Re: Igassik Morphophonology

On Mon, 8 Jan 2001, jesse stephen bangs wrote:

> dirk elzinga sikayal:
>
> > > I suppose `syllables' are something not so easy to define then.
> > > Listening to people from Georgia, I decided that the two v, the r and
> > > the i make syllables.  So I counted four, thinking the second was
> > > stressed: v-'prts-kv-ni.
> > >
> > > But I don't know whether that's accepted theory.  It was what I heard
> > > and how I could pronounce the word so that people said: `Yes,
> > > right.'. :-)
> >
> > This is an interesting question, and one which different natural
> > languages handle differently. In Berber, for example, there are
> > no restrictions on letting obstruents (stops, fricatives,
> > affricates) be syllabic (or at least, very few restrictions), so
> > syllable structure is rather simple. Here are some examples
> > (syllable peaks are in capitals):
> >
> >         .rA.tK.tI.  'she will remember'
> >         .bD.dL.     'exchange!'
> >         .tF.tKt.    'you suffered a sprain'
> >         .tzMt.      'it (f) is stifling'
> >         .tR.gLt.    'you have locked'
> >
> >         etc
>
> I didn't think that stops could be syllable peaks.  It seems impossible to
> me, since it's impossible to prolong the articulation of a stop and still
> have it be stop-like.  Unless, of course, the symbols above are actually
> fricatives.
Nope, they're actually stops. While my knowledge of Berber
phonology is pretty rudimentary, I haven't yet come across a
long syllabic stop, so there are likely other syllabification
constraints operating that I haven't mentioned. But all of the
sources on Berber phonology are unanimous in stating that stops
can be syllable peaks in the language.

About three years ago, I started sketching a conlang with a
Berber-like phonology. I imposed the restriction that stops
could not be syllable peaks (unlike Berber), and the low vowel
[a] could not be a syllable margin (like Berber). The inventory
was:

        p   t   c   k   kw
        f   s   S   x   xw
        m   n
            l r i       u
                    a

Morphologically it was an unremarkable effort (except for an
agreement system that I thought was pretty neat), and I didn't
get far enough into it to discover what the syntax was like. I
did work on numerals, though, and I developed a dual counting
system--an "old" system which was kind of base-12, and a "new"
system which was base-10. The base-12 system had alternatives
for many of the numbers below 12; they could be expressed as an
integer or as a fraction of 12. Thus, 'three' or '1/4 twelve'.

The phonology was fun to work out, but the rest was forgettable.

> > (snip fascinating and impronounceable examples)
> >
> > My impression from the literature is that syllabification in
> > Georgian is actually more like Bella Coola than like Berber;
> > i.e., consonants pile up into clusters to avoid being made
> > syllable nuclei. So the form you cite above would be a single
> > phonological syllable, since there is only one segment which is
> > eligible to be a syllable nucleus (/i/). How it is actually
> > articulated is another matter; even the most nimble tongue will
> > have transitional spaces between consonants which will sound
> > like vowels.
>
> Is the idea of "phonological syllable" even valid?  I've always thought of
> syllables as a feature of phonetics and articulatory factors.
The examples from Bella Coola show that phonological syllables
are essential to the description of the language. How else do
you explain the restrictions evident in the reduplication
patterns? At least that was Bruce Bagemihl's argument (the
source of the Bella Coola examples--see his '91 article in
Linguistic Inquiry for all of the gory details), and I find it
convincing in this context.

I should mention that there is at least one serious phonological
theory (Government Phonology) which does without the notion of
'syllable'. I have to confess that I don't know much about GP,
though; And knows more about it than I do.

Dirk

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