Re: CHAT: Importance of stress

Nik Taylor wrote:

>> (1)  VCV is almost always syllabified as V.CV, not VC.V
>
>Unless there's some resyllabification rule, like English /'h&p.i/ (which
>can be shown by the fact that it's ['h&pi] and not [h&:p_hi])
See my comment below...

>> For example, when asked to break up the word "happy" [haepi] into
>> syllables, many native speakers will hesitate between [hae.pi] and
>> [haep.i].
>
>But it's pronounced as the second, as shown by the allophones.
I generally pronounce it the first way - or at least, I find the
first syllabification *much* more intuitive.

>Syllable-initial voiceless stops are aspirated, and vowels followed by
>voiceless obstruents are shorter.  /h&.pi/ would be [h&:p_hi], while
>/h&p.i/ is [h&pi], the way it actually is.
Syllable-initial voiceless stops only aspirate when they are
word-initial and/or followed by a stressed vowel.  And anyway,
if the /p/ were entirely syllable-final, then it would be unreleased,
as it is in "capture", and for me it's definitely not unreleased.

I think the source of the ambisyllabicity effect here has to do with
conflicting phonotactic constraints in English:  On the one hand, there's
a constraint against having a vowel-initial syllable following a
consonant-final syllable, which forces the [h&.pi] analysis.  On
the other hand, there's a constraint against syllables which end in
short lax vowel, which forces the [h&p.i] analysis.  My guess is that
for some speakers (like me), these constraints are equally ranked,
and so the conflict between them is resolved by treating the
consonant as ambisyllabic.  For other speakers (like you, I would
suppose), one constraint ranks more highly than the other, and
so there's no impression of ambisyllabicity.

Matt.