Some Notes on Bounding

Cedric Boeckx
(Harvard University)

Boeckx, Cedric. (2007). Some Notes on Bounding. Language Research 43.1,

35-52.

In this paper I argue for a specific way to understand successive cyclic

movement by showing that (i) the conceptualization of successive cyclicity I
examine requires a ban on movement that is too short, and (ii) the ban re-
quired is the one that is empirically superior to recent alternative ways of de-
fining lower bounds on movement. Empirical arguments come from the
domains of applicative and psych constructions.

Keywords: anti-locality, applicative, locality, psych-verb, successive cyclicity.

1. Introduction

Much research within the Minimalist Program tries to offer a better under-
standing of phenomena and generalization that research of the past 30 years
has firmly established. One such phenomenon is successive cyclic movement,
the fact that movement steps have an upper bound, that movement cannot be
"too long" (see Chomsky (1973) and much subsequent work). In this paper I
offer an argument in favor of a specific conception of successive cyclicity
within the minimalist program on the basis of considerations having to do
with lower bounds on movement steps, that is, the idea that movement cannot
be "too short." Specifically, I argue for a version of successive cyclic movement
that minimizes chain links (Takahashi 1994, Boeckx 2003, among others) be-
cause that version requires a ban on movement internal to the projection from
which movement originates. I show on the basis of data from applicative and
psych constructions that this ban on movement is superior to alternative ways
of defining lower bounds on movement.

2. Successive Cyclic Movement

There is substantial empirical evidence for the existence of successive cyclic

movement in natural languages. That movement indeed proceeds in short
steps (alternatively, that chains consists of short links) can be seen on the basis
of various tests, for both A- and A-bar movement. Consider the following
36 Cedric Boeckx

binding facts in the case of A-movement. Castillo, Drury, and Grohmann

(1999: 94) discuss the following paradigm as evidence for successive cyclic A-
movement.

(1) a. Johnjseems to Mary [t'~ y to appear to himselfi [(tt) to be [(ti) happy]]J

b. *Mary seems to Johnj to appear to himselfi to be happy
c. *Maryk seems to Johnj [t':' to appear to himselfi [(t:') to be [(tk) happy]]]

Standard assumptions about binding tell us that the binding of the reflexive in
(la) is unproblematic since John has raised from its base position over the re-
flexive to the specifier of to appear and then subsequently raised to its surface
position. Thus we understand the reflexive to be locally bound by virtue of the
trace/copy in the intermediate position (indicated by .f). (lb), on the other
hand, is ruled out by virtue of a kind of blocking effect since Mary, by hy-
pothesis, has raised through the specifier of to appear as in (lc). Thus, typical
binding requirements could be said to rule out (lb) on the assumption that the
intermediate movement really takes place.
That A-bar movement also proceeds in short steps can be seen from exam-
ples like (2).

(2) a. [Which pictures ofhimselfi/j] does Johnj think ~ that Bi1lj bought
b. Who said that Johnj thinks that Bi1lj bought pictures of himself'i/j

Movement of the wh-phrase in (2a) brings the anaphor to a position where it is

c-commanded by John but not by Bill (position indicated by .f).
Additional arguments for successive cyclic movement can be constructed on
the basis of Quantifier Float data (under Sportiche's 1988 influential analysis
of Q-stranding, and McCloskey's 2000 eXtension of it to the A-bar domain).
(Data in (3) are from standard English. Data in (4) come from West Ulster
Irish English.)

(3) a. All the boys seem to appear to like ice cream

b. The boys seem all to appear to like ice cream
c. The boys seem to appear all to like ice cream
d. The boys seem to appear to all like ice cream

(4) a. What all did you get for Christmas

b. What did you get all for Christmas
c. What all did John say that Peter ate for breakfast
d. What did John say that Peter ate all for breakfast
e. What did John say all that Peter ate for breakfast
Some Notes on Bounding 37

3. Conceptions of Successive Cyclicity

Although the facts around successive cyclic movement are clear, the under-
lyjng cause is much less so. Consider McCloskey's (2002: 184-185) telling
quote:

"If locality conditions are at the heart of syntax (as increasingly seems to be the
case), then the existence of apparently unbounded dependencies like [long-distance
wh-movement] represents an anomaly. Since Chornsky (1973), it has come to be
widely believed that the apparently distant connection between antecedent and vari-
able position in such cases is mediated by a sequence of more local connections. (... )
In all variants of this core idea, the specifier of CP is one of the crucial left-peripheral
positions establishing these connections (... ) Movement is always at least this local.
(... ) A much harder question is what makes this true - what property of language-
design determines that this is how things work."

The conceptual problem posed by successive cyclicity has been around ever
since the advent of the rrrinimalist program and its insistence on movement as
last resort. Put simply, there doesn't seem to be obvious features triggering in-
termediate steps of movement.
Various featural options have been tried, but I agree with McCloskey (2002:
186) that all of them boil down to "spurious," or "pseudo-"features (Q, Op,
Wh, etc.) - movement-triggering features optionally present on intermediate
landing sites, whose presence is required neither by lexical requirements or by
considerations of interpretability.
It has sometimes been suggested (see, e.g., Hornstein (2001: 119» that the
checking of 4>-features should be implicated in the formulation of successive
cyclic movement on the basis of so-called wh-agreement phenomena in lan-
guages like Chamorro (see Chung (1998». However, this conception of wh-
agreement (agreement triggered by successive cyclic movement of the wh-
phrase) appears to rest on a factual misunderstanding (noted by various ex-
perts on the languages exhibiting overt 'wh' -agreement). As noted in Boeckx
(2003: 57), building upon observations in Chung and Georgopoulos (1988),
Georgopoulos (1991), and Chung (1988), for Palauan and Chamorro (see also
Rackowski and Richards (2003) on Tagalog, Pearson to appear on Malagasy,
and Finer (1997) on Selayarese), 'wh-agreement' is only indirectly conditioned
by overt wh-movement. That is, although wh-movement induces a morpho-
logical change on intermediate verbs (verbs along the wh-movement path), the
morphological change refers to a special kind of agreement between the verb
and the clause from which the wh-phrase has been extracted. In particular,
when overt wh-movement takes place, the verbs along the way to the ultimate
[+wh] SpecCP bears the morphology they would bear if the complement
38 Cedric Boeckx

clause out of which wh-movement took place were extracted. 1,2

Boskovic (2002) provides additional arguments that intermediate landing sites
are not checking sites. The clearest piece of evidence comes from the generaliza-
tion going back to Lobeck (1995) and Saito and Murasugi (1990), according to
which ellipsis is licensed in the complement of a head taking part in Spec-Head
agreement. Boskovic argues that the unacceptability of (5) is puzzling if Spec-
Head agreement (feature checking) takes place in the intermediate C position.
By contrast, if no checking takes place, (5) is excluded on a par with (6b, d).

(5) *John met someone, but I don't know who Peter said (that) John met

(6) a. John's talk about the economy was interesting, but Bill's talk about
the @conomy was boring

I I here set aside differences among the relevant languages pertaining to whether 'wh-
agreement' on intermediate verbs is obligatory (Tagalog) or optional (Chamorro) de-
pending on the nature of the moving wh-phrase (D-linked or not). I also set aside the
fact that in the relevant languages, the moving wh-phrase directly affects the morphol-
ogy of the most deeply embedded verb. The latter point is largely orthogonal to the issue
of successive cyclicity.
2 The only instance of genuine long-distance wh-agreement (where morphology co-varies
with the featural specification of the wh-word, not the clause containing it) I am aware
of is found in Kinande. As Schneider-Zioga has illustrated in a series of papers (Schnei-
der-Zioga 2000, 2002, 2004; see also Rizzi 1990: 55), the language expresses the noun
class and number of the wh-phrase on the complementizer (lfocus-marker) immediately
adjacent to the wh-phrase (i), as weH as on complementizers along the wh-path (ii) ..
(i) a. Iyond! yO Karnbale alangIra
Who.l that. 1 Kambale saw
'Who did Karnbale see'
b. AbahI Bo Kambale alangIra
Who.2 that.2 Karnbale saw
c. EkIhI kyO Karnbale alangIra
What.7 that.7 Karnbale saw
'What did Kambale see'
d. EBIhI ByO Kambale alangIra
What.8 that.8 Karnbale saw
(ii) EkIhIj kyOj Yosefu a-kabula [,mp nga-kyoj rip a-kalangira ~ ]]
what FOC J. wonders if -FOC agr.sees
'What does Yosefu wonder ifhe sees?'
To analyze Kinande in a way consistent with the claim that there is no wh-agreement triggered
on intermediate landing sites, Boeckx (2004) foHows Davies' (2003) analysis of long-
distance wh-questions in Madurese and Javenese. Davies argues that apparent long-
distance wh-movement are instances of iterative prolepsis, as schematized in (iii). (See
also McCloskey (2002: 199) for an independent argument in favoring of allowing a
strategy like (iii) based on Irish.)
(ill) Whi [Opi C .... <Opi> [Opi C ... <Opi>]]
(It is interesting to note that Kinande wh-extractions, like Madurese and Javanese wh-questions,
have a cleft-like nature.)
Some Notes on Bounding 39

b. *A single student came to class because the SffiGem thought that it

was important
c. John met someone, but I don't know who John met
d. *John believes that Peter met someone, but I don't believe that Pet€f
met someone

On the basis of arguments of this sort I conclude that successive cyclic move-
ment steps are not feature-driven. This conclusion is also embraced by Chom-
sky in his recent writings (Chomsky 2000, 2001, 2004, 2005a-b). There, suc-
cessive-cyclic movement is forced by the PlC (11). To avoid being trapped in a
phase, an element must move from phase-edge to phase-edge. The mechanism
that ensures this is given in Chomsky (2000):

(7) At the end of phase HP, the head H may freely be assigned an EPP-
feature, forcing overt movement of a phrase into SpecHP

There are two important concepts in Chomsky's formulation, both of which

are actually characteristic of all versions of successive cyclic movement since
Chomsky (1973): it is not a forced option (cf., 'freely assigned'), and it happens
for "EPP"-reasons. As Lasnik (2001) has emphasized, the EPP in Chomsky's
system is not a feature in the technical sense of the term, the way, say, [wh] or
[<1>] features are (i.e., things that are being valued or 'checked'). For instance,
the [EPP] feature can never be checked in situ. It just seems that "the EPP ( ... )
demands that certain functional heads have a specifier" (Lasnik 2001a). As
such, it is more adequate to speak on an EPP-property, since the way to satisfy
this property is quite distinct from valuing features at a distance ('checking').
Needless to say, the EPP so construed is no more than a descriptive charac-
terization. As Epstein and Seely (2002: 86) aptly note, the EPP is "a represen-
tational macro-tree description, demanding explanation in terms of lexical
features and their mode of combination."
In addition to appealing to an unexplained property of the grammar (EPP),
Chomsky's technical implementation of successive cyclic movement takes it to
be an optional movement, which, from a minimalist point of view, just begs
the question. 3
The most principled account of successive cyclicity I know of is Takahashi's
1994 analysis, based on Chomsky and Lasnik's (1993: 90) Minimize Chain
Links Principle, which requires that each chain link be as short as possible.4

3 For additional arguments against phase-based derivations in Chomsky's sense, see

Boeckx and Grohmann (to appear) and H Ko (2005).
4 In this paper I set aside the interesting, but orthogonal, question of whether Takahashi's
representational version of the Minimize Chain Links is superior to Boskovic's (2005)
recent derivational rendering of the same intuition that successive cyclic movement is a
40 Cedric Boeckx

Takahashi claims that the Mirrimize Chain Links Principle captures the cy-
clic (local) nature of movement while avoiding the pitfalls of spurious features
posited in intermediate sites to drive successive cyclic steps.
Takahashi's core idea is that successive steps are taken not in order to check
some feature in intermediate sites, but simply due to the requirement that
chain links must be minimized (a reflex of economy). In Takahashi's (1994)
terms, each link of a chain must be as short as possible. The requirement forces
any element X undergoing movement of type Y to stop at every position of
type Y on the way to its final landing site independently of feature checking. It
is worth noting that Takahashi assumes that the relevant operation underlying
movement is Form Chain. In so doing, Last Resort is relevant only to the for-
mation of a chain, not links of a chain. In other words, formation of a chain
must have feature-checking motivation, but formation of chain links needs not
Building upon a suggestion made in Manzini (1994), Boeckx (2003: 8) modi-
fied Takahashi (1994), and argued that a moving element adjoins to the
maximal domain of each head on its way to its ultimate landing site (see also
Boskovic (2002: 186), Boskovic (2005), Fox and Lasnik (2003), Fox (2000),
Richards (2002) for similar claims; see also the percolation mechanisms in
frameworks like HPSG, or in Neeleman and van de Koot (2002)).5 The moti-
vation for this idea was twofold. First, ever since the principle of the cycle was
proposed, the number of cyclic nodes (originally restricted to S and NP) in-
creased, and just about every node became a cyclic node (see already Williams
(1974)). This strikes me as the simplest assumption. Second, the notion of
movement type (AI A-bar) has no clear status in current syntactic theorizing,
which makes it very difficult to define in a non-arbitrary way what a landing
site of the relevant type is.

property of the movement operation, and is not due to the checking of (pseudo-)features
in intermediate positions. As far as this paper is concerned, any version of successive
cyclicity that encodes this intuition and that requires a lower bound on movement will do.
Here I rely on Takahashi's version because it is fairly we11-known, and because the need
for a lower bound on movement once we adopt Takahashi's version has been fleshed out
in the literature (Boskovic 1994).
5 One may say that under Boeckx's version of successive cyclic movement every maximal
projection is a phase (as argued for in Epstein and Seely (2002) and Boskovic (2005)),
but it is important to note that were one to do so, one would be using a very different no-
tion of phase from the one used by Chomsky. For Chomsky, every phase induces a PLC
effect. If every projection were a phase, no extraction would be possible, as the comple-
ment of any phrase would have to move to the edge of that phrase/phase, a movement
step that would count as too local under the version of 'anti-locality' that I entertain in
this paper. For this reason I refrain from using the term 'phase' in connection with the
conception of successive cyclicity argued for here.
Some Notes on Bounding 41

4. On Movement That Is 'Too Short'

I would like to offer a new argument for a Takahashi-style conception of

successive cyclicity. My argument is based on Boskovic's (1994) observation
that some condition is needed to prevent the Minimize Chain Links Principle
from forcing a phrase in an adjoined position to keep adjoining the same node.
Put differently, some condition is needed to prevent chain links from being too
short. (Although Boskovic was referring to Takahashi's version of successive
cyclicity, his point applies with equal force to Boeckx's amendment discussed
above.)
Perhaps the first explicit proposal to the effect that chain links cannot be too
short comes from Murasugi and Saito (1995), who formulated (8).

(8) A chain link must be at least of length 1

A chain link from A to B is of length n iff there are n "nodes" (X, X',
or XP, but not segments of these) that dominate A and exclude B

The empirical reason for positing (8) is given in (9).

(9) *I think that hp John, hP <John> likes Mary]]

Lasnik and Saito (1992) observe that if (short) subject topicalization (adjunc-
tion to SpecIP) were allowed, (10) would be predicted to be on a par with (11),
contrary to fact.

(10) *JOhni thinks that himself; ti likes Peter

(11) JOhni thinks that himself; Peter likes 4

Based on such facts, Lasnik and Saito (1992) conclude that movement from
within one projection, more precisely for them, movement from SpecIP to the
IP-adjoined position, must be disallowed. To explain this, Murasugi and Saito
(1995) propose (8), which they argue is reducible to an economy guideline, viz.
the ban on superfluous steps (possibly related to Chomsky's (1986b) ban on
vacuous projections).
Boskovic (1994) (see also Boskovic (1997: 184, n28» argues that (8) has con-
siderable motivation. In particular, he notes that (8) rules out adjunction of X
to its own XP and substitution of X to SpecXP (situations that Chomsky
(1995: 321) referred to as 'self-attachment').
More recently, Kayne (2005) has independently proposed to rule out
movement of the complement of X to the specifier position of XP, and sug-
gests that this condition could be derived in feature-checking terms if upon
42 Cedric Boeckx

Merge the maximal set of matching features must be checked.

Another instance of the same idea is provided by Bobaljik (2000), who ar-
gues, after examining a wide range of facts, that V-to-I movement never takes
place in situation (12a), but becomes possible in (12b).

(12) (a) IF (b) IF

~ ~
I VP I XP
I ~
V X VP
I
V

The desired result can be deduced from (8) if we take the label of X to be a
copy of X (Chomsky 2001, Har1ey 2004, Boeckx 2004). Accordingly, (8)
would forbid movement of V to I in (12a), since a copy of V (VP) is already in
a local relation with I (it is the complement of 1). By contrast, no such local
relation exists prior to V-movement in (12b), due to the presence of XP, which
renders movement possible.
In a similar vein, Abels (2003) notes that, though mobile in general, IFs may
not move and strand their selecting CPs.

(13) *Frank saw a play that «a play> was long and boring> yesterday
<a play> was long and boring

(14) *John is a fool is believed that <John is a fool>

(15) *John is a fool, Mary told herself that <John is a fool>

Likewise, Abels shows that, though mobile in principle (in some languages),
VPs never strand VD. Based on this, Abels proposes the following principle:

(16) Given a, the head of a phase,

always *[a t]

Given the requirement that domains called phases only allow movement of
their specifiers (not their complements) (see Chomsky (2000); the PlC in (7)
above), (16) in effect amounts to a version of (8); i.e., no phrase can be both
specifier and complement of the same head (Always: *[aP l3i [a ti]].
In sum, there seems to be good reasons to adopt some version of (8). Impor-
tantly, those reasons are independent of the fact that something like (8) is
needed to guarantee the adequacy of any version of a Takahashi-style ap-
Some Notes on Bounding 43

proach to successive cyclicity.

In the remainder of this paper I would like to argue that (8) is in fact supe-
rior to an alternative conception (Grohmann 2000,2003) of what the lower
bound on movement is. Interestingly, as we will see shortly, Grohmann's con-
ception is stricter than (8), and it is more than what a Takahashi-style analysis
requires. My goal is to show that Grohmann's conception is also less
empirically adequate than (8). So, what we will end up with is an empirical
argument for (8), which in turn becomes a conceptual argument for a
Takahashi-style analysis, since (8) is precisely what the latter requires, nothing
more, nothing less.

5. 'Too Short' and Too 'Too Short'

Based on a wide range of considerations, Grohmann (2000, 2003) formu-

lates (17).6

(17) Anti-locality hypothesis

Movement must not be too local

Grohmann conjectures that movement is too local if an element K has two

occurrences within a given domain <x. 7 For Grohmann, <X ranges over the-
matic ("VP"), inflectional ("IP"), and discourse-related ("CP") domains. Ac-
cordingly, no movement cannot take place within, say, the verbal domain
(unless resumption takes place, see note 4). I will now show that such a ban is
too strong. Specifically, I will show that movement within the thematic do-
main is required. That movement can be either feature-driven, or not (inter-
mediate movement step).
The evidence against Grohmann's conception of (anti-)locality comes from

6 Grohmann embeds his ban on movement that is too short (lower bounds on movement) into a
theory of locality. This is clearly the most desirable move: bounding and locality should go hand
in hand.
7 Grohmann notes that just like movement that is "too long" can be "salvaged" (impressionisti-
cal1y speaking) by resumption (i), movement that is too short can, too. For justification of the
derivations in (ii), see Grohmann's own work.
(i) ? Which woman did you claim that Peter met the man who saw <whlch 'l.~ her
(ii) a. John [yp<John> likes-<Jelm> -+ himself]
b. [xpDiesen Mann, [xp <diesefl: MaBfl> -+ den kenne ich nicht] German
This.acc man that-one. ace know I not]
'This man, I don't know him'
For alternative views on resumption (including reflexivity) that don't require Grohmann's spe-
cific version of Anti-locality, but instead rely in agreement and case, see Boeckx (2003) and
Hornstein (2001), respectively.
44 Cedric Boeckx

two constructions: applicatives and psych predicates. I discuss each in turn.

5.1. Applicatives

Empirical considerations have led several researchers to claim that we need

to distinguish at least two kinds of double object structures, a low-
applicativel dative structure like (18), and a high-applicativel dative structure
like (19) (see PyIkkanen (2002), Anagnostopoulou (2003), McGinnis (2001), J-
E Lee (2004), Y Jeong (2004), Miyagawa and Tsujioka (2004)). (I have modi-
fied the structures in (18) and (19) slightly, eliminating structural details that
are irrelevant to the issue at hand.)

(18) [yp Subj V [yp 10 V DO]]

(19) [yp Subj V [yp 10 V [yp V DOm

The latter line of research has reached a conclusion that is puzzling at first.
McGinnis (2001) in particular has provided compelling empirical evidence
suggesting that languages employing the structure in (19) correspond to what
previous research had called symmetric languages, that is, languages that treat
both objects alike for a variety of syntactic purposes such as passivization, cli-
ticization, etc. British English is often described as one such language, on the
basis of (20).

(20) a. Mary was given a candy

b. A candy was given Mary

By contrast, languages that use a structure like (18) correspond to asymmetric

languages, such as American English. Witness the contrast in (21).

(21) a. Mary was given a candy

b. *A candy was given Mary

McGinnis's conclusion amounts to this surprising statement: the closer the

objects are structurally upon base merge, the more asymmetrically they behave
Conversely, the more distant they are upon base merge, the more symmetri-
cally they behave. From a phrase structural perspective this is surprising. One
would expect that greater distance between the two objects would increase the
asymmetry between them. But natural languages just do not seem to work that
way. From the point of view of locality (specifically, Relativized Minimal-
ity/intervention effects), it is easy to understand why 10 is always passivizable
(pace independent factors that might block such movement, see Y Jeong
Some Notes on Bounding 45

(2004)); 10 always start off higher than DO, i.e., closer to T". What remains to
be explained is how DO can circumvent the intervening 10.
Ura (1996), McGinnis (1998, 2001) and Anagnostopoulou (2003) converge
on the idea that passivization of DO is rendered possible as a result of DO
moving to the edge of the projection hosting 10. Once that movement step has
taken place, DO is higher than 10 and can be passivized.
Note that this movement step must be blocked in so-called asymmetric lan-
guages, since these languages lack DO-passivization. J-E Lee (2004) (see also
Y Jeong (2004)) suggests an interesting way of doing precisely that. She as-
sumes that Ura, McGinnis, and Anagnostopoulou are correct in saying that
passivization of DO in a language that adopts (19) allows movement of DO to
the edge of the VP hosting 10. At that point, DO is higher than 10 and can be
passivized, as schematized in (22).

(22) DO T" [vp t'IO [vp V t]]

To prevent this derivation in languages making use of (18), J-E Lee appeals to
anti-locality to block movement of DO to the edge of the VP hosting 10. The
key factor here is that the VP hosting 10 also hosts DO. Since 'anti-locality'
considerations rule out movement of the complement of X to the specifier of
XP, DO cannot become higher than 10. Therefore, only 10 can be passivized,
as illustrated in (23)-(24).

(23) 10 T' [vp t V DO]

(24) *DO T" [vp t'IO V t]

Notice that J-E Lee's reasoning only goes through if anti-locality only applies
within one projection, not within a given domain, such as VP / vP, as in Groh-
mann (2000, 2003). If movement within vP (the thematic domain) were
banned, one would not be able to distinguish between symmetric and asym-
metric languages. In fact, Grohmann predicts all languages to be of the
asymmetric type, since the movement step that obviates minimality, 'Leap-
frogging' in McGinnis' terms, would violate Grohmann's notion of anti-
locality.

5.2. Psych Verbs

An argument similar to J-E Lee's can be made on the basis of psych verb
data, although this time, we'll see that the intervention-obviating step is fea-
ture-driven.
As is well-known, Belletti and Rizzi (1988) distinguish among three types of
46 Cedric Boeckx

psych-verbs on the basis of Italian data:

(25) Gianni teme questo

Gianni fears this

(26) Questo preoccupa Gianni

This worries Gianni

(27) a. A Gianni piace questo

To Gianni pleases this
b. Questo piace a Gianni
This pleases to Gianni

The three verb classes are Subject Experiencer NP verbs like temere ('fear');
Object Experiencer verbs like preoccupare ('worry'), and verbs like piacere
('please'), which allow PP-Experiencer subjects.
Much attention has been devoted to ·Object-Experiencer verbs, in particular
to the backward binding facts like (28).

(28) a. rumors about herself; worried Maryi

b. rumors about hisi mother upset everybody/nobodYi
c. the assertion that she; was unfit to serve worried no female candidate;

To account for such facts, Belletti and Rizzi argued that the subjects of object-
experiencer psych verbs start off as complements of V. This assumption al-
lowed for a straightforward account of "backward binding" facts. If (surface)
subjects of object-experiencer psych verbs start off as complements of V, the
(surface) object experiencer is able to bind the subject prior to movement (al-
ternatively, backward binding can be treated as a result of reconstruc-
tion/interpretation of the copy of the subject left by movement). The basic
derivation is given in (29).

(29) hP [rumors about herself]i T" fxp worrie<1j [ypMary [;, ti]]]]

To this day, Belletti and Rizzi's analysis of backward binding remains the con-
ceptually most appealing. 8

8 It has sometimes been suggested that the importance of the backward binding facts has
been overestimated. For example, Landau (2003) suggests that the binding examples are
instances of logophoric binding, which do not require the strict licensing conditions that
regular binding does. Landau's point may be correct for examples like (28a) (picture-NP
cases more generally), but it is hard to see how logophoricity could be involved in in-
stances of binding by universal or negative quantifiers (28b-c).
Some Notes on Bounding 47

Despite its conceptual appeal, Belletti and Rizzi's analysis was heavily criti-
cized, especially by Pesetsky (1995). The major criticism raised by Pesetsky is
that there are good reasons to believe that object experiencer psych-verbs are
not to be represented as unaccusative structures. The gist of Pesetsky's analysis
is that the surface subjects of object experiencer psych verbs behave themati-
cally like external arguments (causers).
Pesetsky offers several arguments in favor of this conclusion. Perhaps the
most compelling one is the fact that in contrast ot unaccusatives (30), Object
Experiencer verbs passivize (31).

(30) a. The lamp sat on the desk

b. *The desk was sat on by the lamp

(31) a. The article in the Times angered Bill

b. Bill was angered by the article in the Times

If, as is now standardly assumed, passivization boils down to absorption of an

external theta role, one must conclude on the basis of (31) that there is an ex-
ternal theta-role in object experiencer verbs.
Here I will follow Hornstein and Motomura (2002) and amend Belletti and
Rizzi's (1988) proposal in a way that preserves its conceptual appeal while ac-
commodating Pesetsky's claim. Hornstein and Motomura revise (29) along the
lines of (32).

(32) b [rumors about herself]i T" [vp t/v"-worrie~ [vpMary [~ti]]]]

According to (32), the surface subject starts off as the complement of V (where
it receives the target theta-role; see Pesetsky (1995: 58», a position that will
allow a Belletti-Rizzi style account of backward binding. On its way to SpecTP,
the subject stops by SpecvP and collects an external theta-role, which allows
Hornstein and Motomura to capture Pesetsky's results. (Horlli!tein and Mo-
tomura assume that the specifier and complement of the same projection are
equidistant, which allows rumors about herself to cross Mary on its way to
SpecvP without violating Minimality.)
A derivation like (32) is possible as soon as one allows for movement into
theta-position, an assumption that I will simply adopt here (for extensive dis-
cussion, see Hornstein (1999), Boeckx and Hornstein (2004) and references
therein).
As Hornstein and Motomura note, the intermediate step of movement tar-
geting SpecvP obviates the intervention effect of the experiencer for movement
to SpecTP. But the question now arises as to why the experiencer doesn't
block movement of the innermost object to SpecvP. (This question did not
48 Cedric Boeckx

arise in the context of applicatives, as movement of the lower object targeted a

position that was within the same projection as the position already occupied
by the higher object.)
Hornstein and Motomura simply assume that the specifier and complement
of the same projection are equidistant, which allows rumors about herself to
cross Mary on its way to SpecvP without violating Minimality, but once Equi-
distance is adopted we lose the account we had for asymmetric applicative
structures. If the members of a projection were equidistant, DO-passivization
would be possible in (24). ,
Instead of appealing to Equidistance (an odd minimalist notion anyway), I
would like to capitalize on the fact that cross-linguistically experiencers are
never agents. For example, in Icelandic no quirky subjects (typically experi-
encers) bear an agent theta-role. I therefore claim that by bearing an experi-
encer theta-role, the relevant NP cantlot check the agent theta-role. (The ban in
question is easy to encode in featural terms, but I confess that I would like to
derive this ban from something deeper. I haven't succeeded so far. So, pending
deeper understanding of the content of theta-roles, I leave this ban as a stipula-
tion.) If experiencer and agent theta-roles are incompatible, and if, as is stan-
dard (see Boeckx and Jeong (2004), Rizzi (2004), Starke (2001)) Minimality is
defined in terms of feature matching), the experiencer will not block move-
ment of the innermost object to SpecvP as it doesn't have the right kind of
matching feature to interfere. 9
If the modified version of Hornstein and Motomura's analysis proposed
here is tenable, derivations like (32) provide another argument against Groh-
mann's ban on movement inside the thematic domain (with no co-occurring
resumption), this time based on theta-driven movement.

9 Similarly, if we assume that PPs don't (match in features, hence don't) intervene (witness John
seems to Mary [t to be smart}]), movement of the innermost object in piacere class could proceed in
one fell swoop, as in (i).
(i) h1' DDi T" [~ V'- V; [vp PP-ID [~t;llJ]
This would explain why in the piacere verb class no agentivity is detected.
(I assume that when the PP-experiencer is in subject position, we are actually dealing not with a
genuine Pp, but with an NP associated with an overt case-marker. Put differently, the optionality
witnessed in the context of piacere-type verbs is the result of the ambiguous full preposition/case-
marker status of the a associated with the experiencer.)
Some Notes on Bounding 49

6. Conclusion

The double object and psych verbs data demand a notion of anti-locality
that confines the effect of the condition to a single projection (as opposed to
involving the notion of 'domam), something like (33).

(33) The complement of X cannot move anywhere within XP.

Equivalently:

(34) Movement internal to a projection counts as too local, and is banned.

(33)-(34) is exactly what a Takahashi-style analysis of successive-cyclic move-

ment requires. The picture that emerges is the following: Long-distance
movement can't be too long, chain links must be kept short. In particular, if
Boeckx (2003) is correct, a moved element must adjoin to each and every
maximal projection on its way to its final landing site, except the projection it
originates from (in that case movement would be too short). This convergence
of conceptual and empirical results establishes a certain complementarity be-
tween the notions of locality (upper bound on movement steps/chain links)
and anti-locality (lower bounds on movement steps/chain links), which is
highly desirable from the theoretical perspective of the minimalist program.

References

Abels, Klaus. (2003). Successive-cyclidty, Anti-locality, and Adposition Stranding. Doctoral

dissertation, University of Connecticut.
Anagnostopoulou, E1ena. (2003). The Syntax of Ditransitives. Berlin: Moutonl de
Gruyter.
Belletti, Adriana and Luigi Rizzi. (1988). Psych verbs and theta-theory. Natural Lan-
guage and Linguistic Theory 6,291-352.
Bobaljik, Jonathan D. (2000). The rich agreement hypothesis in review. Ms., McGill
University.
Boeckx, Cedric. (2003). Islands and Chains. Amsterdam: John Benjamins.
Boeckx, Cedric. (2004). Bare syntax. Ms., Harvard University.
Boeckx, Cedric and Kleanthes K. Grohmann. (to appear). Phases in perspective. Syntax
Boeckx, Cedric and Norbert Hornstein. (2004). Movement under control. Linguistic
Inquiry 35, 431-452.
Boeckx, Cedric and Youngmi Jeong. (2004). The fine structure of intervention in syn-
tax. In Issues in Current Linguistic Theory: A Festschrift for Hong Bae Lee, Chungja
Kwon and WonbinLee, eds., 83-116.
50 Cedric Boeckx

Boskovic, Zeljko. (1994). D-structure, 8-criterion, and movement into 8-positions. Lin-
guistic Analysis 24, 247-286.
Boskovic, Ze1jko. (1997). The Syntax of Non-finite Comp!ementation: An Economy Approach.
Cambridge, Mass.: MIT Press.
Boskovic, Zeljko. (2002). A-movement and the EPP. Syntax 5, 167-218.
Boskovic, Zeljko. (2005). On the locality of move and agree. Ms., University of Con-
necticut.
CastilIo, Juan Carlos, John Drury, and Kleanthes K. Grohmann. (1998). Merge-over-
move and the EPP. In University of Maryland Working Papers in Linguistics 8, 63-103.
University of Maryland, College Park.
Chomsky, Noam. (1973). Conditions on transformations. In AfestschrififOr Morris Halle, S.
Anderson and P. Kiparsky, eds., 232-286. New York: Holt, Rinehart, and Wmston.
Chomsky, Noam. (1986). Barriers. Cambridge, Mass.: MIT Press.
Chomsky, Noam. (1995). The Minimalist Program. Cambridge, Mass.: MlT Press.
Chomsky, Noam. (2000). Minimalist inquiries: the framework. In Step by Step, R. Mar-
tin, D. Michaels and J. Uriagereka, eds., 89-155. Cambridge, Mass.: MIT Press.
Chomsky, Noam. (2001). Derivation by phase. In Ken Hale: A Life in Language, M.
Kenstowicz, ed., 1-52. Cambridge, Mass.: MIT Press.
Chomsky, Noam. (2004). Beyond explanatory adequacy. In Structures and Beyond, A.
Belletti, ed., 31. Oxford: Oxford University Press.
Chomsky, Noam. (2005a). Three factors in language design. Linguistic Inquiry 36, 1-22.
Chomsky, Noam. (200Sb). On phases. Ms., MlT.
Chomsky, Noam and Howard Lasnik. (1993). Principles and parameters theory. In
Syntax: An International Handbook of Contemporary Research, J. Jacobs, A. von Ste-
chow, W Stemefeld and T. Vennemann, eds., 506-569. Berlin: de Gruyter.
Chung, Sandra and Carol Georgopoulos. (1988). Agreement with gaps in Chamorro
and Palauan. In Agreement in Natural Language: Approaches, Theories, and Description,
M. Barlowand C. A. Ferguson, eds., 251-267. Stanford, CaIi£: CSLI.
Davies, William. (2003). Extreme locality in Madurese wh-questions. Syntax 6, 237-259.
Finer, Dan L. (1997). Contrasting A-bar dependencies in Selayarese. Natural Language
and Linguistic Theory 15, 677-728.
Fox, Danny. (2000). Economy and Semantic Interpretation. Cambridge, Mass.: MlT Press.
Fox, Danny and Howard Lasnik. (2003). Successive-cyclic movement and island repair:
The difference between sluicing and VP-ellipsis. Linguistic Inquiry 34, 143-154.
Georgopoulos, Carol. (1991). Syntactic Variables: Resumptive Pronouns and Binding in Pa-
lauan. Dordrecht: Kluwer.
Grohmann, Kleanthes K. (2000). Prolific Peripheries: A Radical View from the Left. Doc-
toral dissertation, University of Maryland.
Grohmann, Kleanthes K. (2003). Prolific Domains. Amsterdam: John Benjamins.
Harley, Heidy. (2004). Merge, conflation, and head movement: The first sister principle
revisited. In Proceedings of NELS 34, K. Moulton and M. Wolf, eds., University of
Massachusetts, Amherst: GLSA.
Some Notes on Bounding 51

Hornstein, Norbert. (1999). Control and movement. LinguisticInquiry 30,69-96.

Hornstein, Norbert. (2001). Move! A Minimalist Theory of Construal. Oxford: Blackwell.
Hornstein, Norbert and Mitsue Motomura. (2002). Psych verbs, theta roles and recon-
struction. In Proceedings of 2002 LSK International Summer Conference Vol. IT, 39-58.
Linguistics Society of Korea.
Jeong, Youngmi. (2004). The landscape of ditransitives. Ms., University of Maryland.
Kayne, Richard S. (2005). Some notes on comparative syntax, with special reference to
English and French. In The Oxford Handbook of Comparative Syntax, G. Cinque
and R. Kayne, eds., 3-69. Oxford: Oxford University Press.
Ko, Heejeong. (2005). Syntactic Edges and Linearization. Doctoral dissertation, MlT.
Landau, ldan. (2003). The locative syntax of experiencers. Ms., Ben Gurion University.
Lasnik, Howard. (2001). A note on the EPP. Linguistic Inquiry 32, 356-362.
Lasnik, Howard and Mamoru Saito. (1992). Move a. Cambridge, Mass.: MIT Press.
Lee, Ju-Eun. (2004). Ditransitive Structures and (Anti-)locality. Doctoral dissertation, Har-
vard University.
Manzini, Maria Rita. (1994). Locality, minimality, and parasitic gaps. Linguistic Inquiry
25,481-508.
McCloskey, James. (2000). Quantifier float and wh-movement in an Irish English. Lin-
guistic Inquiry 31,57-84.
McCloskey, James. (2002). Resumption, successive cyclicity, and the locality of opera-
tions. In Derivation and Explanation in the Minimalist Program, S. D. Epstein and T.
D. Seely, eds., 184-226. Oxford: Blackwell.
McGinnis, Martha. (1998). Locality in A-movement. Doctoral dissertation, MIT.
McGinnis, Martha. (2001). Variation in the phase structure of applicatives. In Linguistics
Variation Yearbook #1,105-146.
Miyagawa, Shigeru and Takae Tsujioka. (2004). Argument structure and ditransitive
verbs in Japanese. Journal of East-Asian Linguistics 13, 1-38.
Murasugi, Keito and Mamoru Saito. (1995). Adjunction and cyclicity. In Proceedings of
WCCFL 13, 302-317, Stanford, CA: CSLI.
Neeleman, Ad, and Hans van de Koot. (2002). The configurational matrix. Linguistic
Inquiry 33, 529-574.
Pearson, Matt. (in press). The Malagasy SUbject/topic as an A-bar element. Natural
Language and Linguistic Theory.
Pesetsky, David. (1995). Zero syntax. Cambridge, Mass.: MIT Press.
Pylkkanen, Liina. (2002). Introducing Arguments. Doctoral dissertation, MIT.
Rackowski, Andrea and Norvin Richards. (2003). EPP and extraction: A Tagalog case
study. Ms., MIT.
Richards, Norvin. (2002). Very local A-bar movement in a root-first derivation. In Deri-
vation and Explanation in the Minima/ist Program, S. D. Epstein and T. D. See1y, eds.,
227-248. Oxford: Blackwell.
Rizzi, Luigi. (1990). Relativized Minimality. Cambridge, Mass.: MIT Press.
Rizzi, Luigi. (2004). Locality and the left periphery. In Structures and Beyond, A. Belletti,
52 Cedric Boeckx

ed., 223-251. Oxford: Oxford University Press.

Schneider-Zioga, Patricia. (2000). Anti-agreement and the fine structure of the left pe-
riphery. University of California Irvine Working Papers in Linguistics ,Vol. 6.
Schneider-Zioga, Patricia. (2002). The case of anti-agreement. Presented at AFLA 8,
MIT.
Schneider-Zioga, Patricia. (2004). Anti-agreement, shortest chain and minimal-
ity: the syntax of dislocated subjects. Ms., UC Irvine.
Sportiche, Dominique. (1988). A theory of floating quantifiers and its corollaries for
constituent structure. Linguistic Inquiry 19, 425-449.
Starke, Michal. (2001). Move Dissolves into Merge: A Theory of Locality. Doctoral disserta-
tion, University of Geneva.
Takahashi, Daiko. (1994). Minimality of Movement. Doctoral dissertation, University of
Connecticut, Storrs.
Ura, Hiroyuki. (1996). Multiple Feature-checking. Doctoral dissertation, MIT.
Williams, Edwin. (1974). Rule Ordering in Syntax. Doctoral dissertation, MlT.

Cedric Boeckx
Department of Linguistics
Boylston Hall, Thlrd Floor
Harvard University
Cambridge, MA 02138
USA
E-mail: [email protected]

Received: April 1, 2007

Revised version received: June 18, 2007
Accepted: June 25, 2007