[Lingtyp] Structural congruence

[William Croft]

I haven't had the time to comment on the contributions for a couple of days, and many interesting issues have been raised in the meantime - though only lately has the discussion turned to what I consider the most important issue of all. I'll comment on a couple of other issues, leaving the most important to last.

1. The term "comparative concept" has been used to cover both "purely semantic" (I would say, functional) concepts and the hybrid formal/functional comparative concepts that Haspelmath (2010) argues for; or just the latter. Haspelmath (2010) has the more inclusive use. The use of purely functional concepts as the basis for crosslinguistic comparison appears to be much less controversial than the hybrid concepts as Martin conceives them (though not entirely so; see Croft 2001, ch. 3 for discussion).

Comparative concepts are invented by humans for human purposes, as Jan notes. As Martin notes, good comparative concepts are those that are useful for the purpose. I see the purpose of comparative concepts (in the broad sense, or even just referring to the hybrid formal/functional concepts) for typologists as aiding us in explaining why human languages are the way they are, as Matthew notes.

I think that typologists need at least two classes of (hybrid) comparative concepts, what I call "constructions" and "strategies" (Croft 2014, 2015), meaning more or less what typologists mean when they use these terms (assuming they are defined in a crosslinguistically valid way and applied consistently across languages). Constructions are functional equivalents, not unlike analogy in evolutionary biology (contrasting with homology, forms with a common ancestry -- apparently homology is used differently in microbiology). Strategies are constructions that share a relevant crosslinguistically valid formal property. I don't think there's an equivalent to that in evolutionary biology. But these appear to be useful types of comparative concepts for typologists.

The example from Martin's 2010 article that Östen cites can be thought of as a strategy for a construction. The construction is how languages encode the recipient participant of a predicated physical transfer event (one must describe the function without reference to formal concepts such as 'argument' and 'verb'). The strategy is to encode that function with a different form than is used to encode the theme participant. As a strategy, it won't look like analogy in evolutionary biology. I think that 'dative' is not a good label for a strategy, as opposed to a construction; but that's a terminological issue, not a theoretical one.

2. Dan commented that the use of terms in grammatical description is not always clear, in particular that grammatical and formal categories are confused. This is a major and serious problem. Related to it is that (hybrid) comparative concepts are confused with language-specific concepts, or are poorly defined -- this is the sense in which someone uses the term "adjective" or "ergative", or more typically argues that a language does not have "adjectives" or is not "ergative", in some crosslinguistic sense. As David Beck notes, this is usually a pointless or meaningless exercise. Some typologists going back to the 1970s (Gilbert Lazard, Bernard Comrie, Joan Bybee) proposed capitalizing terms for language-specific categories vs. crosslinguistic categories (whether purely functional or hybrid), but not many have followed this suggestion consistently. And then we should also clearly distinguish purely functional comparative concepts from the hybrid ones we find useful. People very commonly interpret typologists' talk of comparative concepts as language-specific grammatical categories or word classes.

3. The most important issue is the relationship between typology and language documentation/description. I agree with Peter's insistence that there ought to be a close relationship between the two. Typological research shouldn't end up being as irrelevant to language description as generative syntax has turned out to be. So while I agree that language-specific grammatical categories are just that, I don't believe that what typologists have discovered about languages is irrelevant to language description.

I think that assumptions about what a "grammar" is, which are independent of the recognition that grammatical categories are language-specific and the recognition of the value of comparative concepts (even the hybrid ones), lead to the seeming sharp divergence between typology and language description. Matthew takes the view that "grammar" and "usage" are completely separate (see also Newmeyer 2003), and appears to say that "grammar" forms a "system" (in the structuralist sense?) that has nothing to do with crosslinguistic categories that typologists use to formulate language universals about things like word order or the past perfective. 

I disagree with those assumptions about what I would call a grammar, that is, what a speaker knows about her language (hence the scare quotes in the preceding paragraph). The crosslinguistic diversity of grammatical behavior that led typologists to conclude that grammatical categories are language-specific is mirrored by the diversity of grammatical behavior of words and phrases across constructions that leads to the conclusion that categories are not just language-specific but construction-specific (Croft 2001 etc.). The same patterns that in one language are (or seem to be) categorical distinctions are found in other languages as token frequency differences; a well-documented example is Grev Corbett's data on his Agreement Hierarchy. These observations break down the barrier between "grammar" and "usage", and also mean that language-specific categories are not fundamental building blocks of grammatical description but one side of a rich and complex relationship between constructions and the units that fill the roles in the constructions.

I agree with Dryer's (1997) argument that (a)-(c) exist, but not (d):

a. categories and relations in particular languages
b. similarities among these language-particular categories and relations
c. functional, cognitive and semantic explanations for these similarities
d. categories and relations in a cross-linguistic sense

But I believe that (c) is part of what a speaker knows about her language, which is what I would call the speaker's grammar; that is, grammar is symbolic. But (c) is what typology is trying to achieve. 

Finally, grammars are not fixed, neither across speakers, across time, or even across the lifetimes of individual speakers. The sorts of factors that cause grammars to change, and underlie grammars, are the sort of factors that underlie typological universals, which are themselves the product of language change across languages.

Bill


Corbett, Greville G. 1983.  Hierarchies, targets and controllers: Agreement patterns in Slavic. University Park, Penn.: Pennsylvania State University Press.

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

Croft, William. 2014. Comparing categories and constructions crosslinguistically (again): the diversity of ditransitives [Review article on Studies in ditransitive constructions: a comparative handbook, ed. Andrej Malchukov, Martin Haspelmath and Bernard Comrie]. Linguistic Typology 18.533-51.

Croft, William. 2015. Grammatical categories, semantic classes and information packaging. Morphosyntax: constructions of the world's languages (Cambridge University Press, to appear), chapter 1. Available at http://www.unm.edu/~wcroft/WACpubs.html.

Dryer, Matthew S. 1997. Are grammatical relations universal? Essays on language function and language type, ed. Joan Bybee, John Haiman and Sandra A. Thompson, 115-143. Amsterdam: John Benjamins.

Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in cross-linguistic studies. Language 86.663-87.

Newmeyer, Frederick J. 2003. Grammar is grammar and usage is usage. Language 79.682-707.