Prevailing approaches do not have a computational lexicon (fwd)

[Andrew Carnie]

Hi All,

I asked Heidi Harley to pop in on Carl's queston about the non-lexicalist
approaches to MP. Here's her response (forwarded with permission).

AC
---------- Forwarded message ----------
Date: Wed, 25 Sep 2002 12:37:20 -0700
From: Heidi Harley <hharley at email.arizona.edu>
To: Andrew Carnie <carnie at U.Arizona.EDU>
Cc: pollard at ling.ohio-state.edu
Subject: Re: Prevailing approaches do not have a computational lexicon (fwd)

Dear Andrew, Carl --

I hesitate to venture into deep waters here, but
as Tanya Reinhart casually referred to the non-lexicalist
view as the 'prevailing' one (which does surprise me
considerably!), someone oughta be able to explicate
what that prevailing view is...

>I suppose some would say this is a logical place to end up
>starting from "Remarks on nominalization".

Some wouldn't, however. The most strongly worded statement
of the apparently prevailing view is an article by Alec
Marantz that appeared in a UPenn Working Paper in 1997,
called "No escape from syntax: Don't try morphological
analysis in the privacy of your own lexicon".  The full
reference is author and title as above, UPenn WPL 4.2,
A Dimitriadis et. al, eds, pp. 201-225. It's an argument
against lexicalism starting from premises laid out
in Remarks.

I'd be happy to go over the general picture, although it
was pretty lucidly stated by Andrew (though picking DIE
as the example of a Root was maybe not the most historically
felicitous example to start with). In fact, if you were
to press me on it, I'd say that the roots don't even
come into the picture until Late Insertion comes along --
up until then, the syntax doesn't know whether it's
working with 'cat' or 'dog' or what (as, indeed, it
shouldn't). The syntax DOES work with the things it
cares about -- morphosyntactic features, plus *a very
few* semantically 'heavier' things, basically the verbalizers
and nominalizers Andrew mentioned. Spell-out is constrained
by regular interpretive-morphology ideas about
feature-matching and competition, i.e. Panini's principle.
Roots, when they're inserted, are subject to licensing
restrictions -- they've got subcategorization
frames on them.

But I don't know whether you're interested in all
that. It certainly surprised ME to hear it was the
prevailing view!

best, hh








---------------------------------------------------------------------
Heidi Harley
Department of Linguistics
Douglass 200E
University of Arizona
Tucson, AZ 85721
Ph: (520) 626-3554
Fax: (520) 626-9014
hharley at u.arizona.edu