accusative and ergative languages

Larry Trask larryt at
Mon Aug 9 15:25:43 UTC 1999

On Tue, 3 Aug 1999, Patrick C. Ryan wrote:

> Oh, so "our" type of languages, accusative-type, can be *split-free*
> but "their" type of languages, ergative-type, cannot be. Akkusativ
> ueber alles!

No need for the rhetorical flourishes -- and this really is rhetoric.

There are lots of accusative languages which show no trace of
ergativity.  But it's very hard -- perhaps even impossible -- to find an
ergative language which shows no trace of accusativity.  Possibly
without exception, a language which exhibits ergativity at all is
ergative only in certain respects and accusative in other respects.

That's just the way it is.  This is an empirical result, not an
ideological position.  If you know of even a single language which is
100% ergative and 0% accusative, let's hear about it.  Nobody else seems
to have located one.

> What you seem not to be able to grasp because of your unfamiliarity
> with languages like Sumerian is that the ergative "subject" is
> frequently NOT EXPRESSED.

Perhaps, but so what?  Lots of languages frequently don't bother to
express subjects overtly -- Spanish, Italian, Japanese, hundreds of
them.  Why is it significant that Sumerian is another one?

In Basque, not only ergative subjects, but also absolutive subjects,
direct objects and indirect objects are commonly omitted when they would
otherwise be pronominal.  Is this significant?

Even in English, in which a well-formed sentence normally requires an
overt subject, it is common in ordinary speech to omit the subject.
Look at my very first sentence above for an example, and consider other
common locutions like these:

	Beats me.
	Don't know what you mean by that.
	Can't say.
	Looks like rain.
	Seems we have a problem.
	Been drinking?

Why is the omissibility of subjects of any interest at all?

> And, I am not even sure that "subject" is a useful term to apply
> to relationships between ergative and nominative languages.

Oh, `subject' is a *very* useful term, even though there appear to be
languages in which grammatical subjects are poorly developed or maybe
even absent altogether.

Anyway, the concept of `subject' is not even essential to demolish the
"passive" view of ergative languages.  All ya gotta do is to list the
syntactic properties of the three major NP-types -- S, A and P in
Dixon's notation -- and see which two pattern together.  It's almost
always S and A, which therefore constitute a single syntactic category,
whether or not we choose to call this the class of subjects.  And it's
hardly ever S and P, as required by any version of the "passive"

Facts are facts.

Larry Trask
University of Sussex
Brighton BN1 9QH

larryt at

More information about the Indo-european mailing list