Content uploaded by Ning Zhang
Author content
All content in this area was uploaded by Ning Zhang on Jul 14, 2015
Content may be subject to copyright.
folia
linguistica
30 2 2006
folia
linguistica
Folia Linguistica 41/1–2 (2007), 135–169.
ISSN 0165–4004, E-ISSN 1614–7308 © Mouton de Gruyter – Societas Linguistica Europaea
The syntax of English comitative constructions
1
Niina Ning Zhang
National Chung Cheng University
This paper argues that in comitative constructions, two Determiner Phrases (DPs)
and the word with form a complex nominal [DP
1
with DP
2
] in their base-positions.
Moreover, it establishes a distinction between two types of comitative constructions:
symmetrical and asymmetrical. Unlike symmetrical comitatives, DP
2
in asymmetrical
comitatives denotes an appurtenance of the referent of DP
1
. Other differences
between the two types are that symmetrical comitatives always contain a plural
feature and that in asymmetrical comitatives the cluster [with DP
2
] is optional and
the cluster [DP
1
with DP
2
] may occur in preverbal subject positions and A-bar
positions. My proposal is that in symmetrical comitatives the word with has features
of [D, Plural, Case assigning], and takes DP
2
as its complement and DP
1
as its
Specifier. In asymmetrical comitatives, by contrast, the word with is a regular
preposition and the cluster [with DP
2
] is a Prepositional Phrase and an adjunct to
DP
1
. Finally, the paper also argues against an extraposition analysis of discontinuous
surface word orders in comitative constructions.
Keywords: comitative constructions, coordination, extraposition, Case, plural,
categorial features
1
Earlier versions of this paper have received help from many people: David Adger,
Philippa Cook, Laura Downing, Tom Ernst, Paul Law, James Myers, Benjamin
Shaer, Peter Svenonius, Chris Wilder, the editor and two anonymous reviewers of
Folia Linguistica, and the participants in the 5th GLOW in Asia (New Delhi, October
5–8, 2005) and NELS 36 (University of Massachusetts Amherst, October 28–30,
2005). I am grateful to them all, as also to the National Science Council of Taiwan
for generous financial support via grant 93–2411–H–194–027.
136 Niina Ning Zhang
1. Introduction
This paper studies a type of with construction in English in which the
Determiner Phrase (DP) immediately following with (DP
2
henceforth) has
the same animacy value as another DP in the clause (DP
1
henceforth). For
instance, in the two examples in (1), DP
2
Kim is animate, so is DP
1
Robin; on
the other hand, in the two examples in (2), DP
2
the milk is inanimate, so is
DP
1
the beer. Such constructions are known in the literature as COMITATIVE
CONSTRUCTIONS.
(1) a. Robin conferred with Kim.
b. Robin sang with Kim.
(2) a. Peter compared the beer with the milk.
b. Robin drank the beer with the milk.
In a comitative construction, the two DPs are either both necessary for a
certain eventuality, or have an accompanimental relation in the same
eventuality. In (1a), both Robin and Kim are necessary for the conference
event, and in (2a), both the beer and the milk are necessary for the
comparison event. In (1b) and (2b) DP
1
and DP
2
have an accompanimental
or togetherness relation in the event.
The syntactic properties of English comitative constructions have been
studied by Lakoff & Peters (1966, reprinted 1969; henceforth L&P), Fillmore
(1968), Buckingham (1973), Walmsley (1971), Seiler (1974), Kayne (1994),
and Stolz (2001), among others.
2
However, one basic syntactic issue is still
not clear: what is the structural relation between DP
1
and the cluster [with
DP
2
]?
Considering the plural form of the predicate is friends in (3a), Kayne
(1994: 66) suggests that its subject should also encode a plural entity. The
surface subject of the sentence is however John, which encodes a singular
entity. Kayne therefore assumes that the two nominals in this sentence, John
and Bill, form a complex nominal in their base positions, as shown in (3b).
This assumed complex, which denotes a plural entity, is the real subject of
the plural predicate friends. He claims that the surface order of (3a) is derived
by the movement of John from this complex, as shown in (3c).
2
From a typological perspective, comitatives have now been studied by Lehmann &
Shin (2005) and Stolz, Stroh & Urdze (2006).
The syntax of English comitative constructions 137
(3) a. John is friends with Bill.
b. [John with Bill]
c. John
i
is friends [t
i
with Bill]
But on the same page (p. 66), he mentions a second possibility: the with in
such examples might be introducing a verbal or adjectival complement,
which means that [with Bill] in (3a) might not be part of any nominal
complex, but rather a modifier of the predicate is friends.
In this paper, I will first adduce new arguments to support Kayne’s
constituency analysis in (3b), where DP
1
and the cluster [with DP
2
] form a
complex nominal in their base-positions, and thus the cluster [with DP
2
] is
not a VP adverbial. Secondly, I will distinguish
SYMMETRICAL comitatives
from
ASYMMETRICAL comitatives, based on a series of syntactic differences
between them. In order to account for the differences, I will propose that the
two DPs of a symmetrical comitative are base-generated as in the
complementation structure in (4a), as claimed in Kayne (1994), whereas the
two DPs of an asymmetrical comitative are base-generated as in the structure
in (4b), where the PP is a right adjunct of DP
1
.
(4) a. XP b. DP
DP
1
X’ DP
1
PP
X DP
2
P DP
2
with with
My analysis differs from Kayne’s (1994) in two major respects. One is that
I identify two types of comitative constructions and provide an explanation
for their differences. The other is that for the structure in (4a), Kayne (1994)
proposes that the XP is a coordinate phrase, headed by either and or with,
whereas in my analysis, the XP is a DP headed by with, which has the features
[D, Plural, Case assigning]. Note also that this with shares [D] and [Plural]
with regular nominals, and [Case assigning] with regular prepositions.
Like Kayne, I claim that in the constructions where DP
1
is not adjacent to
the cluster [with DP
2
], as in (1) and (3), it is raised to the surface position. I
will accordingly argue against an extraposition analysis of comitative
constructions, such as that in (5).
138 Niina Ning Zhang
(5) [John t
i
] is friends [with Bill]
i
The paper is organized as follows. Section 2 provides new evidence for the
constituency in (3b). Section 3 presents the syntactic contrasts between
symmetrical and asymmetrical comitative constructions. My explanations for
the contrasts are put forward in Section 4. Section 5 examines comitative
constructions with discontinuous word orders and argues against an
extraposition analysis. Section 6 is a summary.
2. Comitative and coordinate nominal constructions
In this section, I argue that comitative constructions differ syntactically from
coordinate constructions, but like the conjuncts of a coordinate nominal, the
two DPs of a comitative construction are contained in a complex nominal in
their base-positions.
In the generative literature, the relationship between comitative and
coordinate constructions has not been made clear. In Fillmore (1968),
coordinate constructions are derived from comitative constructions, whereas
in L&P it is just the reverse. In Kayne (1994), both constructions are derived
from the same base-structure. It is claimed that the two nominals of a
comitative construction behave like conjuncts, and with behaves likes a
conjunction. I will argue against this coordination analysis in §2.1. However,
I will support in §2.2 the claim made in Fillmore (1968), L&P, and Kayne
(1994) that the two nominals in a comitative construction are base-generated
in a complex nominal.
2.1. Differences
Comitative constructions differ from coordinate constructions in at least
three respects, as detailed below.
2.1.1. DP
1
raising and the Coordinate Structure Constraint
In a comitative construction, it is always possible for DP
1
to be separated from
the cluster [with DP
2
]:
The syntax of English comitative constructions 139
(6) a. John is friends with Bill.
b. John drank beer with Mary.
c. The apple is compared with the orange.
According to Kayne (1994), in this kind of examples DP
1
is raised out of the
assumed coordinate complex. However, as pointed out by van Oirsouw (1987:
13), this raising of DP
1
out of the XP violates Ross’s (1967) Coordinate
Structure Constraint, which states that no conjunct may be moved.
2.1.2. The A-bar positions of the cluster [with DP
2
]
The mobility of the cluster [with DP
2
] in comitative constructions, as seen in
(7), suggests that the cluster is different from the combination of a
conjunction and a conjunct, which cannot move, as shown in (8) (Sledd 1959:
101, Huddleston & Pullum 2002: 1291).
(7) a. With which apple did Mary compare the orange?
b. (Together) with his mother, John went to a Chinese restaurant.
(8) a. John and his mother went to a Chinese restaurant.
b. *And his mother, John went to a Chinese restaurant.
This contrast in behaviour between the cluster [with DP
2
] and the cluster
[and DP
2
] does not support Kayne’s hypothesis that the structure of a
comitative construction is the same as that of a coordinate construction.
In §4.1.2 and §4.2.2 below I will argue that in cases such as (7) the
element that undergoes the A-bar movement is not the constituent [with
DP
2
] alone, but the whole complex [DP
1
with DP
2
], after DP
1
has moved out.
This is a kind of
REMNANT MOVEMENT, as described in den Besten &
Webelhuth (1990) and Baltin (2002, 2006: 734): a part of a phrase is
extracted, and then the rest of the phrase is moved.
If with and coordinators behave the same way, one wonders why (8b)
cannot be generated by the same derivation.
2.1.3. The occurrence of together to the left of with
In some comitative constructions, the adverb together may precede with. By
contrast, genuine conjunctions can never be modified by together:
140 Niina Ning Zhang
(9) a. A mother (together) with her two kids came.
b. A mother (*together) and her two kids came.
We conclude that with in a comitative construction is never a conjunction,
and therefore English comitative constructions are not syntactically
coordinate (contrary to the conclusion drawn from other languages by
Camacho 2000 or Vassilieva & Larson 2005: 103, among others). Further
differences between comitatives and coordinate constructions with respect to
information structure will be adduced in §5.2.
2.2. Similarities
Despite the differences discussed above, comitative constructions and
coordinate nominal constructions share a set of important properties that
indicate that the two DPs are base-generated in the same complex nominal,
so that the cluster [with DP
2
] is not a VP adverbial. Among the properties
listed in this subsection, the second and the third have been reported in L&P.
2.2.1. Overt presence of the cluster [with DP
2
]
If the first conjunct is singular and the verb or predicate selects for a plural
nominal, the presence of the cluster [and DP
2
] is obligatory. This is shown in
(10a) and (11a). Similarly, if DP
1
is singular and the relevant verb or predicate
requires a plural nominal, the presence of the cluster [with DP
2
] is obligatory.
This is shown in (10b) and (11b).
(10) a. John *(and Bill) conferred.
b. John conferred *(with Bill).
(11) a. John compared Mary *(and Bill).
b. John compared Mary *(with Bill).
This shared property of coordinate nominals and comitative constructions
shows that the cluster [with DP
2
] in neither subject nor object comitative
constructions is a VP adverbial.
The syntax of English comitative constructions 141
2.2.2. The theta-role unification
In a coordinate nominal, the conjuncts cannot have opposite thematic roles
(Johannessen 1998: 253ff, among others): an agent and a patient cannot be
conjoined. Similarly, comitative constructions show unification of theta-roles
(L&P: 120). The theta-role unification of Bill and John in (12a) is explained if
the two DPs are base-generated in a DP complex, which has a single thematic
relation with the verb kill. Similarly, the theta-role unification of Bill and John
in (12b) is explained if the two DPs are base-generated in a DP complex,
which has a single thematic relation with the passive verb.
(12) a. John killed with Bill. (Like John, Bill was a killer)
b. John was killed with Bill. (Like John, Bill was the victim of a killer)
The theta-role unification of the two DPs in a comitative construction
distinguishes the accompaniment reading of the construction from other
readings of non-comitative with constructions. In (13), the amusing effect
comes from the availability of two structural relations for with his wife: it can
either be in construal with the agent Heinrich Christian Schwan, to form a
comitative construction, or with the instrument with cookies, colored ribbons,
nuts, and candles.
(13) Heinrich Christian Schwan, the newly appointed pastor, chopped down an
evergreen in the forest near his parsonage, decorated it with his wife,
Emma, with cookies, colored ribbons, nuts, and candles, and, according to
the church’s Web site, placed it in a prominent spot in the chancel.
(Quoted as a filler in The New Yorker, Dec. 18, 2006, p. 94)
If with his wife is in construal with Heinrich Christian Schwan, both have an
agent reading. In the syntactic literature the terms “comitative role” or
“comitative case” are sometimes applied to the semantic function of DP
2
(e.g.
Fillmore 1968, Walmsley 1971). However, this semantic role is never
independent of the role of DP
1
in the comitative construction in question.
2.2.3. The do so test
Like a conjunct in a coordinate nominal or any other component of complex
nominals in general, but unlike adverbials, the [with DP
2
] of comitative
constructions cannot be stranded in the VP proform do so.
142 Niina Ning Zhang
The form do so has generally been assumed to be a proform of VP, which
contains a verb and its arguments (Lakoff & Ross 1966, Ross 1970, among
others; see also Hallman 2004, who argues that the words do and so are
realizations of v and VP, respectively). In (14a), did so replaces read the book.
(14b) is ungrammatical because, as did so pronominalizes the whole VP,
magazine is not licensed.
(14) a. John read the book on Monday, and Mary did so on Tuesday.
b. *John read the book on Monday, and Mary did so the magazine.
Returning now to comitative constructions, consider the following data;
(15b,c) and (17) are from L&P (p. 119, fn. 5), (16c) is from Grosu (1976: 645),
and the rest were confirmed with native speakers of English:
(15) a. *John compared A with B, and Mary did so with C. [object
comitative]
b. *I drink milk with meat and John does so with fish. [object
comitative]
c. *I stole the warden’s wallet with his keys and John did so with his
glasses. [object comitative]
(16) a. John ate the dinner with good cheer, and Harry did so with no
cheer.
b. John ate the dinner slowly, and Harry did so with great speed.
c. John sliced a large piece of cake with a shining new knife before
Mary had a chance to do so with a rusty old blade.
(17) John killed a man with Bill and Harry did so with Tom. [subject
comitative]
The sentences in (15) are object comitative constructions whereas the
sentences in (16) are not. The with-PPs in the latter group are manner
expressions, which are adjuncts of the predicates. If we analyzed the [with
DP
2
] in (15) as adjuncts of the predicates as well, we would not be able to
explain the contrast in acceptability between (15) and (16). However, the
contrast can be accounted for if [A with B] in the first clause in (15a) is
treated as a complex object. The intended object in the second clause is [A
with C], and this makes the occurrence of the VP proform does so
problematic, since part of the object, [with C], is stranded, and the
well-established syntax of do so does not allow any stranding of objects or
The syntax of English comitative constructions 143
object fragments, as seen in (14b). The unacceptability of (15b) and (15c)
can be accounted for in the same way.
In (16a) and (16b), however, the with-phrases are manner adverbials, and
in (16c), the with-phrases are instrumentals. The phrase did so in (16a)
legally substitutes for the VP, which contains the transitive verb ate and its
object the dinner. Thus the acceptability of the sentence is expected. (16b)
patterns with (16a), except for the fact that the manner expression in the first
clause is an adverb (slowly) rather than a with-PP. The phrase do so in (16c)
legally substitutes for the VP sliced a large piece of cake. Thus the acceptability
of the sentence is also expected.
In (17), with Tom in the second clausal conjunct is not part of the object,
either. Instead, it is semantically associated with the local subject, Harry. The
do so legally substitutes for the VP, which contains the transitive verb kill and
its object a man. Thus the acceptability of the example is expected. I will
discuss the derivation of this sentence in §5.1 below (cf. (63a)).
2.2.4. The gapping test
Like a conjunct in a coordinate nominal or any other constituent of a
complex nominal and unlike adverbials, the [with DP
2
] of comitative
constructions cannot be stranded by gapping.
Gapping is subject to the Major Constituent Condition (Hankamer 1973:
18), whereby “[a] ‘Major Constituent’ of a given sentence S
0
is a constituent
either immediately dominated by S
0
or immediately dominated by VP, which
is immediately dominated by S
0
.” Roughly speaking, complete subjects,
objects, or adverbials are major constituents, whereas a fragment of any of
them is not. The Major Constituent Condition states that the remnants of
gapping must be major constituents. In (18a), the remnants of gapping are
Akira and with a brush. One is a complete subject and the other is a complete
adverbial. Thus the sentence is acceptable. In contrast, (18b) is
ungrammatical because a brush is part of the instrumental adverbial with a
brush.
(18) a. John writes with a pencil and Akira _ with a brush.
b. *John writes with a pencil and Akira _ a brush.
144 Niina Ning Zhang
Consider the contrasts in acceptability between the following examples;
(21c) is cited from Hankamer (1973: 18), and the rest were confirmed with
native speakers of English:
(19) a. I enjoy beer with pretzels and he _ beer with peanuts. [object
comitative]
b. *I enjoy beer with pretzels and he _ with peanuts.
(20) a. *I drink milk with meat and John _ with fish. [object comitative]
b. *I stole the warden’s wallet with his keys and John _ with his
glasses.
(21) a. John ate the dinner with good cheer, and Harry _ with no cheer.
b. Max eats with chopsticks, and Albert _ with a fork.
c. Max spoke fluently, and Albert _ haltingly.
(22) I went to London with Bill and Harry _ with Tom. [subject comitative]
The sentences in (19) and (20) are object comitative constructions,
whereas the sentences in (21) are not. The with-PPs in the latter are adjuncts
of the predicates. If we analyzed the with-PPs in (19)–(20) also as adjuncts of
the predicates, we would not be able to explain the contrast in acceptability
between (19a) and (19b), nor that between (19b)–(20), and (21). However,
the contrasts can be accounted for if [beer with pretzels] in the first clause of
(19b) is treated as a complex object. If so, then the intended object in the
second clause should be [beer with peanuts]; since with peanuts is part of the
comitative object, it is not a major constituent and therefore gapping as in
(19b) is disallowed. Similarly, in (20a), if with fish is part of a complex object,
it is not a major constituent, and thus it cannot be retained after gapping.
The same applies to with his glasses in (20b), which is also part of a complex
object. This accounts for the ungrammaticality of (19b), (20a) and (20b).
In (21a) and (21b), however, the with-phrases are manner and instrument
adverbials respectively, just like the adverb in (21c). The gapping in these
sentences is legal, since the adverbials are major constituents.
In (22), with Tom is not part of the object, either. Instead, it is
semantically associated with the local subject Harry and can be regarded as a
major constituent, containing the trace of the surface subject Harry. I will
discuss the derivation of this sentence in §5.1 (cf. (63b)).
The syntax of English comitative constructions 145
2.2.5. Summary
The results yielded by the application of the do so (§2.2.3) and gapping
(§2.2.4) tests show that the comitative object [DP
1
with DP
2
] is
base-generated as a constituent. If DP
1
alone were the object and the cluster
[with DP
2
] were an adjunct of the predicate, we would not be able to account
for the above-mentioned contrasts in acceptability.
The other two items of evidence presented in §2.2.1 and §2.2.2 relate to
both subject and object comitative constructions (see also §3.2 below) and
also support my claim that comitative with-clusters are not adverbials, a
claim implied in Kayne (1994) but not explicitly argued for.
3. Two types of comitative constructions
Having reviewed evidence that the two DPs of a comitative construction are
base-generated inside a complex DP, I now examine a number of systematic
variations that can be observed in comitative constructions.
I have found that there are two types of comitative constructions, namely
symmetrical and asymmetrical, depending on whether DP
2
plays a role of
equal importance as DP
1
in the relevant eventuality. Specifically, if the two
DPs together satisfy the selection requirements of a collective verb (e.g.
compare, mix) or predicate (e.g. be friends), DP
2
plays a role of equal
importance as DP
1
in the eventuality, and thus the construction is a
symmetrical comitative construction. Witness (23):
(23) a. John is friends with Bill. [sym: selected]
b. John mixed the rice with the powder. [sym: selected]
If the verb or predicate is not collective, there are two possibilities. If DP
2
functions as an appurtenance
3
of DP
1
, such as an assistant (cf. his staff officers
in (24a)), a follower (cf. their blind mother in (24b)), or an accompaniment
(cf. rice in (24c)), the comitative construction is asymmetrical. Otherwise, it
is symmetrical.
3
For the term APPURTENANCE in connection with with constructions such as (24b)
see Lindstromberg (1998: 209–210).
146 Niina Ning Zhang
(24) a. The general (together) with his staff officers will inspect the camp.
b. Two small children (together) with their blind mother came
through the door.
c. Curry with rice is my favorite dish.
With non-collective verbs or predicates, the occurrence of the word
together to the left of with signals an asymmetrical reading. Thus, as pointed
out by both my informants and a discussion in http://forum.
wordreference.com/showthread.php?t=163085, (25a) encodes that John and
Mary are symmetrical partners, whereas (25b) encodes that John played the
major role.
(25) a. John baked a cake with Mary. [sym]
b. John baked a cake together with Mary. [asym]
Similarly, the occurrence of the phrase by X’s side to the right of [with
DP
2
], where X is a pronoun coreferential with DP
1
, also signals an
asymmetrical reading.
(26) a. John will drink beer with Bill. [sym]
b. John will drink beer with Bill by his side. [asym]
Symmetrical and asymmetrical comitatives are syntactically different in at
least four ways, as discussed in what follows.
3.1. The optionality of [with DP
2
]
As pointed out above, verbs or predicates that select for plural dependents are
excluded, by definition, from asymmetrical comitative constructions. With
this kind of verbs and predicates, if DP
1
is singular, the cluster [with DP
2
] is
obligatory. By contrast, in asymmetrical comitative constructions the cluster
[with DP
2
] is always syntactically optional. This is shown in (27) and (28).
(27) a. A mother came through the door (together with her two children).
[asym]
b. John is friends *(with Bill). [sym]
(28) a. John ate the apple (together with the orange). [asym]
b. John compared Berlin *(with Paris). [sym]
The syntax of English comitative constructions 147
3.2. The cluster [DP
1
with DP
2
] in preverbal subject position
In symmetrical comitatives the cluster [DP
1
with DP
2
] cannot surface as
preverbal subject, unlike in asymmetrical comitatives:
(29) a. *John with Bill are friends. [sym: selected]
b. *John with his wife collided. [sym: selected]
c. *John with Bill baked a cake. [sym]
d. *John with Bill will drink beer. [sym]
e. *John with Bill were killed in the accident. [sym]
(30) a. John together with Bill baked a cake.
b. John with Bill by his side will drink beer.
c. A mother with her two kids were killed in the accident.
d. Sally’s rude behavior together with Bill’s polite reactions
(obviously) amused John.
e. Fish (together) with wine sauce tastes good.
The fact that the cluster [with DP
2
] can occur preverbally distinguishes it
from VP adverbials:
(31) a. *John with his good appetite ate the dinner. (without a pause
before with)
b. *John with his glasses saw the insects.
Examples such as (30) can be uttered without a pause before or after the
with-cluster. However, examples such as (31a,b) are only possible when there
is a clear pause before and after the with-PP, as seen in (32). The pauses
indicate that the PP is a parenthetical and not part of the subject nominal.
(32) John, with his good appetite, ate the dinner.
The non-adverbial status of the cluster [with DP
2
] is also shown by the fact
that its position is to the left of modals, auxiliaries, and sentential adverbs, as
seen in (30b), (30c), and (30d), respectively. It should be remembered that
VP adverbials cannot occur to the left of these three types of elements.
That the preverbal cluster [DP
1
with DP
2
] is a constituent in asymmetrical
comitative constructions is also attested by the fact that it can be conjoined
with another DP, as in (33a), or with another cluster of the same structure, as
in (33b):
148 Niina Ning Zhang
(33) a. A teacher and a student with his father arrived at the same time.
b. A father with his son and a mother with her daughter all came
through the same door.
Finally, the subject status of the preverbal cluster [DP
1
with DP
2
] in
asymmetrical comitatives is shown by its ability to control PRO, as in (34a)
and (34b). Such a property is typical of subjects.
(34) a. A father (together) with his two children traveled ten miles PRO to
attend the class.
b. A mother with her two kids tried PRO to move the piano
themselves.
3.3. A-bar movement of the cluster [DP
1
with DP
2
]
The cluster [DP
1
with DP
2
] cannot undergo A-bar movement in symmetrical
comitatives, unlike in asymmetrical comitatives:
(35) a. *The apple with the orange, Mary compared. [sym: selected]
b. *Which apple with the orange did Mary compare? [sym: selected]
c. *John with Bill, Mary saw. [sym]
d. *Which guy with Bill did Mary see? [sym]
e. *This is the boy with Mary that I saw. [sym]
(36) a. A mother (together) with her two kids, Mary saw.
b. Fish (together) with wine sauce, Mary ate.
c. Which mother (together) with her two kids did Mary see?
d. Which kind of fish (together) with wine sauce did Mary eat?
e. This is the fish with wine sauce that I bought.
3.4. The presence of a plural feature with a singular DP
1
Symmetrical comitative constructions always exhibit a plural feature, even
when DP
1
is singular. In contrast, asymmetrical comitative constructions do
not have this property. The default plural feature of a symmetrical comitative
construction can be determined in two ways.
First, symmetrical comitative constructions can satisfy the plurality
requirement of collective predicates and verbs, even when both DPs are
singular. In (37a), for instance, the collective predicate confer requires its
external argument to be a non-singular element. This requirement is not
The syntax of English comitative constructions 149
satisfied by either the surface singular subject John, as shown by the
unacceptability of (37b), or the singular nominal Bill.
(37) a. John conferred with Bill.
b. *John conferred.
Similarly, in (38a), the comitative complex Berlin with Paris satisfies the
plurality requirement of the verb compare, although neither Berlin nor Paris
alone can do so. (38b) confirms this same point.
(38) a. Peter compared Berlin with Paris.
b. John combined butter with sugar.
Second, symmetrical comitative constructions can license the plural
feature of plural nominal predicates, even when both DP
1
and DP
2
are
singular (Kayne 1994: 66f). This is illustrated in (39). The noun friends could
be replaced by mates, pals, buddies, twins, enemies, rivals, and so on.
(39) a. John is friends with Bill.
b. *John is friends.
The comitative construction in (39a) contrasts with the unacceptable (39b).
The plural nominal predicate friends is not licensed in the latter, even though
both examples have a singular surface subject. Note that no plural agreement
is triggered on T in (39a) simply because the Specifier of T is filled by the
singular DP
1
alone.
This default plural feature is not observable in asymmetrical comitatives.
By definition, collective verbs or predicates are excluded from asymmetrical
comitatives. Yet it is still possible to check whether a default plural feature is
attested when DP
1
is singular. As pointed out in §3.2, the cluster [DP
1
with
DP
2
] can surface as preverbal subject in asymmetrical comitative
constructions. If these also had a plural feature by default, we would expect
that the complex [DP
1
with DP
2
], when used as preverbal subject, would
trigger plural agreement, regardless of the number features of DP1. However,
this is not the case. If DP
1
is singular, then the verb is singular too:
150 Niina Ning Zhang
(40) a. Jack (together) with several of his noisy friends was drinking till
after 2 in the morning.
b. Sally’s rude behavior together with Bill’s polite reactions amuses
John. (L&P: example (34))
Thus the number feature of the complex [DP
1
with DP
2
] in asymmetrical
comitative constructions is determined by DP
1
alone. In other words,
asymmetrical comitative constructions with a singular [DP
1
] have a singular
feature available by default.
The four contrasts presented in this section are summarized in (41):
(41) Four contrastive properties of symmetrical and asymmetrical comitative
constructions
[with DP
2
] is
always
optional
[DP
1
with DP
2
]
occurs in
preverbal subject
position
[DP
1
with DP
2
]
occurs in A-bar
positions
plural feature,
even when DP
1
is singular
symmetrical - - - +
asymmetrical + + + -
4. The with-cluster in relation to complementation and adjunction
structures
In order to account for the contrasts outlined above, in this section I propose
different syntactic structures for symmetrical and asymmetrical comitative
constructions.
In Kayne (1994), the word with in comitative constructions is treated as a
head element, and the two DPs in the construction are analyzed as the
Specifier and Complement of with, as in (42a). Another possible structure for
comitative constructions is the nominal adjunction structure in (42b). Here
the PP is an adjunct of DP
1
; unlike in (42a), in this adjunction structure with
is not the head of the complex nominal.
The syntax of English comitative constructions 151
(42) a. XP b. DP
DP
1
X’ DP
1
PP
X DP
2
P DP
2
with with
In both structures, DP
1
c-commands DP
2
. Since binding is licensed by
c-command (see Baltin 2006 for a recent clarification of this issue), the
acceptability contrast between (43a) and (43b), and between (44a) and (44b),
can be accounted for by either structure:
(43) a. I compared every boy
i
with his
i
girlfriend.
b. *I compared his
i
girlfriend with every boy
i
.
(44) a. The opponents
i
sat down with each other
i
’s examiners.
b. *Each other
i
’s examiners sat down with the opponents
i
.
However, the syntactic facts presented in §3.1 through §3.4 suggest that both
(42a) and (42b) are needed to adequately account for comitative
constructions. Specifically, I claim that symmetrical comitative constructions
have the complementation structure in (42a), whereas asymmetrical
comitative constructions, which Kayne (1994) does not discuss, have the
adjunction structure in (42b).
4.1. The complementation structure of symmetrical constructions
4.1.1. The configuration
My proposed structure for symmetrical comitative constructions is similar to
Kayne’s (1994) structure in (42a) in the sense that DP
1
is the specifier and
DP
2
is the complement of with. This analysis of symmetrical comitative
constructions as complex DPs serves to explain the non-adverbial features of
the cluster [with DP
2
] presented in §§2.2.1–4. The proposed
complementation configuration also captures the first contrastive property
of symmetrical and asymmetrical comitatives: the cluster [with DP
2
] is not
always optional, and thus cannot be an adjunct (cf. §3.1).
152 Niina Ning Zhang
4.1.2. The formal features
In order to account for the other three properties of symmetrical comitatives
reported in §3.2 through §3.4, I claim that with in symmetrical comitatives
carries three important formal features, which are shared with both D and P.
They are [D], [plural], and [+ Case assigning].
(45) DP [-Case]
DP
1
D’
D DP
2
with
[D]
[ϕ:
PL]
[Case assigning]
First, I claim that the word with in symmetrical comitatives has the categorial
feature [D], and is the head of the whole complex nominal.The D feature of
with accounts for the category of the projected complex. Under the
assumption that c-selection must be satisfied locally (i.e., by either a
complement or a specifier, rather than any element inside a complement or a
specifier), the D feature of with enables the whole complex in its
base-position to satisfy the c-selection of selecting verbs (e.g. combined and
drank in (46a)-(46b)), predicates (e.g. friends in (46c)), or prepositions (e.g.
of in (46d)).
(46) a. John combined butter with sugar.
b. John drank beer with milk.
c. John is friends with Bill.
d. a picture of John with his fans
The D feature of with in symmetrical constructions also accounts for the
fact that unlike the conjunction and, which does not possess intrinsic
categorial features (Zhang 2006), with never projects any non-D complex: the
complex headed by with must be a DP. By contrast, in (47a), the conjunction
and simply conjoins the two VPs, cut the bread and melted the sugar. Similarly,
in (47b), the conjunction and conjoins the two nominal predicates, a good
The syntax of English comitative constructions 153
father and an excellent husband (remember that nominal predicates are
generally assumed to be NPs rather than DPs).
(47) a. John cut the butter {and/*with} melted the sugar.
b. My uncle is [a good father {and/*with} an excellent husband].
Moreover, like prepositions but unlike other D elements, the with in
symmetrical comitatives has a [Case assigning] feature. The feature is
checked with the Case feature of its complement, DP
2
. This can be seen from
the accusative Case of any pronominal DP
2
(e.g. with {him/*he}). After the
checking, with does not possess a Case feature any more, and thus the
projected DP does not possess a Case feature either, unlike regular DPs. This
accounts for the fact reported in §3.2, namely, that in symmetrical comitative
constructions, the cluster [DP
1
with DP
2
] cannot surface as preverbal subject
(see (29) above). Preverbal subject positions in English are generally assumed
to be Spec of TP, and have a Case relation with T. Now since the cluster [DP
1
with DP
2
] does not have any Case feature, it cannot surface at Spec of TP.
Instead, DP
1
alone is raised to Spec of TP to check the Case feature of T. This
is further supported by the fact that all the unacceptable examples in (29)
become acceptable if only DP
1
occurs in the preverbal subject position,
leaving the rest behind (see §5.1 for more discussion of the raising of DP
1
):
(48) a. John is friends with Bill. (cf. (29a)) [sym: selected]
b. John baked a cake with Bill. (cf. (29c)) [sym]
The symmetrical cluster [DP
1
with DP
2
] cannot occur in the Accusative
Case-checking position of an Exceptional Case Marking construction, either.
Kayne (1994: 64) observes that examples like (49a) are not the only ones
unacceptable, but also those like (49b) and (49c):
(49) a. *John with Bill will collide.
b. *I consider John with Bill to have collided.
c. *John is considered with Bill to have collided.
The comitative constructions in (49) are all symmetrical, since the predicate
collide is collective. In (49b) and (49c), the embedded clause is a non-finite
version of (49a). All three examples show that if the representation of a
simple clause does not allow any movement chain of the whole complex John
with Bill, the representations of any complex constructions that contain the
154 Niina Ning Zhang
chain will also be rejected. I assume that the syntactic representations of (49)
are as in (50). In each structure in (50), the whole complex [DP
1
with DP
2
] is
moved to a Case position. In (50a), the complex is supposed to check the
nominative Case of T and in (50b), the complex is supposed to check the
accusative Case of the matrix v; in (50c), additionally, DP
1
alone moves
further from a structure similar to (50b).
(50) a. *[
DP
John with Bill]
i
will collide t
i
.
b. *I consider [
DP
John with Bill]
i
to have collided t
i
.
c. *John
j
is considered [
DP
t
j
with Bill]
i
to have collided t
i
.
In each example in (50), the representation of the initial movement of the
constituent indexed with i is unacceptable. Kayne (1994) claims that the
failure to raise of the whole complex [DP
1
with DP
2
] in data like (49) is
related to Case. He states that DP
1
cannot be Case-licensed in the [DP
1
with
DP
2
] complex simply by virtue of the whole complex being in a licensed
position. Therefore, DP
1
in the complex must be raised alone to get its Case
licensed.
I adopt the basic idea of Kayne’s Case analysis. But I emphasize that in
symmetrical comitative constructions, the base-generated cluster [DP
1
with
DP
2
] does not have any Case feature, just like a PP, and that is why it cannot
occur in a Case position. Thus, whenever the symmetrical cluster [DP
1
with
DP
2
] is merged as an argument, the Case-checking task of the argument is
achieved by DP
1
alone. I claim that the difference between the cluster and a
regular complex nominal is due to the mixed properties of with in the
construction: it behaves like both a D (it projects a nominal) and a P (it
assigns Case).
The discussion so far has dealt with agent comitatives of transitive verbs
and with comitatives of intransitive verbs. I now extend the analysis to
patient comitatives of transitive verbs. Following Johnson (1991) and Bowers
(2002), among others, I assume that Accusative Case checking of v also
requires raising of the direct object. If the patient argument of a transitive
verb is a symmetrical cluster [DP
1
with DP
2
], only DP
1
is raised, as happened
in the case of agent comitatives. This can been in the quantifier floating in
(51), as pointed out by an anonymous reviewer:
(51) John saw the boys [all t with Sam].
The syntax of English comitative constructions 155
Moreover, in both subject and object symmetrical comitative constructions,
after raising of DP
1
, DP
1
and [with DP
2
] do not form a constituent any more.
Therefore, there is no A-bar movement of the symmetrical cluster [DP
1
with
DP
2
], thus accounting for the fact presented in §3.3.
The proposed [Plural] feature of with accounts for the fourth contrastive
property of symmetrical and asymmetrical constructions presented in §3.4.
Like many D elements, the with in symmetrical comitatives has number
features, and the value of the feature is plural. This plural feature satisfies the
plural selection of collective verbs or predicates, regardless of the number
feature value of either DP
1
or DP
2
. In (52), for instance, both an orange and
an apple are singular, and neither of these alone can satisfy the plural
selection of compared. If with as a head has a plural feature and the feature is
projected to the whole complex an orange with an apple, the selection of the
verb is satisfied.
4
(52) John compared an orange with an apple.
In addition to the four properties examined in §3, the complementation
structure proposed for symmetrical comitatives also covers the fact that the
cluster [with DP
2
] may occur in an A-bar position, as shown in (53) and (54).
(53) a. With whom did you discuss linguistics? [sym: selected sbj]
b. With whom did John eat the dish? [sym: sbj]
(54) a. With which apple did Mary compare the orange? [sym: selected
obj]
b. With which kind of liquid did Mary mix water? [sym: selected obj]
c. With which kind of beer did Mary drink milk? [sym: obj]
I assume that what undergoes the A-bar movement in data like (53) and (54)
is the whole complex DP that contains the trace of DP
1
. It is a kind of
remnant movement (see §2.1.2 above and Baltin 2002, 2006: 734).
4
Conjunctive coordinate complexes are also plural, their plurality reading coming
either from the plurality feature of and itself (Zhang 2006: §4.3), or from the Set
Product denotation of and (Heycock & Zamparelli 2005). Note in this connection
that since both conjunctions and disjunctions (such as or) introduce coordinate
nominals, the fact that only conjunctive coordinate nominals have a plural feature
indicates that coordinate constructions themselves do not carry a plural feature;
instead, it is the specific lexical item and (but not or) that does so.
156 Niina Ning Zhang
4.2. The adjunction structure of asymmetrical constructions
4.2.1. The configuration
In §2.2 and §3.2, I reached the conclusion that the two DPs of a comitative
construction are base-generated inside a complex DP, and the cluster [with
DP
2
] is not a VP adjunct. The complex DP structure which I proposed in
(42b) above for asymmetrical comitatives reflects this non-adverbial status of
[with DP
2
]; it is repeated here for convenience as (55a).
Moreover, we have also seen in §3.1 that in the case of asymmetrical
comitatives the cluster [with DP
2
] is always optional. This optionality is
captured, too, by (55a), where DP
1
is the projecting element of the whole
complex nominal and the cluster [with DP
2
] is its adjunct.
Furthermore, as discussed in §3, the adverb together can left-adjoin to this
PP, and similarly, the phrase by X’s side can right-adjoin to the PP.
(55) a. DP b. DP c. DP
DP
1
PP DP
1
PP DP
1
PP
P DP
2
together PP PP by X’s side
with with DP
2
with DP
2
4.2.2. The formal features
In this section, I account for the remaining three properties of asymmetrical
comitative constructions presented in §3.2 through §3.4.
The proposed nominal-internal adjunct is headed by the preposition with.
As assumed generally, the categorial features of the preposition with are [-N,
-V], and the preposition has a Case checking relation with its complement,
DP
2
.
The syntax of English comitative constructions 157
(56) DP [+Case]
DP
1
PP
P DP
2
with
[-N, -V]
[Case assigning]
In this adjunction structure, the D category of DP
1
is the category of the
whole complex. The category of the complex must satisfy the c-selection of
any selecting predicate (e.g. drink beer in (57a)), verb (e.g. drank in (57b)), or
preposition (e.g. by in (57c)), and the PP [with DP
2
], being a
nominal-internal adjunct, does not affect the category of the complex as a
whole.
(57) a. [John with Bill by his side] will drink beer. [sbj]
b. Mary drank [beer together with milk]. [obj]
c. It was done by [John together with Harry]. [obj of by]
The proposed [Case assigning] feature in (56) accounts for the ability,
discussed in §3.2, of [DP
1
with DP
2
] to occur in preverbal subject position.
Since the preposition with has a Case checking relation with its complement,
DP
2
, the projected PP does not have any Case feature, just like a regular PP.
When the complex [DP
1
with DP
2
] is base-generated in an argument position,
it is always DP
1
that fulfills a Case checking task. Since DP
1
is the projecting
element and the PP [with DP
2
] is its adjunct, the Case features of the whole
complex [DP
1
with DP
2
] are those of DP
1
. Thus, the complex has Case
features and is able to occur at Spec of TP, checking the Case feature of T (see
(30) above). Moreover, since the complex [DP
1
with DP
2
] is a regular
nominal, it can undergo A-bar movement. This accounts for the fact
presented in §3.3. Furthermore, since [with DP
2
] is an adjunct of DP
1
, it does
not affect the ϕ-features of DP
1
. Then the number features of DP
1
are those of
the whole DP complex. If DP
1
is singular, it triggers singular agreement by
default. This accounts for the fact in §3.4.
In addition to the four properties examined in §3, the nominal-internal
adjunction structure proposed for asymmetrical comitatives also explains the
158 Niina Ning Zhang
fact that the cluster [with DP
2
] may undergo A-bar movement, as seen in
(58):
(58) a. (Together) with his mother, John went to a Chinese restaurant.
[sbj]
b. With Bill by his side, John will drink beer. [sbj]
c. Together with which kind of beer did Mary drink milk? [obj]
d. Together with wine sauce, John ate the fish. [obj]
The fronted cluster can be either a complex nominal containing the trace of
DP
1
, or the adjunct PP [with DP
2
] alone. Either analysis could capture the
word order in (58).
Summing up, I have argued in this section that both the complementation
structure in (42a) and the adjunction structure in (42b) (= (55a)) are found
in comitative constructions in English, before any movement affects the
integrated elements. The complementation structure is that of symmetrical
comitative constructions and the adjunction structure corresponds to
asymmetrical comitative constructions. My explanations of the four contrasts
existing between the two types of constructions are summarized in (59).
(59) The contrasts between symmetrical and asymmetrical comitative
constructions
[with DP
2
] is
always
optional
[DP
1
with DP
2
]
occurs in
preverbal subject
position
[DP
1
with DP
2
]
occurs in A-bar
positions
plural feature,
even when DP
1
is
singular
symmetrical -, because it is
not an
adjunct
-, because it has
no Case
-, because it is not a
constituent after
DP
1
moves
+, because the
[ϕ: PL] of with is
projected to the
whole complex
asymmetrical +, because it
is an adjunct
+, because it has
Case
+, because it can be
a constituent
-, because the
number feature
of DP
1
is
projected
5. Deriving the surface positions of the cluster [with DP
2
]
I have argued so far that in an English comitative construction, DP
1
and DP
2
are base-generated in a nominal complex [DP
1
with DP
2
]. The nominal
The syntax of English comitative constructions 159
complex can function as a subject, direct object, or complement of of, as seen
in (60a), (60b), and (60c), respectively.
(60) a. A mother with her two small children came through the door.
b. John combined butter with sugar.
c. a picture of John with his fans
Following Kayne (1994), I have also claimed that when DP
1
is separated
from the cluster [with DP
2
], as seen in John drank beer with Mary, DP
1
has
undergone leftward movement, as shown in (61a). An alternative analysis
that has not been discussed in the literature is whether the construction
could be derived by rightward movement of the cluster [with DP
2
], as in
(61b).
(61) a. John
i
drank beer [t
i
with Mary] [DP
1
raising]
b. [John t
i
] drank beer [with Mary]
i
[extraposition]
In this section, I will argue against the extraposition analysis in (61b).
5.1. The raising of DP
1
in comitative constructions
Like Kayne (1994), I claim that the separation of the subject DP
1
from the
rest of the complex in the surface representation is derived by raising.
Specifically, the tree structure of (61a) is the following:
(62) IP
John
i
I’
…
drank
beer vP
DP v’
t
i
D’ v VP
D DP V DP
with Mary
160 Niina Ning Zhang
The raising of DP
1
in comitative constructions has a parallel in nominals
involving
FLOATING QUANTIFIERS such as all (e.g. “All the children can do it”
/ “The children can all do it”), as discussed by Sportiche (1988) and others.
The cluster [with DP
2
], like all, demarcates argument positions.
5
Another
analogue to this kind of DP Specifier raising from a complex DP can be
found in possessor raising constructions in languages such as Hebrew (see
Landau 1999: 10, among others).
In my analysis, the cluster [with DP
2
] occurring to the right of the object
in examples such as (61) is the remnant of a base-generated subject complex.
The raising of the object/complement and the selecting verb in (62) is
compatible with the hypothesis proposed in the literature that in English
both objects and their selecting verbs have been raised out of their
theta-domain, vP, to a position below IP (see Johnson 1991, Koizumi 1995,
Runner 1995, Bowers 2002, and Baltin 2002, 2006 for various
implementations of this view). Thus any vP-internal element that is not
raised out of vP, including elements in the base-position of a subject,
accordingly surface to the right of objects.
In addition to (61a), the DP raising can also be illustrated by the
derivations of the do so sentence in (63a) (= (17)), and the gapping sentence
in (63b) (= (22)).
(63) a. John killed a man with Bill and Harry did so with Tom.
b. I went to London with Bill and Harry _ with Tom.
I assumed in §2.2.3 that the base-generated subjects of the two clauses in
(63a) are the two nominal complexes [John with Bill] and [Harry with Tom],
respectively. If John is raised out of the first complex and surfaces at the
SpecIP of the first clausal conjunct, and if Harry is also raised out of the
second complex and surfaces at the SpecIP of the second clausal conjunct,
the surface order of the sentence is derived. In the second clausal conjunct,
the VP is realized by the VP proform did so. Every step is fine and the
resulting sentence is acceptable.
In (63b), with Tom survives after gapping, indicating that it behaves like a
Major Constituent in the sense of Hankamer (1973; see also §2.2.4 above). I
assume that gapping is a deletion operation in the phonological component,
5
I am indebted to the anonymous reviewers for bringing this parallelism to my
attention.
The syntax of English comitative constructions 161
which is implemented after syntactic operations (for recent advocates of this
deletion approach, see Schwarz 1999, Lin 2002; also Johnson 2002: 114,
fn.14). In the case under discussion it is implemented after the raising of
Harry from the complex DP [
DP
Harry with Tom] to SpecIP. The complex DP
that is at SpecvP now contains with Tom only. The remnant DP can be
regarded as a major constituent. This accounts for the acceptability of (63b).
If the whole complex nominal is base-generated as an object, DP
1
can also
be raised alone, leaving the with-PP at the base-position. In (51) above we
considered evidence from floating quantifiers supporting this raising. In
(64a), with Sandy is associated with the object Lee. In this sentence, the
complex DP Lee with Sandy is base-generated at Spec VP, and to London is
base-generated as the complement of V. After the raising of Lee out of the
complex object, and after the raising of the verb sent and the complement to
London out of the VP, with Sandy occurs in sentence-final position. The
derivation is shown in (64b).
(64) a. Kim sent Lee to London with Sandy.
b. IP
Kim I’
sent …
Lee
to London vP
DP v’
v VP
DP V’
t
i
D’ V PP
D DP
with Sandy
162 Niina Ning Zhang
5.2. Against an extraposition analysis of the surface position of the
cluster [with DP
2
]
I have claimed that the separation of DP
1
from the cluster [with DP
2
] of a
comitative construction is the result of the leftward movement of DP
1
.
Instead of this, could the separation be accounted for in terms of the
rightward extraposition of [with DP
2
], as in (65a), a type of operation similar
to the one undergone by about the Soviet Union in (65b)?
(65) a. [John t
i
] drank beer [with Mary]
i
b. [
DP
Many books __
i
] will appear soon [
PP
about the Soviet Union]
i
In this section, I present four arguments against the extraposition analysis in
(65a).
First, if an extraposed element right-adjoins to another element, it should
be an island for extraction, as shown in (66b) (Ross 1967).
(66) a. A man entered the room who was wearing a fez.
b. *What did a man enter the room who was wearing _?
However, in (67a), who is extracted from the sentence-final [with DP
2
],
and the sentence is fine. The acceptability of the extraction indicates that the
sentence-final with-cluster is not an extraposed element.
(67) a. Who did a mother come through the door with _?
b. *Who(m) did a mother with _ come through the door?
The unacceptability of (67b) can be accounted for by the constraint that
nothing can be extracted from a raised subject (Takahashi 1994, Stepanov
2001). In this example, the subject a mother with whom is raised from its
theta position to SpecIP, and thus whom cannot be extracted.
Second, extraposed constituents are typically syntactically and
phonologically “heavy,” whereas the [with DP
2
] in comitative constructions
does not have to be heavy. In fact, the situation is just the opposite. In
symmetrical comitative constructions, the cluster [with DP
2
] is usually short,
yet it has to occur to the right of the clause, as in (68); by contrast, in
asymmetrical constructions the cluster is comparatively longer, yet it need
not occur at the right position, as shown in (69) (see §3.2).
The syntax of English comitative constructions 163
(68) a. *John with Bill will drink beer. [light PP]
b. John will drink beer with Bill.
(69) a. John with his two children will drink beer. [heavy PP]
b. John will drink beer with his two children.
My third argument against an extraposition analysis of the surface
position of the cluster [with DP
2
] has to do with information structure.
Assuming with Rochemont & Culicover (1990), Huck & Na (1990) and
others that extraposed elements are semantically foregrounded and thus
focused, we may now consider what is the information status of the [with
DP
2
] in comitative constructions.
The fact that the two DPs in a comitative construction are not parallel in
interpretation has been discussed in Dougherty (1970: 858) and Stassen
(2000: 6). Compare (70a), where a coordinate nominal functions as subject,
with the comitative construction in (70b):
(70) a. Mary and John are in love.
b. Mary is in love with John.
The meanings of the two sentences are not identical. The reading of (70a) is
that Mary and John love each other, whereas in (70b) the possibility exists
that John does not love Mary, even though the construction is in fact a
symmetrical comitative. Yet (70b) is symmetrical only in the sense that the
two DPs, Mary and John, are necessary for satisfying the plural selection of
the relational predicate in love, and John is not an appurtenance of Mary in
the eventuality (see §3). As regards asymmetrical comitatives, the absence of
an exact parallelism between the two DPs is of course a defining property of
that type of construction. For instance, the reading of (71) (Svenonius
forthcoming: example (41c)) is that it is we, rather than the audience, that
sprayed the dog.
(71) We sprayed the dog with an audience of boy scouts.
This difference in interpretation between the two DPs in both symmetrical
and asymmetrical comitatives can be captured by the generalization that in
the relevant eventuality, the individual expressed by DP
1
is foregrounded,
while the one expressed by DP
2
, which is introduced by a preposition (or a
preposition-like element), is backgrounded (Svenonius forthcoming).
164 Niina Ning Zhang
The fact that the two DPs of a comitative construction are not identical in
terms of information structure also accounts for the unavailability of
reciprocals in comitatives. I have argued that like the conjuncts of a nominal
coordinate complex, the two DPs of a comitative construction are
base-generated within a complex DP. Baker (1992: 46) observes that
reciprocals can be licensed by conjoined nominals, as in (72a), but not by
comitatives, as in (72b):
(72) a. John and Mary kissed each other.
b. *John kissed each other with Mary.
I claim that reciprocal eventualities are disallowed in comitative
constructions simply because the two DPs in a comitative construction have
different statuses with respect to information structure.
In some cases, however, the contrast between the two partners in a
comitative construction with respect to the degree of affectedness is not so
striking. Consider the following:
(73) a. They eat meat with dairy products.
b. A person with his daughter came in.
Obviously, the two encoded entities, meat and dairy products, are both
affected in the ingestion event expressed by (73a). Similarly, both a person
and his daughter must take part in the coming event expressed by (73b). This
means that the difference discussed in this section between the interpretation
of [DP
1
and DP
2
] and [DP
1
with DP
2
] can be captured by Seiler’s (1974)
generalization that comitative constructions leave unspecified the extent of
the participation of the backgrounded element (i.e. the referent of DP
2
) in
the action in question. In other words, the backgrounded element
participates in the action to varying degrees, from mere “accompanying” to
full-fledged “partnership”. I conclude from the above that the backgrounded
reading of comitative with-clusters is not compatible with the foregrounded
status of extraposed elements.
Finally, a fourth argument against an extraposition analysis is that
extraposed elements occur at the right edge of sentences:
(74) a. A man came into the bar who we knew in school.
b. *A man seemed who know the truth to be late.
The syntax of English comitative constructions 165
However, the [with DP
2
] in comitative constructions does not need to be
at the right edge. For instance, in (75a) with Harry precedes the
subject-oriented depictive naked, and in (75c) with Bill precedes about
politics.
(75) a. John
i
ate the fish with Harry naked
i
.
b. John
i
ate the fish naked
i
with Harry.
c. John chatted with Bill about politics.
6. Summary
I have argued that in comitative constructions, the two DPs are both
contained in a complex nominal in their base-positions, and thus the cluster
[with DP
2
] is structurally different from a VP adverbial.
I have also argued that two different types of comitative constructions can
be distinguished: symmetrical and asymmetrical. The proposed base-
structures for each of these are (76a) and (76b), respectively:
(76) a. DP [-Case] b. DP [+Case]
DP
1
D’ DP
1
PP
D DP
2
P DP
2
with with
[D] [-N, -V]
[ϕ:
PL] [Case assigning]
[Case assigning]
It can be seen that with has different formal features in the two structures
in (76). In (76b), with is a regular preposition. In (76a), however, with shares
features with both nominals and prepositions. Its [D] and [ϕ:
PL] features are
shared with nominals, and its [Case assigning] feature is shared with verbs or
prepositions. Since syntactic elements are bundles of formal features
(Chomsky 1994: 7), the special kind of with postulated in (76a) is
conceptually possible.
I have presented four differences between symmetrical and asymmetrical
comitative constructions and have put forward relevant explanations for
166 Niina Ning Zhang
them. First, the cluster [with DP
2
] is always optional in asymmetrical
comitative constructions, but not in symmetrical constructions. This contrast
can be explained in view of the structural differences between the two
constructions: the cluster is base-generated as an adjunct of DP
1
in
asymmetrical constructions, but not in symmetrical ones. In the latter, DP
1
is
the Specifier of with and DP
2
is its complement. Second, the cluster [DP
1
with
DP
2
] may occur in preverbal subject position in asymmetrical constructions,
but not in symmetrical constructions. This is because in asymmetrical
constructions the syntactic properties of the cluster are identical to the
properties of DP
1
; hence the cluster may occur in a Case position, just like a
regular DP. However, in symmetrical constructions, the syntactic properties
of the cluster are projected from the word with. Since the Case feature of with
has been checked with its complement, DP
2
, the complex projected by with,
like a regular PP, has no Case feature and thus is not able to occur in any Case
position. Thus whenever the cluster [DP
1
with DP
2
] is base-generated at an
argument position, it is always DP
1
that is raised to a Case position. Third, the
cluster [DP
1
with DP
2
] may occur in A-bar positions in asymmetrical
constructions, but not in symmetrical constructions. My account for the
contrast is that in symmetrical constructions DP
1
is raised alone, so that after
movement it is no longer adjacent to [with DP
2
]; the cluster [DP
1
with DP
2
],
therefore, can never surface as a constituent. By contrast, in asymmetrical
constructions the whole cluster may undergo either A or A-bar movement,
and thus surface as a syntactic constituent. Fourth, a plural feature is always
present in symmetrical constructions, even when DP
1
is singular, whereas
asymmetrical constructions lack such plural feature. My explanation for the
contrast is that in symmetrical constructions, the word with, like plural D
elements, has a plural feature, and thus the projected complex nominal is
always plural; however, in asymmetrical constructions, the projecting
element is DP
1
, and if DP
1
is singular, then the whole projected complex
nominal is singular. Finally, in addition to examining the various differences
between symmetrical and asymmetrical comitatives, I have also argued (cf.
§5) against an extraposition analysis of discontinuous surface orders in
comitative constructions.
References
Baker, Mark. 1992. “Unmatched chains and the representation of plural pronouns”.
Natural Language Semantics 1: 33–73.
The syntax of English comitative constructions 167
Baltin, Mark. 2002. “Movement to the higher V is remnant movement”. Linguistic
Inquiry 33: 653–659.
Baltin, Mark. 2006. “The nonunity of VP-preposing”. Language 82: 734–766.
Bowers, John. 2002. “Transitivity”. Linguistic Inquiry 33: 183–224.
Buckingham, Hugh W., Jr. 1973. “The comitative and case grammar”. Foundations of
Language 10: 111–121.
Camacho, José. 2000. “Structural restrictions on comitative coordination”. Linguistic
Inquiry 31: 366–375.
Chomsky, Noam.1994. “Bare phrase structure”. MIT Occasional Papers in Linguistics
5.
Chomsky, Noam. 2006. “Approaching UG from below”. Unpublished ms. [Available
at http://www.cbs.polyu.edu.hk/tang/papers/Chomsky%202006.pdf]
den Besten, Hans & Gert Webelhuth. 1990. “Stranding”. In: Gunther Grewendorf &
Wolfgang Sternefeld, eds. Scrambling and barriers. Amsterdam & Philadelphia:
John Benjamins, 77–92.
Dougherty, Ray C. 1970. “A grammar of coordinate conjoined structures, Part I”.
Language 46: 850–898.
Fillmore, Charles J. 1968. “The case for case”. In: Emmon Bach & Robert T. Harms,
eds. Universals in linguistic theory. New York: Holt, Rinehart & Winston, 1–88.
Grosu, Alexander. 1976. “A note on Subject Raising to Object and Right Node
Raising”. Linguistic Inquiry 6: 642–645.
Hallman, Peter. 2004. “Constituency and agency in VP”. In: Vineeta Chand, Ann
Kelleher, Angelo J. Rodríguez & Benjamin Schmeiser, eds. Proceedings of the 23rd
West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Press,
304–317.
Hankamer, Jorge. 1973. “Unacceptable ambiguity”. Linguistic Inquiry 4: 17–68.
Heycock, Caroline & Roberto Zamparelli. 2005. “Friends and colleagues: Plurality,
coordination, and the structure of DP”. Natural Language Semantics 13: 201–270.
Huck, Geoffrey J. & Younghee Na. 1990. “Extraposition and focus”. Language 66:
51–77.
Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the
English language. Cambridge: Cambridge University Press.
Johannessen, Janne Bondi. 1998. Coordination. Oxford: Oxford University Press.
Johnson, Kyle. 1991. “Object positions”. Natural Language and Linguistic Theory 9:
577–636.
Johnson, Kyle. 2002. “Restoring exotic coordinations to normalcy”. Linguistic Inquiry
33: 97–156.
Kayne, Richard S. 1994. The antisymmetry of syntax. (Linguistic Inquiry Monograph
25). Cambridge, MA: MIT Press.
Koizumi, Masatoshi. 1995. Phrase structure in minimalist syntax. Doctoral
dissertation, MIT.
168 Niina Ning Zhang
Lakoff, George & Stanley Peters. 1966. “Phrasal conjunction and symmetric
predicates”. Mathematical Linguistics and Automatic Translation, Report no.
NSF–17, pp. VI/1–49. Cambridge, MA: Harvard University. [Reprinted in 1969
in: David A. Reibel & Sanford A. Schane, eds. Modern studies in English.
Englewood Cliffs, N.J.: Prentice-Hall, 113–142]
Lakoff, George & John Robert Ross. 1966. “A criterion for verb phrase constituency”.
Mathematical Linguistics and Automatic Translation, Report NSF–17, pp. II/1–11.
Cambridge, MA: Harvard University.
Landau, Idan. 1999. “Possessor raising and the structure of VP”. Lingua 107: 1–37.
Lehmann, Christian & Yong-Min Shin. 2005. “The functional domain of concomitance.
A typological study of instrumental and comitative relations”. In: Christian
Lehmann, ed. Typological studies in participation. Berlin: Akademie Verlag, 9–104.
Lin, Vivian. 2002. Coordination and sharing at the interfaces. Doctoral dissertation,
MIT.
Lindstromberg, Seth. 1998. English prepositions explained. Amsterdam &
Philadelphia: John Benjamins.
Rochemont, Michael S. & Peter W. Culicover. 1990. English focus constructions and
the theory of grammar. Cambridge: Cambridge University Press.
Ross, John Robert. 1967. Constraints on variables in syntax. Doctoral dissertation,
MIT.
Ross, John Robert. 1970. “Gapping and the order of constituents”. In: Manfred
Bierwisch & Karl Erich Heidolph, eds. Progress in linguistics. The Hague: Mouton,
249–259.
Runner, Jeffrey T. 1995. Noun phrase licensing and interpretation. Doctoral
dissertation, University of Massachusetts at Amherst.
Schwarz, Bernhard. 1999. “On the syntax of either ... or”. Natural Language and
Linguistic Theory 17: 339–370.
Seiler, Hansjakob. 1974. “The principle of concomitance: Instrumental, comitative,
and collective (with special reference to German)”. Foundations of Language 12:
215–247.
Sledd, James. 1959. A short introduction to English grammar. Chicago: Scott,
Foresman & Company.
Sportiche, Dominique. 1988. “A theory of floating quantifiers and its corollaries for
constituent structure”. Linquistic Inquiry 19: 425–49.
Stassen, Leon. 2000. “And-languages and with-languages”. Linguistic Typology 4:
1–54.
Stepanov, Arthur. 2001. “The End of CED?” In: Karine Megerdoomian & Leora
Anne Bar-el, eds. Proceedings of the 20th West Coast Conference on Formal
Linguistics. Somerville, MA: Cascadilla Press, 524–537.
Stolz, Thomas. 2001. “To be with X is to have X: Comitatives, instrumentals, locative,
and predicative possession”. Linguistics 39: 321–350.
The syntax of English comitative constructions 169
Stolz, Thomas, Cornelia Stroh & Aina Urdze. 2006. On comitatives and related
categories: A typological study with special focus on the languages of Europe. Berlin
& New York: Mouton de Gruyter.
Svenonius, Peter. Forthcoming. “Adpositions, particles, and the arguments they
introduce”. In: Tanmoy Bhattacharya, Eric Reuland & Giorgos Spathas, eds.
Argument structure. Amsterdam & Philadelphia: John Benjamins. [Ms available at
http://ling.auf.net/lingBuzz/000042]
Takahashi, Daiko. 1994. Minimality of movement. Doctoral dissertation, University of
Connecticut.
van Oirsouw, Robert R. 1987. The syntax of coordination. London: Croom Helm.
Vassilieva, Masha & Richard K. Larson. 2005. “The semantics of the plural pronoun
construction”. Natural Language Semantics 13: 101–124.
Walmsley, John B. 1971. “The English comitative case and the concept of deep
structure”. Foundations of Language 7: 493–507.
Zhang, Niina Ning. 2006. “On the categorial issue of coordination”. [Available at
http://www.ccunix.ccu.edu.tw/~lngnz/index.files/Cat%20LinBuzz.pdf]
Author’s address
Institute of Linguistics
National Chung Cheng University
Min-Hsiung, Chia-Yi 621, Taiwan
e-mail:
Lngnz@ccu.edu.tw
received: 29 August 2006
revision invited: 28 November 2006
revised version accepted: 5 March 2007
folia
linguistica
30 2 2006
folia
linguistica
