accusative and ergative languages

[Jens Elmegaard Rasmussen]

On Tue, 29 Jun 1999, Larry Trask wrote:

[...]

> There are two separate issues here, the synchronic one and the
> diachronic one.

> (1) Is a particular ergative construction "really" passive in nature?

That was not meant to be my question.

> (2) Does a particular ergative construction descend by reanalysis from
> an earlier passive?

That was what I was driving at, albeit only as a possibility, not as a
pervasive solution of all ergative.

> For the diachronic question (2), the answer "yes" has been defended in
> some particular cases, including Indo-Iranian.  But a passive origin for
> the Indo-Iranian ergative constructions has also been disputed, and I am
> not aware that there exists a consensus among specialists.

> Since Indo-Iranian is a rare case in which we have millennia of
> documentation of the intervening stages, and since the right answer to
> the question is still not obvious, then it must be very much harder to
> answer the question in respect of other cases for which little history
> is available.  At present, there appear to be few cases in which the
> origin of an ergative construction is fully understood and beyond
> controversy.
[...]

I'd say the cases I have seen of ergativity in Indic and Iranian languages
so clearly reflect underlying/earlier passive circumlocutions that
controversy is absurd. Yet, in that field controversy is to be expected
over anything.

I grant you (quoting Dixon) that Hittite is not this way: making neuters
animate when they are subjects of transitive verbs is indeed not the
reflex of a passive transformation. I have no knowledge of the Amazon
language Pari, let alone of its history. But note that I never claimed
that ergative always comes from passive, only that there are very clear
examples that it sometimes does - at least, that was what I meant.

Jens