We provide a critical review of the distinction between “comparative concepts” and “descriptive categories”, showing that in current typological practice the former are usually dependent on the latter and are often vague, being organized around prototypes rather than having sharp boundaries. We also propose a classification of comparative concepts, arguing that their definitions can be based on similarities between languages or on differences between languages or can also be “blind” to language-particular facts. We conclude that, first, comparative concepts and descriptive categories are ontologically not as distinct as some typologists would like to have it, and, second, that attempts at a “non-aprioristic” approach to linguistic description and language typology are more of an illusion than reality or even a desideratum.
In polysynthetic West Caucasian languages, the morphological verbal complex amounts to a clause with all kinds of participants cross-referenced by affixes. Relativization is performed by introducing a relative affix in the cross-reference slot that corresponds to the relativized participant. However, these languages display several crosslinguistically rare features of relativization. Firstly, while under the view of the verbal complex as a clause this affix appears to be a relative pronoun, it is an unusual relative pronoun because it remains in situ. Secondly, relative affixes may appear several times in the same clause. Thirdly, relative pronouns are not expected to occur in languages with prenominal relative clauses. Fourthly, in the Circassian branch, relative pronouns are identical to reflexive pronouns. These features are explained by considering relative prefixes to be resumptive pronouns. This interpretation finds a parallel in the neighboring East Caucasian languages, where reflexive pronouns also show resumptive usages. Finally, since in some West Caucasian languages the relative affix is a morpheme with a dedicated relative function but still shows properties of a resumptive pronoun, our data suggest that the distinction between relative pronouns and resumptive pronouns may not be as clear as is usually assumed.