designations for certain body parts.''] Borchers, DoÈrte (2003). ``Sprachenvielfalt und Sprachpolitik in Nepal.'' GoÈttinger BeitraÈge zur Sprachwissenschaft 8:25±37. [in German: ``Language variety and language politics in Nepal.''] Ð (in press). ``Reasons for language death: myths and counterevidence.'' Roland Bielmeier & Felix Haller (eds.), (title to be announced). Berlin / New York: Mouton de Gruyter (Trends in Linguistics). Bowerman, Melissa, Penelope Brown, Sonja Eisenbeiss, Bhuvana Narasimhan & Dan Slobin (2002). ``Putting Things in Places: Developmental Consequences of Linguistic Typology.'' Eve V. Clark (ed.), The Crosslinguistic Encoding of Goal-Directed Motion in ChildCaregiver Discourse. Stanford: Center for Studies in Language and Information. 1±122. Cardona, George and Dhanesh Jain (eds.) (2003). The Indo-Aryan Languages. London / New York: Routledge. (Routledge Language Family Series). Daladier, Anne (2002). ``De®niteness in Amwi: Grammaticalization and Syntax.'' A. Zribi Hertz & A. Daladier (eds.), Syntaxe de la de®nitude. Paris: Presses Universitaires de Vincennes. (Recherches linguistiques de Vincennes, 31). 61±78. Ð (in press, a). ``Khasi.'' EncyclopeÂdie des Sciences du Langage, Dictionnaire des Langues. Paris: Presses Universitaires de France. [in French]. Ð (in press, b). ``Kinship and Spirit Terms Renewed as Classi®ers of ``Animate'' Nouns and their reduced Combining Forms in Austroasiatic.'' Proceedings of the 28th BLS Annual Conference, Volume dedicated to James Matiso¨. Berkeley: Berkeley University Press. Driem, George van (2001). Languages of the Himalayas. An Ethnolinguistic Handbook of the Greater Himalayan Region, containing an Introduction to the Symbiotic Theory of Language. (two volumes). Leiden / Boston / KoÈln: Brill (Handbook of Oriental Studies, Section Two: India, Volume 10). Ð (2002a). ``Tibeto-Burman replaces Indo-Chinese in the 1990s: Review of a decade of scholarship.'' Lingua 111:79±102. Ð (2002b). ``Tibeto-Burman phylogeny and prehistory: Languages, material culture and genes.'' Peter Bellwood & Colin Renfrew (eds.), Examining the Farming / Language Dispersal Hypothesis. Cambridge: McDonald Institute for Archaeological Research. 233±49. Ð (2003). ``Mahakiranti revisited: Mahakiranti or Newaric?'' Tej Ratna Kansakar and Mark Turin (eds.), 21±6. Ebert, Karen H. (1993). ``Kiranti subordination in the South Asian areal context.'' Karen H. Ebert (ed.), Studies in Clause Linkage. Papers from the First KoÈln-ZuÈrich Workshop. ZuÈrich: Allgemeine Sprachwissenschaft, UniversitaÈt ZuÈrich. (ASAS, 12). 83±110. Ð (1999). ``Non®nite verbs in Kiranti languagesÐan areal perspective.'' Yogendra P. Yadava & Warren W. Glover (eds.), Topics in Nepalese Lingustics. Kathmandu: Royal Nepal Academy. 371±400. Ð (2003a). ``Kiranti Languages: An Overview.'' Graham Thurgood and Randy LaPolla (eds.), 505±17. Ð (2003b). ``Camling.'' Graham Thurgood and Randy LaPolla (eds.), 533±45.
Europe
141
Ð (2003c). ``Equivalents of `conjunctive participles' in Kiranti languages.'' Tej Ratna Kansakar and Mark Turin (eds.), 27±47. Garzilli, Enrica (in press). ``The Flowers of Rgveda Hymns: Lotus in V.78.7, X.184.2, X.107.10, VI.16.13, and VII.33.11, VI.61.2, VIII.1.33, X.142.8''. Indo-Iranian Journal 46/ 4:293±314. Gatzla¨-HaÈlsig, Margot (ed.) (2002). HandwoÈrterbuch Hindi-Deutsch, unter Mitarbeit von Lutz Baganz, Barbara BoÈrner-Westphal, Hannelore Bauhaus-LoÈtzke und Christina Oesterheld. Hamburg: Buske. [in German: ``Hindi-German Hand Dictionary.''] Ð (2003). Grammatischer Leitfaden des Hindi. 5, durchgesehene und mit einer aktualiserten Einleitung versehene Au¯age. Hamburg: Buske. [in German: ``Grammatical Manual of Hindi, 5th revised edition, with an updated introduction'']. Huber, Brigitte (2003a). ``Relative clauses in Kyirong Tibetan.'' Linguistics of the TibetoBurman Area 26/1:1±14. Ð (2003b). ``Tone in Kyirong Tibetan: A conservative phonological development.'' Tej Ratna Kansakar and Mark Turin (eds.), 65±80. Ð (in press). The Lende Subdialect of Kyirong Tibetan: A Grammatical Description with Historical Annotations. Bonn: VGH Wissenschaftsverlag. Kansakar, Tej Ratna & Mark Turin (eds.) (2003). Themes in Himalayan Languages and Linguistics. Heidelberg: South Asia Institute and Kathmandu: Tribhuvan University. Khokhlova, Liudmila Viktorovna (2003). ``The Distribution of Analytic and Synthetic Passives in Indo-European Languages of Western India.'' Ritva Laury, Gerald McMenamin, Shigeko Okamoto, Vida Samiian & Karumuri V. Subbarao (eds.), Papers in Honor of P.J. Mistry. New Delhi: Indian Institute of Language Studies. 139±57. Ð (in press). ``Infringement of Morphological and Syntactic Operations' Pairing in ``Second Causative'' Formation (Hindi-Urdu, Punjabi, Gujarati, Rajasthani).'' Indian Linguistics 64. Kulikov, Leonid (2002a). ``TipologicÏeskie zametki o morfologicÏeskom reciproke v vediyskom sanskrite.'' V.A. Plungian & A.J. UrmancÏieva (eds.). Yazyki mira. Tipologiya. Uralistika. Ï danovoy. Stat'i i vospominaniya. Moskva: Indrik. 276±99 [in Russian: ``TypoPamjati T. Z logical remarks on reciprocal verbs in Vedic Sanskrit.''] Ð (2002b). ``TipologicÏeskie zametki ob imennoy inkorporacii v indoevropeyskom: frizskiy i sanskrit.'' Tatyana Agranat & Olga KazakevicÏ (eds.), LinvisticÏeskiy bespredel: Sb. statey k 70-letiju A.I. Kuznecovoy. Moskva: Izd-vo MGU. 262±73 [in Russian: ``Typological remarks on noun incorporation in Indo-European: Frisian and Sanskrit.''] Ð (in press). ``Typological notes on the Vedic re¯exive.'' Proceedings of the International Conference ``Re¯exive and Middle'', Tunis, March 2001. Leuven: Peeters. Lehmann, Thomas (1998). Old Tamil. Sanford S. Steever (ed.), The Dravidian Languages. London: Routledge. 75±99. Liperovsky, V.A. (2003). ``Kontur grammaticÏeskogo stroya yazyka hindi.'' V.M. Alpatov (ed.), VostocÏnoye Yazykoznanoye. Moskva: Izdatel'stvo VostocÏnaya Literatura. 167±77. [In Russian: ``A Grammatical Sketch of Hindi.''] Martynyuk, Stanislav (2002). ``Statistical Approach to the Debate on Urdu and Hindi.'' The Annual of Urdu Studies. http://www.urdustudies.com/Issue18/i18home.html Mazaudon, Martine (2003a). ``Tamang.'' Graham Thurgood and Randy LaPolla (eds.), 291±314. Ð (2003b). ``Les principes de construction du nombre dans les langues tibeÂto-birmanes.'' J. FrancËois (eÂd.), La PluraliteÂ. MeÂmoires de la SocieÂte de Linguistique de Paris. Paris: Peeters. 91±119. [In French: ``The principles of construction of number in Tibeto-Burman languages.''] Ð (in press). ``From discourse to grammar in Tamang: topic, focus, intensi®ers, and subordination.'' David Bradley, Randy LaPolla, Boyd Michailovsky & Graham Thurgood (eds.), Language Variation: Papers on variation and change in the Sinosphere and in the
142
John Peterson
Indosphere in honour of James A. Matiso¨. Canberra: Australian National University. (Paci®c Linguistics). Ð (to appear). ``Tamang.'' S. Auroux (eÂd.), EncyclopeÂdie des Sciences du Langage. Dictionnaire des Langues. Paris: Presses Universitaires de France. [in French]. Michailovsky, Boyd (2002). Limbu-English Dictionary of the Mewa Khola Dialect, with EnglishLimbu Index. Kathmandu: Mandala Book Point. Ð (2003). ``Hayu.'' Graham Thurgood and Randy LaPolla (eds.), 518±32. Ð (to appear, a). ``Su½x-runs and counters in Kiranti time-ordinals.'' David Bradley, Randy LaPolla, Boyd Michailovsky & Graham Thurgood (eds.), Language Variation: Papers on variation and change in the Sinosphere and in the Indosphere in honour of James A. Matiso¨. Canberra: Australian National University (Paci®c Linguistics). Ð (to appear, b). ``Limbu.'' Djamel Kouloughli & Alain Peyraube (eÂds.), EncyclopeÂdie des Sciences du Language. Dictionnaire des Langues. [in French]. Ð (to appear, c). ``TibeÂto-birman.'' Djamel Kouloughli & Alain Peyraube (eÂds.), EncyclopeÂdie des Sciences du Language. Dictionnaire des Langues. [in French]. Montaut, Annie (2002). ``La Particule to en hindi-ourdou.'' Ch. Bonnot, A. Montaut & S. Vassilaki (eÂds.). Cahier de linguistique de l'Inalco. 111±134. [in French: ``The Particle to in Hindi-Urdu.''] Ð (2003a). ``Aoriste et contrefactuel en hindi moderne.'' Cahiers Chronos 11 (Modes de repeÂrage temporel ). [in French: ``Aorist and counterfactual in modern Hindi.]. 55±70. Ð (2003b). ``Comment une langue `libeÂreÂe' et libeÂratrice en vient aÁ agir comme langue d'oppression: le cas du hindi.'' A. M. Laurian (eÂd.), La Langue libeÂreÂe, EÂtudes de sociolexicologie. Bern / Berlin / Paris: Peter Lang. [``How a `liberated' and liberating language comes to act as a language of oppression: The case of Hindi.'']. Ð (in press, a). ``Oblique main arguments in Hindi as localizing predications: questioning the category of subject.'' Peri Bhaskararao (ed.), Non-Nominative Subjects in South Asian Languages. Amsterdam: Benjamins. Ð (in press, b). ``Quotatif, meÂdiatif et miratif dans les langues d'Asie du sud: une ou plusieurs cateÂgories?'' Ph. Jeanne (ed.), MeÂdiatiques, La meÂdiation en langue et en discours. Paris: L'Universite de Laval. [in French: ``Hearsay, mediative and mirative in the languages of South Asia: One or many categories?'']. Narasimhan, Bhuvana (2003). ``Motion events and the lexicon: a case study of Hindi.'' Lingua 113/2:123±60. Narasimhan, Bhuvana, Richard Sproat & George Kiraz (in press). ``Schwa-deletion in Hindi Text-to-Speech Synthesis.'' International Journal of Speech Technology. Neukom, Lukas (2001). ``Nomen/Verb-Distinktion im Santali.'' Bernhard WaÈlchli & Fernando ZuÂnÄiga (eds.), Sprachbeschreibung und Typologie. Publikation zum Workshop vom 16. Dezember 2000 in Bern. Bern: Institut fuÈr Sprachwissenschaft. [In German: ``Noun/VerbDistinction in Santali.''] Neukom, Lukas & Manideepa Patnaik (2003). A Grammar of Oriya. ZuÈrich: UniversitaÈt ZuÈrich (ASAS, 17). Oberlies, Thomas (2002). ``Language economy: `Short(ened)' case-endings in Indo-Aryan.'' Bulletin d'EÂtudes Indiennes 20:193±7. Ð (2003a). ``AsÂokan Prakrit and PaÅli.'' George Cardona & Dhanesh Jain (eds.), 161±203. Ð (2003b). A Grammar of Epic Sanskrit. (Indian Philology and South Asian Studies, 5). Berlin / New York: Walter de Gruyter. Ð (in press). Historical Grammar of Hindi. The genesis of its morphological system. Reinbek: Dr. Inge Wezler. Peterson, John (2002a). ``Morphological and Semantic Finiteness in Kharia: A ®rst look''. Michael Bommes, Doris Tophinke & Christina Noack (eds.), Sprache und Form. Festschrift fuÈr Utz Maas. Wiesbaden: Westdeutscher Verlag. 63±73.
Europe
143
Ð (2002b). ``The Nepali Converbs: A holistic approach''. Rajendra Singh (chief editor). Yearbook of South Asian Languages and Linguistics, 2002. Dehli: Sage Publications. 93± 133. Ð (in press). ``The ``totality'' morpheme -may in Kharia.'' To appear in Ram Dayal Munda (ed.), Felicitation Volume for Norman Zide. Ð (no date). Bibliography of Endangered and Seldom Studied South Asian Languages. http:// www.SouthAsiaBibliography.de/ Plaisier, Heleen (2003). ``Lepcha.'' Graham Thurgood and Randy LaPolla (eds.), 705±16. Saxena, Anju. (2002a). ``Request and command in Kinnauri: The pragmatics of translating politeness.'' Linguistics of the Tibeto-Burman Area 25/1:165±90. Ð (2002b). ``Using the Kinnauri corpus in research and teaching.'' Nordisk Sprogteknologi. AÊrbog for Nordisk SpraÊkteknologisk Forskningsprogram 2000±2004. 75±8. Ð (2002±2003). ``FlerspraÊkighet och spraÊkkontakt i Sydasien.'' Orientalia Suecana LI±LII. International Journal of Indological, Iranian, Semitic and Turkic Studies. Special Volume in Honor of Bo Utas and Gunilla-Gren Eklund. [In Swedish: ``Multilingualism and Language Contact in South Asia''.] Ð (to appear, a). ``Speech reporting strategies in Kinnauri narratives.'' Linguistics of the Tibeto-Burman Area. Ð (to appear, b). ``On discourse functions of the ®nite verb in Kinnauri narratives.'' To appear in Anju Saxena (ed.). Ð (to appear, c). ``Linguistic synchrony and diachrony on the roof of the worldÐthe study of Himalayan languages.'' To appear in Anju Saxena (ed.). Ð (ed.) (to appear, d). Himalayan Languages. Past and Present. Berlin et al.: Mouton de Gruyter. (Trends in Linguistics). Ð (to appear, e). ``Context shift and linguistic coding in Kinnauri narratives.'' Roland Bielmeier & Felix Haller (eds.). (title to be annouced). Berlin et al.: Mouton de Gruyter. (Trends in Linguistics). Schmidt, Ruth Laila (2003). ``Urdu.'' George Cardona & Dhanesh Jain (eds.), 286±350. Ð (forthcoming, a). ``Converbs in a Kohistani Shina narrative.'' Acta Orientalia 64. Ð (forthcoming, b). ``A grammatical comparison of Shina dialects''. Werner Winter, ed. (title to be announced). Berlin: Mouton de Gruyter. (Trends in Linguistics). Schmidt, Ruth Laila & Razwal Koistani (2001). ``Nominal in¯ections in the Shina of Indus Kohistan.'' Acta Orientalia 62:107±43. Shackle, Christopher (2003). ``Panjabi.'' George Cardona & Dhanesh Jain (eds.), 581±621. Thurgood, Graham & Randy J. LaPolla, eds. (2003). The Sino-Tibetan Languages. London / New York: Routledge. (Routledge Language Family Series). Turin, Mark (2002a). ``Ethnonyms and Other-nyms: Linguistic Anthropology among the Thangmi of Nepal.'' Katia Bu¨etrille & Hildegard Diemberger (eds.), Territory and Identity in Tibet and the Himalayas. Brill: Leiden (Brill's Tibetan Studies Library). 253± 70. Ð (2002b). ``Call Me Uncle: An Outsider's Experience of Nepali Kinship.'' Contributions to Nepalese Studies, XXVIII, 2:277±83. Ð (2003). ``A Geolinguistic Analysis of Historical Writings on the Thangmi People and Language of Nepal.'' Wayne H. Finke & Leonard R.N. Ashley (eds.), Geolinguistics 29. New York: Cummings & Hathway. 71±92. Volkart, Marianne (2003). ``Types of compounds in Written Tibetan.'' Tej Ratna Kansakar and Mark Turin (eds.), 233±248. Zakharyin, Boris (2002). ``RazmezÏevanie fonetiki i morphonologii v rabotax drevneindiyskikh lingvistov.'' Yazykoznanie v teorii i eksperimente. Moskva, ISAA pri MGU. 216±38. [In Russian: ``The Di¨erentiation between phonetics and morphonology in the works of ancient Indian linguists''].
144
John Peterson
Ð (2003a). Patanjali `Paspasha', ili vedenie v nauku o yazyke i lingvo®losophiyu Drevney Indii. Moskva, ISAA pri MGU. [In Russian: Patanjali `Paspasha', Introduction to Ancient Ç Indian Linguistics and Linguophilosophy.] Ð (2003b). ``The semantic category `relationship' (sambandha) of PaÅninõÅyas and some of the Ç Ç Indologica Taurinensia corresponding syntactic phenomena in Old and New Indo-Aryan.'' 26:193±206. Zoller, Claus Peter (in press). A Grammar and Dictionary of Indus Kohistani. Part I: Dictionary. Berlin: Mouton de Gruyter.
An Interpretive Survey of Tamil Studies in Tamil E. ANNAMALAI
1. Sources for Publications `Studies on language' gives more ¯exibility in coverage than the designation `research' and avoids the question of methodological rigour and cultural neutrality. This survey by necessity is not a bibliographical one, not even of works that are landmarks. There are bibliographies speci®c to Tamil language such as An International Bibliography of Dravidian Languages and Linguistics, volume 2ÐTamil Language and Linguistics (Ramaiah 1995), Tamil Dictionaries: A Bibliography (Dhamotharan 1978), bibliographies on Tamil of which the language is a part such as A Reference Guide to Tamil Studies (Thaninayagam 1966), A Bibliography of Tamil Reference Works (Mathaiyan 1978, in Tamil) and bibliographies on Dravidian of which Tamil is a part such as Materials for a Bibliography of Dravidian Linguistics (Andronov 1964), A Bibliography of Dravidian Linguistics (Agesthialingom and Saktivel 1973). Though these bibliographies focus on publications in English, they also randomly include publications in Tamil. There are catalogues of specialized libraries such as the Tamil University Library (Thanjavur), Saraswathi Maha Library (Thanjavur), Tamil Sangam Library (Madurai), Tamil Sangam Library (Karanthai), Maraimalai Adigal Library (Chennai), and Rojah Muthiah Research Library (Chennai), which together contain the majority of publications in Tamil on Tamil language. In addition to contemporary publications, these libraries have archival materials for the study of the Tamil language. There are also the catalogues, though disorganized and incomplete, of depository libraries, including Connemara library in Chennai and National library in Kolkatta, which list publications in Tamil on Tamil language, among others. There are indices of serial publications such as the Index of aayvukkoovai (`Research Collection', which is the title of the Annual Proceedings of the All India Conference of University Tamil Teachers) from 1969 to 1978 (Saktivel n.d.), in which there are sections (10 and 11) on grammar and linguistics.
146
E. Annamalai
Since this paper focuses on studies on Tamil language published in Tamil, it leaves out studies which are in English and other languages, databases of Tamil texts developed for research as well as language products like dictionaries, pedagogical materials and the like using the results of these studies. Not being a bibliographical survey, the focus will be not on listing publications but on the trends and the nature of Tamil language studies. The paper attempts to interpret these studies as to their impulses and impacts. Sources for publishing in Tamil studies done on Tamil are not wanting. There is one journal, moZiyiyal (Linguistics), published by the All India Tamil Linguistics Association, that is devoted to linguistics and there is another, pulamai (Scholarship), heavily oriented towards papers on language. The old journals centamiZ-c-celvi (Damsel of High Tamil) and tamiZ-ppozil (Garden of Tamil), edited and patronized by traditional Tamil scholars, publish grammatical research by them. The departments of Tamil in the universities in Tamilnadu permit M.Phil and Ph.D. dissertations on Tamil language to be written in Tamil. Indian University Tamil Teachers Association and the All India Tamil Linguistics Association were active for some years in organizing annual conferences, where papers are presented. Author-publications by scholars are common in Tamilnadu. All these sources ensure a large quantity of publications in Tamil on Tamil language. These publications, however, vary in their standard of research, as there is no quality control mechanism (like peer review and editorial scrutiny). 2. Interface with Traditional Grammar Tamil language studies inspired by modern linguistics fall into the following four categories: (1) interface with traditional grammatical research, (2) linguistic analysis of grammar, (3) hyphenated and applied linguistic work, (4) study of Tamil as a cultural icon. The characteristics of the studies in each of these categories will be exempli®ed from the studies from the nineteen ®fties, when modern ( American) linguistic training was provided in Pune with funds from the Rockefeller Foundation and when Tamil became the o½cial language of the state bringing in the state as a player in setting the tone of Tamil studies. The same characterization of studies on Tamil language applies to the studies done in Sri Lanka during this period. Tamil has a long written grammatical tradition from the time of Tolkappiyam, just after the Common Era. The tradition is basically hermeneutical, though it admits changes in the forms of language and in the mode of analysis induced by the changes in the language as well as by the changes in the nature of relationship with the Sanskrit language and its grammatical
An Interpretive Survey of Tamil Studies in Tamil
147
tradition. Modern linguistics initiated reinterpretation of traditional grammatical analyses in the light of its methods and concepts. It is, however, rare to ®nd serious linguistic studies on the basic concepts of traditional grammars that are at variance with modern linguistics (such as Annamalai 1989) in order to understand their conceptual base. Much of the reinterpretation is to see in the traditional grammar foreshadowing Bloom®eldian and later Chomskyan linguistics selectively and in fragments (Agesthialingom and Murugaiyan 1972, Agesthialingom and Balasubramanian 1974), and none of it is to inform modern linguistics by any insights of the traditional grammar. Nevertheless, reinterpretation became a legitimate linguistic enterprise resulting in many doctoral dissertations in several linguistics departments. There are also conferences and journal publications on the subject. A bibliography (Ramaiah 1995) lists 295 tiles of linguistic studies on the two important traditional grammars, Tolkappiyam and Nannul, about a quarter of which are in Tamil (this is an underestimate because this bibliography is less comprehensive in its coverage of materials in Tamil). There is a book-length study of all studies on Tolkappiyam (Krishnamoorthy 1990). The linguistic interpretation of traditional grammars by linguists incidentally has expectations of a local political gain of winning over the traditional grammatical scholars by making them tolerant of modern linguistics, who are skeptical that it is doing the same work as theirs but in English and that it is subversive in elevating the colloquial language to be the subject of grammatical analysis. The real gain of linguistic reinterpretation, however, is that it makes the traditional grammars and their commentaries comprehensible to modern linguists in their own jargon and in their way of conceptualizing language. It also introduces a critical approach to the tradition in spite of a strong current of laudatory approach by linguists co-existing with it. The series of books by S.V. Shanmugam (1980, 1986) and Rajaram (1992) on traditional grammatical concepts and analyses are an example of this critical, but accommodating, approach. Such studies by linguists parallel contemporary studies of traditional grammars by language scholars in their framework embedded in the grammatical tradition (Ilakkuvanar 1971). There is, however, no sustained professional dialogue between these two groups of scholars studying language other than one seminar in which both groups participated and the language scholars presented papers in Tamil and linguists in English (Agesthialingom and Murugaiyan 1972). Modern linguistics brought two new ways of approaching the traditional grammar. One is to compare the language described in grammatical works with the language of their contemporary literary works and point out divergences. Though such studies would have had some impact on language scholars, there are very few studies in Tamil (such as Natarajan 1977a, which is based on his published doctoral dissertation in English (Natarajan
148
E. Annamalai
1977b) comparing the language of Sangam literature and of Tolkappiyam and its grammatical rules). The impact of the publication of such divergence on language scholars might have been on their view that the great among the old grammarians are infallible. Another new way introduced by modern linguistics is to look at the old grammatical works historically and thus introduce a ®eld of study of grammatical thought, which includes the in¯uence of the descriptive models of other languages, particularly Sanskrit. Rajalakshmi's (1980) unpublished thesis, History of Tamil Grammar and T.P. Meenakshisundaran's (1974) Foreign Models of Tamil Grammar are two examples of this type of study. Both are in English, but there are some studies, albeit on a smaller scale in Tamil, on historical and comparative grammatical thought (Shanmugam 1969, Mallika 1981). These works are by linguists; the studies by language scholars are not on historical development of grammatical thought, but rather on the chronological history of traditional grammars (Ilavarasu 1963, Ilankumaran 1988) and comparison between commentaries on sutras in grammars (Aravindan 1968). A third way in which there is some interest among a few linguists, like S.V. Shanmugam, is to study the sociology of grammars by describing the social milieu in which grammars emerge in di¨erent periods of time in the history of language. The emphasis in modern linguistics on authentic language data for grammatical analysis, coupled with the divergence mentioned above between grammars of di¨erent periods and the language of the literature and inscriptions of those periods, initiated the study of literary and inscriptional texts for the purpose of writing their grammars. This became the common topic of research at master's and doctoral levels. Series of text grammars were produced covering most of the period of Tamil language. But there are a very few grammars of a historical period as a whole like Agesthialingom's (1983 in Tamil) and Rajam's (1992 in English) grammar of the language of the Sangam period based on all the literary texts of that period. Linguists noted that the language of inscriptions deviated from the language of their contemporary literary texts and attributed the divergence to the in¯uence of the spoken language on the language of the inscriptions (Veluppillai 1971 in Tamil, Veluppillai 1966b and Agesthialingom and Shanmugam 1970 in English). This attribution makes the modern linguists believe that there are data, through incomplete, available for spoken Tamil throughout the historical period. This contributes to the theory that diglossic division of Tamil has a considerable history. The text grammars are the inputs for dealing with questions of language change historically in Tamil and they contribute to writing the history of the Tamil language with information on the spoken language from inscriptions. The culmination of research on the language of texts is T.P.
An Interpretive Survey of Tamil Studies in Tamil
149
Meenakshisundaran's (1965) A History of Tamil Language in English. Orientation to historical change in terms of the grammar of the language in di¨erent periods is found in Veluppillai (1966a), a Sri Lankan linguist. It is a re¯ection of the state of research on historical research on Tamil language by Tamil linguists that no other major work has been done in this area after these books other than a few doctoral dissertations in English such as Mathaiyan (1980). It took twenty two years (1987) to have a Tamil translation of the book by Meenakshisundaran. There are, however, studies on the history of the Tamil language in the framework of treating Tamil as a cultural icon (see below), of which the leading one is Pavanar (1967). The monumental Tamil Lexicon of the thirties under the editorship of S. Vaiyapuri Pillai (1824±1939), a language scholar, is partially historical, but it antedates these text grammars and is based on the knowledge of the texts of the traditional scholars. A major concern of Tamil scholars is to ®x the dates of Tamil literary works published from palm leaf manuscripts. The language of these works is an obvious source for clues to decide on the relative chronology of literary works. Vaiyapuri Pillai, who was criticized blithely by traditional language scholars for his Sanskrit bias, was chief among the Tamil scholars to make use of this technique in his editing work and study of literary works. The linguists are not interested in this kind of use of linguistic evidence for literary history, or to use their professional knowledge of historical changes in Tamil language for this purpose. That is, the historical studies of the Tamil language by linguists are not applied fully to the historical questions of Tamil literature. Being dissertations in universities, these text grammars are mostly in English. Research publications in Tamil in good number on textually supported historical changes in the Tamil language would have had an e¨ect on language scholars, who believe in the constancy of Tamil and limit the changes in the language in its recorded history of two thousand years to a few that are authenticated in traditional grammars. This is a failure of modern linguists in Tamilnadu, who did not contribute to changing the way of thinking in traditional grammars and by language scholars about language change being exceptional. They did not contribute to general historical linguistics either by pointing out new phenomena or by providing evidence from Tamil. 3. Linguistic Analysis of Grammar The traditional way of doing grammar is to analyze and describe the language as a whole. The modern period brought the practice of choosing speci®cally selected, narrow topics for description because of the changes in the
150
E. Annamalai
pattern of learning and communication. Many such selected topics have been discussed in the two traditional journals mentioned above by traditional scholars of grammar. This trend is accentuated by modern linguistics. As a matter of fact, full grammatical description of the Tamil language has become a rarity. After ®fty years of modern linguistics, there is no comprehensive grammar of Tamil of the modern or any other period, except perhaps the Sangam period. The available grammars in English are on one or another part of the grammar, have one or another theoretical or descriptive orientation, and soon become dated or serve a limited purpose such as use in second language teaching. The two grammars written in Tamil are Paramasivam (1983) and Kothandaraman (2002), both of which aim at bridging traditional grammatical description and modern linguistic description, but fall between two stools. The latter is more a collection of facts of modern Tamil, many of whose equivalents were ignored in traditional grammar, than an analytical description of the language. There is no full ¯edged grammar of spoken Tamil in Tamil, which re¯ects the hold on linguists of the grammatical tradition and social beliefs about grammar held by the speakers of Tamil. Like the grammars of texts mentioned earlier, there are many grammars of dialects, which are primarily descriptions of their phonology and morphology. These dialect grammars, being `standardized' productions, do not add to the knowledge of the grammatical structure of Tamil, other than bringing attention to the variable phonemic inventories and morphological forms in di¨erent dialects based on caste and region. There is no consolidation of the large amount of work on the grammars and lexicons of dialects to produce a variable grammar of the whole language or to produce a dialect atlas of Tamil. The study of dialects of Tamil is, by and large, descriptive, looking at each dialect in isolation as if the dialects are independent entities; it is also open ended in the sense that the speech of each caste, even sub-castes in each region, or even a village, will be a separate dialect. There is no attempt to determine which social and geographic boundaries based on sociological and geographic properties are likely to produce distinct dialects, and what kind and extent of linguistic di¨erences will make a distinct dialect. At the same time, the study of dialects is not sociolinguistic in methodology or theory, and there is not much work in the nature of variation studies. Almost all the studies of dialects are in English, as are the ones on spoken Tamil. These publications in Tamil may not have attracted the students in Tamil departments. There are a number of papers and collections of papers dealing with speci®c points of the grammar of spoken as well as literary Tamil. In the proceedings of the All India University Tamil Teachers' annual conference for the period of ten years (1969±1978), 186 papers are listed under the research
An Interpretive Survey of Tamil Studies in Tamil
151
category traditional grammar and 333 under linguistics (Saktivel n.d.), which include papers on formal grammar, that form the majority, and on dialect features and language use in society and various genres such as literature and media. The authors of papers in each category include linguists (those who have at least master's level training in the subject) and language scholars (those who have at least master's training in language and literature and diploma level or short term training or self-learning in basic linguistics). The linguists make up nearly a hundred percent of the second category (linguistics), and their presence is not a small minority in the ®rst category (traditional grammar), as could be expected from the preceding section. Roughly two in three papers on `linguistic topics' are written in English, and one in eight on `grammar topics' is in English. This shows the kind of audience the authors writing on the two categories of topics intend to reach. The papers on linguistic topics, however, have not drawn the attention of linguists outside Tamilnadu because of their super®cial treatment from a theoretical point of view and a narrow focus on TamilÐthey never suggest parallels of the grammatical phenomenon in other languages reported in linguistic literature, not even the grammatical phenomenon in the languages in India. They end up as research by Tamil-speaking linguists for their compatriots. Their contribution to knowledge is fragmentary and it does not substantially increase the knowledge of the grammar of Tamil. One will be hard put, for example, to list ten Tamil grammatical phenomena about which the knowledge was substantially increased by modern linguistic research. Such theoretical research is seldom published in Tamil. The advancement in research from modern linguistics is in formulating the problem in the idiom of linguistics; as the idiom comes from the prevailing theoretical model of the time, the formulation tends to lose its relevance with the change of the model. This is true of studies published in English. The doctoral dissertations on Tamil language, which may be in the range of 500 in the period under the survey, deal with larger areas of grammar like tense and aspect, case system, relative clause, auxiliary verbs etc and assemble a range of data not scrutinized before, but they also su¨er from a mechanical application of theories, which change without any input from research on Tamil language. The dissertations are not valuable for the analyses they make as for the data they marshal. The trend in Tamil research is not to subject the same grammatical phenomenon to new analysis with new theoretical insights and advancement, but to move to a new phenomenon that has unexplored data. The dissertation topics are generally not pointed enough to force the researcher into a deep understanding of the conceptual issues and critical understanding of previous research on the topic; they are allowed to be broad enough to cover a large area of the grammar and to be amenable to collection of a huge amount of data.
152
E. Annamalai
The linguistic research opened new areas for detailed studies in the domain of grammar such as syntax and contrastive grammar. The traditional grammar also deals with syntax, but it is grounded in morphology as found in its description of subject-verb agreement, case markers and the overlap between them, endings of non-®nite clauses. The modern syntactic studies, most of which are grounded in the generative grammar model, try to seek evidence in Tamil for the ®ndings in that model from time to time. This approach leads to selectivity of data, giving preference to data that are of interest to the model at a given time and are parallel to the data widely considered in that model, mostly from English. The Tamil grammatical research is happy to be in the periphery responding to the questions raised at the centre, mostly a½rmatively, occasionally with minor modi®cations to the answers provided at the centre, which, however, are largely ignored. The desire for the nod from the centre explains the fact that much of this research is published in English. This contrasts with the traditional grammatical study of earlier times, which dealt with the Sanskrit tradition at the centre by resisting being in the periphery and by synthesizing the ideas from the centre with its own. None of the grammatical studies of Tamil was written in Sanskrit. The synthesis is evidenced in the earliest grammar of Tamil, Tolkappiyam, but its conceptual and analytical traditions of Tamil have remained distinct; further synthesis over time has not obliterated the original distinctions even when there were e¨orts from the medieval period to draw concepts and modes of analysis from Sanskrit for wholesale application to Tamil data. When the centre shifts to the west in the modern period, the linguists use its concepts and modes of analysis not for synthesis with the grammatical tradition, but for application to Tamil data. The tradition does not inform the current ways of doing grammatical analysis and has become sidelined or understood merely in terms of the ideas of the western centre. Though the grammar of Sanskrit and the ways of analyzing it were in the back of the mind of the traditional Tamil grammarians, and descriptions of some grammatical phenomena of Tamil could be traced to the analyses of Sanskrit grammar, explicit contrasts between aspects of Tamil and Sanskrit grammars are rare. Research contrasting grammars of two languages is new in the modern period motivated by its usefulness in second language teaching (of Tamil to other language speakers and other languages to Tamil speakers) (see section 11 of Ramaiah 1995, which lists 112 titles, all of which are in English). The preferred language for contrast is English, though there are contrastive studies with Hindi and other Indian languages. Contrastive study, which is a favourite topic for doctoral research in Tamilnadu universities, generally has a large canvas covering a major part of the grammars of the two languages. There is no theoretical interest
An Interpretive Survey of Tamil Studies in Tamil
153
in the implications of grammatical contrast between two languages. The interest in contrastive grammars has not developed into, or has not provided impetus for, the study of the typology of grammar, probably because of its focus on applied linguistics, which is interested in surface grammatical differences. There is actually no respectable research in English or in Tamil on typological issues raised by Tamil grammar. The achievement of modern linguistic studies of Tamil grammar, besides expanding the language base to include the spoken language and nonliterary written language mentioned earlier, lies in showing the inadequacy of the traditional grammar in itself (for example, regarding aspect and mood, co-reference, subordination etc) and in reframing the questions of grammar outside the purposes of literary appreciation and moral philosophy. The grammar is studied for its own sake or for understanding human culture, cognition and uniqueness. There is very little in the study of Tamil grammar in English or in Tamil in the latter frameworks. The understanding has increased somewhat in the former framework as regards grammatical systems (like phonology, tense, adjective etc, which are di¨erent from grammatical processes (like compounding, modifying, conjoining etc), about which traditional grammar does not provide any advanced knowledge. 4. Research in Hyphenated and Applied Sub-Fields Modern linguistics extends the study of language to include its relation to the culture and society of its speakers and its acquisition by its speakers. These extensions require bringing linguistics together with its sister disciplines of anthropology, sociology and psychology. Tamil linguists do not have any training or critical reading in these disciplines. Nevertheless, they study issues in ethnolinguistics, sociolinguistics and psycholinguistics and publish mostly in English. Of these, sociolinguistic research is the largest. Ethnolinguistic research is restricted to kin terms and occupational vocabulary, which by and large consists of lexical collection and classi®cation. Psycholinguistic research is on ®rst language or bilingual acquisition (in which Tamil is one of the languages) mostly based on linguists' own children. Studies on language pathology in a clinical context are done at the All India Institute of Speech and Hearing, Mysore. All this research does not amount to much in English, and it is almost non-existent in Tamil. Psycholinguistic research and ethnolinguistic research do not have separate sections in Tamil bibliographies (Ramaiah 1995). The bulk of sociolinguistic research consists of dialect studies. However, this is not sociolinguistics in theory or methods, as mentioned earlier, except that it is about dialect variation and it includes speech varieties of one kind
154
E. Annamalai
of social group or caste. A bibliography (Ramaiah 1995) lists 288 titles under Dialectology, almost all of which are in English, except the ones on Ja¨na dialect and Muslim dialect. These studies focus on identifying dialects and describing them as static entities. They are not strictly dialectological studies, as they do not focus on the spread of isoglosses and mapping them. They should be called descriptive grammars of dialects. They are not even variation studies of language use by people of di¨erent social and regional backgrounds. A very few studies of this type (in English) are based in urban settings. They are co-relational like Labov's, not interactivecommunicative like Gumperz's. Their interest is neither in studying the Tamil speech community and its use of language variation nor in the speech event describing ethnographically who speaks what to whom, when and how like Hyme's, but in studying the language in its variant forms. Tamil diglossia is a favourite subject of research, most of which is published in English. It is studied as a property of the language, as originally proposed by Ferguson, focusing on the grammatical and lexical distance between H variety and L variety, the mutual in¯uence between them and the heterogeneity of each variety. The diglossic distinction is often con¯ated with the style distinction between speaking and writing since the distance between spoken Tamil and written Tamil is large and they are functionally exclusive of each other. Diglossia is also studied as a property of the speech community, as suggested by Fishman, focusing on social contexts in which H and L varieties are used and the expansion of contexts for using the L variety. The study of language contact is about the contact of the minority languages in Tamilnadu with Tamil and the consequences of the contact to the minority language and its grammar. The former includes studies, mostly in English, of language shift comparatively across generations, and the latter includes studies of convergence identifying the grammatical features and systems of Tamil in Saurashtri, Marathi, Telugu and Kannada spoken in Tamilnadu for centuries. There is no big corpus built of their actual speech. Code-mixing is another area of research of bilinguals in Tamil and English, which describes the lexical and grammatical nature of mixing and the social functions of mixing. There is again no corpus of actual mixing of codes. There is no report in English or Tamil of mixing of Tamil with minority languages by their speakers, which must be happening along with the convergence of latter's grammars with Tamil grammar. Longitudinal demographic studies of multilingualism using the statistical data of censuses are missing to learn about the language shift or language attitude / ethnic identity shift or change in the multilingual composition of the state. After the formation of the state in 1956 on the principle of having Tamil as the majority language, Tamil legally became the o½cial language of the
An Interpretive Survey of Tamil Studies in Tamil
155
state and needed to adapt itself to the new role. This inspired studies on development and modernization of Tamil, which consist of documenting changes, mostly lexical, in modern domains of language use like the media, evaluating the use of Tamil in o½cial domains and suggesting ways for successfully implementing the policy. Strangely, the studies by linguists published in Tamil are a minority. Language scholars have published more in Tamil on this subject. The major task is considered to be creation of terms and there is political debate about the source for drawing the needed terms in full form or coining them from available roots and words, whether it should be old Tamil literature, inscriptions, dialects or English or a combination of all of these. With the English source, the question is about choosing between translation and transliteration. The debate is on ideological lines like purism and legacy. It extends to script reform to regularize irregular characters, alphabet expansion to allow new characters for the new sounds in English terms and changing spelling conventions about word initial and ®nal consonants. The linguists have joined language scholars as well as non-linguists in the debate and published books and papers (Shanmugam 1978). The linguists' writings have ideological underpinnings too like others, which include their claim to scienti®c objectivity. In such matters, the linguist and the speaker of Tamil coalesce even when the linguist claims to write as a professional. An interest in words is paramount in Tamil society. Words de®ne the language for the Tamil speakers. This is re¯ected in the work of linguists also. Their contribution to term creation mentioned above is the study of terms used in public administration, law, science and the media as well as the study of origins and history of technical terms. The Tamil University has a department of Tamil in Science (aRiviyal tamiZ), whose faculty members have training in linguistics and which publishes studies on the above aspects of term creation. A major preoccupation of language scholars and hobbyists is to `re®ne' the terms, which includes replacement of loans with pure Tamil words, preserving etymological isomorphism with English originals and shortening phrases into compounds. Their excitement is shared by linguists despite their training. The same Tamil University has a project to compile a dictionary of pure Tamil words in its Department of Lexicography. Discovering loan words in Tamil is an obsessive pursuit for language scholars. The linguists are also interested in loan word studies, but they are more tolerant of commonly used loans and new loans from English in technical ®elds. They, however, accept the ideology of purism when it comes to Sanskrit, whether it is coining or borrowing new terms, as an additional source for lexical enrichment of Tamil. The studies of the etymology of words, embedded in purist ideology, culminate in the project of the
156
E. Annamalai
government to make an etymological dictionary of Tamil, with G. Devaneya Pavanar, a cultural nationalistic philologist, as the founding editor. The word studies of linguists have not resulted in compilation of a historical dictionary of Tamil or a dialect dictionary using all research outputs from the descriptions of literary texts and dialects. The interest in words, on the other hand, extends to the study of names of place and persons in which linguists along with language scholars get engaged. The lexical interest does not extend to legitimate linguistic subjects like word formation, lexical relations and lexical meaning as much as one would expect. The linguists specializing in lexicography as an academic discipline have written papers on these, but not much in Tamil. This is noteworthy given the interest of Tamil speakers in words, mentioned above. The recent interest among linguists in these topics in the context of preparing tools for natural language processing does not translate into publication of their work in Tamil. That would be of great interest to language scholars and hobbyists. The academic lexicography specialists have not engaged themselves in making practical dictionaries, which help to understand the lexical semantic nature of the language and will ful®ll the needs of the language users. It is left to private initiatives like those of CreA, a publishing ®rm and Mozhi, a public trust, to create, using the expertise of linguists, Dictionary of Contemporary Tamil and Dictionary of Idioms and Phrases of Contemporary Tamil respectively, which are linguistically solid and methodologically sound. These dictionaries have become databases for lexical semantic research by linguists in a limited way. The interest of linguists in technical terms takes them to exploring issues of translation at the lexical level, and by extension at the text level. Linguists o¨er courses in translation in their departments, but their studies of linguistic problems of translation is anecdotal and is con®ned mostly to culture speci®c words. The interesting thing about teaching and research on translation by linguists is that they are not addressed to the students in the literature departments and are not done in Tamil. It is a good opportunity lost to make translation really an applied branch of linguistics by good linguistic research on the subject and ful®ll the perceived need to enlarge the skill set of the students of literature, text book translators, news translators, copy editors in advertisement and publishing. Linguistic research on discourse analysis, conversational analysis, linguistic concepts and methods in literary analysis will take linguistics beyond the closed wall of linguists to those interested in literature, culture and society, who are looking for new ways of doing their work. This kind of research relating to Tamil is almost absent both in Tamil and in English. The linguistic analysis of literature is done by scholars in literature departments by self-learning linguistics with no help from linguists beyond giving
An Interpretive Survey of Tamil Studies in Tamil
157
perfunctory and prepackaged orientation or refresher courses in basic linguistics to teachers from other disciplines. Linguists in India became teachers of second languages by historical accident when they got training in modern linguistics in American universities while they taught their languages to American students. This match of expertise continues in India joined by the body of teachers of English as second language for adults speaking Indian languages. This extends to teaching second languages to children in schools, but it does not extend to ®rst language teaching by training or experience. Linguists take upon themselves to write on ®rst language teaching by analogy of their second language teaching experience. Almost none of the writings by linguists on second (or ®rst) language teaching is in Tamil. It was already mentioned that no contrastive study of grammars has been published in Tamil and the same is true of error analysis and other topics of research in this branch of applied linguistics. Linguists have published introductory books in Tamil in many branches of linguistics such as descriptive linguistics, transformational linguistics, semantics, sociolinguistics, dialectology, stylistics and lexicography. These books are derivatives of western books in English with Tamil examples to illustrate speci®c points. They are not a synthesis of research done on Tamil nor do they contain ideas that will spur new research on Tamil. They have been useful to non-linguists interested in language to give them some familiarity with linguistics and to the students of linguistics at master's level (There is no under-graduate programme in linguistics in Tamilnadu) as supplementary materials to understand the text books in English prescribed for their courses. There is a Tamil glossary of technical terms in linguistics to help textbook writers and students. A dictionary with de®nitions of linguistic terms started in the Tamil University is not completed. 5. Tamil as a Cultural Icon The Tamil language is more than a medium of communication for its speakers. It is an icon of their culture to rally around and protect. This is contrary to the conceptualization of language by linguists and they ignore this aspect of speakers' relation with their language when they do scienti®c linguistics. Sociolinguists do not go beyond looking at Tamil as a symbol of social identity relevant for their study of language behaviour and language con¯icts. It is left to historians to study Tamil as a cultural icon and the passion that it inspires (Ramaswamy 1997). Being in English, it is for others to know about Tamil's iconic relationship with their language, but such a study in Tamil will draw the language scholars into the discursive analytical universe and get their inputs. The view of their language as their cultural
158
E. Annamalai
icon is of paramount importance to its speakers not only to ®ght their political battles but also to de®ne their cultural existence. Many studies have been published on this aspect, most of them in Tamil, by language scholars. There is one historical study of the attributes used to characterize Tamil in the literature, which are indicative of the emotive cultural conception of Tamil by its litteÂrateurs. Some of these studies have been written in English also by language scholars (Pavanar 1966) in order to get the attention and acceptance of non-Tamils, primarily westerners, of their agenda of language study, which they believe will give it the legitimacy of serious work, or at least a recognition of it as an alternative stream of language study. Studies of this kind fall into two categories. One is to make use of scienti®c linguistic research by others for building their case and constructing the icon. The other is to make their own study using the facts believed to have been ignored by others or reinterpreting di¨erently the facts used by others. An example of the former is the writings on Lemuria, a land mass believed to have existed connecting Asia and Africa, where the Tamils and their language are believed to have originated and spread to the rest of the world. This story takes Tamil to be the mother of all languages (like Sanskrit is claimed to be by counterparts in the northern part of India). At a di¨erent level, Tamil is equated with Dravidian, making use of Caldwell's Comparative Grammar of Dravidian or South Indian Languages (1856), speci®cally the structural proximity of Tamil to Proto-Dravidian in Caldwell's reconstruction and his demonstration of absence of genetic relation of Tamil with Sanskrit. An example of the latter category is the reinterpretation of the Indus script and its other artifacts with regard to the language and culture of Indus civilization. Such interpretations claim that the Indus language is Tamil and the culture is Tamil (as opposed to interpretations relating both to Sanskrit). This culturally motivated research runs in parallel with the mainstream research both of which are evidence based, but they di¨er in the acceptance of cannons for interpreting evidence in research. The motive to prove the primacy of Tamil among the languages of the world (Pavanar 1953) naturally stimulates etymological studies. One aspect of these studies is to prove that Sanskrit loans considered crucial to Tamil culture are of Tamil origin etymologically. It countervails the Sanskritic group of researchers, who claim any word similar in Sanskrit and Tamil is of Sanskrit origin. This aspect of research draws inspiration from the research of Indologists like T. Burrow, which shows words of Dravidian origin in Vedic texts. For the Tamil scholars, Dravidian and Tamil are synonymous. The other larger aspect of this etymological study is to prove that all languages of the world are derived from Tamil. There are amateur etymologists who have published derivations of words in English and other European as well as other languages including Mayan from contemporary
An Interpretive Survey of Tamil Studies in Tamil
159
Tamil words. The di¨erence between such work and professional works like the historical-comparative study of Tamil and Japanese (by Ohno 1980) is in the degree of arbitrariness and randomness in phonological and semantic correspondence and in the extent of the discarding of methodological consistency. The Tamil scholars use contemporary Tamil data to make claims of historical relationship of a few millennia old because, in their belief (i.e. axiom), Tamil has the unique quality of unchanging over time (i.e., remaining a virgin, in their metaphor). The studies of the primacy of Tamil are a leap from the studies of the antiquity of Tamil that try to show that the Tamil language (and culture) is as old as Sanskrit, if not older. These studies done in the context of Tamil nationalism during colonial times were the works of people trained in the disciplines of history and law, who turned themselves into language scholars. They used literary evidence to extrapolate linguistic antiquity. From such studies derive the studies of language identity. The studies of language identity by language scholars di¨er from identity studies by sociolinguists in following respects. The sociolinguists relate language identity with Tamil ethnicity and contemporary political divisions. For language scholars, relating language identity with Tamil ethnicity is in the realm of high culture and is much more historical. It is more speci®cally in contrast with Sanskrit, and derivatively in contrast with Hindi when it comes down to modern politics. This contrast with Sanskrit can be sensed through the literary history of Tamil from the beginning. The historical strand of the Tamil identity that is inclusive of divinity and antiquity of language can be found in the studies of modern language scholars of Tamil. Their studies relate Tamil identity with Tamil religion, which, in their view, is a variety of Saivism fostered in Tamilnadu, as evidenced in the studies of scholars like Maraimalai Adigal. Such a correlation is not a part of language identity studies by sociolinguists, whose correlations with Tamil identity are temporal or secular. Because of the fundamental di¨erences between sociolinguists and language scholars in their concept of language and in their analytical tools, the studies of the latter are placed in the disciplines of social history or cultural politics, and not in linguistics. The divergence in the studies of these two groups remains un-bridged. If linguists do research on language as a cultural object or institution also using their analytical tools, they may have some impact on the language studies of language scholars in this area.
6. Conclusion The linguistic studies on Tamil published in Tamil by linguists exhibit a tension between addressing local and non-local concerns about language.
160
E. Annamalai
This tension is seen surfacing in the international conferences sponsored by the International Association of Tamil Studies in any country which has a strong Tamil community. The linguists try to resolve the tension by ignoring the area of language as a cultural icon, by not publishing in Tamil their work on hyphenated and applied sub-areas of linguistics and by their grammatical research that will illuminate traditional grammatical questions. The last mentioned area of research has helped the interested linguists in understanding the issues of Tamil grammar formulated in the traditional grammar by reformulating them in the framework of linguistics they are trained in at a given period of time. It cannot be said, however, that it has changed the way the traditional grammarians do their grammatical study. It has perhaps increased the acceptance of linguistics among the younger generation of students studying traditional grammars in Tamil language departments. The grammatical studies published in Tamil on speci®c aspects of Tamil grammar in the framework of modern linguistics remains fragmentary and insubstantial. To really conclude, the studies available in Tamil do not o¨er any new approach or ®ndings that will inform one's research on any language including Tamil, except in one area, which is the history of grammatical thought in the world. References Agesthialingom, S. (1983). canka-t-tamiZ, vols. 1, 2, 3. (Tamil of Sangam (literature)). Annamalainagar: All India Tamil Linguistics Association. Agesthialingom, S. and K. Balasubramanian (1974). ilakkaNa aayvu-k kaTTuraikaL (Essays on Grammatical Research). Annamalainagar: Annamalai University. Agesthialingom, S. and K. Murugaiyan (eds.) (1972). tolkaappiya moZiyiyal (The Linguistics of Tolkappiyam). Annamalainagar: Annamalai University. Agethialingom, S. and S. Saktivel (1973). A Bibliography of Dravidian Linguistics. Annamalainagar: Annamalai Univerisity. Agesthialingom, S. and S. V. Shanmugam (1970). The Language of Tamil Inscriptions. Annamalainagar: Annamalai Unviersity. Andronov, M. S. (1964). Materials for a Bibliography of Dravidian Linguistics. Tamil Culture 11: 3±50. Annamalai, E. (1989). tolkaappiyarin muunRu moZiyiyal karuttukkaL (Three linguistic concepts of Tolkappiyar). moZiyiyal. Aravindan, M. (1968). uraiyaaciriyarkaL (Commentators). Chidambaram: Manivasagar Noolagam. Caldwell, Robert (1856). A Comparative Grammar of Dravidian or South Indian Family of Languages. Revised and edited by J. L. Wyatt and T. Pillai Ramakrishan (1913, third edition). London: Kegan Paul, Trench and Trubuk. Dhamotharan, A. (1978). Tamil Dictionaries: A Bibliography. Wiesbaden: Franz Steiner Verlag. Ilakkuvanar, S. (1971). tolkaappiya aaraaycci (Research on Tolkappiyam). Pudukottai: Valluvar Publications (second edition).
An Interpretive Survey of Tamil Studies in Tamil
161
Ilankumaran (1988). ilakkaNa varalaaRu (History of (Tamil) Grammars). Chidambaram: Manaivasagar Noolagam. Ilavarasu, Soma (1963). ilakkaNa varalaaRu (History of (Tamil) Grammars). Chidambaram: Manivasagar Noolagam. Kothandaraman, Pon (1992). ikkaala-t-tamiZ ilakkaNam (A Grammar of Modern Tamil). Chennai: PuumpoZil. Krishnamoorthy, G. (1990). tolkaappiya aayvin varalaaRu (History of Research on Tolkappiyam). Madras: University of Madras. Mallika, P. (1981). ilakkaNa marapil iTaiccoRkaL (`Grammatical Morphemes' in (Tamil) Grammatical Tradition). Palghat: Baskaran Publishers. Mathaiyan, P. (1980). Adverbial Participles in Tamil: A Historical Study. Madras: University of Madras (Unpublished thesis). Mathaiyan, S. (1978). tamiZ nookkunuul aTaivu (A Bibliography of Tamil Reference Works). Madras: Sakuntala. Meenakshisundaran, T.P. (1965). History of Tamil Language. Pune: Deccan College. Ð (1974). Foreign Models in Tamil Grammar. Thiruvanthapuram: Dravidian Linguistics Association. Natarajan, T. (1977a). canka kaala eZuttiyal (Phonology (of Tamil) in the Sangam Period). Madurai: Sarvodya Ilakkiya Pannai. Ð (1977b). The Language of Sangam Literature and Tolkappiyam. Madurai: New Century Publishing House. Ohno, Sususmu (1980). Sound Correspondence between Tamil and Japanese. Tokyo: Gahashuitt University. Paramasivan, K. (1983). ikkaalat-t-tamiZ marapu ((Grammatical) Conventions of Modern Tamil). Sivagangai: Annam. Pavanar, Devaneya G. (1953). mutal taaymoZi (The Primary Mothertongue). Madras: Saivasiddhanta Kazhagam. Ð (1966). The Primary Classical Language of the World. Katpadi: Nesamani Padippagam. Ð (1967). tamiZ varalaaRu (History of Tamil). Katpadi: Nesamani Padippagam. Rajalaksmi, J. (1980). History of Tamil Grammar. Annamalainagar: Annamalai University. Unpublished Ph.D. thesis. Rajam. V.S. (1992). A Reference Grammar of Classical Tamil Poetry. Philadelphia: American Philosophical Society. Rajaram, S. (1992). viiracooZiya ilakkaNa-k kooTpaaTu (Grammatical Theory of Virachozhiyam). Nagercoil: Iragavenil. Ramaiah, L.S. (1995). An International Bibliography of Dravidian Languages and Linguistics. Vol. 2. Tamil Language and Linguistics. Madras: T.R. Publications. Ramaswamy, Sumathi (1997). Passions of the Tongue: Language Devotion in Tamil India, 1891±1970. Berkeley: University of California Press. Saktivel, S. (n.d.). aayvukkoovai nuultokuti (Index of aayvu-k-kooavai volumes 1969±1978). Annamalainagar: Indian University Tamil Teachers Association. Shanmugam, S.V. (1969). iTaikkaala ilakkaNankaL: neeminaatam (Medieval (Tamil) Grammars: Neminatham). tamiz-p-poZil 42.6. Ð (1978). eZuttu-c-ciirtiruttam (Orthography Reform). Annamalainagar: All India Tamil Linguistics Association. Ð (1980). eZuttilakkaNa-k-kooTpaaTu (Theory of Phonology). Annamalainagar: All India Tamil Linguistics Association. Ð (1986). collilakkaNa-k-kooTpaaTu (Theory of Morphology). Annamalainagar: All India Tamil Linguistics Association. Thaninayagam, Xavier (1966). A Reference Guide to Tamil Studies. Kuala Lumpur: University of Malaya Press.
162
E. Annamalai
Vaiyapuri Pillai, S. (ed.) (1924±1939). Tamil Lexicon (6 volumes). Madras: Madras University. Veluppillai, A. (1966a). tamiZ varalaaRRilakkaNam (Historical Grammar of Tamil). Madars: Pari Nilaiyam. Ð (1966b). Study of the Dialects in Inscriptional Tamil. Trivandrum: Dravidian Linguistics Association. Ð (1971). caacanamum tamiZum (Inscriptions and Tamil). Kanchi: National Printers.
North America TEJ K. BHATIA
1. Introduction In the aftermath of the September 11, 2001 terrorist attack, the United States found itself unprepared to confront the future challenges of globalization and international education. The General Accounting O½ce reported that four agenciesÐthe Army, the FBI, The State Department and the Commerce Department's Foreign Commercial ServiceÐfound themselves ill-equipped to meet such challenges, largely due to the acute shortage of sta¨ with foreign language skills (see Stephen Barr. Looking for people who can talk the talkÐin other languages. Washington Post, March 12, 2002). The long-neglected languages and area study of South Asia, the Middle East, and other regions gained new strategic importance. With this came a new major turning point in the areas of teaching, learning and research of South Asian languages and linguistics in North America. These new developments and perspectives are given in the section (3). Although this survey focuses on material published since the dawn of the new century, we also provide, in an appendix, brief resumes of some recent unpublished doctoral dissertations. Given the fact that North America is as diversi®ed, proli®c and productive as South Asia, we make no exhaustiveness claims. The report is presented in ®ve major sections: (1) Theoretical Linguistics; (2) Psycho-and Neuro-linguistics; and (3) Sociolinguistics and Applied linguistics; (4) South Asian language pedagogy; and (5) Reference and Collected Works. 2. Theoretical Linguistics Syntax, Semantics and Universal Grammar Word order variability, the tense-aspect agreement system together with ergative and causative constructions in South Asian languages continue to draw attention from a variety of theoretical and typological perspectives. Lee (2003) deals with the phenomenon of word order freezing and the
164
Tej K. Bhatia
dynamics of hierarchy alignment which leads to freezing. The role of the general phenomenon of ``markedness reduction'' in ruling out the highly marked prominence hierarchies is identi®ed. Based on the analysis of the scrambing languages such as Hindi and Korean within the Optimality Theory framework, this work proposes a universal typological approach to account for word order variation in terms of free and ®xed order on a crosslinguistic basis. Davison (2002) presents a comparative account of tense, aspect, & agreement in two Indic languages, standard Hindi/Urdu, & Kurmali, an ``Eastern Hindi'' language spoken in Orissa. The paper explores the consequences in the di¨erence of agreement systems of the two varieties for Indic languageszations. Lahiri (2002) presents a semantic analysis of the Scope Marking structures with reference to ``expletive wh'' in Hindi and goes on show the inadequacy of the direct dependency analysis to account for ``Scope Freezing'' e¨ects. Morphology and Phonology The works of Singh and Stump deal with the morphology of South Asian languages at the synchronic and diachronic level, respectively. Some notable recent publications on Whole Word Morphology, the model of morphology proposed by Singh, in which the model is applied to various aspects of morphology are: Dasgupta et al (2001), Singh (2001 and 2002), Singh and Neuvel (2003) and Singh and Starosta (2003). Singh's main claim is that only a truly word-based morphology, a morphology that neither entertains nor allows operations on units smaller than the word, can explain what needs to be explained in morphology. Practically all these publications draw heavily on South Asian facts. Historical and diachronic studies are primarily devoted to Sanskrit and Pali morphology. Stump (2001) deals with four analogical developments inthe declensional morphology of early Indic: the full integration of pseudoradical Ãõ-stem nominals into the derivative ii-stem declension in Epic Sanskrit; the emergence of two new, hybrid declensionsÐone for radical ii-stem nominals, the other for feminine i-stem nominalsÐin innovative varieties of Epic Sanskrit; and the full integration of radical Ãõ-stem and feminine i-stem nominals into the derivative ii-stem declension in Pali. It is claimed that these developments all embody a single, general tendency in human language: a preference for declensional systems in which a nominal's membership in a particular declension class is both a necessary and a su½cient correlate of its membership in a particular gender class. In particular, the paper argues that the direction of the analogical in¯uences in a language's declensional system is in¯uenced if not fully determined by this preference: the Epic Sanskrit developments serve to heighten the deducibility
North America
165
of a nominal's gender from its declension-class membership; the Pali development heightens the deducibility of a nominal's declension-class membership from its gender (together with its stem form). In practice, it is common for the paradigm of morphological forms associated with a lexeme's root to be identi®ed with the paradigm of syntactic atoms associated with that lexeme; indeed, this practice is so usual that one might easily think of dismissing the need for any theoretical distinction between a root's morphological paradigm and a lexeme's syntactic paradigm. Based on synchronic and diachronic evidence from Sanskrit and Pali, Stump (2002) argues that the distinction is an important one; in particular, it is shown that the phenomena of heteroclisis and deponence entail a nontrivial linkage between paradigms of the two sorts. The paper speci®cally proposes how to represent this linkage in an inferential-realizational theory of morphology. Janda and Joseph (2002) provide a substantial critique of standard generative accounts of the interaction between Grassmann's Law and Bartholomae's Law, a critique that challenges the standard generative practice of appropriating some of what belongs to morphology to phonology. 3. Psycho- and Neuro-Linguistic Studies Memory, sentence processing and aphasia are the main subjects of the psycho- and neuro-linguistic studies. The writing system of Devanagari exhibits the properties of syllabic as well as alphabetic scripts; therefore it is hybrid in nature. In addition, its non-linear properties of vowel signs render the disparity between spatial & temporal sequencing, which would incur a processing cost. Vaid and Gupta (2002) designed three experiments to examine this issue of processing time in terms of naming speed, accuracy, and writing order for the level at which words in Devanagari are segmented. The results support `a partly phonemic & partly syllabic level of segmentation, consistent with the structural hybridity of the script.' Bhatnagar et al (2002) presents a clinical pro®le of Hindi-speaking stroke patients with aphasia. The interactional e¨ect between age and gender with aphasia type in 97 Hindi-speaking right-handed individuals, the majority of them with a con®rmed diagnosis of a cerebrovascular accident, is identi®ed. The two notable ®ndings are as follows: (1) Although the pattern of the age-aphasia is essentially the same as in the western world, with a notable di¨erence in the mean age of Indian patients which was signi®cantly lower; and (2) `in addition to gender and literacy related di¨erences, many clinical symptoms that are known to co-occur with aphasia were not readily reported by subjects with stroke.' Kanada-English bilinguals' aphasia is accounted for in Ijalba, Obler and Chengappa (2004).
166
Tej K. Bhatia
4. Sociolinguistics and Applied Linguistics A large body of research devoted to sociolinguisitics deals with language maintenance and shift both in native and diasporic contexts. In spite of the recent substantive developments in the area of language maintenance and shift, key questions such as what is meant by the term language shift and language maintenance, and when is language shift believed to have taken place, remain to be investigated. In attempt to answer these key questions, Kumar (2001) examines the complex sociolinguistic contexts which determine the language shift from a minority language, Maithili, to a majority language, Hindi in India. The paper identi®es functional (e.g. formal situations such as School, O½ce, Public Speeches) and formal syntactic domains of language use that facilitate such a shift. The latter case of shift super®cially appears to be code-mixing. The maintenance of Tamil in Malaysia is the focus of Schi¨man (2002). Since the national language policy of Malaysia favors national integration through Bhasha Malaysia and Islam, there is little support for non-Malay schooling beyond the elementary level. Therefore, the future of Tamil seems bleak in Malaysia. Furthermore, even the intense language loyalty does not lead to language maintenance because Tamil prefers corpus planning over status planning. The divisive e¨ects (language loyalty, religious identity etc.) of scripts are underscored in King (2001) with reference to the classic Hindi vs. Urdu controversy. The question of identity and diaspora continues to generate a lot of attention in sociolinguistic studies. The role of religion and media in the spread of bilingualism and identity formation is intriguing. Pandharipande (2001) focuses on following four aspects of language use in the diasporic Hindu speech community in the U.S.: (1) the rationale for choosing religion as the marker of identity, (2) the role and the patterns of language(es) used in the religious discourse, (3) the change in the (Hindu) religious discourse in the U.S., and (4) the issue of ``authority'' which licenses the change in the religious practices (including language use) in the diasporic community. The paper shows that like identity, authority is not a static concept. The linguistic markers of Hindu identity together with the determinants of religious authority and authenticity are uncovered in this pertinent work. Bhatia (2001) presents a pro®le and pattern of Indian diaspora together with the causes which led to the emigration of Indians from India. The role of language and media in identity formation of diasporic Indians is highlighted. The third notable feature of sociolingustic research is discourse in general and advertising discourse in particular. Steever (2002) deals with direct and indirect discourse in Tamil. Ladousa (2002) examines printed school advertising in Banaras, a North Indian city, creates correspondences and argues that in the educational domain, Hindi and English perform complimentary
North America
167
functions. While English and the Roman script mark the ``central'', Hindi and the Devanagari exhibit ``peripheral'' power index. The primary concern of second language acquisition studies is to examine the role of transfer. The debate on the initial stage grammar of second language acquisition is the central objective of Bhatt and Hancin-Bhatt (2002). Based on the pattern of CP (complementizer phrase) acquisition of the Hindi learners of English as a second language, this paper sets out to examine the predictions of two transfer theories: Full transfer (Schwartz & Sprouse, 1994 & 1996) and minimal trees/partial transfer (MT/PT) hypothesis (Vainikka & Young-Scholten, 1996). The data obtained shows that the L2 learners start out without a CP stage and later go on to acquire the CP stage. These ®ndings show the inadequacies of the two theories/hypotheses. It is proposed that structural minimalityÐthat clausal projections are in¯ectional phrasesÐas a hypothesis on the initial state of L2 acquisition. The main objective of Helms-Park (2001) is to examine the role of transfer in the acquisition of verb properties by second language (L2) English learners of Hindi-Urdu and Vietnamese. The study is comparative in nature and was carried out in Toronto, Canada. The study reveals that Hindi-Urdu and Vietnamese groups di¨ered signi®cantly in several semantic contexts. However, both groups show that verb properties are transferred selectively and that transfer plays a role in the di½culty or ease involved in shedding overgeneralized lexical rules. 5. South Asian Language Pedagogy Although the teaching of South Asian languages in North America is a rich and complex problem, not much serious, systematic research has been done on the subject. Some general observations may, however, be useful: 0
0
0
There is a shortage of centers and instructors to teach such languages. Native speakers and instructors quali®ed to teach the language are not available in many communities, and where they are, they can attend to only a limited number of students. Instructional materials for South Asian languages are currently noninteractive, consisting for the most part of textbooks, audiotapes, and paper-based workbooks. The instructor is responsible for classroom teaching, scoring of student work, and preparation of feedbackÐall traditionally labor-intensive tasks that may be di½cult for the instructor to do well or at all. Multimedia training through use of desktop computers is a technology alternative available at some centers that teach South Asian languages. It requires students to use the programs in computer labs, which entails
168
0
Tej K. Bhatia
inconveniences for the students and facilities issues for the college. Furthermore, such programs are still in their infant state. In addition, one of the serious challenges that confront South Asian languages is how to respond to the communicative needs of professionals. The commercial sector has not responded to this need due to low demand and the high production cost associated with non-western scripts. The result is a critical shortfall in our national capacity to educate pro®cient and advanced users of strategically crucial languages of South Asia.
One of the major developments in the capacity building in the ®eld of the teaching of South Asian languages is the establishment of SALRC (South Asian Language Resource Center) funded by the U.S. Department of Education. SALRC (http://salrc.uchicago.edu) is likely to have a major impact on the scope and the nature of the South Asian language pedagogy. Those interested in the teaching of South Asian languages in North America, may also wish to look at The UCLA Language Materials Project, which provides a wealth of information on South Asian language resources (http://www.lmp.ucla.edu). 6. Reference and Collected Works In this section, we take up four recent reference works that contain important contributions to the study of South Asian languages. The 2001 special issue of the Yearbook (Bhaskararao and Subbarao 2001) is devoted to the phenomenon of language contact, convergence and typology in South Asia. It contains most of the papers presented at the international symposium organized by the Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies, Japan in 1999. Syntactic, morphological and phonological typology of South Asian languages with a wide variety of theoretical underpinnings and methodological orientations is covered in this excellent volume. Methodological and de®nitional issues pertaining to the concept of `linguistic area' are addressed by Masica. Other topics covered by the volume are: word order (Hock), morphological borrowing (Singh), code-mixing (Bhatia and Ritchie), compound verbs (Hook), and ergative marking in Hindi, Gujrathi, and Pashto (Starosta), Indo-Aryan and Dravidian contact (Sojberg), Munda (Zide and Anderson) and creolization (Smith). The role of nonlinguistic factors (e.g. religion) in contact and convergence is the subject of the paper by Pandharipande. In 20 chapters, Cardona and Jain (2003) o¨er a linguistic and sociolinguistic pro®le of Indo-Aryan languages spoken in India, Pakistan, Nepal, Bangladesh, Sri Lanka, and the Maldives. The language contact situation
North America
169
of the Indo-Aryan languages with Dravidian, Munda, Tibeto-Burman, Semitic languages (Arabic) and others is also taken into account. North American contributions include the following chapters: the writing system (Salomon), Sanskrit (Cardona), Prakrits and Appabhramshas (Vit Bubenik), Hindi (Shapiro), Magahi (S. Verma), Bhojapuri (M. Verma), Konkani (Miranda), Nepali (Riccadi), Marathi (Pandharipande), and Dardic (Bashir). Out of the 191 chapters in Gary and Rubino (2001), 24 are devoted to South Asian languages. The South Asian languages pro®led in the reference work are the following (when the author is a North-American, his/her name is given after the name of the language-chapter contributed by him/ her): Balochi, Bengali, Bhojpuri (Shukla), Gujarati (Mistry), Hindi (Shapiro), Kannada (Sridhar), Konkani (Miranda), Maithili, Marathi (Southworth), Nepali (Verma), Oriya (Ramachandran), Pali, Pashto (Inozemtsev), Punjabi (Bhatia), Rajasthani, Sanskrit (Witzel), Santali (Anderson), Shona (Carter), Sindhi (Cole), Sinhala, Saraiki, Tamil (Steever), Telugu (Pelletier), Urdu (Siddiqi). Kani¨ka (2001) presents cross-cultural dimensions of `grammatical traditions in contact.' The ®rst part of the book deals with the Indian subcontinent and Tibet. The phenomenon of contact and convergence of Indic and non-Indic grammatical traditions is accounted for with special reference to Sanskrit (Deshpande, Raster, Verpoorten), Tamil (Steever, Niklas), Hindi (Bhatia) and Tibetan (Kaschewsky and Rose) in eight chapters. The fascinating journey of the Indic grammatical tradition to outside India (Arabia, Europe) is accounted for in various chapters throughout the book. Conclusion In a relatively short span of four years, the South Asian linguistics has registered important gains. The research in areas such as morphology and hyphenated ®elds of linguistics (socio-, psycho- and neuro-linguistics) shows signs of vigour and interdisciplinary vitality. It also seems to be a good time to rethink South Asian language pedagogy in North America. Appendix: Some Doctoral Dissertations 1. Coelho, Gail Maria. 2003. A Grammar of Betta Kurumba. Austin: University of Texas at Austin. This dissertation provides a description of the language of the Betta Kurumbas, an indigenous (aadivaasii) ethnic group (population: > articiocco g Intl artichoke > Arabic [È/ard.i ÈSo:k(i)] QòÜí¸ Qò¿1# What is interesting here is that this type of hybridization is not X Y, wherein X is native and Y is (originally) foreign and combined they form a hybrid. Rather, it is X A Y, i.e., the entire expression is simultaneously both native and foreign. I call this phenomenon ``Phono-Semantic Matching''
244
Reviews
(PSM) and de®ne it as ``multisourced neologization that preserves both the meaning and the approximate sound of the parallel expression in the source language (SL), using pre-existent target language (TL) lexemes or roots'' (see Zuckermann, Ghil'ad 2003, Language Contact and Lexical Enrichment in Israeli Hebrew, LondonÐNew York: Palgrave Macmillan; cf. Zuckermann, Ghil`ad 2004, ``Cultural Hybridity: Multisourced Neologization in ``Reinvented'' Languages and in Languages with ``Phono-Logographic'' Script.'' Languages in Contrast 4.2). The phenomenon is widespread and sometimes the same internationalism is phono-semantically matched in two distinct languages, e.g. AIDS entered Modern Standard Chinese as 1ËÅ aõÁzõÅ bõÁng, lit. `love cause/ neutralize disease', i.e. `a disease caused by (making) love'; and Icelandic as eyDni, from eyDa `eliminate, destroy' ni, nominal su½x. Finally, Kuczkiewicz-FrasÂ's book is a much-needed project. I look forward to further research in the ®eldÐboth ``vertical'' (ferreting out the theoretical implications of Kuczkiewicz-FrasÂ's data-base of Perso-Arabic hybrids) and horizontal (such as a volume which could be entitled ``PhonoSemantic Matching'' Hybrids in Hindi: A Socio-Linguistic and Structural Analysis).
Hindustani as an Anxiety Between Hindi±Urdu Commitment* S. IMTIAZ HASNAIN and K.S. RAJYASHREE
1. Prelude There is an obliquity about Hindi and Urdu, a curious slant anchored to its inceptual rudimentsÐjourneying through the pulsations of loaded historicityÐmanifesting itself in the creation of a space with margins shared by no other Indian language. Both Hindi and Urdu provide ample evidence suggesting how interesting cross-currents are at work when a shared linguistic domain is treated as a site for identity politics. Historically, Hindu-Muslim antagonism has not only generated tension between HindiUrdu, between the conceded response of the non-committed and communally untainted speakers, on the one hand, and the calculated response of the demographers and the fanaticists committed to divisiveness, on the other, but has also helped people isolate the two languages and relate them to certain social roles and group identitiesÐHindi as Hindu, Urdu as Muslim. The antagonism has even coerced both the languages to acquire a loaded position, vis-aÁ-vis the other, and in relation to all other languages of India as a potential national language. Hindustani emerged as a response to this turmoilÐa contemplated decision urging towards openness devoid of any fanatic deviousness in its usage. Its adaptation by both Hindi and Urdu speakers alike re¯ects a positive a½rmation. It proves itself as an e½cient move for counter-hegemony against the pristine posture acquired by Hindi and Urdu, against the machinations of communal constructionists of both Hindi and Urdu. Speakers' claims regarding the name of the language they speak and the consequential crystallization of identities emanate from this anxiety between Hindi±Urdu commitment, which is more predicated upon the events which are political rather than linguistic. 2. Situating the Anxiety We begin with a look at anxiety displayed in an autobiographical note given by Rama Kant Agnihotri and a seemingly biographical commitment
248
S. Imtiaz Hasnain and K.S. Rajyashree
emanating out of a ®lial obligation and privilege experienced by Alok Rai. Both writings, drenched with a strong sense of lament, convey similar voices of anguish. Agnihotri, and others belonging to his age, have witnessed a fond sharing of language and culture by the past generations of Hindus and Muslims, followed by the gradual disintegration of those ties: ``All my education was in the naagarii or from the middle school onwards in the Roman script. Yet the so-called Urdu speakers and writers were a very active and meaningful part of my world. From a generation which could share in normal conversation Kabir, Premchand, Ghalib, Faiz, Nirala and Keats among several others and have sustained discourse on them irrespective of whether they read in the naagarii or the Perso-Arabic script or both, I felt we were entering a world of increasing polarization and focusing of social, cultural and linguistic norms. In the small sleepy town of Nahan, in the Sirmaur district of Himachal Pradesh, my parents, very liberal otherwise, forbade me to associate any longer with two of my very dear schoolmates, Shaheed and Anwar. That was the time when I think I lost touch with the idiom of the street and could never return to ¯ying kites and making ®recrackers. My father constantly told me that Muslims have been great friends, always sincere and honest, but `you know' etc. He was also convinced that in the interest of national unity all of us should speak `pure Hindi' and write in the naagarii script though he could never do either himself. . . Later, when I came to Delhi I made more friends speaking di¨erent languages. Very often some friends would get together in the evening in the lawns of India Gate, recite poems and critically evaluate them and in the end almost always return to the increasing separation between the language of the people and the new `pure Hindi' of All India Radio, nascent Doordarshan, Hindi newspapers and magazines and most of all, of Hindi literature. Though literature from the other side of the border was not easily available, I at that time assumed that comparable developments must be taking place there as well. Of course, I subsequently realized that Urdu/Hindi was the native language of a very small group in Pakistan even though Partition was considerably predicated on the association of Muslims with Urdu. In 1970, I found a copy of the poems of Faiz (1969) in the naagarii script. I was simply overwhelmed by the simplicity and power of some of his poems and asked my friends and myself whether he was simply not still the greatest Hindi / Urdu poet. I think it is in Faiz, Kai® Azmi, Dhumil and other similar poets, in both India and Pakistan, that we may trace the natural descendants of Kabir or Tulsi . . . Are poems like ham jo taariik raaho me maare gaye, yah fasl umiido kii hamdam, shaam, kab Thahregaa dard, paas raho etc., Hindi or Urdu poems? Decided for yourself: jab terii samandar aakho me yah dhuup kinaaraa, shaam Dhale milte hai dono vaqt jahaan jo raat na din, jo aaj na kal
Hindustani as an Anxiety Between Hindi±Urdu Commitment
249
pal bhar ko amar, pal bhar me dhuaan is dhuup kinaare, pal do pal honTon kii lapak baahon kii chanak yah mel hamaaraa, jhuuTh na sac kyon raar karo, kyon dosh dharo kis kaaraN jhuuThii baat karo jab terii samandar aakhon men is shaam kaa suraj Duubegaa sukh soyenge ghar-dar vaale aur raahii apnii raah legaa'' (Agnihotri, 2002: 31±32) (This verge of light where at sunset, night and day meet; it is neither night nor day; neither today nor tomorrow; immortal for a moment and ever transient; at this threshold of light, for a moment or two, there is that sensuous movement of lips and clinking of bangles; this union of ours is neither true nor false; why complain then and why blame anybody; why lie; when this evening's sun descends in your ocean eyes, men of the world will sleep peacefully and the traveler will be on his way.)
Alok Rai, as many others, seeking to save Hindi from the politics of nationalism, has been arguing that if Hindi has to become a national language of communication then it has to liberate itself from its own repressed history and dissociate itself from its deformed other: ``There are two, widely di¨erent reactions whenever the matter of Hindi is broached: all those who are connected with Hindi-Urdu in any way whatsoever fall instantly into passionate contention; others, who are outside this furious circle are totally bemused . . . Even the simplest questions beget further controversy, but no clari®cation. Thus, consider the following elementary queries: are Hindi and Urdu two names of same language, or are they two di¨erent languages? does Urdu become Hindi if it is written in the Nagari script? is Hindi Hindu? is Urdu Muslim, even though Muslims in distant Malabar have been known to claim it as their mother tongue? The only reasonable, and maddening, answer to all these questions is, well, yes and no. In respect of neither Hindi nor Urdu can one give an unambiguous answer: one has to go into the historical detail to explain how/why it isn't; and then, in the space of a few decades, why it is''. (Alok Rai 2000: 4)
What uni®es the approach of Agnihotri and Alok Rai is their passionate defence of the language of every day life, synonymously referred to as Hindi, Urdu, or Hindustani- not as three distinct linguistic entities, but constituting a single entity aiming towards mutual intelligibility and sharing the social and personal inner space. Both intervene in this contentious domain in defence of the common language of the common man, resisting the
250
S. Imtiaz Hasnain and K.S. Rajyashree
designs of communal constructionists trying to subvert the social mind that enjoys Hindi-Urdu speech and literature in spite of a ``pre-partition partition'', and a narrative of intimate destabilization and dispossession. 3. Brief Historical Background of Hindi-Urdu Antagonism1 The origin of the divide may be traced back to the establishment of Fort William College (henceforth, referred as FWC). With its establishment, the process of divergence between Hindi and Urdu at the linguistic, sociocultural, and literary levels assumed a new dimension. Khadi boli was communalized. Hindu and Muslim writers from far-¯ung places were called to write prose in two styles of Khadi boli making use of two di¨erent scripts: Devanagari and Perso-Arabic. While Lallu Ji Lal and Sadal Mishra were assigned the task of writing Khadi boli prose in the Devanagari script by using words of Sanskrit origin, Meer Amman, Haidar Baksh Haidari, Sher Ali `Afsos' were under instructions to write Khadi boli prose in the PersoArabic script by using Perso-Arabic words. Prem Sagar (`The Ocean of Love') of Lallu Lal (which was begun in 1804 and completed in 1810) in Hindi and Bagh-o-Bahar (published in 1803) of Meer Amman is the most representative literary works of FWC. However, it is refreshing to note that tangential to such concerted e¨orts to create schisms are sporadic evidences where the expunging of tatsam expressions had a di¨erent motive altogether, as is su½ciently evident in the poetry of Wali. Wali's shift from Dakani to Rekhta during the early eighteenth century may have a semblance of initiating a cultural divide. However, this shift was based on outcome of a cherished ideal of an individual eager to disseminate his talent to a larger literary audience by adopting the re®ned language of the elite connected to the royalty. When Wali, in his ®rst visit to Delhi, met Shah Saadullah Gulshan and read out his poems, Shah Saadullah Gulshan o¨ered him the following advice: ``Make your Dakani language, now obsolete, conform with the Rekhta, which is in accordance with the Urdu-e-Mualla Shahjahanabad. Doing this will give you fame, keep you in step with the manner of the times and make your work admired by men of the ®nest minds and the most impeccable taste.'' (qtd. in Amrit Rai 1984: 226)
This advice brought about a cataclysmic change in the literary attitude of Wali. There was a complete transformation in his style of diction, which at one level was a manifestation of the linguistic engineering carried out at the behest of the social elite, and at another it was indicative of the demands of language standardization which had come into being in the Khadi boli style of Urdu at that time. His couplet written in Khadi boli based Urdu:
Hindustani as an Anxiety Between Hindi±Urdu Commitment
251
mu¯isi sab bahaar khotii hai mard kaa aitebaar khotii hai (In penury one loses all his joy his trust also gets eroded) is a ®ne instance of this transformation re¯ective of the present day Standard Urdu. At this juncture it is important to mention that Wali's conversion was the outcome of a desire to give expression to his writings by adopting a more aesthetically appealing and re®ned form devoid of any sectarian in¯uences. Unfortunately, this was not the case with FWC. Lallu Ji Lal and Sadal Mishra of the FWC `created' a new language called ``Modern High Hindi'' or ``Standard Hindi'' on sectarian lines2 expelling words of Persian and Arabic origin from Urdu. Almost the same view has been expressed by Keay (1920/1960) when he wrote: ``Under the direction of Dr. John Gilchrist he [Lallu Ji Lal] and Sadal Mishra were the creators of modern `High Hindi'. Many dialects of Hindi were, as we have seen, spoken in North India, but the vehicle of polite speech amongst those who did not know Persian was Urdu. Urdu, however, had a vocabulary borrowed largely from the Persian and Arabic languages, which were specially connected with Muhammadanism. A literary language for Hindi-speaking people which could commend itself more to Hindus was very desirable, and the result was produced by taking Urdu and expelling from it words of Persian and Arabic origin, and substituting for them words of Sanskrit or Hindi origin.'' (Keay 1920/ 1960: 83, italics ours)
In the words of Alok Rai, ``. . . the important thing that emerged from FWC is the idea of two-ness, of linguistic duality. Fort William College gave institutional recognition to the notion that there were in fact two ways of doing HindustaniÐone which used the available and mixed language, and another from which the Arabic-Persian words (i.e. words of `Muslim' origin) had been removed in order to produce a language (register? idiom?) more suitable to Hindus.'' (2000: 22) Thus the establishment of FWC brought about the overt policy of divergence between Hindi and Urdu language, and the covert and subtle policy of a divide between Hindus and Muslims. This seemingly innocuous literary venture attributed Hindi to Hindus and Urdu to Muslims. What became signi®cant in this literary venture was the deliberate use of Devanagari script3, which incited the minds of the Hindi revivalists. Hindi was then considered as the most suitable language for the expression of their socio-cultural thought. Consequently, the development of Hindi as the medium of modern education became a major agenda for the Hindu revivalists. In addition to the process of linguistic division with respect to Hindi and Urdu initiated by the pedants of FWC, political contingency also
252
S. Imtiaz Hasnain and K.S. Rajyashree
contributed towards the revival of Hindi linguism. The political value of Hindi as a symbol of anti-colonialism combined with Hindi-Urdu di¨erentiation. Hindi was perceived by its supporters as the symbolic instrument for ®ghting colonialism and English, as Urdu was considered to be an extension of Mughal imperialism and was perceived by them basically as an instrument for preserving Muslim self-identity. Further, the decision made by the English administration to use Urdu as the court language reinforced the suspicion of Hindi supporters with regard to Urdu and even questioned its suitability for the anti-colonial struggle. Between Hindi and Urdu there was primarily an issue concerning selfidenti®cation, and issues pertaining to apparent linguistic similarities and di¨erences became rather secondary. Self-identi®cation basically involves issues like attitude towards the di¨erences, the functional roles assigned to them, and the uses made of them. Two distinct scripts and deliberate use of Sanskritised Hindi and Persianised Urdu accentuated the question of self-identi®cation and it subsequently gave rise to the Hindi movement. Agitation of the Hindus of North-Western provinces and Oudh between 1868 and 1900 over the replacement of Persian by Urdu ®nally culminated in the o¨ensive against Urdu. All this linguistic antagonism led to the reinforcement of the communal divergence between Hindus and Muslims. The linguistic and communal di¨erences between Hindi and Urdu were further exaggerated during the late nineteenth and early twentieth century. Both these languages became the means as well as the symbol of community creation. It was during this period that Nagari Pracharini Sabha was formed (1893), Hindi Sahitya Sammelan was founded (1910), Anjuman Taraqqi-e-Urdu was established (1903), Arya Samaj's Urdu campaign was initiated and the Anjuman-e-Islam of Lahore waged a counter-attack against the Hindi movement. The setting up of the Benaras Hindi University (BHU) in 1915 twinned with what was to become a few years later (1920) a Muslim counterpart in Aligarh called Aligarh Muslim University (AMU) as a modern institution with a religio-cultural agenda strengthened the self-image of the Hindi/Urdu literati and contributed immensely towards the success of their cultural agenda. The early twentieth century also marked the deliberate creation and crystallization of Hindi speakers' identity mediated through Hindi curriculum and texts. At this point, the contribution of Acharya Ramchandra Shukla, who started teaching Hindi at the Benaras Hindu University in 1919, is worth mentioning. To quote Kumar, ``A man of extraordinary talent and energy, Shukla shaped not only the format that the syllabi of Hindi in colleges continue to follow to a great extent to this day, he also de®ned the heritage of Hindi language and literature in a manner that few have dared to quarrel with'' (Kumar 1990: 180).
Hindustani as an Anxiety Between Hindi±Urdu Commitment
253
The deliberate construction of Hindi-Hindu identity has further been cogently highlighted by Kumar when he says: ``Shukla went well beyond the territory of the literary historian, and took a strong ideological position indicating the irrelevance of the Urdu-Persian tradition for the development of modern Hindi. He ignored major Urdu poets of the eighteenth and nineteenth centuries in his otherwise meticulous chronology . . . In an autobiographical essay, Shukla had written that his father had good knowledge of Persian and he used to enjoy mixing lines of Persian poetry with the lines written by Hindi poet. Ramchandra Shukla gave no signs, either in his history or in his other proli®c works of literary criticism, of having either taste or tolerance for this kind of mixture. By denying the literary works written in the mixed Hindi-Urdu tradition a valid place and status in Hindi's literary history, he performed a decisive symbolic act in shaping the cultural identity of college/ educated men and women for generations. The identity Shukla gave to the Hindi heritage was a distinct Hindu identity. His appreciation of a Muslim poet like Jaisi, and his acknowledgement of the achievement of Premchand, who symbolized the con¯uence of Hindi and Urdu at a time when the two had traveled far apart, made little di¨erence to this.'' (Kumar 1990: 180)
Thus, Shukla not only prepared the syllabi for Hindi language and literature that contributed towards the ``Hindi literati's cultural agenda'' but, as a creator of modern Hindi's self-consciousness, also strengthened the tendency to identify Urdu as the language of the Muslims and Hindi as the language of the Hindus. It is around the same period that these linguistic identities merged with national identity. We defer the issue of nationalism and language identity in the context of Hindi-Urdu and instead move on to look into the dynamics inherent in the machination of communal constructionists of both Hindi and Urdu. 4. Strategy of Communalist Construction As mentioned above Shukla believed that ``There is no necessity to use Arabic and Persian words in speaking Hindi . . .'' His contention, although it seems to be very simple and innocuous, is quite in tune with the processes of Sanskritization and Perso-Arabicization, for merely conveying an ostensibly unproblematic and true statement about language choice, is not just a re¯ection of a benign question of individual language choice. Rather, it is suggestive of a maneuver designed for the communal constructionists. However, it is noteworthy that the emergence of a speci®c communalist construct is not entirely attributable to the existence of the other. These diverse constructions are neither directly sequential nor purely reactionary in nature. Thus Perso-Arabic is in no way a counter o¨ensive o¨shoot of Sanskritization.
254
S. Imtiaz Hasnain and K.S. Rajyashree
There is an added dimension to these discursive practices which are transparent in nature and absorbed in truth-value statements, where the discourse of identity formation is located in a political space, not neutral but neither just negative and more so indicative of an ideology which is exclusionary forming a part of cultural practice, which is both constrained by and develops as a response to rules and conventions of speaking culture. As part of cultural practice it is both restrictive and productive. Thus in the Foucauldean sense both pristine Urdu and pristine Hindi possess regulatory and dominating aspects of power where each is trying to produce categories of speech community displaying certain linguistic dispositions which are believed to be both mimetic and emblematic of cultural history. It produces its own opposition, i.e. oppositional categories from where people work and think against it. It also creates ideologies of linguistic differentiation by recognizing or ``misrecognizing'' di¨erences among linguistic practices, locating, interpreting, rationalizing sociolingusitic complexity and identifying linguistic varieties with ``typical'' persons and activity. The ideologies of linguistic di¨erentiation work through iconicity, recursiveness and erasure. These semiotic processes not only establish links between linguistic forms and social phenomena, thus providing the means by which people construct ideological representations of di¨erences in linguistic practices, but also focus on linguistic di¨erences for de®ning `self' against some imagined `other' in the identity formation. By recursively projecting an opposition and meaningful distinction between imagined homogeneous speech community (achieved through erasure which simpli®es the ®eld of linguistic practices, features or varieties by imagining language as homogeneous entity devoid of internal linguistic variations) an iconic relationship is established between linguistic practices and social groups. This process of identity formation is well known in the social sciences. It assumes that ``there is some intrinsic and essential content to any identity which is de®ned by either a common origin or a common structure of experiences or both'' (Grossberg 1996: 89). Telecasting of Ramanand Sagar's serial on the Ramayana and perhaps several other mythologyÐbased tele-serials support the illustration of this essentialist model of identity formation. Telecasting of the Ramayana legitimized the stateÐsponsored and patronizing endorsement of a uniform, homogenized culture.4 Notwithstanding its popularity, the long term e¨ects of such a serial have perhaps been to ``project what the new culture should be'' and to ``expunge diversities and present a homogenized view . . .''. (Thapar 1989: 72) Its telecast not only encouraged a cultural hegemony by allowing the state, as a patron, ``to adopt a particular cultural stream as the mainstream'' but also posed a threat to creativity since the ``culture is being represented through the authoritative stateÐ controlled media''5. (Thapar 1989: 72)
Hindustani as an Anxiety Between Hindi±Urdu Commitment
255
Besides representing a social, religious or cultural group as a homogeneous entity, even a language may be imagined as homogeneous. Its internal linguistic variations may be ignored and there may be a conscious selection of those expressions, which only exaggerate and magnify already existing di¨erentiation. There is a totalizing vision in such linguistic ideology. Use of chaste Hindi in the Ramayana, Chanakya, etc. further vindicates our point. In fact the anxiety of the communal constructionists towards Hindi-Urdu commitment is not symmetrical. In Urdu, the fanaticism is of a vanishing tribe who simply uses Perso-Arabic expressions in defence. In Hindi, the fanaticism is of an emerging culture, yearning for the revival of the Golden Age, a nostalgia that induces a false sense of linguistic identity. Use of chaste Hindi is chosen here not as part of simple individual language choice, but it is re¯ective of a language design to create a simulation of identity. The Sanskritized expressions, deliberate and contrived, were consciously used to simulate language of the Ancient period in order to imbibe the identity of the people of the Golden Age. 5. Politics of Linguistic Engineering of Language Patterns Linguistic engineering is often undertaken at the behest of a select social elite either to appropriate political power or to maintain the desired status. In such cases a premium is always placed on the language of a social elite. The emergence of separatist tendencies in linguistic engineering not only created the `Muslim-Urdu' and `Hindu-Hindi' equation, but also set into motion forces of Sanskritization and Perso-Arabicization within the secular Hindu world and Muslim world. Let us substantiate it by citing a particular empirical study carried out by Agnihotri (1977). While examining the oral, written, normative and attitudinal behavior of native speakers of HindiÐ mostly University of Delhi lectures teaching Hindi and other subjects in humanities and social sciencesÐAgnihotri found striking di¨erences in all aspects of linguistic behavior and clear emergence of `pure and standardized Hindi' which has disowned a considerable part of its lexical heritage. Of course much of the discourse during the national movement revolved around the issue of selecting Hindi as ``the language in which the citizens of the future nation could speak to each other . . . [for] inter-regional communication.'' (Alok Rai 2000: 107). It had the support of Bankim Chandra and Gandhi. Although it was not clear what exactly this Hindi would be, it was unquestionably people's vernacular Hindi that had the appeal and support for becoming centerpiece of national struggle. But over the period of time ``the discursive space of the people's vernacular Hindi . . . was progressively usurped by Sanskrit `Hindi'.'' (Alok Rai 2000: 108) What remained common in this transformation is only the continuity in the name. The
256
S. Imtiaz Hasnain and K.S. Rajyashree
internal linguistic forms underwent radical change. ``Sanskritized Hindi'' came into existence. Alok Rai has rightly pointed out that this act of Sanskritizing the people's vernacular was carried out with twin objectives. At one level it was e¨ectively used as ``an important tactic in the regional power struggle'', and, at the other level it helped ``Sanskritized Hindi'' catapult to national level. It helped it go beyond its regional territorial a½liation and also enabled it to assume ecumenical acceptance as the rashtrabhaÅshaÅ Hindi characterized by its uniformity and absence of local hue and shade and capable of carrying out national struggle with romanticÐ ethnic nationalism. Thus, besides being perceived as the symbolic instrument for ®ghting colonialism and English, Hindi, according to Kumar, also ``represented a far more ambitious program, that of crowning the emergent vision of an independent India with a pan-Indian language'' (1990: 176). Hence the seeds of Hindi nationalism were inadvertently (or perhaps not so inadvertently) sown. There is yet another dimension to the machinations of language patterns arising out of linguistic engineering of the constructionists. At one level there is a projected claim of the constructionists that ``Hindi'' is the real representative of ancient Indian culture and the only substitute for a raÅshtrabhasha capable of producing nationalist struggle against English and English imperialism, notwithstanding the fact that behind the veneer of struggle against English imperialism there lies the Hindi imperial project.6 At another level these constructionists use English for sustained comprehensibility of contrived expressions. A strange mind set indeed which is willing to share with imperialist bedfellow but unwilling to have any liaison with shared and composite heritage. This mind set evokes a queer ambivalence, as both opposition and collusive operation can be discerned from the relationship of nominal hostility between the Hindi and the English elites. This is apparent from the neologisms created in imitation of English words. Both pre®xes like an-, up- and su½xes like- (õÅ) karan (in words like raÅshtrõÅyakaran `nationalization', aÅdhunikõÅkaran `modernizations', etc) are utilized as parallels for English `un'-, `sub'-, and `-isation', respectively. But what is important to mention here is that mere familiarity with Sanskrit lexicon may not necessarily guarantee any comprehensibility of neologisms. Comprehension is achieved only if the underlying English model is perceived. There are also instances where, in spite of the availability of existing terms in Sanskrit, consciously English words (or ideas based on English words) are used for coinages of neologisms. For example, Hindi has borrowed words describing eight-way distinction for direction as `loans' from Sanskrit, such as uttardakshin/puÅrv-pakchim and vaÅyuvya-nairuty/õÅsaÅnya-aÅgneÅya for `North-South/ East-West' and `SouthWest/NorthEast-SouthEast', respectively. These are not only `loans' from Sanskrit but they also have attestation in Sanskrit
Hindustani as an Anxiety Between Hindi±Urdu Commitment
257
usage. Still neologisms like uttarpuÅrvõÅ for `NorthEast', etc are created on the patterns of English. Sometimes too much of irrationality prevails over the coinages literally verging at the point of ridicule. For example, duÅr puÅrvõÅya deshon `Far East countries'. These countries may be `far' for the English speakers but not for us. Certainly the `hidden hand' of English is playing an important role in the development of modern Hindi style. As Snell has rightly pointed out ``though Sanskrit and English may seem odd bedfellows in the context of modern Hindi, their illicit relationship is proving extremely productive to an extent which would have been unthinkable for either one of the pair acting alone'' (1993: 79). What su¨ered most from this Trojan horse of English was the character of Hindustani. 6. Hindustani and the Politics of Nationalism Hindustani, the considered choice towards openness with no fanatical underpinnings of rigidity, became a victim of both the separatist tendencies in the linguistic engineering of language patterns and the idea of a single national language during the freedom movement. Rhetoric used during the freedom struggle for ``policing the boundaries of language'' (Sridhar Pathak cited in Rai 2000: 02) or even e¨orts made to merge language boundaries with the physical boundaries of the nation (Ganesh Shankar Vidyarthi cited in Rai 2000: 02) gave a new meaning to nationalism. It was publicized through the religio-nationalist and linguistic slogan of ``Hindi, Hindu, Hindustan'' given by the then Hindi stalwart, Pandit Pratap Narayan Mishra: cahuhu jusco nij kalyaÅn to sab mili BhaÅrat santaÅn ! japo nirantar ek jabaÅn hindõÅ, hindu, HindustaÅn ! (If your well-being you really want, O children of Bharat! Then chant forever but these words Hindi, Hindu, Hindustan). A three-fold assertion of the identity of language, religion and motherland, which did consolidate the communal mobilization through its mischievous genius of poetry, but failed to wipe out the trace of lexical heritage (zabaÅn in place of bhaÅshaÅ ). Con¯ation of language with nationalism and nation-states formation ultimately buttressed the claim of Hindustani. It even sharpened the divide between Hindi-Urdu on religious lines and heightened the nationalist rhetoric for the divide between the two languages. Alok Rai believed that the partition killed Hindustani. The immediate context of partition and
258
S. Imtiaz Hasnain and K.S. Rajyashree
Pakistan could not let Hindustani ``command any signi®cant constituency'' (Alok Rai 2000: 113). Certainly there were voices su½ciently vociferous to suggest that Hindustani became a victim of partition. One voice was that of R. V. Dhulekar who ``advised'' Maulana Hifzur Rahman to refrain from making a case for representing Hindustani. To quote Alok Rai; ``Today if you speak for Hindustani, it will not be heard. You will be misrepresented, you will be misunderstood. . .'' (2000: 113). Although speci®c regional politics and identi®cation of Hindustani with ``a status quo-ist defence of privilege'' have been considered to be the reason for such prevailing attitudes towards it, Hindustani unfortunately was not even spared of suspicion by its so called status quo-ist defenders of privilege. Advocates of Urdu and votaries of partition treated it as yet another name for Hindi. Thus in a letter written to Patel on behalf of Liaqat Ali Khan, early in 1947, Ghazanfar Ali Khan pointed out that ``Urdu and Urdu alone is the common language of both Muslims and Hindu in North India. Replacing it with Hindi, even under the pseudonym of Hindustani, is an attempt to suppress those parts of Indian culture, which are not exclusively Hindu but in the making of which Muslims too have had a hand'' (Lelyveld 1993a: 678). This hermeneutics of suspicion may be traced to the colonial knowledge which had its roots deeply entrenched in European intellectual practices and realized in state policy-colonial knowledge ``not by way of denying the rich and complex history of the ways in which Indians created their own language and literatures but rather to understand how language was construed by the power of a foreign regions into bounded institutions and communities with de®ned roles in an overarching political structure.''7 (Lelyveld 1993a: 668) Insofar as the British discovery of the languages of India was tied up with the acquisition and exercise of territorial domination,8 arguments about language became pivotal and were, therefore, central in producing and buttressing European claims to di¨erentiation from the rest of the world and their claim to superiority. The colonial knowledge which explored the possibility of identifying a single language, Hindustani, subsuming numerous linguistic varieties and literature, has had the tradition located in Herderean notion of nationalismÐthe romantic-ethnic nationalism of the nineteenth century. Eventually the de®nition of Hindi, Urdu and Hindustani was placed on the nationalist agenda. 7. Consequences In discussing the educational problems of Anita, a working-class schoolgirl in Delhi, Poonam Kaul (1989) attributes a large proportion of Anita's di½culties to language:
Hindustani as an Anxiety Between Hindi±Urdu Commitment
259
``All of the textbooks are couched in highly Sanskritised language very far from spoken or ordinary written Hindi. The arithmetic texts for primary classes (part four) has the following list of contents: Samay; guna aur gunankhand; sarvagunakhand aur sarvagunaj; . . . The science texts for class ®ve . . . [contains] one chapter entitled ayatan, bhaÅr aur ghanatva (`volume, weight and solidity'), which none of the women at Munushi, including graduates, postgraduates and habitual Hindi speakers, could translate without a dictionary''.
A press review has the following Persianized and Arabicized Urdu lexical items, which have been underlined: / is bare sa GõÅr me amrõÅ kõÅ fOjõÅ madaÅxlat kisõÅ Ase shaXs ke liye taÅ ?jubxez nahõÅ he jis ne amrõÅkõÅ jeograÅ®yaÅÅõ paÅlisõÅ kõÅ peshe raft ko nazar me rakhaÅ ho / Although these are supposed to have been written in a language of public discourse, they remain too inscrutable for public gaze. R. Ahmad (1997) examined the mutual comprehensibility of Hindi and Urdu news bulletins among `Urdu' (Muslim) and `Hindi' (Hindu) speakers, respectively. He found that the comprehensibility of the former was only 49% as against 77% of the latter. He concluded: ``It is clear that the process of Sanskritization of Hindi has resulted in a substantial loss of population that hitherto understood and participated in the life of Hindustani. It appears that Urdu has not closed its doors to the outside in¯uence. It has not only retained the native Hindustani words but has also borrowed freely from English. Not surprisingly, Hindi speakers ®nd the Urdu News bulletin well within their range of comprehension. Hindi on the contrary is increasingly closing its doors to outside in¯uence. Even words and constructions that are a part of the heritage of common man are being replaced by unfamiliar words. It is therefore not surprising that Urdu speaking informants, though living in a predominantly `Hindi World', ®nd the Hindi news bulletin beyond their comprehension''. (R. Ahmad, 1997: 47)
Besides incomprehensibility, loss of literary sensitivity is yet another drawback of this linguistic engineering. To quote Agnihotri: ``By the 1990s, the linguistic consequences of these primarily political moves were becoming increasing transparent. 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. It was only with some di½culty that one could explain Dhumil or Nagarjun to them; the staple diet was Bombay ®lm Hindi. The damage that inevitably accompanies the loss of literary sensitivity in a community is there for everyone to see''. (Agnihotri, 2002: 33)
260
S. Imtiaz Hasnain and K.S. Rajyashree
These are not isolated cases of incomprehension. There are innumerable examples (Hasnain 1991, 1995) suggesting incomprehensibility on account of linguistic engineering not because the innovations in language were di½cult but because these innovations were not in consonance with the general background of the common users of language. Consequences of anxiety are all around us in the form of creation of ``Standard Hindi/Urdu'' or ``Sarkari Hindi/Urdu'' where the creation is far removed from the created ones. Language has lost the potentiality or genius for social relations- the cadence, the vibrance or the entire poetry. It is this poetry, which becomes a justi®cation of the tongue of the tribe in the Mallarmean sense. The language has been robbed of its intimate possession ``some thing that one possesses in the same measure that one is possessed by it''. Besides suggesting the problems of comprehension, the innovations used here (and in all other cases as well) speak of power and o½cialdom and also, in a way, suggest the predicament of public in the language sphereÐbe it Hindi or Urdu, for it is Hindi or Urdu, not English, the language of the people, the vernacular of the masses. 8. Modern Cultural Sphere Hindi-Urdu-Hindustani forms a continuum of speech repertoire where the identities are crystallized. Both Hindu-Urdu form one end of the continuum where they survive by acquiring the language of aggression and identi®cation with the ideology of the perpetrators of communal construction. The writers of Hindustani, on the other hand, are situated on the other end of the continuum that merely wants to portray reality and not dictate survival. Their use of Hindustani is a re¯ection of `voluntary inner exile' consciously chosen to obtain the freedom to distinguish reality from the language of ideologyÐthe desire to vacate the inscrutable gaze of private sphere. This brings us close to the notion of publicÐprivate in the modern cultural sphereÐtaking a cue from Habermas' notion of private sphere vs public sphere. Habermas de®ned the public sphere as ``a realm of our social life in which something approaching public opinion can be formed. Access is guaranteed to all citizens. A portion of the public sphere comes into being in each conversation in which private individuals assemble to form a public body.'' (Habermas, qtd in Orsini 1999: 409) The public sphere is the space where arguments and reasons about the shared real world and hypothetical literary worlds are made and contested, given and taken in a manner which is democratic and civil. The private sphere, on the other hand, is the space where reasons need not be given, arguments need not be contested. Public sphere is the domain of civilityÐ a domain where there is a possibility of `conversation of mankind' (a la
Hindustani as an Anxiety Between Hindi±Urdu Commitment
261
Rorty) or `dialogism' (a la Bakhtin)Ðwhere a discourse can be simultaneously hermetic, as well as mimeticÐwhere each can confront the other on a shared ground of reasons. It is through the public sphere that individuals can exercise political control and create a discourse and a space for a social mind that enjoys HindiUrdu speech and literature. But in case of Hindi-Urdu it has been di½cult to locate any such shared grounds. Any discussion with regard to the matter of Hindi and Urdu is futile. Both Urdu and Hindi-wallahs feel ``paranoid and dispossessed''. A sort of a closure and a blindspot prevails with regard to the matter of Hindi and Urdu as the protagonists of both Urdu and Hindi have largely appropriated any space for public discourse. While the Hindi-wallahs ardently justify the case of Hindi as the case of National language and feel convinced that anyone who attempts to ``open the matter up to rational and historical consideration'' is probably an enemy. For an Urdu-wallah any attempt to bring the matter of Urdu to public scrutiny is tantamount to an attack on their identity and, consequently, disturbing the placidity of religious equilibrium. There is yet another dimension of public and private sphere particularly with regard to Urdu. Habermas traces the history of the division between public and private in language and philosophy and sees the public sphere as developing out of the private institution of the family he calls the ``literary public sphere''. Urdu, when situated in the private institution of the UrduÐspeaking family, marks an impossible ``culturally schizophrenic situation.'' (Ahmad 1989) This situation arises because Urdu has been relegated to the private sphere of family, and, therefore, on the one hand, there is clamor from the UrduÐspeaking family to send their children to learn Urdu for cultural longivity, while on the other, the market forces have sublimated the importance of Urdu resulting in a kind of linguicism where both the parents as well as the children feel that there is really no incentive to learn Urdu. The ideologues of Hindi and Urdu, thanks to the ingenuity of linguistic engineering, have their target audiences, well de®ned constituencies, and well maneuvered divisive strategies to divide the shared linguistic heritage of central India into Hindi and Urdu resulting from ``the intrigues of feudal and imperial powers to keep the common people divided and stay in power.'' (Agnihotri 2002) Lack of literary sensitivity has set in and ideologically motivated works of literature have not only failed to touch each other but have consciously distanced each other from popular modes and shared linguistic domains of interaction. Hence the anxiety of the loss of public sphere. In a situation when fraction becomes fractious, claims to a common language become inevitable. Alok Rai has pointed out ``the democratic legitimacy of the fractious faction's . . . derives from their claim to the
262
S. Imtiaz Hasnain and K.S. Rajyashree
common language'' (2000: 103). This common language could be Hindi, Urdu or Hindustani-any one ``without inverted commas'' (Alok Rai 2000: 15). Alok Rai comes out in his Hindi Nationalism with a programmatic agenda for resurrecting this lost language. However, despite his touch of sublime and sincerity of intention the programmatic project cannot be extirpated from the moment of past and the present. One is conscious of the fact that delving too much into history will not be congenial, for history opens up wounds that need to be sealed if people are to live together. However, if language is forced to descend abjectly from its being the instrument of the dominant elite at one point of time to being associated with poverty, illiteracy and backwardness at another and even further down to being identi®ed with terrorism, then what-ever be the extent of distancing from the languages within inverted commas, any attempt to recover the lost ground would still remain a part of a painful anxiety between Hindi-Urdu commitment. Though said in the context of the life of an Arab Palestinian in the West, perhaps it may not be too incongruous to quote Said (1978) when he writes: ``The life of an Arab Palestinian in the West, particularly in America, is disheartening. There exists here an almost unanimous consensus that politically he does not exist, and when it is allowed that he does, it is either as a nuisance or as an oriental. The web of racism, cultural stereotypes, political imperialism, dehumanizing ideology holding in the Arab or the Muslim is very strong indeed, and it is this web which every Palestinian has come to feel as his uniquely punishing destiny''. Notes * This is a revised version of the paper presented in ZICR (ZIst Century Reality: Language, Culture and Technology) organized by Mahatma Gandhi International Hindi University, New Delhi (October 29±31, 2002). We are grateful to Prof. Beg, Preeti and Shagufta. 1. There are some references dealing with the historical background of Hindi-Urdu antagonism which are not cited speci®cally in the text but were useful for the article, like Beg (1996), King (1994), Matin et al (2001), Orsini (2002), Sonntag (2001), etc. 2. `Created' within a single quote because Gilchrist is reported to have said that Lallu Ji Lal wrote Prem Sagar keeping in view the sole bene®t of the Hindu community of British India. (See Shiti Kanth Mishra 2013 Sam.: 3). 3. According to Rai (2000), the impulse to divide also came out of intransigence of Nagari/Hindi protagonists in the matter of Kaithi. To quote Rai, ``. . . this must be said unambiguously, the impulse to divide came from elsewhere . . . the key to the whole Nagari/Hindi movement was, ultimately, to establish and gain o½cial recognition . . . The necessary intransigence of the Nagari/Hindi protagonists becomes all too evident in the matter of the Kaithi script . . . a cursive variant of the Nagari scriptÐbut its very memory has been all but erased . . . Yet, till but a
Hindustani as an Anxiety Between Hindi±Urdu Commitment
4.
5. 6.
7.
8.
263
century back, this script was better known and much more widespread than Nagari . . . But Kaithi was unacceptable to the Nagari/Hindi propagandists . . . It was perceived to have some association with Hindustani rather than with Sanskrit. It was, moreover, known to Hindus and Muslims alike . . .'' (2000: 51±52) Telecasting of the Ramayana coincided with the emerging forces of Hindutva and Ramjanambhoomi agitation. One wonders if it is a coincidence. Kumar has rightly captured this in the following words: ``What gave this telecast a distinct political signi®cance was the parallel build-up of the movement to `reclaim' Ram's alleged birthplace (or rather spot) from a mosque at Ayodhya. This movement has acted as the biggest symbolic drama staged in India since Gandhi's salt satyagraha. The contrast between the narrow communal ends of the Ramjanambhoomi movement and the secular message of the salt satyagraha is a measure of the pressure liberalism in India has had to bear over the last 60 years''. (Kumar 1991: 45) Whatever may be, it certainly re¯ects a perfect unison of culture and politics in such a way that ``L.K. Advani's rath in his yatra of 1990 was a copy of the raths in B.R. Chopra's tele-serial Mahabharata.'' (Bajpai 1999: 53) For further illustration of this argument, particularly with reference to the Ramayana, please see Thapar (1989). This is evident from the fact that the form of Hindi that came into existence during the national movement not only denied ``the Urdu heritage its share, but also closed itself vis a vis the powerful spoken varieties of the region including Awadhi, Bundeli, Chhatisgarhi, Bhojpuri and the several tribal languages of central India.'' (Kumar, 1990: 187) Perhaps it is a truism to say that everything that happens in colonial societies can be traced to colonial intervention. Even the non-West indulges in productive discourses, which create new kinds of knowledge, political practice and subjectivity. The orientalist critiques of colonial knowledge have been making constant plea for systematic theorization of the relations of culture and power before colonialism. Although it may appear to be a di½cult task as it demands applying modern cognitive framework to pre modern world, a number of studies have come about showing how colonialist power altered notions of political space, subjectivities and collective identities in South Asia. Pollock's studies (1993, 1995a, 1995b) are, in fact, attempts in this direction. For a detailed discussion regarding Indian languages and the creation of a discursive formation as part of British discovery of India's languages, please see Cohn (1985).
References Agnihotri, Ramakant (2002). On a pre-partition partition: The question of Hindi-Urdu. Pangs of Partition: The Human Dimension. Vol. II. Ed. by S. Settar and I.B. Gupta. New Delhi: Indian Council of Historical Research and Manohar. pp29±46. Agnihotri, Ramakant (1977). Choice of Styles in Hindi. Papers in Linguistic Analysis. Vol. 2. 43±52. Ahmad, Aijaz (1989). Some re¯ections on Urdu. Seminar 359.23±29.
264
S. Imtiaz Hasnain and K.S. Rajyashree
Ahmad, Rizwan (1997). News Bulletins in Hindi and Urdu: A Study in Mutual Comprehensibility. M. Phil Dissertation, University of Delhi, Delhi. Bajpai, Shailaja (1999). Culture and television. Seminar 475.52±57. Beg, Mirza Khalil A. (1996). Sociolinguistic Perspective of Hindi and Urdu in India. New Delhi: Bahri Publications. Cohn, Bernard S. (1985). The command of language and the language of command. Subaltern Studies IV, Ed. by Ranajit Guha, Delhi: Oxford University Press, 276±329. Dua, Hans R. (1995). Sociolinguistic processes in the standardization of Hindi±Urdu. Standardization and Modernization: Dynamics of Languages Planning. Ed. by S.I. Hasnain, New Delhi: Bahri Publications, 177±96. Grossberg, Lawrence (1996). Identity and culture studies: Is that all there is?, Questions of Cultural Identity. Ed. by Stuart Hall and Paul du Gay, New Delhi: Sage, 87±107. Hasnain, S. Imtiaz (1991). Language standardization and linguistic norms: case of Urdu. Aligarh Papers in Linguistics, 1.1.17±37. Ð (1995). Innovations in language: An experiment in incomprehensibility with reference to Urdu in mass media and education. Standardization and Modernization: Dynamics of Languages Planning. Ed. by S.I. Hasnain, New Delhi: Bahri Publications, 213±26. Kaul, Poonam (1989). Anita, a working schoolgirl. Manushi 53.30±35. Keay, Frank E. (1920/1960). A History of Hindi Literature (The Heritage of India Series). Calcutta: YMCA Publishing House. King, Christopher R. (1994). One language Two Scripts: The Hindi Movement in Nineteenth Century North India. Bombay: Oxford University Press. Kumar, Krishna (1990). Hindi revivalism and education in North-Central India. Social Scientist 18.10.173±195. Ð (1991). Foul contract. Seminar 377.43±46. Lelyveld, David (1993a). Colonial knowledge and the fate of Hindustani. Journal of Comparative Studies in Society and History 35.4.665±82. Ð (1993b). The fate of Hindustani: Colonial knowledge and the project of a national language. Orientalism and the Postcolonial Predicament: Perspectives on South Asia. Ed. by Carol A. Breckenridge and Peter van der Veer, Philadelphia: University of Pennsylvania Press, 189±214. Matin, Abdul, Piyush K. Mathur and S. Imtiaz Hasnain (2001). Hindi-Urdu construct: Analyses of antagonism. Linguistic Structure and Language Dynamic in South Asia: Papers from the Proceedings of SALA XVIII Roundtable. Ed. by Anvita Abbi, R.S. Gupta and Ayesha Kidwai, Delhi: Motilal Banarsidas Publishers Pvt. Ltd. 197±206. Mishra, Shiti Kanth (2013). Sam Khadi Boli ka Andolan [The Movement of Khadi Boli]. Kashi: Nagari Pracharini Sabha. Orsini, Francesca (1999). What did they mean by `public'? Language, literature and politics of nationalism. Economic and Political Weekly Feb.13.409±16. Ð (2002). The Hindi Public Sphere 1920±1940: Language and Literature in the Age of Nationalism. Delhi: Oxford University Press. Pollock, Sheldon (1993). Deep orientalism? Notes on Sanskrit and power beyond the Raj. Orientalism and the Postcolonial Predicament: Perspectives on South Asia. Ed. by Carol A. Breckenridge and Peter van der Veer, Philadelphia: University of Pennsylvania Press, 76±133. Ð (1995a). Literary history, region, and nation in South Asia: Introductory note. Social Scientist 23.1±7. Ð (1995b). Literary history, Indian history, World history. Social Scientist 23.112±42. Rai, Amrit (1984). A House Divided: The Origin and Development of Hindi-Urdu. Delhi: Oxford University Press. Rai, Alok (2000). Hindi Nationalism. Hyderabad, Delhi: Orient Longman.
Hindustani as an Anxiety Between Hindi±Urdu Commitment
265
Said, Edward, W. (1978). Orientalism. New York: Vintage Books. Snell, Rupert (1993). The hidden hand: English lexis, syntax and idiom as determinants of modern Hindi usage. Institutions and Ideologies: A SOAS South Asia Reader. Ed. by David Arnold and Peter Robb, London: Curzon Press Ltd. 74±90. Sonntag, Selma (2001). The politics of linguistic sub-alternity in North India. Linguistic Structure and Language Dynamic in South Asia: Papers from the Proceedings of SALA XVIII Roundtable. Ed. by Anvita Abbi, R.S. Gupta and Ayesha Kidwai, Delhi: Motilal Banarsidas Publishers Pvt. Ltd. 207±22. Thapar, Romila (1989). The Ramayana syndrome. Seminar 353.71±75.
The Anxiety of Hindustani HARISH TRIVEDI
Towards the end of their article ``Hindustani as an anxiety between HindiUrdu commitment,'' as S. Imtiaz Hasnain and K.S. Rajyashree gird up to conclude, they produce a short, simple and heart-felt sentence such as seldom encountered in academic discourse: ``Any discussion with regard to the matter of Hindi and Urdu is futile'' (261). This self-re¯exive, soul-searching acknowledgement of despair and defeat may be read as both a point of entry and a point of no-entry into the vital discourse with which they engage. Yes, of course, it is in a sense futile to ¯og again this long-dead linguistic horse and perhaps worse than futile to want to reopen (in another poignant formulation by Hasnain and Rajyashree) this historical wound which under the half-formed crust still obviously festers. But then again, if Hasnain and Rajyashree have themselves chosen to do just that, it must be with the intention of seeking to extend the dialogue. Surely it is only by contributing to the existing stock of facts and opinions and by collating and hopefully correcting our respective slants and biases that we can hope to arrive at a fairer and truer understanding of the matter. Hasnain and Rajyashree's article is, in the main, a lament for the lost cause of Hindustani, the notionally ideal middle ground between Hindi and Urdu whose diminution or even disappearance has served to polarize Hindi and Urdu in post-colonial India (to say nothing, as they do, of Pakistan). The thrust of their argument is that this polarization was initiated by the British at the College of Fort William beginning in 1800 and has been pushed almost beyond the reach of any possible reconciliation by the increasing and intransigent Hindi/Hindu nationalism. So far as it goes, this may seem to be a readily acceptable and indeed widely accepted position. But the question to ask may be: does it go far enough in either covering the historical ground or in accepting the logic of its own argument? There are two broad and largely con¯icting narratives of the life and death of Hindustani which it may be as well to restate here in their bare bones. The ®rst of these runs to the e¨ect that (a) before the arrival of the British, Hindi and Urdu were one and the same language or at least
268
Harish Trivedi
mutually intelligible variants of one and the same language properly called Hindustani, (b) the British at the College of Fort William divided this language into two so as be able the more e¨ectively to rule India, and (c) this rift was later widened by the entrenched opposition of Hindi to Urdu and its growing Sanskritization. In the alternative account, (a) there was no common language of the Hindus and the Muslims before the British arrived unless it was Avadhi/Brij in which major Muslim poets such as Kutuban, Manjhan, Jayasi, Rasakhan, Rahim and Alam wrote between the 15 th and the 18 th centuries, (b) the British more or less invented Hindustani (rather than dividing it) for the good reason that they needed a functional common language in which to govern the country, and (c) because the language of command before they came had been Persian, the Hindustani they encouraged to evolve had a large proportion of well established Perso-Arabic terms so that Hindustani leant much more towards Urdu than Hindi. In this latter account, Urdu emerged as a language in North India (as distinct from the quite di¨erent Dakani in Central-South India) only in the eighteenth century, and shortly afterwards, by the last quarter of that century, it systematically purged itself of indigenous (desi) or Sanskrit-derived (tadbhava) words in common currency while importing Perso-Arabic vocabulary, and thus fashioned itself as both a constituent and an icon of a separate Muslim identity, indeed as ``the insignia of the Muslims,'' in Sir Syed Ahmad Khan's memorable phrase (qted in Amrit Rai 1982: 260). The ®rst of these versions, in which Hindi is rather more to blame, has been extensively documented by Christopher King in his One Language Two Scripts: The Hindi Movement in Nineteenth Century North India (1994), while the second version, in which it is Urdu which is argued to have veered away over a century before the British could play imperial mischief at the College of Fort William, has been most persuasively argued by Amrit Rai in his A House Divided: The Origin and Development of Hindi/ Hindavi (1984). King himself acknowledges that Rai's analysis is ``convincing'' but goes on to add that after Urdu had broken away, Hindi reacted by causing ``the other side of the divide''Ðas if a divide had not two sides to begin with (King 1994: 12±13, 175±77). The point of interest, therefore, in this new intervention by Hasnain and Rajyashree, would be to try and see, ®rstly, how they push forward this long-standing and richly documented debate by adducing any new facts or fresh interpretations, and secondly, just how they insert themselves in this debate in terms of their own slant and preferences. They begin by citing at some length a personal elegy on the loss of a shared language, Hindustani, by the linguist Ramakant Agnihotri (who is, incidentally, a colleague and friend of mine at the University of Delhi, as is another source cited by Hasnain and Rajyashree, Alok Rai; the three of us probably are in a tiny
The Anxiety of Hindustani
269
minority on our campus of persons who care about and have written on Hindi, Urdu and English equally). Agnihotri recalls how his parents asked him to disassociate himself from Muslim friends and to use ``pure Hindi'' (though he does not historicize the mindset in this aberrant respect of his parents who were ``very liberal otherwise'') and how he discovered as a young adult poetry by Hindu and Muslim writers which was written in a common language so that one could not decide whether it was Urdu or Hindi. To counter testimony with testimony for the sake of argument, I had no Muslim friends in the various small towns of Uttar Pradesh where I grew up or at least did not think of them as Muslim except one named Intezam Ali (who drew attention to his name by calling himself variously Arrangement Ali and Prabandha Ali as it took his parodic fancy), though I do recall distinctly some Muslim teachers who were distinguished by their elegance, courtesy and dignity, i.e. shaistagi. And in our ancestral village near Kanpur, where the Muslims were about a quarter of the total population, we all spoke undi¨erentiated Baiswari, a local variety of Awadhi. The ®rst Urdu I encountered was at the age of 14 when I bought out of my pocket money a copy of Divan-e-Ghalib just published in the devanagari script by the newly launched Hind Paperbacks (1961), and while I still treasure my tattered copy, I am not sure I have yet ®gured out the heavy Perso-Arabic complexity of the very ®rst couplet in the volume: Naqsh fariyaadii hai kiskii shokhi-e-tehriir kaa Kaaghazi hai pairahan har paikar-e-tasviir kaa I still ``love'' Urdu in a way one does not love one's mother tongue and sometimes wonder whether it is not at least partly because I am glamoured by its ``other''-ness. Agnihotri's juxtaposition of Nirala with Faiz for me only highlights this mutual otherness, for I do doubt whether Faiz himself would have understood very much of the opening couplet of what is probably Nirala's greatest poem: Ravi huaa ast; jyoti ke patra par likhaa amar Rah gayaa Raam-Raavan kaa aparaajeya samar. Agnihotri fondly recalls people a generation or two ago who could have understood both Faiz and Nirala, and perhaps there are some of us left who can still have a fair shot at it, but were the very di¨erent cultural and literary matrices of Faiz and Nirala ever part of a common heritage? Nostalgia, notoriously, is nothing if not highly selective and utopian. Hasnain and Rajyashree move on from the personal witness of Agnihotri to Alok Rai who, they say, has in his recent book Hindi Nationalism sought
270
Harish Trivedi
to ``save Hindi from the politics of nationalism.'' (249) In my own reading, Alok Rai's main concern in his ``tract'' is not so much with depoliticizing Hindi as with adjudicating its ``democratic'' claim for space in the colonial period against Urdu and now in the post-colonial period against English, both languages spoken by a small but privileged minority. It is in this evolving context that one needs to see what Hasnain and Rajyashree describe as a ``defence of the common language of the common man'' (249) that in their view both Agnihotri and Rai put forward. But was Hindustani ever a ``common language,'' as Hasnain and Rajyashree assume it to have been? Wasn't it always too close to urban and courtly Urdu, rather than to dusty and rural Hindi, to have had a mass circulation, for did not the huge majority of the population which lived in the villages, whether Hindu or Muslim, always speak the local dialect of Hindi, as depicted in telling detail, for example, in Rahi Masom Raza's novel, Aadhaa Gaaon? Even at the height of the last-ditch movement to project Urdu as a potential (but not actual) common language in the late 1920s and early 1930s through an institution such as the Hindustani Academy, did not its champions repeatedly claim that Hindustani was an ``aam-faham'' and ``mushtarkaa'' language, blissfully unaware of the contradiction that these very words (which mean ``commonly understood'' and ``mixed or shared,'' respectively) could not be understood except by those who spoke Urdu? No wonder that even an expert in the matter such as Christopher King, in the ``Glossary'' in his book One Language Two Scripts, de®nes ``Hindustani'' as meaning primarily and simply ``Urdu'' (King 1994: 198). Even the foremost champions of the cause of Hindustani, such as Mahatma Gandhi and Premchand, were to ®nd out that what the Urduspeaking Muslims understood by ``Hindustani'' was simply Urdu and not a mixture of Hindi and Urdu written in either the Persian or the Nagari script. When Gandhi sought to correct this bias at the ®rst (and last) annual conference of the Bharatiya Sahitya Parishad in Nagpur in 1936 by suggesting the adoption of the nomenclature ``Hindi or Hindustani'' for the projected common language, a champion of Urdu, Maulana Abdul Haque (later acclaimed as Baba-e-Urdu, i.e. the Father or Grand Old Man of Urdu), accused Gandhi of betrayal, reminding him of the days when Gandhi had ``written in his own hand a letter to Hakim Ajmal Khan in Hindustani, that is, in the Urdu language and in the Persian script'' (quoted in Amrit Rai 1982: 356; emphasis added). If this is what the Urdu-wallahs' notion of ``Hindustani'' was, Gandhi seems to have been fully justi®ed in calling the common language ``Hindi or Hindustani'' so as to pull it closer to the mid-way point between Hindi and Urdu. ``Hindustani'' thus seems to have been either a well meaning name for unadulterated Urdu or a utopian dream for a common language that seems
The Anxiety of Hindustani
271
never actually to have existed, at least not beyond a basic street vocabulary of say 500 words. Alok Rai seems to get it about right when he writes of Hindustani as a ``ghost'' of some dead entity, or more accurately of something never even born, of our ``recurrent yearning'' for Hindustani and indeed for a ``possibility of Hindustani'' (Rai 2003). As Faiz put it in another context: ``Yun na thaa, maine faqat chaahaa thaa yun ho jaaye'' i.e., ``It wasn't so, I merely wished it had been so.'' In lamenting the passing away of Hindustani, Hasnain and Rajyashree are thus mourning not the passing away of a phase of harmonious history but of a utopian dream which never came anywhere near ful®llment. No wonder their account begins in sentimental lament, goes on to heap unfair and excessive blame on the head of Hindi, and is throughout marked by signi®cant exclusions. For example, when they say that Wali switched from the mixed linguistic form of Dakani to ``a more aesthetically appealing and re®ned form [of language] devoid of any sectarian in¯uences'' (251), they are camou¯aging the fact that this more re®ned form happened to be distinctly more Perso-Arabic as well. When they quote Keay and italicise the part where the new Hindi forged at the College of Fort William is characterized as a language that ``could commend itself more to Hindus,'' they quietly play down Keay's explicit statement in the previous sentence that ``Urdu [already] had a vocabulary borrowed largely from the Persian and Arabic languages, which were specially connected with Muhammadanism'' (251). When they mention the date of the founding of the ``Benaras Hindi [sc. Hindu] University'' as 1915 and of its ``Muslim counterpart'' the Aligarh Muslim University as 1920 (252), they suppress the vital fact that the latter was merely an upgradation of the Muhammadan Anglo-Oriental College which had been founded in Aligarh by Sir Syed as far back as in 1877 (an elision hardly ascribable to ignorance as one of the co-authors of the article, Hasnain, himself teaches at the Aligarh Muslim University). Similarly, while Hasnain and Rajyashree quote Krishna Kumar on Ramchandra Shukla as having undervalued, in his de®nitive literary history of Hindi, works ``written in the mixed Hindi-Urdu tradition'' and having given ``the Hindi heritage . . . a distinct Hindu identity,'' they omit to cite any corresponding observation such as that by the eminent historian of Urdu literature, Muhammad Sadiq, that even Premchand, a Hindu who wrote exclusively in Urdu for the ®rst half of his career and is still claimed to be the greatest novelist in Urdu, continued to be treated ``more or less as an outsider'' by his Muslim readers in Urdu (qtd. in Trivedi 2003: 975). When they cite A. Ahmad as reporting that the proportion of Urdu-speakers who can follow Hindi news bulletins is substantially lower than that of Hindispeakers who can follow Urdu news bulletins (10), they fail to see that this may be so not only because Urdu is a more open language than Hindi but
272
Harish Trivedi
equally because Hindi-speakers are more widely receptive to Urdu vocabulary than Urdu-speakers to Hindi vocabulary. And when they animadvert on ``the machinations . . . of linguistic engineering'' (256), it is only Hindi that they have in mind and not Urdu as well. To cite just one example of how perverse some supporters of Urdu could be, after the devanagari script had been permitted to be used in judicial and administrative business alongside the already prevalent Urdu script in U. P. in the year 1900, Raza Ali, a member of the provincial council, brought an amendment in 1917 proposing that Hindi in the devanagari script may be thrown out and replaced by ``French, Russian, Italian or Persian'' (cited in Singh 1992: 170). Hasnain and Rajyashree cannot see that the story they tell is only half the story, for they do not seem to be interested in registering the matching record of Urdu hostility against Hindi. Two other major limitations their discussion seems to su¨er from are, ®rstly, that it is not the historically extinguished question of Hindustani but the burning issue of the rampant spread of ``Hinglish'' which both Urdu and Hindi now need to address with some urgency, and, secondly, that by becoming more Sanskritic, while Hindi may have drifted away from Urdu it has at the same time come distinctly closer to nearly all the other major Indian languages, including the Dravidian languages which too share a large Sanskritic vocabulary. Writing out of a nostalgic backward-looking mindset, Hasnain and Rajyashree cannot see that the ground beneath the feet of Hindustani, doubly treacherous as it always was, has now shifted clearly away as does a riverbed sometimes, and even when seeking to reclaim the old desideratum we may be slipping into newer realities. Thus, on the last page of his Hindi Nationalism, a work Hasnain and Rajyashree cite with approbation, Alok Rai arrives at the optimistic conclusion that Sanskritic or sarkari Hindi has not been able to triumph over ``the people's vernacular.'' The latter is characterized by Rai, in a clinching quotation from the foremost Marxist critic in Hindi, Ram Bilas Sharma, as follows: Ye sarakaari Hindi nahiin hai, ye parinishthhit Hindi nahiin hai, kintu raajnitigyon kii svaarth-niiti kii chintaa na karke jo Hindi raashtrabhaashaa bani hai vah yahii Hindi haiÐanek rupaa, aparinishthhit, jiivanta . . . This is not o½cial Hindi, this is not standardised HindiÐbut this is the Hindi that has become the national language in spite of the sel®sh intrigues of the politiciansÐthis Hindi, various, anarchic, alive . . . (Rai 2000: 122)
What is fascinating here is that the ``people's vernacular'' that both Rai and Sharma support is being acclaimed here in a Sanskritic vocabulary which, in an irony that seems to have escaped both Sharma and Rai, goes clearly against what it seeks to advocate, for the medium here ¯agrantly contradicts
The Anxiety of Hindustani
273
the message. In particular, that ``[a]parinishthhit'' takes the cake for this word is as high Sanskritic as Hindi can possibly get; the Sankshipt Hindi Shabda-saagar includes over 200 tatsam words beginning with the Sanskrit pre®x pari- but ``[a]parinishthhit'' is so rare and recondite a Sanskrit formation that it is not to be found among them. Rai himself translates it as ``not standardised'' and then, raising the stakes, as ``anarchic'' (for it occurs not once but twice in that short quotation) but the word itself is as highstandardised and conservative as possible. Time was, some seven or eight decades ago, when the projected common language of the people was called, in a comparable Perso-Arabic overkill, ``aam-faham'' and ``mushtarkaa;'' it is now apparently called ``aparinishthhit.'' We are still caught, despite our best intentions, between the devil and the deep sea, while the fabled common ground of Hindustani remains submerged and out of sight. References Hasnain, S. Imtiaz, and K. S. Rajyashree (2004). ``Hindustani as an Anxiety Between Hindi± Urdu Commitment.'' This yearbook: 247±265. King, Christopher R. (1994). One Language Two Scripts: The Hindi Movement in Nineteenth Century North India. Bombay: Oxford University Press. Rai, Alok (2000). Hindi Nationalism. New Delhi: Orient Longman. Ð (2003). The Persistence of Hindustani. India: A National Culture, ed. Geeti Sen. New Delhi: India International Centre and Sage Publications, New Delhi. Pp. 70±79. Rai, Amrit (1982). Premchand: A Life, tr. Harish Trivedi. New Delhi: People's Publishing House. Ð (1984). A House Divided: The Origin and Development of Hindi/Hindan. Delhi: Oxford University Press. Singh, Kripashankar (1992). Hindi-Urdu-Hindustani: Hindu-Muslim Sampradayikta [Sectarianism] aur Angrezi Raj [and British Rule] 1800±1947. Delhi: Prasangik Prakashan. Trivedi, Harish (2003). The Progress of Hindi Part 2: Hindi and the Nation. Literary Cultures in History: Reconstructions from South Asia, ed. Sheldon Pollock. Berkeley: University of California Press. Pp. 958±1021.
Announcements
277
The Gyandeep Prize We are happy to re-announce the continued availability of this annual prize. It will be awarded to the most outstanding student contribution to The Yearbook. YB-2005 We are happy to announce that the next issue of The Yearbook(YB-2005) will be a special issue on South Asian syntax. Guest-edited by Dr. Tanmoy Bhattacharya of the University of Delhi, it will contain contributions by Bayer, Dasgupta, Davison, Gair, Hook, Kidwai, Kiparsky, Lidz, Madhavan, Pfau and Zeeshan, Saleemi, Subbarao, and Yadava. Housekeeping We are happy to note that Professors Rama Kant Agnihotri, Josef Bayer and Peter Hook have now joined the editorial-board. From the next issues, all of them will participate fully in the editing of The Yearbook. As it is our intention to bring out future issues earlier than September/ October of each year, potential contributors to The Yearbook should get in touch with one of the editors as soon as possible. Our new deadlines are: November 1: initial submission March 1: ®nal 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, 2004, for example, will be considered only for the 2006 Yearbook. Potential contributors are encouraged to send their initial submissions as word document ®les to the Chief Editor. Those whose papers are accepted must, however, submit both hard and soft copies before the second deadline (March 1). We reserve the right not to process papers requiring un-necessary 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
E. Annamalai, Ph.D. (Chicago) is currently a visiting professor at Yale University. He was, until recently, the director of the Central Institute of Indian Languages, Mysore. In 1996, he was Visiting Professor, Indian Languages and Cultures, Tokyo University of Foreign Studies, Tokyo. He now lives in Mysore. He has published extensively on syntax, semantics, pragmatics, and language-planning. ([email protected]) Peter Bakker, Ph.D. (Amsterdam) teaches linguistics at Aarhus University, Aarhus, Denmark. He has written extensively on mixed languages, his main research area. ([email protected]) Tej K. Bhatia, Ph.D. (Illinois) is Professor of Linguistics at Syracuse University, Syracuse, New York. His areas of interest include language and cognition, media (advertising) discourse, the bilingual brain, Sociolinguistics, and the structure of English and South Asian languages. (tkbhatia@ mailbox.syr.edu) Vit Bubenik, Ph.D. (Brno) is professor of General and Historical Linguistics at Memorial University of Newfoundland. His current research interests include Indo-Aryan Linguistics, Ancient and Medieval Greek, Slavic Linguistics, and Morphology and Syntax. ([email protected]) Patricia Donegan, Ph.D. (Ohio State) teaches in the Department of Linguistics, The University of Hawaii. A co-architect of Natural Phonology, she has published extensively. ([email protected]) S. Imtiaz Hasnain, Ph.D. (Jawahar Lal Nehru University) is Professor of Sociolinguistics in the Department of Linguistics, Aligarh Muslim University, Aligarh. Before joining A.M.U., he was a Fellow at the Central Institute of Indian Languages, Mysore, where this paper came into existence. ([email protected]) B. A. Hussainmiya, Ph.D. (Peradeniya) is a historian specializing in the history of the Malays of Sri Lanka, Sri Lanka Malay literature, and Brunei
280
Notes on Contributors
Darussalam. He is the author of Lost Cousins: the Malays of Sri Lanka (1987) and Orang Regimen: the Malays of the Ceylon Ri¯e Regiment (1990). ( [email protected]) Thiru Kandiah, Ph.D. (London) is Professor Emeritus and Senior Consultant at the University of Peradeniya, Sri Lanka. Before his retirement, he taught the National University of Singapore. He has published extensively on syntax and on New Varieties of English. ([email protected]) Jouko Lindstedt, Ph.D. (Helsinki) is Professor of Slavonic and Baltic Languages and Literatures at the University of Helsinki. An Esperantist, his areas of interest include language-contact and morphology and syntax, particularly of Baltic and Slavonic languages. ( jouko.lindstedt@helsinki.®) Scott Paauw, M.A. (York) is a doctoral student in linguistics at York University. He specializes in Malay linguistics and historical linguistics. His interests include Austronesian linguistics, Malay and Indonesian, and contact varieties of Malay. His current work focuses on developing a typology of Malay varieties, and research on Malay varieties of East Indonesia as well as Sri Lanka Malay. ([email protected]) Soma Paul, Ph.D. (Hyderabad) is a Research Scientist at the Indian Institute of Information Technology, Hyderabad. Her main area of research is Semantics. ([email protected]) John Peterson, Ph.D. (Kiel) He is currently working as an assistant professor at the department of Linguistics at the University of OsnabruÈck. Together with Tej Bhatia (Syracuse) he co-manages the email-discussion group VyaÃkaran, which he founded, and also manages the `Bibliography for Seldom Studied and Endangered South Asian Languages' under http://www. southasianbibliography.de ( [email protected]) K. S. Rajyashree, Ph.D. (Pune) is ProfessorÐcumÐDeputy Director at the Central Institute of Indian Languages, Mysore. Her main area of research is Sociolinguistics. ([email protected]) Suchita Singh, M.A. (Calcutta) studied Sanskrit and Pali at Sanskrit College and Calcutta University. She has worked on a project on Language Movements at the M.S. University of Baroda and Centre for Social Studies, Surat (1980±83), and taught in di¨erent schools. She has also authored a 10-volume book, `Arise Awake' (Renommee, New Delhi, 17 th Reprint, 1996). ([email protected]) Udaya Narayan Singh, Ph.D. (Delhi) is Director of the Central Institute of Indian Languages, Mysore, India. A playwright-poet in Maithili (several volumes) and a poet-critic in Bengali (several books and papers), he has
Notes on Contributors
281
also published extensively in syntax, morpho-phonology, typology, translation, stylistics, sociolinguistics and language planning. ([email protected]. soft.net) Ian Smith, Ph.D. (Cornell) is Associate Professor of Linguistics at York University, Canada. He specializes in sociolinguistics and historical linguistics. His current work focuses on contact-induced language change. His areas of language specialization include South Asian languages, Australian languages, English, and Romance languages. He is the author of Sri Lanka Creole Portuguese Phonology (1978) and Higgins the Phonetics Tutor (1999). ([email protected]) David Stampe, Ph.D. (Chicago) teaches at the University of Hawaii. Known as the father of Natural Phonology, he has been interested in Munda since his student days. ([email protected]) Harish Trivedi, Ph.D. (Wales) is Professor of English at the University of Delhi and has been a visiting professor at the University of Chicago and the University of London. He is the author of Colonial Transactions: English Literature and India (1993, reprint 1995) and has co-edited Interrogating Postcolonialism (1996) and Postcolonial Translation (1999). (trivedih@ vsnl.com) Shravan Vasishth, Ph.D. (Ohio State) teaches Computational Linguistics at Saarland University. An Associate of the U.K. Institute of Translating and Interpreting, he has published articles on `honori®cs' and computational linguistics. His current research interests are: the syntax-semantics interface and computational linguistics. ([email protected]) Ghil'ad Zuckermann, Ph.D. (Oxford) is Gulbenkian Research Fellow at Churchill College, University of Cambridge (UK). He has taught in Singapore, Israel, the UK and the USA, and has held research posts in Italy, Japan, and Australia. His recent book, Language Contact and Lexical Enrichment in Israeli Hebrew, came out with Palgrave Macmillan in 2003. ([email protected])
Notes for Contributors
Submissions to the yearbook should be written in English. Contributors whose native language is not English should have their manuscripts read by a native speaker before submission. All contributions should be sent to the editor-in chief. Four copies of each manuscript should be submitted and should be double-spaced throughout (including notes and references) on one side of A4 or letter-size paper, leaving wide margins. All pages should be numbered serially. The authorás name should appear on a separate sheet only. Authors are requested to label at the ®rst occurrence and clearly identify any special characters used. Non-English words should appear in italics or be underscored if no italic typeface is available. Use only double quotation marks throughout, with the exception of translations, when single quotation marks should be used. Contributors are requested to submit the ®nal version on disk as well as a hard copy. If possible one of the common word-processors should be used. Examples not in English must have aligned interlinear glosses and an idiomatic translation. Figures and maps must be reproducible originals and should be submitted on separate sheets, carefully numbered and labeled. They should be referred to in the text and the approximate position should be indicated. Notes should be kept to an absolute minimum and be as brief as possible. They may contain no tree diagrams or tables. They should be numbered consecutively and indicated in the text by a raised (superscript) number following any punctuation marks. Citations in the text should give the name of the author/editor, the year of publication, and, in the case of quotations, the page reference, all in parentheses, for example: (Smart 1974: 22). Use ``et al'' in the case of more than two authors. Abbreviations such as ``ibid.'' and ``loc. cit.'' etc. should not be used. The reference section should contain all works cited in the text, and only those, and they must be listed in alphabetical order of author/editor, with the full ®rst name wherever possible, and with complete bibliographical details (including publisher); in cases of multiple authorship the names of all authors must be given. Journal and book titles must be given in full and must be italicised. Page references must be given for articles in books and journals.