typology and categorial particularism

[Martin Haspelmath]

Thanks, Wolfgang, for your comments. I did not understand all of them, 
but here are some reactions:

Wolfgang Schulze wrote:
>
> For instance, it is not clear to me why Martin assumes that e.g. 
> 'position' and 'special' are 'general concepts'
>
'Position' is a general concept because it also occurs outside of 
language structure, so it could not be specific to individual languages. 
This makes it easy to compare languages with respect to position (this 
is one reason why constituent order typology has ben so successful -- 
order is an eminently comparable property of language structure).
>
> and that 'syntactic construction' is a universal formal concept. What 
> is the model that motivates this assumption?
>
I have no "model", but I assume that this much is uncontroversial: All 
languages have constructions, and 'construction' is a property of 
language form (not a conceptual-semantic property).
>
> The same holds for ... the formulation: "A/n ergative case is a 
> morphological category that has among its functions the coding of the 
> agent of typical transitive clauses, when this is coded differently 
> from the patient/" (by itself a problematic definition). What is the 
> theory that explains to us the notion of 'agent' and of is a 'typical 
> (sic!) transitive clause'?
>
Again, I don't need a "theory" (whatever that my mean), but of course I 
need a definition of 'agent' and of 'typical transitive clause' (e.g. in 
terms of the typical transitive verb 'kill', cf. note 10). I think a 
definition along these lines is used by all typlogists. And crucially, 
such a definition in terms of universal notions is quite different from 
a language-specific descriptive category.
>
> In other words: I am left with the impression that Martin's 
> comparative concepts do much the same t.c.'s do that are explicitly 
> derived from a specific model of language, although they 
> (occasionally) seem to be more impressionistic than the latter ones.
>
I don't know what exactly is meant by a "specific model of language", 
but I'm quite happy to equate my comparative concepts with your tertia 
comparationis. It's probably just a different terminology. We seem to 
agree that tertia comparationis are different from language-particular 
descriptive categories, and that tertia comparationis are created by 
linguists (whether on the basis of a "theory" or a "model" or just on 
the basis of what seems reasonable to them).

Martin