Re: THEORY: Active case-marking natlangs

J Matthew Pearson <pearson@...> writes:

> The proper generalisation for Tokana seems to be that it is active (a la
> Smith-Rosta), but with quasi-semantic (i.e., 'fluid') case-marking of the
> S/A.
> More explicitly, Tokana has an active alignment (in Marcus's sense,
> where alignment
> of grammatical relations is in principle separable from case-marking),
> and within
> that active alignment, the case-marking of the S/A function is partially
> dependent on animacy/volitionality and definiteness:
>
>    Grammatical   Case-marking
>    relation
>
>    P/O           absolutive
>
>    S/A           nominative (if definite and volitional)
>                  ablative (if indefinite and/or animate non-volitional)
>                  instrumental (if inanimate)
This is pretty much similar to how I interpret the case marking in
Nur-ellen.  It distinguishes S/A from P/O by case marking, and within
the active alignment, S/A arguments receive different cases dependent on
animacy and volition:

 Relation   Animacy      Degree of volition                 Case

 S/A        animate      volitional                         agentive
 S/A        animate      accidental                         dative
 S/A        animate      under external force               instrumental

"S/A"       inanimate    ---                                instrumental

 P/O        any          ---                                objective

Technically, an S/A must always be animate; the inanimate instrumental
"subject" is actually an oblique complement on an impersonal sentence
(i.e. one where the S/A slot is left empty).

Jörg.