LINGUIST List 10.1976

Mon Dec 20 1999

Disc: What Exactly Are Allophones?/Last posting

Editor for this issue: Karen Milligan <>


  1. Dan Moonhawk Alford, Re: 10.1947, Disc: What Exactly Are Allophones?

Message 1: Re: 10.1947, Disc: What Exactly Are Allophones?

Date: Fri, 17 Dec 1999 11:18:11 -0800 (PST)
From: Dan Moonhawk Alford <>
Subject: Re: 10.1947, Disc: What Exactly Are Allophones?

On Tue, 14 Dec 1999 21:51:48 -0100 (GMT), Celso Alvarez Caccamo
<> wrote:
> Dan Moonhawk writes,
> > "Allo-" only works *between* levels.
> I disagree, and that's the root of our overall disagreement. I am aware
> that the standard description is "[X] is an allophone of /y/", meaning
> 'one among several types of realizations of /y/'. However, "allophone"
> literally means nothing but 'another, a different phone', and that's what
> I'm trying to focus on. 

I believe the technical term for another, a different phone is "a
different phone". 

Okay, let's see if we have or do not have agreement on this: If you are
doing straight phonetics, translating (impressions of) sounds into some
form of IPA, and discussing it with someone else, would the term
"allophone" ever come up, unless you changed levels and began talking
about phonemes?

For me and others, it would/could never come up when talking straight
phonetics -- not as "another, a different phone". It only occurs when we
begin looking vertically, "down" from phoneme to phone, never merely
horizontally, phone to phone. I.e., only between levels. Then the
"variant" meaning comes to bear, meaning more than one typical

> Of course, we can say "an allophone OF /Y/", meaning 'a different
> phone OF phoneme /Y/',

No, I can't. Relationship is about sameness, and "allo-" is a relationship
term, so the word "different" above causes me cognitive dissonance. 
"Variant" works better for me, which then suggests others as well in this
particular relationship with a phone.

> but this pressuposes the existence of a previous phone from which
> another one is different.


> I believe that in the coinage of the term "allophone" initially
> underlied the notion that each phoneme has one canonical type of
> realization (a representative phone, e.g. [p] as canonical of /p/)
> and, therefore, a variant type of realization should be regarded as
> 'another phone'. (At the level of abstraction I'm working with to
> babble about this, a phone is a type of sounds, not each of the
> tokens).

So you've moved phones to the mental level instead of the physical
level. Is that a common view these days, or are "phones" more generally
seen by the profession as speech sounds?

> But the (vertical?) relation between a phoneme and a phone is actually
> one of realization-representation; whereas it is the (horizontal)
> relation between phone-types which entails otherness, difference -- it
> is an allo-phonic relation. Strictly speaking, then, a given segment
> [X] is a phone OF /y/, which may be allophonic TO (different from)
> another phone [Z] OF /y/.

WHAT IF we're both right? What if it is a vertical term that needs
horizontal difference in order to come into play?

> All I'm trying to say in my limited English is that the use of the
> standard description "[X] is an allophone OF /y/" shouldn't preclude
> us from understanding that "allo" pressuposes difference between
> comparable objects (phones), 


> and that this difference is THE allophonic relation between phones of
> a given phoneme.

OR it's more complex -- it's vertical/horizontal simultaneously.

> If we replace "allophones" with "complementarily distributed phones"
> this relation becomes clear: allophones are complementarily
> distributed to each other.

... in relation to a particular phoneme. 

Beyond all this, of course -- or perhaps underneath it -- we can find a
professional pandora's box of theoretical positions regarding the
phoneme, which we have taken as a given in these discussions.

As one (anonymous) linguist pointed out to me, are the different
responders to this allophone thread thinking of "phoneme" as a
psychological reality, as a physical reality such as a class of sounds, or
as a Trubetskovian functional unit?

Perhaps we've ridden this pony far enough.

warm regards, moonhawk
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue