Lakota wa- 'variety object'

[Koontz John E]

On Wed, 10 Dec 2003, REGINA PUSTET wrote:
> (1) thi-w-í-wa-'uN
>     house-things.PAT-paint-1SG.AG-paint
>     'I paint the house'
>
> (2) thi-'í-wa-'uN
>     house-paint-1SG.AG-paint
>     'I paint the house'
>
> The structural difference between (1) and (2) is that (1) contains the
> non-specific patient marker *w-*, whose full form is *wa-*. But there is
> also a subtle semantic difference between the two examples: according to
> my Lakota speaker, (1) actually means 'I paint the house in many areas',
> while (2) simply means 'I paint the house'.

I wonder if this wouldn't account for the situation.  Suppose that i?uN
'to paint' takes two objects, the thing painted, and the paint or maybe
(also?) the place on the thing painted that gets painted.  The paint or
place painted is governed, essentially, by the i- "locative," while the
thing painted is governed by the whole stem (or maybe just by ?uN?).  If
you omit the w- and use thi?i?uN then thi is clearly the thing painted and
the unspecified paint is understood to be one specific color (or place,
presumably the whole thing).  If you include w(a)- to produce thiwi?uN,
then this w(a)- necessarily refers to the colors painted with or places
painted on.  Or perhaps we should say that w(a)- doesn't so much refer to
these places and indicate that they are unspecified.

This is the sort of usage of wa-, of course, that must underlie the use of
wa- as a plural marker in other Siouan languages.

> Moreover, *wa-* can, obviously, appear more than once per finite verb:
>
> (5) itówapi ki   hé  wa-w-í-wa-'uN
>     picture   the that  WA-WA-paint-1SG.AG-paint
>     'I am painting that picture with different colors'
>
> The extra *wa-*, according to my speaker, refers to 'different colors'
> here.

What puzzles me here is that while I'd agree that one wa- refers to the
colors, I'd say the second wa- was the thing painted, but I didn't think a
wa- object marker could be combined with an NP for the reference.

> This analysis is substantiated by the following examples:
>
> (6) sápa w-í-wa-'uN
>     black  WA-paint-1SG.AG-paint
>     'I paint it black'
>
> (7) *sápa wa-w-í-wa-'uN
>     black   WA-WA-paint-1SG.AG-paint
>     'I paint it black'
>
> The ungrammaticality of (7) can be blamed on the fact that in (7), *wa-*
> and the color term *sápa* 'black' fill the same syntactic slot.

I'd have said that (7) was ungrammatical because sapa acted as the 'paint
or place painted' object while the w(a)- acted as the thing painted
object.

> The following examples should bring this out even more clearly:
>
> (8)  wa-yúha
>      WA-have
>      'he has all kinds of things'
>
> (9)  wó-ha
>      WA+YU-have
>      'he has things/everything (like a rich person)'

> The form *wó-* in (6) results from contraction of *wa-* with the
> instrumental prefix *yu-*. *wó-* conveys the meaning of "regular"
> non-specific patient, while *wayú-* seems to indicate a variety object.
> Not every verb that starts with *yu-*, however, admits the two
> contrasting expression formats ...  So far, all tested yu-verbs that do
> not have alternating *wa-*- forms have a *wayú-*-form but not a
> *wó-*-form, with one exception: *yúta* 'to eat'. *wóta* is fine but
> *wayúta* is not grammatical.

I'm not sure if this is necessarily the same thing.  I think the usual
explanation is that the wo- contract forms are conservative and have less
transparent meanings, while the wa-yu- forms are regularized and
transparent.  The only wo-form that I was really aware of was wota, which
also has first person forms wa-ta A1 and ya-ta A2 (yu- missing).  I don't
think the other wo- contracts have this propert either, right?  This has
one comparable correspondent outside of Dakotan.  Winnebago has ruuc^,
inflected ha-c^ A1, ra-c^ A2.  I don't recall that there's anything
unusual about the *wa + ru with this stem, but I'll check.  As far as I
know, this is the only verb like this in Winnebago.

There may be something a bit unusual going on with 'eat', but perhaps the
rest of the examples are simply more transparent vs. less transparent
cases, rather than varietal objects?

> A further argument for keeping *wó-* and *wayú-* forms apart, not only
> semantically, can be derived from the fact that for 1st and 2nd person,
> *wó-* and *wayú-* forms inflect differently:
>
> (10)  wa-blús^taN
>       WA-1SG.AG.finish
>       'I finish a lot of things'
>
> (11)  wó-wa-s^taN
>       WA+YU-1SG.AG-finish
>       'I am done'

The pattern of inflection is interesting, since it suggests that once w-y-
contracts to wo- it becomes morphologically individual.  I suppose it
infixes because w- is not a permitted stem initial, leading to an infixing
template being selected.

> The two main questions which are implicit in these data are:
>
> (a) Does a distinction between a marker for "plain" non-specific object
> vs. a similar or identical marker for variety object surface in other
> Siouan languages as well?

Not to my knowledge.

> What are the exact semantic functions of the equivalent of *wa-* in
> other Siouan languages?

In Dhegiha there are two wa- markers:  one acts like the usual perception
of wa- in Dakotan - a detransitivizer or valence reducer, with the
peculiarities we've discussed in the past - namely that it seems to be
more of an indefinite object (or subject) than an elimination of the slot
per se.  The other use in Dhegiha is analogous to wic^ha- in Dakotan - as
a third person plural object.  I'm not counting wa-a- as the patient form
of the inclusive marker.  That is, Dakotan uNk- equates to OP aNg- (A12)
and wa-a- (P12).

I've eliminated the other questions for now, since it seems like they
depend on a yes answer to the first one.