Matt Pearson <[log in to unmask]> wrote:

>I like it!  Some questions:  (1) Is case 3 used only for 'canonically'
>non-volitional agents, or for *all* non-volitional agents?  Can you
>ever get case 2 and case 3 co-occurring?  In other words, if the
>cat accidentally killed the mouse, which of the following would
>you get?
>        cat.1 mouse.2 kill
>        mouse.2 cat.3 kill
>        mouse.2 cat.4 kill
>        cat.3 mouse.4 kill

Canonical non-volitional agents. The verb is conventionally associated
to a pattern. Not so arbitrarily as, say, prepositions to certain verbs
in IE langs, but arbitrarily based in one core meaning: 'kill', for
example, implies actual intent. So it would be

    cat.1 mouse.2 kill  'the cat killed the mouse'
    mouse.2 cat.4 kill  'the mouse was killed by the cat'

If it was unintentional, you need an adverb to clarify, or maybe another
verb. ('Non-volitional' and 'volitional' are probably not good labels;
if anyone wants to suggest another ones, be my guest. Not doing something
on purpose doesn't mean one isn't the agent; even if the cat didn't want
to kill the mouse, the non-volitionality of this action is not equal to
the non-volitional act of seeing or hearing, for example -- the cat *did*
something (maybe just playing) which resulted in the mouse being killed.

(Now I'm trying to resist the idea of adding another case to the scale
for this kind of actions... I hate my mind when it does these things.)

>(2) Is there a semantic difference between the following sentences,
>other than which noun phrase is the topic?
>        cat.3 mouse.4 see
>        mouse.2 cat.3 see

Hmm... It seems to me that the first one focuses (in the non-technical
sense) equally on the topic (the cat) and the *verb*, while the second
one emphasizes the patient/topic (the mouse) only. That's my impression
so far; I could be wrong, and maybe this is only used to mark the topic
and for syntactic purposes (like coordination):

        mouse.2 cat.3 see it.4 eat
        'The mouse was seen by the cat and eaten by it.'

since you can't do this:

       * cat.3 mouse.4 see it.2 eat
        'The cat saw the mouse and ate it.'

Wow, I just discovered that! Thank you! :)
In such phrases, the two verbs must agree with the topic (of
the first one) but hierarchy must be preserved if want to
have ellipses like these. Hence if you 'expand' the ellipsis
in example 2, you would have

       * cat.3 mouse.4 see cat.3 it.2 eat

which is doubly wrong (case 3 before 2, and used with a
volitional verb). (And 'cat.3' *must* be first, if it's
the topic...)

>A suggestion:  Use case 2 to denote objects which are somehow
>'affected' by the action of the verb (i.e. true patients), and use
>case 4 to denote objects which are not (completely) affected by
>the action.

That's a good idea. Maybe it was what I was thinking above,
though I'm not sure.

>(3) I assume that case 3 is used to express goals (motion
>towards) while case 4 is used to express sources (motion
>from).  But which case is used to express locations?
>How are instruments handled in this system?

Same cases (the lowest ones, 3 and 4) with or without adpositions
(I think post-). It's not an accident that case 4 is also known
as oblique; it may be used as a catch-all case (much like _ni_
in Japanese, AFAIK). I'm trying not to add more cases to the
system (no matter if they're outside the hierarchy); what did
Latin do with locations?

--Pablo Flores
  ... I cannot combine any characters that the divine Library
  has not foreseen, which in some of its secret tongues do not
  bear some terrible meaning. No-one can articulate a syllable
  not filled of caresses and fears; which is not, in some one
  of those languages, the powerful name of a god...
                   Jorge Luis Borges, _The Library of Babel_