Lakota wa- 'variety object'

Tue Dec 16 19:17:40 UTC 2003

(Regarding the difficulties with the format of some of my previous messages, I'm working with Yahoo, and I'm not sure if there is a way of getting around html format with Yahoo. I'm now trying to convert my message into txt format -- please let me know if this looks nicer.)

Here's some more data on Lakota wa-. This time, I'm mainly interested in constructions with multiple wa-, but I'm afraid that the following data complicate, rather than clarify, the situation. Using iyuNg^a 'to ask' for the purpose of demonstration, in Lakota, this verb may take three (seemingly direct) objects:

(1)        John itowapi ki taku ota iyuNg^a-pi

            John  picture the things many ask-PL

            'they ask John many things about the picture'

(2)        John iyuNg^a-pi

            'they ask John'

(3)        taku ota iyuNg^a-pi

            'they ask him many things'

(4)        itowapi ki iyuNg^a-pi

            'they ask him about the picture'

although there is a tendency to mark the 'about' phrase explicitly by means of a postposition:

(5)        taku ota John itowapi ki el/etaN/uN/thaNtahaN iyuNg^a-pi

            'they ask John many things about the picture'

With w-iyuNg^a-pi (WA-ask-PL), the acceptability of objects seems to be more restricted. Putting all three objects is ungrammatical:

(6)        *John itowapi ki taku ota w-iyuNg^a-pi

            John  picture the things many WA-ask-PL

            'they ask John many things about the picture'

In all wiyuNg^api-examples elicited so far, the NP John may appear, which is not surprising since John, as an animate referent, can never be coreferential with wa-, which requires inanimate referents. The same is true for the wa-w-iyuNg^a-pi-examples (WA-WA-ask-PL). But only one of the other two object NPs is admissible. So a possible working hypothesis is that with wiyuNg^api, either the 'about'-phrase or the 'object of the question'-phrase (for lack of a better term) are eliminated by the presence of w-.

'object of question'-phrase present:

(7)        John taku ota w-iyuNg^a-pi

            John things many WA-ask-PL

            'they ask John many things'

However, if the 'about'-phrase is explicitly expressed, it must apparently be coded by means of a postposition:

(8)        John itowapi ki thaNtahaN w-iyuNg^a-pi

            John picture the about WA-ask-PL

            'they ask John about the picture'

(9)        *John itowapi ki wiyuNg^api

            'they ask John about the picture'

One more thing worth checking is if the presence of a postposition such as thaNtahaN in the 'about' phrase would make the triple-object example (6) acceptable. On the basis of the above working hypothesis, we might expect now that with double wa-, both inanimate object NPs, i.e. the 'about'-phrase and the 'object of question'-phrase, will be "banned" from the clause as independent constituents. However, very much to my surprise, the following example is grammatical on several trials:

(10)      John itowapi ki    taku    ota      wa-w-iyuNg^a-pi

            John picture  the  things  many  WA-WA-ask-PL

'they ask John many things about the picture'

At this point, the hypothesis about the object-eliminating function of wa- collapses since, unless the verb iyuNg^a 'to ask' can take some mysterious additional types of semantic object whose exact nature could, so far, not be determined by my elicitation techniques, we have to interpret wa- and w- in (10) as coreferential with the full NPs itowapi ki 'the picture' and taku ota 'many things', respectively. I realize that these data do not really take us in the direction in which they hoped they would take us, since my initial idea was that by putting full object NPs and checking which wa-s would be eliminated by their presence, we'd get some more insight into the semantic reference of the wa-s. In this context, I also experimented with Rory's suggestions about wa-:

(11)      sapa i'uN  'to paint black'   is grammatical, and so is

(12)      itowapi ki he wi'uN 'to paint that picture'.

I also elicited

(13)      John itowapi ki he zi i'uN

      'John paints that picture yellow'

(14)      John itowapi ki he zi wi'uN

      'John paints that picture yellow'.

In other words, again, it does not seem to matter much if the wa-s are there or not, they do certainly not eliminate the full object NPs they should be coreferential with. So, structurally speaking, these wa-s do not really detransitivize verbs, although I agree with Bob saying that in other cases, via "piecemeal grammaticalization", wa- might actually function as a true detransitivizer.

As for Carolyn's Osage data, in with wa- is glossed by 'repetitive action', I think they are really good examples of something that is in the semantic range of "variety object". So if variety wa- in Lakota indicates diversification of the object in the material world, the Osage version expresses diversification of action in time. I tried to track down this usage in Lakota as well, but I always ended up with examples in which repetitive action is coded by the progressive/continuous/repetitive marker -hAN/-he, rather than by wa-:

(15)      thi-'i-wa-'uN-he

            house-paint-1SG.AG-paint-REP

            'I paint the house over and over, I paint and paint the house'

Regina

---------------------------------
Do you Yahoo!?
New Yahoo! Photos - easier uploading and sharing
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/siouan/attachments/20031216/167bcf8c/attachment.htm>