Jörg Rhiemeier wrote:

>Another question about active languages: How typical is it for active
>to bar inanimate nouns from filling the A slot?

I couldn't say.  The active languages I have most experience with do not
have agreement with the third person, so it would be impossible to tell if
inanimates are allowed as A or not.  (Without a deal of abstraction and
theorizing, of course).

>   In Nur-ellen, inanimate
>nouns and pronouns have only the objective case, but no agentive case.
>And I seem to have read that in Dakota, the agent must also be animate,
>but I am not entirely sure.

Mohawk appears to be the same in typical sentences, but under certain
conditions, you can get inanimate A's (but not inanimate agents).  Here's
an example:

Wa-w-ate-nohare-' ne atya'tawi
FACT-NsA-SRFL-wash-PUNC ne dress
'The dress became clean.'

The NsA means neuter (the gender of 'dress') singular "agent".  (Nobody
really knows what 'ne' means -- lots of theories, all with holes).

>  My feeling about active
>is that inanimate A's are a no-go, because an inanimate item cannot
>be an agent.

But you can have non-agent A's, as I just showed in the Mohawk example.

And what about cases where the agent is in the P slot?  It does
happen.  The Mohawk words for "throw" and "bump" require that their agents
be marked as P's, and that their patients are not marked at all -- often
they get incorporated into the verb.

>   That's why inanimates in Nur-ellen have no agentives, and
>because some linguist once said anything about it.

Good solid reasoning, IMO.  That is the way your language works, so that is
the way it should be.  Not all languages fit into the
accusative-ergative-active paradigm that everybody always talks
about.  Someone would have to come up with very good arguments to convince
me that the Algonquian languages are any of those three.

>Yes, just about every single active language I have seen does it
>it seems.  And are there active natlangs which grammaticalize more than
>degrees of volition?

Yup.  Muskogean languages differentiate the person experiencing the verb
from the agent and patient.  To give the most common Chickasaw example:

chokma 'good'
chokma-li 'I act/behave good.'
sa-chokma 'I am good/nice'
am-chokma 'I feel good.'

This division is also true of Choctaw and Alabama.

>And that's why so many Amerindian languages have switch reference,
>in order to disambiguate the situation.  If I understood it correctly,
>that is.

There is no correlation between active marking and
switch-reference.   Muskogean languages have both.  Iroquoian languages
have activity but not SR.  Yuman, Hokan, Eskimoan, and Uto-Aztecan
languages have SR but not activity.  Souian languages have activity, but I
don't know about SR.

Marcus Smith
AIM:  Anaakoot
"When you lose a language, it's like
dropping a bomb on a museum."
   -- Kenneth Hale