The claims that there are few universals of language are based on a narrow and unwarranted definition of the notion itself (arbitrarily excluding universals that could be eventually explained away by functional considerations, perhaps having operated long ago during cognitive evolution) and of the epistemology of linguistics (which should then be surprisingly different from the models adopted in other successful sciences). Once these and other terminological misunderstandings are eliminated, certain proposed universals, e.g. constituency, appear to indeed hold crosslinguistically. Historical explanations of language similarities and diversity should certainly be an increasingly central concern of cognitive science, and can be successfully pursued precisely within a framework assuming the existence of some universals and of variation constrained in a principled way.
