schwa-raising - crossing the phonemic quantum gap

[Mark J. Jones]

----------------------------Original message----------------------------
Dear all,

Cecil Ward comments that:

> It really does seem that these phenomena are not simply (or solely)
> "phonetic environment-governed".

and gives the example of the pronunciation [fælkIn] ="falcon" on the BBC.
The claim for a phonological process is apparently supported by the comment
that the schwa vowel here is "not in contact with a syllable containing /i/
or /j/", i.e. this is the conditioning environment. Thereafter, the speaker
apparently said [falk at n].

I would like to make clear that my own instrumental study identified
surrounding coronal consonants (not therefore /i/ or /j/) as the phonetic
conditioning factor, in which case 'falcon' could qualify, having a final
/n/. There will also be some nasalisation of the vowel, which could also
contribute to differences.

There may be other phonetic factors which contribute. I determined a lower
jaw height to be crucial in 'raising' schwa. When speaking loudly there is
a tendency to open the mouth more. This can have the effect of lowering the
vowels, especially the central ones. When teaching a practical phonetics
class demonstrating the IPA transcription of vowels, it is hard to get [@]
and not RP /^/ (IPA turned A) if projecting the voice in a large room.
Cecil Ward's speaker may therefore have varied this aspect of speech
(particularly if involved in a 'performance' of some sort for broadcast),
but the very fact that the same speaker said [@] thereafter suggests that
for them, this vowel is underlyingly /@/.

I've already alluded to the possibility of there being variation between
speakers here, and hinted that as the unstressed vowels are so short, their
quality may be hard to determine precisely. This goes as much for
perception as for instrumental analysis. Vowel perception is a tricky
business at the best of times, with issues like normalisation across
speakers and environments, and the utility (or not) of coarticulatory
features to surrounding consonants playing a role.

At the end of the day these impressionistic comments are all very well, but
as anyone who has done any acoustic analysis will confirm, the gulf between
what we think we hear and what is there in the signal can be immense.
Speech perception is for this very reason a complex subject. By this, I do
not mean to suggest that Bob Rankin (who commented on [I]-like realisations
previously) and Cecil Ward are wrong, merely that they may have, under
certain conditions, attributed a vowel showing a particular phonetic effect
to the vowel /I/. This is arguably how sound change happens, and so there
is nothing surprising in it, but I feel, as I have said before, that a
larger instrumental production study, possibly involving perceptual
identification and discrimination experiments on vowels, is the best way to
resolve this matter.

I end by saying once again that the only (relatively) objective data on
this issue so far, on my own speech, shows that 'raised schwa' is not
phoentically equivalent to /I/, but is probably best transcribed IPA
reverse E (secondary cardinal 19).

Mark

Mark Jones
Department of Linguistics
University of Cambridge