Chomsky -- Put up or blah blah

bill fancher fancher at pacbell.net
Sat Apr 1 09:33:35 PST 2000


(Sorry, this is not more timely. ISP problems...)

on 3/28/00 1:07 PM, Scott Martens at smartens at moncourrier.com wrote:


> ...The "poverty of stimulus" agrument is one of the good things I am willing
> to give Chomsky credit for.
>
I always understood that argument to be: look how fast kids learn language, therefore behaviorism is false and language is innate. Did I miss some subtlety?


> ...There aren't many people left who don't agree, in princile, with the
> inateness hypothesis of language.
>
See "Rethinking Innateness" by Elfman, Bates, Johnson, Karmiloff-Smith, Parisi, and Plunkett to hear from six who don't. (They don't find "poverty of stimulus" very convincing either.)


> For the most part, the damage has been restricted to linguistics. Education
> theory never really relied on generative grammar anyway (what it does rely on
> often isn't any better though.)
>
Generative grammar, to my knowlege, has had no practical application, but education COULD gain a great deal from behaviorism. This was empirically demonstrated by the Department of Education Project Follow Through study, which showed curricula informed by behavioral principles superior to cognitive approaches.


>...NLP probably never could have gotten very far without better computers.
>
NLP lacks a theoretical foundation. It should have been an application of GG, but GG and its descendants haven't progressed to the point of presenting an actual theory; instead they are concerned with what such a theory might be like were one to exist.


> AI actually has made a great deal of progress...
>
Only since _Parallel Distributed Processing_, which was a break with the rule based approach of traditional AI. The "classical" approach owes a great deal to Chomsky (though he distances himself from it). Much AI related work still has baggage from cognitive psychology. (There are notable exceptions, e.g., Rodney Brooks' "Intelligence Without Representation".)


> Psychology's problems are their own. Blaming Chomsky for that just isn't
> true.
>
Chomsky's arguments against Skinner are typically part of day 1 in any Intoduction to Cognitive Psychology course, e.g. a quick search turns up <http://comp.uark.edu/~lampinen/LEC1.html>


> generative grammar is a dead end
>
.

-- bill



More information about the lbo-talk mailing list