Funny W

[Koontz John E]

On Wed, 25 Oct 2006, Rory M Larson wrote:
> Is it *W that is involved in the Dhegihan positional ama'/aba'/apa' ?
> I.e.
>
>   Dh. *aWa' => OP ama', Ks aba', Os apa'   ?

Yes, though this set is restricted to Dhegiha, and involves an unusual
position (medial) and context (enclitic).  Compare also the OP =ma, Os
=pa, etc., set, in which the *W is initial.  Both these sets are
positional articles (but not positional in nature) with different plural,
non-focus readings.

> Would a single phoneme, if one could be found, be acceptable on other
> grounds as a reconstruction for the consonant in the above set?

Certainly, and presumably the same would apply to the various other *W
sets.  We don't need to have one eplanation for all *W sets.  For example,
it is clear that many Dh *R sets reflect *pr-, e.g., OP nu(ga) 'male
(animal)', Os to(ka), but Teton bloka'.  However, continuing with *R,
where the evidence is clearer, cases like OP nez^e 'urine', Te lez^e', ...
or OP negi' 'mother's brother', Te lek(s^i(t)), show no evidence of *pr in
Mississippi Valley.  Because some *R sets in some subgroups seem to
reflect clusters, it's tempting to assume that all *R are clusters, and I
think this explains T. Kaufman's *?r, which is essentially a cluster with
a stealth initial element.  My *W and *R are just placeholders saying "I
don't know what it is, but it's like a w, etc."

Finding a single phoneme to propose instead of *?r and *?w, etc., is
difficult only because the territory is crowded.  The only options I see
are m and n, though we have some m and n that behave like w and r before
nasal vowels that make better candidates and we generally tend to expect m
and n to be conditioned allophones of *w and *r.  If we don't rule out
clusters, then mb and nd are also possibilities.