Capturing Object Sharing in Ditransitives
Recent work at the syntax-semantic interface (Pylkkänen 2002, McGinnis 2001, 2004) has
reached the conclusion that multiple object constructions (applicatives, ditransitives) split into
`high' and `low' constructions, combining insights from previous analyses (see especially
Marantz 1993, Pesetsky 1995). High constructions (HC) express relations between an individual
(AO; applied object) and an event (1), and Low constructions (LC), relations between two
individuals, IO and DO, indirect and direct object, respectively) (2). HC and LC have the Phrase
Structure representations in (3) and (4), respectively.
In this talk I reconsider the nature of LC of the English type (John sent Mary a book), and argue
that the representation in (4) is incomplete. (4) correctly captures the resultative part of the
meaning of LC: the fact that if `John sent Mary a book', Mary got the book. ((4) is very close to a
small clause/possessive DP structure (cf. Harley 1995; 2003)) But (4) fails to express what
Pietroski (2003) calls the transfer part of the meaning of LC: the fact that if `John sent Mary a
book', John sent the book. For this aspect of meaning, DO is clearly an argument of the verb, a
fact that a standard representation like (5) readily captures.
I argue that both (4) and (5) are needed to fully account for the properties of LCs. Specifically, I
propose that (4) and (5) be combined in a way that is very reminiscent of serial verb
constructions: as an object sharing structure. As is well-known, in serial verb constructions (e.g.
(6)), an object has a dual thematic status, a fact that has often been captured in terms of the object
of Verb-1 necessarily binding a phonetically null argument (pro) of Verb-2 (7; see Collins 1997).
Argument sharing is also a property of control structures, which is standardly captured in terms of
binding of an empty argument (PRO). Following Hornstein (1999) I argue that argument sharing
is the result of movement driven by the checking of a theta-feature. Specifically, for LCs, I
propose a derivation like (8). Here, DO starts off in a projection where it thematically related to
IO (this is essentially (4)). DO then moves to SpecVP, where it becomes a direct argument of the
verb (essentially as in (5)). In order for DO to reach SpecVP, it has to cross IO, in apparent
violation of Relativized Minimality (Rizzi 1990). To circumvent the minimality problem I appeal
to Chomsky's (2001) proposal that locality is computed not strictly derivationally, but at the
phase-level (in this case, at the vP-level). Specifically, Chomsky argues that no
minimality/intervention effect will result at the phase level if the potential intervener raises
beyond the landing site of the element whose movement it may block, as schematized in (9).
Applying this reasoning to (8) leads me to conclude that IO necessarily raises beyond SpecVP in
LCs (10). Interestingly, several authors have recently argued for obligatory IO-raising on
independent grounds (scope freezing effect (11), Bruening 2001; ban on (sub-)extraction of
(/from) IO (12), Landau 2005)). I propose that IO raises to receive structural case in SpecvP (an
Object Shift-like operation). This operation is not blocked by the presence of DO in SpecVP if we
assume that DO receives inherent case from V (see Boeckx and Hornstein 2005 for independent
evidence to this effect), and inherent case renders an element inert for attraction/intervention
(McGinnis 1998).
In sum, I argue that the recent treatment of ditransitives of the English kind as LCs is incomplete.
It only captures half of the thematic properties of the construction. To remedy this problem I
argue that ditransitives involve object-sharing, captured via theta-feature-driven movement. I
furthermore claim that in order for this movement to be possible, IO must necessarily undergo
movement, for which there is independent evidence. We thus seem to have both syntactic and
semantic reasons for the position argued here, which reinforces the idea that syntax and semantics
work in unison.
(1) Mukasa ya-som-e-dde Katonga ekitabo [Luganda]
Mukasa 3G.PAST-read-APPL-past Katonga book
`Mukasa read a book for Katonga' (i.e, M read a book and it (reading a book) was for K)
(2) John sent Mary a book
(3) [vP Subj vº [ApplP IO Applº [VP Vº DO]]]
[vP Subj vº [VP Vº [ApplP IO Applº DO]]]
(4)
(5) [vP Subj vº [VP DO [Vº IO]]; alternatively: [vP Subj vº [VP IO [Vº DO]]
(6) M-a da fufu du (Ewe)
1sg-Fut cook fufu eat
`I will cook fufu and eat it'
(7) [M-a [[da fufui] [du proi]]]
(8) [vP Subj vº [VP DO Vº [ApplP IO Applº tDO]]]
*: [Phase X ... [Y ... [W ...[ tY]]]]
(9)
|_____*____|
OK: [Phase W X [Y ... [tW ... [tY]]]]
|__________|
(10) [vP IO [Subj vº [VP DO Vº [ApplP tIO Applº tDO]]]]
(11) John gave some girl every candy (some > every; *every > some)
(12) a. ?*Who did John send a medal
b. *Who did John give [friends of __] a medal
References
Bruening, B. 2001. QR obeys superiority: ACD and frozen scope. LI 32:233-273.
Chomsky, N. 2001. Derivation by phase. In Ken Hale: a life in language. MIT Press.
Collins, C. 1997. Argument sharing in serial verb constructions. LI 28, 461-497.
Harley, H. 1995. Subjects, events, and licensing. PhD, MIT.
Harley, H. 2002. Possession and the double object constrcuction. Linguistic Variation Yearbook
#2, 29-68.
Hornstein, N. 1999. Movement and control. LI 30, 69-96.
Landau, I. 2005. The locative syntax of experiencers. Ms., Ben Gurion University.
Marantz, A. 1993. 1993. Implications of asymmetries in double object constructions. In Aspects
of Bantu Grammar. CSLI.
McGinnis, M. 1998. Locality in A-movement. PhD, MIT.
McGinnis, M. 2001. Variation in the phrase structure of applicatives. Linguistic Variation
Yearbook #1, 101-142.
Pesetsky, D. 1995. Zero syntax. MIT Press.
Pietroski, P. 2003. Small verbs, complex events. In Chomsky and his critics. Blackwell.
Pylkkänen, L. 2002. Introducing arguments. PhD, MIT.
Rizzi, L. 1990. Relativized Minimality. MIT Press.