Re: Chickasaw (was Re: Argument Structures)

On Fri, 25 Aug 2000, [iso-8859-1] Jörg Rhiemeier wrote:

> It looks like that though being active, Ch. does not deny the agentive
> case
> to inanimate nouns.  I always had the impression that it is typical of
> active languages to only allow animates in this role.  In Dakota, for
> example,
> there is never a personal agreement marker if the corresponding NP is
> inanimate, and if one of two NPs in a transitive sentence is animate and
> the other is inanimate, the animate NP is taken as subject no matter
> whether it goes first or not; in my conlang Nur-ellen, inanimate nouns
> have no agentive case.
Chickasaw actually has no agreement marker for 3rd person at all.  There
is even evidence that they are not phonetically null morphemes.  They just
plain don't exist.  Animate and inanimate does not play a role in this
aspect of the grammar.

> From an active language with number verbs, one would rather expect that
> these number verbs behave like stative verbs ought to: treat their
> argument as P, not as A.  (If I had decided to have numbers as verbs in
> Nur-ellen - I did not -,
> I would surely have had them take their argument in objective rather
> than agentive case - otherwise one couldn't count inanimate things!)
Yes indeed.  This is one reason that I don't believe that active marking
is semantically based, popular view to the contrary.  There are just too
many regular exceptions.

> BTW, I have heard more than once that Chickasaw was NOT active, but
> accusative, and the usage of number verbs you describe looks more like
> this.
Chickasaw has a split system.  Verbal agreement follows an active pattern:
there is no doubt about this at all.  Nominals follow an accusative
pattern.

> Does it perhaps exhibit some kind of arbitrary lexical ergativity (i.e.,
> some intransitive verbs treat their S like an A, others like a P, but in
> a way not obviously related to semantics)?
The semantics of A vs. P in general follows what you would expect in an
active system.  It even displays the common split between control vs.
uncontrolled action.  For example, "to cough" can be A or P subject
depending on whether the coughing was intentional or not.

One peculiarity that doesn't completely match with any other active system
I've seen (not a whole lot of experience in this area though) is the
existance of "dative" agreement.  The dative applicative may license what
will become the subject of the sentence for "experience" verbs, like "feel
good".

A, P, and D(ative) can all be the marking for subjects, an example being
_chokma_ "good".

chokma-li
good-1sA
"I act good"

sa-chokma
1sP-good
"I am good"

an-chokma
1sD-good
"I feel good"

An overt pronoun in these contexts would all have nominative marking, as
we would expect in an accusative system.

  If this is the case, it
> should not be called "active", but merely "lexically split-ergative".
I'll have to look into this a little more.  It is intriguing.

Marcus