Re: syllable cut prosody [was: Re: Swedish vowel phonemes]

At 7:17 PM -0500 2/12/03, Roger Mills wrote:
>Dirk Elzinga wrote:
>> If consonant length is predictable (gemination occurs to make a light
>syllable heavy under stress), and vowel length is predictable (vowels are
>lengthened to make a light syllable heavy under stress), maybe the real
>distinction is syllable cut. Syllables with loose contact will lengthen the
>vowel under stress, while syllables with close contact will geminate the
>consonant. Syllables which are closed are characterized by close contact in
>any case and can be stressed without any type of lengthening.
>>
>> A pair of words from Norwegian may be helpful in illustrating the
>distinction:
>>
>> /hake/ 'chin' becomes ['ha:.k@] -- loose contact
>> /hake/ 'pickaxe' becomes ['hak.k@] -- close contact
>>
>> Both forms have the same underlying segmental structure; they differ only
>in the syllable cut distinction. With syllable cut prosody, you get to keep
>your small vowel inventory, since lengthening is predictable under stress
>for syllables in loose contact; and you don't need to commit to underlying
>geminates, since gemination is predictable under stress for syllables in
>close contact.
>>
>Admittedly I'm mired in the generative tradition, but where/when would one
>specify that one "hake" has close contact, the other "hake" has loose?
I'm in the same tradition, but I got the Optimality Theory update a few years ago
:-). I assume that close/loose contact would be specified in the UR. I played
around with some representational devices, and decided that a relatively
transparent representation would use the tie bar (which normally indicates
simultaneous articulation or affricates) to represent close contact, and the
absence of the tie bar to indicate loose contact. Alternatively, you could use
an underscore to indicate loose contact, and the absence of underscore to
indicate close contact. It depends on which value of contact you feel is
unmarked (in the Praguean sense; IIRC, Trubetzkoy considered loose contact to
be unmarked).

>Seems to me that's adding another, rather squishy, feature to the ULR, so
>why not just opt for shortness (of V) OR tenseness (of C); or even
>plus/minus TENSE for vowels, with a morpheme structure rule that specifies
>+tense cons. (and vice versa), or a cluster, after -tense V?
I tend to agree. I favor the underlying geminate solution; that way, you can
distinguish between open syllables, in which vowels are lengthened under
stress; and closed syllables, which don't have vowel lengthening. However,
there is a small phonetic literature which shows that the distinction between
short/lax and long/tense vowels in German correlates well with the interval of
time between the end of the vocalic opening gesture to the beginning of the
consonant closing gesture. This interval of time is practically non-existent
for short/lax vowels, but is about 40 ms for long/tense vowels. In other words,
the articulatory opening gesture for tense vowels peaks before the consonantal
closure, while the opening gesture for lax vowels peaks at about the same time
as the consonantal closing gesture.

So there may be some phonetic correlate of close/loose contact after all.

Dirk
--
Dirk Elzinga                                               Dirk_Elzinga@byu.edu

"It is important not to let one's aesthetics interfere with the appreciation of
fact." - Stephen Anderson