Message-Id: <v03007800aea2fcfdb157@[18.162.0.29]>
In-Reply-To: <9611011627.AA15105@emunix.emich.edu>
Date: Sun, 3 Nov 1996 21:40:14 -0500
To: Linguistics Conference <LINCONF@tamvm1.tamu.edu>
From: David Pesetsky <pesetsk@MIT.EDU>
Subject: Two comments on KEYNOTE ADDRESS BY H. LASNIK
Two remarks for Howard:
Remark 1
=========
You reproduce Uriagereka's example:
(15) There arrived two knights on each other's horses
and you consider, in effect, three possible accounts of the binding
possibility:
(A) "Two knights" occupies a VP-internal position and the recipocal is
lower than that position (Larson);
(B) The reciprocal is higher than "two knights", but lower than a position
to which some features of "two knights" has moved. The position of the
formal features counts for binding (Uriagereka)
(C) "Two knights" has moved overtly (i.e. not just its features) to a
position higher than the reciprocal (extension of Lasnik & Saito).
You argue effectively against (B), but I don't see the argument against
(A). The main point seems to be that Larsonian shells do not tell us why
an ECM subject behaves for binding as if it were an upstairs object (the
Lasnik and Saito data). True enough -- but your hypothesis does not (in
turn) address binding phenomena that are the stock-in-trade of Larsonian
shells, e.g.:
(X) Mary gave candy to the children on each other's birthdays.
[See also my Zero Syntax and Colin Phillips' MIT dissertation for more
discussion.]
So isn't the proper analysis of Uriagereka's example (overt raising vs.
Larsonian configurations) underdetermined by the evidence? Aren't the new
data of your current paper merely *consistent* with your hypotheses, rather
than furnishing an argument for them over the Larsonian alternative?
You seem to acknowledge this in your careful phrasing:
> However, several
> of the paradigms examined by Lasnik and Saito (1991)
> cast doubt on this Larsonian approach as a general
> solution.
-- but your proposal is also not a "general solution", right? Since
Uriagereka's datum is explainable both in terms of shells and in terms of
object raising, and since shells and object raising explain partly disjoint
sets of data, the datum doesn't seem to furnish an argument for your
proposal over the Larsonian alternative. Or am I missing something crucial?
Remark 2
=======
In footnote 1, you write about Belletti & Rizzi's "anywhere" proposal for
Principle A in the context of Larson's theory of double object structures:
> (iii) I showed the professors each other's students
> (iv) *I showed each other's students the professors
You note, correctly:
> Larson argues at great length for an analysis of double
> object constructions in which the first object begins
> lower than the second but raises to a position that is
> higher. The grammaticalitity of (iii) is thus
> straightforwardly explained on Larson's assumptions.
> However, as noted by Hoekstra (1991), the account
> falsely predicts symmetrical, rather than asymmetrical,
> binding. Licensing of the higher surface
> reciprocal in (iv) could have obtained in the earlier
> structure where it was actually lower than its
> antecedent "the professors".
But notice that Larsonian shells will offer an account of (iii) even if
Larson's particular proposal about double object structures is wrong.
In this context, it is interesting to note that the Belletti and Rizzi
proposal makes the *correct* predictions in a shell context if we replace
Larson's derivation of the double-object alternation with its *opposite*.
If the GOAL starts out higher than THEME (and not the other way around), we
expect the judgments in (iii) and (iv). The GOAL is not lower than THEME at
any stage of the derivation.
The surface constructions in which THEME is higher than GOAL would then
have to be derived by movement. That, of course, is where the symmetry
shows up, as noticed by Burzio (cited by Belletti and Rizzi; the data are
also discussed in my Zero Syntax):
(Y) I showed the professors to each other's students.
(Z) I showed each other's students to the professors. [?]
Aoun and Li offered scope evidence that dovetails with the binding evidence
in (Y) and (Z) [LI circa 1990, I think].
P.S. Examples like (Z) are generally more easily accepted if you replace
'students' with an inanimate (though this makes full paradigms impossible
to construct), e.g. "I gave copies of each other's books to the professors".
--David Pesetsky