[Lingtyp] Probabilistic typology vs. typology-based grammatical theory

[Matthew Dryer]

Jan makes a number of interesting points, but I think the picture is 
somewhat more complex.

First, although his characterization of some of my work as probabilitic 
typology is not inaccurate, I not only share his taste expressed by “So 
for me the statistically insignificant variants in languages across the 
globe are at least as interesting and relevant as the statistically 
significant ones”, but, as I often communicate to others, the “rara” are 
to me far more interesting than the common features.

Second, “grammatical theory” means different things to different people. 
I think what Jan means is not only a theoretical framework in which to 
analyse languages, but a theoretical framework that makes empirical 
claims about what we find in language. To me and many typologists, 
grammatical theory primarily means explanations for why “languages are 
the way they are”. The theoretical framework for describing languages 
doesn’t matter a whole lot as long as it provides tools for adequately 
describing languages and does not impose things on the language for 
which there is no language-specific motivation. In that sense, it isn’t 
really theory at all. But I would also include as theoretical, 
typological work that characterizes that range of ways in which 
different languages do things.

Thus, from the perspective of my metatheoretical assumptions, Jan’s 
statement “A theory that can account for both common and unusual 
phenomena is superior to a theory that can only handle common 
grammatical phenomena” does not make sense, since theories for me do not 
“account for individual phenomena”. The rara are relevant to typological 
work in that they are crucial for demonstrating the range of ways that 
languages do things, and in so far as that is theory, they are of 
theoretical importance. But they are not particularly relevant to the 
theoretical goal of explaining why languages are the way they are, which 
I think is primarily explaining why the dominant patterns are dominant.

My point is not to argue that Jan is wrong, but simply that we need to 
be clear how ones views depend on one’s underlying metatheoretical 
assumptions.

Matthew
On 1/25/16 11:52 AM, Jan Rijkhoff wrote:
> I think I should mention an important point, which is perhaps so obvious
> that no one thought of bringing it up so far: *some of us are doing
> probabilistic typology *(or are not committed to grammatical theory)*,
> others use typological data to test a grammatical theory (Rijkhoff
> 2002). Which group one belongs to makes a big difference regarding the
> way cross-linguistic data are identified and processed.*
>
> *Probabilistic Typology*
>
> For example, it is one of Matthew’s goals to find word order
> correlations (or rather ‘constituent order correlations’), so if the
> formal, morpho-syntactic aspects of the members of some semantic
> category are statistically insignificant, he can ignore morpho-syntactic
> variation in his data for the purpose of his research goal.
>
> *Typology and grammatical theory*
>
> It is my goal to contribute to a typology-based theory of grammar that
> can describe and explain both the frequent and the less frequent
> grammatical phenomena in all languages. So for me the statistically
> insignificant variants in languages across the globe are at least as
> interesting and relevant as the statistically significant ones. A theory
> that can account for both common and unusual phenomena is superior to a
> theory that can only handle common grammatical phenomena.
>
> I demonstrated in Rijkhoff (2010: 223) that “… grammatical theorizing
> and linguistic typologizing must go hand in hand and that rare
> typological features play a central role in the interaction of typology
> and theory”. This is, of course, also the reason we (Rijkhoff, Bakker,
> Hengeveld & Kahrel 1993; Rijkhoff & Bakker 2000) developed a sampling
> method that is designed to produce the maximum degree of linguistic
> variation in a sample (‘variety sample’). In a variety sample (as
> opposed to a probability sample) it is very important to have cases of
> the rarest type, since “exceptional types test the theory” (Perkins
> 1988: 367).
>
> So what is statistically not significant for probabilistic typology is
> highly relevant for a complete theory of grammar and since grammatical
> behaviour is due to semantic, formal and functional (interpersonal,
> communicative) factors, functional grammarians need to distinguish
> between semantic, formal and functional categories in typological
> research when they collect and interpret their cross-linguistic data.
> There no need for ‘conceptual categories’ here: since no two
> forms/constructions are completely identical in grammatical behaviour
> incl. usage/frequency (also noted in Edith’s contribution), we should
> instead agree on what counts a ‘sufficiently similar for a responsible
> comparison’ when we collect and process cross-linguistic data.
>
> Hengeveld, Kees and Lachlan Mackenzie. 2008. /Functional Discourse
> Grammar: A typologically-based theory of language structure/. Oxford:
> Oxford University Press.
>
> Perkins, Revere D. 1988. The covariation of culture and grammar. In M.T.
> Hammond, E.A. Moravcsik and J.R Wirth (eds), /Studies in Syntactic
> Typology/, 359-378. Amsterdam: Benjamins.
>
> Rijkhoff, Jan. 2002. On the interaction of Linguistic Typology and
> Functional Grammar. In /The Interaction of Data, Description, and Theory
> in Linguistics: Functional perspectives/. Special issue of /Functions of
> Language/ 9-2, William B. McGregor, (ed.), 209–237.
>
> Rijkhoff, Jan. 2010. Rara and grammatical theory. In J. Wohlgemuth and
> M. Cysouw (eds.), /Rethinking Universals: How rarities affect linguistic
> theory/(Empirical Approaches to Linguistic Typology 45), 223-240. Berlin
> and New York: Mouton de Gruyter.
>
> Rijkhoff, Jan and Dik Bakker. 1998. Language sampling. /Linguistic
> Typology/ 2–3, 263–314.
>
> Rijkhoff, Jan, Dik Bakker, Kees Hengeveld and Peter Kahrel. 1993. A
> method of language sampling. /Studies in Language/ 17-1, 169-203.
>
> ------------------------------------------------------------------------
>
>
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>