Professional Documents
Culture Documents
1 The Verb-Particle Combination in Present-Day English
1 The Verb-Particle Combination in Present-Day English
English
The English language has a rich array of verbs that consist of more than one word.
Commonly known as phrasal verbs, they are composed of a verb and a particle and this
technique is creatively employed by speakers to express one (complex) meaning. As the set
of examples in (1) illustrates, they come in many different types and forms.
(1) a. He cannot seriously expect his hamster to pick up the newspaper for him.
b. The four unmanageable kids chucked their parents right out.
c. Her boyfriend does nothing but veg out and sleep these days.
d. Their efforts to cheer up the singer who lost her voice were useless.
e. The artistic impulses of the zoo-keepers really cheesed the purple monkeys
off.
f. The hairdresser tried to calm down his mistakenly permed customer.
g. The frustrated athlete put the weights down on the instructors foot.
h. The escape to the country failed miserably because of a badly thought-out
plan.
i. The slug in her lettuce was a bit of a turn-off.
j. Clubs have the right to refuse entrance to girls who are not tarted-up.
k. These determined lookers-on are not likely to leave before dawn.
l. His goal in life is to become a successful sucker-upper.
m. The switching off of the lights continues to fascinate the night watchman.
n. The burglar could not believe his luck when he discovered the bank was
extremely breakinable.
English particles constitute a closed class, which centrally includes up, out, down, off, in, as
well as the less frequent back, on, away, over, around, about, along. Many particles have the same
phonological form as prepositions, but they are distinctly a separate class on the basis of
their syntactic distribution, which differs considerably from that of prepositions. The
examples in (2) illustrate this.
The constituency test in (2a), fronting of the particle together with the NP object, yields an
ungrammatical sentence and thus shows that the particle and the NP object do not form a
constituent. In (2a), on the other hand, fronting of down together with the NP the mountain is
possible, indicating that they form a constituent. Thus, while we are dealing with a
preposition in (2a), (2a) contains not a preposition but a particle. A second distributional
difference between particles and prepositions reveals itself when the order of the elements
is altered. In (2b), the particle and the NP object have been inverted, which gives a
grammatical sentence. Inversion produces an ungrammatical sentence when a preposition is
involved, as (2b) shows. Particles and prepositions also pattern differently with respect to
gapping, witness the examples in (2c, c) and (2d, d). In the coordinated sentence in (2c),
the verb is omitted, and this results in an ungrammatical sentence. This indicates that the
verb and the particle form a close unit, which does not allow gapping of one of its parts.
The bond between the verb and the particle is also apparent from the irreplaceability of the
particle, e.g. Olivia bottled up/*down/*in/*off her anger and frustration (cf. Haegeman and
Guron 1999: 254). By contrast, omission of the verb in the coordinated sentence in (2c) is
allowed. This follows from the looser tie between verbs and prepositions compared to that
between verbs and particles. Rather than with the verb, prepositions form a constituent
with their object. The observation that particles form a unit with verbs predicts the
possibility of gapping the verb together with the particle. This prediction proves true, as the
example in (2d) shows. Similarly, we expect such gapping to be impossible with
prepositions, since they form a constituent with an NP.
The distributional differences can be linked directly to the distinct argument structures of
particles and prepositions. Particles, unlike prepositions, do not have a complement.
Svenonius (2002a) captures this observation by demonstrating a robust semantic pattern
concerning the type of constituents English particles and prepositions combine with. The
complement of prepositions are typically Grounds (Talmy 1978), i.e. entities expressing a
THE VERB-PARTICLE COMBINATION IN PDE 11
location with respect to which the Figure is located (Svenonius 2002a: 433). A Figure
(Talmy 1978) expresses the entity in motion or at rest which is located with respect to the
Ground (Svenonius 2002a: 432). The example in (3) is taken from Svenonius (2002a: 433),
his example (10b).
(3) The cook twisted the lid [PP off the jar]
Figure Ground
The example shows that prepositions relate an entity (in (3) the Figure) to a certain location
(the Ground). Particles denote the endstate or location of a Figure and do not combine
with a Ground, (4).
(5) a. PP b. PP c.
ei ei
Figure P Figure P PP
the lid ei the lid g ei
P Ground P P Ground
off the jar off * off the jar
The structure for prepositions in (5a) represents what Svenonius calls transitive
prepositions. Prepositions can also be unaccusative, in which case they combine with a
Ground, (5c). This is not possible with example (3), assuming that the jar is not twisted off
of something else (Svenonius 2002b: 6), but can be illustrated with the example The helicopter
flew up the mountain (Svenonius 2002b: 6). Although there is no Figure in unaccusative PPs, it
seems that a Figure is not entirely absent. The matrix subject, in this case the helicopter, is the
Figure (cf. the helicopter is up). Particles are treated as intransitive prepositions: they do not
combine with a Ground, therefore have no internal argument, (5b). They do have an
external argument, i.e. the Figure. Under standard X-theory, particles are structurally
defective in the sense that they do not have a complement.1
Although VPCs normally leave their Ground implicit, there are examples in which the
Ground is expressed. In such cases, the Ground has been promoted to object of the entire
VPC (see Svenonius 1996b, 2002a; Zeller 2001b; McIntyre 2003). Examples are given in (6)
(from McIntyre 2003: 128130).
1 Svenonius (1996a) proposes that the complement position of English particles is occupied by an
abstract complement , which incorporates into the particle. See Chapter 2 for discussion.
12 CHAPTER 1
(6) a. wipe the dust off the table (i) wipe the dust off
Figure Ground (ii) wipe the table off
b. pour water out (of) the bucket (i) pour water out
(ii) pour the bucket out
c. squeeze the juice out (of) the orange (i) squeeze the juice out
(ii) squeeze the orange out
In the (i) examples, the particle predicates over the Figure, expressing the Path of the
activity, while in the (ii) examples, the subject of the particle is a Ground, denoting the Place
(see also Svenonius 2002a: 433). According to Svenonius (1996b: 31), examples like the
ones in (6) often involve a Ground which is a Theme or Patient argument of the transitive
verb of the VPC, e.g. The waiter wiped the table. He also notes that the suppressed Figure often
denotes something that is removed from the Ground. McIntyre (2003: 130) observes that
VPCs with an explicit Ground (reference object in McIntyres terminology) have a strong
holistic effect. Pour the bucket out, for example, has the interpretation that the bucket is
emptied completely. The same effect is noted for similar cases in German by Zeller (2001b).
He observes that an object such as a beer in John had a beer can refer simultaneously to the
Theme (beer) or to the Container holding the beer (a glass, bottle) (Zeller 2001b: 279). This,
according to Zeller, causes examples like the ones in (6) to have a strong completive
interpretation. I suggest that the suppression of the Figure (i.e. the theme argument) is
caused by semantic (and maybe pragmatic) factors. The completion of the event can be
given more emphasis by expressing the Ground.
(7) a. He cannot seriously expect his hamster to pick up the newspaper for him.
b. The artistic impulses of the zoo-keepers really cheesed the purple monkeys
off.
Particles can occupy the position immediately following the verb, as in (7a), or may follow
the object which intervenes between the verb and the particle, as in (7b). Crucially, all VPCs
occur in this word order alternation, hence pick the newspaper up and cheesed off the purple
monkeys are the possible alternates for (7a) and (7b) respectively.2 In relation to this, it is
interesting to note the semantic correspondence between the two word orders. The fact
that there is no obvious difference in meaning between the two suggests that the choice
between the two word orders is entirely optional. If this is indeed the case, the question is
2 It is important to note here that there are VPCs which do not show the alternation. These concern
idiomatic cases whose order has become fixed. An example is to sing up a storm sing loud, sing
vigourously.
THE VERB-PARTICLE COMBINATION IN PDE 13
raised how such optionality can be theoretically implemented. In view of their semantic
equivalence, an analysis in which the two structures are somehow related to each other
seems most obvious. Such an analysis needs to determine how the two orders are related
and what determines which of the two orders is chosen. Several linguists have pointed out,
for instance, that the two word orders have different information structures and that the
choice of word order is determined by the kind of information expressed by the particle and
the object (Bolinger 1971, Svenonius 1996a, Deh 2002). The role of pragmatics is also
implied by the so-called definiteness effect. Bolinger (1971: 64), for example, points out that
a definite noun phrase will occur in between a verb and an adverb, You must do the job again
versus *You must do again the job, because of their semantic content. By definition they are not
newsworthy as they refer to a previously mentioned element. He notes that VPCs are not
subject to this restriction and that the word order may alternate depending on the relative
semantic weight of the elements involved.
English particles may also be separated from their verb by an adverbial modifier, (8a), or
by an object and an adverbial modifier, (8b), repeated from (1b).
Strikingly, however, modification of the particle is impossible when the particle precedes the
object, (9).3
(9) *The four unmanageable kids chucked right out their parents.
This demonstrates that nothing may intervene between the verb and the particle when the
object follows the particle and raises questions about the structural status of the verb and
the particle in this order. While it is clear from the separability facts we have seen so far that
the particle is an independent syntactic element, the ungrammaticality of (9) casts doubt on
this view.
Other distributional facts add to the uncertainty about the independent status of English
particles. Consider the example involving coordination in (10).
In (10), the VPC clean up is coordinated with a simplex V, dust. On the assumption that
coordination is only possible with like structures (i.e. head with head, phrase with phrase),
examples such as this one suggest that the particle forms a (complex) V together with the
verb. Note, however, that coordination of two particles is also possible, (11).
3 Kayne (1985: 127) notes that modification of the particle in this position is possible when the object
is heavy, cf. (i) (from Kayne 1985: 127, his example (137)).
(11) Our cat Daisy kept switching the light on and off.
In this example, the coordinated particles occur separated from the verb, which is a first
indication of their independent status. Moreover, the fact that modification of both particles
is possible, i.e. Daisy switched the light right on and right off, also argues in favour of an analysis in
which particles are independent elements (XPs).
The set of elements that can intervene between the verb and the particle is restricted to
the ones we have discussed so far, namely objects (full DPs or pronouns) and adverbial
modifiers such as right, straight (cf. Den Dikken 1995), or a combination of these two. A
familiar quirk in the element order of VPCs is the unavailability of the VPrtNP order
when the object is pronominal, (12).
In contrast with modification, separating of the verb and the particle is obligatory. When
the pronominal object receives stress, on the other hand, it is allowed to follow the verb
and the particle (cf. Bolinger 1971; Fraser 1976; among others). The examples in (13) are
from Fraser (1976: 17), his example (1x58). The example in (14) is from Dickens Our
Mutual Friend (from van Kemenade and Los 2003: 85, their example (15)).
(14) If you force your confidence upon me, Mr. Headstone, Ill give up every
word of it. Mind! Take notice. Ill give it up, and Ill give up YOU. I will!
(Dickens, Our Mutual Friend, 673)
These facts show that an explanation of the distribution of English particles should also
take into account prosodic factors.4
English particles may be fronted to sentence-initial position, (15).
Particle topicalisation is employed to bring the particle into focus (cf. Cappelle 2002: 57;
Deh 2002: 110). Not all particles can be fronted in this way, however. Consider the
examples in (16).
4 Recall that this was also suggested by the expansion of the modification possibilities when a heavy
The factor ruling out (16a) and (16b) is semantic in nature, as has been pointed out by
several linguists (Bolinger 1971; Wurmbrand 2000; Cappelle 2002; Zeller 2002). The particle
up in (16a) has a transparent (i.e. directional) meaning and can be topicalised. In (16a), the
meaning of the particle up is not transparent and fronting is not allowed. The same contrast
is demonstrated by examples (16b) and (16b), where the meaning of the particle down is
transparent in one case, (16b), and non-transparent in the other, (16b).5
In his discussion on particle preposing, Cappelle (2002) argues that particle fronting is
only allowed with particles that are semantically independent (see also Ldeling and de Jong
2002: 316). One test for semantic independence he gives is the possibility to use the
particle with the same meaning outside verb-particle constructions (e.g. after the copula be)
(Cappelle 2002: 57). This ties in with his definition of particles with a transparent (Cappelle:
literal) meaning. He assumes that a particle is literal if its meaning is constant across
different verb-particle constructions, in other words, if its meaning is not dependent on the
particular verb it combines with. Defined this way, a literal particle does not have to express
a direction in physical space (Cappelle 2002: 56). My definition differs from Cappelles in
that mine does refer to their directional meaning: the directional meaning of particles (their
meaning outside the VPCs) is their transparent meaning.6 Cappelle shows that particle
preposing is not restricted to particles with a directional meaning, hence adopts a definition
in terms of semantic independence. Other tests for semantic independence of the particle
mentioned by Cappelle are the existence of a contrastive counterpart (updown, inout, onoff)
and the possibility of reduplicating it (Cappelle 2002: 57).
Cappelle notes that the semantic condition on topicalisation is directly related to the
focus/background structure of sentences. The topic position of a sentence is a prominent
position in terms of information content (Cappelle 2002: 58) and hosts foregrounded
elements. An element can only be foregrounded when it has semantic content of its own.
This explains why transparent particles, but not non-transparent particles, can be fronted.
On the general assumption that only phrases can topicalise, the topicalisation facts
presented here indicate that transparent particles should be analysed as phrases.
Importantly, the fact that non-transparent particles cannot topicalise does not prove that
they are not phrases. The inability to topicalise is related to semantic factors and does not
provide evidence for the structural status of non-transparent particles (cf. Ldeling and de
Jong 2002: 316).
5 It has been observed that preposed particles pattern like PPs in locative inversion constructions, e.g.
Down the hill rolled the baby carriage (Cappelle 2002: 58; Jackendoff 2002: 75; Zeller 2002: 244).
6 My definition is (partly) based on the historical origins of particles. In Old English, particles
predominantly had a transparent meaning. In the Middle English period, non-transparent meanings
developed from these. I therefore consider the directional (transparent) meaning of particles as their
original meaning.
16 CHAPTER 1
English VPCs can be the input to word formation, as the examples in (17), repeated from
(1), show.
(17) a. The escape to the country failed miserably because of a badly thought-out
plan.
b. The slug in her lettuce was a bit of a turn-off.
c. Clubs have the right to refuse entrance to girls who are not tarted-up.
d. These determined lookers-on are not likely to leave before dawn.
e. His goal in life is to become a successful sucker-upper.
f. The switching off of the lights continues to fascinate the night watchman.
g. The burglar could not believe his luck when he discovered the bank was
extremely breakinable.
h. The team showed great bouncebackability.7
Examples (17a), (17c) and (17g) contain adjectives formed from VPCs, examples (17b),
(17d), (17e), (17f) and (17h) all contain nouns formed from VPCs. The fact that VPCs can
take part in word-formation processes adds to the puzzle posed by their structural status.
On the traditional assumption that word-formation targets words, the facts presented in
(17) provide evidence for an analysis which treats VPCs as words. Apart from the problems
such an analysis would face in dealing with the syntactic separability of VPCs, the form of
the VPCs deviates from that of other morphological constructs, which questions their
status as morphological words.
The possibility of word formation not necessarily points to word status, given that word-
formation can target phrases too (cf. Booij 2002). This is illustrated by the examples in (18);
the Dutch examples in (18b) are taken from Booij (2002: 27), his example (11).
In principle, then, the fact that VPCs can be converted into nouns and adjectives does not
necessarily provide evidence for the claim that they are words.
7 Bouncebackability has recently been added to the online OED (draft entry June 2006). The entry states
that it is chiefly found in sports commentaries and defines it as the capacity to recover quickly or
fully from a setback, bad situation, etc.. The first possible example cited in the OED is from 1961,
(i).
(i) The Tribe demonstrated its bounce-back ability in a three-game series with Washington,
taking the set 2-1.
(Times Recorder (Zanesville, Ohio) 18 Apr. 2B/1)
THE VERB-PARTICLE COMBINATION IN PDE 17
An analysis which treats VPCs as words is also problematic in view of the order of the
verb and the particle. If VPCs are words, the syntactic ordering of the verb and the particle
(yielding a head-initial construct) is unexpected, given the form of regular English
compounds, which are head-final, (19).
(19) a. No
ei
Ao No
black board blackboard
b. [ [Po under] [No world] ]No underworld
c. [ [Ao high] [Vo light] ]Vo highlight
d. [ [Po over] [Ao ripe] ]Ao over-ripe
VPCs would thus violate the RHR, which indicates that they are not morphological
constructs. Another argument against treating VPCs as words is that the stress pattern of
VPCs differs from that of compounds (Emonds 1993: 243 fn. 27). Compare the examples
in (21a) and (21b), where the stressed syllable is underlined.
In English compounds stress is placed on the left-hand member, (21a). This contrasts with
VPCs, whose right-hand member receives stress, (21b).8
The status of converted VPCs is not clear-cut either. English compounds generally do
not allow internal suffixation (e.g. *hands-bag versus handbags), but VPCs standardly do, (22).
The internal suffixation displayed by English VPCs is problematic for an analysis treating
(converted) VPCs as words. Examples such as the ones in (22) suggest that conversion into
a noun or adjective takes place after suffixation.
The picture is further complicated by examples such as in (23), in which the suffixes
have been attached to the particle rather than to the verb.
8 When VPCs are converted into nouns or adjectives, however, stress placement shifts from the right-
(23) a. takedownable
b. put-outer9
The adjectival suffix able normally appears on the particle to form an adjective. The
position of the agentive suffix er is more variable (cf. looker-on) and the suffix may even
appear more than once on the same converted VPC, (24).
As the examples in (24) show, the agentive suffix er can show up on the verb and on the
particle at the same time. These curious facts have not yet been studied in great detail and in
my view are illustrative of the ambiguous nature of English VPCs, which straddle the
boundary between syntax and morphology.11 Suffixation on the verbal part follows syntactic
principles, whereas suffix placement on the predicate part signals an interpretation of the
VPC as a word. Reduplication of the suffix accommodates the principles of both syntax and
morphology.
Svenonius (2004) provides an explanation of reduplication with VPCs in terms of Halle
and Marantz (1993) Distributed Morphology. For examples like picker-upper he argues that
the suffix er has to bind two thematic roles, namely SOR subject of result (assigned in the
specifier position of the particle) and SOP subject of process (assigned in the specifier
position of the verb). Apparently, this binding takes place twice in English (by reduplication
of the suffix er), rather than once, after the verb and the particle have combined.
Svenonius does not mention the even more exotic examples involving retriplication, which
require an additional explanation.
English VPCs may appear in nominalisations. As the examples in (26) show, they can occur
with or without of.
(26) a. [The washing out of the red wine stains] proved an impossible task.
b. [The shaking down the apples] angered the neighbour.
9 From the Harry Potter books by J.K. Rowling. It is a small magical object that looks like a cigarette
lighter. When clicked, the Put-Outer turns off a street light. When reversed, balls of light fly back out
of the Put-Outer and return to the street lights, The Harry Potter lexicon Magical items and
devices (http://www.hp-lexicon.org/magic/devices_n-r.html).
10 Winner of The Apprentice 2006, in an interview for Pure T4 (Channel 4) on 13 May 2006.
11 McIntyre (2004) mentions such cases in a Linguist List issue (http://www.sfs.nphil.uni-
Adopting Chomskys (1970) terminology, the mixed nominalisation in (26a) differs from the
gerundive nominalisation in (26b) in that it does not allow the particle and the of-phrase to
invert: *The washing of the red wine stains out, versus The shaking the apples down. This reflects the
syntactic situation, where the verb and the particle cannot be separated by a prepositional
phrase either: *My sister washed the red wine stains of the carpet out (cf. Kayne 1985; Johnson
1991). Chomsky argues that gerundive nominalisations are derived from sentences, while
mixed nominalisations are listed in the lexicon. According to Chomsky, the
ungrammaticality of *The washing of the red wine stains out follows from the fact that mixed
nominalisations are lexical, since syntactic operations cannot apply to it. The impossibility
of the shift does not necessarily prove that mixed nominalisations are lexically listed,
however, given that the inversion is also impossible in the syntax. Moreover, there are
idioms which are lexically listed, but which still show sentential syntax in that idiom chunks
can be moved (cf. Fraser 1976; Jackendoff 1997).
Present-Day English has a range of prefixed verbs, nouns and adjectives some of which are
historically related to the VPCs under investigation in this thesis. Some are of later origin
than Old or Middle English. They predominantly consist of up, out, down or over and a verb
or noun. Consider the verbs with out in (27).
The prefix out in the examples in (27) adds a sense of comparison to the meaning of the
verb (Fraser 1976: 29). Thus, to outrun means running faster or longer than someone else
and to outgun means winning a battle because you have more weapons than the other side.
Fraser (1976: 29) points out that the addition of the prefix has a transitivising effect. The
verb to run is intransitive, but to outrun requires a direct object, otherwise there is nothing or
no-one to outrun. At the same time, the prefix out may combine with transitive verbs as
well (e.g. outthrow). The function of the prefix out is similar to that of particles in VPCs,
which also add a meaning to that of the verb and also have the ability to transitivise verbs.
Prefixation with out- seems to be fairly productive, in that out can freely be combined with
new or existing verbs. Historically, some of these prefixed verbs are related to the prefixed
verbs (inseparable complex verbs) that existed next to the VPCs in the Old English period
(roughly 5001100 A.D.), while others are of later origin. In the early Modern English
period, the prefix out- meaning surpassing/exceeding is often attached to verbs (OED
Online).
The prefix out also appears in nouns and adjectives, and in these contexts it can express
several meanings. In nouns like out-take and outcast, out denotes removal from a certain
point or place and they often have a verbal counterpart (the VPCs take out, cast out in this
case). In outfield and outskirts, out adds the meaning not central. In some cases, out appears
to have an ingressive meaning, as in outbreak and outburst. Some of these prefixed nouns and
20 CHAPTER 1
adjectives appear to have been derived from VPCs, e.g. out-take (take out), outcast (cast out),
outbreak (break out), outburst (burst out).
The prefix up usually adds the meaning upwards, as in uphold, uproot and upgrade.
Prefixed verbs with up are not as frequent as the out type prefixed verbs. Apart from
verbs, there are a number of nouns with up as well, e.g. uprise/-ing, uproar, upsurge, upswing,
uptake, upturn. In such examples, the meaning of up is increase. These prefixed nouns are
formed differently from the prefixed verbs. The nouns are nominalised VPCs, cf.
uptake/take up, uprise/rise up. The prefixed verbs, on the other hand, are formed in the old
inseparable particle pattern, which is no longer productive (with the exception of the prefix
out, see above), but which has resulted in several lexicalised cases.
The stress pattern of English prefixed verbs differs from that of English compounds,
and is similar to that of English VPCs, e.g. outrun, upgrade. In prefixed nouns and adjectives,
on the other hand, stress is placed on the prefix and the pattern is similar to that of
compounds. Like English compounds, English prefixed verbs, nouns and adjectives are
head-final and therefore obey the RHR. This points to an analysis in which they are treated
as words. This contrasts with the structural ambiguity of VPCs.
Roeper (1999: 5) observes a difference in argument structure between prefixed nouns
and adjectives and VPC nominals and adjectives. The prefixed variant allows a complement
and is non-agentive, (28a), while VPC nominals and adjectives do not permit a complement
and involve a notion of agentivity, (28b).
The argument structure of prefixed nominals like the ones in (28a) is unaccusative, i.e. non-
agentive. By contrast, VPC nominals, e.g. (28b), have an implied agent. Note that the VPC
counterpart of outbreak, breakout, does allow a complement. Roeper (1999: 9, footnote)
indeed acknowledges that there is a small number of counter-examples to the claim that
VPC nominals do not allow complements. Nevertheless, the complement-pattern is very
robust in general.
The various particles of the English language all contribute their own special meaning to the
verb they combine with. In this sense, English VPCs have a compositional meaning, even
when the meaning of a VPC is completely idiomatic. The chart in (29) lists the most
important meanings of the set of particles introduced at the beginning of this chapter. The
classification is partly based on Kennedy (1967), but includes more particles and sometimes
employs a different division.
THE VERB-PARTICLE COMBINATION IN PDE 21
(29)
Particle Meaning Example
UP 1. upward motion spring up, throw up (a ball)
2. emergence dish up, wake up, show up
3. initiation start up (a car), hurry up
4. cessation slow up
5. completion (perfective) clean up, sober up, warm up
OUT 1. outward direction spread out, stick out
2. removal or separation clean out, pick out
3. disappearance
4. completion (perfective) blow out, fade out, work out, flesh out
DOWN 1. downward direction fall down, tear down
2. transferral hand down, cut down (the costs)
3. cessation quiet down, break down
OFF 1. movement from (away) dash off, send off
2. separation
3. relief or release cut off, tear off, cool off, fire off
IN 1. inward movement bring in, come in
2. inclusion box in, fit in
ON 1. continuation egg on, pass on, carry on
2. connection log on
3. possession take on
4. appearance turn on
AWAY 1. movement from blow away, scare away, run away
2. continuously write away, chat away
3. gradually melt away, waste away
BACK 1. return hand back, put back
OVER 1. across hand over
2. down fall over, knock over
3. again start over
ALONG 1. forward walk along, sing along
AROUND/ 1. in this place push about/around, run around, lie
ABOUT about
As the chart shows, English particles have many meanings beside their core directional
meaning. There are often still traces of the transparent meaning in the non-transparent
meanings and the distinction is sometimes difficult to draw. In some cases, it may be hard
to establish whether it is really the particle that contributes a certain meaning, or whether
that meaning is already present in the verb itself. It is indeed the case that the meaning of
the verb and the particle overlaps in certain cases. For example, in the VPC start up, both
the verb and the particle carry the meaning initiation. It is also possible that two different
particles can contribute the same meaning. Thus, both up and down may express cessation, as
22 CHAPTER 1
in slow up and quiet down. In fact, there is also a VPC slow down, whose meaning is (roughly)
the same as that of slow up. The choice of particle, then, is not always straightforwardly
predictable.
The addition of the particle up not only contributes perfectivity in the sense that the work
on shop-window is finished, it also indicates a change of state. The telicity of particles can
be tested in several ways, thanks to Verkuyls (1972, 1993) work on telic aspect. A well-
known test is the for X time/in X time test, (31).
(31) a. His hamster picked up his newspaper *for two minutes / in two minutes.
b. Her boyfriend vegged out for a month / *in a month.
The adverbial phrase for X time expresses a duration, whereas in X time indicates that the
action or event has an endpoint. Thus, the fact that the sentence in (31a) allows
modification by in X time indicates that it has telic aspect. This telic aspect is expressed by
the particle, since the verb to pick on its own is not telic: *Jenny picked wild berries in one hour;
instead, Jenny picked wild berries for one hour. The adverbial phrase for X time, on the other
hand, signals an a-telic sentence, (31b).
Another test for telicity (Verkuyl 1972, 1993) involves the conjunction of two time
adverbials, (32).
(32) a. His hamster picked up the newspaper in the morning and in the afternoon
(unambiguous)
b. Her boyfriend vegged out in the morning and in the afternoon (ambiguous)
Sentence (32a) has the unambiguous reading that the event of picking up the newspaper
took place at two independent points in time, i.e. in the morning and in the afternoon. This
signals a telic predicate. The sentence in (32b) has two possible readings, the first of which
is that the event took place at two different points in time (with an interval). The alternative
reading holds that the event took place continuously for the period of time denoted by the
two time adverbials. The ambiguity indicates that the sentence is a-telic.
THE VERB-PARTICLE COMBINATION IN PDE 23
The particle up in (33a) signals completion of the action of tearing the lottery ticket. Due to
the absence of the particle in (33b) nothing indicates the completion of the tearing action.
Despite the connection between telicity and resultativity, a-telic sentences may in fact be
resultative, as the example in (34) shows (cf. Goldberg and Jackendoff 2004).
(34) a. The hairdresser tried to calm down his mistakenly permed customer.
b. The hairdresser tried to calm down his mistakenly permed customer for an
hour.
b. *The hairdresser tried to calm down his mistakenly permed customer in an
hour.
As the modification possibilities show, cf. (34b) and (34b), the VPC calm down is a-telic.
Interestingly, the VPC does have a resultative interpretation. The particle down expresses the
result of the action/process denoted by the verb calm. This state of affairs arise as a result of
the fact that the change of state expressed by the particle down is not bounded by an
endpoint. Thus, while a change of state has been established, the endpoint has not yet been
achieved.
While the bulk of English VPCs have a resultative interpretation, there are some notable
exceptions. Consider the uses of the particles away, on, and along in (35).
(35) a. The inattentive audience chatted away throughout the entire play.
b. Despite his disqualification, the driver raced on.
c. The stray parrot tagged along.
In these examples, the particles away, on and along indicate the continuity of the action and
express durative/progressive aspect. Durative/progressive VPCs are typically unergative
and a possible transitive counterpart, such as The bands new hit single helped along their albums
sales, has the effect of yielding a resultative reading. The differences in semantics between
resultative and durative/progressive VPCs suggest that their syntactic structures are not
alike. This will be discussed in Chapter 3.
24 CHAPTER 1
The meaning of the particle sometimes overlaps with that of the verb, as is the case in fall
down. For this reason, it has been claimed that some particles do not add any meaning to
that of the verb and are therefore redundant (Jackendoff 2002: 76). Typical examples are
presented in (36).
The verb to finish inherently implies completion and the particles up and off contribute
precisely this meaning. The same goes for open up and close up, where the verbs to open and to
close by themselves imply complete opening and closure, respectively, which is also
expressed by the particle up. Despite the overlap in meaning, the particles in these VPCs are
by no means redundant, as Jackendoff implies. They serve to emphasise the endstate of an
inherently telic activity, or, as Los (2004: 88) puts it, they draw attention to the
effectiveness of the action.
The classic studies on verbal semantics by Vendler (1957) and Dowty (1979, 1986) have
proposed four classes of event types, including states, activities, accomplishments and
achievements. The temporal properties of these verb classes are presented in the chart in
(37), which is based on Vendler (1957).
(37)
No duration Duration
Achievement Accomplishment
Telic
e.g. arrive, resign e.g. create, tell
State Activity
Atelic
e.g. feel, love e.g. write, run
Accomplishment verbs and activity verbs are distinguished from achievement verbs and
state verbs in that the former express duration. Accomplishment verbs differ from activity
verbs in that accomplishment verbs are bounded, whereas activity verbs are not.
Closer inspection of the semantics of the verbs in English VPCs reveals that they are
mostly activity verbs, (38a), or accomplishment verbs, (38b).
In VPCs containing activity verbs, (38a), the particle adds telic aspect to the entire event
(the verb itself is atelic). Particles that combine with accomplishment verbs, as in (38b), add
telic aspect too, but the difference is that the verb itself is already telic. The particle lends
more expressivity to the boundedness of the event.
The verb in VPCs is predominantly monosyllabic and is often a light verb, such as
come/get/give/go/put/take, which can combine with more than one particle, (39).
(39)
VPC
light verb particle
come up/in/down/out/off
get up/in/down/out/off
give up/in/out
go up/in/down/out/off
put up/down/out/off
take up/in/down/out/off
Van Kemenade and Los (2003: 86) observe that a VPC verb is not restricted to a light verb
but may be any verb that can indicate the means or manner by which the result was
reached, or even a verb that lexicalizes the predicate itself. Verbs of the means-type signal
the instrument used to reach the endstate (van Kemenade and Los 2003: 93) and include
cases such as the ones in (40).
(40) belt up, butter up, chalk up, cork up, dish up/out, lace up, mop up, polish
up/off, elbow out, hammer out, iron out, nose out, flag down, throttle
back/down, clock out, fence off, mouth off, seal off
Manner verbs indicate the manner in which the endstate is reached (van Kemenade and Los
2003: 93). Examples are given in (41).
(41) bundle up, carve up/out, cook up, shake up/off, sharpen up, freeze out,
portion out, rub out, chop down, shoot down, rattle off, wipe off
Sometimes the verb is a conversion of the predicate itself (van Kemenade and Los 2003:
93). In such cases, an adjectival or nominal predicate has been converted into a verb. An
example is cool down, in which the verb cool has been converted from the predicate adjective
cool (cf. the food is cool). Other examples are given in (42) and are taken from van Kemenade
and Los (2003: 93), their example (32).
(42) back off/away, brazen out, cheer up, clean up/out/off/away, clear
up/out/off/away, crack up, free up, gloss over, open up/out, parcel out,
pretty up, round up/off
26 CHAPTER 1
In some cases, the particle has become a verb. This is the case in examples like Mum offed the
television, The butcher upped the prices, The hooligan downed the beer.
Some VPCs contain a verb that does not exist outside the combination with the particle.
Examples are veg out to relax and spend time doing very little, (1c), cheese off to annoy, (1e),
tart up make more attractive by putting on make-up etc., (1j).
English VPCs express a complex event, which consists of an activity or process (denoted by
the verb) and an endstate (expressed by the particle). An example is given in (43).
The object our tree-house is affected by the activity denoted by the verb and reaches the
endstate denoted by the particle. The particle clearly functions as a resultative secondary
predicate, analogous to adjectives and prepositional phrases in resultative constructions,
(44).
In both examples, there is a resultative predicate (red and down the loo) which predicates over
the direct object, denoting an endstate (red) or an end-location (down the loo). Particles, like
down in (43), have the same secondary predicate function, but they have a different syntactic
distribution. Resultative constructions such as the ones in (44) do not show the word order
alternation available to VPCs. This is illustrated in (45) and (46).
The distributional difference also shows up in nominalisations. Particles must occur before
the of-phrase in nominalisations, (47a), but adjectives have to appear after the of-phrase,
(47b).
(47) a. the tearing down of the tree-house / *the tearing of the tree-house down
b. *the painting red of his nose / the painting of his nose red
There are a few exceptions to the lack of the word order alternation with resultative
constructions. The adjectives clean, open, free, short can appear immediately adjacent to the
THE VERB-PARTICLE COMBINATION IN PDE 27
verb, just like particles. As expected, these adjectives appear before the of-phrase in
nominalisations, just like particles.
In Chapter 3, I will argue that the differences in distribution between particles on the one
hand and adjectives and prepositional phrases on the other follow from differences in their
structural status. I will propose that resultative adjectival and prepositional predicates are
always phrases, whereas particles are ambiguous between head and phrase. When they are
analysed as heads, they appear immediately adjacent to the verb.
The complex semantics of English VPCs include transparent meanings. Some examples of
transparent VPCs are presented in (49), repeated from (1).
(49) a. He cannot seriously expect his hamster to pick up the newspaper for him.
b. The four unmanageable kids chucked their parents right out.
c. The frustrated athlete put the weights down on the instructors foot.
In the group of transparent VPCs I include VPCs whose meaning can be inferred from the
sum of the meaning of its parts. The meaning of the verb and of the particle in a
transparent VPC corresponds to their meaning when used in isolation. Thus, the meaning
of the VPC to pick up, to lift something using your hands etc., in (49a) can be deduced from
the meaning of the verb to pick remove combined with that of the particle up towards a
higher position, upwards. The same procedure applies to the VPCs in (49b) and (49c).
A large number of English VPCs has a non-transparent meaning, but it is often hard to
determine whether the meaning of a VPC is truly non-transparent or not. This is sometimes
due to the fact that only part of the VPC is non-transparent. Consider for example the
VPCs in (50), repeated from (1).
(50) a. Their efforts to cheer up the singer who lost her voice were useless.
b. The artistic impulses of the zoo-keepers really cheesed the purple monkeys
off.
The particle up in the VPC cheer up in (50a) does not express the meaning it has in isolation
and is therefore non-transparent. The meaning of the verb to cheer, on the other hand, is
transparent and the meaning of the entire VPC is compositional in the sense of Nunberg,
Sag and Wasow (1994), mentioned in the discussion of Dutch Separable Complex Verbs
(SCVs) by Blom (2005: 82). In their view, compositionality is defined as the degree to
28 CHAPTER 1
which the meaning of a construction, once known, can be analysed in terms of the
contributions of its constituent parts (Blom 2005: 82, citing Nunberg, Sag and Wasow
1994: 498). Under this definition, the meaning of cheer up is clearly compositional, even
though the meaning of the particle up is non-transparent. Both the meaning of the verb and
that of the particle contribute to the meaning of the entire VPC. Thus, VPCs with a non-
transparent meaning can still have compositional semantics.
In (50b), the verbal part of the VPC cheese off, i.e. cheese, does not exist as a verb outside
the VPC and its meaning in the VPC is non-transparent. It does not refer to the food made
from milk, but its meaning is rather the result of the combination with the particle off. The
meaning of the particle off is not liking and is fairly transparent, because its isolated
meaning, movement away from, can still be detected. Unlike the meaning of the VPC in
(50a), the meaning of cheese off is non-compositional under Nunberg, Sag and Wasows
(1994) definition of compositionality. It appears to be the particle that contributes most of
the meaning to the meaning of the entire VPC, with the verb almost seeming a lexical
dummy.
The semantics of English VPCs, which are often non-transparent, suggests that VPCs
are listed in the lexicon as a unit. If this is the case, an explanation is needed for the
separability of verb and particle in the syntax. Jackendoff (1997) proposes that idioms (i.e.
for example non-transparent VPCs) have internal structure, which explains the syntactic
movability.
Another option would be that the verb and the particle are listed separately in the
lexicon and are inserted into syntax as two separate lexical items. This view must include an
account for the complex event semantics of VPCs which are often non-transparent.
In Chapter 3, I will propose a lexical decomposition analysis of English VPCs and will
claim that these lexically decomposed structures are in the syntax rather than in the lexicon.
The meanings of VPCs are nevertheless stored in the lexicon, which I assume hosts
idiosyncratic information (words and phrases). I assume that idiomatic constructs have
internal structure (cf. Jackendoff 1997) which allows me to account for the syntactic
behaviour of English VPCs.
The bulk of the VPCs discussed in this chapter so far take an internal argument. Apart from
transitive VPCs, English has a number of VPCs which do not have an internal argument.
Some examples are given in (51).
Many VPCs without an internal argument are unaccusative. Burzio (1986) defines
unaccusative verbs as verbs without an external argument, which therefore are unable to
assign structural case. This is formulated in Burzios Generalisation, (52).
The generalisation holds that the underlying direct object moves to the subject position
(where it receives nominative case), because it cannot receive case in object position.
Another group of verbs which lack an internal argument is the unergative class. The subject
of unergative verbs is semantically and syntactically a real subject, in contrast with subjects
of unaccusative verbs, which are objects underlyingly and therefore do not actively initiate
the action of the verb. There are several unaccusative/unergative diagnostics for English
(cf. Levin and Rappaport Hovav 1995; Alexiadou, Anagnostopoulou and Everaert 2004)
that can be used to determine the argument structure of VPCs without an internal
argument.
English intransitive verbs generally allow a cognate object, whose semantic content is
more or less identical to that of the verb. Consider the example in (53).
Whereas unergative verbs in English allow a cognate object, (54a), unaccusative verbs do
not, (54b).12
The explanation for the difference in grammaticality between (54a) and (54b) lies in the case
properties of unergative and unaccusative verbs (Burzio 1986). Unaccusative verbs do not
allow a cognate object, because they are unable to assign structural case. This problem does
not arise with unergative verbs, hence unergative verbs do allow cognate objects, cf. (54a).13
English change of state (or position) verbs have an intransitive and a transitive
(causative) use. This is known as the causative alternation, illustrated in (55).
(55) The piece of chalk broke. > The pupil broke the piece of chalk.
12 According to McIntyre (2004: 528), the example in (54a) is ungrammatical (cf. his example (7b)). A
Google search yields several examples in which sing along is accompanied by the cognate object a song,
however.
13 McIntyre (2004: 556) notes that intransitive cases which nevertheless allow an argument, as in (54a),
involve a change in semantics: the presence of the object forces a resultative reading.
30 CHAPTER 1
Since unaccusative verbs in English often express a change of state, it follows that they can
undergo the causative alternation, (56a). Unergative verbs, on the other hand, are normally
agentive and therefore do not allow the causative alternation, (56b).
(56) a. Her glasses fogged up. > The heat fogged her glasses up.
b. The parrot looked on. > *The cat looked the parrot on.
The availability of the causative alternation with unaccusative verbs, (56a), follows from the
thematic structure of unaccusative verbs (cf. Levin and Rappaport Hovav 1995: 80). They
have a patient or theme argument and lack an agent argument. Thus, the event expressed by
unaccusative verbs is brought about by an unexpressed agent. When this agent is expressed,
the patient or theme remains in object position. Not all unaccusative verbs participate in the
causative alternation, however, e.g. The applause died down > *The presenter died down the
applause. The alternation is not available for unergatives, cf. (56b), since subjects of
unergatives are real agents, which cannot be demoted to object position.
Some resultatives force the presence of a reflexive object, (57).
In (57), it is the resultative adjective crazy which selects the reflexive object, given that the
verb to laugh is intransitive. The appearance of a reflexive object follows from a condition on
resultative predication, formulated in the Direct Object Restriction (DOR) (Levin and
Rappaport Hovav 1995: 34).
Thus, when a resultative phrase combines with an intransitive verb, the appearance of a
reflexive object is necessary for the resultative predication to be successful. Simpson (1983)
was the first to note that the presence of a reflexive object signals that the verb is unergative
(Levin and Rappaport Hovav 1995: 34). Unergative verbs do not have an internal argument
that a resultative phrase can predicate over, hence a reflexive object appears (cf. DOR).
Unaccusative verbs, on the other hand, have an internal argument which is promoted, so no
reflexive object is required when a resultative phrase is present.
This test can be applied to resultative VPCs. Consider the examples in (59).
The examples in (59a) require the presence of a reflexive object, which indicates that the
verbs are unergative. The VPCs in (59b) are unaccusative, as is evident from the fact that a
reflexive object is not allowed. We may conclude, then, that all resultative VPCs without
reflexive objects are unaccusative.14
Unaccusative verbs allow there-insertion, (60a), unlike unergatives, (60b).
Again, the agentive role of the external argument of unergatives blocks demotion to object
position. Not all unaccusative verbs allow there-insertion. It works especially well with verbs
of appearance (e.g. arise) and existence (e.g. exist) (cf. Levin and Rappaport Hovav 1995:
149).
The differences in argument structure of VPCs indicate that VPCs can not all be
assigned the same syntactic structure. This is captured by Chomskys (1995a) framework, by
the assumption that the external argument is introduced by the verbal head v and is
therefore not present in the structure of unaccusative verbs, since they lack an external
argument.
The direct object the fridge in (61) is an argument of the verb and the secondary predicate
empty. VPCs are also complex predicates in the sense that there are two predicates which
express a complex event and also involve argument sharing.
The object his bicycle in (62) is the shared argument of the verb and the particle. Both the
verb to slow, which denotes an activity, and the particle down, which expresses the endstate,
affect the object his bicycle.
14Cf. Spencer and Zaretskaya (1998: 23), who point out that English resultatives cannot occur with
unergative verbs, since resultatives cannot predicate over an external argument. Note that unergative
verbs do occur in resultatives and VPCs with unselected objects, e.g. He slept his way to the top, She
chatted up the footballer.
32 CHAPTER 1
Particles seem to have the ability to alter the argument structure of the verb they combine
with, witness examples such as (63).
The presence of the direct object the garden gnome is unexpected given the intransitivity of the
simplex verb to run go quickly. Since the verb does not select an internal argument, we
conclude that it is the particle that licenses the direct object.15 This is supported by cases in
which the simplex verb and the VPC are both transitive, but occur with different objects, as
in (64) (from Los 2004: 85, her example (2)).
The claim that unselected objects are selected by the particle is further supported by the fact
that resultative phrases (i.e. the particle) must be predicated of a direct object. The subject-
predicate relation is confirmed by the possibility of paraphrasing the direct object and the
particle making use of the copula be: The garden gnome is down. This indicates that the
direct object is the external argument of the particle.
The status of unselected objects is not entirely undisputed. Levin and Rappaport Hovav
(1995) (henceforth L&RH) argue that unselected objects are not genuine arguments. They
do so on the basis of the fact that unselected objects fail three argumenthood tests (put
forward by Carrier and Randall 1992, henceforth C&R), the nominalisation test, (64a), the
adjective-passive test, (64b), and the middle-formation test, (64c).
(64) a. Nominalisation
the painting of the door red
*the drinking of the pub dry
b. Adjective-passive
the wiped-clean table (L&RH 1995: 43, ex.(27a))
*a drunk-dry teapot (L&RH 1995: 43, ex.(27b), from C&R 1992, their
example (73c))
c. Middle-formation
this metal pounds flat easily (L&RH 1995: 43, ex.(26a))
*this type of pavement runs thin easily (L&RH 1995: 43, ex.(26b), from C&R
1992, their example (69a))
15Alternatively, it could be argued that it is the entire VPC that selects the argument. This is in line
with an analysis in which the verb and the particle are a morphological object and listed as a single
lexical item in the lexicon. I reject such a view on the basis of the difficulties it has in accounting for
the syntactic properties of VPCs.
THE VERB-PARTICLE COMBINATION IN PDE 33
Spencer and Zaretskaya (1998: 9) show that these tests are not completely reliable. With
respect to the nominalisation test, they suggest that the unacceptability of nominalisations
involving unselected objects is stylistic. They note that the use of unselected objects is
generally restricted to the conversational register, and that the use of ing nominalisations
normally does not appear in that register. Moreover, grammaticality slightly improves when
the resultative predicate is placed in front of the unselected object and immediately adjacent
to the verb, as in (65) (from Spencer and Zaretskaya 1998: 9, their example (29)).
As Spencer and Zaretskaya (1998: 9) point out, this is reminiscent of VPCs, which only
allow nominalisation in the order VPrtNP: the looking up of the information versus *the looking
of the information up.16 They further note that adjectival passive formation with unergative
resultative predicates yields better results when other word orders are allowed. This is
illustrated by the example in (66), which is grammatical (from Spencer and Zaretskaya 1998:
10, their example (31)).
(66) The pub, drunk dry by an entire football team, looked a sorry place.
Such relaxation of the word order is not even required for VPCs with unselected objects, as
is shown by the grammaticality of cases like a written-off car, the looked-up information, the dug-up
bone. Spencer and Zaretskaya (1998: 10) also mention Goldbergs (1995) point that the
middle-formation test is not very reliable, since middle formation is highly restricted in
English. At the same time, some VPCs with unselected objects do in fact allow middle
formation, e.g. That plant digs up easily. The facts presented here prove wrong L&RHs (1995)
claim that unselected objects are not arguments.
An analysis treating unselected objects as arguments yields a case problem. It is not clear
how they should receive case, given that the intransitive verb does not have structural case
to assign and given that particles are not case assigners. Zeller (2001: 218-225) proposes that
the verb can inherit the case assignment property of P, because it is structurally adjacent to
P. The particle has to transfer this property to the verb, because it cannot assign case itself
due to the lack of functional structure. Zeller assumes that particles have the case assigning
property because they are related to prepositions, which are case assigners.
A special type of VPC in which the particle triggers the presence of an argument is the
so-called time-away construction (Jackendoff 1997, 2002), (67).
16 Spencer and Zaretskaya (1998: 10) observe that the non-transparent meaning of many English
VPCs makes it difficult to determine whether an unselected object is involved. This is easier with
transparent VPCs, such as Fido dug a bone up versus *Fido dug a bone (Spencer and Zaretskaya 1998: 10,
their example (33)).
34 CHAPTER 1
The pair in (67) indicates that the particle away selects the argument the afternoon. On its own,
the verb to write does not allow the object the afternoon, (67a), but the presence of the particle
away allows a direct object, (67b). Interestingly, the particle away only licenses arguments that
expresses a period of time (e.g. the afternoon), given the ungrammaticality of *Pam wrote the love
letter away (versus Pam wrote the love letter).
The discussion so far has focused on resultative particles, which select an (external)
argument. English also has a set of particles (on, along, away) which express duration rather
than a result. Durative particles seem to block the argument of the verb they combine with.
The examples in (68) are taken from Deh, Jackendoff, McIntyre and Urban (2002: 14),
their example (19).
The argument of the simplex verb is blocked by the presence of the particle. Durative
particles highlight the time span of the activity expressed by the verb, rather than the entity
that is affected by the activity. This also explains why arguments expressing a period of time
are allowed, as these fit in with the temporal semantics of the particle. The time-away
construction (cf. (67)) is an example of this. The presence of an object would create a telic
event and thus would trigger a resultative interpretation.
In her study on first language acquisition, Clark (1993: 28) notes that particles with a
transparent meaning (i.e. expressing space and motion) belong to the set of early words that
children acquire. This is in line with the three important factors in acquisition she mentions,
namely transparency of meaning, simplicity of form and productivity (Clark 1993: 15). At
first, children make use of particles without any accompanying verb. The data presented by
Clark provide evidence for an analysis in which particles are analysed as predicates. At a
later stage in the acquisition process, children begin to combine the particles with verbs.
The first verbs they put to use in VPCs are light verbs (such as do, make, get, go), i.e. general-
purpose verbs which express an activity or action (Clark 1993: 29-30). Later on, more
specific verbs are added to the inventory. At an even later stage, children acquire VPCs with
a non-transparent meaning, which are not productive and have to be listed.
The intricate characteristics of English VPCs presented in this chapter provide a real
challenge to analyses of the construction. The prime concern is to account for the
ambiguous behaviour of VPCs. Particles act as independent elements in that they are
THE VERB-PARTICLE COMBINATION IN PDE 35
separable from the verb, but at the same time seem to form a unit with the verb when they
cannot be separated, when the VPC has a non-transparent meaning and a unique argument
structure. The ambivalent nature of English VPCs not only raises questions about their
structural status (i.e. whether they are words or phrases), their puzzle also bears upon
general linguistic issues such as the boundary between syntax and morphology.
The rich array of VPC meanings, which may vary on a scale from transparent to
completely non-transparent, must not be neglected. There is a mismatch between the syntax
and semantics of English VPCs, in that non-transparent VPCs have the same syntactic
possibilities as transparent VPCs (an exception is topicalisation, which involves a semantic-
prosodic condition). Moreover, resultative and non-resultative VPCs show distinct
argument structures, which must be reflected in the syntax.
These and other issues will dealt with in the light of existing analyses in Chapter 2.