Biloxi nominal markers

[Koontz John E]

On Fri, 2 Mar 2007, David Kaufman wrote:
> Also, according to info sent to me by John Boyle, Hidatsa uses -ri for
> the same (focus) purpose.  Apparently Hidatsa -ri, Mandan -(r)e, and
> Biloxi -di all derive from PS -ri as a focus marker.  It's use in Biloxi
> as a focus marker is apparently in free variation with 0 marking, since
> first mention occurs both with and without it.

I suspect the marker is *e and that the *-r- reflexes occur only after
vowel-final stems.  In essence this is what the Mandan formula -(r)e
means.

> "Rankin, Robert L" <rankin at ku.edu> wrote:
> That's right. It's Biloxi -di, or -ni after a nasal V.
>
> Sara Trechter has a somewhat similar function for -(r)e in Mandan, even
> though earlier workers portrayed it as "epenthetic". Mandan use isn't
> apparently restricted to first mention.

I believe Kennard called e an article.  I don't remember how he handled
the intrusive r and h and sometimes ? before it.  I don't think Hollow
offered any gloss for it at all.  In essence he treats it as a marker
added to independent nouns, i.e., an absolutive marker.  It appears to be
missing from the initial elements of compound lexical entries, and it is
not clear whether it is also sometimes missing from independent forms.  It
sounds like it is.

In the case of Biloxi I don't recall examples off hand of just -i added to
a CVC noun stem, but there is a declarative morpheme for verbs that
clearly occurs in -(d)i form.  In some cases of the Biloxi declarative, I
recall, other intrusive elements like -h- occur.

It would be interesting to know if Hidatsa also has -(r)i, or if it is
simply -ri there.  In Hidatsa do first mentions of the same noun occur
with and without -ri?  In other words, -ri occurs with some first
mentioned, but not all?

I recall that in Lipkind's Winnebago grammar he talks about -ra (which is
invariant) and -re (ditto) as subordinators.  The first is the article.
My recollection of his examples is that they looked like they occurred
with relative clauses (nominalizations) with object heads, while -ra
occurred with simple nouns and with nominalizations with subject heads.
However, I was examining a very small corpus!

I said that -re was invariant, but, of course, I was dealing with the
examples Lipkind noted for "-re.".  I've also noticed that nouns like
was^c^iNk 'rabbit' add -e- before the "distal" or "respect" element -ga,
as in Was^c^iNgega 'The Rabbit'. That's the only example of this for
Winnebago that I recall, but it may provide -(r)e there as well.

Of course, this is all bound up with the question of noun-finals, and for
those who prefer to see all noun roots as vowel final we have to note that
was^c^iNk ~ was^c^iNge(ga) corresponds to OP mas^tiNge (PDh *mas^tiNke).
In other words, in this case an others like it, perhaps we should explain
the *e as part of the noun and the *-re as something else.  In MVS it
would be hard to argue against this if it weren't for the cooperative
co-occurring patterns with *-a and *-(r)a across the board.  However, this
is why I find the clearer *-(r)e behavior outside of MVS so interesting.

I would argue that in essence we have *-e and that it appears as *-re
after vowels (shifting to -ri in Hidatsa and -di in Biloxi) because there
is an epenthetic -r- there:

  0 => r / V = __ e

We find a few other intrusive elements like -h- or -?- because nouns that
end in h and ? lose those elements if nothing follows.  If a noun has the
form *CVh, then it appears as CVhe when *e is added, but as CV when it is
not.  We know that some things end in *-h (Rankin's discovery) because
this explains cases of final -ua (and -ue?) in Crow-Hidatsa.  I think only
one of these languages shows this, but I forget which at the moment!  In
addition, Rankin uses -h to explain the doublet suffixes -kha and -ka in
Dakotan (often with a sense 'sort of') and the corresponding -kka ~ -ga in
OP (and other Dhegiha languages).

I was initially (no pun intended) bothered that Mandan -r, -h, -? and even
-?r finals only partly matched up with the behavior of cognates in other
branches.  It dawned on me recently that this is because the finals have
become arbitrary morphological patterns in Mandan.  So, in some cases a
form has been transferred from one final class to another, or has acquired
a mixed pattern like -?r.  In fact, I remember a paper by Dick Carter from
some time back in which he pointef out evidence that the final behavior of
some stems in Mandan varied with the dialect.  We have only a few cases of
this, but, of course, we have only as few pieces of evidence of dialect,
in the form of Maximillien's mention of Nuptadi (ruNptare) forms different
from the usual Mandan forms.  I don't remember the examples off the top of
my head, I'm afraid.

This behavior of Mandan finals is like the behavior of -Cia passives in
Polynesian, where -C is historically the final consonant of the preceding
*CVCV(C)  stem.  The final *C is lost in final position, but retained
before *-ia PASSIVE.  However, the *C that is attested with a given stem
tends to vary a bit between languages.  In some languages a few stems
change from one kind of *-C to another at random.  In other languages one
particular *-C becomes favored and replaces others right and left.