NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/bit/listserv/csgl/1607

Text File | 1992-11-24 | 7.1 KB | 136 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!europa.asd.contel.com!paladin.american.edu!auvm!EID.ANL.GOV!GABRIEL
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.csg-l
Date:         Mon, 23 Nov 1992 17:28:30 CST
Sender:       "Control Systems Group Network (CSGnet)" <[e-mail]>
From:         John Gabriel <[e-mail]>
Subject:      foundations of language (B nevin 921123 09:58:11)
Lines: 125

[From gabriel 921123 17:11 CST]

>[From: Bruce Nevin (Mon 921123 09:58:11)]

>No time to keep up, but as I sneaked a glance at the most recent
>this AM this caught my eye:

>(Bill Powers (921122.2000) ) --
>/*****************************************************************************
>*                                                                            *
>* I think the deaf community would have a lot to teach about what happens    *
>* when a group of people is brought together without any initial means of    *
>* communicating. I claim that someone would ALWAYS invent a language,        *
>* using whatever perceptions are easily manipulable as the tokens. In        *
>* other words, I claim that it isn't language that's inherent, but           *
>* ingenuity.                                                                 *
>*****************************************************************************/

>When has a deaf community invented a language in the absence of
>an environing community of people possessed of language who
>intercommunicate with the deaf folks by whatever means they are
>able?

Don't fish communicate non verbally? e.g. the UK stickleback, or
lots of tropical species while nesting, to say nothing of Bluegill?

>You made a similar point, Penni. However, you went on to say:

>Penni Sibun <[e-mail]>
>/*****************************************************************************
>*                                                                            *
>* communication is a terribly important part of what human beings (and       *
>* all organisms!) do.  language may or may not involve symbol                *
>* manipulation.  but language is fundamentally communcation.  wouldn't       *
>* it be a good idea to start studying a language at its foundations?         *
>*****************************************************************************/

>I disagree that language "is fundamentally communication".  Even
>more basically, people use language to create information that
>could not exist without language and could not have existed prior
>to language.

I'm really an amateur at this game, but it seems to me that a Shannon
channel certainly communicates what the probable transitions of the
source are. What this means depends on context. Many of the things
that interest me seem to be descriptions of traversals of real or
hypothetical decision trees, the real traversals often being to solicit
the an estimate of probable truth of a proposition determining which
branch to take at the sender's current node. The hypothetical decision
trees (penni's very good example - X's room is down the hall on the right)
being to establish context about an "aforesaid" which can hencforth
be abbreviated by the symbol sequence "X's room" (avery's comment).
Neither Chomsky or any of the programming language people like myself
have any trouble with this. Question - probably trivial to answer -
are the programming languages of interest to linguists as languages
for communication with machines? I'm reminded of a comment of mine from
years ago - a program is a contract between people and machines
specifying what each must do to achieve some result desired by people.


>This is not to say that the perceptions ("meanings") with which
>words are correlated could not have existed, though some
>categories may well not be universal, but rather that
>higher-level organizations of those perceptions (which are
>themselves perceptions, of course) are not merely communicated by
>means of language, but are consensually created by means of
>language.

One of the nice things about programming languages is that there are
ANSI standards for their meanings and working models for their
understanding by machines. Yes, they are truly dull for linguists,
but perhaps looking at a context free grammar might simplify the
discussion because the meaning of a term cannot change subtly in the
middle of a paragraph - or perhaps it can, but only in well defined
ways. A dumb question perhaps, but is it worth thinking about? For
example the elementary concepts of LISP 1.5 (the undefined terms of`
axiopmatic theories) can be used to build artifacts having apparently
complex behaviours - e.g. LIZA. Are these as worthy of study as the
PCT ARM, and if so why (or if not why). And any real light PCT could shed
on the process of making good software will make its discoverer rich
and famous.

>So I am arguing that communication is not at the foundations of
>language but rather antecedent to them.  To be sure,
>communication has an important role in the story Harris proposes
>about the evolutionary origins of language (in _Language and
>Information_ and in _A Theory of Language and Information: a
>mathematical approach_).  But to get at the foundations of
>language you must study something that is unique to language.

>Just to head off a red herring, note that Shannon/Weaver
>communication theory is not about communicating in this sense
>either, any more than its near synonym information theory is
>about meaning in any general sense of the word.

Again, perhaps a dumb comment. The original Shannon Weaver paper
was a very abstract affair. When I look at either ENIGMA (the
biography of Turing, or the account of Bletchley Hall 1938-1945)
or back issues of CRYPTOLOGIA, these make me inclined to argue that
the red herring is not perhaps as much of a false trail as you might
think. But it seems clear to me that there are subtleties in bruce's
view of language I will not appreciate until we have more shared
experience. Some of the recent articles about Rosetta Stone
like problems that are still open might be of interest to linguists,
and have at least to me a VERY Shannon flavour. Bruce will know of
Bill Woods I am sure from Bill's long association with BBN. How much
difference really is there between Shannon's traversals of a Markov
Diagram, and an individual's traversals of one of Bill's semantic nets, or
successive retrievals from penni's KB?.  Or for that matter the
successive retrievals from a Kanerva database as the algorithm seeks
the closest match to a given pattern - Well, perhaps that's a bit
far fetched. But how about two very similar Kanerva Databases, and
the mapping between them by retrieval of closest matches to a
succession of different patterns!! Perhaps that question IS
worth some thought.

With Greetings to all, and malice to none. Avery's excerpt from another
SIG suggests that our flames are low temperature. Perhaps the diversity
of interests here is a damper for the free radicals of flame propagation.
No pun intended.

Well, I probably need to find time to read Harris as bruce suggests,
before succumbing to fatal hoof in mouth disease.

                        John Gabriel ([e-mail])