From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!ogicse!milton!forbis Tue Jan 28 12:15:28 EST 1992
Article 2988 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:2988 sci.philosophy.tech:1931
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!ogicse!milton!forbis
>From: forbis@milton.u.washington.edu (Gary Forbis)
Newsgroups: comp.ai.philosophy,sci.philosophy.tech
Subject: Re: Table-lookup Chinese speaker
Message-ID: <1992Jan22.045102.20148@milton.u.washington.edu>
Date: 22 Jan 92 04:51:02 GMT
Article-I.D.: milton.1992Jan22.045102.20148
References: <1992Jan20.182835.5307@spss.com> <1992Jan21.191924.18205@aisb.ed.ac.uk> <1992Jan21.182524.7880@husc3.harvard.edu>
Organization: University of Washington, Seattle
Lines: 54

In article <1992Jan21.182524.7880@husc3.harvard.edu> zeleny@widder.harvard.edu (Mikhail Zeleny) writes:
>I can't understand how this thread ever managed to get so far without
>considering a very basic problem, that of context dependence of indexicals,
>including pronouns, demonstratives, anaphora, and tensed verbs.

I think they are included.  This is so by the assumption that the context is
conveyed within the medium used, in this case, written text.  How are humans
able to convey such context through the written medium if machines cannot?

>Given this
>consideration, it is easy to see that the truth-value of the predicate "X
>is a meaningful Chinese sentence" depends not only on the sentential
>antecedents of X, but also on the concrete non-linguistic context of the
>entire discourse.

The Turing text limits one to a purely syntactical exchange.  If this is a
problem it will remain a problem though any medium.

>Thus the notion of a predefined table of meaningful
>conversations is inherently unrealizable, being that the infinitely
>variable context can't *in principle* be fixed in advance.

I could accept this if I could accept that any particular entity can be
exposed to an infinitely variable context rather than a very large finitely
variable context. That is, is there any evidence that any particular
individual can participate in more than a finite number of syntatical
exchanges?  I think the evidence is overwhelmingly that none can.  This is
true even when raised eyebrows, frowns, slaps, extended fingers, etc. are
added.

>In this way,
>the problem reduces to that of reference, for even if you equip your
>artificial speaker with any number of sensors estabilishing a putative
>connection between the content of its sentences, the operational success of
>such connection will depend on the machine's capability of synthesizing a
>single representation of an external object on the basis of infinitely
>variable quasi-phenomenal data caused by its presence.  Since I have
>previously discussed ad nauseam the intractability of this problem by
>algorithmic means, I trust that the readers will excuse me from making
>further arguments to that effect.

I have failed to see any which did not first require me to accept an
non-recursive, infinte context.  It's probably a problem with me.  Sorry.

Do any think less of me just becuase I am not omniscient?  Do I syntnesize
single (correct) representations of external objects on the basis of 
infinitely variable quasi-phenomenal data caused by their presence (without 
recoginizing I am doing so)?  Does my inability to do so prevent me from
connecting the content of my sentences?

>: Mikhail Zeleny                                                     :
>: email zeleny@zariski.harvard.edu or zeleny@HUMA1.BITNET            :

--gary forbis@u.washington.edu


