Dhegiha Plurals and Proximates

[Koontz John E]

On Wed, 11 Jun 2003, Rory M Larson wrote:
> Yes!  That's a very interesting point.  It appears that all the cases
> of future + positional cause the tte to ablaut to tta.
>
>   tta miNkHe      I-future
...
>   tta tHe         constrained future
>
> What's especially interesting to me here is that the a-grade ablaut
> occurs, not just before (probably) every positional, but even before
> the conjugated form of each positional.
>
> Wouldn't this suggest that what is causing ablaut is not the passive
> front end of a particle that happened to begin with a- (e.g. an
> aboriginal *=api), but rather a separate particle *=a- that once stood
> regularly in front of particles of a definite grammatical class?

Bob said I might say something about this, but I've hesitated, because
I've said an awful lot about it in the past, and I think it may be rather
in oversupply for the market.  However, I've thought of a way to say
things a bit differently, so I'll try it.

I'd have to say that =api (plural/proximate) is a rather different case
from "positionals in various uses," and need not be governed by the same
factors.  After all, there are only so many vowels to work with:
different cases of "a"  need not arise from the same factors, or, if they
do, may arise at such different times in the history of the source of "a"
in question as to be effectively unrelated.

However, I agreee that the positional cases in particular suggest an
independent morpheme -a-.  I suspect the context is less of a case of what
follows than of what precedes, e.g., a nominalization.  What I've argued,
in effect, is that =a is a nominalizer of some sort, perhaps originally an
article, but now perceived (by linguists anyway) as conditioned by the
following later layer of nominalizers (the positional articles), but
originally present in its own right.

The progress of the phenomenon would be something like this.

Stage I:  Add =a to mark nominalizations (or some of them).  We're
necessarily a little vague about specific contexts, as the remnants of the
system are much worn down and probably extensively reformulated.

Stage II:  Some sort of vowel change (e > a) occurs at the end of
(some) nominalized forms.

Stage III:  Add positionals to mark some sorts of nominalizations.

I'm saying nominalizations, but of course, it's really "nouns," and the
nouns are nominalizations only if they happen to be clauses ending in
verbs.  The actual marker of nominalizations is probably "nothing."  But
once you have a noun you can add an appropriate "post-nominal" particle
(like an article).  If you add the same particle to all nouns it is
beginning to function something like a nominalizer, especially when added
to a verb.

In the nature of things Stages I and II and II and III, maybe even all
three must overlap, though Stage I begins before II begins before III.
In the end, however, you have a system with Stage III and the observed
rule that certain forms change e to a when a positional follows them.

In some cases the rule about adding -a or changing to it may have been
lost.  Certainly a great many nouns do not end in -a or reflexes of *-ra
(with epenthetic -r- after vowels).  In fact, a-final nouns with clearly
added -a are only really common in Dakotan.  Elsewhere those nouns today
end in -e or nothing.  In other cases the -a- may have been accidentally
reassociated with the following positional (e.g., the OP proximate subject
articles ama and akha).  In other cases =ra (< *-ra) may still be taken to
be something like an article, and positionals may still fall something
rather short of being articles, though occurring with some NPs, this being
essentially the case in Winnebago.  In Dakotan -a and -ya (< *-ra) are
variable endings of some nouns (when the nouns are not incorporated or
when -e doesn't replace -a in possessed forms) and =ki is the (main)
article, while positionals are mostly used only with verbs, as
auxiliaries.

I keep offering this as a hypothesis, though it hasn't exactly taken folks
by storm.  Many details need to be worked out.  It doesn't account for the
core of -e ~ -a phenomena in the verb, with pluralization, for example,
and there may or may not be good examples of *a as a demonstrative.
Also, as there are maybe some traces =ki in Southeastern, figuring out
whether and how *=ki enters into things is also a definite consideration.
I've sometimes wondered if -e as a noun final might not be from =a=ki with
intervocalic loss of -k-.  If *=ki was lost throughout Mississippi Valley
and only, perhaps, restored analogically in Dakotan from some sort of
relict context, that might explain a lot.

I'm pretty sure now that the Dhegiha progressives are originally
nominalizations, i.e., what was originally (and still is structurally)
"the horse's eating of the corn" or, better, "the horse which is eating
the corn"  is now functionally "the horse is eating the corn."  This sort
of progression occurs commonly in language.  These nominalizations do not
have -a-.  The verbs involved do not ablaut, though a-ma and a-kha seem to
have the -a- as a prefix, and the the obviative forms thaN, etc., get an
unexpected a- in the inclusive inflected forms.

Though I don't know why, I think those Dhegiha futures in =tta=miNkhe, ...
are also nominalized, from *=tk=a=POSITIONAL.  This is probably an old
pattern, because I suspect that those Hidatsa and Crow inflected futures
are worn down remnants of the same structure, with only accented syllables
of the auxiliary remaining.  It would be nice if a piece of *=tk remained,
but I gather it doesn't.