linguistics wars

Tue Apr 22 02:56:56 UTC 1997

It seems that every so often another one of these
generativists-vs.-functionalists debates gets going on Funknet (or
Linguist); people talk at cross-purposes for awhile, argue about
terminology, etc.; then the debate fades away. Does anyone ever go away
with a changed mind?

Generativist theories of lg. acquisition claim that the range of
languages a child can learn is constrained by _innate_ __syntactic__
structure. Therefore, children only need sufficient evidence to figure
out which of the possible grammars they are encountering.

I believe this is the claim that functionalists disagree with. I think
(you all correct me if I'm wrong) most functionalists would agree that
children are born with _something_ _innate_ that constrains the types of
grammars they can learn, but that this something is not modularly
__syntactic__. It is related to (or may consist solely of)
semantic and possibly more general innate predispositions concerning cognition.
Hence the consistent appeals to semantics and more-general processing
strategies by functionalists.

As to evidence, we need to sort out what kind of evidence can falsify the
generativists' claim, whether that evidence has been presented or not,
and, if it has, whether anybody is ignoring it because they are too
firmly attached to their theoretical position. I can't speak
authoritatively on these issues.

I think that Dan Everett should be cautious when saying that Givon and
Lakoff both think syntax is necessary. I think they wouldn't agree that
__autonomous__ syntax is supportable.

This whole discussion got started because of questions about rules and
formalisms (those old bugbears). I am wondering how familiar contributors
to the discussion are with network models of knowledge of language (a la
Langacker, Bybee, Hudson and Lamb) and the fact that rules can be read
off of the network via schematization at ever more abstract levels, plus
the connections that reside in the network between experienced
forms/meanings and others that share features of form/meaning. Rules
do not have to be listed separately from their outputs (they are immanent
within them), yet are always available for the computation of novel

Johanna Rubba   Assistant Professor, Linguistics              ~
English Department, California Polytechnic State University   ~
San Luis Obispo, CA 93407                                     ~
Tel. (805)-756-2184  E-mail: jrubba at      ~

More information about the Funknet mailing list