Date: Mon, 04 Nov 96 21:24:59 EST
From: Howard Lasnik <LASNIK@UConnVM.UConn.Edu>
Subject: comments on Rappaport's comments on Lasnik
To: linconf@tamvm1.tamu.edu
Message-Id: <961104.212638.EST.LASNIK@UConnVM.UConn.Edu>
In-Reply-To: <9611031635.AA06787@dragon.emich.edu>
Rappaport offers a nice summary of my paper, and then
raises a number of interesting questions about the
architecture of the theory. With respect to the summary,
I would just add one more point: that direct objects and
ECM subjects behave for binding purposes as if they had
overtly moved, so I conclude that they have, in fact,
raised overtly to Spec of Agr(o), parallel to the raising
of the associate of an expletive.
As to the question of where the binding conditions are
satisfied, for the kinds of phenomena I discussed, LF can
uniformly be the relevant level. The major problems for
this - Chomsky's Condition C effects with r-expressions
inside WH-in situ or QP, and Barss's Condition A effects
in similar circumstances - disappear if LF movement is
strictly of formal features, while binding involves other
properties (in addition?). That is, in relevant respects,
LF will not be different from S-Structure. (This is the
conclusion that Rappaport apparently draws from my
argument, and it is the one I intended.)
Rappaport raises a question about the Barss type examples,
such as my
(8) John wonders which picture of himself Mary showed to
Susan
With Barss, I find it quite acceptable. For those who
reject it, there would seem to be some sort of forced
'reconstruction' at work. Chomsky's proposal of a few
years ago about minimizing the restriction in an operator
(which Rappaport alludes to) could be relevant. And since
that constraint is rather arbitrary, if it exists at all
it might be a likely place for individual variation.
Beyond that, I really don't have anything to say.
Rappaport is exactly right that the discussion immediately
following my example (44) should refer to (43), not to
(44), since the former is the one with the in situ WH-
phrase.
Finally, I assume that Case is checked in Spec of Agr(o)
in just the way Chomsky first proposed: V raises to
Agr(o), and the complex does the checking. That was how
Chomsky distinguished verbs that did or didn't have an
Accusative Case feature. That account straightforwardly
extends to the range of more idiosyncratic Case features
of verbs. As far as I can tell, any distinction that the
head-complement relation in V' would have made is still
available in the Spec-(complex) head configuration in Agr
on this view.