=(b)(i) in Dhegiha

[Koontz John E]

On Mon, 3 Jun 2002, Catherine Rudin/HU/AC/WSC wrote:
> Do we know what -(p/b)i was originally?

The only function - I think - that has been commented upon in comparative
studies has been plurality.  All Siouan languages have one or more
pluralizers used in about the same way, but *=pi is restricted to
Mississippi Valley in that shape and any attempts to recognize it in
pluralizers elsewhere are speculative.  I've done some of that before, but
I won't repeat it here.

> ... In these Assiniboine constructions it (and -kta) really look like
> an "infinitive" marker of some kind -- a complementizer? modality
> head?  same-subject marker?  In any case it introduces a clause with
> obligatory same subject and no person-marking. Not unlike -- dare I
> say it? -- "to" in English.

This is one reason I thought it might clarify the nominalizing =pi in
Dakotan.

> If something like this (with deleted or bleached matrix verb) is the source
> of Omaha future and proximate forms, it seems just a little odd that we
> don't get the -(b)i and -ta on complements of verbs like "want" ...  Not
> terribly odd, since it might have survived only in fossilized corners of
> the grammar, but still a tiny bit odd.

We do get =bi in the context of reported complements and other clauses.
Off the top of my head, it occurs with e=...dh=e=gaN 'to think' and in
egaN and kki clauses.  And under quotative ama.  Also under the the and
khe evidentials, when followed by ama.  These are admittedly all cases in
which same-subject is generally not even a possibility, and the =bi acts
more like a marker of indirect speech combined with - I think -
proximateness.

It did occur with 'want' in the Assiniboine data, but that seemed a bit of
an outlier in the glosses, I think.  In OP I'm not sure I've noticed much
in the way of complementizers.  Are there any?  Usually both main verb and
subordinate are both inflected and come in sequence.

You might be able to argue that the first co-verb, gaN in OP
...gaN=...dha, was a complementizer, perhaps based on the gaN in things
like e=gaN 'like that', though we know that this is actually a contraction
of gi dative (?) and aN 'do'.

> If I remember correctly the pi vs. kta split in Linda's Assiniboine data
> was positive/negative (or realized/unrealized or some such), not
> present/past.

I'll have to check that.

JEK