Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!uknet!festival!edcogsci!usenet
From: mwc@cogsci.ed.ac.uk (Matt Crocker)
Subject: Re: Computational Psycholinguistics
In-Reply-To: vosse@ruls41.LeidenUniv.nl's message of 7 Feb 1995 13:23:12 GMT
Message-ID: <MWC.95Feb9120104@burns.cogsci.ed.ac.uk>
Sender: usenet@cogsci.ed.ac.uk (C News Software)
Nntp-Posting-Host: burns
Organization: Centre for Cognitive Science, University of Edinburgh
References: <3goda4$8i9@pipe3.pipeline.com> <3h1938$fh8@cat.cis.Brown.EDU>
	<3h7s80$sil@highway.LeidenUniv.nl>
Date: Thu, 9 Feb 1995 12:01:04 GMT
Lines: 50

In article <3h7s80$sil@highway.LeidenUniv.nl> vosse@ruls41.LeidenUniv.nl (Theo Vosse) writes:

   Mark Johnson (mj@cs.brown.edu) wrote:
   : In my opinion, the fundamental problem is that the psycholinguistic
   : data available today really doesn't enable one to identify with any
   : certainty even basic architectural properties of the human parser, and
   : it is not clear what experiments one could do to clarify the issue.

   Well, IMHO it will never ever be possible to determine these
   properties of the human parser with whatever experiment.

Possibly not with any *one* experimental paradigm, but by combining
evidence form, say, eye-tracking and event related potentials with
more 'off-line' evidence from aphasia, I think it is (and will be
increasingly) possible to identify and reduce the *space* of possible
architectures. It does, however, mean that computational
(psycho)-linguists will have to start getting their hands dirty,
taking into consideration a broad range of empirical findings :-)

   : There is even controversy as to whether humans construct (in any sense
   : of the word) something resembling the syntactic analysis trees posited
   : by linguists, and if so, what information they use in doing so.

   I don't think there is very much controversy. Can you name someone
   who really claims we don't use syntax (and who has built a model to

	   Theo Vosse

But Mark never said 'we don't use syntax', he simply said we may not
construct syntactic representations explicitly. Mark's own paper in
the Journal of Psycholinguistic Research (1989) demonstrated this
point very nicely. Of course one could argue that the representation
are real, bu simply manifest in the memory/control structures of the
parser, rather than constructed as explicit outputs of the parse
procedure.

Matt Crocker

-------Dr. Matthew Crocker-----------------------------------------
   Centre for Cognitive Science    |        CCS: +44 -31 650 4589
     University of Edinburgh       |        FAX: +44 -31 650 4587
        2 Buccleuch Place          |       Secr: +44 -31 650 4432
   Edinburgh, Scotland, EH8 9LW    |         mwc@cogsci.ed.ac.uk

--
-------Dr. Matthew Crocker-----------------------------------------
   Centre for Cognitive Science    |        CCS: +44 -31 650 4589
     University of Edinburgh       |        FAX: +44 -31 650 4587
        2 Buccleuch Place          |       Secr: +44 -31 650 4432
   Edinburgh, Scotland, EH8 9LW    |         mwc@cogsci.ed.ac.uk
