Noun/pronoun split ergativity (was Re: SVO vs SOV [...])

Paul Bennett wrote at 2003-10-13 17:46:50 (-0400)
 >
 > On the subject of doing weird things with pronouns, there are some
 > languages (Australian IIRC) which are Nom/Acc in nouns and Erg/Abs
 > in pronouns. Let me just go dig out my copy of Dixon _Ergativity_
 > ...
 >
 > Damn. I can't find the exact reference. I know it's in there
 > somewhere. Either there or Payne _Describing Morphosyntax_
 >
 > If anyone's interested, I can dig out the exact reference, with
 > quotes and examples.
 >

I'm sure it's in the Dixon too, but it is in Payne.  You have it
backwards, though.  It's pronouns that are accusative, and nouns
ergative. And only the 1st and 2nd person pronouns, in the example
given (Dyirbal, so I'm pretty sure it'll be in Dixon).  The examples
are on p.155.

This is given as a universal, incidentally (in _Describing
Morphosyntax_), that the more animate, the more topic-worthy the
entities denoted by a grammatical category are, the more likely it is
that that category will behave in a nom/acc pattern, and the less
topic-worthy the more likely it is to be erg/abs.  This makes sense,
in that topical/animate entities are more likely to be active agents,
and thus require a marked case for when they have something done to
them (accusative) while less topical/animate entities are more likely
to be patients, and require a special case for when they're agents
(ergative). There's a heirarchy given, on which pronouns are
inherently more topical than nouns, and SAPs* more topical than 3rd
person.  A simplified version looks something like this:

agreement > pronouns > proper nouns > animates > inanimates

For any language that shows this kind of split ergativity between
types of nominal argument, you're supposed to be able to mark the
split at some point on this scale and everything to the left will be
nom/acc and everything to the right will be erg/abs.

So, if we accept this reasoning, nom/acc marking on nouns and erg/abs
on pronouns is functionally unlikely.  (I just bring this up because I
thought it was pretty interesting when I first read it.)

* Speech Act Participants, that is, 1st and 2nd person.  It also lists
  1st as more topical than 2nd. but I think I've heard of languages
  with an inverse system that considers 2nd more agentive than 1st, so
  maybe we shouldn't take that as gospel.

Replies