form, function, data, description

[Scott Delancey]

On Fri, 25 Feb 2000, Dan Everett wrote:

> The grammar of Hidatsa, by Hu Matthews, is still worth reading - Chomsky
> (1965) said it was worth 1,000 descriptive grammars uniformed by theory.
> I don't know, but it is good.

Peculiar example to choose--I've only ever heard this mentioned (by
Americanists, Siouanists in particular) as a particularly egregious
example of how useless most grammars written in an explicitly generative
framework are.  I can't claim to have read through it myself, but I
did, many years ago, try to use it as a resource for something I was
working on, and found it, to put it diplomatically, a very frustrating
experience.  "Worth", of course, is a two-argument predicate; there's
no such thing as intrinsic worth, only worth to somebody--Matthews'
grammar may well be worth a lot to Chomsky, but I honestly don't think
it would be on many people's lists of really useful grammars.

Scott DeLancey