subjects and comps

[Miriam Butt]

With regard to the further discussion contributed by Alex, Tara and Mo
on the question of whether sentential subjects and objects should be
represented as SUBJ/OBJ or COMP/XCOMP --- here my two cents (not to
worry, I'm not starting on OBJ vs. OBJtheta again).

I came across this problem while working on complex predicates vs.
control constructions in my dissertation.  Essentially, what I did
there was to assume that XCOMPs followed the same linking rules as
OBJ, but that they were represented as Ev(ent) at the level of
argument structure, not as th(eme).  I didn't go so far as to try to
tackle sentential subjects. 

However, my approach essentially follows the idea the Mohanans and
Alex Alsina had, but might perhaps allow one to avoid stating
subcategorization in terms of syntactic categories like S, NP, PP,
etc.

On the other hand, Peter Sells did argue somewhere on the basis of
Japanese that syntactic subcategorization information like S, NP, PP
should in fact not be left by the wayside, as is often done, but
is in fact crucial for an understanding of the entire pattern of
subcategorization.

Miriam