A very good question, Peter. Handbooks all
too often ignore such problems.
There is a similat phenomenon in Japanese:
the high vowels /i/ and /u/ may be devoiced and virtually lost in casual speech
between voiceless obstruents, and sometimes also word-finally after a voiceless
obstruent. The point is that the phenomenon is a predictable fast-speech process
and that the "stray onsets" left after vowel weakening are still perceived as
"moraic" and as constituting separate syllables by the Japanese (this is why
they perceive English "sport" as a word of three syllables and four moras;
they wouldn't do so if Japanese had different syllabification principles). In
slow and careful pronunciation the vowels demanded by the preferred syllable
structure are all realised; [spoot] is impossible as a slow-speech pronunciation
in Japanese.
I know precious little about Maori, but I
suspect the situation is similar there. A constraint that requires the reduction
and/or deletion of unstressed vowels (in casual speech?) is in conflict with the
syllable-structure constraints and with the shape of the underlying form. The
conflict is resolved in a language-specific way. Let's imagine that the
underlying form of "Christ" in Maori is /kraiti/ rather than /karaiti/. If the
"no-complex-onset" constraint is more important ("higher-ranking") that the
demand that the lexical and surface segments should correspond in one-to-one
fashion, [karai-] or [k@...] will be preferred to [krai-] as a phonetic
realisation.
Of course once final-vowel deletion becomes
possible as an optional surface process, this opens the way for the eventual
elimination of such vowels from lexical representations and for the wholesale
reorganisation of the syllable-structure system. Polish allows
monstrous consonant sequences now, but at the beginning of its
career as a distinct Slavic dialect it was still an open-syllable language
allowing only fairly simple onsets:
PSl. *drU.gno~.ti 'to jerk' (3 syllables)
> Polish drgna,c' [drgnon'ts'] (1 syllable)
English-speakers often perceive (and
imitate) foreign words like Gdansk or Knesset as containing an extra syllable.
Why? Because the English constraints on syllable structure intervene, thus
manifesting their reality. Even if the underlying form is /gdansk/,
surface-structure constraints demand vowel epenthesis to avoid initial
[gd-].
The interaction of two types of conflicting
constraints -- the structural ones that impose demands concerning the
surface form, and the "faithfulness" constraints, which require that the surface
form should map the underlying form as closely as possible -- is what
"optimality theory" is all about. The theory has been particularly successful in
explaining syllable structure. If you're interested in it, have a look at John
Hutton's website (I posted the reference earlier today), or try this
book:
Kager, René. 1999. Optimality Theory (the
Cambridge Textbooks in Linguistics series). Cambridge: CUP.
----- Original Message -----
Sent: Sunday, November 26, 2000 8:03 PM
Subject: Re: [tied] Consonant-verb regularities
We may learn the rules of syllables in languages such as
Japanese, but how
well are these rules expressed in actual speech?
My
question comes from my experience of Maori, where the syllable rules
demand
open syllables in all cases, and the literature refuses to
acknowledge
anything but open syllabes. But in actual speech, one often
hears
closed syllables. One well known example is the name
Karaiti
(Christ), pronounced /karait'/ or even /k@...'/.
Of course
this /-t'/ can be analysed as a surface representation of an
underlying final
/-ti/, but I think that misses the point.
How far do syllable "rules" in
other languages actually represent surface
speech?
Peter