Hi Bob:

  If what you mean by barred I vs barred U is a similar articulation except 
for lip rounding in U, then I'd go for lowered formants for the U case. 
Look particularly (but not necessarily only) at F2 and F3. The extent to 
which F3 is affected, relative to the extent to which F2 is affected, 
should depend on how far back the main constriction really is, and 
probably also on whether there's much lowering associated with either. If 
the constriction is quite front, and given the same cross-sectional area 
at the constriction for both vowels, then F3 should be lowered more than 
F2 when the lips are rounded. When the constriction is farther back, F2 
should be lowered more than F3.
    I don;t know what happens in the middle range, and I don't know whether 
there's much variation in the range of constriction areas used in the 
Scottish sounds you're interested in. Some of the early work on, say, 
Swedish vowels might help guide you (eg Fant shows quite a variety of 
spectra, e.g. in Malmberg, but I haven;t looked at them before writing 
this.)
   You can easily check by measuring a few yourself: keep the constriction 
location and area the same as far as possible, and phonate steadily as you 
change the lip rounding. If you hold one steady state, then change the 
lips reasonably slowly to the other, you'll see the formants shifting 
slowly during the lip rounding, and be convinced. Then try with other 
constriction locations, and other jaw/tongue heights (i.e. constriction 
cross-sectional areas!).

  However, if you are working from other people's transciptions, or from 
speech spoken in natural contexts rather than citation form in neutral 
contexts, I wouldn't necessarily make the above assumptions. First, as you 
know, lip rounding can vary enormously, with coarticulation due to local 
phonetic context, with speech rate, and various so-called paralinguistic 
things (smiling, whether you're talking to a baby or a pet), some of which 
may be communicatively significant in more clearly linguistic ways, as 
demonstrated by variationists. Second, depending on the orientation of the 
transcriber, transcriptions may use the nearest 'phonemic' equivalent. 
Thus if the word is 'good' the sound resulting from a particular 
articulation may be transcribed with barred U even if it was produced 
without lip rounding and in another word might have been transcribed as 
barred I. I know the two sound different, but a person not focussing on 
that particular distinction may not capture it in a transcription.

hope this helps

Sarah





On Mon, 19 Jun 2006, D R Ladd wrote:

> Dear Phonet people
>
> So far I've had two replies to my query - from Jim Scobbie and Jane 
> Stuart-Smith - and both included caveats along the lines of this from Jim:
>
>> My guess is that this distinction between estuary and scottish versions of
>> the vowel in "good" is too complex for an MSc to make a lot of headway
>> unless some basic background work is done. The vowel will be very variable
>> in both accent types, and though there still may be a simple acoustic
>> relationship between formants and perceived Scottishness that can be found,
>> what would that tell you, if you find it?
>
> So I should clarify, in case anyone else wants to respond, that I am NOT 
> interested in directly comparing Scottish and Estuary English renditions of 
> _good_.  I was only using those as examples of the kind of vowels I'm 
> interested in.
>
> To answer Jim's further question: what I'm aiming at is, for Scottish 
> English, something like Labov's index of "Philadelphian-ness" based on the 
> acoustic realization of specific vowel phonemes. The basic idea is to explore 
> the Labovian distinction between lexical set membership and phonetic 
> realization (in order to have authentically Philadelphian lexical sets for 
> _bad, glad, sad_ etc., you have to have Philadelphia-native parents, but in 
> order to sound Philadelphian you just have to modify the phonetic realization 
> of _caught_).  We want to compare Scottish speakers' lexical set intuitions 
> with their acoustic realizations of certain phonemes.
>
> I think that the frontness of /u/ would be a useful thing to measure in this 
> theoretical context (along with the backness of /a/, the shortness of /i/ and 
> /u/ before nasals and voiced stops, etc.), and what I was asking about in my 
> original query was simply how to distinguish, in acoustic space, between 
> "barred U" (which should sound "Scottish") and "barred I" (which shouldn't). 
> It was a very practical question, and I'm sorry I gummed it up by dragging 
> Estuary English into it.  Can anyone help me now?
>
> Thanks
>
> Bob Ladd
>
>
> PS: Jim adds
>
>> And what sort of judgement task is
>> the right one for this sort of study? I think this needs to be clear to
>> start with, and think that a strong methodological and descriptive angle
>> would be really useful for other researchers.
>
> If we find anything useful out, I'll be sure to let everyone know later in 
> the summer.
>

______________________________________________________
  Sarah Hawkins, Professor of Phonetic Sciences, Dept. of Linguistics
  University of Cambridge, Sidgwick Avenue, Cambridge CB3 9DA, U.K.
  Phone: +44 1223 33 50 52          http://kiri.ling.cam.ac.uk/sarah