[Lingtyp] comparative concepts

[William Croft]

This discussion has ranged even farther and wider in the last 48 hours, expanding to fundamental questions in epistemology and philosophy of science. While I agree that it would be better if we were more knowledgeable about philosophy, I don't feel qualified to address those issues. I will return to the somewhat narrower topic of comparative concepts, language-specific categories, and the relationship between typology and language description.

It seems that there is some confusion or disagreement about what language-specific categories and comparative concepts are, when one gets down to specific examples, real or hypothetical. Before commenting on the discussions, I should make explicit what I take these two things to be:

--Comparative concepts are either functional ("purely semantic") concepts, or hybrid. The only hybrid comparative concepts that I see to be useful as this point are those that describe how function is expressed in morphosyntactic form, i.e. constructions and strategies as defined in my previous email.

--Language-specific categories are classes of words, morphemes or larger units that are defined distributionally, that is by their occurrence in roles in constructions of the language ("construction" defined broadly as in construction grammar: a morphosyntactic unit of any size paired with its meaning).

I will have more to say about those later, but let me turn to the specific examples first.

(1) The hypothetical example described by David Gil:

"let's consider three hypothetical (and somewhat simplistic cases of) languages that Matthew would classify as having SVO basic word order:
 
Language A:  has well-defined Ss and Os, and specific linearization rules that put the S before the V and the O after it.
 
Language B:  has well-defined Ss and Os, but no linearization rules that refer to them; instead it has specific linearization rules that put the A before the V and the P after it.
 
Language C:  does not have well-defined Ss and Os, but has specific linearization rules that put the A before the V and the P after it."

Languages A and B are described as having "well-defined Ss and Os". But if the languages have these categories, they are language-specific, and hence not S and O in the comparative sense that the languages are SVO. And so Language A's "well-defined Ss and Os" are not the same as Language B's "well-defined Ss and Os". That is, in this description S and O are being used to describe both language-specific categories of languages A and B (and hence should not even be compared to each other), and to describe comparative concepts, presumably something like A and P respectively.

(2) French and Italian "adjectives". Here it seems there is a similar ambiguity between using "adjective" to describe language-specific categories, let us call them French Adjective and Italian Adjective, and to describe a comparative concept 'adjective', which is not defined in the posts. Ekkehard's brief discussion of distributional differences (leaving aside Nigel's remark that it is a simplification of the facts) led to Edith responding that both the French and Italian forms could be adjectives in a comparative sense if they share at least two properties. But these two language-specific categories share none of the properties that Ekkehard provides: French Adjectives are defined by distribution in French constructions and Italian Adjectives in Italian constructions. A genuine comparative concept of 'adjective' has to be based on crosslinguistically valid criteria, e.g. 'property concept functioning as a modifier', or some other such criterion that you find to be revealing about the nature of this part of language.

(3) "Nounlike" and "verblike" "adjectives" and word order. One crosslinguistically valid comparative concept pertaining to form is word order; word order can be compared abstracting away from language-specific categories. I assume that Matthew means something like 'property concept modifier' for "adjective"; I am not sure if Jan means the same thing. Essentially, the debate between Matthew and Jan is an empirical one: do "nounlike" and "verblike" adjectives [property concept modifiers] behave alike across languages with respect to word order, or not? But in order to evaluate that question empirically, there has to be a crosslinguistically valid, consistently applied definition of "nounlike" and "verblike". I don't see that in either Matthew's or Jan's posts (though perhaps they are defined elsewhere); and I don't know if Matthew and Jan are using the same definition for their comparative concepts of "nounlike" and "verblike". (Stassen (1997) has pretty good definitions for "nouny" and "verby" predicated property concepts, but predication is not the same as modification.)


Returning to what comparative concepts and language-specific categories are, and hence the relevance (if any) of typology to documentary/descriptive linguistics. In my previous email I argued that language-specific categories are construction-specific. Hence they are overlapping in many complex ways and in some sense are epiphenomenal. Just as recognizing that categories are language-specific frees us to describe languages "as they are", recognizing that categories are construction-specific frees us to describe grammatical categories of languages "as they are" -- for each individual word, morpheme, or larger unit, what constructions it occurs in and what ones it doesn't; and why, to the extent that we can motivate their distribution in functional terms.

Comparative concepts aren't defined in the same way. In fact, the comparative concepts that typologists usually use pick out forms used for a fairly narrowly-defined function. I used narrow semantic classes in my parts of speech analysis (Croft 1991), and so did Stassen (1997) in his study of intransitive predication. In other words, in practice comparative concepts are a selection of points in multidimensional conceptual space - hopefully, ones whose formal expression across languages yield interesting patterns. Some typological work such as Östen's tense-aspect questionnaire survey (Dahl 1985), or the spatial (Bowerman-Pederson; Levinson et al. 2003) and cutting/breaking studies from Nijmegen (Majid et al. 2008) provide a denser distribution of comparative concepts in particular regions of conceptual space. Those studies reveal that linguistic categorization is even more variable than we believed, and that the conceptual space is fine-grained and probably continuous in many dimensions (Croft and Poole 2008; Croft 2011). Comparative concepts don't form large or even medium-sized classical (Aristotelian) categories any more than language-specific categories do (see Gross 1979), albeit in different ways.

Of course, typology would not exist without language documentation and description. But all language documentation and description is finite, whereas a language is an open-ended product of all linguistic interactions in a speech community (which itself is a complexly structured, constantly evolving entity). Typological research using comparative concepts can help prioritize what regions of conceptual space are most likely to yield a good sample of the diversity of constructions and construction-specific categories in a language, and what the salient formal properties of constructions are most likely to be. Of course, both typologists and documentary/descriptive linguists have to be open to seeing things that prior research did not discover. (I guess that makes me Heideggerian.)

Bill


Croft, William. 1991. Syntactic categories and grammatical relations: The cognitive organization of information. Chicago: University of Chicago Press.

Croft, William. 2008. Relativity, linguistic variation and language universals. CogniTextes 4.303 [http://cognitextes.revues.org/303/]

Croft , William and Keith T. Poole. 2008. Inferring universals from grammatical variation: multidimensional scaling for typological analysis. Theoretical Linguistics 34.1-37.

Dahl, Östen. 1985. Tense and Aspect Systems. Oxford: Basil Blackwell.

Gross, Maurice. 1979. On the failure of generative grammar. Language 55.859-885.

Levinson, Stephen C., Sérgio Meira, and the Language and Cognition Group. 2003. ‘Natural concepts’ in the spatial topological domain—adpositional meanings in crosslinguistic perspective: an exercise in semantic typology. Language 79.485-516.

Majid, Asifa, James S. Boster and Melissa Bowerman. 2008. The cross-linguistic categorization of everyday events: a study of cutting and breaking. Cognition 109.239-50.

Stassen, Leon. 1997. Intransitive predication. Oxford: Oxford University Press.