Nominal Ablaut, Noun Theme Formants, and Demonstratives

[Koontz John E]

On Wed, 5 Sep 2001, Rankin, Robert L wrote:
> > However, I'm not at all worried about the potential for something like
> *s^unk(...) e 'the aforesaid canine' existing in Proto-Mississippi Valley
> Siouan.
>
> That's what I do worry about. I think that any such grammaticalization
> clines we assume have to be demonstrated very carefully, and hopefully with
> different extant languages showing different stages. It's easy to say "oh,
> demonstratives, they can be grammaticalized anywhere...". It might even be
> true in some languages, but I'd want to spend a lot of time looking at lots
> of data before I'd want to accept it.

The paper by Greenberg that first led me to consider articles as a source
of noun theme formants and gender markers is:

Greenberg, Joseph H.  1978.  How does a language acquire gender markers?
pp. 47-82 in Universals of Human Language, Vol. 3.  Joseph H. Greenberg,
Charles Ferguson, and Edith Moravcsik, eds.  Stanford University Press,
Stanford, CA.

Greenberg had the advantage over us of working especially (for this paper)
with Niger-Congo, especially the Bantu subfamily, which has a great many
more languages than the Siouan family.  He didn't restrict himself to NC.
He also considered various other language families, mostly African, but
including Uto-Aztecan, with respect to the absolute markers used with
nouns there, and Indo-European, with respect to the long forms of the
adjectives in Slavic.  One of my favorite examples is Berber, since the
feminine marker, especially, shows up clearly fore-and-aft, as in amazigh
'Berber man' vs. tamazight 'the Berber language', for example.  In most of
the Moroccan dialects the t is actually theta, and the prefix is ta- for
the feminine, and a- for the masculine.

> >> In addition, here, we are dealing with two particles that share
> neither function nor basic meaning; i.e., they are not a "pair" in any
> discernable sense.
>
> >They do occur in oppositions like e=na(N) 'that many, so many' and a=na(N)
> 'some quantity' or 'how many', which willy nilly yokes them into a pair.
>
> It's true they occur consistently in Dhegiha in WH/TH pairs, but not as
> members of the same class of objects.  They are all WHich one/THis one, how
> many/this many, how long/that long, etc. pairs.

The boundary between demonstratives and interrogatives is particularly
weak in Siouan languages.  In particular, the interrogatives consistently
serve as indefinites, e.g., 'what' = 'something', 'who' = 'somebody',
'where' = 'somewhere', 'what quantity' = 'some quantity', etc.  This is
the only reason I allowed myself to consider these WH-forms in this light.
That, of course, and the form being right.  I do consider the h-initial in
Kaw, Osage and Quapaw to be a potentially serious problem, which is why
I'm interested in that initial correspondence.

> >Of course, it's a pair that's part of a larger set, including also
> dhe=na(N), s^e=na(N), and ga=na(N),...
>
> These only occur in a paradigm with the TH pair of the set. None occurs in
> the WH set. It is the TH group, namely, ?ee, *ree, *shee, *ga that form a
> set. Ha- is not a member of that set.

When every enclitic pattern that occurs with the dhe/she/ga and e
demonstratives occurs with the (h)a interrogative/indefinite it is
difficult to see a major barrier within the paradigm of TH/WH + enclitic
forms.  All of dhe=naN, e=naN, and a=naN, for example, occur.  I think I
may be misunderstanding the application of paradigm here.

There is, of course, a major barrier with the bare forms, for though the
TH members of the set do occur post-nominally, the WH members do not,
except perhaps in the case of duba/juba 'some', if that is connected with
the *tV interrogative/indefinite.  In fact, as far as I know, the (h)a
"element," never occurs as a bare stem in Dhegiha or elsewhere, and so it
does not occur either prenominally or postnominally.

I suppose we could wonder about the a-question particle that appears
sentence finally, but let's agree not to, since that's not clearly
demonstrative, even though there is some parallel with e occurring
sentence finally.

I agree that it would be very helpful to have an instance of, say OP "a
NOUN" or "NOUN a" in the sense of 'some NOUN' or 'is it a NOUN'.  I'm
pretty sure that the question particle is as close as I'm going to get on
that and I don't see that as strong enough to press.  Ironically, an
example of "a NOUN" would be more helpful here, even though I'm trying to
place a in a postnominal context.

> But we're losing track of what we're being asked to believe. Personally, I
> just think it extremely unlikely that the Ha that forms WH questions in
> Dhegiha (and sometimes maybe Winn./Chiwere too) and is always a root turns
> up as a suffix -a on a few nouns, has nothing to do with Q-formation, is
> some sort of demonstrative and explains noun ablaut.

I read this as:  "where it has nothing to do with Q-formation," right?