universality of gfs

[Avery Andrews]

Rachel wrote ...
 
> The claim for Jaminjung, as far as I understand it, is that all of the
> syntactic and morphological facts of the language can be explained
> without reference to grammatical functions such as "subject" and
> "object".  There are no controlled clauses, valence-changing
> operations, etc. that operate on these grammatical functions.  There
> is argument cross-referencing on the verb, but this can be explained
> as cross-referencing "agent" and "patient" (rather like the Acehnese
> data Joan mentioned in an earlier post). (Note however, that this
> cross-referencing system, along with the distribution of absolutive
> case, can be used to make a term/oblique distinction such as Avery
> mentioned in his earlier message.) 

How well does it really work to do it in terms of strictly semantic
Agent and Patient concepts?  What about the treatment of a verb
like `see', for example?  Dixon invented A and O in part because
the semantics of Agent and Patient weren't always really being obeyed,
so maybe it might be an idea to ask the investigator how A, S & O would
work out (basically as a way of smoking out more data; these terms carry
slightly different theoretical baggage than subject & object).

What I expect to see is Agent-like arguments going one way, and Patient-like
arguments going another, without any simple & solid definition of Agent that
will explain it (Acehnese is like this, self-propelled vehicles are for
example treated grammatically as Agents).  If there's a reasonable
amount of this kind of arbitrariness, you can say that there have to be
something like A and O rather than just Agent and Patient.  And if there
really isn't, that's very interesting too.

 Avery.Andrews at anu.edu.au