Theiling Online    Sitemap    Conlang Mailing List HQ   

THEORY: storage v computation (was: RE: Language revival)

From:And Rosta <a.rosta@...>
Date:Thursday, November 25, 1999, 22:57
[I was going to reply privately to Ed, but since several people
have posted on this topic, I decided to post it to the list.]

> My reasoning comes down to two factors: > > 1. I learned from reliable sources that recall is, in general brain > function, vastly superior to computation. > > 2. I believe that language uses essentially similar brain processes > to any other human activity, not different ones in kind. > > (2) was explicitly rejected by the Chomskian tradition, and that > tradition has formed many of the foundational assumptions of language > study for the past several decades.
*some* language study; there are plenty of unmarginalized dissenters.
> Indeed, the opposite idea from my > claim -- that computation is preferable to storage, and so we should > always describe patterns in language in terms of "rules" (implying > that the results of these "rules" are computed on-line) -- has been > taken for granted for a long time in linguistics.
partly this is because they're not models of processing or performance
> So it's not at all > surprising that my claim sounds controversial to the point of being > intuitively false. > > And yet, given (1) and (2), which I hold, it follows inevitably.
In fact, your claims strike me as pretty much the received wisdom. The point you make about the Chomskyan tradition is true enough, but most people in that tradition don't see themselves as doing psycholinguistics & building processing models. And quite rightly so. I am, though, aware of the existence of recent studies purporting to show that no irregular English verb form is faster than any regular verb from (allowing for frequency effects). I have been told about these studies (cited in a recent Pinker book - _How the mind works_, maybe) and do not have the references. But the lesson is clear: in production, irregularity appears to confer no benefits (if the experimental findings are correct). --And.