Nominal Ablaut, Noun Theme Formants, and Demonstratives

[Koontz John E]

On Mon, 10 Sep 2001, Rankin, Robert L wrote:
> Dhegiha looks to me to have the opposite pattern; the DEM tends to follow
> the N only if there IS an article (which is always deverbal).

Without having done any statistics on this, I'd say that was consistent
with the numbers of examples I was noticing.  I'd say that the numbers
were something like NOUN DEM=ART > DEM NOUN > NOUN DEM.  I'm not sure
where things like DEM NOUN=ART fall in this, but I think near DEM NOUN,
i.e., fairly common.

In regard to the choice of demonstrative, ga, of course, was fairly rare
are we noticed earlier, with dhe and s^e (and e) accounting for most the
demonstratives.  Some of the more prominent examples fo ga involve cases
of explicit pointing to remote objects.

> >Demonstratives follow and are written as enclitic in Mandan.  A posiitonal
> >can follow the demonstrative.
>
> Oddly positionals tend to follow 'this' but not 'that' in Kennard.

I hadn't noticed that.  I thought it particularly interesting that
positional precede demonstratives in Winnebago.  That tends to suggest to
me that the historical processes leading to grammaticalization of
positionals have been fairly independent across the family.

> >The list I provided gave samples of most possibilities, though not of
> things like NOUN=ART DEM=ART and so on.
>
> I think in every one of my Kaw examples of this construction, the DEM-ART is
> a predicate.

I think Catherine Rudin had numerous non-predicative examples of this sort
of thing.

> > So, of course, when the demonstrative precedes I see that as a sort of
> > extraction.
>
> The pattern is so prevalent in my data that I have a hard time looking upon
> it as an extraction.  Unless the DEM follows the N and forms an NP distinct
> from the N, DEM-N looks normal to me in Kaw and, I expect, in Dhegiha
> generally.  This sort of thing can happen in languages. In Romanian you can
> say either "omul acesta" or "acest om" 'this man' but in Spanish "este
> hombre" is the rule and I can't get speakers to accept *"hombre este" as an
> NP.

I'd think that even if extraction is a marked or additional sort of thing
processually, it might come to be the more common (less marked)
alternative over time.  However, I don't know which order is considered
historically primary in Romance!  I do have the impression that DEM > NOUN
is normal in most of the older IE languages.

> Again, does any Siouan language have anything resembling a consistent noun
> "ablaut"? Or, are what little there is in the way of rules/alternations in
> the extant languages found in matching environments? (the sort of thing one
> might expect of relic forms.) If not, then for now at least the article
> theory of ablaut is still DOA for me.

I'd be interested in this, too, of course.  The only cases I'm aware of
without appealing to comparative data (Da s^uNka vs. OP s^aNge, e.g.) or
to extensions matching *-ra (e.g., Da -ya) are:

- Dakotan, fairly well discussed in Boas & Deloria and in Shaw's diss.
  (Maybe in Carter's diss., too?)

- OP, not in the literature, involving cases like ppahe' :: ppaha'=di

- Crow citation forms in e for -i and -a nouns.

I think the Dakotan cases are considered to be somewhat non-productive
(earlier) to non-productive (contemporary), though, in fact, the
conditioning environments are rather specialized and derivational, e.g.,
tha=NOUN conditions the e-grade, vs. NOUN, the a-grade as with s^uNka, so
that it would be difficult to imagine the process as every being
productive in these specific terms, by comparison with verb ablaut, which
comes on on nearly a clause by clause basis.

The OP cases are clearly somewhat moribund, too, as exceptions occur for
the same nouns in Dorsey's texts of the 1890s.

The OP data tend to firmly link the pattern with the intrusive -a- some
postpositions, whereas intrusive -ya- before postpositions falls under the
heading of *-ra extensions in Dakotan.  The OP data are one reason why I
associate the two phenomena in Dakotan.  (Though I developed that tendency
before I noticed the ablaut in OP nouns!)

My impression from Randy Graczyk's discussion of the Crow phenomena are
that they are fully productive.

I don't know if Hidatsa has anything like the Crow pattern.

With a slight loosening of criteria one could include cases like:

- Mandan's -e suffix that comes and goes and gets different analyses from
  different linguists

- Dakotan final -a and -e in alternation with C-final form

- Dakotan -ya(N) extensions on some nouns (bound vs. free instances)

- Dakotan -ya- extensions on some nouns before some postpositions

- Dhegiha (OP only?) -a- extensions on some nouns before some
postpositions

- Winnebago -ra article

- Winnebago was^c^iNiNk vs. was^c^iNge'ga ('rabbit' vs. 'the Rabbit')

- Perhaps some cases of noun :: noun-ka variation in Mandan

- Perhaps some cases of noun :: noun-ke (or ge) variation in Winnebago

I'm restricting alternations to synchronic (if sometimes fossilized)
alternations in particular languages, so I am omitting cases of final
alternations of *a :: *e, 0 :: *ra, 0 :: *ka across languages, though, as
a matter of fact, it's the same roots that participate in both the
within-language and cross-family sets.

Again, I'd again like to point out that recognizing some of the individual
alternations as patterns has nothing to do with seeing them as old
articles.  The patterns just there.  You don't have to make articles of
them.  You do have to recognize them, even if you then regard them as
chance or analogies of form, etc.

Even associating some or all of the various patterns in a greater pattern
doesn't entail seeing the greater pattern as due to old articles. But,
obviously, as you do extend the set of patterns you are willing to
attribute to a greater pattern you approach at least a possibility of a
morphological analysis.  You might consider the extensions to be theme
formants of the form -a ~ -e ~ -ra ~ -ka (~ -re?), for example.  I guess
you could consider them to be old parts of the root analyzed off by
analogy, but then the awkward question arises, by analogy to what?  That
approach actually seems to presuppose the existence of suffixed morphemes
to analogize with.

If we are looking for a non-articular - inarticulate doesn't quite work! -
solution we don't need to look any further than Indo-European.
Indo-Europeanists, of course, are quite happy talking about noun and verb
thematic affixes without passing to any issue of functional origins.  In
fact, traditionally the issue of functional origins is more or less
explicitly denied - a thematic affix is simply a semantically empty
gesture - though even the same sources consider PIE *-ske/o- to be
inceptive, -ye/o- causative, and so on.  (I'm thinking of Meillet, but my
recollections are rather vague at this point.)

Anyway, it's clear to me than a non-articular analysis of the phenomena is
possible.  It seems to me that it may omit the obvious, but I can see that
there are steps to take from seeing the individual patterns, to
integrating them, to deducing a source for the integrated pattern. At any
point (after recognizing the individual patterns), you can stop or take a
different turning.