Ablaut (RE: Obviative/Proximate and the Omaha verb system)

[Koontz John E]

On Thu, 30 Aug 2001, Koontz John E wrote:
> We tend to think of ablaut in terms of verbs only, but Dakotan has nominal
> ablaut, too, ...

> Given this we have a bit more than a set of arbitrary unrelated facts.
> We have some sort of process that allows e ~ a after stems of the form
> CV'C and perhaps CV.  Either both of the vowels are affixal in some way
> (I've suggested articles) or one is organic or epenthetic and the other
> morphemic, etc.  ...

I perhaps should have said that my suggestion was that nominal ablaut and
related cross-language patterns were to be explained by postulating two
"articles" *e 'specific' and *a 'generic' that acted as enclitics to nouns
in, say, Proto-Mississippi Valley Siouan and were widely preserved with
what had been CVC-stem nouns as well as with some CV nouns (+> CV-r- with
epenthetic r), as well as with verb citation forms in Dakotan and in the
various fossils like the Winnebago =ra article and =re relativizer, in
those intrusive -a- linkers with postpositions in Dakotan and Omaha-Ponca,
and so on.

I thought that *e was clearly the *e demonstrative (or third person
pronoun), and that *a was the *a demonstrative that is used to form
indefinites and interogatives widely in MVS (but not in Dakotan), cf. OP
e=naN 'that many' :: a=naN 'how many, some number or other'.  A
significant morphological problem I recognized later with this is that the
*a demonstrative surfaces as *ha in Dhegiha outside of Omaha-Ponca.  I
still think there may be something in the general approach, but I'm a bit
worried about that h, even though it is a bit of an odd-ball.

Incidentally, Dakota nominal ablaut does clearly seem to have a-variants
for less-specific cases and e-variants for more specific.  This holds for
nouns where the e-grades are for possessed forms, and I had the impression
it accounted for anomalous cases of e vs. a in nominalized verbs, too.

You can compare that a :: ha set with 'day', incidentally.  That has h in
Osage, say, but 0 in OP and Dakotan:  Da aNpe=(tu), OP aNba(=dhe), Os
haNba (I seem to recall, but can't look up at the moment).

I think the h continues on into Ioway-Otoe and Winnebago.

Something similar occurs with the first person *wa, which is a in Dhegiha
(throughout), but ha in IO and Wi.  In Wi there is reason to believe that
h is epenthetic before initial short vowels, but this doesn't explain IO.

Another nominalizing, probably demonstrative particle that gets attached
to nouns is *ka, which sure looks like the 'yon' demonstrative.  In
Winnebago it looks like *a after velars gets shifted to *e and all final
*e after simple finals is deleted, though it is preserved after clusters.

Mandan and Winnebago are hotbeds of *ka affixation, though it occurs in
other branches as well.

JEK