My pointing this out to him in Manchester in 2000, though acknowledged as "new", never provoked any substantial response either then or later in his writings.

      Boas and Sapir disagreed over strategies of investigation of genetic relationships . Boas was against demonstrating distant relations and emphasised intra-family work. Sapir, on the other hand, emphasised exactly the opposite. Thus in this early stage of the debate, one can easily see that Boas is for areal linguistics and against linguistics based on genetic, language-family related relationships. Given the development of the idea of an UG much later, one can reduce this view to say that Boas was against any notion of UG. The fact that many of us still run to the neighbouring languages to seek support for our generalisations can thus be found an early precedence. Diffusion, contact, is a "reality", like code alteration is, whereas that- t is not a reality because we can't see it. So the boundaries of which we are merely victims were drawn much earlier.

      Boas, in particular, argued for a separation between diffused and inherited languages . The following quotation from Boas is revealing from this perspective:

      Neighbouring dialects have shown a lot of similarity in morphology but not vocabulary...; beyond a certain point the geneological question has no meaning since it would lead to several sources and an arbitrary ancestral selection when languages are influenced by each other to a large extent. ... Languages influence each other so much that we can no longer speak of a single origin. In other words, the theory of Ursprache must not be believed until we have found that languages are not originated through acculturation.

(Boas 1940 [1920] : 217)

      This is as definitive a statement one can get against the notion of UG, the "truth", for Boas, must be visible. This, as we know, is the driving force behind the enterprise of Typology and Areal Linguistics in general: the truth must be visible . In other words, knowledge must be obtained directly.

      I object to this reductionism in the scientific enterprise and will show in section ? 1.4 that Typology is meaningless unless we also find a syntactic explanation for the facts listed by the typologist. In fact, at that level of operating, we begin to essentially see a unification of the two; if unification is a subversion of a sort, then, I believe, we should find ourselves there before it's too late.

      Sapir too, to begin with, allowed for areal influences (1916 [1949] ) but later as a reaction to Boas, argued against structural influences to be traced to areal diffusion. The following quotation clearly shows his preference, mild though his language may be:

"...if there is no really convincing example of profound morphological influence by diffusion, we shall do well not to put too much reliance on diffusion theories."

------------------------------------

See The languages of Native America : Historical and Comparative Assessment edited by Lyle Campbell and Mariane Mithun, University of Texas press, Austin/ London . 1979.

See Darell, R. and J Sherzer. 1971. Areal Linguistics in North America: A historical perspective. IJAL 37, 20-28 for a re-cap

Boas, Franz. 1940. Race, language and culture . New York : Macmillan.

Sapir, Edward. 1949. Selected Writings in Language, Culture, and Personality . Berkeley : University of California Press.

(Sapir 1921a: 206 )

      Thus by 1921 Boas and Sapir were proclaimed quarrelling over separability of areal and geneological similarity in languages. Boas believed their separability to be dubious making genetic research questionable. In contrast, Sapir believed that the effects of diffusion would not be profound, making proposals of remote relations brighter .

      This debate, unnoticed, has been re-enacted in our times through the work embodied, surprisingly, in one person. Richard Kayne, whose Linear Correspondence Axiom (LCA ) though is one of the most outrageous statements in generative grammar, it is still in line with the general logic of UG. This is so because the logic of UG is to look for broad, and in the opinion of this essays, sweeping, generalisations that languages are basically similar. At the same time though, his stance in general regarding the microparametric approach is most clearly identifiable with the Boas school in the Boas versus Sapir debate of 1920s mentioned above. Kayne (1996 ) in his introduction remarks that a micro-linguistic trend of comparing closely related languages is a meaningful one. That is, he takes the position (obvious but not apparent in the UG view of language) that comparing geographically closely related languages will lead to more interesting results than comparing it with genetically more distantly related languages. His cooperative work on Northern Italian and Sardinian dialects are efforts towards this direction. Is this then a case of oil and water finally coming together, if not in spirit at least in person?

 

•  India Not as Linguistic Area

      Notwithstanding Boas' take on areal linguistics and Kayne's faith in microparametric syntax, we go back to the main title of this paper: India Not as a Linguistic Area (INLA). In this section, I will point out merely two examples for illustrative purposes from Bangla and Hindi-Urdu and assume that it is obvious that otherwise similar differences can be seen within contiguous languages and in several other domains.

•  Complement Clauses in Bangla/ Hindi-Urdu

Consider for example a simple clause like the following:

( 2 ) John knows [that mother will come].

---------------------------------------------

Sapir, Edward. 1921. Language . New York: Harcourt, Brace and World.

See also Jakobson, R. 1944. Franz Boas' Approach to Language. IJAL 10.1 88-95

Trivialising for the non-specialist, LCA reduces to the statement that there is only one word order available in UG and it is the Head-Medial order . For definition and elaboration, see Kayne, Richard. 1994. The Antisymmetry of Syntax . Cambridge, MA: MIT Press. For an elaboration of LCA may apply to south Asian languages in the domain of NPs and Wh movement, see the following:

Bhattacharya, Tanmoy. 1999. The Structure of the Bangla DP . London: University College London dissertation.

Simpson, Andrew and Tanmoy Bhattacharya. 2003. Obligatory Wh- Movement in a Wh-in-situ Language. Linguistic Inquiry 34.1: 127-142.

Kayne, Richard. 1996. Microparametric syntax: some introductory remarks. Microparametric Syntax and Dialectic Variation , ed. by J. R. Black and V. Motopanyane, ix-xviii. Amsterdam/ Philadelphia: John Benjamins.

In Bengali and Hindi-Urdu, this translates as follows:

( 3 ) a. John jane [je ma aSbe]. Bangla

b. John jantaa hai [ki maa aayegii]. Hindi

      The English that and Bangla je and Hindi-Urdu ki are the typical clause connecting complentizers. On surface these three sentences look quite similar although the latter two languages have been classified as SOV or Verb-final languages.

      However, differences begin to appear when one looks at the relative mobility of the complement clause (in square brackets) in these languages. Bangla seems to show a greater degree of freedom in moving the complement around:

( 4 ) a. John [ma je aSbe] jane.

b. [ma je aSbe] John jane.

c. * John [maa ki aayegii] jantaa hai.

d. .* [maa ki aayegii] John jantaa hai.

      Notice, crucially though that this order in Bangla is only possible if the words within the complement are re-arranged, both incidentally are again ungrammatical in Hindi-Urdu:

( 5 ) a.* John [je ma aSbe] jane

b.* [je ma aSbe] John jane

     Up to this point what I have done is typology, I have simply listed the difference bwteeen the two languages. The next step of asking why this should be so and what properties of je and ki dictate this difference is a matter of syntax proper. The analysis of this phenomenon in Bhattacharya (2000a et seq ) relates it to the broader domain of focus accent in Bangla. The point that I wish to make is that without the syntactic explanation of this difference, the difference as displayed above is merely a listing strategy and therefore meaningless. The difference is not a parametric if parametric differences are considered to be reducible to matters of Morphology; rather, the difference is available due to the structural pressure that the language in question faces. I will come back to this in section 1.5 .

-----------------------------------------------

Bhattacharya, Tanmoy. 2000a. Comp-internal Clauses: Derivation by Phase. Klausurtagung , Gro b bothen.

Bhattacharya, Tanmoy. 2000b. Peripheral and Clause-internal Complementizers in Bangla: A Case for Remnant Movement. Western Conference in Linguistics (WECOL) 2000 , Fresno , CA (published as 2002c)

Bhattacharya, Tanmoy. 2001a. Bhattacharya, Tanmoy. 2001. The Puzzle of Bangla Comp-Internal Clauses. SNIPPETS 3 . Universite di Urbino, Italy

Bhattacharya, Tanmoy. 2001b. Breaking GROUND, Klausurtagung, Gro b bothen

Bhattacharya, Tanmoy. 2002a. Breaking ground: the syntax of two-Accent sentences in Bangla, Architecture of Language , CIEFL, Hyderabad

Bhattacharya, Tanmoy. 2002b. Focus Accent in Bangla in a Complex Model of Discourse, 24 th LSI Conference, Mysore

Bhattacharya, Tanmoy. 2002c. Peripheral and Clause-internal Complementizers in Bangla: A Case for Remnant Movement. Proceedings of Western Conference in Linguistics 2000, 100-112, Fresno , CA .

Bhattacharya, Tanmoy. 2002d. Focus accent in Bangla complex sentences, 31 st SALA Meeting, Univ of Iowa.

 

•  Wh Scope in Bangla/ Hindi-Urdu

      Yet another area where the two languages seem to substantially differ and where yet again the difference in principle is of an entirely different nature than parametric differences is in the domain of Wh scope.

      If a wh -phrase occurs in an embedded clause and is intended to have matrix clause scope, the CP has to occur in the pre-verbal position, as in (6 ) and the gloss in (6 i).

( 6 ) Ora [ CP ke aS-be] Sune-che. Sub [ CP ..wh..] V

they who come-FUT.3 hear-PAST.3

i. Who have they heard will come?

ii. They have heard who will come.

 

In ( 7 ) where the same CP occurs in a post-verbal position it is no longer possible for the wh -subject to take matrix scope and only the indirect reading in gloss (ii) is possible:

 

( 7 ) Ora Sune-che [ke aS-be ]. Sub V [ CP ..wh..]

they hear-PAST.3 who come-FUT.3

(i) #Who have they heard will come?

(ii) They have heard who will come.

 

This state of affairs simply does not exist in Hindi-Urdu:

( 8 ) a. us-ne sunaa hE ki [kaun aayegaa]

s/he- erg heard is comp who come. fut.3.m

(i) S/he heard that who will come.

(ii) #Who has s/he heard will come?

b. *us-ne [kaun aayegaa] sunaa hE

     That is, in Hindi-Urdu, there is no way of obtaining wide scope for the Wh unless one uses a complex Wh expression. In Simpson and Bhattacharya (2003), we wanted to give an account of restrictions on the scope of wh-phrases and the observation that a wh-phrase in the CP complement of a verb can only have matrix interrogative scope and a direct questioned interpretation if the CP containing it occurs in the pre-verbal position in Bangla, and not in post-verbal position. Our syntactic explanation related this to the word order being SVO in Bangla. Again, this cannot be a parametric difference (as there is no Morphology involved) but to do more with syntactic structures available in the language concerned.

 

•  Typology or Syntax?

      What I have done so far is to show that (i) there are crucial syntactic differences between closely related languages which may be reducible to parametric differences, and (ii) that mere listing of these differences is rarely useful. In this section, I will elaborate a case which makes us rethink the need for Typology at all.

--------------------------------------------------------------

Simpson, Andrew and Tanmoy Bhattacharya. 2003. Obligatory Wh- Movement in a Wh-in-situ Language. Linguistic Inquiry 34.1: 127-142

      In my past work on DPs , putative DP-universal such as the following were proposed which were found syntactic explanation:

( 9 ) DP-Internal Universal: H-Final languages show NP movement rather than N movement inside the DP

      This, at once, identifies something like LCA operative in a so-called head-final language by forcing leftward NP movement inside the DP, and, by extension, similar XP movement at the clausal level as well. As a corollary to this universal, non-head-final languages must show N movement as a natural consequence of the LCA.

      In a very similar vain and an important review of Antisymmetry Cinque (1996) in fact demonstrates that LCA derives something like (9 ) above. Cinque (1996: 452) demonstrates that Greenberg's (1966) Universal 20 (as in (10 ) below) is accommodated naturally within the Antisymmetry framework of Kayne (1994).

( 10 ) When any or all of the items (demonstrative, numeral and descriptive adjective) precede the noun they are always found in that order. If they follow, the order is either the same or its exact opposite.

      A clue to this asymmetry comes from Hawkins (1983) . Hawkins observes that in prepositional languages if the Dem follows the N then Adj follows the N too and if the Num follows the N then Adj follows the N too. These are interpreted as follows by Cinque:

( 11 ) a. N-Dem & N-A (Swahili, Fulani, Bahasa Indonesian, ...)

b. Dem-N & N-A (Maori, Baure, Douala , Tunen, ...)

c. Dem-N & A-N (Greek, Maya, Norwegian, ...)

d* N-Dem & A-N

( 12 ) a. N-Num & N-A (Swahili, Douala , Tunen, ...)

b. Num-N & N-A (Maori, Baure, Bahasa Indonesian, ...)

c. Num-N & A-N (Greek, Maya, Norwegian,...)

d.* N-Num & A-N (Cinque 1996: 453)

----------------------------------------

Cinque proceeds to show that this can be explained if we adopt the structure ( 13 ) as the base structure and derive different orders by raising N to higher functional heads.

Bhattacharya, Tanmoy. 1998. DP-Internal NP Movement. UCL Working Papers in Linguistics 10 225-252.

Bhattacharya, Tanmoy. 1999a. Specificity in the Bangla DP. Yearbook of South Asian Language and Linguistics 2 , ed. by R. Singh, 71-99. New Delhi / London : Sage Publications.

Bhattacharya, Tanmoy. 1999b. The structure of the Bangla DP . PhD. University of London

Bhattacharya, Tanmoy. 1999c. In search of the vague 'One'. Proceedings of ConSOLE 7 , ed. by Tina Cambier-Langeveld, Anikó Lipták, Michael reford and Erik Jan van der Torre, 33-48. Leiden : SOLE.

Bhattacharya, Tanmoy. 2000a. Gerundial Aspect and NP Movement. Yearbook of South Asian Language and Linguistics 3 , ed. by R. Singh, 123 - 146. New Delhi / London : Sage Publications.

Bhattacharya, Tanmoy. 2001b. NUM/ Q-CLA as a Complex Head. Henk van Riemsdijk and Norbert Corver, eds. Semi-Lexical Categories . 191-222, Berlin : Mouton de Gruyter.

Cinque, Gulielmo. 1996. The 'antisymmetric' programme: theoretical and typological implications. Journal of Linguistics 32 447-464.

Hawkins, J. 1983. Word Order Universals . New York: Academic Press.

 

      The head N either remains in-situ or moves to W as in 1 (in Maori) and derives the order in ( 11 b) and ( 12 a); Y as in 2 (in Douala ) giving the orders ( 11 b) and ( 12 a); X as in 3 (in Swahili) giving the orders ( 11 a) and ( 12 a).

      Hawkins' predictions for postpositional languages translate into the following patterns:

( 14 ) a. N-Dem & N-A (Selepet, Mojave, Digueño, ...)

b. Dem-N & N-A (Burmese, Kabardian, Warao, ...)

c. Dem-N & A-N (Burushaski, Hindi, Japanese, ...)

d.* N-Dem & A-N

( 15 ) a. N-Num & N-A (Selepet, Mojave, Kabardian, Warao, ...)

b. Num-N & N-A (Burmese, Hixkaryana, Ubykh, ...)

c. Num-N & A-N (Burushaski, Hindi, Japanese, ...)

d.* N-Num & A-N (Cinque 1996: 455)

 

      This can be explained in terms of the same base structure as in ( 13 ), and then either nothing moves to derive the order Dem>Num>Adj>N as in Hindi, or there are a number of leftward movements of the complements of the functional Z, W, Y of ( 13 ) to spec positions on intermediate XPs :

------------------------------------

If postpositional languages are considered to be left branching, with specs on the right and with rightward movement, as shown in (i), one would expect that DemN implied AN, thus ruling out the postpositional languages in (14 b):

(i) [NP N] Z ZP] Adj ZP] W WP] Num WP] Y YP] Dem YP] X XP] XP] ...

      Given that successive leftward XP movement is typically found in head-final languages (see proposals in Kayne (1994: 52ff) for deriving final complementisers), these sets of movements are licit. The fact that intermediate steps in (58) are also attested supports the derivation. For example the orders Dem-N-Adj-Num is found in postpositional languages like Kabardian and Warao and Dem-Num-N-Adj attested in Burmese, Kokama and Ubykh (see Hawkins (1983) for details). Although many questions remain, especially with regards to the status of the AgrPs in ( 16 ), it nonetheless establishes the consequence that in head-final languages XP movement (of which NP movement is a special case) instead of N movement inside the DP is the expected norm. The absence of NP movement in prepositional languages is attested by absence of orders such as: Dem Adj N Num or Num Adj N Dem which involve moving the spec XPs Dem, Num and Adj.

      There are a few exceptions as well. For example, unpredictable orders in Welsh, Berber, Hebrew, Basque among others, are discussed in Cinque (1996: 456, 20ff) and noted in the various typological studies like Hawkins (1983), Dryer (1988) , and may be derived in terms of a different set of movements (e.g. skipping step 2 would derive the Num>N>Adj>Dem order found in Basque) or a lower origin of Dem, but they remain contentious nonetheless.

 

Is this typology or syntax?

•  Pressure Points: A View from the Interfaces

      As stated in the introductory section, all is not lost. In fact, I suggest we discover the pressure points created in each language which derives the form and shape of specific constructions or of specific syntactic features, seen otherwise (by typological or parametric criteria) as the so-called inter-language differences. Furthermore our job as syntacticians is to locate these pressure points at the Interfaces.

•  What are Interfaces?

      Very briefly, Chomsky's view is that FL (Language Faculty) is embedded into already existing "external" systems: the sensorimotor system and some kind of system of thought (conception, intention etc.) both of which are somewhat independent of language. The language faculty has to interact with these systems, otherwise it won't be usable (like, if the liver produced something else, not bile, that is not usable by the rest of the organs of the body, then it wouldn't be useful). So the question we ask is: Is it well designed for interaction with those systems? Then we get a different set of answers and conditions. Given that language is essentially an information system, the only condition that clearly emerges is that the information it stores must be accessible to the other systems with which it interacts. So we ask whether language is well designed to meet the condition of accessibility to the systems in which it is embedded.

      In the rest of this section, I take up two specific cases (Classifiers in Bangla and Cleft questions in Meiteilon) and look at them from the point of view of Interfaces pressure points. In particular we ask the simple question why these constructions are there in the language.

•  Why classifier ?

Consider the fact that the following is ungrammatical in Bangla:

( 17 ) a. *it book

     This is so because you need an abstract representation and the substantive part (an "equal to" sign), i.e., there should be some way to express the thought that it represents/ stands for book but is not book itself. The copula in English does this job. However, in Bangla, the following is possible:

•  o-Ta boi

How/ why is (17 b) possible? Notice that (17 c) is not possible:

c. *O boi

      That is, the classifier -Ta has to be used with the representation of the thing (book). On the other hand (17 d) is inappropriate for this expression since it is merely an NP, or not a complete thought.

d. O boi-Ta

      The interface question that I wish to raise here is as follows: Why is it the case that a language chooses to use a nominal element (classifier) to complete a thought? The answer is not so simple. First, syntactically the copula in Bangla existential sentences can be dropped in the present tense. This is merely a syntactic observation, it doesn't approach the interface question at all. However, we can still argue on the basis of this syntactic observation that it is not the case that a nominal element completes a thought in this language, there's always an underlying verbal element. This underlying verb shows up in the past:

( 18 ) O-Ta boi chilo

      However, the observation can also be approached in a more minimalist manner. Is it possible that there is an intricate relationship between the absence of the copula and the presence of the classifier. In other words, is it the case that either (i) the classifier is needed to express the distinction between the thing and its representation because the be verb can be dropped in certain cases, or (ii) the presence of the classifier triggers that deletion of the copula because they are both performing the same function? Given the minimalist assumptions, this hypothesis seems quite likely. Moreover, the same function referred to above could easily be the job of maintaining the distinction between the two nouns, so that the hearer can distinguish the message that one is a representation of the other. The situation is somewhat like the following:

( 19 ) a. X = book

b. is (X, book ) [English, Hindi-Urdu; X = it / yah ]

c. X- def/spec book

      That is, English, Hindi-Urdu do it by putting something in between the two nouns (an "equal to" sign) and Bangla does it by making one of them more definite/ specific. This is a plausible explanation also because either is or hE does not play the role of marking time on the sentence but rather of what I have said so far, equivalence between the two nouns. Whereas in the case of the past tense, the time information is required and therefore Bangla must resort to the copula to indicate purely the time information. Viewed this way, in the past was or thii in English and Hindi express both time and the equivalence information.

 

•  Why cleft ?

      In connection with the Interface issue, the specific empirical case that is taken up in this subsection is with regards to cleft questions in Meiteilon. In particular, in the course of the following discussion, it will emerge that in Meiteilon a specific interface requirement of de-emphasizing governs the functioning of the cleft question strategy . The strategy of clefting, as a part of the FL in Meiteilon, is a response to this particular requirement at the interface. The work reported here is the striking result obtained in Bhattacharya and Thangjam (2003) with regards to cleft questions in Meiteilon exemplifies dramatically the importance of interface phenomena.

Consider the following example of a cleft Wh -questions in Meiteilon:

( 20 ) a. Tombi-nA kAna ukhi-ge?

Tombi-Nom who saw-Q

'Whom did Tombi see?'

b. tombi-nA ukhi-bA Adu kAna no?

Tombi-Nom saw-Inf/Nzr Det who Q

'Whom was it that Tombi saw?'

c. tombi-nA ukhi-bA mi Adu kAna no?

Tombi-Nom saw-Inf/Nzr person Det who Q

'Whom was it that Tombi saw?'

      Since questions have focus semantics and clefting is a form of focussing and plus the fact that wh- words in interrogatives normally attract phonological focus, it seemed rather strange that the element which attracts phonological focus (the wh-word) be again marked for focus syntactically by clefting. Such a reiteration of the same information (i.e. of emphasis) is unlikely and doesn't follow the kind of logic that languages seem to follow. In this light, it was conjectured whether it is really a doubling of information. This led to the intonation experiments conducted and the discovery that the language lacks phonological focus entirely.

      Without going into the details (to be found in Bhattacharya and Thangjam (2003)), the conclusion that the language lacks phonological focus marking, led to the logical hypothesis that the language therefore compensates for this lack by marking it morpho-syntactically via the cleft strategy. From the experimental results it was concluded that absence of focus contour (phonological focus) is the reason that the language uses clefting to assign focus syntactically to the question word in cleft Wh - questions.

      Questioning can be thought of as a part of the C-I interface that imposes certain restrictions on the FL as to how a question is to be formed and uttered. Of importance is the latter point about utterance. C-I, by these terms, seems to impose a requirement on the A-P interface as well, namely, to apply a special question intonation onto the expression to be uttered to ask a question. To elaborate further, question semantics seems to be coming from C-I and question intonation is a diktat of the A-P interface that is involved in crucial ways to the requirement at the C-I interface.

      We suggested that the requirement at the A-P interface, namely, De-emphasize is the raison d'être for the clefting strategy of question formation. That is, because the obligatory requirement of layering a proposition by question intonation in order to turn it into an interrogative is met with a sudden block in Meiteilon at the A-P interface in the form of a larger global requirement to de-emphasize every proposition, the narrow syntax of Meiteilon reacts by devising the cleft strategy to convey emphasis. This, we propose, is how the requirement that a proposition be marked for emphasis to turn it into an interrogative is met with in Meiteilon. In short, because Meiteilon imposes the A-P restriction that no proposition be marked phonologically for emphasis, it nevertheless maintains the C-I restriction to mark a question by emphasis syntactically by clefting.

•  The Final Word

      Finally, it is hoped that that the cases discussed above show why these very typical syntactic constructions in each language (Classifiers in Bangla and Cleft Questions in Meiteilon) are a result of some tectonic movement within the language concerned and nothing to do with parameters or observable (and therefore merely listable) differences between neighbouring or genealogically related languages. In particular, it is claimed that these movements within a language are a result of pressure through the Interfaces. In other words, the Interfaces put requirements on the FL which result in turn to observable phenomena in that language.

Dryer, Mathew S. 1988. Object-verb order and adjective-noun order: dispelling a myth. Lingua 74 185-217.

Bhattacharya, Tanmoy and Thangjam Hindustani Devi. 2003 [2004]. SALA XXIII Procedings. Doug Bigham, Mark Brown, Q Wan Kim and Sadia Rahman (eds.), Stanford: CSLI Publications.