complements

[Joan Bresnan]

In response to TAM's (=Tara, Alex, Mohanan's) lateest posting, I would
like simply to point out that when a theory changes, the meanings of
the theoretical constructs implicitly or explicitly defined by it also
change.  For example, lmt changed the meanings of the functions
subject, object, secondary/restricted object, and oblique at one fell
swoop (by redefining "semantically restricted" [r], as discussed at
length in earlier postings involving Lori Levin).  O2 is not the same
as O-theta; likewise, OBL-theta differs substantially under the old
and new theories, even though its nomenclature is unchanged.
In the same way, as TAM  point out, the meaning of "complement" is changed, if
it now includes both XCOMP and COMP.  But the simple fact is, that
there *are* non-object complements other than S, which (like S) have
not yet been integrated into lmt (except for the proposal of Zaenen
and Engdahl I referred to earlier).  Thus TAM's various proposals for
dealing with S (such as calling it an OBJ and treating its non-OBJ
properties with features at other levels) do not have the desirable
effects of eliminating redundancy that they suppose.