Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!Germany.EU.net!news.dfn.de!news.rwth-aachen.de!nntp.gmd.de!dearn!barilvm!news.datasrv.co.il!adam.tovna.co.il!dr
From: dr@tovna.co.il (Daniel Radzinski)
Subject: Re: Definiton for NL & NLP
Message-ID: <D1264E.64t@tovna.co.il>
Organization: Tovna Translation Machines Ltd.
X-Newsreader: TIN version 1.2 PL2
References: <D0My3L.44J@ritz.mordor.com> <3cgofu$f5n@hobbes.cc.uga.edu>
Date: Mon, 19 Dec 1994 12:45:50 GMT
Lines: 50

Michael Covington (mcovingt@ai.uga.edu) wrote:
: Niranjan Hira (merlin@ritz.mordor.com) wrote:
: : 'ello fair gentry.

: : 	The subject just about says it all.  I'm writing a paper and
: : would love to see a definiton (not necessarily formal) for _Natural
: : Language_ and one for _Natural Language Processing_.

: : 	Can any gurus enlighten me?

: "Natural language" simply means an ordinary human language, not an
                                              ~~~~~~~~~~~~~~
: artificial language made up for special purposes.  That is, it means
: things like English, French, Hindi, Kwakiutl, etc., not Fortran,
: BASIC, Pascal, or symbolic logic.

: (Nothing subtle is being said about "naturalness" here.  All we mean
: is that it's not an artificial, consciously-made-up language.)

: NLP is simply the computer processing of information that arrives
: expressed in a natural (= human) language.
                 ~~~~~~~~~~~~~~~~~~~~~~~~~~

: --
: < Michael A. Covington, Assc Rsch Scientist, Artificial Intelligence Center >
: < The University of Georgia, Athens, GA 30602-7415 USA  mcovingt@ai.uga.edu >
: < Unless specifically indicated, I am not speaking for the University. >  <><

True as a first approximation. However, recall Postal and Langendoen's (1984)
highly controversial Vastness argument which goes like this: NLs contain
strings of infinite (vs. merely indefinite or unbounded) length and are
therefore not recursively enumerable. Thus NLs are not human languages, since
human languages do not, and can not, contain strings of infinite length.

I, for one, interpret the Vastness argument as merely a clarification for
distinguishing between the notions of "natural language" and "human language."
In other words, a HUMAN language L is the countably infinite set of
finite-length strings within the larger uncountable NATURAL language L'.

We can, of course, always simply reject Postal and Langendoen's premiss re
strings of infinite length and then the definitions of NL and HL would be the
same. Just to clarify that some distinguish between these two notions, however
unacceptable/unpopular/unknown the distinction may be.


--
Daniel Radzinski
Tovna Translation Machines
Jerusalem, Israel
dr@tovna.co.il
