From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!wupost!darwin.sura.net!jvnc.net!nuscc!hilbert!smoliar Tue Nov 26 12:31:50 EST 1991
Article 1531 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:1531 sci.philosophy.tech:1076
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!wupost!darwin.sura.net!jvnc.net!nuscc!hilbert!smoliar
>From: smoliar@hilbert.iss.nus.sg (stephen smoliar)
Newsgroups: comp.ai.philosophy,sci.philosophy.tech
Subject: Re: Natural languages are formal systems?
Message-ID: <1991Nov24.003959.2482@nuscc.nus.sg>
Date: 24 Nov 91 00:39:59 GMT
References: <9myTBB1w164w@depsych.Gwinnett.COM>
Sender: usenet@nuscc.nus.sg
Organization: Institute of Systems Science, NUS, Singapore
Lines: 75

In article <9myTBB1w164w@depsych.Gwinnett.COM> rc@depsych.Gwinnett.COM (Richard
Carlson) writes:
>
>Why is it a mistake to treat concepts as mental entities?  That's
>what they seem to be.
>
I do not think there is anything to be gained from repeating the attack on
introspection which was just posted here, but Richard's response to his own
question is pretty flimsy.  If we wanted to be methodical about it, we might
want to turn the question around and ask what could concepts be OTHER than
mental entities.  If we then shot down all the viable alternatives we could
come up with, we might feel a bit more secure in our position.

Unfortunately, I am not that confident that we can easily make a case for
associating the way we use the word "concepts" with any kind of ENTITIES
at all!  Treating them as entities puts them in the same class as abstract
data types--meaning that, in a very rigorous way, we can talk about attributes
which they do and do not possess, relationships among them, and operators for
manipulating them.  What makes Wittgenstein's PHILOSOPHICAL INVESTIGATIONS so
interesting is that prospect that the way we REALLY USE the "concept of
concept" is just to elusive to stand up to that sort of rigor.  Even more
important is the observation that we can still get along quite well in the
world even if we are not supported by that rigorous foundation.

IS it a mistake to treat concepts as entities?  Well, on the other side of the
coin, treating them as such as probably prerequisite to talking about them at
all.  This is what Minsky means when he says that we have to "thing-ify"
certain aspects of our experiences before we can talk about them.  However,
if those "things" about which we talk are not REALLY THINGS (which is the
direction Wittgenstein was trying to investigate), then we have to be very
careful about what we say.  In particular, we should not treat the sentences
we use as formal assertions in a logical calculus, endowed with a truth value
which may then be associated with truth values of other such formal assertions.
To oversimply the situation, the sentence "They are playing a game" simply
cannot be subject to the same rules as the sentence "For all integers x and
y, x + y is equal to y + x."  Once we recognize that caveat, we can treat
concepts as any sort of entity we wish.  (Note, by the way, that, unlike
Mikhail, I do not believe that the views of PHILOSOPHICAL INVESTIGATIONS
"exclude AI."  They simply caution us against taking the wrong premises
for granted.)

>The pride and joy of structuralist linguistics is Jacobson's
>structuralist phonology.  It is as precise and testable as a
>theory in physics and perhaps the only really universal,
>cross-culturally valid phenomenon, describing the phonemes that
>may exist in any language the way the periodic table describes
>elements, in the "human sciences."  It is based on Saussure's
>basic notions of the bipolarity of human language.  It reveals the
>individual "phonemes" of a language to be not primitive building
>blocks or units, but themselves comprised of "features" which are
>"distinctive."  The human phonetic system is about as well
>understood as any aspect of reality and seems to be a good
>candidate to serve as a model for other, as yet not fully grasped,
>aspects of language.  And it is not a "formal"-looking system but
>a dynamic _structure_ built on the kind of bipolar oppositions
>Saussure said we would find in every aspect of natural language.
>
This is all very well and good as far as formal systems are concerned.
However, the proof of any phonological pudding lies in its application
to either the production or the perception of speech.  Does anyone have
any reports of the influence of Jacobson's system on the implementation
of effective systems for either speech synthesis or speech recognition?
As I understand it, the physical properties of the speech signal seem to
be at considerable odds with most theories of phonology, which is to say
that it is often impossible to isolate phonemes by examination of the acoustic
wave-forms.  This may mean that phonemes are a POST HOC phenomenon arising from
our ability to explain what we here--a product of "thing-ification" applied to
speech, perhaps.  While they may be very useful for certain approaches to the
study of spoken language, as is the case with the "concept of concept," we had
better keep our eyes open with regard to their limitations.
-- 
Stephen W. Smoliar; Institute of Systems Science
National University of Singapore; Heng Mui Keng Terrace
Kent Ridge, SINGAPORE 0511
Internet:  smoliar@iss.nus.sg


