On Wed, Nov 21, 2001 at 09:36:05AM -0500, John Cowan wrote:
> H. S. Teoh wrote:
> >Well, perhaps I didn't make things obvious enough. It *is* necessary to
> >still mark the incorporated noun for case -- for example, the verb
> >_lww'ma_ means "to carry", but the attached noun could be the carrier or
> >the thing which is being carried.
> In other words, case-marking on a verb-incorporating noun shows the
> role of the noun wrt the incorporated verb, rather than its role
> in its own clause?

Well, in this particular construction, it shows *both*. It's as if the
sentence has multiple main verbs.

> If that is so, how does one show the "native" role?  For example,
> in
>        dog-AGT bit carry-man-AGT
>        The dog bit the postman.
> "postman" is a patient in its own clause, but is agent of the
> incorporated verb "carry".  How is its patienthood shown?

Well, your example is handled differently in my conlang. A full subclause
must be used here:
        gha'ng0 kyy'kh nu    lww'me da      pii'z3du.
        monster harms  <sub> carry  </sub>  man
        (org)   (v)    (rcp) (v)    (instr) (rcp)
        "The monster harms the carrying man."
        The last three words _nu lww'me d0 pii'z3du_ forms a receptive
        noun clause, which is the receptive argument of the main verb
        _kyy'kh_. It could be interpreted as "the man who carries", or
        "the delivery man", which is what you want.

If verb-incorporation were used:
        gha'ng0 kyy'kh lww'mupiiz3du.
        monster harms  carry-man
        (org)   (v)    (rcp)

This sentence would have a very weird meaning of "the monster harms the
man to whom it (the harming) is carried." In other words, the act of
harming on the monster's side is the carrying of something to the man.
Semantically nonsensical, though grammatically valid :-)


Never ascribe to malice that which is adequately explained by incompetence.
-- Napoleon Bonaparte