Newsgroups: sci.lang,alt.usage.english
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!mvb.saic.com!eskimo!rickw
From: rickw@eskimo.com (Richard Wojcik)
Subject: Chomsky's Theorem
X-Nntp-Posting-Host: eskimo.com
Message-ID: <D60BG3.5qp@eskimo.com>
Followup-To: alt.usage.english
Sender: usenet@eskimo.com (News User Id)
Organization: Eskimo North (206) For-Ever
Date: Sat, 25 Mar 1995 17:33:39 GMT
Lines: 158

Mark Israel:  MI>
Rick Wojcik:  RW>

Mark and I have been going back and forth on the significance of "Chomsky's
Theorem".  What follows is a lengthy response to some of his comments.

     [re: the passage from the encyclopedia of linguistic terms]
MI>> Thus, not only is Chomsky's Theorem mentioned:  the immediate
MI>> context talks about natural languages!

RW> Read the passage again, Mark.  "A theorem of Chomsky" is not 
RW> "Chomsky's Theorem".

MI>   Oh, so you meant that I must have pulled the *names* of the 
MI>theorems out of a textbook.  Nope, I didn't do that either.  And I 
MI>said that the encyclopaedia mentioned the theorem, not that it 
MI>mentioned it by a particular name.

It was the names of the theorems in the context of a remark about general
linguistics that originally led to this thread.  It seems that you did some
research and couldn't find any reference to them in any work, although I
grant you that they exist in some references somewhere.  It should tell you
something that you didn't even find them in Bright's encyclopedia.  If you
ask a linguist what "Chomsky's Theorem" is, you are likely to get the same
response that I gave you, i.e. "you tell me".

[snip]
RW> No linguist promotes the view that "Chomsky's Theorem" or 
RW> "Greibach's Theorem" have anything to do with natural language,

MI>   Then why does the encyclopaedia juxtapose a sentence about one 
MI>with a sentence about the other?  The article says, "Many useful 
MI>theorems have been proved about CFLs."  It doesn't say "useful for 
MI>artificial languages only"; on the contrary, it talks about the use 
MI>of recursive transition networks in natural language processing 
MI>systems.

You are correct about what the article doesn't say, not about what it does
say.  Those theorems play a role in computational linguistics and compiler
theory, which is why you learned about them in the first place from your
computer science course, not your introductory linguistics course.
Although natural language processing has gradually become a part of the
linguistic curriculum, it is still a fairly peripheral field of study for
linguistics students.  The encyclopedia mentions "theorems by Chomsky" in
passing because this work has been important to non-linguists and to
linguists who specialize in studying the mathematical properties of
grammars.

MI>> "Traditional grammarians" may not have been linguists, but
MI>> "structuralist grammarians" (Leonard Bloomfield et al.) certainly
MI>> were.  And the encyclopaedia considers these demonstrations
MI>> important.

RW> This is completely lost on me.  Why are you mentioning Bloomfield 
RW> here?  He has nothing to do with Chomsky.

MI>   The topic was the attention linguists have paid to CFGs.  
MI>Structuralist grammarians (I mentioned Bloomfield merely to remind
MI>people that structuralists were *linguists*) are said to have 
MI>tacitly assumed that natural languages were CFLs.  Chomsky argued
MI>otherwise.  And the encyclopaedia regards these contentions as
MI>important.  Got it now?

The topic was not the attention that linguists have paid to CFGs, but the
relevance of two theorems to general linguistic theory.  The fact that
those theorems are about CFG does not make them relevant to the kind of
immediate constituent analysis that pre-Chomskyan linguists were noted for.

MI>> The generative grammars are binary:  either a string is in the 
MI>> language or it is not.  Real natural languages have (as the
MI>> OED editors sagely put it) "a well-defined centre but no
MI>> discernible circumference."

RW> This is total gobbledegook.

MI>   Jacques Guy had no difficulty understanding this "total
MI>gobbledegook".  Did anyone besides Richard find it hard to 
MI>understand?

Jacques was responding to your words, not the relevance of your words to
the point of the discussion.  I rather enjoyed his comment about the little
holes, though.  ;-)  

[snip]
MI>   For sci.lang readers who have just joined this thread, I shall 
MI>elucidate.  Richard wrote:  "The terms 'prescriptive' and 
MI>'descriptive' are frequently misunderstood. [...]  In the 
MI>'prescriptive' world, you should be able to create a definitive 
MI>dictionary that never needs to be updated.  The prescriptivist never 
MI>needs to update the dictionary because it is the dictionary, not 
MI>current usage, that defines the standard."

MI>   I asked him if he could produce a single real prescriptivist who
MI>believed this.  He evaded my challenge several times before finally
MI>saying, "you misconstrued my original statement as being about real 
MI>people."

What I said was that my argument was a reductio ad absurdum.  My use of the
generic phrase "the prescriptivist" was not intended to refer to any
particular individual.  My remarks were based on the fact that
prescriptivists tend to believe in a "pure form" of the language, and I
have since backed up that interpretation with passages from introductory
texts (not posted in sci.lang).  The fact that you accept the need for
linguistic change and then call yourself a prescriptivist does not mean
that prescriptivism can be defined as an approach that embraces linguistic
change and the need for frequent updates of dictionaries to bring entries
into greater harmony with current usage.  

   [responding to Cameron Majidi]
>   My opponent, Richard Wojcik, has spent his life studying linguistics.  

Not quite.  I feel that I have spent a good part of it getting myself
bogged down in meandering debates on usenet.  ;-)

>   I took one course in it, learned some related stuff in CompSci courses,
>and have done some additional reading and held some conversations.

I have never challenged your credentials, Mark.  :-)

MI>   I learned Chomsky's Theorem in a CompSci course that I took many years 
MI>ago.  When Richard challenged me on its relevance to linguistics, I did
MI>bother to look it up in a linguistics encyclopaedia.  Would you rather I
MI>hadn't?

Thank you, Mark.  After repeated attempts to get this information out of
you, you finally spill it to Cameron.  How did you do that, Cameron?  I
would like to remind you, Mark, that the encyclopedia contained no
reference to a theorem called "Chomsky's Theorem."  That encyclopedia
article was the best you could do to back up a rather flippant remark.

CM>> Note Mr. Israel's helplessness when his knowledge of the technical
CM>> details of the theories in question is called into question.

MI>   I grant that it may be possible to stump me with a question about the
MI>technical details (although Richard has so far failed to do so).  The 
MI>topic, however, was not the technical details of the theories, but rather 
MI>the importance that linguists attach to the theories.

I hope that you didn't expect me to try to stump you.  It should be
remembered that you brought up the technical details yourself and
misconstrued their significance to linguistic theory.

>> Don't let Mr. Israel fool you.
>
>   Don't let *anyone* fool you!  As Richard himself said (but not to me), 
>"I'm a bit nonplussed by your insistence on authority [...].  Even 
>geniuses and Ph.D. linguists can be wrong, can't they?  You should demand 
>an explanation from everyone."

As long as you are quoting me out of context, let me embellish my remarks.
Demand an explanation, but pay attention to it.  There is nothing wrong
with consulting authorities, but there is something wrong with dismissing
their advice and explanations on the basis of who they are rather than what
they said.  And this is something that I am saying directly to you, Mr.
Anti-Linguist.  
-- 
Rick Wojcik  rickw@eskimo.com     Seattle (for locals: Bellevue), WA
             http://www.eskimo.com/~rickw/
