Fw: Error Condition Re: Re: transitivity, etc,

[Koontz John E]

On Mon, 30 Sep 2002, Shannon West wrote:
> > Whatever principles or diagnostics enable you to select as subject the
> > agent if present, and the patient if not, will probably support the
> > experiencer as subject, too.
>
> Thanks John. I haven't needed to use semantic roles for the most part, and
> likely won't be (gotta work within theory restrictions, and for now that's
> Minimalism).

So, if I may ask, how does one Minimally determine what the subject is for
"ordinary clauses" without double statives?  Is it a question of instinct/
definition, or, as Bob puts it, translating into English and noting what
the subject is there?  I don't ask [just] to be difficult.  I'm actually
curious how this is done.  When I used the phrase "whatever principles of
diagnostics" I actually had this in mind as a possibility.  My feeling is
that *whatever* scheme one uses to identify subject, the experiencer of
experiencer verbs will turn out to be the subject.  In short, if you're
allowed to gloss this over for "ordinary clauses,", your advisors are
going to be perfectly satisfied with glossing it over for experiencer
clauses.  Or, if not, then whatever scheme you use is probably going to
come up subject for experiencers, too.

In any event, even if subjects are effectively givens of a theory there
should be extratheoretical heuristics to identify them, as Keenan made his
reputation by observing.

To air my own dirty linen, the way I do it in Omaha-Ponca is that, by way
of a shorthand for Bob's translational scheme, I define subject as the
agent of transitive and active verbs, and the patient of stative verbs.
In doing so I have neglected the experiencer verbs, of course, or, rather,
I have swept them up with the statives, making that term also cover two
argument verbs that can only have one patient marker, and taking that
argument that can be so-marked as the subject.  These cases turn out to be
what we've been calling experiencer or sometimes, somewhat innacurately,
dative-subject verbs.  I haven't addressed two-patient verbs, because I
haven't elicited or discovered any.

In support of this arbitrary definitional approach, I can point out that
the class of arguments so defined is the same class that takes the
proximate subject articles akha and ama and (as recently illustrated)
governs agent concord in various auxiliaries, and miNkhe/niNkhe/akha/ama
concord in certain others. Of course, there are those awkward cases of
"non-subject" or "obviative subject" articles, in which articles more
typically used with objects appear with subjects, and there are also some
oblique uses of akha/ama, too, but in Dhegiha circles these are both
reasily accepted as special cases, so, I have a heuristic approach or two,
as well.  Unfortunately, the article heuristic won't work for Dakota (or
anything but Dhegiha), though the auxiliary-concord one might.

These heuristics apply within clauses.  I don't at present have any
heristics arising from behavior across clauses (argument identifications,
extractability, etc.), though I suspect such arguments exist and would be
useful in Dakota, too.

I suppose I get away with this because I've largely ignored syntax to
date!

JEK