Re: "The phoneme concept" [was: Re: Claims [was: Re: Drifting phonemes [was: Re: The AmE 'o' sound]]]

From: Greg Lee (greg_at_ling.lll.hawaii.edu)
Date: 11/14/04 

Date: 14 Nov 2004 08:58:54 GMT

In sci.lang Jacques Guy <jguy@alphalink.com.au> wrote:
> Ben Zimmer wrote:
>
> > Jacques Guy wrote:

> > > I have no English example to offer, so I am reduced to turning
> > > to my own language: French. French has long 'a' contrasting
> > > with short 'a'. But no minimal pair I could find.
>
> > The usual English example is the distinction between /N/ and /h/. Even
> > though they're in complementary, not contrastive, distribution (/N/ only
> > occurs after vowels, not before, while /h/ only occurs before vowels,
> > not after), everybody knows that they're distinct phonemes.

> That is a different aspect of the question. In the French case I
> mentioned there is no clear contrastive distribution, yet the two
> cannot be reasonably reduced to a single phoneme, even though
> they are phonetically very close.

> In the English case you mention, [N] and [h] are in mutually exclusive
> distribution and therefore, fulfill the condition for being considered
> as a single phoneme. However, they are too distant phonetically to
> make for credible allophones of the same phoneme.

You don't need to appeal to phonetic distance, since the method
you used to demonstrate the contrast between long and short [a]
in French works here, too. There are no plausible phonological
processes of English that could derive [N] from [h] or [h] from
[N], so this must be a phonemic contrast.

The problem with the phonetic distance criterion is that, in
addition to being rather vague, phonological processes can sometime
produce very different allophones. English [t] becoming glottal
stop before another consonant, e.g.

> Mind you, a uvular approximant is as phonetically different from
> an alveolar trill as you could imagine. Yet both are /r/ in French,
> depending on the region. So...?

Or initial /r/ in Portuguese becoming velar fricative. Or syllable
offset /r/ in English becoming a central glide or disappearing. I'm
sure you can think of many such cases.

There isn't any simple methodological rule of thumb that will work
as well as formulating a phonological theory. 

-- 
Greg Lee <greg@ling.lll.hawaii.edu>

Relevant Pages

  • Re: "The phoneme concept" [was: Re: Claims [was: Re: Drifting phonemes [was: Re: The AmE
    ... "Punctuated equilibrium" was invented by Gould and Eldridge. ... "Phoneme" was not invented by me. ... >> necessary) for establishing a phonemic contrast. ... What concept of phoneme excludes semantics from consideration? ...
    (sci.lang)
  • Phonetic contrast between lenis unvoiced and lenis voiced?
    ... Are there languages where it is the norm* to phonemically contrast ... lenis voiced and lenis unvoiced stop phonemes? ... unvoiced stop phoneme. ... there can be a contrast between the realizations of a voiced stop ...
    (sci.lang)
  • Re: why cant the BBC
    ... What I said was, if they contrast, even if they differ by only 1 Hz, ... Except that a contrast has to be noticeable to be significant. ... How would one acquire a phonemic distinction between auditorily ... French e as a phoneme different from null. ...
    (sci.lang)
Quantcast