argument structure of k'u 'give'

[Koontz John E]

On Wed, 17 Nov 2004, REGINA PUSTET wrote:
> The fact that Lakota has separate benefactive/possessive person markers
> can be taken to indicate that the language has a distinct notion of
> benefactivity. And any (?) other Lakota verb that expresses
> benefactivity does so by means of those benefactive/possessive
> paradigms, k'u 'give' being the only exception that I'm aware of with
> certainty right now.

> So what puzzled me is that if Lakota conceptualizes ditransitive verbs
> the same way as English in all but one or maybe a few cases, in using
> patient markers for (English) patients and benefactive markers for
> (English) benefactives, why not k'u 'give -- the most prototypical of
> ditransitive verbs, semantically speaking -- as well? It's the lack of
> analogy with the rest of the system that's the real eye-catcher.

But does Lakota use patient markers for patients and benefactive markers
for benefactives?  Apart from the unusual "gave me to them" case cited,
which I assume will be discussed separately, I have the impression that
there is only one set of patient agreement markers present typically.  If
there is a benefactive prefix present, the markers signify the
beneficiary; if not, the patient.  The morphology of the benefactive
causes the patient morphemes to assume a different shape - miN- vs.
maN-, etc., when they occur with it, but the potential for both kinds of
marking to occur is absent or minimal.  All the exceptions I've ever heard
of involved the third person plural. It's certainly typical of Dhegiha to
mark one kind of object or the other, not both, and I had always thought
the same was true also of Dakota, barring a few marginal examples in
either language.

You can definitely have a nominal reference to a patient in a transitive
clause with a benefactive verb, however.  I take the noun to be
essentially a "chomeur."  In some languages this would entail some kind of
adposition or case marking, but this is not typical in Siouan langauges.
As far as I know you can never mention a dative in this way - as a noun or
postpositional phrase only, with no marking in the verb.  Thus, in the
strict sense, you can say "I gave him the bread."  or "I gave bread." but
not "I gave bread to him."  This is the pattern I associate with what I've
seen called primary vs. secondary object marking, albeit in Siouan
languages there is a ranking constraint that requires a verb to make a
primary object of any dative (or benefactive) argument.  I don't think
that constraint is always present with primary vs. secondary schemes.

Rephrasing this, the pattern I thought I was seeing in Siouan languages is
that their verbs code one object.  Non-dative verbs - "simple transitives"
- code a patient and no uncoded dative object nominal can be mentioned in
the clause.  Some kinds of more remote objects can be introduced with
locative prefixes.  Dative verbs code a dative and an uncoded patient
object nominal can be included in the clause.  Most dative verbs involve
some specific marker of dativeness, which is usually manifested to some
degree in a modification of the form of explicit patient pronominals.

The unique thing about k?u and any similar verbs in this context is that
they behave as dative verbs without benefit of any special marking - they
are inherently dative.  A Siouan parallel for this (for OP) would be
dhiNge' 'to lack' which inherently agrees with the experiencer of the
lack, whereas git?e' 'one's kin to die' requires the help of the dative
gi-prefix to do this.

hiNbe' aNdhiN'ge
shoes  I don't have

iz^iN'ge iNt?e'
son      I had die

(I wonder if wiz^iN'ge is required here?)

There is one interesting thing about k?u as an "unmarked dative verb."
It does start with k.  Maybe the verb was *k-?u in Proto-Siouan.  There's
apparently no reason to see it in these terms in any of the modern
languages.  I don't know of any occurrences of hypothetical underlying
*?u, either.