LINGUIST List 11.228

Thu Feb 3 2000

Disc: Phonemic Analysis

Editor for this issue: Lydia Grebenyova <>


  1. Larry Trask, Phonemic Analysis

Message 1: Phonemic Analysis

Date: Sun, 30 Jan 2000 13:48:47 +0000
From: Larry Trask <>
Subject: Phonemic Analysis

Jorge Guitart writes:

> I never thought I would see anybody advocating autonomous phonemics in the
> year 2000.

OK. I confess I've been too busy to follow this discussion closely, but I'm 
rashly going to jump in here.

I don't know what is meant by the word "advocating" above.

But I would like to point out that autonomous phonemics is a wonderfully
appropriate tool in a number of circumstances, such as language teaching,
lexicography, and the devising of practical orthographies for previously
unwritten languages.

I am currently teaching a course in English pronunciation to our English
Language majors. The goal of the course is to introduce the students to the
range of English accents which occur around the world, and to give them an
understanding of the phonological and phonetic differences among these.

For this purpose, I rely heavily upon an autonomous-phonemic approach,
complete with autonomous phonemes, allophones, phonotactics, and morpheme-
structure constraints. This, I think, is the only practical approach.
Submerging the students in a sea of contemporary phonological theory would,
for my purposes, be little more than a bewildering waste of time, and an
almighty distraction from the business at hand.
> We should not try to find out what the phonemes of the language are
> starting from the sounds: biuniqueness is not an inviolate constraint of
> any known human language and using minimal pairs is cheating because you
> are assuming that sounds are part of words but you said you were starting
> from the sounds. 

But this is not a criticism of autonomous phonemics generally. Rather,
it is only an attack on *one particular approach* to autonomous phonemics.

It is perfectly possible, and in many quarters usual, to adopt a version of
autonomous phonemics which abandons biuniqueness in favor of neutralization.
And it is also perfectly possible, and probably usual, to start the analysis 
in terms of lexical items and word-forms taken as given. That is, in order 
to do autonomous phonemics, you don't have to accept the most extreme positions
ever suggested by the American structuralists.

> Rather, we should find out what phonemes are starting
> from the lexical units, then see what happens to them under conditions
> of distortions, and then propose a theory of what mediates between lexical
> units and their physical realizations.

Well, I'm not happy with that word "distortions". Otherwise, though, what
is suggested here is *precisely* what I do in my presentation of autonomous
phonemics -- except that I don't think I'd care to dignify my account of
phonological processes with the label "theory".

Autonomous phonemics may not find much favor with contemporary phonological
theorists. But it remains an indispensable tool in a wide variety of
linguistic enterprises. It does not deserve to be dismissed as no more
than the laughable and fossilized relic of more primitive days.

Finally, and more generally, I would, as usual, like to raise my voice against
any suggestion that the business of linguistics instructors is to teach
their students the latest and most fashionable theories. In linguistics,
theories come and theories go, and this week's fashionable theory may be
old hat by the time the students graduate. Our business is to teach our
students to analyze linguistic data, using whatever tools may be convenient.
We should not be urging them merely to hop onto the latest theoretical
bandwagon. To do that is to fail them.

Larry Trask
University of Sussex
Brighton BN1 9QH

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue