LINGUIST List 13.2076

Mon Aug 12 2002

Disc: New: Re Sum: Tense and Lax i

Editor for this issue: Karen Milligan <karenlinguistlist.org>

Directory

1. Mark Jones, Re 13.2053 Tense and Lax i - acoustic data

Re Linguist 13.2053

Dear All,

I'm going to put the cat among the pigeons a bit, especially for those
who have commented on British English. I'm a 30 year old native
speakers of British English. I've moved around a bit within Britain,
but I'd say my accent is fairly standard British English. However, my
/i/ definitely sounds tenser before velar nasals and phonetic engma in
/ink/ words. I don't know how common this is, but if I have it, I'm
willing to bet many other SBE speakers do too.

I've been wondering about a phonetic explanation. One which occurs to
me is to relate this to the position of the tongue root: English tense
and lax vowels have been linked (fairly unsuccessfully) with Advanced
Tongue Root (ATR) vowels in West African languages (Ladefoged and
Maddieson 1996:302-306). However, these data seem to suggest that a
link is plausible, maybe only for some speakers/dialects.

I've done some acoustic measurements of my own speech. In order to get
a lengthy vowel for measuring purposes, I've used utterance final
tokens of 'sin', sing' and 'seen', and 'bid', 'big', 'bead' and the
nonsense word 'beag'.

The measurements show that, for my speech at least, the formants of
/I/ before a velar and /i/ are not identical as predicted, but neither
are the formants of /I/ in 'sin/sing' and 'bid/big'. It appears that
in comparison with alveolar contexts, /I/ before a velar has a lower
F1 and a higher F2 throughout, a more radical F2 movement (average of
318 Hz versus 131 for /I/ + alveolar) and a closer F2 and F3 at all
times, culminating in a 'velar pinch' as one would expect. Both
dynamic and static measurements show a similarity between /Ig/ and
/id, ig/.

Average formant distances (in Hz) are as follows:

 /Ig/ /Id/ /ig, id/
F1-F2 1788 1560 2194
F2-F3 390 687 423

These data for nasals and stops indicate that this is a tongue root
effect, not an effect due to acoustic coupling of the nasal cavities
or interaction between the tongue musculature and the musculature for
velic lowering.

Although the data are for one non-naive speaker, there does appear to
be a difference which makes /I/ before /g, ng/ intermediate between
/i/ and /I/ in terms of these formant data, and it clearly isn't
*just* a co-articulatory effect, but persists throughout the vowel. I
would say that the non-identity of /I/ in 'seen' and 'sing' indicates
that I wasn't prejudiced by the task, but clearly more work, with more
speakers, would be needed to confirm this effect for SBE.

Mark Jones

Department of Linguistics
University of Cambridge