> Wait, that's not true, is it? Chomsky's UG is not the English grammar
> we know and hate, but a more abstract one which will probably be
> satisfied by literal transcripts, etc. No?
UG is better thought of as the set of formal constraints (if any) on the grammars of actual languages. English has its own grammar (not the same thing schoolmarms teach, of course) which is, as Chomsky has said at least at one time, UG plus specific parameters (to oversimplify quite a lot).
And what kids hear is a mishmash. People don't finish their sentences. They forget what the subject was. They get lost in a subordinate clause and never climb out. How does the kid distinguish the badly-formed utterances from the well-formed ones?
It's a bit like the deconvolution algorithm for sharpening a blurry image. It works quite well *if* you have some information about the way the image was created in the first place. The UG postulate is that necessary information information about how the image was created. (This is my analogy not Chomsky's.)
> That's a fair summary of what I was saying. One more thing: just as in
> the Creator theory, but to a lesser degree, the postulated entity
> offers no explanatory benefit, and if some child behavioural
> psychologists are to be believed, it has negative predictive benefit.
Those of us who find the poverty-of-input argument persuasive believe that it explains something which otherwise needs explaining. Are there competing explanations these days? Since I haven't really kept up, there's a lot of recent work that I won't have read.
--
Michael Smith mjs at smithbowen.net http://stopmebeforeivoteagain.org