accusative and ergative languages

[Larry Trask]

On Fri, 9 Jul 1999, Jens Elmegaard Rasmussen wrote:

> Excuse my interruption, but I think your discussion is missing the
> only point of any interest in this context: Can "Mutila jo zuen" not
> *come from* something which *originally* meant, not 'the boy was
> hit' pure and simple, but specifically 'the boy was hit by him
> (e.g., by the one we're talking about)'? Nobody is claiming that the
> passive constructions out of which the ergative grew in a number of
> languages were restricted to impersonal use. Incidentally, if they
> were, there would not have been anything to put in the case that
> subsequently got interpreted as a "transitive-subject, i.e. ergative
> case".

Well, this suggestion is possible, I suppose, but I know of no evidence
to support it, while there is some evidence against it.  Let's look at
the structure of the Basque sentence.

Here <mutil> is `boy', and <-a> is the article.  This NP, being a direct
object, stands in the absolutive case, which has case-suffix zero.

The item <jo> is the perfective participle of the verb `hit'.

And <zuen> is the auxiliary verb-form.  Now, third person is usually
marked by zero in Basque verbs.  So, the usual absolutive agreement
slot, which is the first slot in the auxiliary, cannot be filled by an
agreement marker, and it is filled instead by <z->, a redundant marker
of past tense.  The next element, <-u->, is a reduced form of the verbal
root <-du->, from *<edun> `have'; this is the usual transitive
auxiliary.  Again, the ergative agreement slot, which follows the verbal
root, is empty, because the ergative (subject) NP is third-person.
Finally, <-en> marks past tense.

The whole thing is thus this:

	boy-the-Abs hit-Perf Past-have-0-Past

Or, roughly, `(She/he/it) had hit the boy.'

But it translates English `S/he hit the boy' (before today).  This is a
periphrastic form comparable to the ones we find in Romance and
Germanic.  Nobody knows if such periphrastic forms are calqued on
Romance or are an independent development in Basque.

Now, I can see little scope here either for an original passive
interpretation of the form or for any way of reading `by him' into the
auxiliary form.

> I don't know a first thing about Basque, though I have been
> intrigued by it on many occasions, especially since it offers such a
> good parallel to Old Irish in the verb where you apparently have to
> memorize practically all forms (which are many) to be able to say
> even the simplest of things - and that of course was also what made
> me stop every time I got started. >From the primitive and casual
> books at my disposal I do see that "zuen" and "zuten" mean 'he had
> him' and 'they had him' resp. I also believe I see that such
> auxiliaries are combined with a particularly short form of the
> participles, referred to by Schuchardt as the root of the
> participle;

It is simply the perfective participle of the verb.  Most participles in
Basque carry an overt suffix proclaiming them as such, but <jo> happens
to be one of the exceptions: it has no participial suffix, though I
suspect that it once did, and that the suffix has been lost by a
combination of phonological change and analogical readjustment.

> and "jo" is 'stick; beat' in its shortest form, says my little
> dictionary;

Actually, `hit', `strike', `beat' -- not `stick', which is a noun.

> and "mutil-a" is 'boy' with the article "-a", but without case or
> number marking.

Not quite.  The form <mutila> is marked as absolutive by its zero
suffix, and as singular by its singular article <-a>.  The plural
article is <-ak>, and `the boys' is <mutilak>, in the absolutive.

> Therefore my persistent question: Why can't "mutila jo zuen" and
> "mutila jo zuten" reflect a construction that was earlier meant to
> express 'the boy, he had him hit', 'the boy, they had him hit'?

Well, I can't rule that out, but I can't see any evidence to support it.
Note in particular that <mutila> exhibits *no* subject properties in
modern Basque, or in Basque of the historical period.  If it ever was a
subject, as this proposal requires, the reanalysis must have been
carried to completion a long time ago.

Note also that intransitive verbs are likewise conjugated
periphrastically but with the intransitive auxiliary <izan> `be'.

> Schuchardt also gives "zen" to mean 'he was', so that if you gloss
> "mutila jo zen" as 'the boy was hit', it seems there is quite a bit
> of agreement that the verbal root is a participle by itself.

The lexical verb stands in the form of its perfective participle in all
periphrastic past-tense forms, and also in all periphrastic perfects.
The perfect form corresponding to <Mutila jo zuen> is <Mutila jo du>,
which differs only in that the auxiliary is now present-tense.  This
form translates both English `He has hit him' and English `He hit him'
(earlier today) -- much as in European Spanish.

> I do not see in what way this makes the *diachronic* interpretation
> of "mutila jo zuen" any different from the Hindi preterites that are
> based on Sanskrit constructions of the type "A-Nominative +
> B-Genitive + PPP/nom." meaning earlier "A was (verb)-ed by B", but
> now simply "B (verb)-ed A." Where am I wrong?

Well, in the Indic case, we have several thousand years of texts to
consult, so that we can get an idea how the ergative construction arose.
With Basque, we are not so lucky.

Note also that, in Hindi, as in Indic generally (I think), the ergative
occurs only in the past tense, as is common with ergatives that have
arisen from perfective or passive constructions.  In Basque, however,
the ergative construction is used in all circumstances, without
exception.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk