accusative and ergative languages

[Larry Trask]

On Thu, 29 Jul 1999, Patrick C. Ryan wrote:

[on cases like this one:]

>> `John hit me and went away' *must* mean `John went away'.

> Another word for a "rule" is a "convention". Again, we appear to be
> chasing our tails (definitions).

There is nothing conventional about it.  We are looking at a rule of
English syntax.  If anything is a rule of English syntax, this is.

A convention, by definition, can be changed by agreement.  We do not
have that option here.

[on this one:]

>> `John hit Bill and went away'.

>> Now, in English, it is John who went away, not Bill.  However, according
>> to my understanding of Dixon, if you say what looks like the literal
>> equivalent of this in Dyirbal, it is *Bill* who went away, not John.
>> This is one of the ways in which syntactic ergativity manifests itself
>> in Dyirbal.

> Is it really that simple?

Yes.

> Would we not be coming close to describing both 'nominative' and
> 'ergative' phenomena if we said that "a null-NP frequently refers to
> the foregoing NP with which it agrees in case"?

No, not at all.  Quite apart from that slippery `frequently', this
doesn't remotely work.

First, the English example has no case-marking -- yet the facts are
clear.

Second, in Basque and in other ergative languages, the null NP in such a
construction co-refers to the NP with which it DISagrees in case -- or,
rather, with the NP with which it would disagree in case if the null NP
were overtly present and bore the required case-marking.

Much of the point here is that, in a morphologically ergative language,
the morphology -- including the case-marking -- is at odds with the
syntax.  And `subject' is a syntactic notion, not a morphological one.

[on ergative languages]

> First, as you know, I have questioned whether Dyirbal is, in fact,
> "split", and have proposed a different explanation for the data.

Dyirbal is about as thoroughly ergative as any language I have ever
seen, but, even so, Dixon makes it quite clear that a split exists.

> Second, the point that Leo is making about distinguishing syntactic
> from morphological subjects needs to be addressed by you in terms of
> "subject" properties.

I have already discussed this at length.  In a Basque transitive
sentence, it is the ergative NP, and not the absolutive NP, which shares
the subject properties of the absolutive NP in an intransitive sentence.
As far as I am concerned, that is the end of the matter.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk