On Fri, 10 Nov 2000, Jeff Jones wrote:

> Uh, excuse me, but what I understand phonemes to be is along the lines of
> what you say in the *following* paragraph, *not* the *preceding* one.

I snipped the paragraphs. There seems to be a looseness in how the
term 'phoneme' is being used and understood. This is understandable,
since the American structuralists didn't agree, either. Here are some
quotes from some of the introductory texts I have on my shelf:

Among the gross acoustic features of any utterance, certain ones are
distinctive, recurring in recognizable and relatively constant shape
in successive utterances.These distinctive features occur in lumps or
bundles, each one of which we call a phoneme. (Bloomfield)

A list or table of the phonemes of a language should therefore ignore
all non-distinctive features. (Bloomfield)

The phonological system of a language is a network of differences
between sounds. A phoneme is an element of such a system. (Hockett)

A phoneme is a functional unit of sound. (Robert A. Hall, Jr)

A phoneme is a class of sounds. (Gleason)

So a phoneme is: 1) the bundle of features which remains invariant
from one context to another (Bloomfield), 2) a contrastive element in
a phonological system (Hockett, Hall), or 3) a group of sounds which
are related to each other (Gleason).

There was also a big rift between those who subscribed to a
"mentalist" view of the phoneme (Sapir and his students) and those who
didn't. However, what all these definitions seem to have in common is
the function of contrast which a phoneme must fill at some abstract
level of structure.

An additional requirement imposed by the American structuralists on
phonemic theory is that (a) given a phonemic representation, there
will be exactly one phonetic utterance which it gives rise to; and (b)
given any utterance, there is exactly one phonemic representation
which underlies it. This is known as the bi-uniqueness condition. It
was shown early on to be an unrealistic ideal, but it remained a
guiding principle of structuralist phonemics. Generative phonology
(including Optimality Theory) abandoned clause (b) but maintained that
given an underlying form, it should be possible to predict its surface
realization. This changes what the phoneme is supposed to do, and thus
what the phoneme is. So if the phoneme is still said to exist at this
point, it is not the same phoneme of the American structuralists. In
fact, in his introductory textbook (the first to be written from a
generative point of view), Ronald Langacker nowhere invokes the term
'phoneme'; he uses 'underlying form' or 'underlying representation'
instead to underscore the fact that we are no longer dealing with
American structuralist phonemes.

As John noted early on in this thread, if the phoneme is dead, it's
dead like Newtonian mechanics is dead. It still provides a good model
for analysis, and I certainly teach something like the phoneme to my


Dirk Elzinga
[log in to unmask]