On Fri, 5 Nov 1999, Daniel Andreasson wrote:

> Hello all.
> Once again I'm thinking about implementing a really
> cool feature in Rinya, but as usual it's so weird and
> against everything that I've written on Rinya so far
> that it seems impossible to go through with it.
> But I still want to try it and see what you guys
> think of it. The idea comes from an article by Thomas Payne.
> Here goes:
> Imagine a 'natural' hierarchy of 1st, 2nd and 3rd persons.
> Like this: 1 > 2 > 3prox > 3obv
> The third person is split in two: proximate and obviative.
> If two 3p interact, the proximate is the topic.
> Ok. If a person to the left of another person in the
> hierarchy acts on the person to the right...
> e.g. 1 acts on 2. 'I look at you'
> ...then the normal use of the cases would be used.
> il-in   yulo     im                 (i)
> 1sg-ERG see:PRES 2sg:ABS
> 'I look at you'
> But if a person acts on a person to the left of the first
> one, then you use an inverse form of the verb. Like this:
> il-in   yulo-vo      im             (ii)
> 1sg-ERG see:PRES-INV 2sg:ABS
> 'You look at me'
> That is, the '-vo' suffix inverses the cases.
> One more example:
> thel    yulo     e-gedh-ando        (iii)
> dog:ABS see:PRES OBV-bird-ABL
> 'A dog sees a bird'
> thel    yulo-vo      e-gedh-ando    (iv)
> dog:ABS see:PRES-INV OBV-bird-ABL
> 'A bird sees a dog'
> Examples (iii) and (iv) illustrate several things.
> 1) The most problematic thing. Rinya is an active
> case marking lang. It marks semantic roles with
> cases, and volitionality. The dog is non-volitional
> and therefore is in the absolutive case. The thing seen,
> (the bird) then is in the ablative case. This may cause
> confusion when the roles are suddenly reversed.
> (That is also why examples (i) and (ii) are translated
> 'look at' since '1sg' is volitional and in the ergative.)

Hmmmm. I've always associated inverse marking with languages
with a hierarchical argument structure (Georgian notwith-
standing), and you're proposing just such a hierarchy (1 > 2 > 3
> 3'). Languages to look at for ideas are Algonkian, Yuman, and
Navajo. All of these languages use something like inverse
marking as well as the proximate/obviative distinction, even if
it isn't called that in the sources (though I believe this is
the conventional term in Algonkian studies; in Navajo studies
the obviative is referred to as the 4th person). None of these
languages case-marks nouns.

AFMCL (you knew this was coming :-), Tepa has inverse marking as
well. The person hierarchy is different from the one you're
proposing--I used 2 > 1 > 3 > 3'--but the mechanics are the
same. Argument structure in unmarked transitive sentences is
understood as obeying this hierarchy. Any changes to this
hierarchy must be marked by the inverse grade of a verb (which
is formed by infixing /l/ in the verb stem).

> 2) The speaker decides that the dog is the topic
> and marks 'bird' with an obviative prefix. When
> the bird then (iv), which isn't the topic, sees the dog,
> one must use the inverse suffix on the verb.

The proximate/obviative distinction is useful for tracking
participants through discourse--that is, across sentence
boundaries. It doesn't get a chance to shine in isolated

> 3) Maybe the verb should be first in the sentence
> so the listeners understands right away what is
> going on. Rinya is usually OVS, (though not in
> the examples above, since that probably would cause
> even more confusion) so if I make it VOS in
> inversed sentences, maybe I don't even need a suffix
> on the verb. The inversity is marked with word order.

Tepa is VOS, but that was an early decision which was not
influenced by the later appearance of proximate/obviative.

> Since this crashes with the active case marking, I
> thought I might just use it when two 3p interact.

Again, it seems to me that inverse is incompatible with case
marking of any stripe. (I wish I knew more about Georgian!)
Maybe you could make it work, though!


Dirk Elzinga
[log in to unmask]                 "All grammars leak."               -Edward Sapir