LINGUIST List 10.1885

Wed Dec 8 1999

Disc: What Exactly Are Allophones?

Editor for this issue: Karen Milligan <>


  1. Jorge Guitart, Re: 10.1879, Disc: What Exactly Are Allophones?

Message 1: Re: 10.1879, Disc: What Exactly Are Allophones?

Date: Tue, 7 Dec 1999 14:14:57 -0500 (EST)
From: Jorge Guitart <>
Subject: Re: 10.1879, Disc: What Exactly Are Allophones?

M. Hubey wrote: 
 Those who bring
> up
> the concept that allophones, phones, phonemes are "abstract"
> things
> really mean that they are "sets" (more or less). The reason for
> this,
> we are told, is that nobody hears a phoneme for it is an abstract
> thing
> and we only hear specific instantiations (sample functions in
> probability
> theory, or "tokens" in computational linguistics and computer
> science).

Jorge Guitart responds

I think that a phoneme is abstract but I surely do not mean that it is a
set in the sense of set that you are using. So you can exclude
me from the group of those who mean that when they say that a phoneme is
abstract. A phoneme is a unitary thing, yes, a thing, though an abstract
thing, a mental unit if you wish, but it is a unit and I would suspect
that it is represented as a unit in the brain. I would say that the word
CAT is composed of three discrete segments in the mind (alphabetic writing
is a reflection of that, by the way). If I replace the first unitary thing
by another, say by P, I get a new sequence that does not mean the same as
the first. 

A phoneme is not a set of its allophones.It is not, as some structuralists
would have it, 'a class of sounds'. Allophone A of
phoneme P" is a REPRESENTATION of P. 

To me, to say 'a certain allophone of a certain phoneme' is more like
saying 'a certain photograph of a certain house' than like saying 'a
certain photograph of a certain type' (a daguerrotype,say). I think the
latter is the kind of analogy that you have in mind when you say that
those who say that a phoneme is abstract are saying that a phoneme is a
Once more with feeling: a phoneme is an invariant pychological unit
(invariant in perception: it is always what you think you hear! (You heard
Empire State Building even though your interlocutor said Empire Stape
Building.) It may have infinite representations, yes, infinite--no two
tokens of even the same allophone are exactly alike.

Now, everyone is free to have her or his own definitions of phoneme and
allophone but I would like to say that mine have always worked with smart
students or at least with those that do not have a mental block against
analyses that posit entities that cannot be inspected directly.

My best to all

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