Stylistic Inversion in English A Reconsideration

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/225827349
Stylistic Inversion In English: A Reconsideration
Article in Natural Language & Linguistic Theory · May 2001

DOI: 10.1023/A:1010646417840
CITATIONS READS
90 3,015
2 authors:
Peter W Culicover Robert Levine

The Ohio State University The Ohio State University
52 PUBLICATIONS 1,015 CITATIONS 48 PUBLICATIONS 741 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Ellipsis in Hybrid Type-Logical Categorial Grammar View project
Languages & linguistics View project
All content following this page was uploaded by Robert Levine on 27 October 2015.
The user has requested enhancement of the downloaded file.

PETER W. CULICOVER and ROBERT D. LEVINE
STYLISTIC INVERSION IN ENGLISH: A RECONSIDERATION
ABSTRACT. We argue that the phenomenon described and discussed in the literature
as locative or stylistic inversion in English is actually a conflation of two quite different
constructions: on the one hand, light inversion (LI), in which the postverbal NP element
can be phonologically and structurally extremely simple, possibly consisting of a single
name, and on the other hand heavy inversion (HI), where the postverbal element is heavy
in the sense of Heavy NP Shift.1 We present evidence that the preverbal PP in LI patterns
with subjects but the PP in HI is a syntactic topic, using a variety of tests which distinguish
A-positions from Ā-positions. Other significant differences between HI and LI, such as the
classes of verbs which support these two constructions respectively, and the differential
behavior of HI and LI with respect to adverbial placement, provide support for interpreting
HI as a case of Heavy NP Shift applying to subject constituents.
1. I NTRODUCTION
Levin & Rappaport Hovav (1995) have recently argued against the view
that stylistic inversion is a diagnostic for unaccusativity.2 Rather, they
suggest, stylistic inversion occurs with a wide range of verbs, including
unaccusatives, passives, and, crucially, unergatives. We demonstrate in the
following discussion that the argument of Levin and Rappaport Hovav
does not go through, because they, along with all other students of styl-
istic inversion, fail to observe that there are actually two stylistic inversion
constructions in English. One construction, which we call light inversion
(LI), is restricted to unaccusatives; the other, which we call heavy inversion
(HI), is not (we explain this technology shortly). In general, it has been
evidence of HI that has been used to argue that stylistic inversion is not
restricted to unaccusatives.
1 An earlier version of this paper was presented at the Colloque de Syntax et Se-
mantique, University of Paris VII, October 1995. We thank the participants at that
conference for their comments, as well as various other audiences elsewhere which have
provided us with helpful feedback, including the University of Girona. In addition, we wish
to express our appreciation for the care and effort evident in the responses to our paper of
several anonymous referees for NLLT.
2 Throughout we use the term ‘stylistic inversion’. Another term commonly found in
the literature is ‘locative inversion’.
Natural Language & Linguistic Theory 19: 283–310, 2001.

© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
284 PETER W. CULICOVER AND ROBERT D. LEVINE
We begin by adducing evidence in §2 that in LI the fronted PP is a

subject, that is, occupies the Spec position associated with IP. In §3 we
elaborate our claim that there are two stylistic inversion constructions,
presenting a wide range of evidence that stylistic inversion with ‘light’
subjects is possible only when the verb is unaccusative; when the verb is
unergative or even transitive, stylistic inversion is possible, but only with
a ‘heavy’ subject. The notion of ‘heavy’ here corresponds exactly to the
one that is relevant to Heavy NP Shift (see (Arnold et al. 2000) for detailed
discussion of the factors which heaviness comprises). We assume that in
the case of light inversion (LI), the subject is in situ in VP, while in the
case of heavy inversion (HI), the subject appears in [Spec, IP] at some
point in the derivation and subsequently postposes to the right of VP. For
concreteness we assume the following derivations.
(1) LI:
[IP e I [VP V NPsubj PP . . . ]] →
[IP PP I [VP V NPsubj t . . . ]]
(2) HI:
[IP e I [VP NPsubj V PP . . . ]] →
[IP NPsubj I [VP tsubj V PP . . . ]] →
[IP tsubj I [VP tsubj V PP . . . ] NPsubj ] →
[IP PP [IP tsubj I [VP tsubj V tPP . . . ] NPsubj ]]
We stress at the outset that the main focus of this paper is that there are
two constructions. Space considerations prohibit us from exploring in sat-
isfactory depth all of the technical questions bearing on the specific details.
We do assume, following proposals of Coopmans (1989) and Hoekstra &
Mulder (1990) among others, that the subject NP in (1) is selected as a
sister of the unaccusative verb. Either it or the PP moves into the higher
specifier position, which we assume to be [Spec, IP]. The apparent op-
tionality of such movement is an obvious problem from the perspective
of a theory of movement triggered by the need to discharge features (e.g.,
Chomsky 1995) but we will not pursue this aspect of the analysis here.
More controversially, we assume that the sentence-final subject in (2)
is necessarily in [Spec, IP] at some point in the derivation, and that it ends
up in final position through movement. If this NP moves to the right, as we
assume in (2), then this clearly raises important questions in the light of
the proposal of Kayne (1994) that there are no rightward movements. For
recent commentary on this as well as other aspects of Kayne’s proposal,
see the papers in Beermann et al. (1997). It is conceivable that the proper
STYLISTIC INVERSION IN ENGLISH: A RECONSIDERATION 285
derivation of HI does not involvement movement of the subject to the right,

but rather movement of everything else to the left. We will not be able to
develop and evaluate here an analysis along these lines.
An additional complication is that movement of the NP to the right
leaves a trace which must be licensed. It is generally claimed that in the
configuration that t Infl . . . , the trace of the subject is not licensed (see, for
example, Rizzi (1997)). The question then arises as to why the subject trace
would be licensed in the configuration PP t Infl . . . , as in (2). Our hypo-
thesis is that the licensing of the subject trace is not a strictly grammatical
phenomenon, but rather a processing effect. Again, to develop such an idea
in satisfactory depth would take us far afield and away from the primary
focus of the paper.
In the following section, we briefly touch on the claim that LI occurs
only when the verb is unaccusative. The facts turn out not to be entirely
simple, but the generalization can be sustained more or less in this form.
We support this claim by providing a number of syntactic contexts in which
LI is impossible, but where HI yields a structure which creates the illusion
of an ordinary stylistically inverted form. §4 summarizes our conclusions
and notes several important issues which our conclusions raise, but which
we have not been able to address within the confines of this paper.
2. PP IS A S UBJECT
Frequently cited evidence that the PP in stylistic inversion is a subject is

the following. First, long extraction of the PP produces a that-t effect, as
first noted in Bresnan (1977); see also Culicover (1993). This generaliz-
ation extends to other complementizers (e.g., whether-t, extraction from
gerundives) that show a Comp-t effect (Pesetsky 1982). We illustrate the
relevant data in examples (3)–(7):
• that-t:
(3)a. Into the room Terry claims (∗ that) t walked a bunch of gorillas
b. Into which room does Terry claim (∗ that) t walked that bunch
of gorillas?
(4) That bunch of gorillas, Terry claims (∗ that) t walked into the
room.
• whether-t:
(5)a. ?In to this room, Terry wonders whether a bunch of gorillas had
walked t.
∗
b. Into this room, Terry wonders whether t had walked a bunch of
gorillas.
• gerundive:
(6)a. Terry imagined a bunch of gorillas walking into the room.
b. Into the room Terry imagined a bunch of gorillas walking.
c. Into the room Terry imagined t walking a bunch of gorillas.
d. Into which room did Terry imagine a bunch of gorillas walking?
e. Into which room did Terry imagine walking a bunch of gorillas.
f. How many gorillas did Terry imagine t walking into the room?
(7)a. Terry thought about a bunch of gorillas walking into the room.
b. ?Into the room Terry thought about a bunch of gorillas walking.

∗
c. Into the room Terry thought about t walking a bunch of gorillas.
d. ?Into which room did Terry think about a bunch of gorillas

walking?
∗
e. Into which room did Terry think about walking a bunch of
gorillas?
∗
f. How many gorillas did Terry think about t walking into the
room?
But this argument is far from conclusive, because it crucially assumes that
it is the fronted PP and not the postverbal subject which is responsible for
the trace in subject position. We argue later that the postverbal subjects
in such examples are exclusively heavy, in precisely the sense that distin-
guishes constituents eligible to undergo heavy NP shift from those which
are not, and hence must be moved to their surface position from [Spec, IP].
What the starred examples then show is that that-t is indeed ill-formed, but
not that the extracted PP is linked to the subject trace.
Second, the fronted PP in stylistic inversion appears to undergo Raising,
suggesting that it is a subject.
(8)a. A picture of Robin seemed to be hanging on the wall.
b. On the wall seemed to be hanging a picture of Robin.

These sentences can be derived on the approach taken in Rochemont
& Culicover (1990) (RC), in which the adverbial is topicalized and the
combination of modal and main verb or tensed verb is moved into second
position. However, as Levine (1989), Culicover (1991) and Coopmans
(1992) point out, it is necessary in such cases to extend RC’s I/V-raising
to a less principled restructuring operation, as in (9). In this derivation,
a succession of lexical heads must raise into higher Infl nodes to form a
complex I/V category which can then undergo a subsequent raising to the
highest Infl node as a single unit, finally undergoing moving to C to give
rise to the distinctive inversion. On this analysis, a category [I/V seemed to
be hanging] must be formed, join seem in a higher I/V category, and at last
raise to the matrix Comp to yield [I/V/C seemed to be hanging], taken to be
a single complex head:3
(9) [IP e [Infl seemed] [IP [a picture of Robin] [Infl to] [VP be hanging
on the wall]]] →
[IP e [Infl seemed] [IP [a picture of Robin] [I/V to bei [VP ti
hanging on the Wall]]] →
[IP e [Infl seemed] [IP [a picture of Robin] [I/V to bei hangingj ]
[VP ti tj on the wall]]] →
[IP [a picture of Robin]k [Infl seemed] [IP tk [I/V to be hanging]
[VP . . . on the wall]]] →
[IP [a picture of Robin]k [I/V seemed [to be hanging]l ] [IP tk tl
[VP . . . on the wall]]] →
[IP [VP . . . on the wall]m [IP [a picture of Robin]k [I/V seemed to
be hanging] [IP tk tl tm ]]] →
[IP [VP . . . on the wall]m [I/V/C seemed to be hanging]h [IP [a
picture of Robin]k th [IP . . . ]]]
The RC analysis thus becomes distinctly implausible once more complex
structures are involved in the inversion process.
3 In order to minimize notational complexity, we replace strings of traces with ellipses
where appropriate.
Regardless of the particular details, it turns out that Raising cannot be

taken as evidence that the PP is a subject if, as we argue later, the postverbal
subject can arrive in this position by rightward movement from [Spec, IP]
in some cases of stylistic inversion. The correct analysis, we claim, is that
the postverbal subject is what undergoes Raising, PRIOR to its movement
to the right. The extracted PP is then in a topic position in these examples.
In fact it appears that the PP in the case of LI CANNOT undergo Raising,
in spite of the fact that it is in [Spec, IP].4
Consider the following contrasts.
(10)a. Into the room appeared to be walking slowly a very large

caterpillar.
b. Into the room walked Robin slowly.5
c. ∗ Into the room appeared to be walking Robin slowly.
(11)a. Slowly into the room walked Robin boldly.
b. ∗ Slowly into the room appeared to walk Robin boldly.
(12)a. Into the room singing walked Robin slowly.
b. ∗ Into the room singing appeared to walk Robin slowly.

4 It is not entirely clear why the PP in LI does not undergo Raising. If the PP can raise
from VP to [Spec, IP] in the first place, then we might expect that it would satisfy the
conditions for raising from a lower [Spec, IP] into a higher [Spec, IP]. Thus it appears
that the answer to the question must be a semantic one. If, for example, seem and appear
predicate of [Spec, IP], then only a referential PP could be in this position as the surface
subject of seem and appear. Contrast the following:
(i)a. Under the table is a good place to put the beer.
b. Under the table rolled the beer slowly.
(ii)a. Under the table seems to be a good place to put the beer.
b. ∗ Under the table seemed to roll the beer slowly.
5 Note that on our current analysis, slowly into the room must be a constituent. This
contrasts with the view taken by RC, which is that slowly into the room is the remnant of a
VP from which the V has raised. This analysis is ruled out on the present account, due to
the presence of the subject NP within VP.
The presence of the adverb after the subject forces the LI analysis. We see
that in this case, a simple PP, or a more complex XP (a V-less VP in the
RC analysis, cannot undergo Raising to a higher subject position.
Yet while these two arguments ultimately fail to support the treatment
of PPs in stylistic inversion as subjects, there is a significant set of data,
reflecting a systematic differences between A- and Ā-positions, which
confirms the subject status of the preverbal PPs in (light) stylistic inversion,
and which is not consistent with PP moving directly into a topic position
in these cases, viz., the fact that true stylistic inversion, which we refer to
as Light Inversion (LI), does not produce weak crossover (WCO) effects
(just like Raising, and in contrast with wh-movement).6 The basic contrast
we appeal to here is shown in (13):
(13)a. Whoi appears to hisi mother [ti to be a genius]?
b. ?Whoi is hisi mother grilling ti obsessively? (WCO).
c. ??Whoi does hisi mother think [ti is a genius]? (WCO).
d. ?To whomi did hisi students give ti a teaching award?
While the last three examples here are not altogether impossible, they are
far less well-formed than the Raising example in (14a), which is impec-
cable.7 A strikingly parallel contrast is evident between PPs in stylistic
inversion on the one hand and straightforward topicalization on the other,
where the PP contains a quantified NP that is to be interpreted as binding
a pronoun in the post-verb NP:
(14)a. ∗ Into every dogi ’s cage itsi owner peered. (Topicalization, WCO)
b. Into every dogi ’s cage peered itsi owner. (Stylistic inversion, no

WCO)
6 Prior claims for the subject status of the subject status of the PP have been made
by Bresnan (1994) and, in somewhat more complex form, in Stowell (1981), where the
PP moves through subject position en route to a final topic position. We stress that, as
indicated, we do not take all the evidence cited in such sources as genuine support for the
analysis of PPs as subjects, though we agree on the conclusion.
7 Note that whether the fronted wh element is fronted on its own or is pied pied, the
effect is the same – as we would expect, given, on the one hand, index percolation at S-
structure, and reconstruction of the preposition back to its D-structure location at LF on
the other.
∗
(15)a. Itsi owner stood next to none of the winning dogsi . (WCO at
LF)
b. Next to none of the winning dogsi stood itsi owner. (Stylistic

inversion, no WCO)
∗
c. Next to none of the winning dogsi itsi owner stood. (Topical-
ization, WCO)
d. ∗ ?Next to none of the winning dogsi did itsi owner stand.

(Negative inversion, WCO)
(16)a. In every dogi ’s cage hung itsi collar.
b. ∗ In every dogi ’s cage hung on a hook itsi most attractive and

expensive collar.8
The relevant point is highlighted in (14): the quantified NP within the

PP in the stylistic inversion example (14b) can bind the possessive pro-
noun within the VP, parallel to the raised subject who in (13a), while the
quantified NP within the topicalized PP in (14a), like the wh-moved NP
in (l3b–c), cannot bind the corresponding pronoun. The difference in the
status of the inversion and topicalization example respectively shown here
follows immediately on the assumption that in (16a) the PP is in a A-
position and the subject is in VP, while in (16b) the PP is topicalized and
the subject is linked to [Spec, IP]. A particularly clear demonstration of
8 An anonymous reader judges examples (16a) and (16b) to be indistinguishable in
grammaticality. We suspect that the relative acceptability of (16b) is due to a reading of
every as each, which does not produce weak crossover violation. Compare (16b) with (i).
(i) In each dog’s cage itsi most attractive and expensive collar was sitting on a
hook.
Replacing every by no should sharpen the judgment for those speakers for whom the
difference between (16a) and (16b) is minimal.
(ii)a. In no dogi ’s cage hung itsi collar.
b. ∗ In no dog ’s cage was hanging on a hook its most attractive and expensive
i i
collari .
c. ∗ In no dog ’s cage its most attractive and expensive collar was hanging on a
i i
hook.
the contrast we find between the two kinds of case emerges from the fact
that the postverbal quantifier no dog produces a WCO violation when it
binds the pronoun in the PP in A-position in ??∗ In itsi cage sat no dogi ,
just as a quantifier in a direct object produces a WCO violation when the
pronoun is in an NP subject, as in ∗ Itsi master criticized no dogi . Again,
the stylistic inversion cases pattern in a fashion parallel to examples with
a quantified subject uncontroversially in [Spec, IP]. Example (16b), on the
other hand, falls together with the case in which the PP is topicalized and
the subject is in [Spec, IP], as in (13d), or, e.g., ∗ To every instructori hisi
students gave a teaching award. In this case, as we have already noted,
the PP behaves as though it is reconstructed into the postverbal position.
Compare the examples in (15), which show the same pattern.
The WCO data we have adduced thus points strongly to the conclusion
that the fronted PP is a syntactic surface subject (that is, is in [Spec. IP]) or,
at the very least, is in a superior A-position with respect to binding, weak
crossover, and so on. This hypothesis is consistent with Bresnan’s (1993)
proposal that PP is assigned the SUBJ function under an LFG treatment.
3. L IGHT AND H EAVY I NVERSION
We turn now to our claim that there are two types of stylistic inversion. To
launch the discussion, we repeat an example cited by Levin and Rappaport
Hovav that is intended to demonstrate that stylistic inversion occurs with
unergatives.
(17) In the enclosure, among the chicks, hopped the most recent chil-
dren of Nepomuk and Snow White. [M. Benary, Rowan Farm
287] (LRH’s (78): 257)
It will be noted that the subject in this example is relatively complex. When
we replace it with a less complex simple NP, the sentence becomes a good
deal less natural; it is considerably improved if the NP is made prosodically
more prominent.
The difference between the heavy and light subjects is made still
sharper when we introduce more material into the VP. As noted in Kathol
& Levine (1992), a simple subject NP cannot appear at all after a V-
Adverb, but a focused or heavy NP can, when the verb is unaccusative;

thus, compare the ill-formed (18c) with (d) and (e):
(18)a. Into the room walked Robin
b. Into the room walked Robin carefully.
c. ∗ Into the room walked carefully Robin.
d. Remember Robin? Well, into the room walked carefully,

. . . ROBIN!
e. Into the room walked carefully the students in the class who had
heard about the social psych experiment that we were about to
perpetrate.
The pattern is comparable to that of Heavy NP Shift in VP, as illustrated in

(19):
(19)a. Carefully I addressed Robin.
b. I addressed Robin carefully.
c. ∗ I addressed carefully Robin.
d. I addressed carefully . . . ROBIN!
e. I addressed carefully the students in the class who had heard

about the social psych experiment that we were about to
perpetrate.
f. I addressed the students in the class who had heard about

the social psych experiment that we were about to perpetrate
(extremely) carefully.
It is important to note that the intended judgments are difficult if not

impossible to make unless the sentences are spoken with the proper in-
tonation. Embedding the examples in context may help to produce the
intonation, but to reinforce the point, we illustrate the intonational phrasing
of some of the crucial examples in (19). Specifically, in the HI example
(18d), there are three intonational phrases, one for into the room, one for
walked carefully, and one for Robin.9
(20)
This “HI-intonation” is the intonation of example (18d) and all of the

examples of HI in this paper.10 What is crucial here is the segmentation
of the intonational pattern into three phrases; the precise contour of each
phrase may vary to some extent.
When the verb is unergative, the light NP cannot appear postverbally at
all, while the heavy NP can appear after the VP, but not before it, as we see
in (21).
(21)a. ∗ In the room slept Robin.
b. ∗ In the room slept Robin fitfully.
c. ∗ In the room slept fitfully Robin.
d. Remember Robin? Well, in the room slept fitfully . . . ROBIN!
e. In the room slept fitfully the students in the class who had
heard about the social psych experiment that we were about to
perpetrate.
f. In the room slept the students in the class who had heard about
the social psych experiment that we were about to perpetrate
(very) fitfully.
Here the crucial contrast is between the c. example and the d./e. examples.
Such contrasts follow immediately if a sentence such as (21e) is derived by
movement of the heavy NP subject to the right, as suggested by RC. If this
approach is correct, we would expect the heavy NP to appear exclusively
external to the VP, since it would then be moving across the entire VP
from [Spec, IP], perhaps adjoined to IP as shown in (22), and the contrast
9 We are grateful to Mary Beckman for her help in notating the HI intontation.
10 As Mary Beckman has pointed out to us (p.c.), this phrasing correlates rather nicely
with what we claim to be the constituent structure of such examples (see footnote 2).
between (21e) and (21f) indeed shows that this subject must be in a po-
sition adjoined outside the VP slept fitfully, just as the scenario we have
outlined requires.
(20)
But then the pattern seen in connection with unaccusative verbs, for ex-
ample (18b), where the subject is light and cannot appear in the adjoined
position occupied by the heavy NP in (22), must have a different deriva-
tion, one in which the subject is licensed in a VP-INTERNAL position.11
11 It should be pointed out that when the verb is unaccusative and the subject is heavy.
there is really no way to tell whether the subject is in situ in VP as in the LI construction,
or whether it has moved to the right from [Spec, IP] as in the HI construction. Such a
sentence will display all of the properties of both constructions (since, on our account. the
conditions for each of the two homophonous structures will be satisfied) and will therefore
have no diagnostic utility vis-à-vis the proposed analysis.
Such a position is available only to the subject of verbs like walk, given
the difference with sleep that is illustrated here.12,13
12 We conjecture that there is a correlation between this structure, in which the unaccus-
ative subject originates as the direct object of the verb, and the interpretation of “movement
along a path” that is typical of the unaccusative construction. Note that this correlation is
constructional, not lexical, given that such an interpretation can be associated with any verb
that can be plausibly used to denote a property of movement along a path:
 

stumbled

 


 
wobbled 
(i) Into the room stormed Fred.

 


 
blustered
 
skidded 
13 As pointed out by two reviewers, the derivation that we propose for HI raises the
question of how it is that topicalization of PP and movement to the right of the heavy
NP can interact. If the heavy NP moves first, we might expect the resulting structure
to be ‘frozen’ (cf. Wexler & Culicover 1980), blocking subsequent topicalization. But if
topicalization applies first, then we might expect there to be a topic-island effect, blocking
subsequent movement to the right of the heavy subject NP.
As pointed out by Johnson (1985), the evidence that heavy NP shift blocks subsequent
extraction is not conclusive. In the following example, the PP must extract from a VP to
which Heavy NP Shift has applied.
(i) the refrigerator [into which]i I put tj ti after I got home [all of the beer that I
had bought]j
Moreover, compare the following examples.
(ii)a. Whoi did you give all of your money to ti ?
b. ∗ Who did you give t to t [all of your money] ?

i j i j
c. the person whoi I mentioned tj to ti over the phone [the decision to close the
factory]j
Example (iic) suggests that the problem with (iib) is not strictly speaking a matter of a
grammatical constraint that blocks extraction. Rather, it appears to have to do with the
identification of the trace of the wh-phrase, which is facilitated when there is material
intervening between it and the postposed heavy NP. For related ideas, see Fodor (1978)
and Jackendoff & Culicover (1972).
Regarding the possibility that there is a topic-island effect, again we suggest that in some
cases extraction across a topic presents problems for language processing, particularly
when the topic and the extracted constituent are of the same syntactic category. Extraction
of a subject to the right when there is a PP topic does not present comparable difficulties.
In fact, we note that the Adverb Effect of Culicover (1993) constitutes evidence that a PP
topic actually ameliorates problems caused by extraction of a subject to the left.
Presentational there constructions are standardly taken to illustrate the

existence of a class of unaccusative verbs in English, as in e.g., Coopmans
(1989). But there phenomena also provide independent motivation for
movement of the subject to the right, and for the observation that the
locative PP need not move to the left. Thus, RC argue that movement of
a heavy NP subject to the right produces Presentational there Insertion
(PTI), as in (23):
(23)a. There slept fitfully in the next room a group of the students in
the class who had heard about the social psych experiment that
we were about to perpetrate.
b. In the next room there slept fitfully a group of the students in

we were about to perpetrate.
Our proposal, however, is that the appearance of there in subject position

is not the only way to license such a rightward displacement of the subject;
rather, what must be the case is that the empty subject position be licensed,
either by filling it with there, or by preposing the PP. If this proposal is
correct, then the fact that HI appears to be a type of stylistic inversion
is in part an illusion. We claim that what seems to be stylistic inversion,
via Heavy NP Shift of a subject, exists in numerous contexts where light
inversion is impossible – a state of affairs making it a priori very unlikely
that a single mechanism subsumes both HI and LI. Consider the following:
(i) Heavy NP Shift derives the illusion of stylistic inversion in infinitival

complements (as in (24b, c)):
(24)a. I expected Robin to walk into the room.
b. ∗ I expected t to walk Robin into the room/∗ I expected t to walk

into the room Robin.
c. I expected t to walk into the room . . . ROBIN! [HI intonation]
d. I expected t to walk into the room a group of the students in the

class who had heard about the social psych experiment that we
were about to perpetrate. [HI intonation]
e. I expect t to preach from this pulpit a close associate of the great

Cotton Mather. [HI intonation]
(25)a. ∗ Into the room I expected t to walk Robin.
b. Into the room I expected t to walk . . . ROBIN! [HI intonation]
c. I didn’t expect ROBIN to walk into the room; rather, into the
room I expected t to walk a group of the students in the class
who had heard about the social psych experiment that we were
about to perpetrate [HI intonation, where indicates a marked
juncture].
d. Q: Who did you expect to preach from this pulpit? A: From this
pulpit I expected t to preach a close associate of the great Cotton
Mather. [HI intonation]
(ii) Heavy NP Shift derives the illusion of stylistic inversion in gerundives,

as in (26)–(28).14
(26)a. I was speculating about who would walk into the room. First, I
imagined Robin walking into the room.
b. I imagined into the room t walking ∗ (. . . ) Robin.
14 The reader may find these data somewhat surprising, in that on our analysis the well-
formed inversion examples are analyzed as instances of PP topicalization. yet it is well-
known that topicalization within nonfinite clauses is typically extremely degraded. But
this is far less true in the case of gerundives than infinitives. Compare, for example,
(i) T HAT solution Robin having already explored t and rejected t, she decided to
see if she could mate in six moves with just the rook and the two pawns.
(ii) ∗ I really want ∗ THAT solution Robin to explore t thoroughly.
It thus appears that gerundive clauses are rather more tolerant of topicalization than in-
finitive clauses; in fact, this is essentially what we would predict if the Case-assignment
properties of gerundives are as analyzed in Reuland (1983). where the subject of gerunds
is governed by the verbal affix and thus an internal source of Case is available to such
subjects, as opposed to infinitive clauses, whose overt subjects must be in all cases be
externally governed in order to receive Case. We grant however that they are probably not
up to the standard of normal finite clause complementation and might therefore strike some
readers as less than fully natural.
c. I was speculating about who would walk into the room, and I
imagined into the room t WALKing a group of the students in
we were about to perpetrate. [HI intonation]
d. I was having a fantasy about what had happened in this church,

and I imagined from this pulpit t PREACHing a close associate
of the great Cotton Mather. [HI intonation]
(27)a. I decided to let no one into the room; in fact, ∗ I prevented t from
walking into the room Robin.
b. I prevented t from walking into the room . . . ROBIN! [HI

intonation]
c. I prevented t from walking into the room a group of the students

in the class who had heard about the social psych experiment
that we were about to perpetrate. [HI intonation]
(28)a. I decided to let no one into the room; in fact, ∗ into the room I
prevented t from walking Robin.
b. Into the room I even prevented t from walking . . . ROBIN!
c. Into the room I even prevented t from walking a group of the

students in the class who had heard about the social psych
experiment that we were about to perpetrate. [HI intonation]
d. I decided to allow no one to do anything in this church; in

fact, from this pulpit I even prevented t from preaching a close
associate of the great Cotton Mather. [HI intonation]
(iii) Heavy NP Shift corresponds to control of PRO by an ‘invisible’

subject coindexed with the postverbal heavy NP, as in (29d, e).
(29)a. Robin expected PRO to walk into the room.
b. Into the room Robin expected PRO to walk.
c. ∗ Into the room t expected PRO to walk Robin.

d. Into the room t expected PRO to walk . . . ROBIN! [HI

intonation]
e. We had set up the protocols perfectly to “trick” the students,

so that into the room t fully expected PRO to walk a group of
the students in the class who had heard about the social psych
experiment that we were about to perpetrate. [HI intonation]
f. Preaching from this pulpit is a great achievement and people

come from near and far hoping to do it. In fact, from this pulpit
t expected PRO to preach a number of close associates of the
great Cotton Mather himself. [HI intonation]
(30)a. Robin avoided PRO walking into the room.
b. Into the room Robin avoided PRO walking.15

15 The following example is ill-formed on normal intonation:
(i) Remember Robin and her fear of windows? ∗ Well, predictably, into the room
t avoided PRO walking Robin.
But note that the following examples appear to be well-formed with the appropriate
prosody:
(ii) They said that not everyone would recklessly walk into the room, and,
predictably, into the room t avoided PRO walking . . . ROBIN! [HI intonation]
(iii) We had set up the protocols perfectly to “trick” the students. But for some
reason, into the room t AVOIDED PRO walking a group of the students in the
class who had heard about the social psych experiment that we were about to
perpetrate. [HI intonation]
(iv) Preaching from this pulpit was known by many to be terribly unlucky; in fact,
from this pulpit t studiously AVOIDED PRO preaching any sane associate of
Cotton Mather/even the LEAST superstitious of Cotton Mather’s associates.
[HI intonation]
These and the previous examples raise the obvious question of how the ECP is to be satis-
fied with respect to the trace in subject position. The question is actually more complicated,
in view of the problems noted in Culicover (1993) in accounting for the that-trace effect
in terms of the ECP. For an interesting approach to these problems, see Rizzi (1997); full
discussion of the possible sources of the that-trace effect and their interaction with the
structures we are positing for Heavy Inversion would take us well beyond the scope of the
present paper, and we leave investigation of this issue for future work.
Clearly, it is extremely unlikely that the PP is interpretable as the controller

of PRO in these cases; the simplest assumption is that PRO is somehow
controlled by the heavy postverbal NP. But since PRO subjects in comple-
ments of expect, for example, are obligatorily controlled by the subject of
expect, it follows that there is a subject of expect in the examples in (29)
coindexed with the heavy NP, but invisible – exactly what follows from
our HI analysis.
(iv) Heavy NP Shift derives the illusion that stylistic inversion occurs in
the complement of a perception verb, as in (31)–(32).16
(31)a. ∗ We saw into this room run Robin.
b. It was terrible to be in the hotel during the Tolstoy convention;

we actually saw into this room RUN a ravenous horde of
angry Tolstoy scholars. [HI intonation]
c. We heard from this pulpit preach a close associate of Cotton

Mather. [HI intonation]
(32)a. ∗ Into this room we saw run Robin.
b. Into this room we saw run a ravenous horde of angry Tolstoy

scholars. [HI intonation]
c. From this pulpit we heard preach a close associate of Cotton

Mat her. [HI intonation]
Cf.
(33) We saw go totally ballistic that ravenous horde of angry Tolstoy

scholars. [HI intonation]
16 It is not clear to us how the PP gets into topic position in (31b), in view of the
ungrammaticality of
(i) ∗ We saw into the room an angry horde of Tolstoy scholars run.
We leave this question as an unsolved problem. It is possible that the phenomenon seen
here is related to that of French exceptional case marking. where the subject of an infinitival
cannot appear in situ but can be extracted if it is an interrogative or a clitic pronoun (Kayne
1983, Chapter 5).
(v) Heavy NP Shift derives the illusion that the postposed subject of styl-
istic inversion can be the antecedent of a floated quantifier in the Aux, as
for example in (34):
(34)a. Everyone seemed very hungry today. For example, into the
cafeteria have BOTH gone the two students that I was telling
you about. [HI intonation]
a. From this pulpit have both preached Cotton Mather’s two

closest and most trusted associates. [HI intonation]
By contrast, when the subject is light, as in (35), it cannot be the antecedent

of the floated quantifier, as (36) and (37) illustrate.
(35)a. Both the students have gone into the cafeteria.
b. The students have both gone into the cafeteria.
(36)a. Q: Who went into the cafeteria? A: Into the cafeteria have gone
both (of the) the students, I think.
b. Q: Who went into the cafeteria? ∗ A: Into the cafeteria have both
gone the students, I think.
(37)a. Into the mists of history are quickly disappearing both my

heroes.
b. ∗ Into the mists of history are both quickly disappearing my

heroes.
The evidence thus suggests, once again, that the heavy subject is moving
to the right from [Spec, IP], while the light subject is in situ in VP.
There are several other differences between LI and HI that do not
involve the subject NP directly:
(vi) HI but not LI allows long extraction of the XP from a tensed

complement.
(38)a. ∗ Into the room I claim/believe walked Robin.
b. ∗ Into the room I claim/believe/expect t will walk Robin.
c. ∗ From this pulpit I claim/believe/expect t will preach Robin.

(39)a. Into the room I claim/believe/expect ti will walk . . . ROBINi !

[HI intonation]
b. From this pulpit I claim/believe/expect t will preach (elo-

quently) . . . ROBIN! [HI intonation]
(40)a. Into the room I claim/believe/expect ti will walk [a ravenous

horde of angry Tolstoy scholars]i . [HI intonation]
b. From this pulpit I claim/believe/expect ti will preach (incoher-

ently) [a series of ravenous Tolstoy scholars]i . [HI intonation]
The key point here is the contrast between (38) on the one hand and (39)–
(40) on the other, pointedly demonstrating the difference in extraction
possibilities that hinges on the lightness or heaviness of the postverbal
NP. Moreover, simple Heavy NP Shift of the subject of the tensed S
unaccompanied by topicalization of the PP is ungrammatical:
(41)a. ∗ I claim/believe/expect [t will walk into the room this minute a

horde of angry Tolstoy scholars].
b. From this pulpit I claim/believe/expect ti will preach

. . . ROBINi !
(42)a. I claim/believe/expect [there will walk into the room this minute
a horde of angry Tolstoy scholars.]
a. I claim/believe/expect [there will preach from this pulpit all

week a series of increasingly angry Tolstoy scholars.]
(vii) Extraction from a subject in the LI (immediate post-verbal) position

is better than extraction from a subject in the HI (VP-final) position.17
17 Note that if the light subject is in situ, the awkwardness of extracting from it must
be due to the fact that it is the logical and not the syntactic subject of the sentence. This
observation recalls the proposal of Culicover & Wilkins (1984) that extraction from the
antecedent of a predicate diminishes acceptability, regardless of the syntactic configur-
ation in which the antecedent appears. This specific effect need not, and apparently is
not, universal, given that extraction from postverbal unaccusative subjects is fine in other
languages such as German and Italian. But the language-specific nature of such restrictions
is unsurprising and well-attested elsewhere; thus, in English, gaps within subjects are only
sanctioned as part of parasitic gap constructions (modulo a limited class of examples noted
in Ross (1967)), while in Icelandic such gaps may occur freely even without coindexed
gaps elsewhere in the clause, as noted in Sells (1984).
(43)a. ?Who did you say that into the room walked offensive friends of
t waving rude signs? [HI intonation]
b. ∗ Who did you say that into the room walked waving rude signs
offensive friends of t? [HI intonation]
∗
c. Who did you say that from this pulpit preached waving rude
signs offensive friends of t?
This difference is consistent with the view that the light subject is in situ
in VP, while the heavy subject is in an adjoined position. It is equally
ungrammatical to extract from a shifted heavy direct object, for example
(Wexler & Culicover 1980).
∗
(44) Whoi did you say that you saw ti yesterday [offensive friends
of tj ]i
(vii) HI but not LI (marginally) allows where.

We begin with the general observation that while a relative PP produces
stylistic inversion, both relative and interrogative where block inversion, as
illustrated in (45)–(52).
(45)a. the place to which Robin went
b. the place where Robin went
(46)a. the place to which went Robin
b. ∗ the place where went Robin
(47)a. the city in which all my relatives live
b. the city in which live all my relatives
(48)a. the city where all my relatives live
b. ∗ the city where live all my relatives
Similarly for interrogative PP vs. where.
(49)a. To which place did Robin go?
b. Where did Robin go?

(50)a. To which place went Robin?
b. ∗ Where went Robin?
(51)a. In which city do all your relatives live?
b. Where do all your relatives live?
(52)a. In which city live all your relatives?
b. ∗ Where live all your relatives?
These facts appear at first sight to be totally mysterious. Notice, however,

that the ungrammatical examples are greatly improved by introduction of
an adverb, an apparent instance of the Adverb Effect (Culicover 1993).
(53)a. This is the city where for the most part live all my relatives.
b. This is the city where for most of the year live all my relatives.
c. ?Leslie asked me where, at that point, had gone the thieves who
had taken my money.
d. ?(Leslie was wondering) where for most of the year live all of
your most favorite relatives.
Significantly, however, there is no improvement unless the postposed

subject is relatively heavy.
(54)a. ∗ This is the city where for the most part lives Robin.
b. ∗ This is the city where for most of the year lives Robin.
c. ∗ (Leslie asked me) where at that point went Robin.
d. ∗ (Leslie was wondering) where for most of the year live your
kids.
The efficacy of the Adverb Effect when there is HI, but not when there
is LI, once again strongly suggests that there are two different structures
for the two constructions. More precisely, it appears that the landing site
for where in HI is the complementizer position or [Spec, CP], producing
a C-t effect that is ameliorated by the Adverb Effect. But apparently there
is no landing site for where in LI. If, as we have suggested, LI involves

movement of a PP into [Spec, IP], we can explain the absence of a landing
site by positing that where is not a PP in the required sense, but an NP since
NPs – for reasons that of course need to be explained – fail to participate
in LI.18,19
To sum up, we have observed several distinct syntactic phenomena that
support the claim that there are two SI constructions, HI and LI, notably:
• that it is possible to postpose only a constituent corresponding to a
heavy subject in the cases of various kinds of nonfinite complements
(see (i)), gerundives (see (ii)), configurations of control (see (iii)), and
complements of a perception verb (see (iv));
• that only the heavy antecedent of a floated quantifier can postpose (see
(v));
• that only the heavy subject of an embedded complement can postpose
when a constituent of that complement has been fronted to the matrix
(see (vi));
• that it is less acceptable to extract from a postposed heavy subject than
from the subject of LI (see (vii)).
Furthermore, as we have already pointed out in §1, only heavy NPs can
appear in what have standardly been taken to be instances where a PP
18 Note, for example, that where can be a tough subject, in spite of the fact that PPs are
typically ruled out as subjects of tough predicates: ??∗ In which room would be easiest to
hold the exam?, but Where would be easiest to hold the exam?.
19 We can only consider briefly here the restriction that allows only PPs in [Spec, IP] to
trigger locative inversion. Suppose that NP movement paralleled PP movement to create
inversion structures. Consider the following contrast:
(i)a. Robin ran into the room.
b. Into the room ran Robin.
c. [e] I [[ran Robin] into the room]
(ii)a. Robin ran the race.
b. ∗ The race ran Robin. [on the same reading as a.]
c. [e] I [Robin [ran the race]]
The (a) examples show the consequence of moving the subject into [Spec, IP]. The (b)-
examples show what happens when we move the non-subject out of VP into [Spec, IP].
The approximate underlying structures are given as the (c)-examples, where (ic) follows
(1) in the text.
We assume that a DS subject in [Spec, VP] will be assigned an agentive θ-role.
inversion subject has been raised. In each case, we have strong evidence
from the possible appearance of adverbial material intervening between
the postposed subject and the verb that these subjects are in adjoined po-
sitions outside the VP, just where heavy-shifted objects appear in HNPS.
Thus, all of these differences follow from the view that HI is derived by a
generalization of Heavy NP Shift to subjects, while in LI the subject is in
situ in VP. It is crucial to note that the role of heaviness here is not simply
that of preventing light postposed subjects from appearing to the right of
adverbial or other material, for were this the case, it would be possible to
interpret what we are calling heavy inversion simply as the occurrence of
heavy NPs in the LI structure, followed by heavy shift of the postverbal
NP. Such an interpretation of the LI/HI distinction is however precluded
by the fact that there can be no light NP inversion at all in the nonfinite
cases noted, which would be inexplicable under the assumption that the
heavy NPs which do appear in these constructions originated in postverbal
position, as we are claiming for the light NPs. That is, on the assumption
that both HI and LI correspond to the structure
(55) [IP PPi I [VP V NPsubj ti . . . ]]
there would be no way to block the possibility of the PP raising to root

subject position in the LI as well as the HI cases, in spite of the fact that as
noted above such apparent ‘raising’ cases are restricted to HI, and similarly
for the various other examples we have given of inversion possibilities
allowed only for HI constructions.20
20 One additional piece of evidence that the postverbal NP is in situ comes from the
fact that superiority effects: LI, unlike wh-movement, does not produce strong Superiority
violations, as shown in (i)–(iv):
(i)a. Who did what?
b. ∗ What did who do?
(ii)a. Who came out of which room?
b. ∗ Out of which room did who come?
c. (?)Out of which room came who?
(iii)a. Who did you claim t did what?
b. ∗ What did you claim who did t?

4. C ONCLUSION
If the arguments presented in the preceding discussion are on the right

track, it is necessary to reassess the data standardly cited by syntacticians
offering accounts of English stylistic inversion, so that such accounts are
to be expected to correctly predict the well-formedness status of inversion
just in case the subject can be light. In support of this claim, we have
presented evidence from weak crossover phenomena that preposed PPs
in (light) inversion are genuine subjects, rather than topicalized constitu-
ents, and then provided several strands of evidence, involving intervention
effects, infinitival and gerundive complements and associated control phe-
nomena, perception verb complementation, quantifier float, and a variety
of other phenomena, which clearly sort the possible appearance of post-
verbal heavy NPs from those which allow light NPs. The data which have
in the past been used to argue that fronted PPs are subjects which can
undergo Raising are a further case in point, since as we showed earlier
these examples are only well-formed when the post-verb NP is heavy. The
simplest account of these effects, we believe, is to recognize the possibility
that subjects as well as objects can heavy-shift. Such a conclusion in turn
raises several important theoretical questions.
– What licences the trace in subject position when HI heavy-shifts the
subject? The general impossibility of heavy-shifting subjects of finite
clauses would lead one to conclude that the resulting subject traces
(iv)a. Who did you claim came out of which room?
b. ∗ Out of which room did you claim who came?
c. Out of which room did you claim came who?

(Cf. ?Which man saw who?)
Again we see that the PP in stylistic inversion displays subject rather than topic or wh-
moved properties: (iic) is essentially comparable in acceptability to (iia), while (iib),
containing a wh-moved PP, displays the strong unacceptability of a classic Superiority
Effect violation. An anonymous reader writes that some speakers find it difficult to perceive
the intended difference between (ivb) and (ivc), although to our ears it is quite sharp. let us
replace who by how many people:
(v)a. ∗ Out of which room did you claim how many people came?
b. Out of which room did you claim came how many people?
In our judgment this move strengthens the Superiority effect in (ivb) to the point that the
sentence is virtually uninterpretable but leaves (ivc) unchanged.
are not properly governed, giving rise to an ECP effect. But as we

have noted earlier, reducing the that-t effect to the ECP is not entirely
straightforward (see note 15).
– Why is HI as well as LI incompatible with an overt object? In the case
of LI, it seems reasonable to take this property as a reflection of the
restriction of LI to unaccusative verbs, which of course do not take a
direct object in addition to their surface subject; but why should the
same restriction carry over to HI, whose derivational history should
make it irrelevant whether or not an object is present? On the contrary,
it is standardly assumed that no examples of stylistic inversion are
possible with direct objects:
(56)a. A bunch of teenagers in funny hats had put some gum into
the gas tank of our motorcycle.
b. ∗ Into the gas tank of our motorcycle had put some gum a
bunch of teenagers in funny hats.
We believe that any full discussion of this point must take into account
the fact that, although awkward, there are examples of HI containing
direct objects which we believe to be grammatical:21
(57)a. In the backyard were quietly sunning themselves a group

of the largest iguanas that had ever seen in Ohio.
b. The economist predicted that at that precise moment

would turn the corner the economics of half a dozen
South American nations.22
c. In the laboratory were dying their various horrible deaths

the more than ten thousand fruit flies that Dr. Zapp had
collected in his garden over the summer.
d. Outside in the still upright hangar were heaving deep sighs

of relief the few remaining pilots who had not been chosen
to fly in the worst hurricane since hurricanes had names.
21 As above, we indicate with the notation a major prosodic juncture. Such junctures
appear in what we take to be acceptable utterances of these examples.
22 Unquestionably, turn the corner is at least semi-idiomatic. Nonetheless, the fact that
this idiomaticity is preserved under passivization (e.g., The corner was finally turned on
July 10, when the Ostrogoth economy finally emerged from its deep recession) indicates
that the corner is indeed an internal syntactic argument of the verb, which can therefore
Our analysis predicts that such examples should exist; what remains at
issue is the distinction between cases such as (57) on the one hand vs.
(56b) on the other. We note that the direct objects in the examples in
(64) are not referential. This fact suggests that what allows such cases
is that the verb phrases are thematically intransitive, that is, no θ-role
is assigned to the direct object. Sun oneself means ‘to sun’, turn the
corner in this case is an idiom that means ‘improve’, die a horrible
death means die horribly, and heave a sigh means ‘sigh deeply’. Pre-
cisely why the absence of an object θ-role allows inversion to occur is
a question for future research.
– Finally, why does true stylistic inversion – that is, LI – seem, bey-
ond its pragmatically presentational impact, to be restricted to verbs
which can be interpreted as expressing either motion to a point or
maintenance of a particular physical orientation at some location?
It is well beyond the scope of the present paper to provide detailed discus-
sion of these issues. In view of the evidence presented above, however, we
believe that there is good reason to reassess much of the literature devoted
to inversion constructions and to treat stylistic inversion proper as a far
more restricted phenomenon than it has previously been considered.
REFERENCES
Arnold, Jennifer, Thomas Wasow, Anthony Losongco, and Ryan Ginstrom. 2000. ‘Heavi-
ness vs. Newness: The Effects of Complexity and Information Structure on Constituent
Ordering’, Language 76, 28–55.
Beermann, Dorothee, David Leblanc, and Henk van Riejmsdijk (eds). 1997. Rightward
Movement, John Benjamins, Amsterdam.
Bresnan, Joan. 1977. ‘Variables in the Theory of Transformations’, in Peter W. Culicover,
Thomas Wasow, and Adrian Akmajian (eds.), Formal Syntax, Academic Press, New
York, pp. 157–196.
Chomsky, Noam. 1995. The Minimalist Program, MIT Press, Cambridge, MA.
Coopmans, Peter. 1989. ‘Where Stylistic and Syntactic Processes Meet: Locative Inversion
in English’, Language 65, 728–751.
Coopmans, Peter. 1992. ‘Review of Rochemont and Culicover’, English Focus Construc-
tions and the Theory of Grammar, Language 68, 206–210.
Culicover, Peter. 1991. ‘Topicalization, Inversion and Complementizers in English’, in OTS
Working Papers: Going Romance and Beyond, University of Utrecht, Utrecht.
Culicover, Peter W. 1993. ‘Evidence Against ECP Accounts of the that-t Effect’, Linguistic
Inquiry 24, 557–561.
hardly be regarded as exhibiting intransitive, much less unaccusative argument structure

here. Similar observations hold for (57d), e.g., After the crisis brows were mopped, deep
sighs of relief were heaved and then everyone got back to work.
Culicover, Peter W. and Wendy Wilkins. 1984. Locality in Linguistic Theory, Academic
Press, New York.
Fodor, Janet Dean. 1978. ‘Parsing Strategies and Constraints on Transformations’, Lin-
guistic Inquiry 9, 427–473.
Hoekstra, Teun and René Mulder. 1990. ‘Unergatives as Copular Verbs: Locational and
Existential Predication’, The Linguistic Review 7, 1–79.
Jackendoff, Ray and Peter Culicover. 1972. ‘A Reconsideration of Dative Movement’,
Foundations of Language 6, 197–219.
Johnson, Kyle. 1985. A Case for Movement, Ph.D. dissertation, MIT Press, Cambridge,
MA.
Kathol, Andreas and Robert D. Levine. 1992. ‘Inversion as a Linearization Effect’, in
Proceedings of NELS 23, Department of Linguistics, University of Massachussetts,
Amherst, pp. 207–221.
Kayne, Richard. 1983. Connectedness and Binary Branching, Foris, Dordrecht.
Kayne, Richard. 1994. The Antisymmetry of Syntax, MIT Press, Cambridge, MA.
Levin, Beth and Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax-Lexical
Semantics Interface, MIT Press, Cambridge, MA.
Levine, Robert D. 1989. ‘On Focus Inversion: Syntactic Valence and the Role of a
SUBCAT List’, Linguistics 27, 1013–1055.
Pesetsky, David. 1982. ‘Complementizer-Trace Phenomena and the Nominative Island
Condition’, The Linguistic Review 1, 297–343.
Reuland, Eric. 1983. ‘Governing -ing’, Linguistic Inquiry l4, 101–136.
Rizzi, Luigi. 1997. ‘The Fine Structure of the Left Periphery’, in Handbook of Generative
Syntax, Kluwer, Dordrecht, pp. 281–338.
Rochemont, Michael and Peter W. Culicover. 1990. English Focus Constructions and the
Theory of Grammar, Cambridge University Press, Cambridge.
Ross, John R. 1967. Constraints on Variables in Syntax, Ph.D. dissertation, Massachussetts
Institute of Technology; published as Infinite Syntax!, Ablex, Norwich, NJ, 1986.
Sells, Peter. 1984. Syntax and Semantics of Resumptive Pronouns, Unpublished Ph.D.
dissertation, University of Massachusetts, Amherst.
Stowell, Tim. 1981. Origins of Phrase Structure, Unpublished Ph.D. dissertation, Mas-
sachussetts Institute of Technology.
Wexler, Kenneth and Peter W. Culicover. 1980. Formal Principles of Language Acquisi-
tion, MIT Press, Cambridge, MA.
Received 1 March 1996

Revised 17 August 2000
Department of Linguistics
Ohio State University
222 Oxley Hall
1712 Neil Avenue
Columbus, OH 43210-1298
U.S.A.
[email protected]
[email protected]
View publication stats

Stylistic Inversion in English A Reconsideration

Uploaded by

Copyright:

Available Formats

Stylistic Inversion in English A Reconsideration

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stylistic Inversion in English A Reconsideration

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Stylistic Inversion In English: A Reconsideration

Article in Natural Language & Linguistic Theory · May 2001

Peter W Culicover Robert Levine

SEE PROFILE SEE PROFILE

Ellipsis in Hybrid Type-Logical Categorial Grammar View project

Languages & linguistics View project

The user has requested enhancement of the downloaded file.

STYLISTIC INVERSION IN ENGLISH: A RECONSIDERATION

Natural Language & Linguistic Theory 19: 283–310, 2001.

We begin by adducing evidence in §2 that in LI the fronted PP is a

derivation of HI does not involvement movement of the subject to the right,

Frequently cited evidence that the PP in stylistic inversion is a subject is

(6)a. Terry imagined a bunch of gorillas walking into the room.

b. Into the room Terry imagined a bunch of gorillas walking.

c. Into the room Terry imagined t walking a bunch of gorillas.

d. Into which room did Terry imagine a bunch of gorillas walking?

e. Into which room did Terry imagine walking a bunch of gorillas.

b. ?Into the room Terry thought about a bunch of gorillas walking.

d. ?Into which room did Terry think about a bunch of gorillas

b. On the wall seemed to be hanging a picture of Robin.

Regardless of the particular details, it turns out that Raising cannot be

(10)a. Into the room appeared to be walking slowly a very large

b. Into the room walked Robin slowly.5

c. ∗ Into the room appeared to be walking Robin slowly.

(11)a. Slowly into the room walked Robin boldly.

b. ∗ Slowly into the room appeared to walk Robin boldly.

(12)a. Into the room singing walked Robin slowly.

b. ∗ Into the room singing appeared to walk Robin slowly.

(i)a. Under the table is a good place to put the beer.

b. Under the table rolled the beer slowly.

b. ∗ Under the table seemed to roll the beer slowly.

(13)a. Whoi appears to hisi mother [ti to be a genius]?

b. ?Whoi is hisi mother grilling ti obsessively? (WCO).

c. ??Whoi does hisi mother think [ti is a genius]? (WCO).

d. ?To whomi did hisi students give ti a teaching award?

b. Into every dogi ’s cage peered itsi owner. (Stylistic inversion, no

b. Next to none of the winning dogsi stood itsi owner. (Stylistic

d. ∗ ?Next to none of the winning dogsi did itsi owner stand.

(16)a. In every dogi ’s cage hung itsi collar.

b. ∗ In every dogi ’s cage hung on a hook itsi most attractive and

The relevant point is highlighted in (14): the quantified NP within the

(ii)a. In no dogi ’s cage hung itsi collar.

3. L IGHT AND H EAVY I NVERSION

Adverb, but a focused or heavy NP can, when the verb is unaccusative;

(18)a. Into the room walked Robin

b. Into the room walked Robin carefully.

c. ∗ Into the room walked carefully Robin.

d. Remember Robin? Well, into the room walked carefully,

The pattern is comparable to that of Heavy NP Shift in VP, as illustrated in

(19)a. Carefully I addressed Robin.

b. I addressed Robin carefully.

c. ∗ I addressed carefully Robin.

d. I addressed carefully . . . ROBIN!

e. I addressed carefully the students in the class who had heard

f. I addressed the students in the class who had heard about

It is important to note that the intended judgments are difficult if not

This “HI-intonation” is the intonation of example (18d) and all of the

(21)a. ∗ In the room slept Robin.

b. Into this room we saw run a ravenous horde of angry Tolstoy