testing

[Rankin, Robert L]

> Looking at the data you present, I would be inclined to read these stems as phonemically CVC in proto-Siouan, but as operating within a phonological system that required a small, meaningless, schwa-like vocalization after a final consonant to clarify that final sound.

But your “schwa-like” epenthetic vowel is [–e] in 12 or 13 languages in several subgroups over a 2000 mile stretch.  This is the equivalent of reconstructing *-e in these cases.  You’re just reconstructing *-e as a “rule” or process instead of as an “item”.  But with a c. 3000 time depth, we don’t have any way to distinguish the two equivalent “solutions”, and the phonological result is the same either way.

> . . . it would stay consonant final, as in Winnebago or Mandan. 

The so-called “consonant finals” in Mandan are not real in the sense that they are in Winnebago.  A final –e in these stems is actually pronounced.  They seem to be a creation of Bob Hollow, who tried the “all final –e in Mandan are epenthetic” solution in his dissertation.  He fell into the trap of the Dakotacentric “consonant-final stems” because he couldn’t hear the long/short vowel distinction in Mandan.  Carter and Mixco cleared this up.

This is another type of analysis that I distrust.  What you and Hollow are saying, in effect, is that all short unaccented vowels can occur word-finally except the most common, namely, -e.  And for some unfathomable reason, short unaccented -e alone can’t.  This trick was toyed with in the ’70s as a means of creating additional “economy”.  But it does so at the expense of badly skewing the vowel distribution and basic syllable structure.  Theoretically, of course, in ANY language with a requirement of open syllable structure, it is, in fact, possible to “predict” the statistically most common vowel syllable-finally.  But this sort of parsimony has generally been considered spurious.  

> From what work I have done with Omaha, I think these final -e sounds receive much less stress than previous vowels in the stem, and the speakers sometimes seem a little ambivalent about whether they should be pronounced -a or -e.  

I respectfully doubt that this would true for Omaha-dominant speakers.  Speakers can normally hear/produce phonemic distinctions 100 times out of 100.  I can see English-dominant speakers producing schwas and the like.  But I have to say I didn’t get that sort of doubt from Ponca, Osage and Kaw speakers.  

> When I try to get them to choose one, I can usually make them agree that it's -e, but perhaps I'm the one imposing something on the language that isn't actually there.  

Given the cognate sets, plus my limited experience with Omaha, I think your hearing is just fine.  So, in summary, I believe that, (a) whether you reconstruct *e- to Proto-Siouan as a “rule” or an “item”, either way you’re reconstructing *-e, and (b) the “epenthesis” solution creates a lopsided short vowel distribution and an “economy” (feature saving) that, although technically possible, is not genuine for most phonologists.

And if vowel length is properly perceived and recorded, the CVC stems or roots are not needed in order to establish accentual pattern.  Only in Dakota, which has apparently lost the length distinction, is it a useful analysis.  And this brings me to a basic problem in Siouan linguistics, whether historical or synchronic:  Because Dakota was the earliest and best documented Siouan language, it has served as a model for subsequent studies of the other Siouan languages.  This is understandable, but it has held the discipline back in certain ways because Dakota alone lost vowel length and has simplified in many ways grammatically.  

Bob