Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!news.sprintlink.net!howland.reston.ans.net!nntp.crl.com!pacbell.com!gw2.att.com!nntpa!mac-118.lz.att.com!user
From: rte@elmo.lz.att.com (Ralph T. Edwards)
Subject: Re: Chomksy, Significance, and Current Trends
Message-ID: <rte-1608951053260001@mac-118.lz.att.com>
Sender: news@nntpa.cb.att.com (Netnews Administration)
Nntp-Posting-Host: mac-118.lz.att.com
Organization: AT&T Bell Labs
References: <DD18rC.AK0@actrix.gen.nz> <jguy.12.302EB65E@trl.oz.au> <DDAKqG.E9D@actrix.gen.nz> <40q490$bbg@mailnews.kub.nl> <1995Aug15.100147.1@ctdvx5.priv.ornl.gov> <40qkp2$1pm@senator-bedfellow.MIT.EDU>
Date: Wed, 16 Aug 1995 15:53:26 GMT
Lines: 61

In article <40qkp2$1pm@senator-bedfellow.MIT.EDU>, David Pesetsky
<pesetsk@mit.edu> wrote:

Re Michelson-Morley
> No, they took an empirical claim, put it to an experimental test, and got a 
> negative result. That's more or less what Chomsky did to Skinner's proposal 
> as well (though his review involved only Gedanken experiments).
> 

A gedanken experiment isn't an experiment, it's the examination of what
the theory predicts in a particular case.  It's useful in examining the
self-consistancy of the theory, but no substitute for experiment.

> 
> P.S. By the way, am I misremembering, or is it the case that Michelson and 
> Morley themselves didn't believe their result, and were sure it must be due 
> to experimental error?

Wouldn't matter, as long as they had the courage to publish their
unexpected result.


I think the problem with linguistics is there are far too few experiments.
Things are asserted with plausibility arguments rather than experiments to
support them.  It's more like political science or interpretation of
history than science.

For example Chomsky asserts that folks carry around an underlying version
of the latinate roots in English with all the vowels in their stressed
form, which then express themselves when the stress shifts from one
syllable to another when various endings are added.  Maybe.  Or maybe they just
learn each word.  Or maybe they learn classes of words with similar behavior.
Point is I heard him assert this as fact.  It's not fact to me until I
hear the evidence for his opinion.

I can imagine an experiment.  Pick words that have an established pronunciation
which is known by very few.  Include some which are regular, some which are
irregular in how the rules allegedly work.  Ask a large group of people
to attempt to pronounce them.  I suggest they will show little competence unless
they know the word or several directly analogous words
(same vowels, same endings).
Ask them to define the target and analogous words to see if they know them,
correlate competence with knowledge of the word or completely analogous words.
Has anyone done anything like this?

It's also probably true that people vary considerably in their ability to do
this for reasons other than which words they know, including how they were
taught to read.  My wife is likely to give an unknown word a German or
French
pronunciation (although a native English speaker) because she was taught the
infamous see-and-say hieroglypic method for reading English.

Now I recognize that my experiment above doesn't really get at the issue
of whether people carry around underlying representations for words that
recapitulate the last 2000 years of Latin and English phonology, because
it mixes in ability to read.  Which raises the question, can anyone devise
an experiment to test this allegation?  If not, it's just an amusing
thought, not fact.

-- 
R.T.Edwards rte@elmo.att.com 908 576-3031
