Raising Expletives
|
|
Jacek Witkoś (Poznań) |
[This paper is a report on a joint
project with Norbert Hornstein, UMD: Yet Another Approach to Existential
Constructions, to appear in Festschrift for Anders Holmberg.]
Existential constructions (ECs)
display several contradictory properties, repeatedly exploited in numerous
generative approaches (e.g. Chomsky 1986, 1995, 1999; Lasnik 1995, 1999 and
Bošković 1997).
For example, the
relation between there and someone/a beer in (1a,b) shows the same
locality properties as a chain of movement between someone/a beer and the trace in (2a,b).
(1)
a.
*There
seems that someone is in the room.
b.
*There
is the man drinking a beer
(2)
a.
*Someone
seems that t is here.
b.
*A
beer is the man drinking t
The associate acts ‘as if’ it were in
the position of the expletive as regards agreement.
(3)
There
is/*are a man in the room
b.
There
*is/are dogs in the park
There is a one to one correlation
between expletives and associates. And there is the well-known definiteness
effect. These well known facts all point to the same conclusion; that the
associate and expletive form an (A-)chain at some point in the derivation; a
standard implementation assumes that a LF structure like (4b) underlies (4a).
(4)
There
is someone in the room
b.
[There+someone
[is [someone [in the room ]]]]
Yet, other data indicate
that the associate is interpreted at LF in its overt position. For instance Den
Dikken (1995) shows that in (5a), many
people scopes under negation, in (5b) under the modal, in (5c) under seems, and (5e) does not license ACD
ellipsis that is licensed in (5d):
(5)
There
aren't many people in the room.
b.
There
may be someone in the room.
c.
There
seems to be someone in the room.
d.
John
expect someone that I do to be in the room.
e.
*John
expects there to be someone that I do to be in the room
Consider another interesting fact.
Specifiers of associates are less adept at binding than are specifiers in
"regular" DPs, e.g. the
binding indicated in (6a,c,e) is not possible in (6b,d,f). Why not?
(6)
Yesterday,
someone's1 mother was saying that he1 was handsome.
*Yesterday,
there was someone's1 mother saying that he1 was
handsome.
When
I walked in, nobody's1 father was talking to him1
When
I walked in, there was nobody's1 father talking to him1
Nobody's/Somebody's1
father was kissing his1 mother.
*There
was nobody's/somebody's1 father kissing his1 mother.
More facts pointing in the same
direction concern defective agreement patterns in Ecs. They are not identical
to what we find in their non-EC counterparts, e.g. we can find less than full agreement in (7a,b) nut not in
(7c,d):
(7)
(?)There
seems to be men in the garden.
There
is a dog and a cat on the roof.
*Men
seems to be in the garden.
*A
dog and a cat is on the roof.
Why should this be so?
We would like to resolve these problems
by rejecting the assumption that associates move at LF:
(8)
Expletives
cannot check theta roles.
Associates
never move.
Given this, a questions arises: how
to derive the chain properties of Ecs if associates do not move. Our proposal:
we have a derivation like (9a) for ECs.
(9)
There
is someone in the room.
[There
is [ [there someone] in the room]].
If there first forms a unit with the associate and then distances
itself from the associate by movement, we can retain both the assumption that
the expletive and associate are in a chain relation. The defective agreement
pattern in (7a,b) makes sense if the predicate directly agrees with features of
there rather than those of men or a dog and a cat. More
concretely, let's say that there need
not agree in number with its complement.
If so, when there agrees with
finite T0, it is a default form for number that is manifest.
References:
Bošković, Željko. 1997. The Syntax of Nonfinite Complementation.
Cambridge: MIT Press.
Chomsky, Noam. 1986. Barriers. Cambridge: MIT Press.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT
Press.
Den Dikken, Marcel. 1995. Binding, Expletives and Levels. Linguistic Inquiry 26: 347-54.
Lasnik, Howard. 1995. Case and
Expletives Revisited. Linguistic Inquiry
26: 615-633.
Lasnik, Howard. 1999. Chains of
Arguments. In Working Minimalism, eds
Samuel David Epstein and Norbert Hornstein, 189-215. Cambridge: MIT Press.