[Lingtyp] comparative concepts

[William Croft]

David's response makes an assumption that I, and possibly Martin, would not agree with: that comparative concepts are distributionally-defined categories just like language-specific categories are. David proposes a hypothetical world in which two languages have identical distributions of equivalent units across equivalent constructions, and suggests that these would be comparative concepts in Martin's and my, or at least my, sense. David already raises a major problem: the world isn't like that, so why should we waste time theorizing about it? But there is another problem: how do we decide that two languages have equivalent constructions and equivalent units whose distributions match? The only way we can do that is with comparative concepts of the types that Martin described in his 2010 paper and I described in earlier emails (and references cited therein).

I did look at your 2000 paper, David. The paper constructs a categorial grammar model to define crosslinguistically valid language-specific categories. Just one symbol, S, is assumed as a starting point. But to define that crosslinguistically, you appeal to Boas' semantic definition of 'a complete idea'. That's a comparative concept in Martin's and my sense, hence not purely distributional (and also too vague to apply consistently). S is supplemented with categorial grammar slash, which describes distribution (cf. Croft 2001:48-49), and with levels not unlike X' theory, e.g. S^0, S^1, etc. These levels are not given a definition as far as I can tell, and so I did not see a crosslinguistically valid definition of them. Finally, you appeal to the notion of a syntactic head and list a few tests for headhood. But these distributional "tests" are mismatched both within and across languages (Croft 2001, ch. 7), and anyway appeal to concepts like agreement and government which require crosslinguistically valid definitions -- i.e. must be comparative concepts in Martin's and my sense, if they can be constructed at all (I offer a comparative concept of head in that chapter).

It may seem like I'm picking on David. In fact it shows that it is very hard to construct crosslinguistically valid comparative concepts; it is much easier to criticize others' attempts. Even some of the examples in Martin's 2010 paper are problematic in my view (not just terminologically). I try hard to define all the comparative concepts in my morphosyntax textbook consistently, but I wouldn't be surprised to be taken to task about some of them when it is published. Language-specific assumptions are buried very deep in grammatical description.

Finally, David discusses language-internal variation. But his solution is to slice up a language into homogeneous grammars in order to preserve some kind of categorial consistency. This is the same analysis used by generative grammarians for diglossia. Taken to its logicla conclusion, this leads to a model of each speaker having zillions of grammars in their head in order to capture all the variation across all situations of use (Croft 2000:51-53). Instead we should embrace a single variable grammar for a speaker to model her knowledge about language-internal variation (ibid.). We must accept it for distributional variation, as I argued earlier. It is only natural to accommodate socially-determined variation into this model. This of course requires abandoning the notion of classical (Aristotelian) language-specific categories, which I suspect is another assumption underlying some of the disagreements in this debate.

Bill


Croft, William. 2000. Explaining language change: an evolutionary approach. Harlow, Essex: Longman.

Croft, William. 2001. Radical Construction Grammar: syntactic theory in typological perspective. Oxford: Oxford University Press.


On Jan 24, 2016, at 5:25 AM, David Gil <gil at shh.mpg.de> wrote:

> On 24/01/2016 13:46, William Croft wrote:
>> 1. David Gil argues that syntactic categories "based exclusively on distributional criteria, and...blind to the semantics" can be defined as comparative concepts. But distributional criteria are criteria defined by a construction or constructions in a specific language, and so by definition are not comparative: the constructions are constructions of particular languages. This is what makes grammatical categories language-specific, and in fact construction-specific.
> Under Martin's definition of comparative concept, Bill is right in saying that "distributional criteria are criteria defined by a construction or constructions in a specific language, and so by definition are not comparative: the constructions are constructions of particular languages."  But I would like to suggest that the kinds of categories that we typologists deal with (and which I somewhat loosely referred to as comparative concepts) in fact do not provide a perfect fit to Martin's definition of comparative concept.
> 
> Martin raised the hypothetical case of two languages with identical grammars differing only with respect to their lexicons.  Now let's imagine a hypothetical world in which these were the only two languages in existence.  In such a hypothetical world, language-specific categories would be identical to cross-linguistic categories.  Now obviously our real world is very different from such a hypothetical world, but how different exactly?  At the opposite extreme one can imagine another world with several thousand languages each of which is endowed with its own individual constructions incommensurate with those of each and every other language.  Many linguists would say that this is the actual world that we inhabit.  I would like to suggest (a) that the positioning of our real world of languages in relation to these two extremes should be considered an empirical question, not a matter of belief or philosophical persuasion; and (b) that the factually correct answer is that our real world of languages actually lies in-between these two extremes, albeit probably closer to the several-thousand-incommensurate-languages extreme than is commonly assumed not just by generativists but also by many other linguists belonging to various other camps.
> 
> In other words, I argue that there do exist categories that are, simultaneously, manifest in the grammars of individual languages and also relevant to cross-linguistic comparisons.  My (2000) paper referred to in my previous posting was an attempt to define some syntactic categories which fit this bill.  Since these categories yield useful typological insights, I somewhat sloppily referred to them in the preceding discussion as comparative concepts, which is what Bill correctly took issue with above.  But my response is to suggest that the current definition of comparative concept may not be the most useful one.  Of course, the specifics of that proposal could easily be wrong, but it's the principle that is at issue here.
> 
> But for a simpler example of a category / construction type that is both language-specific and universal, how about "utterance", defined roughly as what occurs in-between two periods of silence.
> 
> Following are two related reasons why we should expect to find language-specific categories which are shared by more than one language, and hence which might be useful also for cross-linguistic categories.
> 
> The first (as pointed out by Eitan earlier) is the phenomenon of bilingualism.  Bilinguals are clearly capable of providing (necessarily imperfect) translations from one language to another; it is hard to see how such an ability can be accounted for without positing certain correspondences between languages — not just at a "deep" level of conceptual representations, but also with respect to more readily observable formal properties.
> 
> The second reason is the phenomenon of language-internal variation, be it geographical, social, idiolectal, or whatever.  We linguists are good at telling our laymen friends that there's no difference between languages and dialects, but we sometimes fail to internalize this message ourselves.  In fact, many if not all of us are competent in a multiplicity of registers or varieties of our native languages — how many such varieties exactly is a moot question since it depends on arbitrary decisions concerning the level of resolution that we wish to adopt.  So if I speak a Home-English and an Office-English, do these have two separate grammars which just happen to be virtually identical, or do they share a single common grammar with (Labovian) specifications for those areas which exhibit variation?  If the former, then it would be absurd not to admit that the two separate grammars share lots of categories.  If the latter, then well, if there's no difference between dialects and languages, then just as my Home-English and my Office-English will share categories and constructions, so, as a bilingual, will my English and my Hebrew (albeit presumably fewer).
> 
> To summarize:  While I agree with many of my typologist colleagues that languages are massively more different from each other than is commonly supposed by many, perhaps most, other linguists, I do not believe — as I suspect some do — that languages may differ without bounds.  Moreover, while I agree that many of the most influential proposals that have been made in the past (e.g. by generativists) for categories that are, simultaneously, language-specific and universally valid are fundamentally misguided, I do not believe that we should throw the baby out with the bathwater and reject the possibility that such language-specific-cum-universal categories may nevertheless exist.
> 
> -- 
> David Gil
> 
> Department of Linguistic and Cultural Evolution
> Max Planck Institute for the Science of Human History
> Kahlaische Strasse 10, 07745 Jena, Germany
> 
> Email:gil at shh.mpg.de
> Mobile Phone (Indonesia): +62-812-73567992
> 
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp