Reflexives as subjects?

[Miriam Butt]

Balthasar Bickel wrote:

>For clarification:
>in her 1994 book, Mohanan claims that BOTH the derived subject and the
>agent (or 'logical subject') can control reflexives (following Subbarao
>1971 and many others since), and I think that's essentially right (I

Yes, I think that is essentially right as well.

But that still does not invalidate my comment on this Gujarati example.

>> > Raaj-thi potaane  naa vakhaaNaayo
>> > R-INSTR  REFL:DAT not praised:PASS
>> > 'Raaj could not praise himself'

Frans Plank was kind enough to lend me the volume the article by Mistry is in,
where he talks about this example.  Mistry says:  "pote coindexes with
dative subjects in a dative construction (39a), in a 'have' construction
(39b), and with the instrumental subject in a passive construction (39c)."
(Mistry 2000:352)

The latter is the example above.  I do not think he is talking about
logical dative subjects (is there such a thing?) and logical instrumental
subjects here.  I take him to be talking about "real" subjects here (in so
far as the notion subject can be real), also because he doesn't make any
mention of logical subjects in the previous part of his article.

So, I want to repeat my point that this construction is probably not a
"real" passive (in so far as the notion of passive is real in any case in
the South Asian context).  This is supported by the un-passivelike
semantics that appear to accompany this construction and Mistry's own
statement that the instrumental is a subject (and not a suppressed agent
oblique).

Just because languages make use of the notion of logical subject does not
mean that every instance of an instrumental NP must necessarily be analyzed
as one:  it could be an oblique that was not a logical subject, or it could
be a subject that was also a logical subject, or it could only be a logical
subject.  All possibilities exist and each construction must be examined
carefully on its own terms for which of these is true.

Miriam