Funny W

[Rankin, Robert L]

> If we don't reconstruct *m, *n in Proto-Siouan, how about a reconstruction of *W = *pw, *R = *tr ?  Would this slot be open?

"Slots" or "pigeon holes" would need to be open, of course, but there is a bigger problem.  The reconstructions have to be made to jibe with what we know of the morphology.  And there just aren't any prefixes that undergo syncope with the shape *pV or *tV.  P and t are just not candidates for the ancestral form of bl-, br-, bdh-, etc. clusters (there aren't viable reconstructions with the dental at all.  'Cat' and 'squash', the only remote possibilities, are only found in some languages and likely diffused).  The only good candidate is *w, because we have *wa- 'inanim. absolutive', wi- 'animate absolutive', wa- '1st sg. actor', etc., all of which undergo syncope.  
 
So we know of prefixal morphemes or portmanteaux with the shapes /wi, we, wa, wo, bu/ but not /pi, pe/ and the instrumentals /pa, po, pu/ don't undergo syncope or had different sources (with /W/).  * /ti, te, ta, to, tu/ prefixes are equally unknown.  
 
Perhaps worst of all, the mysterious /W/ phonemes virtually all tend to fit with the 'absolutive' semantics.  The only exception I can think of is our Dhegiha *-aWa, and it is restricted to a single subgroup and is of unknown origin.  I would agree that it, and -akha, were bi-morphemic, but which two morphemes?  The -ha of -akha is likely a variant of the same -he that we get in -the, -khe, athaNhe, niNkhe, niNkha, etc.  But what about the rest?
 
This would make a nice paper for whoever sorts it out.    :-)
 
Bob