accusative and ergative languages

[Larry Trask]

On Wed, 4 Aug 1999, petegray wrote:

> Consider a co-ordinate sentence with suppressed second subject:
> (a) In an accusative language, or a language with a passive construction, it
> appears as:
>        A(agent)  verb(passive)  B(grammatical subject; logical object) and
> verb
> If the second verb is passive, this should mean that B suffers both actions
> (and presumably A is the agent of both).

Not necessarily:

	`Thomas a Becket was criticized by the King and was murdered.'

> But if the second verb is active,
> it means that B suffers the first action, then performs the second.  For
> example, Latin:
>    a Paula pulsatur Marcus et necatur (odd word order, but you get the point)
>    by Paula is-beaten Marcus(subject) and is-being-killed.
> and
>    a Paula pulsatur Marcus et ridet
>    by Paula is-beaten Marcus(subject) and he laughs.

But, assuming these are grammatical in Latin, surely `Marcus' is the
subject of all the verbs, since it alone stands in the subject case.

> (b) What happens in a truly ergative language?

Don't know what is meant by `a truly ergative language', I'm afraid.

>        A(ergative)  verb(of action)  B(absolute)  and verb
> Am I right in thinking that if the second verb can take an ergative subject,
> A is the agent of both (and B is presumably the object);  whereas if the
> verb cannot take an ergative subject (i.e. it is a description rather than
> an action), then either B must be the subject (as Dixon claims for
> Dyirbal?), or the sentence is meaningless or ungrammatical  (Is this the
> case in Basque, Larry?)

Not all ergative languages have passives.  Basque has one, but the
passive does not permit the overt expression of an agent, so the
examples above cannot be expressed literally in Basque.

In general, when two verb phrases are coordinated in Basque, the subject
of the first is also the subject of the second.  This is so regardless
of case-marking.  If the two VPs differ in transitivity, then the overt
NP subject takes the case-marking appropriate to the closer verb, but
that doesn't interfere with its being the logical subject of the other
verb.

>     I guess that different ergative languages will interpret this in
> different ways, but is there any mileage in pursuing co-ordinate sentences
> in order to reveal a difference between passive languages and ergative
> languages?

Coordination is, in generally, a potentially useful test for
establishing shared syntactic properties, but it doesn't always give
useful results.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk