Re: THEORY: Mixed erg/acc

>I have a question about langs that use ergative/absolutive
>for all sentences, except for 1st and 2nd person, which
>use nominative/accusative (e. g. Dyirbal and my G'amah).
>
>Does this mean that
>
>a) the 1st and 2nd person pronouns are marked NOM or ACC
>   while the other argument is marked ABS or ERG, respectively,
>OR
>b) when the subject is 1st and 2nd person, it's marked NOM
>   while the object is ACC?
Alternative (a).  If the transitive subject is 1st/2nd person and
the object is non-1st/2nd, then the subject is in the NOM form
and the object is in the ACC form (i.e. both are unmarked).  If
the transitive subject is non-1st/2nd and the object is 1st/2nd,
then the subject is in the ERG form and the object is in the NOM
form (i.e. both are marked).

There's another kind of split-ergativity pattern out there as well,
where the choice between NOM/ACC marking and ERG/ABS marking
depends on the aspect of the verb.  If the verb is imperfective,
then you get the NOM/ACC pattern, and if the verb is perfective,
then you get the ERG/ABS pattern:

   John-NOM  book-ACC  read-PROG
   "John is reading the book"

   John-ERG  book-ABS  read-PERF
   "John has read the book"

This kind of split-ergativity is found in Hindi and other Indo-Aryan
languages, and also (I think) in Georgian.  (A linguistics professor
of mine, Anoop Mahajan, has a neat story for explaining this pattern,
but I won't go into that unless asked...)

Matt.