"Thomas R. Wier" <[log in to unmask]> writes:
> > But since some languages have them, it must be possible...
> > on a related note, are there any ergative languages which have
> > applicatives?
> Mayan languages, which are mostly ergative, have them IIRC.

My Qthen|gai has applicatives, too.  This is a recent extension
partially due to this thread.  It already had incorporated pronouns
for any adjunct case, and now, it features a promotion of any adjunct
to patient using an affix.  You can also add a passive voice marker to
further promote it to agent.

BTW, I have antipassive voice, too, demoting an agent to patient.  I
defined that for passive voice, the original agent may be supplied as
a causative adjunct.  In antipassive voice, I defined that the
original patient may be supplied with a serial verb construction (SVC)
with the word 'to do' + patient.  Is that feasible?  Example:

   drink milk.PAT John.AGT
   'John drinks milk.'

   John.CAU drink.PASS milk.AGT
   'Milk is drunken by John.'

   drink.ANTIP John.PAT do.SVC milk.PAT
   'John's drinking is done to the milk.'

Is this feasible?

(The SVC-marker has not yet materialised, I have to think about its
exact surface form, but that's not important for the theory, of