=(b)(i) in Dhegiha

[Koontz John E]

Since several people are interested in this issue, including some
Dakotanists, I'll post an interesting development from the SACC this year.

Incidentally, there is no official abbreviation for this very informal
conference, but SACC = Siouan And Caddoan Conference, while SCALC is, I
think Siouan and CAddoan (or Annual?) Linguistics Conference.

At the meeting Linda Cumberland described a class of Assiniboine verbs
which are characterized by embedding same subject complements that require
=pi on the embedded clause verb in the present and =ktA in the past.  The
embedded verb has no pronominal prefix.  These verbs include forms like
'want' and 'like to/don't like to'.  I don't recall the precise list at
the moment.  I hope Linda will comment further when she has time, since
she is looking for similar cases and suggestions of additional candidate
verbs to look at.

It occurred to me that this might explain two things about Dhegiha syntax
that are a bit perplexing.

1) The future is embedded under a positional article, usually forms of
dhiNkhe 'to sit' in the first and second persons, and forms of akha/ama in
the third person.  For example bdhe=tta miNkhe 'I will go'.

2) Third person singular proximates (not obviatives) are marked with *=pi
(usually > =i in Omaha-Ponca, sometimes =bi, and usually > =p=a/e - adding
male/female declarative particles - in Osage, sometimes =pi, etc.). This
particle is homophonous with the plural marker used with inclusive,
second, and third plurals.

The first case might arise from a use of a construction comparable to the
one Linda reports (but requiring pronominal prefixes) under superordinate
positionals with the original sense 'I sit that I would ...', i.e., 'I
would that ...', subsequently reinterpreted as a simple future.

Note that the inflected suffixal future auxiliary in Crow and Hidatsa
looks like a heavily reduced version of the Dhegiha futures, too, so this
may be something that has happened several times, independently, and no
doubt with different details, within Siouan.

The second case might arise from a construction like 's/he says that s/he
...'.  If the superordinate verb were one that tended to be zeroed for
phonological or structural reasons, then the =(p)i might be left as the
sole marking of this reportative sort of construction and I think it is
not unreasonable to suppose that such a construction would be more
proximate than the non-reportative, leading porentially to a contrast of
=pi (proximate) vs. [zero] obviative.  The obviative forms or the forms
taken as obviative would also naturally occur in embedded contexts, which
is definitely the distribution in Dhegiha as well.

The verb 'to say' in Dhegiha has a complex allomorphy, but the third
person stem, as in Chiwere and Winnebago is e (ablauts to a, e.g., a=i 'he
says' in OP).  Note that an additional vowel, albeit apparently varying
with gender, does actually occur with =pi in Osage and, I think Kaw:
=p=a/e (Kaw =b=e at least).

In addition, as Rory Larson has pointed out, =bi tends to appear in
reportative subordinate contexts in Omaha-Ponca, e.g., under 'think'.
This might also explain the quotative =bi=am(a) as from =bi ab(i) REPORTED
they-say.

It's possible that the superordinate verb might be something like 'to be',
as well, but that is also essentially *e in the third person.

I think that this is an approach to explaining the Dhegiha future and
proximate =pi that is well worth considering.

It might also prove a fruitful way of looking at nominalizing =pi in
Dakotan.

JEK