argument structure k'u etc.

[REGINA PUSTET]

Sorry for joining the discussion so late, but my weekend was rather busy.
I have checked my grammar files, and in there the next-best thing to the hypothetical wawak'u etc. structures David would like to explore are the following examples:

(1)        w-o-'uN-k'u-pi    'they fed us (gave us things)'

(2)        w-o-'uN-ni-c'u-pi-kte  'we'll give you food (=things)'

(3)        w-o-wicha-k'u-pi   'they gave them food (=things)'

However, in all three cases, a potentially valence-increasing locative prefix o- is present, which may explain the possibility of adding the direct object (patient) wa- 'things' to the verb, which also carries person affixes for the indirect object (dative/benefactive/recipient).

In an earlier discussion, I quoted the follwing example from my text data:

(4)        wicha-ma-k'u-pi  'they (my family) gave me to them (my husband's family) in marriage'

This time we're looking at a bare k'u-verb which is inflected for both patient and recipient (and agent, of course).

As for the valence issue, I recommend viewing things systematically from two angles: structure and semantics. On the structural side, a criterion that works very nicely is determining whether an argument (at least in non-passive/antipassive constructions) is omissible or not. Anything that's not omissible is part of the valence of a given verb. The "problem" here is that in some cases, arguments marked by obliques will end up in the valence frame. Locatives in 'put' and similar verbs are an example. Also the 'of'-phrase in 'devoid of'. You can't use this adjective without adding the of-phrase. Personally, I don't consider this a serious problem, so I would classify 'devoid' as transitive. There is literature on this, including some passages in my 2003 book on copulas.

This brings us to the second criterion which has probably been used to define valence much more often, that of the semantics and other properties of the case markers involved. A "well-behaved" argument which is part of the valence frame of some verb better be either a subject/agent, direct object/patient, or at least an indirect object/benefactive, right? And in most cases, probably globally, the arguments which are "inside" valence on the omissibility criterion will function to code these three semantic/syntactic/whatever roles. But there are exceptions, as examples like 'devoid' show.

Regina






Koontz John E <John.Koontz at colorado.edu> wrote:On Sat, 2 Apr 2005, ROOD DAVID S wrote:
> English, "put", e.g., requires a locative (you can't just "put
> something"). Does that mean that the locative is an argument of that
> verb? Or is the absence of prepositional marking part of the
> prerequisite for calling something an argument?

My inclination on this is that a required locative is indeed an argument,
while an optional one is not. I think this amounts to Filmore's
distinction of between central and peripheral obliques. This is the point
at which many languages incorporate the adposition into the verb. I think
it's also the one at which Siouan languages often allow indexing of the
argument governed by a locative prefix, e.g.,

JOD 1890:61.6
aNdhaNna?u (i'..na?u 'pass close to')
i-aN -na?u
pass close to me (IMP)

JOD 1890:165.11
aNdhaNdhikkaN (i'..kkaN 'to contend with')
aN-i-dhi-kkaN
we contend with you

However, for what it's worth, locations don't seem to be mandatory with OP
'to put' forms, which may be one reason they have the alternate reading
'to put away'.

> Bob's comment about the Dryer "primary/secondary" argument typology is
> relevant, except that Lak. has a very robust "dative case" that it can and
> does utilize quite often to show "secondary" argument (i.e. obliquely
> marked) status for recipients and beneficiaries.

It was definitely Dryer's article I was remembering:

DRYER, MATTHEW S. Primary objects, secondary objects, and antidative.
Language 62.808.45 (1986).

> I have somehow internalized the idea that Dryer's theory is a typology
> of systems (some languages do it one way, some the other), not a
> variable for individual verbs. ...

That's what I recall, too.



		
---------------------------------
Yahoo! Messenger
 Show us what our next emoticon should look like. Join the fun.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/siouan/attachments/20050404/20f0a112/attachment.html>