Language universals and linguistic knowledge

One major difference between the typological approach to language
universals and the approach taken within generative grammar and its
offspring concerns the status of language universals in terms of a
speaker's linguistic knowledge, that is, whether or not
language universals are part of grammatical representation in a speaker's
mind. The paper offers a number of arguments supporting
the view that typological universals as such
should not be regarded as
part of a speaker's linguistic knowledge.
The paper also argues that there is no evidence that a speaker's
linguistic knowledge consists of an inventory of universal
grammatical categories and relations that can be defined in formal terms.
Contrary to what is assumed by generative linguists and many
typologically
oriented linguists,
we only have distributional evidence for language-specific and construction-specific grammatical categories and relations, and
universals of grammar are rather found in a number of
principles of correspondence between language form and language
function manifested in these
categories and relations.