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

[Koontz John E]

On Fri, 31 Aug 2001 voorhis at westman.wave.ca wrote:
> A question:  What time relative to Proto-Siouan are we talking about
> here? The ablaut patterns in Dakota/Lakota, Winnebago, and Omaha-Ponca
> seem to be similar enough in most details to warrant simply
> reconstructing ablaut in the ancestor of these languages, at least. ...

It's clear that similar forms of verb ablaut are attested right across
Siouan.  I think that pretty much the only constants are e ~ a.  The
conditioning contexts are always following enclitics and a set of
additional factors like nominalization.  Some specific enclitics are
subject to ablaut themselves in the same contexts.  Perhaps the main ones
that are reconstructable for earlier stages of the language are *ktE or
(?) iNktE, the irrealis marker, and maybe something like *krE as a kind of
third person plural.

The specific grades associated with particular enclitics are not very
constant.  I think perhaps plural stems are always a-grades, but things
like negatives and the irrealis vary wuite a bit, even within particular
dialect continuums.

Ablaut may or may not be associated with something like unaccented final
vowels.  Some stems clearly have accented ablauting vowels, like yA' 'go',
yA 'cause' in Teton.  It's true that at least 'go' may come from something
like (?) *ree'hE which might explain the exceptional pattern away as
secondary.

Verbal ablaut is by its nature a constant, productive factor in Siouan
morphology, since the conditioning enclitics are in constant use.  It can
only disappear by an active change in the morphology or phonology of the
languages.

What does seem clear is that there is a very good chance that ablaut is
some of its primary specific examples (with pi or ktE) can be accounted
for nicely by assuming that the vowel preceding the enclitic is actually
an historical part of the enclitic.  I think this was proposed first (that
I'm aware of) by Wes Jones, and taken up by David Rood, and subsequently
Bob Rankin.  It looks very reasonable to see the plural as *api (across
much of the family) and the irrealis as iNkt(e) (in at least some Dakotan
dialects).

This doesn't account for all of verbal ablaut in any cut and dried way,
because there are so many enclitics, many of them not clearly cognate, and
varying from dialect to dialect, let alone language to language.
Furthermore, in some cases apparently cognate enclitics condition
different ablaut grades in different languages.  I'm think here of the
negative, which is an e-grade in dakotan, but an a-grade in Dhegiha.
It's true that this particular set is quite complex, and seems to involve
a group of associated morphemes *s^(i) ~ *z^(i), *niN*, etc.

I'd say that ablaut as an abstract phenomenon is somewhat self-renewing.
Once you have any situation that results in a common pattern of e ~ a
alternation, any chance circumstance that produces a new e ~ a alternation
gets dragged into the complex.  Probably something like the plural *api in
connection with a rule that merges some V1 + V2 as V2 across enclitic
boundaries and syllabifies the initial vowel of the enclitic with a final
consonant of a host seeded the situation, and additional fuel has been
added to the fire since then by other enclitics at intervals.  I'm also
inclined to feel that there is an additional common source of final vowel
alternations in Proto-Siouan verbs that we haven't yet recognized.

It seems to me that the additional of aN-finals to ablaut and the
inclusion of iN as an ablaut grade are specific to Dakotan and involve
analogical extensions of ablaut in the abstract, albiet fueled in the case
of iN by a new instance of something like the original source of ablaut.

To sum things up, I think verbal ablaut originated in Proto-Siouan, but
not as morphologized and lexicalized ablaut (vowel alternations) per se.
I suppose you could call the stage at which it was still an unpatterned
collection of vowel elicisions or combinations and resyllabifications at
enclitic boundaries Pre-Proto-Siouan, and the stage at which it was more
abstract and arbitrary Proto-Siouan, though I'm not sure that this phase
of the development hasn't occurred more or less independently in the ealy
stages of development of the several branches of Siouan.  Certainly the
system continues to evolve in the various contemporary languages.

Turning to nominal ablaut, either in the form of not very productive
internal relicts, as in Dakotan or Omaha-Ponca, or cross-branch
alternations, as between, say, Dakotan and Ioway-Otoe, it is more
restricted, and may only occur in Mississippi Valley.  It, too, involves e
~ a alternations, but it is usually fairly obscure in conditioning.  I'd
say that the e-grade could be attributed to 'possession', perhaps
specificity, in Dakotan, and that at least some a-grades seem to be
conditioning by enclitics in Dakotan and Omaha-Ponca, but I'm sure not
everyone would agree with this.  Even if 'specificity' marking accounts
for the e-grades in, say, Ioway-Otoe, it's interesting, but maybe less
useful, to be able to say that most e-final nouns in Dhegiha reflect
'specific' forms, while most a-final nouns in Dakotan reflect 'generic'
forms.  The real benefit in comparative terms may be to save us from
looking for a specious phonological basis for the difference in final
vowels.

Nominal ablaut is probably an internal development of Mississippi Valley,
originating in Proto-Mississippi Valley.  I think the conditioning
enclitics in this case were just vowels, though morphemes none-the-less.
It looks superficially like verbal ablaut as an abstract morphological
phenomenon, and has a similar source.  It interacts with it because the
morphemes in question were nominalizers and appear with nominalized verbs,
and because the morphemic basis for it are homophonous with the
phonological grades of verbal ablaut.  But it is not as old as verbal
ablaut and has an independent source.  If the nominalizers in question
have been, say, i and o, then things would have looked quite different.

Nominal ablaut is not productive, primarily because in each of the
branches the renewal of article and/or nominalizing systems has replaced
the relevant morphological systems, and because in Ioway-Otoe and
Winnebago the merger of final *a and *e after velars and the subsequent
loss of *e in final light syllables in Winnebago has eliminated much of
the phonological material of the system.  Ironically, I think that the
Winnebago =ra (and maybe =re) nominal markers and maybe Ioway-Otoe are
'that' may be the best remininants of the system as articles, though they
seem to be post-vocalic allomorphs.

JEK