Ablaut et al

[jhobartkyle at GMAIL.COM]

Sorry to jump in late on this, but I think Bob is correct in that Dakota is  
the 'odd man out' here in that Dakota phonology can be analyzed as having  
epenthesis on some roots. Boas & Deloria were the first (as far as I know)  
to analyze Dakota as having CVC and (CV)CV roots. It turns out that the CVC  
roots will surface with an epenthetic vowel but stress falls on the first  
syllable while bi-syllabic CVCV roots take the stress on the second  
syllable (as do most Dakota words). The two root types also differ in how  
they reduplicate: CVC roots reduplicate the entire CVC and then epenthesize  
a final vowel while CVCV roots reduplicate the final syllable.


[sa'pa] "black" comes from /sap/ and reduplicates as [sap-sap-a]

[waSte'] "good" comes from /waSte/ and reduplicates as [waSte-Ste]


The 'generative' view on this is that stress and reduplication rules apply  
before epenthesis.

This pattern holds up very well for stress and reduplication.  
Unfortunately, ablaut is not so predictable. Some of the epenthetic vowels  
DO ablaut but not all of them and some of the non-epenthetic vowels DO  
ablaut but not all of them.

The fact that the other Siouan languages do not have these CVC type roots  
shows that Dakota has done a bit of reanalysis on their underlying forms.

John Kyle


On , "Rankin, Robert L" <rankin at ku.edu> wrote:
> > Correct. I was disputing the rigid dichotomy you raised in your  
> previous post to make a CVC hypothesis for proto-Siouan seem  
> unreasonable. I was not particularly disputing the substance of your  
> thesis regarding the later development of ablaut in Siouan, and  
> especially Dakotan. By your solution, *-e goes away in the face of a  
> suffixed *-a because it is phonologically weak. By mine, it goes away  
> because it is not really there at all.



> But it IS there in about 11 or 12 languages spread all over the eastern  
> 2/3 of the continent, that's my point. So, once again, EITHER we have to  
> put it there in Proto-Siouan underlying phonology, OR we have to  
> reconstruct a phonological rule in Proto-Siouan that, in effect,  
> says "all 7 other vowels (iaou iN, aN, uN) can occur unaccented  
> word-finally, but we're going to use this rule to "predict" the most  
> common one, (e). That would go against 150 years of phonology UNLESS it's  
> the only way to predict accent, in which case one might argue for it as  
> Pat Shaw, Dick Carter and others have. But since the status of phonemic  
> vowel length has been clarified (by Bruce Hays and by yours truly and  
> others), we can see that the highly exceptional CVC roots are no longer  
> justified except in Dakotan. I think that sums up my view more compactly  
> than before.



> > Thanks for this explanation. I stand corrected on Mandan.



> It had me fooled for a long time too, but when Dick Carter did his work  
> on Mandan in about the early '90s, he found length all over the place  
> along with the final -e's that Hollow had left off (for perfectly good  
> accentual reasons if he relied on Dakotan phonology to provide a window  
> into Mandan).



> > I'm confused here. Can you give me a few examples of widespread old  
> Siouan words with these word-final short unaccented vowels other than -e  
> that we're talking about here?



> I could certainly do that, but the easiest thing to do is to search for  
> them in the Comparative Dictionary MS. If I didn't send you one as an  
> attachment a couple of years back, I apologize. I can get one to you.  
> Just do a search on "PSI[ *" and it will flip from one proto-Siouan  
> reconstruction to the next. You'd get even more using "PMV[ *"  
> (Proto-Mississippi Valley).



> > Also, why would suggesting that _some_ words of the form CVCe are  
> underlyingly CVC imply that _all_ words of that form necessarily are?



> Well, I think that would depend on how seriously you take phonology, the  
> status of underlying vs. surface phenomena, the notion of invariance and  
> a host of other factors that have been cussed and discussed in the  
> literature for several decades. For me, at least, the bottom line is "do  
> I NEED to posit exceptional CVC roots in order to explain accent?" And,  
> outside of Dakotan, the answer is apparently "no." It would just cost us  
> an exceptional syllable canon for no reason. Plus, it would skew vowel  
> distribution where, otherwise, we have a number of neat positive  
> generalizations: Certain vowels tend to occur in initial unaccented  
> syllables, certain ones (all of 'em) in accented syllables, and certain  
> ones in post accentual syllables, etc.



> > I think my hearing is reasonably good too. But my hearing sometimes  
> interprets the sound as -a when they say it spontaneously, though I can  
> often get them to admit that it's -e when I force them to choose. And I  
> know that much of my foundational knowledge of Omaha grammar comes from  
> linguists, not directly from the speakers. Also, that a good deal of what  
> I thought I knew from the former has been convincingly challenged,  
> corrected, or greatly augmented by the latter.



> > Every linguist needs a "hobby horse", as my colleague Keith Percival  
> has always said, and ours is Dhegiha. Nothing clarifies linguistic theory  
> and washes away the bullshit like lots of real data.



> bob


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/siouan/attachments/20110905/95da30c6/attachment.htm>