Re: Ergative?

> Date:         Thu, 27 Sep 2001 14:52:12 -0400
> From: The Gray Wizard <dbell@...>
>
> > From: Vasiliy Chernov
> >
> > On Thu, 27 Sep 2001 00:02:01 -0500, Thomas R. Wier
> > <trwier@...> wrote:
> > >
> > >There are actually some ergative languages where the ergative
> > >marks agents of any kind, including those in passive sentences,
> > >where we'd normally expect something called instrumental.
> >
> > BTW, is there a simple criterion to distinguish the ergative
> > construction from the passive one?
>
> I'm not sure I understand your question.  Ergative constructions and Passive
> constructions are not mutually exclusive.
>
> Ergativity is the discriminatory application of case roles to the core
> arguments of a predicate based on a formal parallel between the P-function
> argument of a transitive predicate and the S-function argument of an
> intransitive one.  Passivity, on the other hand, is a voice operator used to
> modify the valency or argument structure of a predicate.  NPs are typically
> marked for the former while VPs are typically marked for the latter.  While
> antipassive voice is more common among ergative languages, a number also
> have passive forms (my conlang, amman iar, is ergative and has both passive
> and antipassive voice operators).
However, looking at the constructions in isolation, there's not much
to distinguish them, especially if they admit the same word order:

Intransitive: Mary:ABS/NOM sleep:INTR

Active:       John:NOM loves:ACT Mary:ACC

Passive:      (By John):INS (is loved):PAS Mary:ABS/NOM

Ergative:     John:ERG loves:ACT Mary:ABS

(These construction names and role markings are a bit dodgy, I know,
but I hope the meaning comes across).

If you're just presented with one of these constructions, and not
allowed to check what other constructions the language has, how do you
distinguish between passive and ergative? (And you don't know that the
verb is marked for passive if you don't have the unmarked form).

Well, in some cases, you can tell which of the noun phrases it is that
behaves as a grammatical subject -- if it's the patient, you might be
looking at a passive, if it's the agent, it might be ergative. But
this doesn't always work, there are many ergative languages where the
subject-like behaviour is distributed between the noun phrases. And
it can change fairly easily.

In fact, that is what happened in some languages that are now split
ergative: They used to have an active past/perfect (using a synthetic
form of the verb) and a passive one (using a participle). But when the
active construction fell out of use, the remaining one didn't have
anything to be the passive of, so we call it an ergative construction
instead. (And the original instrumental case that's used for the agent
gets to double as the ergative case). I suppose that the coordination
patterns changed as well, at least in some of the languages, to make
the agent the syntactic subject too.

Lars Mathiesen (U of Copenhagen CS Dep) <thorinn@...> (Humour NOT marked)

Reply