Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!noc.near.net!paperboy.wellfleet.com!news-feed-1.peachnet.edu!emory!swrinde!howland.reston.ans.net!pipex!lyra.csx.cam.ac.uk!warwick!uknet!festival!hwcee!greg
From: greg@cee.hw.ac.uk
Subject: Re: Chips & chimps
Message-ID: <CwKvJG.19x@cee.hw.ac.uk>
Originator: greg@thor.cee.hw.ac.uk
Sender: news@cee.hw.ac.uk (News Administrator)
Organization: Dept of Computing and Electrical Engineering, Heriot-Watt University
References: <CwD6qw.1vr@news.Hawaii.Edu>
Date: Fri, 23 Sep 1994 10:23:39 GMT
Lines: 26

From article <CwD6qw.1vr@news.Hawaii.Edu>, by roitblat@uhunix.uhcc.Hawaii.Edu (Herbert Roitblat):
> If you knew the entire corpus ahead of making your model you could
> make some very  uninteresting models of chimp or any other form of
> "language" production.  The question about chimps is whether they
> employ such an exhaustive enumeration or a productive grammar.

I'm sure that the question of exhaustive enumeration versus productive
grammar is of interest to chimp linguists but frankly I don't care if
chimps have 18,000 tiny Searles in their heads, all reading chimp language
instruction manuals. Rather, how would people who work with chimps
& "natural" language respond to a system that had the same "natural"
language performance and competence as a chimp? To test this, any model
satisfying these criteria would do.

People see through ELIZA & PARRY very quickly but many who work with
chimp language seem to think that the chimps really are communicating
effectively with it. Would they think the same of a simulation? Chimps
are organic and, as relativists keep telling us, share 98% of their DNA
with human beings. Maybe the missing 2% is where Searle et al think
that human language semantics comes from...?

Anyway, my speculation is that as chimp language is quite restricted
it should be possible to build such a system and I wondered if anyone
had done so. Alas, it seems that there's not much between SHRDLU's big
red box, English lexicons and parsers without semantics and as yet
unproven mega-projects like CYC.
