accusative and ergative languages

[Jens Elmegaard Rasmussen]

On Thu, 10 Jun 1999, Rick Mc Callister wrote:

> Can you give examples?

> asked in response to my posting:]

>>There are some quite obvious counterexamples: Modern Indic is very close
>>to being ergative; Kurdish too; Lithuanian is well on its way; and
>>Eskimo has accusative remains in the pronouns.

There are others on the list better informed than me, but I'll try, basing
myself on some handbook aterial:

1. In Hindi the perfect preterite has the structure

   A-obl. ne O Vb-(gender/number of obj.)

e.g.: lar.ke ne na:rangi: kha:i: "the boy ate the orange",

where lar.ke is the oblique sg. of lar.ka: 'boy', the oblique case being
obligatory before the postposition _ne_ marking the agent; and the verb is
inflected in the fem.sg. ('orange' is fem.; the masc. is kha:ya:); the
object is unmarked. - Intransitive verbs agree with the subject, e.g.
lar.ka: a:ya: 'the boy came'.

2. Lithuanian has expressions like

   s^itas arklys mano pirktas 'this horse has been bought by me'

(Quoted from Senn: Kleine lit. Sprachlehre, Heidelberg 1929, 107). Since
gen. + nom. is also the general syntgm for 'have', this is pragmatically
congruent with 'I have bought this horse'. Schmalstieg has written
extensively on the subject.

3. Kurdish is much like Lithuanian. J.Blau, in Compendium Linguarum
Iranicarum 331, gives the pair

   mirov-ek-i^ hesp di^t-0 "un homme a vu le cheval"
   kec,k-ek-e^ hesp di^t-in "une fille a vu les chevaux"

-ek- is 'one, a'; -i^/-e^ is the izafet particle of belonging in the masc.
and fem. sg., oblique case (old genitive, I take it); the object is
unmarked; the verb is an old participle in -ta-, in the sg. with zero for
understood 'is', the the pl. with -in 'are'. The original meaning thus
appears to be "(it is) a-man-of-whom the-horse (is-)seen" and "(it is)
a-girls-of-whom the-horses are-seen". Note the number concord in the
auxiliary with the object (which then dispenses with plural marking
itself). Intransitive verbs have agreement with the subject.

4. Eskimo at least has a concept of subject exactly like ours in their
selection of reflexive as opposed to 3rd person, in that the reflexive
always refers to the subject, be it of a transitive or an intransitive
verb, never to the object. With certain semi-pronouns like Greenlandic
tamar-ma '(I) whole', tamar-pit '(you) whole', this means that the
non-reflexive 3rd sg. tama-at '(him) whole' never refers to the object,
while the reflexive tamar-mi '(himself) whole' refers to the subject,
irrespective of the transitivity of the verb, and so the opposition
tama-at : tamar-mik is pragmatically like acc. vs. nom. - Also the variant
forms of the plural of demonstrative pronouns, Greenlandic uku / uku-a
'them'/'they' are reported to have been earlier distributed according to
this parameter, the longer form being nominative (while the later tendency
is rather to reserve the longer form for the ergative/genitive role). The
nom./acc.pl. of pronouns is no pervasive cross-dialectal opposition in
Eskimo, but the reflexive selection according to subject role seems quite
fundamental and certainly reveals an old thinking in terms of quite
traditional categories.

Jens

Date: Fri, 11 Jun 1999 17:49:44 +0200 (MET DST)
From: Jens Elmegaard Rasmussen <jer at cphling.dk>
Subject: Re: accusative and ergative languages
In-Reply-To: <Pine.SO4.4.05.9906111701400.479-100000 at rask>
Message-ID: <Pine.SO4.4.05.9906111742580.479-100000 at rask>

On Fri, 11 Jun 1999, I just wrote:

> [T]his means that the [Greenlandic]
> non-reflexive 3rd sg. tama-at '(him) whole' never refers to the object,
> while the reflexive tamar-mi '(himself) whole' refers to the subject.

I meant of course that the former never refers to the subject, while the
latter always does that.

May I add that the ergative systems exhibited by these languages, plainly
based on old passive reconstructions with participles, are - when not
specially marked - by virtue of their derivatory history quite naturally
restricted to perfective statements. This explains the tense-split pointed
out in Alexander Nikolayev's posting.

Jens