Ablaut ad nauseum.

[Rankin, Robert L]

> I think we accept that there existed CVCv active verb roots in proto-Siouan, where the lower case v represents an unaccented vowel.  We seem to be finding that all, or almost all, CVCv active verb roots were specifically CVCe.  Some of the daughter languages have secondarily derived forms in which the v is some other vowel than -e, as in Omaha ttą́ąðį, 'run'.  But in proto-Siouan, active verb roots of the form CVCv were in general distinguishable from each other by the initial CVC sequence only, and not by the final unaccented vowel.

The exact status of CVCv roots (not stems) in Proto-Siouan is not certain.  We used Dick Carter's database originally at the '84 workshop at CU, and his database apparently included all the roots he regarded as CVC from his dissertation.  Note that these include a lot of STATIVE verbs too.  A couple of them were included in the cognate sets I listed in an earlier post.  I'll post them again if anybody wants me to.  My main point is that we've learned that the selection of lexical verbs isn't random.  

We have tossed out several possible hypotheses to explain this pattern:

> 1) The unaccented final vowel was a separate phoneme that was an integral part of the root.  The fact that it was always or almost always -e is insignificant, because some vowel had to predominate.

No, it's not possible to establish the facts for integral parts of the ROOT except by further internal reconstruction.  But, YES, an integral part of the STEM AND of the LEXEME, i.e., final -e was present as part of the WORD in ALL of the verbs that show -e, or the regular reflex of *-e, in the cognate sets.  We've both cited any number of examples.  You can't say it "wasn't there" in proto-Siouan unless you want to claim that each of the languages innovated an epenthesis rule independently.  

> 2) The unaccented final vowel was the schwa'ed out reduction of any of the eight possible vowels due to a process that affected active verb roots of CVCv type.  In this model, a prior position of phonemic distinctiveness merged together.  CVCa, CVCe, CVCi, CVCo, CVCu, CVCaN, CVCiN and CVCuN all collapsed into something that we reconstruct as CVCe.

I see what you mean.  I don't know that we have evidence for that, but we DO have evidence that, even in proto-Siouan, active verbs could end in more than just -e, (and also, that CVCv stative verbs could end in unaccented -e).  The regularity of such sound changes would affect all lexemes in the language, so nouns, adverbs, etc. would also all end in -e if unaccented.  If you're dealing with real Lautgesetz, the phonology is affected across the board.  It's only found in particular grammatical or lexical categories if you're dealing with ANALOGY or BORROWING, i.e., so called "Labovian sound change".  So I'd say (2), vowel reduction, is ruled out.

3) The CVCe active verb roots were all underlyingly CVC, but received an epenthetic -e either as a requirement for the release of a final consonant or to fill out syllable structure.  I just don't think we can know this for the proto language

This is technically possible, but the phoneme /-e/ is still there in reconstructions of Proto-Siouan vocabulary.  Given the otherwise completely OPEN syllable structure, I personally wouldn't want to give up that important phonological generalization.  It explains accent in Dakotan but not in the rest of the languages. In some phonological theories, "prediction" of the most common vowel wouldn't be acceptable even if it were the only vowel in that environment.  

> 4) The final -e is a separate morpheme, like Spanish -r used to mark the infinitives of verbs.  In this case, the root itself of these CVC-e verbs is composed of CVC without the -e.

Well, not 'infinitive', but I take your meaning.  We could call it a 'stem formative'.  Personally, I don't think this is tenable with our present knowledge.  If EVERY active verb ended in -e, we could argue this, but many don't.  Bear in mind, though, that putative auxiliary I mentioned in my last post, namely *-re.  IF your -e is a morpheme that follows a root-final consonant, then *-re would be the allomorph that would follow root-final VOWELS.  You could try pursuing that hypothesis.  Personally, I just don't know at present whether, e.g., *riN-re would be a proto-Siouan STEM formed from a ROOT with the shape *riN 'be moving'.  If you believe -e is a stem-forming morpheme, then the next step might be to check out the function(s) of *-re.  I've

> I am open to hypotheses 2, 3 and 4, but I find hypothesis 1 to be unexplanatory.  My understanding is that Bob favors hypothesis 1 and rejects hypotheses 3 and 4.

It looks to me as though we have been lumping the notions ROOT, STEM and WORD (or LEXEME).  These are all distinct concepts.  Root is the smallest; word is potentially the largest.  We've generally tried to reconstruct the largest of these units possible, given sound change regularity, in the CSD database.  

Bob, is this a fair statement of what we agree on concerning this question, and where we differ?

Given my caveats, I think so.  I believe we just believe in different degrees of necessity in "interpretation" of reconstructed vocabulary, i.e., in languages other than Dakotan.

Bob