Re: me again

Thomas R. Wier sikyal:

> I thought so. It is not in fact unprecedented for what
> are normally considered consonant clusters to be considered
> unit phonemes (Georgian's so-called "harmonic clusters" like
> /t_s'q_X'/ and /bd/ pattern this way), but they are *vanishingly*
> rare.  Such clusters apparently always agree in laryngeal
> features, like voicing or glottalization. So, your /ps/ and
> /ks/ pass the test, even if I doubt very much whether any
> other conlang has them!
Think again.  Hiksilipsi, my oft-revised tonal language, has /ps/ and /ks/
as unit phonemes, along with the nasal occlusives /mp/ and /Nk/.  But
there is no /ts/ or /nt/, as both of those have already been assimilated
away.

The analysis of these as single phonemes is motivated by the fact that
Hiksilipsi allows no word-final consonants, which implies no consonants in
coda positions--yet words such as /apsu/ are perfectly fine.  The
syllabification must therefore be /a.psu/.  Looking at other words reveals
that the set of allowable onset clusters is quite limited--in fact,
restricted to just four phonetic clusters: [ps ks mp Nk].  The best
analysis, then, and the one I support, is to regard these as unit
phonemes, and to say that Hiksilipsi has a strict prohibition against
onset clusters and coda consonants.

Also, aren't the nasal occlusives [mb nd Ng] fairly common as unit
phonemes?  I believe that they occur in Kiswahili, where they even
constrast with NC clusters, e.g. a hypothetical [a.mbu] could contrast
with [am.bu].


Jesse S. Bangs jaspax@u.washington.edu

"If you look at a thing nine hundred and ninety-nine times, you are
perfectly safe; if you look at it the thousandth time, you are in
frightful danger of seeing it for the first time."
--G.K. Chesterton

Replies