(fwd) from VIctor Golla re. Agent/Patient

[Spike Gildea]

5/13/98

I have a footnote to the recent discussion by Wally Chafe and
Spike Gildea of "indefinite" markers functioning in "nonpromotional
passives".  As Spike says, this is a very common phenomenon, no
less so in North America than elsewhere.  In addition to the Caddo
example Wally cites, most Athabaskan languages similarly whisk the
agent off-stage by means of a bleached-out agent marker.

Athabaskanists (particularly those brought up on Navajo) are in the
habit of referring to this category as "4th person", and it is so
widespread in the family that few doubt that it should be recon-
structed to the protolanguage.  Although Chad Thompson and others
have argued that the Proto-Athabaskan 4th person marker was multi-
functional, if so, marking a nonreferential agent was its central
role, with other functions developing in specific discourse contexts.
The commonest of these secondary functions is the coding of 1st person
plural agents.

However, a much less expectable thing happens to the 4th person
in the California subgroup (Hupa, Mattole, Kato and a few other
dialects).  While Mattole preserves the marker in its general
Athabaskan impersonal role, Hupa and Kato appear to have have
reversed the evolutionary process and have re-coded the prefix
to mark a discourse-salient 3rd person.  The inherited 3rd person
marker, meanwhile, has assumed an obviative function.  Schematically:

General Athabaskan:

 SOMEONE sees you. (= 'you are seen'; 'we see you') [4th person]
 HE/SHE sees me.  [3rd person]

Hupa:

 HE/SHE (whom we're talking about here) sees you.  [4th person]
 HE/SHE/IT (who is not our focus) sees you.   [3rd person]

(It was this pronominal opposition that P. E. Goddard notoriously--and
mistakenly--described as distinguishing adult Hupa men from children,
women, animals, and non-Hupas.)

The formal history of this shift is pretty clear (at least to me),
but the functional motivation is not.  Does anyone know of other
instances of nonreferential "indefinite" agent markers evolving
a salient referential function?

--Victor Golla