Re: THEORY: Ray on ambisyllabicity

On Tue, 17 Oct 2000, John Cowan wrote:

> dirk elzinga wrote:
>
> > Up until now, I have been reluctant to admit the possibility
> > of ambisyllabicity *as a possible structure*, regardless of the
> > dialect under consideration. This reluctance is admittedly based on
> > theory-internal arguments, but there are other analyses available
> > which account for the "ambisyllabic" facts without invoking
> > ambisyllabicity.
>
> It seems to me that there isn't any possible ambisyllabicity that can't
> be accounted for as a covert gemination.  If [X] (some random sound)
> seems to be treated as part of two adjacent syllables, you can postulate
> an underlying [XX] in every case.  And vice versa.
>
> In short, ambisyllabicity and covert gemination are "equivalent" theories:
> either of them can account for every possible datum.
Not at all. Consider the example given by And:

>  hoe          [h@w]
>  holy         [h@wli]   (/l/ only in 2nd syllable)
>  whole, hole  [hOw]
>  wholly       [hOwli]   (/l/ ambisyllabic, triggering vowel
>                          allophone in 1st syllable)
If the [w] of the first pair is part of the nucleus, and the [w] of
the second pair is the allophone of /l/ which occurs in the coda, then
the form 'wholly' shows that /l/ can be "split" by an allophonic
process into a part which undergoes it (yielding [w] in the coda) and
a part which does not (leaving [l] in the following onset). The
simplest explanation is that it was geminate to begin with. If the /l/
was truly ambisyllabic (and therefore a single segment occupying two
places) one would not expect this kind of differentiated behavior.
(This argument of course hinges on my interpretation of the data.)

Another way to distinguish an ambisyllabic consonant from a geminate
is to see if processes of lenition can apply. In English, intervocalic
/t/ is often taken to be ambisyllabic in words like 'atom' (i.e.,
following a stressed lax vowel), and these /t/s (among others) undergo
flapping. Geminates are resistant to lenition processes (a property
dubbed inalterability).

So I don't think that ambisyllabicity and covert gemination are
equivalent theories, since there are processes which can tease apart
the difference.

Dirk

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