RCode vs CCode
R. V. Guha <ai.guha@mcc.com>
Date: Sat, 8 Sep 90 11:22 CDT
From: R. V. Guha <ai.guha@mcc.com>
Subject: RCode vs CCode
To: interlingua@venera.isi.edu
Message-id: <19900908162210.5.GUHA@GAIA.ACA.MCC.COM>
I have this feeling I am missing something here,
but the distinction between RCode and CCode seems
very very fuzzy and use dependent to me. I would
appreciate any clarifications/responses that
straighten me out on this.
The more I think about the RCode CCode distinction,
the more uncomfortable I am about it. My main problem
is as follows - given a string, can I say whether
it is a piece of RCode or CCode? I am not sure
we can. The issue seems to be not whether it is RCode
or CCode, but of what the conventions to used when
interpreting it are.
Lets assume A sends a string to B. There has to be some
shared convention if any sense is to be made of it by B.
At one extreme, these conventions could be
first order semantics (model theory and all that). At another
extreme, the conventions could be that of english. One
extreme seems to correspond to RCode and the other CCode.
There are obviously a lot of intermediate points.
Is there any clear demarcation point dividing things
into CCode and RCode? The one point that struck me
was whether we could give a mathematical charecterization
of the convention - but then anything we accept as an interlingua
will need such a specification.
Pat made a distinction between whether inference could
be done directly on the string sent over. This demarcation
point assumes that by inference one mean strictly something
like deduction. I can claim that the translation from CCode
to RCode is itself some form of inference. (Eg. Consider
a Boyer moore thm prover which translates the fol sentences
into its internal form - both are surely RCode, but the
transformation needs to be done before inference can be done).
Or what if we used something like basic english to communicate and
we have a set of transformation rules (not unlike a proof
method for say Montague's semantics for english) to derive
conclusions from it. Sure, it would be very hard to come
up with these, but how complex these inference rules are is a
fuzzy thing.
I don't like these conclusions because it seems to loose
what seemed to be a nice distinction between communication
and rep. languages and turns the distinction into a much
less sharp distinction between simple and complex communication
conventions.
On another point - Peter says that the interlingua is RCode
because it talks about denotation, etc. I don't get this point.
The linguists worry about the denotation of words
in english - does this make english RCode?
Very confused
Guha