Re: THEORY: vocalic h/voiceless vowels

On Mon, 7 Feb 2000, Ed Heil wrote:

> Raymond Brown wrote:
> > >I think that "syllabic [h]" would be a fair, if somewhat too vague,
> > >description of the "unvoiced vowels" one sometimes finds in American
> > >Indian languages.  (One would also have to specify the quality of the
> > >vowel to make the description at all adequate.)
> >
> > Do these unvoiced vowels act as centers of syllables?  If they do,
> > presumably they have some vocalic coloring (otherwise there'll be only one
> > unvoiced vowel) which seems to me to be a little more than merely "syllabic
> > [h]"
>
> Yes, they serve as centers of syllables.  I don't know my IPA ASCII
> well enough to transliterate the examples I have found in the
> "voiceless vowels" section of the wonderful "IPA Help" Windows program
> (does anybody still have the URL for the Web/RealAudio version?) but
> the examples they give, from Comanche, Enga, Cuaiquer, and Malayo,
> include voiceless turned omega, voiceless o, voiceless u, voiceless
> barred i and voiceless i, the last two in minimal contrast in a single
> language (Malayo).
>
> They do sound like vocalic h, but with appropriate vowel coloring
> (which is very difficult for me to discern, but there.  Perhaps it's
> mostly there in its effect on surrounding segments, but it's there).
Having done extensive fieldwork on a language which has
voiceless vowels (Gosiute Shoshone, closely related to
Comanche), I can say that the quality and discernability of a
voiceless vowel have everything to do with the quality of the
voiced vowel it is derived from. The Gosiute vowel inventory is
(where <y> represents a high central unrounded vowel):

        i       y       u
        e               o
                a

All of these vowels except for [e] can appear as voiceless,
though voiceless [o] is rare. (There is a reason for the lack of
voiceless [e], but that takes us too far afield.) I will
represent voiceless versions of vowels using the corresponding
capital letter; thus, voiceless [i] = [I].

The most easily discernable voiceless vowel is [A]; there is
clearly a vocalic nucleus, and its quality is easily
recoverable. A "hierarchy of discernability" can be constructed,
which runs [A] > ([O] >) [U] > [I] > [Y]. The high vowel
qualities are less distinct than the non-high vowel qualities,
with [y] being very hard to hear. Also, the high vowels are not
as clearly heard as syllabic nuclei. In at least one of my
consultants' speech, etymological [Y] has been deleted
word-finally following a voiceless stop, leaving only
aspiration, and [I] is often apparent only in its fronting or
palatalizing effect on a following consonant.

In my dissertation, I analyzed voiceless vowels as being the
result of assimilation of a following [+spread glottis]
feature, which is realized in isolation as [h]. The assimilation
of [+spread glottis] goes both directions; it devoices a
preceding vowel, and it fricates a following voiceless stop.
Thus, /haintsyh-pai/ 'I have a friend' becomes [haintSYPai] with
voiceless [Y] and a voiceless bilabial fricative [P] both being
the result of the assimilatory power of /h/ = [+spread glottis].

Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu