autonomous syntax

[Ellen F. Prince]

jon aske wrote:

>But the real question when it comes to autonomy vs. non-autonomy, is
>whether the constructions of a language can, or should, be described
>independently of the semantic and pragmatic meanings which they are used
>to express, and independently, for instance, of the iconic and universal
>principles, such as topic-comment or comment-topic, on which they are
>sometimes based.
>
>I don't think they can and I don't think they should.  The simple reason
>for this is that I do not think that that is how humans learn or store
>the constructions of a language.  Form is always stored and intimately
>connected to function and that is how it should be described and
>analyzed.

well, i guess it all depends on where you're looking. for quite a few
years now i've been looking at cases of language contact where the
discourse functions associated with a syntactic form in one language
come to be associated with an 'analogous' syntactic form in a contact
language (and where the analogy is statable in purely syntactic terms)
and where the two forms in question may have originally had totally
unrelated discourse functions. in fact, it is precisely by studying
such cases that i have come to believe in autonomous syntax, since, if
the form-function connection were permanent or driven by iconicity, i
could simply not begin to explain the data.