Double Object Constructions

Barbara Citko
University of Washington, USA

Joseph Embley Emonds

Palacký University, Olomouc, Czech Republic

Rosemarie Whitney
Independent scholar, USA

1 Introduction
2 Structure of double object constructions
2.1 Flat structures
2.2 Asymmetric structures
2.2.1 VP shell structures
2.2.2 Small clause structures
2.2.3 Applicative structures
3 Dative alternation
3.1 Transformational versus lexicalist approaches
3.2 Dative shift
3.3 Status of the “promoted” object
3.4 Productivity
3.5 Meaning differences
4 Constraints on movement
4.1 Symmetric versus asymmetric passives

Note: This chapter grew out of the enterprise to revise and update the chapter that
appeared in The Blackwell Companion to Syntax, 1st ed. Barbara Citko is responsible
for the revision and update of the chapter, which contains significant portions of the
original chapter (especially in sections 2.2., 3.2–3.4, and 4.2). Barbara Citko expresses
her appreciation for the useful and constructive collaboration with the original authors
who, in turn, are grateful to the new author for having accepted the task of revising and
updating this chapter. As the actual author of the present version, Barbara Citko is alone
responsible for any errors and omissions.
The Wiley Blackwell Companion to Syntax, Second Edition.
Edited by Martin Everaert and Henk C. van Riemsdijk.
© 2017 John Wiley & Sons, Inc. Published 2017 by John Wiley & Sons, Inc.
DOI: 10.1002/9781118358733.wbsyncom109
2 Double Object Constructions

4.2 A-bar-movement restriction

4.3 Scope freezing
5 Cross-linguistic variation
6 Conclusion

1 Introduction

Double object constructions have been a source of quite a few debates and
controversies in linguistic literature. These debates involve their structure and
derivational history, the relationship of double object constructions to ditransi-
tive PP datives, the constraints on movement and interpretation, and, last but
not least, cross-linguistic variation in the availability and behavior of double
object constructions. Strictly speaking, the term double object construction refers
only to a subset of ditransitive constructions. Of the examples (1a)–(1d) below, in a
strict sense of the term, only the English (1a) is a double object construction, since
the two objects are morphologically marked in the same way (i.e., with accusative
case).1,2 The Polish counterpart in (1b), due to the fact that the two objects are dis-
tinctly case-marked (accusative Theme versus dative Goal) is not. Neither is (1c),
in which the Goal is realized with a prepositional phrase. In (1d), the presence of
an extra object (Recipient/Benefactive in this case although many other roles are
possible) correlates with an “extra” piece of verbal morphology, the so-called
applicative morpheme and, on some accounts, English-type double object con-
structions are simply treated as a type of applicatives. Since applicatives are a
topic of an entire chapter (Applicatives by Martha McGinnis), we do not treat
them in much detail here. However, since many interesting issues surround the
relationship between double object constructions and PP datives, we discuss both
of them in this chapter and allude to others where relevant.

(1) a. Pat gave Terry a book.

Polish
b. Jan dał Marii książkę.
Jan.NOM gave Maria.DAT book.ACC
‘John gave Mary a book.’
c. Pat gave a book to Terry.
Bukusu
d. wanjala a-a-kul-il-a omu-xasi sii-tabu.
Wanjala 3SS-TENSE-buy-APP-FR CL1-woman CL7-book
‘Wanjala bought the book for the woman.’
(Peterson 2007, 7)

The purpose of this chapter is to outline the issues that have dominated the research
on double object construction, beginning with Fillmore’s (1965) classic study, and to
review the major approaches to the structure of double object constructions as well
as the most influential approaches to the so-called dative alternation (focusing on
the arguments in favor of and against dative shift). However, given the massive
(and ever growing) literature on double object constructions, the current review
of the literature on double objects does not pretend to be exhaustive; it focuses
 Double Object Constructions 3

on the major issues that have dominated the literature and outlines representative
approaches to these issues. The core questions that will form the basis of the discus-
sion that follows are listed in (2a)–(2d).

(2) a. What is the structure of double object constructions?

b. What is the relationship between the double object construction and its PP
dative counterpart?
c. What are the constraints on double object constructions?
d. Why are double object constructions restricted cross-linguistically?

We proceed as follows. Having outlined the empirical domain of investigation, we

turn to examining the structures that have been proposed for double object con-
structions over the years (section 2). In section 3, we focus on what is perhaps
the biggest (still ongoing) debate in the literature on the topic; the debate concerning
what is referred to as the dative alternation (i.e., the alternation between the double
object construction in (1a) and the PP dative in (1c)). The biggest source of debate
here is whether the two are transformationally related or not. The properties that
have played (and continue to play) a big part in this debate concern meaning dif-
ferences between the two variants and the productivity of the alternation. In
section 4, we turn to the constraints that double object constructions are subject
to, focusing on passivization patterns (section 4.1), A-bar-extraction asymmetries
between the two objects (section 4.2) and scope (section 4.3). And, lastly, in
section 5, we turn to the cross-linguistic variation in the availability of double object
constructions. Since, typologically speaking, double objects are fairly restricted
(e.g., in Siewierska’s (1998) sample of 216 languages only 6 percent had them), this
section focuses on the question of what properties might lead to the lack of a double
object construction in a given language. It also examines the different coding of the
two objects in different languages and variation in the order of the two objects that is
likely not the result of dative shift.

2 Structure of double object constructions

2.1 Flat structures

The structure in (3), proposed by Oehrle (1976) and more recently defended by Culi-
cover and Jackendoff (2005, 109) on the grounds of syntactic parsimony, is simple
yet problematic for both theoretical and empirical reasons (as pointed out as early as
in Czepluch 1982, n. 5).

(3) VP

V NP NP

There are also variants of this general ternary-branching structure that do not
treat the two objects as identical. For example, Czepluch (1982) and Kayne (1984)
analyze the indirect object NP as a “covert” PP with a P that remains empty
throughout the derivation, as in (4) below. Czepluch (1982) proposes that ordering
4 Double Object Constructions

of the PP before the NP in (4) is determined by the requirements of case assignment

and the Empty Category Principle, which forces the dummy P to be properly gov-
erned. The empty P attaches to the verb through reanalysis, allowing the verb to
indirectly transmit case to the indirect object NP.

(4) VP

V PP NP

P NP

Since it is somewhat anachronistic to criticize these early approaches based on their

incompatibility with later theoretical developments, such as connectedness (with its
ban on ternary branching), X-bar theory, or Kayne’s (1994) Linear Correspondence
Axiom, we limit our attention here to the empirical predictions such flat structures
make. Perhaps the biggest empirical issue concerns the symmetric relationship of
the objects with respect to each other (and with respect to the verb). First, the struc-
tures (3) and (4) make incorrect predictions with respect to standard c-command
diagnostics, as shown convincingly by Barss and Lasnik (1986). Their diagnostics
involve anaphoric binding, bound variable interpretation, negative polarity item
licensing, weak crossover, superiority, and the so-called each … the other construc-
tion. Below we illustrate for a subset of their diagnostics. The contrast in (5a)–(5b)
shows that the indirect object can bind an anaphoric direct object but not vice versa,
the contrast in (6a)–(6b) shows that a quantified indirect object can bind a direct
object variable but not vice versa, and the contrast in (7a)–(7b) makes a similar point
with respect to negative polarity item licensing; a negative indirect object can license
a negative polarity direct object but a direct object cannot license an indirect object
negative polarity item. Examples (8a)–(8b) show that a direct object cannot cross
over the indirect object containing a coindexed pronoun.3

(5) a. I sent Maryi a picture of herselfi

∗
b. I sent herselfi a picture of Maryi.

(6) a. I sent every account owneri hisi bank statement.

∗
b. I sent itsi owner everyi bank statement.

(7) a. I sent no one anything.

∗
b. I sent anyone nothing.

??
(8) a. Which account owneri did you send ti hisi bank statement?
∗
b. Whose bank statementi did you send his owner ti?

Second, it is not clear how the flat structure in (4) would capture the relationship
between double object constructions and prepositional datives. Third, it is not
clear how the case on the two objects is licensed. A simple and hence optimal case
theory states that case assigners (such as V and P) assign exactly one grammatical
case to a single, adjacent NP (see Pesetsky 1995, 124). In a flat structure, the
 Double Object Constructions 5

assignment of case to the second NP of the double object construction in (3)

becomes problematic. Chomsky (1981, ch. 3) assumes that the V in this situation
assigns an additional “inherent Case” to one of the objects and the usual structural
case to the other (see also Larson 1988). Structures in which one of the objects is
“buried” within the prepositional phrase, such as the one in (4), (and many of its
variants we will discuss in what follows) do not face this problem, as they main-
tain a direct one-to-one mapping between the number of case-assigning heads
(P and V heads) and the number of objects. And fourth, work on double object
applicative constructions in non-Indo-European languages, first in the framework
of Relational Grammar (Chung 1976; Kimenyi 1980; Aissen 1983) and later in
Government and Binding under the impetus of Marantz (1984) and Baker
(1985; 1988a), revealed many syntactic properties which demonstrate that prepo-
sitionless indirect objects function cross-linguistically as surface direct objects. An
additional set of data involving verb particle constructions (Emonds 1972), which
we will discuss in section 3.3, supports the same conclusion for English. This body
of work taken together, has been taken to mean that the indirect object must
be “promoted” (using Relational Grammar terms) from a deep status of an
“initial 3” (a deep oblique complement) to a “final 2” (a surface direct object) in
the course of a derivation. Flat deep structures of the kind given in (4) obscure this
“promotion.” And lastly, it is not clear how a flat structure captures different
restrictions the two objects might be subject to, such as the fact that for many
speakers of American English, only indirect objects may passivize.

2.2 Asymmetric structures

As we saw in the previous section, Barss and Lasnik (1986) established that the indi-
rect object in a double object construction asymmetrically c-commands the direct
object to its right. Their findings thus suggest the following asymmetric structure:

(9)
NP
Goal NP
Theme

There are many different implementations of this general structure, which we

examine in the sections that follow. Crucially, the issue of whether the indirect
object is higher than the direct object is in principle independent of the issue of
whether double object constructions are derived from ditransitive PP construc-
tions, and, if so, whether this is a lexical or a transformational process. This will
be the focus of section 3. The approaches surveyed here differ with respect to what
kinds of elements mediate the relationship between the two objects; an empty verb
in a VP shell structure, a (null) prepositional head, an applicative head, or a func-
tional head of a small clause. They also differ in whether the two objects in a dou-
ble object construction form a constituent with each other (to the exclusion of the
verb) or not.
6 Double Object Constructions

2.2.1 VP shell structures

Larson’s (1988) structure for double object constructions, which has come to be known
as a VP shell structure, has been very influential over the years, and many subsequent
discussions of double object constructions position themselves in relationship to his
proposal. The structure he proposes for double object constructions is intimately tied
to the derivation he proposes for them, in particular to the claim that a double
construction is derived from a PP dative via a syntactic transformation. We present
the gist of Larson’s proposal (as well as some of the questions it raises) here and come
back to it in section 3 (in particular in section 3.2., where we discuss dative shift).
Larson (1988; 1990) follows the insight of Relational Grammar (see Perlmutter
1983; Perlmutter and Postal 1983; and Perlmutter and Rosen 1984) in promoting
indirect objects to direct object status and demoting the original direct object.
The direct object is generated in the specifier of VP, and the indirect object as a
PP complement inside the lower V in the VP shell structure. The empty head of
the higher VP is a landing site for an eventual V-raising.
(10) VP

V VP
Ø
NP V′
Theme
V PP

P NP
Goal
The double object construction is derived by a passive-like movement of the indirect
object.4 In Government and Binding terms, the dative shift operation absorbs case
from the “object” of the verb in (10). The indirect object moves to the “VP subject”
position (i.e., the specifier of the lower VP), as shown in (11). The verb moves to the
higher V position. After the indirect object moves, it asymmetrically c-commands
the second NP, and receives case from the raised verb. The underlying direct object
becomes an adjunct of the V , in a way that parallels what happens to subjects in
by-phrases in passives.
(11) VP

V′

V VP

NPi V′
Goal
V′ NP
Theme

V ti

(adapted from Larson 1988, 353)

 Double Object Constructions 7

The VP shell structure is in principle independent from dative shift. (12) below, for
example, preserves Larson’s basic proposal about the two objects being selected by
two different V heads (with the lower V moving to the higher one) without deriving
the Goal Theme order transformationally.

(12) VP

V′

V VP

NP V′
Goal
V NP
Theme

In the rest of this section, we present some of the issues that have been raised
for Larson’s structure; we come back to them in section 3.2. For example, in
order to explain how a single verb can essentially assign case twice, Larson
assumes that V assigns both an inherent and a structural objective case.
A typical transitive verb is taken to assign both inherent and structural case
to its direct object. In double object constructions, the cases can be split: the
indirect object receives structural objective case and the direct object inherent
objective case. Within this framework, both cases are suppressed for passive
formation, but only structural case for dative shift. Furthermore, case assign-
ment to the promoted indirect object is tied to V reanalysis; the lower V con-
stituent in (11) is reanalyzed as a complex verbal head, which is what allows it
to assign case. Den Dikken (1995, section 3.2) examines the predictions such a
reanalysis makes, and concludes that it faces problems with respect to the Eng-
lish verb particle constructions. In particular, building on Emonds’ (1976)
observations about the distribution of particles in double object constructions,
Den Dikken points out that Larson’s structure cannot account for the distribu-
tion of particles illustrated in (13). If the particle is reanalyzed with the verb
(and undergoes raising) with it, the result is the marginal verb adjacent order.
It is not clear, however, that the medial placement can be derived on Larson’s
structure.

(13) John sent (%off ) Bob (off ) a package (∗off ).

(Den Dikken 1995, 117)

Bowers (1993, sect. 4) presents a critique of Larson’s proposal to assimilate the

dative alternation to that of a passive transformation, as well as a revised analysis
that circumvents some of the attendant problems. His revised proposal uses a var-
iant of a shell structure to account for both the double object and the prepositional
dative constructions, but considers the relation between them to be akin to a lexical
causative alternation; that is, give with double objects realizes the structure of make
IO own/have DO, with give raising from a lower to a higher V position. In compar-
ison, the prepositional construction with give realizes make IO belong to DO, with the
8 Double Object Constructions

double object verb again raising to the V head of the higher VP shell. According to
Bowers, NPs in the double object construction are positioned in the surface as in
(14), with their grammatical roles given as subscripts. In this structure, all three
arguments in a double object construction are in specifier positions. The four bold
nodes in (14) indicate the positions for lexical items in a simple sentence such as We
sent Mary books. Such a structure permits the indirect object to license a floating
quantifier (as in We sent the boys all a book). It also permits the “outer object” to
be modified by secondary predicates (We sent John the food cold), which is not ordi-
narily allowed for adjunct NPs.

(14) [PrP NPSUBJ [Pr’ Pr [VP NPIO V [PrP tIO Pr [VP NPDO V]]]]]

2.2.2 Small clause structures

A number of proposals preserve the insight that the relationship between the two
objects is mediated by a head, but treat this head not as a verbal head, but as a small
clause head (and the resulting constituent as a small clause). In other words, when
XP in a structure like (15) is not a projection of V, it is customary to refer to it as a
“small clause” (i.e., a clause-like structure lacking many – or nearly all – clausal
properties). The choices for XP can be IP (formerly S) or PP, NP/DP, or a functional
element heading a small clause, such as the Predicational (Pr) head of Bowers (1993)
that we alluded to in the previous section.

(15) VP

V XP

NP X′
Goal
X NP
Theme

As we will discuss in detail in section 3.5 (which focuses on meaning differences

between double object and PP dative structures), several authors have noted that
the double object construction indicates a kind of possession of the direct object by
the indirect object, even though verbs such as offer, sing, and read make this somewhat
doubtful. Partly to express this property, Aoun and Li (1989) propose that double
object constructions are small clauses (SC) of indeterminate category, as in (16),
headed by an empty V of possession. The idea that the head of the small clause con-
tributes the meaning of possession is present in many other small clause accounts.

(16) VP

V SC

NP VP
Goal
V NP
Theme
 Double Object Constructions 9

Kayne (1981; 1984) argues in favor of a small clause structure for double object con-
structions based on the parallels between double object constructions and small
clause complements of verbs like consider or believe (often referred to as exceptional
case marking/ECM verbs).5 Kayne (1984, 134–135) thus argues for assimilating the
constituent consisting of the two objects in a double object construction to a small
clause complement of ECM verbs:

(17) a. John gave [‘SC’ Mary a book]

b. John considered [‘SC’ Mary a genius]

His first argument comes from the behavior of double objects in nominalizations, in
particular from the parallelism between small clause subjects and indirect objects in
this respect. The ungrammaticality of (18b) and (19b) shows that neither can nom-
inalize (Kayne 1984, 152). In this regard, double objects contrast with PP datives,
which do allow nominalization of the indirect object, as shown in (19c).

(18) a. John believes Mary to be a genius.

∗
b. John’s belief of Mary a genius

(19) a. John gave Mary a book.

∗
b. John’s gift of Mary of a book
c. John’s gift of a book to Mary

Kayne (1981, 1984) attributes the behavior of both to the fact that “N cannot govern
across a boundary.” Kayne (1983) also uses the ordering restriction illustrated in
(20)–(21), according to which English intransitive [PP P] must precede [PP P − XP]
(see Emonds 1972):

(20) a The secretary sent a memo out to the committee.

b. They are trying to make John out a liar.

∗
(21) a. The secretary sent a memo to the committee out.
∗
b. They are trying to make John a liar out.

He excludes (21a)–(21b) by means of “recursive small clauses” as in make [YP [XP [NP
John] [X’ a liar]] [Y out]]. Since make is not a sister of XP, the latter’s subject is ungov-
erned, hence not assigned case, hence ungrammatical. However, his discussion pro-
vides no indication of how this small clause structure contributes to meaning. Thus,
the different small clauses for the synonymous pair below would apparently have
no effect on meaning:

(22) a. send [SC a memo out] to the committee

b. send [SC [SC the committee out] a memo]

Additionally, the fact that (22b) is good for many speakers disconfirms Kayne’s
account of (21a)–(21b), since his proposal rules out (22b) on exactly the same
grounds as (21a)–(21b).
10 Double Object Constructions

Finally, consider the secondary predication bring John his food cold. While recur-
sive small clause structures (AP inside YP) might accommodate bring [YP [John]
[Y Ø] [AP [his food] [A cold]]], an identical bracketing could also tolerate ∗make John
a liar out. A solution might consist of imposing an ordering restriction on intransitive
P as heads of SCs, thereby bringing us back to the starting point of Emonds (1972).
Consequently, the restriction illustrated in (21a)–(21b) ends up providing no evi-
dence of small clause structure.
The small clause structure has been defended more recently by Beck and John-
son (2004), Harley (2002), among others, who argue that the small clause head con-
tributes an element of possession, associated with the meaning of the double
object construction. For Harley, a double object construction involves a small
clause headed by the preposition which she dubs Phave. This leads her to correlate
the lack of double object constructions to the lack of Phave in a language (see
section 5 for further discussion of cross-linguistic variation in the availability of
double object constructions). The PP datives and double object constructions have
similar structures (both involving a small clause headed by a prepositional ele-
ment). The two are not transformationally derived, however. The small clause
in PP datives is headed by a different type of a prepositional element, the locative
P (Ploc).
Beck and Johnson (2004) focus on the interpretation of again in double object con-
structions. Constructions with the adverb again are generally ambiguous between
the so-called repetitive and restitutive reading, with the two readings paraphrased
in (23b)–(23c) (see Von Stechow 1996). The two readings are taken to reflect different
attachment sites for the adverb.

(23) a. Terry opened the door again.

b. Terry has opened the door before. (Repetitive reading)
c. The door has been open before. (Restitutive reading)

Beck and Johnson note that double objects are similarly ambiguous, as shown in
(24a)–(24b), which receives a natural explanation on the structure in which the
double object construction is a small clause headed by an abstract HAVE
predicate.

(24) a. Thilo gave Satoshi the map again.

b. Thilo gave Satoshi the map and that had happened before. (Repetitive
reading)
c. Thilo gave Satoshi the map, and Satoshi had had the map (Restitutive
before. reading)
(Beck and Johnson 2004, 113)

The ambiguity leads them to argue in favor of the structure in (25), which allows
again to modify either the small clause headed by the abstract have element or
the VP/vP. The former leads to the restitutive interpretation in (24c) and the latter
to the repetitive one in (24b). This is essentially Harley’s (2002) structure, for whom
it is Phave that heads the small clause.
 Double Object Constructions 11

(25) vP

DP v′

v VP

V PP

DP P′
Goal
Phave DP
Theme

Den Dikken (1995, ch. 3), who follows Larson in arguing that double object con-
structions are related to PP datives, provides a carefully designed and defended
small clause analysis of English double objects as complex PPs, with special atten-
tion to their properties when they occur with particles, as in John sent Bob off a pack-
age. The center of his analysis is that English postverbal particles are “unaccusative”
P heads of small clause constituents. Deep structure PPs for double objects with par-
ticles therefore contain a series of small clauses, as in (26a). In the course of the der-
ivation, the PP moves to the specifier of the higher small clause (SC2 first and SC2
next) and the P gets incorporated into the null copula element (the lower V in (26b)).

(26) a. b.
VP VP

V SC1 V SC1

VP VP

V SC2 V SC2
BE BE + Pj
XP PPi XP

X SC3 tj NP X SC3
off Goal
off
NP PP NP ti
Theme Theme
P NP
Goal

(adapted from Den Dikken 1995, 132)

Pesetsky (1995), however, points out some issues for the proposals that rely on a
small clause structure for double object constructions. In his alternative, the rela-
tionship between the two objects is also mediated by a prepositional element but
12 Double Object Constructions

the prepositional element is not heading a small clause.6 Interestingly, in Pesetsky’s

structure, the prepositional element takes the Theme (rather than the Goal) argu-
ment as its complement. The null preposition subsequently incorporates into the
verb, as in (27b). However, overt applicative realizations of incorporated Ps are gen-
erally homophonous with Goal rather than Theme prepositions (e.g., the Chinese
gei ‘to’, discussed by Zhang 1990); this tendency is a bit surprising given the struc-
tures in (27a)–(27b)).

(27) a. b.
VP VP

V PP V PP

DP P′ Gi V DP P′
Goal Goal
P DP P DP
G[+affix] Theme ti Theme

One of the problems Pesetsky points out for the small clause analysis involves bind-
ing. In particular, he points out that run-of-the-mill small clauses count as binding
domains. By contrast, small clauses consisting of the two objects in a double object
construction do not.7 In (28a) the anaphor cannot be bound by an element outside
the small clause containing it, but in (28b) such binding is possible.

∗
(28) a. The boysi made the girls think about each otheri.
b. The boysi gave the girls pictures of each otheri.
(Pesetsky 1995, 159–160)

2.2.3 Applicative structures

A number of analyses of double object constructions assimilate them to applicative
constructions, even in languages with no overt applicative morphology. In these
approaches, the applicative morpheme is present but can be null. Marantz (1993)
extends the analysis of applicatives in Bantu languages to English double object
constructions.8 For Marantz, the null applicative morpheme is a verbal head and
for Baker, Emonds, among others, it is a prepositional element instead. Baker’s
structure for a Chichewa applicative is given in (29b); the underlying
P incorporates into V and surfaces as an overt applicative morpheme.

(29) Chichewa
a. Mbidzi zi-na-perek-er-a nkhandwe msampha.
zebras SP-PAST-hand-TO-ASP fox trap
‘The zebras handed the fox the trap.’
(Baker 1988a, 229)
 Double Object Constructions 13

b. VP

V PP NP

V P P NP trap
hand -iri ti
fox

(adapted from Baker 1988a, 231)

According to Baker, an empty P cannot govern or assign case to its object, so

this NP must receive case from some other source. He achieves this and explains
many other derived direct object properties of promoted indirect objects by
appealing to his general Government Transparency Corollary (Baker 1988a,
64), which states that if a head Y (here P) adjoins to a higher head Z (here
V), then a phrase otherwise governed by Y becomes governed by Z. Since the
complex V (V+P) in (29b) then governs the stranded NP, the latter is assigned
accusative case and treated as a direct object. Incorporation and Government
Transparency thus together account for the fact that deep oblique NPs can behave
as surface direct objects. Baker’s incorporation is supported by several intricate
properties of the applicative systems he examines. Nonetheless, (29b) appears
to have two serious weaknesses. First, the asymmetric c-command required for
the paradigms of Barss and Lasnik (1986) is exactly the opposite of that in
(26b), as observed in Emonds (1993), Marantz (1993), and Pesetsky (1995). Since
the deep direct object asymmetrically c-commands the indirect object at the end
of the derivation, the binding properties holding between double objects should
be the opposite of what holds for English double objects, or at least the second
“outer” NP should act like it c-commands the first. Second, Baker is left without
a general device by which the deep or “demoted” direct objects in (29b) can
acquire case, or at least “pass” the Case Filter.9 Another issue with the structure
in (29b) concerns the fact that overt applicative morphemes generally surface as
suffixes across languages, as discussed recently by Georgala (2012). Baker
(1988a; 1988b) derives applicatives by head movement of P into V. However, in
a clear majority of well-motivated head movements that give rise to single words,
such as noun incorporation and V-raising, a moved head becomes a prefix (not a
suffix). This discrepancy in morpheme order within a V toward structures of the
complex verbal structure of kind given in (30a) for representing applicatives, with
the null applicative morpheme contributing some aspect of meaning (identified
as +PATH below, following Emonds 1993). Assigning applicative suffixes to
the category V also explains why languages fail to have parallel constructions
in other types of phrasal projections; the complex V structure of (30a) but not
the P-incorporation of (30b) accounts for the lack of applicative nouns or
applicative adjectives, for example.
14 Double Object Constructions

(30) a. b.
[V, +PATH] [V, +PATH]

V [V, +PATH] [P, +PATH] V

send, bake, etc. Ø Ø send, bake, etc.

More recently, both overt and null applicative morphemes have been treated as
occupying their own projection, an applicative head of an Applicative Phrase
(see McGinnis 1998; 2001; Pylkkänen 2002; 2008; Jeong 2007; Cuervo 2003; Paul
and Whitman 2010; Bowers 2010; Citko 2011a; 2011b; Kupula 2011; Georgala
2012; among many others). Since applicatives are treated separately in Applicatives,
we refer the interested reader to that chapter (also to Polinsky 2005; Peterson 2007;
and McGinnis 2008, for general overviews). Let us nevertheless mention one impor-
tant distinction, due to Pylkkänen (2002), the distinction between high and low
applicatives, as it will play an important role in the discussion of constraints on
passive formation (and variation in this respect) in section 4.1 of this chapter.
Pylkkänen shows that with respect to a number of diagnostics, different types of
applicatives behave differently. These differences can be between applicatives in
different languages or between different types of applicatives in the same language.
The differences surface in contexts involving intransitive verbs, stative verbs, and
depictive modifiers. English applicatives are impossible in all three contexts: (31a)
shows that an applied argument (the indirect object) cannot be added to an intran-
sitive verb; this example cannot be interpreted as meaning that Mary ran for the
benefit of John. (31b) shows that the applied argument cannot be added to a stative
verb, and (31c) shows that it cannot be modified by a depictive.
∗
(31) a. Mary ran John.
∗
b. Mary held John the door.
∗
c. Mary showed John a movie interested.

In many Bantu languages, on the other hand, the equivalents of all three are per-
fectly grammatical. Pylkkänen derives these differences from the difference in
the position of the applicative head and proposes that there are two types of appli-
catives; a high applicative head takes a VP as its complement and the indirect object
as its specifier, as shown in (32a). A low applicative takes the indirect object as its
specifier and the direct one as its complement, as shown in (32b). Nevertheless, in
both types, the relative relationship between the two objects is the same, with the
indirect object c-commanding the direct one.
(32) a. high applicative b. low applicative
applP VP

DP appl′ V applP

appl VP DP appl′

V DP appl DP
 Double Object Constructions 15

Pylkkänen further shows that this difference in structure is also reflected in the
semantic interpretation of the two types of applicatives; a low applicative head
denotes transfer of possession, whereas a high one is compatible with a wider range
of interpretations (Benefactive, Instrumental, Malefactive, etc.). English double
object constructions involve a low applicative structure and the ungrammaticality
of the examples in (31a)–(31c) follow from this structure. First, since low applica-
tives describe transfer of possession between two internal arguments, they are
not possible with intransitive verbs (which lack one argument). Second, low appli-
catives are incompatible with verbs that describe states (rather than events) because
such verbs are lexically incompatible with the idea of transfer. And third, low appli-
catives are incompatible with depictives modifying indirect objects because the two
are of incompatible semantic types.
The idea that double object constructions involve an applicative structure is not
without its critics. For example, Larson (2010) points out that dissociating the indi-
rect object from the verb (in Pylkkänen’s low applicative structure the indirect
object is not related to the verb) incorrectly predicts that (33a) should entail (33b).

(33) a. John wrote that letter and Bill gave Mary that letter.
b. John wrote Mary that letter.
(Larson 2010, 702)

Given Pylkkänen’s low applicative semantics, the interpretation of both examples

involves a possession relation between Mary and the letter, and in neither is there
any relationship between Mary and the event of writing the letter. Georgala (2012)
also points out that the low applicative structure makes incorrect morphological
predictions regarding the ordering of the applicative morpheme relative to the verb.

3 Dative alternation

3.1 Transformational versus lexicalist approaches

This section focuses on what is perhaps the most exciting ongoing debate in the
literature on double object constructions, going back to the early work on the topic
(see Fillmore 1965), involving the relationship between the double object and PP
dative constructions. The examples in (34a)–(34b) exemplify Fillmore’s (1965) to-
dative alternation and those in (35a)–(35b) his for-dative pattern.

(34) a. Mary sent a present to each family member.

b. Mary sent each family member a present.

(35) a. Carolyn baked that cake for my children.

b. Carolyn baked my children that cake.

This alternation is referred to as dative alternation, and the simple question at the
core of the debate is whether the two are transformationally related or not. The
dative shift approaches, which we outline in section 3.2 below, take the double
object constructions to be derived from the PP dative constructions. Much early
transformational work assumed that the dative alternation resulted from a move-
ment rule, based on the degree of regularity of its operation and the synonymy of
16 Double Object Constructions

the two alternating structures. A typical treatment (in the introductory text of
Akmajian and Heny 1975, 183–186, 220) proposes the following rule of Dative
Movement (see Herriman 1995, ch. 2 for a survey of earlier traditional and gen-
erative analyses of indirect objects).
(36) Structural description: V NP {to, for} NP
1 2 3 4
Structural change: 1+4 2 Ø Ø

The theory of Lexical Functional Grammar (Kaplan and Bresnan 1982) proposes a
similar “surface” analysis for double object verbs. Two distinct sets of grammatical
functions are assigned to the unified predicate argument structure of Agent, Theme,
and Goal:

(37) a. V (SUBJ) (OBJ) (OBJ)

b. V (SUBJ) (OBJ) (to OBJ)

Alternatively, it has been argued that the English double object construction is not
derived by transformation because of the differences between the two constructions
(differences in meaning, productivity of the alternation, to be discussed in sections
3.5 and 3.5, respectively). Rather, double object structures are directly selected by
lexical entries of verbs. The two NPs are then differently interpreted in the semantic
component. Oehrle’s (1976) dissertation is the first detailed argumentation that
dative pairs result from a lexical alternation.

3.2 Dative shift

One of the most influential generative treatments of dative shift is that of Larson
1988, which we discussed in section 2.2 above. The two structures are repeated
below. The underlying structure that serves as input to dative shift is given in
(38a). Recall that Larson analogizes dative shift to a passive transformation; case
absorption forces the Goal DP to undergo movement (for case related reasons).
The result is (38b).

(38) a. b.
VP VP

V′ V′

V VP V VP

NP V′ NPi V′
Theme
Goal
V PP V′ NP
Theme
P NP V ti
Goal

(adapted from Larson 1988, 353)

 Double Object Constructions 17

The Goal argument becomes a derived VP subject and the Theme gets demoted to
an adjunct status. In this respect, his analysis reflects the common insight in
Relational Grammar accounts, where passive and dative shift are also taken to
be a result of the same fundamental operation, 2-1 or 3-2 promotion.
Many of Larson’s arguments in favor of the underlying structure in (38a) come
from standard constituency tests. For example, Larson (1988, 340) points to the
existence of discontinuous idioms in (39a)–(39c), also discussed by Emonds
(1972), as an argument in favor of the structure in (38a), in which the verb forms
a constituent with the PP indirect object, to the exclusion of the direct object.

(39) a. send X to the showers

b. take X to the cleaners/to task/into consideration
c. throw X to the wolves

The facts concerning idioms are somewhat more complicated. Larson acknowledges
the existence of idioms of the kind given in (40a)–(40c) below, which involve the verb
and its Theme argument to the exclusion of the Goal PP argument (see also Jackendoff
1990; Richards 2001; Harley 2002; Bruening 2010b; among others). These are problem-
atic for Larson’s structure as the idiomatic parts do not seem to form a constituent.

(40) a. give one’s all to X

b. give hell to X
c. give the boot to

Larson suggests that as these idioms do not involve the verb, as the idiomatic mean-
ing is preserved with other verbs (such as get). However, the range of verbs that are
compatible with the idiomatic reading of boot is quite limited, which leads Richards
(2001) to attribute the systematic alternation between idiomatic readings with give
and get and the parallelism between give X the boot/flak/the creep and X got the boot/
flak/the creeps to the following lexical decomposition: get is treated as cause to become
and give as cause to have. In both cases, the idiomatic part is HAVE + the boot/flak/the
creeps.10
Harley (2002) points out another issue for Larson’s argument; the fact that idiom
parts do not undergo dative shift:
∗
(41) a. Lasorda sent the showers his starting pitcher.
∗
b. Felix threw the wolves Oscar.
(Harley 2002, 37)

This is not a general prohibition against moving parts of an idiom, as shown by the
well-formedness of constructions like The headway was made or The cat is likely to be
out of the closet.
C-command and coordination also point toward the structure in (38a). Larson
(1988) shows that PP datives exhibit asymmetries indicative of the direct object
c-commanding the indirect one, as shown below for a subset of the Barss and Lasnik
c-command tests:

(42) a. I presented/showed Maryi to herselfi.

∗
b. I presented/showed herselfi to Maryi.
18 Double Object Constructions

(43) a. I gave/sent everyi check to itsi owner.

??
b. I gave/sent hisi paycheck to everyi worker.

(44) a. I sent no presents to any of the children.

∗
b. I sent any of the packages to none of the children.
(Larson 1988, 338)

And (45) shows that the constituent consisting of both the direct object and the indi-
rect object can be coordinated (with a like constituent). However, as pointed out by
Jackendoff (1990), the coordination facts are compatible with coordination of larger
chunks coupled with ellipsis.11

(45) John sent a card to Bill and a present to Jim.

A quite different implementation of the dative shift operation comes from the
work of Emonds (Emonds 1972; 1993). Emonds (1993) reworks his (1972) proposal
that “Dative Movement” is a structure-preserving transformational interchange of a
direct and indirect object (i.e., “3 to 2 advancement” accompanied by demotion of
the direct object).12 Emonds differs from others in that for him both the shifted and
unshifted variant contains an empty preposition (marked as [P, +PATH]). This null
P is realized as an overt preposition only in the absence of dative shift.

(46) a. V′

V NPi PP

send … [P, PATH] NPj

b. V′

V NPj PPj

send … [P, PATH] NPi

Ø …

This structure-preserving interchange of objects accounts for the synonymy of the

dative alternations and the cross-linguistic similarity between these alternations
and applicative constructions. While the analysis does not decide issues about
proper levels for case assignment, it provides two case assigners for two adjacent
NPs. The analysis promotes indirect objects and thus constitutes a basis for explain-
ing their direct object behavior. Finally, the structures in (46a)–(46b) provide
 Double Object Constructions 19

asymmetric c-command of the second NP by the first, in both shifted and unshifted
sequences, and thus account for the Barss and Lasnik (1986) paradigms. In this
account, the underlying PP configuration in (46a) is taken to universally define
an indirect object, independently of a direct object and the category of the head. Late
or PF insertion of Ps such as to and for in the unshifted version of (46a) is obligatory
because of a general prohibition against unlicensed empty categories. Less straight-
forward is the obligatory licensing of an empty P in (only) the shifted or “inter-
changed” derivation. Emonds suggests that an empty head, for example P in
(46a), is coindexed with its sister in underlying structure. The structure-preserving
interchange of NPs, while respecting his Projection Principle, then yields the derived
configuration (46b), where i is the index of the deep direct object and j of the deep
indirect object. His claim is that no preposition can be inserted “late” in this shifted
derivation because a general condition on PF-insertion of a head requires non-
distinct indices between heads and complements.13
The resulting coindexed structure (46b) also accounts for the so-called Restriction
on A -Extraction (which will be the focus of section 4.2). If the NPj in (46b) subse-
quently moves to an A -position, both its trace and the empty-headed PPj are bound
by the same operator. This configuration violates Koopman and Sportiche’s (1983)
Bijection Principle prohibiting an A -constituent from locally binding two empty
categories (here NPj and [P, PATH]j).
A yet different version of dative shift comes from the work of Den Dikken (Den
Dikken 1995; 2006; 2007). We saw Den Dikken’s (1995) structure (and derivation) in
section 2.2.2. Den Dikken (2006) is a more streamlined version; the relationship
between the two objects is mediated by an abstract functional element that he refers
to as a RELATOR, projecting its own phrase, the Relator Phrase. The RELATOR is
the head of the small clause (small clauses being tenseless Relator Phrases in this
approach). The empty preposition incorporates into the verb, after first adjoining
to the RELATOR head.

(47) a. V [RP THEME [RELATOR [PP P GOAL/RECIPIENT]]]

b. V +P [PP tP GOAL/RECIPIENT]i [RP THEME [RELATOR ti]]]
(Den Dikken 2006, 107)

3.3 Status of the “promoted” object

The dative shift approaches reviewed in the previous section raise the question of
whether the promoted object behaves like a standard (i.e., non-promoted) direct
object or not. In Relational Grammar (RG) terms, the promotion is described as “3
to 2 advancement,” where 2 indicates a direct object. And indeed, with respect to
some properties the promoted indirect object does behave like a direct object. For
example, it can passivize, as shown in (48a). For some speakers (and in some lan-
guages), however, the lower object can passivize as well; we will come back to this
issue in section 4.1 below. However, the promoted object resists A-bar-movement (as
shown in (48b). This so-called A-bar Restriction will be the topic of section 4.2.

(48) a. Johni was given ti a book.

∗
b. Whoi did John give ti a book?
20 Double Object Constructions

Haspelmath (2015) points out that framing the discussion in terms of direct object
properties (or lack thereof ) associated with the promoted indirect object might not
be very illuminating as different properties lead to different conclusions. And Postal
(2010) discusses a number of concrete differences between true direct objects and
promoted indirect objects. Even though promoted objects can passivize, with
respect to other diagnostics, however, they do not pattern with direct objects. To
illustrate with a small subset of the properties Postal discusses, promoted indirect
objects do not allow middles (49a), do not allow tough-movement (49b) and cannot
incorporate (49c).
∗
(49) a. Children teach French easily. (with the interpretation ‘It is easy to teach
children French’)
(Postal 2010: 81, citing Everaert 1990)
b. Those people will be easy to send those flowers.
(Postal 2010, 79, citing Farrell 1994)
c. ∗kid telling of stories by teachers.
(Postal 2010: 82, citing Baker 1997)

(49c) is the restriction, observed in Baker (1988a, 278–279; 1988b), that the head indi-
rect object noun in benefactive applicatives cannot “incorporate” into a governing
verb (in contrast to a deep direct object noun). In many languages with productive
dative alternations, an overt “applicative” verbal affix accompanies the advance-
ment to direct object. Marantz (1982) proposes that this situation is typical: when
verbs have supplementary overt morphology corresponding to an oblique argu-
ment NP, that NP and not the usual direct object behaves like a surface direct object.
Baker (1988a, section 5.3.1) refers to this cross-linguistic pattern as Marantz’s
Generalization.
English is poor in diagnostics that syntactically distinguish a surface direct object
among bare NPs. However, some can be constructed. For example, Emonds (1972)
tests the PP-like nature of the direct object in a double object construction by its
interaction with intransitive prepositions (particles). Intransitive prepositions can
follow a direct object, as shown in (50a)–(50b), but not a PP, as shown in (51a)–(51b):

(50) a. The secretary sent a memo out to the committee.

b. Bill fixed a drink up for John at the party.

∗
(51) a. The secretary sent a memo to the committee out.
∗
b. Bill fixed a drink for John up at the party.

For speakers who allow a particle at all within a double object construction, the first
or promoted NP behaves like a direct object in (52a)–(52b) while the second NP
behaves like a PP, as seen by the ungrammaticality of the particles in (53a)–(53b):

(52) a. The secretary sent the committee out a memo.

b. Bill fixed John up a drink at the party.

∗
(53) a. The secretary sent the committee a memo out.
∗
b. Bill fixed John a drink up at the party.
 Double Object Constructions 21

The pattern is considered to have quite different implications in analyses that group
double objects into SCs, such as Kayne (1984) and Den Dikken (1995, ch. 3).
A second test distinguishing English derived direct objects from demoted NPs
(chômeurs) involves shifted quantifiers. Chung (1976, 81–82) observes that in Indo-
nesian, non-adjacent “floated” quantifiers are fully acceptable as modifiers of sub-
jects and direct objects, but not as modifiers of PPs or demoted NPs. The same
(admittedly rather weak) contrast holds in English, where the test concerns not
quantifiers at the end of the clause as in Indonesian, but rather quantifiers shifted
just to the right of the NP they modify. These NPs with floating modifiers can be
either direct objects (54a) or promoted indirect objects (54b), but they cannot be
PPs, as seen in (54c)–(54d):

(54) a. We sent the books all {by mail/to that man}.

b. Mary {brought/sent} the boys {each/both} a present.
c. We sent a refund to those men (??all) by mail.
d. I fixed those drinks for the girls (??both) with ice.

3.4 Productivity
Another question that has featured prominently in the debates on double object
constructions concerns the productivity of the dative alternation. This question
arises irrespective of how the dative alternation is captured. The question is simply
how to account for the fact that some verbs participate in this alternation whereas
others do not. Compared with similar early transformational rules such as passive
formation, the range of conditions on English dative shift suggests to many that it
cannot be a regular syntactic formation. Not all to-datives allow double object var-
iants, as shown by the contrast between the verbs in (55a)–(55b), which do allow
both variants, and the ones in (56a)–(56b), which do not:14

(55) a. Dr. Smith {gave/sent/showed} his collection to the library.

b. Dr. Smith {gave/sent/showed} the library his collection.

(56) a. Dr. Smith {donated/transferred/showed off} his collection to the library.

∗
b. Dr. Smith {donated/transferred/showed off} the library his collection.

There have been many accounts of this difference; relying on semantic, etymolog-
ical, morphological, phonological constraints (or some combination thereof ), a sub-
set of which is given in (57). The list is not exhaustive; it is meant to give the reader
the flavor of the complexity of the issue.

(57) a. Verbs that do not convey transfer of possession do not have double object
variants.
b. Verbs with Latinate (as opposed to Germanic) stems do not have double
object constructions.
c. Verbs with {multisyllabic/multifeet} stems do not have double object
variants.
22 Double Object Constructions

d. Verb stems whose only stress is initial have double object variants.
e. Verb stems lacking secondary stress have double object variants.

Alternatively, while most English double object constructions allow PP dative para-
phrases, a disparate collection of verbs taking double objects does not permit them
(e.g., allow, ask, bet, charge, cost, deny, envy, fine, forgive, pardon, spare):

(58) a. This watch cost Mary fifty dollars.

b. The manager denied the students a chance to explain.

∗
(59) a. This watch cost fifty dollars {to/for} Mary.
∗
b. The manager denied a chance to explain {to/for} the students.

Based on Green (1974), Pinker (1989) and Gropen et al. (1989) divide verbs into two
groups (with the ones listed in (60) participating in the dative alternation and the
ones in (61) not).15 We provide representative examples of both types and refer
the interested reader to these works for a comprehensive list and further discussion.

(60) a. verbs of giving (give, sell)

b. verbs of instantaneous causation of ballistic motion (throw, kick)
c. verbs of sending (send, mail)
d. verbs of continuous causation of accompanied motion in a deictically
specified direction (bring, take)
e. verbs of future having (promise, leave)
f. verbs of instrument of communication (email, fax)
g. verbs of creation (bake, build)
h. verbs of obtaining (buy, find)
(see Gropen et al. 1989, 243–244)

(61) a. verbs of fulfilling (present, credit)

b. verbs of continuous causation of accompanied motion in some manner (pull,
push)
c. verbs of manner of speaking (shout, scream)
d. propositional attitude verbs (say, claim)
e. verbs of choosing (choose, select,)
(see Gropen et al. 1989, 244)

Gropen et al. (1989) point out that not all of the differences can be attributed to the
so-called morphonological constraint, on which monosyllabic verbs with Ger-
manic/native stems allow double object variants and polysyllabic Latinate ones
do not. While verbs of giving (donate versus give), type of communication (tell versus
explain), creation (make versus create), verbs of obtaining (get versus obtain) fall under
the constraint, exceptions include verbs of instrument of communication (email,
radio), which are polysyllabic and allow double object frames) and verbs of future
having (promise, bequeath, assign, prescribe), which are Latinate and polysyllabic, and
yet are also perfectly fine in a double object frame.
Pinker’s (1989) main concern is how children are able to learn the dative alterna-
tion without access to negative evidence. That is, the child must learn without being
told which verbs are unavailable for the double object construction. For example,
 Double Object Constructions 23

how does the learner realize that give is acceptable in this context, but donate is (usu-
ally) not, although the two verbs have very similar meanings? Pinker proposes a
“Grammatically Relevant Subsystem” hypothesis, arguing that “there is a set of
semantic elements and relations that is much smaller than the set of cognitively
available and culturally salient distinctions, and verb meanings are organized
around them” (Pinker 1989, 166). Linguistic processes, including lexical alterna-
tions, are sensitive only to members of this set. He then develops a set of relevant
verbal features and uses linking rules to map certain groupings of features onto cer-
tain syntactic structures. These features are detailed enough to distinguish between
dativizable verbs and similar verbs that do not display the dative alternation
(see Pinker 1989, ch. 5). Although Pinker’s analysis is lexicalist in the sense that
he assumes without argument that “Dative Movement” is a lexical alternation,
nothing in his theory precludes applying it to a transformational theory of datives.
Jackendoff (1990) is another theorist who believes, like Oehrle, that a lack of full
productivity in the dative alternation supports a purely lexical account of double
objects. He reserves transformations for more transparently regular alternations
such as passive. He elaborates on some of Green’s ideas for verb classes, claiming
that the semantic facts of double objects and their verbs militate against a transfor-
mational analysis, at least for English. For Jackendoff, generalizations about theta-
roles do not explain the difference between verbs that allow double objects and
those that do not. Unlike Pinker, he finds no plausible contrast between the roles
assigned by give and donate, which both appear to assign “beneficiary” and “goal”
roles to their complements. He feels rather that the inherent meanings of verbs
determine whether there is a dative alternation. For example, verbs of creation or
preparation such as sing or bake allow double objects, while benefactive NPs with
other verbs such as copy or eat occur only in for-phrases:

(62) a. Enrico {sang/copied} an aria for Luisa.

b. Enrico {sang/∗copied} Luisa an aria.

(63) a. Susan {baked/ate} some vegetables for her mother.

b. Susan {baked/∗ate} her mother some vegetables.

Along similar lines, Jackendoff (1990, 449) favors a lexicalist account for a general-
ization of Pinker (1989), whereby motion verbs that “imply influence of the Agent
continuing throughout the Theme’s trajectory” do not appear with double objects:

(64) a. He {dragged/moved/pushed} the material to Bill.

∗
b. He {dragged/moved/pushed} Bill the material.

(65) a. He {dropped/kicked/threw} the material to Bill.

b. He {dropped/kicked/threw} Bill the material.

An adequate analysis should probably relate this distinction to whether the

P indicates an actual spatial trajectory, as in (64a)–(64b), or whether it simply intro-
duces a third NP argument, as in (65a)–(65b).
A different way to solve the problem of non-productivity is to adopt some variant
of Pesetsky’s (1995, sect. 5.1.3) null P incorporation and derive the constraints on
24 Double Object Constructions

double objects either from the constraints on incorporation or the properties of

the null prepositional element. On one level, this proposal moves the problem of
non-productivity into the lexicon, but this is where both morpheme lists and null
morphemes are known to play a role (Pesetsky 1995, 128–129). In Pesetsky’s
analysis, double object constructions involve incorporation of a null preposition
G. The constraints with productivity will thus have to do with the constraints on
the incorporation of this null preposition. For example, this preposition would be
incompatible with certain verbs, like manner of speaking verbs. Recall that Gropen
et al. (1989, 243–245) list some verb classes that do not occur with double NPs even
though they satisfy the morphophonological constraint. Two types are “manner of
speaking” verbs (scream, whisper, yell) and verbs of continuously caused motion
(drag, move, push). Indirect objects of such verbs are always expressed in to-phrases.
In these cases, following Larson (1988, sect. 5; see also next paragraph), it may well
be that the P’s independent theta-role assigning content causes it to be present
throughout the derivation. In the terms of Emonds (1993) such Ps are lexicalized
in deep structure, which blocks their object NP from advancing to direct object.
Pesetsky (1995) and Den Dikken (1995) propose a similar account in terms of a
lexical P that assigns a theta-role characterized by “continuous imparting of force.”
As a reviewer notes, perhaps the term “indirect object” is simply not appropriate for
objects of these contentful Ps.
For Larson, constraints on productivity also follow from the properties of the
prepositional element. In his analysis, to is contentful and is associated with the
meaning of “Goal of Motion along some path.” The same specification is part of
the meaning associated with the verb give. This means that to can be absorbed/sup-
pressed without violating recoverability of deletion since its meaning is redun-
dantly specified in the meaning of the verb, which in turn means it can be
recovered from it. On this account, the verbs that do not allow dative shift are
the ones that do not have transfer to a Goal as part of their lexical make-up. Verb
particle constructions offer supporting evidence: give away/out no longer describe
transfer of possession, and they also do not allow double object variants:

(66) a. John gave Mary a book.

∗
b. John gave Mary out/away a book.

Particles might play a more general role in explaining the alternation. In this regard,
Fraser’s (1974, 13–15) discussion of which verbs combine with postverbal particles
appears relevant as he concludes that “Surprisingly enough, it is the phonological
shape of a verb that determines to a large extent whether or not it can combine with
a particle … the majority of verbs occurring with particles are monosyllabic and the
remainder are made up primarily of bi-syllabic words which are initially stressed.”
These are also the factors that have been implicated in explaining which classes of
English verbs participate in the dative alternation. If double object frame is limited
to verbs whose only stress is initial (see (57d) above), relatively few double object
verbs have to be listed separately. For the to-dative alternation, the following verbs
with some non-initial stress are listed in Herriman (1995, 61, 104): accord, administer,
advance, afford, allocate, allot, allow, apportion, assign, assure, award, begrudge, bequeath,
concede, deliver, deny, extend, forbid, permit, prohibit, recommend, refund, refuse, reim-
burse, remit, repay, return, telegraph, telephone, and vouchsafe. For the for-dative, she
 Double Object Constructions 25

lists (1995, 124, 146) design, embroider, ensure, furnish, guarantee, occasion, prepare, pre-
scribe, procure, provide, purchase, reserve, and secure. However, most of these “excep-
tions” can be analyzed as containing a prefix and an initially stressed stem. In
standard American speech, verbs that carry secondary or non-initial stress are usu-
ally incompatible with a dative alternation. This suggests that perhaps the relevant
factor is the lack of secondary stress on the verb stem (see (57e) above).16
More recently, Coppock (2009) devises a number of careful experiments to distin-
guish between the types of hypotheses listed in (57), more specifically between the
prosodic weight hypothesis (i.e., only verbs with one metrical foot allow double
objects), the two lexicon hypothesis (i.e., only verbs belonging to the Germanic
(as opposed to Latinate) lexicon allow double objects), the formality hypothesis
(i.e., verbs that are compatible with double objects tend to be less formal). Since con-
structed examples tend to conflate different factors, such as number of syllables and
number of feet or Germanic versus Latinate roots and formality level, she uses
nonce verbs (following Gropen et al. 1989 in this respect). To illustrate briefly, in
one experiment (Experiment 2, discussed on pages 81–90), she keeps the number
of syllables constant but varies the number of feet, contrasting nonce verbs like
’feffam (one foot, stress on the first syllable) with ones like fe’flame (two feet, stress
on the second syllable) and finds no effect of prosodic weight.17 She also finds no
contrast between Germanic and Latinate roots that is not tied to the number of syl-
lables. Formality level was also shown not to be a factor by her experimental results.
A very different view on the productive of the dative alternation is taken by Bres-
nan (2007), Bresnan et al. (2007), Bresnan and Nikitina (2009), who argue against
the approaches that derive the constraints on the dative alternation from the mean-
ing differences, and suggest a more probabilistic approach instead. If change of
possession correlates with a double object frame and change of location/movement
to a goal correlates with PP dative frame, ditransitive constructions that do not indi-
cate transfer of location are predicted to be impossible in a PP dative frame and
ditransitive constructions that do not indicate movement to a location are going
to be impossible in a dative PP frame. The two are illustrated in (67a)–(67b),
respectively.

∗
(67) a. That movie gave the creeps to me.
∗
b. I carried/pulled/pushed/schlepped/lowered/hauled John the box.
(Bresnan et al. 2007, 71)

Bresnan et al. (2007), based on the fact that examples of this general sort are attested
in the corpus (see (68a)–(98b) below), argues that these are only preferences that can
be overridden by independent factors, such as discourse prominence (given pre-
cedes non-given), weight/heavy (light constituents come before heavy constitu-
ents), pronominal status (pronouns come before non-pronouns), animacy
(animates come before inanimates) and definiteness (definites come before indefi-
nites). They thus conclude that “we cannot predict the dative alternation from
meaning alone” (page 75).

(68) a. This story is designed to give the creeps to people who hate spiders but is not true.
b. This life-sized prop will give the creeps to just about anyone!
(Bresnan et al. 2007, 72)
26 Double Object Constructions

They take it to mean that there are no deep syntactic differences between double
object and PP dative frames. Bruening (2010a), however, takes issue with this con-
clusion, based on the syntactic behavior of the examples of the kind given in (68a)–
(68b) above. He argues that these counterexamples are only apparent and that they
are in fact double objects in disguise, as they pattern with double object construc-
tions in a number of respects (e.g., lack of inverse scope).

3.5 Meaning differences

A common intuition, going back at least to Green (1974) and Oehrle (1976) is that the
double object and ditransitive PP constructions are not equivalent in meaning and,
consequently, that the restrictions on productivity discussed in the previous
section can be (at least partially) attributed to the differences in meaning. On this
view, a double object frame implies successful transfer of possession whereas a
PP dative a change of location. Green (1974, 157), for example, claims that learning
necessarily occurs for using (69a) felicitously, but not for (69b). Similarly, Pinker
(1989) claims that possession must change in (70a) but not necessarily in (70b):

(69) a. Mary taught John linguistics.

b. Mary taught linguistics to John.

(70) a. John threw Bill the book.

b. John threw the book to Bill.

For Emonds (1993, 227–228), on the other hand, such examples do not differ in truth
values.18,19 This is also the conclusion of Bresnan et al. (2007), whose work we dis-
cussed in the previous section (in a slightly different context).
Others (such as Pinker 1989; Krifka 2004; among others) point out that the mean-
ing component that is crucial to a double object frame is caused possession rather
than transfer of possession. In examples like give X a headache or give X an idea, it is
clear that there is no literal or metaphorical transfer of possession but there is a
component of caused possession. However, Rappaport and Levin (2008) argue
against such uniform treatments of the difference between the double object and
the PP dative frame, where by uniform we mean that all double object construc-
tions are associated with a single meaning (be it caused possession or transfer of pos-
session) and all PP datives are associated with another meaning. Instead, they
advocate what they dub a verb-sensitive approach to the meaning differences
between the two frames, on which some verbs do indeed have two meanings (cor-
relating with the difference between double object and PP dative frames) but others
maintain a single meaning across the two frames. The verb give always expresses
caused possession (even in the PP dative variant) and lacks a path component.
They point out, for example, that if give had a path component in the PP dative var-
iant, the example in (71), in which the path is expressed overtly, should be gram-
matical, contrary to fact:

∗
(71) Josie gave/handed the ball from Marla (to Bill).
(Rappaport and Levin 2008, 139)
 Double Object Constructions 27

In this respect, give contrasts with verbs like throw, kick, send, or ship, which do have
a path component, expressed overtly below:

(72) a. Jill threw/kicked the ball from home plate to third base.
b. I sent/shipped the bicycle from my house at the beach to my house in the
mountains.
(Rappaport and Levin 2008, 136)

Rappaport and Levin also point out that, depending on the verb, the successful
transfer can be present irrespective of the syntactic frame and in both variants it is
equally easily defeasible (see also Krifka 2004).

4 Constraints on movement

4.1 Symmetric versus asymmetric passives

In this section, we turn to the behavior of the two objects with respect to passiviza-
tion, more specifically to the issue of which of the two objects in a double object con-
struction can undergo passive movement. Passives in which either object can
passivize are commonly referred to symmetric passives and passives in which only
one can as asymmetric passives (see Baker 1988a; Bresnan and Moshi 1990; Woolford
1993; McGinnis 1998; 2001; Citko 2011a; among many others).20
On the asymmetric approaches to double object constructions, a variety of which
we defended in section 2.2, the asymmetric passive in which the higher (i.e., the
indirect) object is the only one that can passivize is the expected pattern given stand-
ard locality considerations; simply put, it is the one closer to the subject position,
identified as the specifier of TP in (73).

(73) TP

DP
Goal DP
Theme

This is what accounts for the following contrast:

(74) a. Johni was offered ti a job.

∗
b. A jobi was offered John ti.

Logically speaking, another asymmetric pattern is possible; the pattern in which

only one object can passivize but it is the direct object instead. This is quite common,
typical in languages that mark indirect objects with inherent (dative) case.
28 Double Object Constructions

We illustrate below with examples from Polish, representative of this class of

languages (see Dziwirek 1994; Citko 2011a; 2011b for further discussion of Polish
double objects and their passivization patterns):
(75) a. Maria poleciła Janowi książkę.
Maria.NOM recommended Jan.DAT book.ACC
‘Maria recommended a book to Jan.’
b. Książkai została polecona (Janowi) ti przez Marię.
book.NOM became.3FEM recommended Jan.DAT by Maria
‘The book was recommended to Jan by Maria.’
c. ∗Jani został polecony ti książkę przez Marię.
Jan.NOM became.3MASC recommended book.ACC by Maria
‘John was recommended a book by Maria.’

Symmetric passives, in which either object can passivize, are known to be possible
in languages like British English (and for some dialects and/or speakers of
American English), Norwegian, Icelandic, among others:21

(76) English
a. Johni was given ti a book.
b. A booki was given John ti.

(77) Norwegian
a. Jon ble gitt boken.
John was given book.DEF
‘John was given the book.’
b. Boken ble gitt Jon.
book.DEF was given John
‘The book was given John.’
(adapted from Holmberg and Platzack 1995, 215)

(78) Icelandic
a. Jόni var gefin bόkin.
John.DAT was given book.NOM
‘John was given the book’
b. Bόkin var gefin Jόni.
book.NOM was given John.DAT
‘The book was given John.’
(Holmberg and Platzack 1995, 189)

The contrast between languages (or dialects) with symmetric passives and those
with asymmetric passives is widely discussed and perhaps best known from the
literature on Bantu languages. In this group of languages, oftentimes the contrast
is not between one language being symmetric and another one asymmetric, but,
rather, between one type of applicative being symmetric and another one asymmet-
ric (see Kimenyi 1980; Woolford 1993; Marantz 1993; Ngonyani 1996; McGinnis
2001; Doggett 2004; and the references therein for further discussion). For example,
in Swahili locative applicatives are symmetric whereas benefactive and goal ones
are asymmetric.
 Double Object Constructions 29

There is some variation regarding the acceptability of direct object passives in

American English double object constructions. While the judgments in (74a)–
(74b) are the ones perhaps most often reported in the literature, some speakers also
allow passivization of the lower (direct) object, as illustrated below. The possibility
of passivizing either object was already reported in the early literature on the topic
(see Fillmore (1965), Emonds (1972), Lees (1963) and Chomsky (1955), for example).

(79) a. Several books were given him by the teacher.

(Postal 2010, 52, citing Chomsky 1955)
b. A letter was given ’im/∗HIM by Mary.
(Larson 1988, 364)

Note that in all of them the (intervening) indirect object is a pronoun, which sug-
gests, as noted by Oehrle (also Larson 1988) that the possibility to passivize the
direct object could be attributed to the weak nature of the (pronominal) indirect
object (and the possibility to cliticize it).
Some of the existing accounts of symmetric passivization patterns (be it in English
or other languages) link it to the availability of word order in which the relative
order of the two objects is reversed (see Ura 1996; Anagnostopoulou 2003; Doggett
2004, among others). Examples of such “alternative double object constructions”
from dialects of British English, more recently discussed by Haddican (2010), are
given in (80).

(80) a. I gave it him.

(Gast 2007, 31)
b. She gave a book the man.
(Hughes and Trudgill 1979, 21)

However, this correlation does not hold up to cross-linguistic scrutiny. Doggett

(2004), for example, points out that Norwegian and Swedish are counterexamples
in that they allow symmetric passives but do not allow both orders. Instead, they
only allow one ordering of the two objects, with the indirect object preceding the
direct one.
Yet other accounts capitalize on movement of the direct object over the indirect
object to a position in which it is the closer target for passivization (see McGinnis
2001 and Anagnostopoulou 2003 for representative examples). For both of them,
the ability for the direct object to move over the indirect one is what is responsible
for symmetric passive formation. Anagnostopoulou (2003) proposes a parameter
that allows the direct object to move to the specifier of the applicative little v, gen-
erated above VP. McGinnis (2001) correlates the availability of symmetric passives
with the availability of a high applicative structure in a language (or for certain type
of applicatives). The gist of McGinnis’ proposal is that the high applicative struc-
ture allows movement of the direct object to the external specifier of the applicative
head, which McGinnis attributes to the high applicative head being a phase head
(which “endows” it with an EPP feature responsible for this movement). From this
position, the direct object is closer than the indirect object to the subject position. To
explain what allows the direct object to passivize as well, we need to assume that the
30 Double Object Constructions

indirect object has the option to tuck in (land below the base-generated position of
the indirect object, making the direct object a closer target for passivization) or that
the presence of the EPP feature is optional or that the two objects are equidistant
(being the specifiers of the same head).

(81) applP

DPi appl′

DP appl′

appl[EPP] VP

V ti

Crucially, analogous movement (of the direct object over the indirect one, schema-
tized in (82)) is impossible in a low applicative structure.

(82) VP

V applP

DPi appl′

DP appl′

appl ti

First, such movement would be movement from a complement position to a spec-

ifier position, which violates anti-locality by virtue of being too short, as argued by
Doggett (2004), Jeong (2007), among others. Second, there is no motivation for this
movement (the low applicative head, by assumption, is not a phase head and lacks
the EPP feature). And passivizing the direct object in a low applicative structure
(i.e., moving the direct object directly to the subject position) would violate standard
locality principles (Relativized Minimality, Shortest Move, Attract Closest, etc.) as
the direct object would cross over the indirect object.

4.2 A-bar-movement restriction

If a promoted indirect object behaves like a standard direct object, the degraded sta-
tus of the examples in (83) is rather mysterious:

??
(83) a. Which sisteri shall we send ti a present?
??
b. Whoi did Carolyn bake ti that cake?
 Double Object Constructions 31

∗
c. This is the friendi (that) {Mary sent ti a present/Carolyn baked ti that cake}.
∗
d. Kidsi are always easy to {tell ti a story/buy ti presents}.
∗
e. Charlie {sent ti a book/baked ti that cake} [the girl who lives next door]i.

We formulate the restriction in (84).

(84) Restriction on A-Bar-Extraction

Movement of a “promoted” indirect object NP unmarked by a P or dative case to a
non-argument position is ungrammatical.

This restriction on what Government and Binding calls “A-bar-movements” holds

for both Goals (to-datives) and Beneficiaries (for-datives). For English, it was first
noted in Fillmore (1965). Ross (1967) observes that indirect objects without
P cannot undergo rightward Complex NP Shift (83e) either. Further examples of
this restriction are presented and discussed in Oehrle (1976), Ziv and Sheintuch
(1979), Hornstein and Weinberg (1981), Whitney (1982; 1983), Czepluch (1982),
Barss and Lasnik (1986), Baker (1988a, sect. 5.4), and Hellan (1991). As Whitney
(1982; 1984) observes for English, the restriction generalizes to all movements to
non-argument positions, including rightward movements, as in (83e).
Before reviewing how authors have treated this restriction theoretically, it is
important to note that this kind of syntactic property strongly suggests that a pro-
moted Goal phrase, even though it has the position and many other properties of a
direct object, is not simply that. Promotion accounts of the kind reviewed in
section 3.2 can distinguish between the wh-objects in (83a)–(83e) and the non-
promoted wh-objects, which are not subject to this restriction. Considering double
object constructions to be base-generated provides no clear reason why they should
behave differently under A-bar-movement.
Oehrle’s (1976) base analysis includes the statement that inner NPs, that is, those
followed by a direct object, are inaccessible to movement rules, but this basically
only rephrases the problem. Moreover, Baker (1988a, 295) points out that the expla-
nations based on the inaccessibility of an inner NP (as defined by the simultaneous
presence of an “outer NP”) encounter a further difficulty. In Chichewa and Cha-
morro, morphologically identifiable applicatives with promoted oblique NPs can
be formed with certain classes of intransitive verbs. Extraction of the applied object
in these cases is just as bad as extraction of the inner object of a transitive verb.
Accounts that depend on the presence of two NPs, rather than on the derivational
status of the promoted NP, are thus inadequate.
Stowell’s (1981) solution for the Restriction on A-Bar-Extraction is somewhat
more attractive. In his analysis, a bare English indirect object (of whatever length)
is incorporated under the V by a “word formation rule.” It then follows that the
inner NP cannot be further moved, as syntactic rules never apply to parts of words.
Larson (1988) also invokes a type of inaccessibility restriction based on a verb form-
ing a deep constituent with an indirect object. Furthermore, for Larson, the reason
why Heavy NP Shift of the indirect object in double object constructions is ungram-
matical is different from the reason other A-bar-movements of direct objects are
ungrammatical. For him, Heavy NP Shift does not involve rightward movement;
instead, it involves reanalysis. And the reanalysis required to derive the Heavy
NP Shifted variant with the promoted indirect object is structurally impossible.
32 Double Object Constructions

Larson also points out that cross-linguistically the two do not pattern together; Nor-
wegian, for example, allows wh-movement of indirect objects but not Heavy NP
Shift thereof. As we saw in the previous section, promoted indirect objects can nev-
ertheless undergo A-movement. Both Stowell and Larson circumvent the problem
involving their ability to move (only) to A-positions by claiming that “dative pas-
sives” in (85a)–(85c) move from an outer, non-promoted position directly to the
subject position. It is not clear, however, what allows this movement not to violate
locality, given the discussion in the previous section.

(85) a. Wei might be given a book ti.

b. A womani was offered the job ti.
c. The guestsi were finally fixed an appropriate lunch ti.

Once English dative passives are not derived directly from double object construc-
tions, however, an important generalization is lost. Early generative analyses
observed that English verbs which do not permit prepositionless double objects
prohibit dative passives as well:

∗
(86) a. Our club might be {distributed/donated/suggested} some books.
∗
b. A woman was {demonstrated/outlined/revealed} the procedure.
∗
c. The guests were finally {concocted/designed/reordered} an appropriate
lunch.

That is, dative passives appear to be derived, in English at least, from double object
constructions. Hoekstra (1995, sect. 3) provides two further arguments against
passivizing indirect objects in their base position. Along the same lines, Romance
languages lack dative passives because they lack bare double objects. Thus, the fol-
lowing descriptive generalization encompasses both English and Romance:

(87) Dative Passive Restriction

Passivization of indirect objects (unmarked by any dative case) results from
moving NPs already promoted to direct object.

Czepluch (1982) takes a different approach to the A-bar-movement restriction. He

analyzes bare datives in terms of a base-generated PP headed by an empty P. He
rules out extraction of the bare dative NP through a constraint on movement out
of empty-headed phrases, which would leave a presumably unacceptable “layered
trace” under a PP. In order to nonetheless allow dative passives, Czepluch must
extend case theory so that these layered traces are unacceptable only under the con-
ditions of A-bar-movement. That is, a sentence like ∗Who did John give the book? is
ruled out because the trace of who cannot be either directly or indirectly case-
marked.
Den Dikken (1995, sect. 4.2) provides a somewhat similar account of the A-bar-
movement restriction that also involves the structure of a PP. Summarizing relevant
data from a range of authors, he concludes that only “short” (clause-bound) wh-
movement of promoted IOs can be grammatical, as in (83a)–(83b), and this only
in some languages, such as Dutch, Norwegian, and certain dialects of English.
 Double Object Constructions 33

For these systems with grammatical “short” overt wh-movement of indirect objects,
which seems to contravene (84), Den Dikken argues that an entire PP headed by an
empty dative-assigning P is fronted. In contrast, he claims that many A-bar-extrac-
tions, including at least those in (83c)–(83e), involve Empty Operator Movement or
more generally simple movement of an NP, and these are always excluded (Den
Dikken 1995, 188).
Baker (1988a, 299) proposes a different refinement of Czepluch’s approach, which
he calls the Non-Oblique Trace Filter. This filter blocks any trace of an A-bar-oper-
ator lacking oblique case if its non-verbal governor incorporates into a higher verb.
By stipulating the presence of an operator in the filter, he automatically exempts
passive traces from it while still blocking A-bar-movements. It is doubtful whether
this ad hoc device works empirically, since it also appears to wrongly rule out
A-bar-movement of a passivized bare indirect object, which is cross-linguistically
permitted (Alsina and Mchombo 1990, sect. 2).
Hellan (1991) illustrates the A-bar Restriction for Mainland Scandinavian,
arguing that the trace of an A-bar-movement of a bare indirect object is not licensed.
In his structure for double objects, the verb governs just the direct object; the bare
indirect object is licensed and receives a theta-role only by virtue of its structural
position (akin to a possessive NP). Moreover, he claims that such lexical NPs do
not need to receive an abstract case, whereas traces of wh-movement uniformly
do require case. Hence, those NPs licensed by virtue of their structural position,
and not by a lexical licenser, are precisely those which cannot undergo A-bar-move-
ment. A virtue of this proposal is that it can account for why possessive NPs in Eng-
lish, for example, are not available for A-bar-movement either. However, it must be
observed that extracting possessive NPs (the Left Branch Constraint of Ross 1967) is
significantly less acceptable than violating (84), and that the Left Branch Constraint
is subject to cross-linguistic variation.
Whitney (1982; 1983) proposes an account of the A-bar Restriction that does not
require extending case theory or special restrictions on movement of the inner NP.
Like Czepluch and Baker, she assumes that indirect objects are generated under a
PP. They then move to their surface position next to the verb, leaving a trace in the
usual way:

(88) V′

V IOk DO PP

[p Ø] [np tk]

All further movements to an A-bar-position, including Complex NP Shift and other

A-bar-movements (Whitney 1983; 1984), leave a trace that serves as a variable in LF.
Further movement of indirect objects is then blocked by Chomsky’s (1982, 31) prin-
ciple that a variable “must be A-free in the domain of the operator that binds it.” The
variable in the position of IOk is under the first branching node V is not A-free. This
analysis correctly predicts that wh-movement, Complex NP Shift, and other move-
ments of bare indirect objects to A-bar-positions are ruled out, while movements to
an A-position, like passive, are acceptable. It also distinguishes base-generated NPs
34 Double Object Constructions

next to the verb, which are extractable, from indirect objects that have been pro-
moted to direct object position, which are not.
Whatever the ultimate fate of these competing proposals for explaining the A-bar
Restriction, the facts do suggest that bare indirect objects are not generated in the
base as direct objects but rather move to their surface position. The restriction on
their further movement, which base-generated direct objects are not subject to,
can be satisfactorily accounted for only if their transformational history is distinct
from that of direct objects.

4.3 Scope freezing

The behavior of double object constructions with respect to scope, in particular the
differences in this respect between double object constructions and PP dative con-
structions, also shows that promoted indirect objects behave differently from run-
of-the-mill direct objects. English is well-known to allow quantified arguments to
optionally take wide scope with respect to other (higher) arguments in the same
clause. In (89), for example, either surface or inverse scope is possible. This is com-
monly attributed to the covert operation of quantifier raising (see May 1977 and
much later work). On the surface scope interpretation, topics are fixed independ-
ently of the students, whereas on the inverse scope interpretation, topics co-vary
with students.

(89) Two new topics might reinvigorate every student in the TWO > EVERY, EVERY > TWO
class.

But, as Aoun and Li (1989; 1993) and Larson (1990), who attributes the observation
to David Lebeaux, note, the double object constructions differ from PP datives in
that they do not allow inverse scope. That is, (90b) below, unlike its PP dative coun-
terpart in (90a) does not typically mean that every student was assigned the same
two topics.22

(90) a. She assigned two new topics to every student in the TWO > EVERY, EVERY > TWO
class.
∗
b. She assigned every student in the class two new TWO > EVERY, EVERY > TWO
topics.

The examples in (90a)–(90b) and (91a)–(91b) provide further illustration; the PP

dative in (91a) allows either scope, whereas its double object counterpart only
allows surface scope, in which all problems were assigned to a single student.

(91) a. The teacher assigned one problem to every student. ONE > EVERY, EVERY > ONE
b. The teacher assigned one student every problem. ONE > EVERY, ∗EVERY > ONE
(Larson 1990, 604)

Bruening (2001) provides a very explicit account of this restriction, and in the remain-
der of this section, we focus on his account. He attributes the lack of inverse scope in
the double object frame to an independent constraint on quantifier raising (QR).
 Double Object Constructions 35

In particular, he assumes, as is pretty standard, that both quantifiers in a double

object construction have to undergo QR, which he takes to be adjunction to the
closest projection of the right semantic type (type t). This movement is subject
to locality; the indirect object, being the higher one, moves first and the lower
one tucks in below (following Richards’ 1997 tucking in principle), which requires
the second movement to a multiple specifier configuration of the same head to
target the inner specifier (i.e., tuck in below the first moved element). Alternative
derivations that would result in inverse scope are not allowed; one would involve
the reverse ordering of the two movements, with the direct object moving first and
the indirect object moving next (and landing in the lower specifier), and the other
one a sequence of movements in which the direct object moves second but lands in
the outer specifier instead. Both would result in (92b), which should yield inverse
scope. However, both are banned by economy (the movement to the outer spec-
ifier is longer than the movement to the inner one). On Bruening’s account, frozen
scope in double object constructions reduces to superiority, which, more gener-
ally, reduces to economy.

(92) a. [vP one studenti [vP every problemj [vP the teacher [v′ v [vp assigned ti tj ] ] ] ]

b. [vP every problemj [vP one studenti [vP the teacher [v′ v [vp assigned ti tj ] ] ] ]

Interestingly, it is not the case that the direct object can never scope over another
element. The examples in (93) show a direct object in a double object construction
can, for example, scope over the subject.

(93) a. A (different) teacher gave me every book. EVERY >A

b. At least two judges awarded me every medal. EVERY > AT LEAST TWO
(Bruening 2001, 243)

The ability for the indirect object to scope over the subject can be accounted for by
allowing the lowest copy of the subject to be the one that “counts” for interpretive
purposes, that is, by interpreting it in its reconstructed position, as shown in (94).

(94) [Tp a different teacher T [vP every booki [vP a (different) teacher [v′ v [vp gave me tj ] ] ] ]

Note that the subject, unlike both objects, moves to the specifier of TP for EPP
related reasons. This movement, being overt (and not related to scope), can recon-
struct. By contrast, QR, being covert (and scope-related), does not reconstruct,
which explains the lack of the option of interpreting the higher copy of the QR-
ed direct object and the lower copy of the QR-ed indirect object, which would give
rise to the non-existent inverse scope interpretation.23
36 Double Object Constructions

5 Cross-linguistic variation

This section focuses on one aspect of cross-linguistic variation with respect to the
availability of double object constructions, in particular on the factors that might
account for the lack of double object constructions (and, relatedly, the lack of a
dative alternation) in a language. In spite of what the vast literature on the dative
alternation might imply, from a typological perspective it is not very common; in
Siewierska’s (1998) sample of 219 languages only 12 (6 percent) exhibited a dative
alternation (see also Haspelmath 2015).24 In this section, we consider three types of
explanations for this gap that have been explored in the literature.
One possibility is to link the availability of double object constructions to morpho-
logical case marking (or the lack thereof ). On this view, languages with morpholog-
ical case marking that distinguish Themes from Goals/Recipients/Benefactives (by
marking the latter with dative case, for example) will not by definition allow a dou-
ble object construction, in which two objects are marked with the same case (such
as the accusative case, the “designated” Theme case). From this perspective, the
German and Polish examples in (95)–(96) are not true double object constructions.
(95) German
a. Ich gab Anna das Buch.
I gave Anna.DAT the book.ACC
‘I gave Anna the book/I gave the book to Anna.’
b. Ich gab das Buch Anna.
I gave the book.ACC Anna.DAT
‘I gave Anna the book./I gave the book to Anna.’
(Holmberg and Rijkhoff 1998, 82)

(96) Polish
a. Jan wysłał Marii paczkę.
Jan.NOM sent Maria.DAT package.ACC
‘Jan sent Mary a package.’
b. Jan wysłał paczkę Marii.
Jan.NOM sent package.ACC Maria.DAT
‘Jan sent Maria a package.’

Thus, the alternation between the (a) and (b) examples in (95) and (96) is most likely
not due to dative shift
Den Dikken (1995, sect. 4.6), building on Müller (1992) argues that the German
has no dative shift and what appears to be dative shift is actually “A -scrambling
of a zero-headed dative PP.” We tentatively agree with this conclusion, although
we cannot fully exclude that languages with overtly case-marked NPs could, upon
investigation, reveal a non-scrambling dative alternation. Nevertheless, construc-
tions of the kind given in (95) and (96) are often referred to as double object con-
structions in the relevant literature, and the structures that have been posited for
them are similar to the structures that have been proposed for English double
objects (e.g., null P structure, applicative structure, VP shell structure, small clause
structure). The questions that this alternation raises are also similar to the questions
that have been at the forefront of the research on the English dative alternation.
 Double Object Constructions 37

Are the two orders (accusative dative and dative accusative) transformationally
related or not? And if they are, which one is basic and which one derived? Not sur-
prisingly, these issues are far from being settled and quite often, different researchers
have argued for different answers to these questions. Dornisch (1998), for example,
argues that the basic order in Polish is accusative dative, whereas Citko (2011a; 2011b)
takes the opposite view. Likewise, Dyakonova (2007; 2009) argues for dative accusa-
tive order being the basic one in Russian, whereas Bailyn (1995; 2012) takes the accu-
sative dative order to be basic. Japanese also allows both orders and marks indirect
objects with dative case. Hoji (1985) takes the dative accusative order in Japanese to be
basic, and the accusative dative order to be derived via scrambling. Miyagawa (1995;
1997), Miyagawa and Tsujioka (2004), on the other hand, argue that the two orders
are not transformationally derived (however, both permit further reordering), but are
a reflex of two different argument structures, corresponding to the English double
object structure and PP dative, respectively. What allows for such an analysis is
the fact that the Japanese particle ni is ambiguous between a dative marker and a
preposition. Interestingly, the two can co-occur:

(97) Japanese
Taroo-ga Hanako-ni Tokyo-ni nimotu-o okutta.
Taroo-NOM Hanako-DAT Tokyo-to package-ACC sent
‘Taroo sent Hanako a package to Tokyo.’
(Miyagawa and Tsujioka 2004, 9)

The lack of the ability of marking the two objects with non-distinct cases is one fac-
tor that might account for the unavailability of double objects in a language. This,
however, will not explain why, French, for example (which lacks case marking out-
side of its pronominal system) lacks double objects.
Kayne (1981; 1984), focusing on the contrast between English and Romance lan-
guages links the availability of double object constructions in language to the avail-
ability of preposition stranding.25 Given the cross-linguistic rarity of preposition
stranding, it would be quite unlikely that only languages with preposition strand-
ing (English, Dutch under certain circumstances, Afrikaans) would allow double
objects. Kayne (1984, 195–196) further states that the reason French does not allow
double object constructions is similar to the reason it does not allow ECM construc-
tions. Kayne assumes that empty prepositions cannot assign Case but can transmit
objective case if prepositions in that language normally assign objective case. Then
objective case transmission by an empty preposition is available in English, but not
in French. Kayne’s proposal is then that P assigns case differently in the two lan-
guages, although neither language has overt case morphology outside the pronoun
system. Larson follows Kayne and derives the lack of double object constructions in
Romance languages from the same parameter setting that derives the lack of prep-
osition stranding. To derive both P-stranding and the to-absorption necessary for
dative shift to obtain, the preposition has to be reanalyzed with the verb. This re-
analysis is blocked if the verb and the preposition assign different cases. For Kayne,
the prepositions in Romance assign oblique case whereas in English they assign
accusative case, which in turn makes reanalysis possible. This, however, seems
to predict that if there are languages in which different prepositions assign different
38 Double Object Constructions

cases, some of which happen to be accusative, in these languages only these prepo-
sitions should allow reanalysis (and thus be strandable and allow dative shift).
Zhang (1990) points out that “under one possible interpretation, Kayne’s and Lar-
son’s theory makes the following claim: Languages that do not permit P-stranding
also do not permit Dative Shift.” Zhang further shows that Chinese contradicts any
such claim, as do Indonesian and other non-Indo-European languages with dative
applicatives. As observed in Van Riemsdijk (1978), preposition stranding is rare and
probably limited to a subset of Germanic languages; hence is unlikely to be a pre-
condition for a dative alternation. Sugisaki and Snyder (2006) examine the predic-
tions Kayne’s model makes for the acquisition of double object constructions, and
show that children acquire double object constructions before they acquire preposi-
tion stranding, which casts doubt on the idea that preposition stranding is a “pre-
requisite” for double object construcions.26
Yet another possibility explored in the literature is to link the availability of dou-
ble object constructions not so much to the existence of another phenomenon (like
morphological case, preposition stranding, or case transmission) but to the availa-
bility of a certain structure or the availability of a certain lexical element. This is the
stance taken by Harley (2002), for example, who links the availability of double
object constructions to the availability of “the possessive prepositional element”
Phave in a language. Languages without this prepositional element (e.g., Irish, Diné
(Navajo)) lack double objects. On this view, there is a correlation between the exist-
ence of double object structures and the expression of possession in a language. Lan-
guages that express possession with a locative structure (Ploc in Harley’s terms) lack
double object constructions. Crucially, the correlation is not between the presence of
the verbal have to express possession in a language and the availability of double
object constructions. Variation among languages that have a verbal have, however,
raises some questions for this view. As Harley notes, Romance languages, which
have have but are generally assumed to lack double object constructions appear
problematic for this view without some extra qualifications.

6 Conclusion

In this chapter, we have outlined the main issues surrounding double object con-
structions that have occupied linguists since the early days of generative grammar.
We hope to have shown what makes double objects interesting, and where the main
points of content and discontent lie. The points that the literature surveyed here has
established, which we take to be fairly uncontroversial, are the following. First, the
indirect object surfaces as asymmetrically “higher” than the direct object. This is
compatible with many structures that have been proposed for the double objects,
which we surveyed in section 2.2. It is also independent of the issue of whether
the indirect object is base-generated in its surface position or gets there via move-
ment; more generally it is independent of the issue of whether the dative alternation
is a lexical or syntactic process (which was the focus of section 3). Second, even
though with respect to some properties (case marking, passive movement), the
indirect object behaves like a true object, with respect to others (A-bar-movement,
scope) it does not. These restrictions (as well as the cross-linguistic variation with
respect to passive movement) were the theme of section 4. While the core empirical
 Double Object Constructions 39

observations remain pretty solid, we have seen that there are many different expla-
nations for these restrictions. And finally, in section 5, we have discussed the pos-
sible explanations for the relative cross-linguistic paucity of double object
constructions.

SEE ALSO: Affectedness; Applicatives; Case, Primarily in Japanese; The Person

Case Constraint; The Spray/Load Alternation

Notes

1. Haspelmath (2015) characterizes a ditransitive construction as “a construction with a

verb denoting transfer of an entity,” where transfer can be literal as with verbs like give
or send, or more metaphorical as with the verb promise. He further distinguishes three
different patterns in the coding of Theme and Recipient arguments, which differ with
respect to whether the Recipient or the Theme argument (or both) pattern with the
Theme of a regular transitive construction.
2. This does not mean that there are no asymmetries between the two objects, as we will
see in sections 3.3 and 4 in particular.
3. Barss and Lasnik consider examples of the kind given in (8a), which involve wh-
movement of the indirect object “less than grammatical” but do not star them. We come
back to the status of such examples in section 4.2.
4. A different execution of this basic idea is proposed in Aoun and Li (1989; 1993).
5. There are many other implementations of small clause structures for double object con-
structions, such as Hoekstra (1988), Aoun and Li (1989), Bowers (1993), Den Dikken
(1995; 2006), Harley (2002), to name a few.
6. The early structure Pesetsky presents on pages 126–127 involves ternary branching. The
structures in (24a)–(24b) are the later binary variants, which preserve his insights about
the G element.
7. Additionally, Pesetsky discusses coordination data that are incompatible with a small
clause structure.
8. This does not imply that there are no differences between the two. Bantu languages
allow multiple (i.e., recursive) applicatives, and allow applied arguments with a wider
range of thematic roles.
9. Consequently, Baker develops two alternatives in his section 5.3.4, one in terms of an
extension of Case Theory and a second in terms of a rather abstract ‘LF incorporation’
of the thematic direct object (thus exempting it from the Case Filter in his system).
10. See Bruening (2010b) for a comprehensive structural theory of idiom formation.
11. Larson (1990, section 5.3) further argues for his constituent structure with a claim that
both gapping and pseudo-gapping can delete a V constituent which unites verbs and
indirect objects. However, finding contrasts in pairs like (i)-(ii) that would support this
claim is difficult; the (a) examples, in which V deletes, do not seem to contrast with
those in (b), in which V is not available:

??
(i) a. He sends love letters to Mary and she [V Ø] dessert recipes.
b. Bill fixes martinis for us more often than he {did/would}[V Ø] manhattans.
??
(ii) a. He sends love letters to Mary and she to Bill.
b. Bill fixes martinis more often for us than he {did/would} for you.

12. An appropriate Principles and Parameters reformulation of the original construction-

specific rule requires that the interchange be sanctioned by UG, but licensed only in lan-
guages with some specific lexical property. His analysis uses a somewhat idiosyncratic
40 Double Object Constructions

understanding of the Projection Principle to predict that dative movement is the only
possible type of argument interchange.
13. It would be perhaps less stipulative to assimilate the empty coindexed P to a sort of a
trace or copy, as movement generally leaves those behind.
14. Similarly, not all for-datives (which appear with characteristic verbs of preparation and
creation) permit the double object variant:

(i) a. Luisa {baked/fashioned/bought} some gingerbread houses for the children.

∗
b. Luisa {invented/devised/dispatched} some new toys for the children.
(ii) a. Luisa {baked/fashioned/bought} the children some gingerbread houses.
∗
b. Luisa {invented/devised/dispatched} the children some new toys.

15. Green (1974) groups verbs that license Dative Movement into different semantic
classes, and relates these classes to different underlying structures. For example,
pass and take are in the “bring” category, feed is in the “give” category, buy and
save in the “selection” group, and so on. She concludes that “Dative Movement”
should actually be divided into a series of interrelated processes that relate dou-
ble object constructions and NP–PP sequences in either direction – from NP PP to
NP NP or from NP NP to NP PP – according to those combinations of semantic
features available for the verb in question. This approach is actually reminiscent
of Harris’ (1957) notion of transformational relations between sets of surface
paradigms.
16. There might also be a difference regarding style, with different conditions on double
object alternations, in restricted and relaxed style:

(i) Restricted Style Constraint: verb stems whose only stress is initial allow double
object frames.
(ii) Relaxed Style Constraint: verb stems lacking secondary stress allow double object
frames.

The Relaxed Style Constraint in (ii) accounts for examples like the following, although
they are still excluded in the restricted style:

(iii) We delivered Mary the package.

(iv) Bill embroidered her a blouse.

17. Interestingly, however, Coppock (2009) showed that prosodic weight did play a role in
response time.
18. Den Dikken (1995, 158–159) accepts the logical non-equivalence of such pairs but argues
that this does not preclude a transformational relation between them; he suggests that
the difference between members of a dative alternation concerns aspect, not theta-roles.
Similarly, Hoekstra (1995, n. 5) maintains that “affectedness” rather than thematic role
distinctions differentiates the pairs.
19. Other authors have tried to distinguish the two versions of the dative alternation on
pragmatic grounds. Erteschik-Shir (1979) claims that the indirect object in double object
constructions tends to be definite (for her, “non-dominant”) and Thompson (1995) that
it is “more topicworthy.” However, such claims seem more concerned with language
use rather than syntax proper.
20. Citko (2011a) (also Citko 2011b) focuses on the factors that allow the lower object to
move over the higher one, especially in languages where this is the only option.
 Double Object Constructions 41

21. According to Holmberg and Platzack (1995, 189), Swedish allows symmetric passives
“more marginally” and Danish does not. Icelandic allows symmetric passives only with
a certain class of verbs, gefa-verbs, (i.e., give-verbs)
22. What might seem related is that wide-scope readings of the higher italic NPs are
also strongly favored in structures derived by subject raising (ia), passive (ib), and,
as pointed out by a reviewer, locative inversion (ic):

(i) a. Two new topics seemed to every student to be unfair.

b. Every student in the class might be reinvigorated by two new topics.
c. On two big billboards were (painted) every party’s slogans.

So perhaps we are dealing with a broader generalization here. However, it does not
seem to be a general factor against lack of inverse scope in A-movement configurations,
as shown by the following example, due to Aoun and Li (1993, 21), which does allow
inverse scope.

(ii) Someone seems to love everyone.

Furthermore, in languages with so-called fixed scope (i.e., Japanese, Chinese, Slavic
languages), A-movement creates new scope possibilities (and gives rise to scope ambi-
guities) rather than removes scope ambiguity.
23. An account of frozen scope in double object constructions that reduces it to superiority
makes an interesting cross-linguistic prediction. Since there are languages that allow vio-
lations of superiority in wh-movement contexts, Bruening’s account predicts that these
languages should also allow violations of superiority in QR contexts. This prediction
was tested by Stepanov and Stateva (2009) (see also Citko 2011a for relevant discussion).
24. These 12 languages (with their genetic affiliations in brackets) are: Dutch (Germanic);
Indonesian, Muna (Western Malayo-Polynesian), Savu (Central Malayo-Polynesian),
Nandi (Western Nilotic), Chichewa (Bantu), Mundari (Munda), Ainu (language isolate),
Kalkatungu (Pama-Nyungan), Kewa (East New Guinea Highlands), Sahaptin
(Sahaptian), Zapotec (Oto-Manguean). Siewierska adds other languages to her original
sample, which also have a dative alternation, which adds up to 38 languages total. Of
the Germanic group, these include English, Frisian, Swedish, Norwegian, for example.
25. He also links the lack of so-called prepositional complementizers to the same parameter.
26. More generally, Snyder and Stromswold (1997) argue for a parametric approach to lan-
guage acquisition, based on the fact that double objects and PP datives are acquired
together with a number of other (related) constructions (i.e., verb particle constructions,
put locatives and causative/perceptual constructions).

References

Aissen, Judith. 1983. “Indirect Object Advancement in Tzotzil.” In Studies in Relational Gram-
mar, 1: edited by David Perlmutter, 272–302. Chicago: Chicago University Press.
Akmajian, Adrian, and Frank Heny. 1975. An Introduction to the Principles of Transformational
Syntax. Cambridge, MA: MIT Press.
Alsina, Alex, and Sam Mchombo. 1990. “The Syntax of Applicatives in Chichewa: Problems
for a Theta-Theoretic Asymmetry.” Natural Language and Linguistic Theory, 8: 493–506.
Anagnostopoulou, Elena. 2003. The Syntax of Ditransitives: Evidence from Clitics. Berlin: De
Gruyter.
Aoun, Joseph, and Yen-Hui Audrey Li. 1989. “Scope and Constituency.” Linguistic Inquiry,
20: 141–172.
42 Double Object Constructions

Aoun, Joseph, and Yen-Hui Audrey Li. 1993. The Syntax of Scope. Cambridge, MA: MIT Press.
Bailyn, John. 1995. “A Configurational Approach to Russian ‘Free’ Word Order.” PhD diss.,
Cornell University.
Bailyn, John. 2012. The Syntax of Russian. Cambridge: Cambridge University Press.
Baker, Mark. 1985. “The Mirror Principle and Morphosyntactic Explanation.” Linguistic
Inquiry, 16: 373–416.
Baker, Mark. 1988a. Incorporation: A Theory of Grammatical Function Changing. Chicago, IL:
University of Chicago Press.
Baker, Mark. 1988b. “Theta Theory and the Syntax of Applicatives in Chichewa.” Natural
Language and Linguistic Theory, 6: 353–389.
Baker, Mark. 1997. “Thematic Roles and Syntactic Structure.” In Elements of Grammar, edited
by Liliane Haegeman, 73–137. Dordrecht: Kluwer.
Barss, Andrew, and Howard Lasnik. 1986. “A Note on Anaphora and Double Objects.” Lin-
guistic Inquiry, 17: 347–354.
Beck, Sigrid, and Kyle Johnson. 2004. “Double Objects Again.” Linguistic Inquiry, 35: 97–123.
Bowers, John. 1993. “The Syntax of Predication.” Linguistic Inquiry, 24: 591–656.
Bowers, John. 2010. Arguments and Relations. Cambridge, MA: MIT Press.
Bresnan, Joan. 2007. “Is Syntactic Knowledge Probabilistic? Experiments with the English
Dative Alternation.” In Roots: Linguistics in Search of Its Evidential Base, edited by Sam Feath-
erston and Wolfgang Sternefeld, 75–96. Berlin: De Gruyter.
Bresnan, Joan, Anna Cueni, Tatiana Nikitina, and Harald Baayen. 2007. “Predicting the
Dative Alternation.” In Cognitive Foundations of Interpretation, edited by Gerlof Boume,
Irene Kramer, and Joost Zwarts, 69–94. Amsterdam: Royal Netherlands Academy of
Science.
Bresnan, Joan, and Tatiana Nikitina. 2009. “The Gradience of the Dative Alternation.” In Real-
ity Exploration and Discovery: Pattern Interaction in Language and Life, edited by Linda Uyechi
and Lian Hee Wee, 161–184. Stanford, CA: CSLI.
Bresnan, Joan, and Lioba Moshi. 1990. “Object Asymmetries in Comparative Bantu Syntax.”
Linguistic Inquiry, 21: 147–185.
Bruening, Benjamin. 2001. “QR Obeys Superiority: Frozen Scope and ACD.” Linguistic
Inquiry, 32: 233–273.
Bruening, Benjamin. 2010a. “Double Object Constructions Disguised as Prepositional
Datives.” Linguistic Inquiry, 41: 287–305.
Bruening, Benjamin. 2010b. “Ditransitive Asymmetries and a Theory of Idiom Formation.”
Linguistic Inquiry, 41: 519–162.
Chomsky, Noam. 1955. “Logical Structure of Linguistic Theory.” MIT Humanities Library.
Microfilm.
Chomsky, Noam. 1981. Lectures on Government and Binding: The Pisa Lectures. Holland: Foris.
Chomsky, Noam. 1982. Some Concepts and Consequences of the Theory of Government and Bind-
ing. Cambridge, MA: MIT Press.
Chung, Sandra. 1976. “An Object-Creating Rule in Bahasa Indonesia.” Linguistic Inquiry, 7:
41–87.
Citko, Barbara. 2011a. Symmetry in Syntax: Merge, Move and Labels. Cambridge: Cambridge
University Press.
Citko, Barbara. 2011b. “A New Look at Symmetric and Asymmetric Passives.” In Proceedings
of the 39th North East Linguistics Society, edited by Suzi Lima, Kevin Mullin, and Brian
Smith, 191–204.
Coppock, Elizabeth. 2009. “The Logical and Empirical Foundations of Baker’s Paradox.” PhD
diss., Stanford University.
Cuervo, Christina. 2003. “Datives at Large.” PhD diss., MIT.
Culicover, Peter, and Ray Jackendoff. 2005. Simpler Syntax. Oxford: Oxford University Press.
 Double Object Constructions 43

Czepluch, Hartmut. 1982. “Case Theory and the Dative Construction.” Linguistic Review, 2:
1–38.
Dikken, Marcel den. 1995. Particles: On the Syntax of Verb-Particle, Triadic, and Causative
Constructions. New York, NY: Oxford University Press.
Dikken, Marcel den. 2006. Relators and Linkers. Cambridge, MA: MIT Press.
Dikken, Marcel den. 2007. “Phase Extension. Contours of a Theory of the Role of Head Move-
ment in Phrasal Extraction.” Theoretical Linguistics, 33: 1–41.
Doggett, Teal Bissell. 2004. “All Things Being Unequal: Locality in Movement.” PhD
diss., MIT.
Dornisch, Ewa. 1998. “Multiple Wh-Questions in Polish: The Interactions between
Wh-Phrases and Clitics.” PhD diss., Cornell University.
Dyakonova, Marina. 2007. “Russian Double Object Constructions.” ACLC Working Papers, 2
(1): 3–30.
Dyakonova, Marina. 2009. “A Phase-Based Approach to Russian Word Order.” PhD diss.,
Netherlands Graduate School of Linguistics.
Dziwirek, Katarzyna. 1994. Polish Subjects. New York, NY: Garland.
Emonds, Joseph. 1972. “Evidence that Indirect Object Movement is a Structure Preserving
Rule.” Foundations of Language, 8: 546–561.
Emonds, Joseph. 1976. A Transformational Approach to English Syntax: Root, Structure-
Preserving and Local Transformations. New York, NY: Academic Press.
Emonds, Joseph. 1993. “Projecting Indirect Objects.” Linguistic Review, 10: 211–263.
Erteschik-Shir, Nomi. 1979. “Discourse Constraints on Dative Movement.” In Discourse and
Syntax, edited by Talmy Givón, 441–467. New York, NY: Academic Press.
Everaert, Martin. 1990. “Movement ‘Across’ Secondary Objects.” In Grammar in Progress:
Glow Essays for Henk van Riemsdijk, edited by Joan Mascaró and Marina Nespor, 126–
136. Dordrecht: Foris.
Farrell, Patrick 1994. Thematic Relations and Relational Grammar. New York: Garland.
Fillmore, Charles. 1965. Indirect Object Constructions in English and the Ordering of Transforma-
tions. The Hague: Mouton.
Fraser, Bruce. 1974. The Verb-Particle Combination in English. Tokyo: Taishukan.
Gast, Volker. 2007. “I Gave It Him: On the Motivation of the ‘Alternative Double Object
Construction’ in Varieties of British English.” Functions of Language, 14: 31–56.
Georgala, Efthymia. 2012. “Applicatives in Their Structural and Thematic Function:
A Minimalist Account of Multitransitivity.” PhD diss., Cornell University.
Green, Georgia. 1974. Semantics and Syntactic Regularity. Bloomington: Indiana Univer-
sity Press.
Gropen, Jess, Steven Pinker, Michelle Hollander, et al. 1989. “The Learnability and Acquisi-
tion of the Dative Alternation in English.” Language, 65: 203–257.
Haddican, William. 2010. “Theme-Goal Ditransitives and Theme Passivisation in British
English Dialects.” Lingua, 120: 2424–2443.
Harley, Heidi. 2002. “Possession and the Double Object Construction.” Linguistic Variation
Yearbook, 2: 29–68.
Harris, Zellig. 1957. “Co-Occurrence and Transformation in Linguistic Structure.” Language,
33: 283–340.
Haspelmath, Martin. 2015. “Ditransitive Constructions.” Annual Review of Linguistics, 1,
19–41.
Hellan, Lars. 1991. “The Phrasal Nature of Double Object Clusters.” In Issues in Germanic
Syntax, edited by Werner Abraham, Wim Kosmeijer, and Eric Reuland, 67–92. Berlin:
De Gruyter.
Herriman, Jennifer. 1995. The Indirect Object Construction in Present-Day English. Gothenburg:
Acta Universitatis Gotoburgensis.
44 Double Object Constructions

Hoekstra, Teun. 1988. “Small Clause Results.” Lingua, 74: 101–139.

Hoekstra, Teun. 1995. “To Have to Be Dative.” In: Studies in Comparative Germanic Syn-
tax, edited by Hubert Haider, Susan Olsen, and Sten Vikner, 119–137. Dordrecht:
Kluwer.
Hoji, Hajime. 1985. “Logical Form Constraints and Configurational Structures in Japanese.”
PhD diss., University of Washington.
Holmberg, Anders, and Christer Platzack. 1995. The Role of Inflection in Scandinavian Syntax.
Oxford: Oxford University Press.
Holmberg, Anders, and Jan Rijkhoff. 1998. “Word Order in the Germanic Languages.” In
Constituent Order in the Languages of Europe, edited by Anna Siewierska, 75–104. Berlin:
De Gruyter.
Hughes, Arthur, and Peter Trudgill. 1979. English Accents and Dialects: An Introduction to the
Social and Regional Varieties in the British Isles. London: Edward Arnold.
Hornstein, Norbert, and Amy Weinberg. 1981. “Case Theory and Preposition Stranding.”
Linguistic Inquiry, 12: 55–92.
Jackendoff, Ray. 1990. “On Larson’s Account of the Double Object Construction.” Linguistic
Inquiry, 21: 427–456.
Jeong, Youngmi. 2007. Applicatives: Structure and Interpretation from a Minimalist Perspective.
Amsterdam: John Benjamins.
Kaplan, Ronald, and Joan Bresnan. 1982. “Lexical-Functional Grammar: A Formal System for
Grammatical Representation.” In The Mental Representation of Grammatical Relations, edited
by Joan Bresnan, 173–281. Cambridge, MA: MIT Press.
Kayne, Richard. 1981. “Unambiguous Paths.” In Levels of Syntactic Representations, edited by
Robert May and Jan Koster, 143–184. Dordrecht: Foris.
Kayne, Richard. 1983. “Connectedness and Binary Branching.” Linguistic Inquiry, 14:
223–249.
Kayne, Richard. 1984. Connectedness and Binary Branching. Dordrecht: Foris.
Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
Kimenyi, Alexandre. 1980. A Relational Grammar of Kinyarwanda. Berkeley, CA: University of
California Press.
Koopman, Hilda, and Dominique Sportiche. 1983. “Variables and the Bijection Principle.”
Linguistic Review, 2, 139–160.
Krifka, Manfred. 2004. “Semantic and Pragmatic Conditions for the Dative Alternation.”
Korean Journal of English Language and Linguistics, 4: 1–32.
Kupula, Mikko. 2011. “A Phase Extension Approach to Double Object Constructions:
Evidence from Modern Greek.” Studia Linguistica, 65: 147–169.
Larson, Richard. 1988. “On the Double Object Construction.” Linguistic Inquiry, 19: 335–391.
Larson, Richard. 1990. “Double Objects Revisited: Reply to Jackendoff.” Linguistic Inquiry, 21:
589–632.
Larson, Richard. 2010. “On Pylkkänen’s Semantics for Low Applicatives.” Linguistic Inquiry,
41: 701–704.
Lees, Robert B. 1963. The Grammar of English Nominalizations. The Hague: Mouton.
Marantz, Alec. 1982. “Affixation and the Syntax of Applied Verb Constructions.” In Proceed-
ings of the First West Coast Conference on Formal Linguistics, edited by Daniel P. Flickinger,
Marlys Macken, and Nancy Wiegand.
Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge, MA: MIT Press.
Marantz, Alec. 1993. “Implications of Asymmetries in Double Object Constructions.” In The-
oretical Aspects of Bantu Grammar, edited by Sam Mchombo, 1: 113–150. Stanford, CA: CSLI.
May, Robert. 1977. “The Grammar of Quantification.” PhD diss., MIT.
McGinnis, Martha. 1998. “Locality in A-Movement.” PhD diss., MIT.
 Double Object Constructions 45

McGinnis, Martha. 2001. “Variation in the Phase Structure of Applicatives.” Linguistic

Variation Yearbook, 1: 105–146.
McGinnis, Martha. 2008. “Applicatives.” Language and Linguistics Compass, 2: 1225–1245.
Miyagawa, Shigeru. 1995. “Scrambling as an Obligatory Movement.” In Proceedings of the
Nanzan University International Symposium on Japanese Language Education and Japanese
Language Studies, 81–92.
Miyagawa, Shigeru. 1997. “Against Optional Scrambling.” Linguistic Inquiry, 28: 1–26.
Miyagawa, Shigeru, and Takae Tsujioka. 2004. “Argument Structure and Ditransitive Verbs
in Japanese.” Journal of East Asian Linguistics, 13: 1–38.
Müller, Gereon. 1992. “In Support of Dative Movement.” In Proceedings of the Third Leiden
Conference for Junior Linguists, edited by Sjef Barbiers, Marcel den Dikken, and Claartje
Levelt, 201–217.
Ngonyani, Deogratias S. 1996. “The Morphosyntax of Applicatives.” PhD diss., University
of California, Los Angeles.
Oehrle, Richard T. 1976. “The Grammatical Status of the English Dative Alternation.” PhD
diss., MIT.
Paul, Waltraud, and John Whitman. 2010. “Applicative Structure and Mandarin Ditransi-
tives.” In Argument Structure and Syntactic Relations from a Cross-Linguistic Perspective, edi-
ted by M. Duguine, S. Huidobro, and N. Madariaga 2010, 261–282. Amsterdam: John
Benjamins.
Perlmutter, David, ed. 1983. Studies in Relational Grammar, vol. 1. Chicago, IL: University of
Chicago Press.
Perlmutter, David, and Paul M. Postal. 1983. “Some Proposed Laws of Basic Clause Struc-
ture.” In Studies in Relational Grammar, vol. 1, edited by David Perlmutter, 81–128.
Chicago, IL: University of Chicago Press.
Perlmutter, David, and Carol Rosen, eds. 1984. Studies in Relational Grammar, vol. 2. Chicago,
IL: University of Chicago Press.
Pesetsky, David. 1995. Zero Syntax: Experiencers and Cascades. Cambridge, MA: MIT Press.
Peterson, David A. 2007. Applicative Constructions. Oxford: Oxford University Press.
Pinker, Steven. 1989. Learnability and Cognition: The Acquisition of Argument Structure.
Cambridge, MA: MIT Press.
Polinsky, Maria. 2005. “Applicative Constructions.” In The World Atlas of Language Structures,
edited by Martin Haspelmath, Mattew S. Dryer, David Gil, and Bernard Comrie, 442–446.
Oxford: Oxford University Press.
Postal, Paul M. 2010. Edge-Based Clausal Syntax: A Study of (Mostly) English Object Structure.
Cambridge, MA: MIT Press.
Pylkkänen, Liina. 2002. “Introducing Arguments.” PhD diss., MIT.
Pylkkänen, Liina. 2008. Introducing Arguments. Cambridge, MA: MIT Press.
Rappaport Hovav, Malka, and Beth Levin. 2008. “English Dative Alternation: The Case for
Verb Sensitivity.” Journal of Linguistics, 44: 129–167.
Richards, Norvin. 1997. “What Moves Where When in Which Language?” PhD diss., MIT.
Richards, Norvin. 2001. “An Idiomatic Argument for Lexical Decomposition.” Linguistic
Inquiry, 32: 183–192.
Riemsdijk, Henk C. van. 1978. A Case Study in Syntactic Markedness: The Binding Nature of
Prepositional Phrases. Dordrecht: Foris.
Ross, John Robert. 1967. “Constraints on Variables in Syntax.” PhD diss., MIT.
Siewierska, Anna. 1998. “Languages With and Without Objects: A Functional Grammar
Approach.” Languages in Contrast, 1: 173–190.
Snyder, William, and Karin Stromswold. 1997. “The Structure and Acquisition of English
Dative Constructions.” Linguistic Inquiry, 28: 281–317.
46 Double Object Constructions

Stechow, Arnim von. 1996. “The Different Readings of Wieder ‘Again’: A Structural
Account.” Journal of Semantics, 13: 87–138.
Stepanov, Arthur, and Penka Stateva. 2009. “When QR Disobeys Superiority.” Linguistic
Inquiry, 40: 176–185.
Stowell, Tim. 1981. “Origins of Phrase Structure.” PhD diss., MIT.
Sugisaki, Koji, and William Snyder. 2006. “The Parameter of Preposition Stranding: A View
from Child English.” Language Acquisition, 13: 349–361.
Thompson, Sandra. 1995. “The Iconicity of ‘Dative Shift’ in English.” In Syntactic Iconicity
and Linguistic Freezes, edited by Marge E. Landsberg, 155–175. Berlin: Mouton.
Ura, Hiroyuki. 1996. “Multiple Feature Checking: A Theory of Grammatical Function Split-
ting.” PhD diss., MIT.
Whitney, Rosemarie. 1982. “The Syntactic Unity of Wh-Movement and Complex NP-Shift.”
Linguistic Analysis, 10: 299–319.
Whitney, Rosemarie. 1983. “The Place of Dative Movement in a Generative Theory.” Linguis-
tic Analysis, 12: 299–319.
Whitney, Rosemarie. 1984. “The Syntax and Interpretation of A’-Adjunctions.” PhD diss.,
University of Washington.
Woolford, Ellen. 1993. “Symmetric and Asymmetric Passives.” Natural Language and Linguis-
tic Theory, 11: 679–728.
Zhang, Shi. 1990. “Correlations between Double Object Construction and Preposition
Stranding.” Linguistic Inquiry, 21: 312–316.
Ziv, Yael, and Gloria Sheintuch. 1979. “Indirect Objects Reconsidered.” In Papers from the Fif-
teenth Regional Meeting of the Chicago Linguistic Society, edited by Paul R. Clyne, William F.
Hanks, and Carol. L. Hofbauer, 390–403. Chicago, IL: Chicago Linguistic Society.