On 8/5/07, T. A. McLeay <[log in to unmask]> wrote:
> Benct Philip Jonsson wrote:
> > caeruleancentaur skrev:
> >>> Benct Philip Jonsson <conlang@...> wrote:
> >>> Of course people do that all the time in normal allegro
> >>> speech, in all languages.
> >> I'm sure that's meant as a generalization, but just to set the record
> >> straight, I always pronounce the /t/ in "crafts" and in "craftsman" no
> >> matter how rapidly I'm speaking.  I just can't imagine saying "crafs."
> >
> > The thing is that people *think* they are pronouncing
> > 'everything' all the time, because our brains are good
> > at filling in the 'missing' parts.  One thing that proves
> > this is not so is what children produce extrapolating from
> > adults' input.  There is no need to get prescriptivist
> > or normativist about the way human speech and speech
> > perception works. Pronouncing [kr&fs] isn't sloppy in
> > any way, it's just what /kr&fts/ becomes at normal
> > speech rates due to the motoric limitations of the
> > speech organs.
> The sort of person who's interested in this might think it's worth
> noting that for me (and I basically categorically lack any "t"), the
> vowel isn't /&/, it's /a:/, which would mean /kra:fts/ has five moras.

It would only have five moras if you insisted on counting each
consonant. To my knowledge, this is never done in moraic phonology.
Rather, it would have at most three moras (for theories which accept
the legality of trimoraic syllables), and probably only two, since the
basic distinction that seems to matter is that between light and heavy

It is true that when the vowel is pronounced long and each consonant
is articulated separately that the raw length of the syllable can be
impressive, but moras aren't about phonetic duration; milliseconds


[1] For languages which have a three-way distinction
(light-heavy-superheavy--Estonian is a famous example), a trimoraic
syllable may seem to be a good idea. But it's surprising how
vigorously this kind of analysis is resisted in the literature.