Skip to content
Publicly Available Published by De Gruyter Mouton December 24, 2015

On the timing of labeling: Deducing Comp-trace effects, the Subject Condition, the Adjunct Condition, and tucking in from labeling

  • Željko Bošković EMAIL logo
From the journal The Linguistic Review

Abstract

The article provides a uniform account of a number of locality effects, in particular, the Subject Condition, the Adjunct Condition, Richards’s (2001) tucking in effect, and the full Comp-trace paradigm, including (in addition to the basic cases) relative and extraposed clauses, the impossibility of short subject topicalization, French que-qui alternation, and the effect of wh-movement on agreement in languages like Kinande. The account is based on a proposal that there is a difference in the timing of labeling between the basic case where a head and a phrase merge and the case where two phrases merge, as well as a particular labeling-based approach to antilocality, which has rather different empirical effects from the previous approaches to antilocality.

1 Introduction

The goal of this article is to provide a uniform account of a number of locality effects, in particular: (a) the Subject Condition (1); (b) the Adjunct Condition (2); (c) the full Comp-trace paradigm, including the basic case in (3), with the improvement with intervening adverbs (4) and relative clauses (6), as well as the unacceptable null C case with extrapositions like (5), French que-qui alternation, and the effect of wh-movement on agreement in languages like Kinande, which will be discussed in some detail (with a comparison of subject wh-movement in Kinande and Kaqchikel as well as subject wh-movement and object wh-movement in Kinande); (d) the impossibility of short subject topicalization (8) and zero subject relatives (7); and (e) Richards’s (2001) tucking in effect.

(1)

?*Who did [friends of t] see Mary?

(2)

?*What did you fall asleep [after Peter had bought t]?

(3)

*Who do you think that t will leave Mary?

(4)

Who do you think that under no circumstances t would leave Mary?

(5)

*Who is it likely t likes Mary?

(6)

the stone that t broke the window

(7)

*John picked up the stone t broke the window.

(8)

*John, t likes Mary.

It will be shown that all these cases can be accounted for in a unified manner in the labeling framework, given the proposal made here that there is a difference in the timing of labeling between the basic case where a head and a phrase merge and the case where two phrases merge, and a particular labeling-based approach to antilocality. I will first discuss the ingredients of the analysis, namely antilocality and the labeling framework (Chomsky 2013, 2015; see also Collins 2014; Epstein et al. 2014; Rizzi 2013; Saito 2014; Shlonsky 2014, among others), and then turn to the account of the cases noted above.

2 Ingredients

2.1 Antilocality

It is standardly assumed that there is an upper bound on movement – movement cannot be too long. A number of authors have argued that movement also cannot be too short. (The ban on movement that is too short is referred to as antilocality in Grohmann 2003). There is a battery of arguments for antilocality in the literature, some of the relevant works being Bošković (1994, 1997, 2005, 2014), Saito and Murasugi (1999), Ishii (1999), Abels (2003), Grohmann (2003), Ticio (2005), Boeckx (2005), Jeong (2006). These works do not all adopt the same definition of antilocality (see Grohmann 2011 for an overview). Thus, Bošković (2005, 2014) argues that Move must cross at least one full phrase (not just a segment). One of the effects of this definition of antilocality is that it blocks complement-to-Spec movement, which deduces Abels’s (2003) generalization that complements of phasal heads cannot move, one of the arguments for antilocality offered in the literature. Another argument, from Bošković (2013b), concerns the unacceptability of extraction of NP-adjuncts, noted by a number of authors (e.g. Huang 1982; Chomsky 1986; Culicover and Rochement 1992) and illustrated by (9). Assuming that these adjuncts are NP-adjoined and that DP is a phase, the derivation in (9) is ruled out by antilocality (movement to SpecDP, required by the PIC, crosses only a segment; the reader is referred to the works cited above for a number of additional arguments for antilocality).

(9)

*From wherei did John meet [DPti [NP [NPgirls] ti]]?

In this article I will argue for a particular view of antilocality that is adjusted to Chomsky’s (2013) system, which allows unlabeled projections.

2.2 Labeling

Chomsky (2013) proposes a theory of labeling where in the case where a head and a phrase merge, the head projects (more precisely, provides the label for the resulting object). [1] Chomsky suggests two ways of implementing labeling in the case where non-minimal projections (i.e. phrases) are merged: through prominent feature sharing or traces, where traces are basically ignored for the purpose of labeling. (10) illustrates the former: when what is merged with interrogative C (actually CP) both the wh-phrase and the CP have the Q-feature; what is projected (i.e. determines the label of the resulting object) then is the Q-feature. [2]

(10)

I wonder [CPwhati [C’C [John bought ti]]]

The latter case is illustrated by (11), with the relevant part of the derivation given in (12).

(11)

Whati do you think [CPt’i [C’that [John bought ti]]]?

(12)

v [VPthink [?what [CPthat [John bought ti]]]]

The timing of labeling here has rather interesting consequences for antilocality, a derivational ban on movement that is too short, which I continue to assume: Move must cross at least one phrase. Chomsky assumes that there is no feature sharing between the intermediate, declarative complementizer that and the wh-phrase that passes through its edge in (11) (which essentially follows Bošković 2002, 2007, 2008). Consequently, labeling through feature sharing is not an option here. The embedded clause then cannot be labeled at the point of movement of what to its edge, as indicated in (12) by using ?-notation. When v is merged, what moves away. The element merged with the CP now being a trace, it is ignored for the purpose of labeling, hence ? is labeled as CP after movement of what. Only at this point the status of t’i in (11) can be determined as the Spec of CP. At the point of movement (i.e. (12)), ? is not a CP, it is simply undetermined regarding the issue in question. Since there is no labeling before movement, at the point of movement there is no crossing of a CP phrase even if that projects after wh-movement, with t’i in SpecCP.

To make the issue clearer, I adopt the following definition of antilocality (cf. Bošković in press), adjusted to the framework that allows unlabeled objects, the intuitive idea here being that movement does not cross B if it involves merger with B. (In effect, (13) requires crossing of a labeled projection.)

(13)

Antilocality: Movement of A targeting B must cross a projection distinct from B (where unlabeled projections are not distinct from labeled projections).

Antilocality is still satisfied in (12) because the movement that targets vP crosses VP. Were VP to be missing in (12), movement of what to vP would violate antilocality. Note that I assume that labeling can take place as soon as it can be accomplished (this will be refined below), otherwise it would not be possible to label structures where both relevant elements move.

Bošković (in press) shows that this approach to labeling/antilocality deduces the Complex NP Constraint, i.e. the ban on extraction out of NPs headed by nouns modified by clauses. Bošković (in press) takes Chomsky’s proposal that vP functions as a phase as indicating that the highest projection in a thematic domain functions as a phase. He argues that there is no theta-marking nP in the traditional Noun Phrase (TNP) of the object nominal in (14) (and more generally complex NPs), which makes NP the highest thematic projection here, hence a phase. As a result, movement in (14) must target the edge of CP and the edge of NP. Since there is no feature sharing between that and the wh-phrase that merges with it, the object that is created by their merger is not labeled at the point it is created. In the next step, rumors merges with this object, the resulting object being labeled as NP via the base step of the labeling algorithm (the head-phrase merger case). The wh-phrase then merges with the NP. This movement, however, violates antilocality. [3]

(14)

??Whati did you hear ti [NPrumors [?ti that [IPJohn bought ti]]]?

At any rate, what is important for our purposes is that under the conception of antilocality argued for here movement must cross a labeled projection (though we will see in Section 4 that this is not always enough). [4]

I will show that the labeling system and antilocality enable us to deduce a number of locality effects, providing a unified treatment for all of them. An important ingredient of the account will be the proposal that labeling via the base step of the algorithm (when a head and a phrase are merged) can be done immediately, while labeling in the case of a merger of two non-minimal projections takes place when the structure is sent to the interfaces.

In Chomsky (2013), labeling in the cases where a head and a phrase are merged (the base step) is done rather differently from the cases where two phrases undergo merger: labeling of the base step occurs via minimal search (MS), the same operation as Agree Closest, a syntactic mechanism falling under minimal computation. MS does not determine the label when two phrases merge. Given the difference, I argue for a timing difference in labeling. I will refer to the proposal below as TOL (timing of labeling). Since the labeling of the base step is done through essentially a syntactic mechanism, it takes place when the relevant configuration is created. Labeling in the case of merger of two phrases occurs when the relevant structure is sent to the interfaces, given Chomsky’s assumption that unlabeled objects are uninterpretable. The issue will be discussed more extensively in Section 6, where it is noted that without TOL, it is not even possible to determine the points of spell-out (basically, phase determination requires TOL); Section 6 will also provide another way of deducing TOL, where labeling in the case of a head-complement merger takes place for a strictly syntactic reason, namely, subcategorization, the underlying assumption being that satisfying subcategorization requires that the element with the requirement to take a complement project (otherwise, there would be no head-complement relation here), see here Chomsky (2000). [5] Pending this discussion, the reader can simply take TOL at face value. The following sections will be devoted to providing arguments for TOL; in particular, I will show that TOL enables us to deduce in a unified manner a number of previously unrelated locality effects, given the labeling approach to antilocality. Thus, Section 3 will show that the labeling system based on these two mechanisms deduces CED effects, i.e. both the Subject Condition and the Adjunct Condition. Section 4 shows that the labeling system in question also deduces Richards’s (2001) tucking in effect, and Section 5 shows that the same holds for the full range of Comp-trace effects and a number of related phenomena, like the effect that wh-movement has on agreement in languages like Kinande. Section 6, which is also the conclusion, returns to the issue of the theoretical underpinnings of TOL.

3 CED effects

3.1 The Subject Condition

The system provides a rather natural account of the traditional Subject Condition, i.e. the ban on extraction out of subjects located in SpecIP. [6] The ban is illustrated by (15).

(15)

*I wonder whoi [friends of ti] left.

Since subjects are phases (being DPs, and DP are phases), whatever moves out of a subject must first move to its edge. Both the merger of who with the subject DP and the merger of the subject DP with the IP involve merger of two phrases. Given TOL, the result of the mergers is labeled only when the structure is sent to the interfaces, not at the point of merger (the subject is given in italics). [7]

(16)

… [?2[?1who[DP subject]] [IP I…[vP

The next step involves merger with C, with C, a head, projecting. The wh-phrase then targets CP. The movement violates antilocality (i.e. (13)). After the movement, ?1 is labeled as DP and ?2 as IP (through feature sharing), but that is too late to save the derivation.

(17)

… [CP C [?2[?1 who [DP subject]] [IP I…[vP

The ban on extraction out of subjects is thus deduced. [8]

3.2 The Adjunct Condition

The analysis is extendable to the ban on extraction out of adjuncts, illustrated by (2), if adjuncts are adjoined to complements of phasal heads, i.e. VP and IP. (18) gives the structure for the former case. Given TOL, since adjunct merger involves merger of two maximal projections, its result is not labeled upon merger. [9] (For ease of exposition, all labels in this section will be given at the right-edge brackets.)

(18)

[[[… VP] K ?] vP]

Assuming adjuncts are phases (CPs, DPs, or PPs, all of which have been argued to be phases, see Bošković 2014, in press and references therein), movement out of an adjunct, given in italics in (19), has to target the adjunct, resulting in an unlabeled object for reasons discussed above. Further movement has to target vP, which violates antilocality.

(19)

[[[… VP] [wh-phrase [K(adjunct)] ?]?] vP]

A question, however, arises regarding wh-adjuncts, as in (20).

(20)

How did John leave?

A number of authors have argued that wh-adjuncts that are located in SpecCP are actually base-generated in that position (see e.g. Law 1993; Uriagereka 1988), and Stepanov (2001) argues that wh-adjuncts are merged differently from their non-wh-counterparts due to the presence of the Q-morpheme, which can be implemented as merging them with the Q. In either case, the problem noted above regarding (17) would not arise in (20) (in the latter case, the adjunct would always cross the QP which is created by the Q-adjunct merger).

Both of these approaches assume that wh-adjuncts and their non-wh-counterparts are not base-generated in the same position. On a more speculatory note, this (i.e. assuming such a difference) in itself opens up another possibility, which is that wh-adjuncts are generated as adjuncts, but not as adjuncts to complements of phasal heads, which would allow them to extract. [10]

4 Tucking in

Based on a variety of cases, Richards (2001) shows that in multiple-specifier constructions, after one specifier is created the second specifier has to be created below the exisiting specifier, tucking in under it, not on top of the existing specifier.

Consider (21), a multiple wh-fronting construction from Bulgarian, a language which places all fronted wh-phrases in SpecCP (see Rudin 1988; for ease of exposition I ignore V-movement and assume that the embedded CP is the only intermediate phase in this example, which means that movement to the matrix clause must proceed via the embedded clause SpecCP).

(21)
a.
Kogoikakvojmisliš[CPti tj če [IPPetkotipopita tj]]?
whowhatthink-2sthatPetkoasked-3s

‘Who do you think that Petko asked what?’

b.

cf. *Kakvo kogo misliš [če Petko popita]?

The indirect object wh-phrase is higher than the direct object wh-phrase prior to undergoing movement. As a result, the indirect object wh-phrase must move first to the embedded clause SpecCP (this is the standard account which basically treats ordering of fronted wh-phrases in Bulgarian as a Superiority effect). An additional CP-Spec is then created by movement of kakvo ‘what’. Richards (2001) argues that the additional Spec is created below the original Spec, i.e. by tucking in under the original Spec. As a result, kogo is still higher than kakvo prior to movement to the matrix clause. Kogo then must move first to the matrix SpecCP, with kakvo tucking in under it. The ordering of fronted wh-phrases in Bulgarian thus illustrates Richards’s tucking in effect.

Richards’s tucking in can also be deduced from the labeling system, in particular, it follows quite straightforwardly without any additional assumptions from the current, labeling-based conception of antilocality and TOL.

Consider again (21) in light of the current system. As discussed above, kogo moves first, merging with the CP headed by that. Since, as discussed above, the relevant elements are not involved in feature sharing, the result of the merger is not labeled (recall that labeling is possible here only after the wh-phrase moves away). We then have (22) prior to the movement of kakvo. Notice now that given the conception of antilocality from (13), kakvo can move to the edge of the CP in question only by merging below kogo, i.e. by merging with the CP. Merging with the kogo+CP object is not an option here. Since this object is not labeled merger with this object inevitably violates (13): unlabeled projections being non-distinct from labeled projections, movement to ? in (22) would not involve crossing a projection distinct from ?, hence it would violate (13). This yields Richards’s tucking in effect. The current labeling system in fact provides a new perspective on the tucking in effect. (13) forces movement to target a labeled category: any movement to an unlabeled object will violate (13). In other words, only labeled categories can be targets of movement given (13). In fact, the problem with movements that do not observe the tucking in effect is that they target unlabeled categories. [11]

(22)

[?kogo [CPče [IPkakvo..]]]?

Consider now the tucking in effect in the matrix clause of (21), or more generally for simple questions like (23). Here, since the relevant C is interrogative, the wh-phrase does undergo feature-sharing with the CP that it merges with. While in (22) the CP cannot be labeled until kogo moves away, this is not the case in (23).

(23)
a.
Kogoikakvojtipopita tj?
Whowhatasked-3s

‘Who did she/he ask what?’

b.

cf. *Kakvo kogopopita?

Recall now the proposal made above: labeling via the base step of the algorithm (when a head and a phrase are merged) can be done immediately while labeling in the case of a merger of two non-minimal projections takes place when the structure is sent to the interfaces. This means that the result of merger of C and the IP can be labeled immediately (at the point of merger), while the result of merger of the CP and kogo cannot be labeled immediately. As a result, (23) has the following structure prior to the movement of kakvo.

(24)

[?kogo [CP C-Q [IPkakvo…]]]?

The only way kakvo can move to a labeled category, which is necessary not to violate (13), is if it targets CP, yielding Richards’s tucking in effect for this case too. As far as I can tell, other cases of tucking in that Richards discussed can also be handled in this manner. I conclude therefore that under the view of labeling and antilocality adopted here, Richards’s tucking in effect follows from the labeling mechanism.

The upshot of the above discussion is that the current system yields the requirement that only labeled categories can be targets of movement, which has the status of a theorem – the requirement follows from antilocality, in particular, the labeling-based conception of antilocality adopted here. “Violations” of the tucking in effect involve movement to an unlabeled category. In fact, given TOL, only movement to an object formed by a head-complement merger has a shot at not violating (13): movement to an object formed by a merger of two phrases (i.e. a traditional phrase with a Spec) inevitably violates (13).

It is worth emphasizing the twist that what appeared to be an innocent re-formulation of antilocality within the labeling system brings in here. Antilocality was originally proposed to ban movements that are too short (see Bošković 1994). Under the current, labeling-based conception of antilocality, even long movements can get “caught” by antilocality: no matter how many phrases movement crosses it will violate antilocality if it targets an unlabeled category. In this respect, the current labeling-based conception of antilocality is very different from all the previous ones. As discussed above, this is desirable: by imposing the requirement that movement targets only labeled categories (which has the status of a theorem), the labeling-based conception of antilocality deduces Richards’s tucking in effect.

5 Local subject movements

I now turn to various types of local subject movements, which exhibit rather interesting patterns crosslinguistically. What I refer to as local subject movement here is movement from SpecTP to the CP that immediately dominates the TP in question. In many cases such movement is clearly blocked. However, there are cases where it seems to be allowed. I am not aware of any attempts at a uniform account of all the relevant cases (see for example (3)–(8)). In this section I will show that given TOL, the labeling system can provide a uniform account of the rather complex paradigm pertaining to the domain of local subject movement. However, given the complexity of the patterns in question, some issues will need to be left unresolved or without proper independent support, which I hope to return to in future work.

A number of works have shown that subject movement to SpecCP cannot proceed via SpecTP (see Bošković 2008; Erlewine 2014; Holmberg and Hróarsdóttir 2003; Rizzi 1990; Rizzi and Shlonsky 2007, among many others for various languages). Thus, in many languages, like Kinande, verbal morphology that arises as a result of agreement between T and the subject located in its Spec cannot be present when the subject undergoes wh-movement, which is standardly taken to indicate that wh-subjects do not move via SpecTP, hence they do not license the usual agreement morphology that occurs with subjects located in SpecTP. The effect is illustrated by the following Kinande paradigm, where the usual subject-agreement morphology from (25a) cannot be present under wh-movement (25b). (It is not possible to simply drop the agreement morphology; rather, a different morphological marker appears on the verb, as in (25c). A labeling-based account of this will be provided in Section 5.5.2.)

(25)
a.
Kambalea.langiraMarya.
Kambaleagr.sawMary
b.
*Iyondi yoa.langiraMarya.
who Cagr.sawMary
c.
Iyondi you.langiraMarya.
who Canti-agr.sawMary
(Schneider-Zioga 1995)

Another argument to this effect is provided by several varieties of Italian. Thus, Brandi and Cordin (1989) show that Trentino and Fiorentino have different agreement with postverbal and preverbal subjects. Wh-movement of subjects is necessarily accompanied by postverbal agreement. The relevant pattern is illustrated below with Fiorentino.

(26)
a.
Leragazzel’hannotelefonato.
thegirlscl.3f.plhas.3plphoned
(Campos 1997)
b.
Gl’-hatelefonatodelleragazze
cl.3sghas.3sgtelephonedsomegirls

‘Some girls have telephoned.’

c.
Quanteragazzeglihaparlatoconte?
how-manygirlscl.3sghas.3sgspokenwithyou

‘How many girls talked to you?’

d.
*Quanteragazzelehannoparlatoconte?
how-manygirlscl.3plhas.3plspokenwithyou
(Brandi and Cordin 1989)

Consider also Icelandic (27). (27a) shows that an intervening experiencer blocks agreement with a nominative object, hence the obligatory singular on the matrix verb in (27a). An NP trace does not exhibit this blocking effect, as shown by (27b). However, the blocking effect is still present in (27c-d). If the experiencer in (27c-d) could undergo the same kind of movement to SpecTP it undergoes in (27a) before undergoing wh-movement, the experiencer blocking effect should be voided in (27c-d) since the intervening element would be an NP-trace, just as in (27b). The Icelandic paradigm in question thus also indicates that wh-movement via SpecTP is not possible.

(27)
a.
Þaðvirðist/*virðasteinhverjummanni[hestarnirveraseinir].
EXPLseems/seemsomeman.datthe.horses.nombeslow

‘It seems to some man that the horses are slow.’

b.
MérvirðasttNP[hestarnirveraseinir].
me.datseem.plthe.horses.nombeslow
c.
Hvaðamanniveistþúvirðist/*virðasttwh[hestarnirveraseinir]?
whichman.datknowyouthatseems/seemthe.horsesbeslow

‘To which man do you know that the horses seem to be slow?’

(Holmberg and Hróarsdóttir 2003)

d.
Hverjummundi/??munduhafavirsttwh[hestarnirvera
who.datwould.3sg/would.3plhaveseemedthe.horses.nombe
seinir]?
slow

‘To whom would it have seemed that the horses are slow?’

(Nomura 2005)

What the above discussion indicates is that the configuration in (28), stated in the traditional obligatory labeling system, should be disallowed.

(28)

[CP whi [IP ti [I’]]]

This in fact follows straightforwardly from the labeling system argued for here, given TOL and the labeling-based approach to antilocality. Given TOL, merger of the wh-phrase and TP does not result in immediate labeling in (29). The object in question is then merged with C (interrogative or non-interrogative; recall that the effect in question is found with both interrogative and non-interrogative Cs). The resulting object is labeled as CP via the base step of the labeling algorithm. The wh-phrase then merges with the CP. The movement, however, does not cross a labeled projection, violating antilocality, i.e. (13).

(29)

[? whi [CP [? ti [IP]]]]

5.1 The that-trace effect

The account can be extended to a number of other local subject movements, including the traditional that-trace effect, illustrated by (30). [12] Movement from the IP-edge to the CP-edge in (30) also involves the configuration in (29), hence it violates antilocality (see (31)). [13]

(30)

*Whoi do you think that ti left Mary?

(31)

Whoi do you think [?ti [CPthat [?ti [IPleft Mary]]]]?

The antilocality account also captures the well-known improvement in (32). Browning (1996) and Watanabe (1993) argue that such cases involve CP-recursion while Culicover (1992) argues that they involve a PolP; what is important for our purposes is that there is a phrase between IP and CP, hence movement of the subject from IP to the highest CP (headed by that), which functions as a phase (assuming that the highest clausal projection functions as a phase, as in Bošković 2014 (see also Bošković in press; Wurmbrand 2014), does not violate antilocality. (All this extends to the (improvement of) Comp-trace effects with other complementizers, see Culicover 1992; note also that the rescuing effect here does not depend on inversion, see Browning 1996). [14]

(32)

Leslie is the person whoi I said [?ti that [CP/PolPat no time C/Pol [?ti [IPconsidered running for public office]]]]

(Browning 1996)

What about cases like (34), which do not exhibit a Comp-trace effect?

(34)

Who do you think t left Mary?

There are several possibilities for analyzing such cases given the proposals that were independently made in the literature for embedded non-interrogative clauses which are not introduced by that in English. Thus, such cases can be treated as in Rizzi (2006), namely as involving truncation of the CP+IP structure (confined to clauses with non-overt subjects (IP-internally) in the V-complement position). The issue that arose in (30) then would not arise in (34) since the relevant structure is missing.

A number of authors have argued that that-less embedded clauses in English are quite generally IPs, even in simple examples like John believes Mary left (see Bošković 1997 and references therein). The IP analysis (with a similar restriction regarding the distribution of the option in question) would also easily capture (34). [15],[16]

Pesetsky’s (1992) account on which that-less embedded clauses are CPs headed by a null C that undergoes movement to V opens up another avenue for analyzing (34). There are a number of well-documented cases where head-movement voids locality violations, in particular, by voiding the phasehood of the phrase whose head undergoes movement (see den Dikken 2007; Gallego and Uriagereka 2007; Bošković 2013a, in press for various approaches to the issue). For example, Galician has a rather interesting phenomenon of D-to-V incorporation which voids islandhood effects (see Uriagereka 1988, Bošković 2013a). To illustrate, Galician disallows movement from definite NPs (35a). However, the violation is voided when D incorporates into the verb (35b). Bošković (in press) argues that the source of the definiteness effect is that movement cannot proceed through the Spec of definite DPs. The problem with (35a) is then that v cannot attract the wh-phrase without violating the PIC. As for (35b), Bošković (in press) argues that movement of the phasal head voids the phasehood of DP, as a result of which v can attract the wh-phrase without violating the PIC. [17]

(35)
a.
*edequéniviche [DPoretrato ti]?
andofwhosaw(you)theportrait
b.
edequénjviche-loi[DPti retrato tj]?
andofwhomsaw(you)-theportrait

‘so, who have you seen the portrait of?’

(Uriagereka 1988)

The phase-voiding effect of head movement provides a straightforward account of (34) under Pesetsky’s (1992) null C incorporation analysis. The incorporation voids the phasehood of the CP hence movement need not proceed through the edge of the CP in (34). As a result, the problem that arose in (30) does not arise in (34). (Pesetsky’s analysis thus enables us to provide a uniform treatment of the contrast in (30)/(34) and the contrast in (35), where head movement voids the phasehood of CP in (34) and the phasehood of DP in (35b)). [18]

Another alternative is provided by Bošković’s (2011) account of the contrast between (30) and (34), which is based on Chomsky and Lasnik’s (1977) proposal that (37a) is derived from (36) via deletion of that, as in (37b).

(36)

Mary thinks that John left.

(37)
a.

Mary thinks John left.

b.

Mary thinks that John left.

There is a long-standing line of research going back to Chomsky (1972) where movement out of an island leads to *-marking of the island, with the locality violation being repaired if the *-marked element is deleted in PF, as in Ross’s (1969) cases where movement out of an island is repaired if the island is elided, illustrated by (38) (see Merchant 2001; Lasnik 2001; Hornstein et al. 2003; Fox and Lasnik 2003; among many others, for this line of research).

(38)
a.

*Ben will be happy if Mary fires one of the students, but she didn’t know which studenti Ben will be happy [if she fires ti].

b.

Ben will be happy if Mary fires one of the students, but she didn’t know which studentiBen will be happy [if she fires ti].

Working within this line of research, Bošković (2013a) argues that with PIC/antilocality violations with phase XP, what is *-marked is the head of the phase. (Bošković 2013a treats (35) in this manner too; see Bošković 2013a; Riqueros 2013; Talić 2014 for evidence for this position).

Consider in this light (30). As discussed above, movement of the subject to SpecCP induces an antilocality violation, which under the proposal made in Bošković (2013a) leads to the *-marking of the head of the relevant phase, namely that.

(39)

*Whoi do you think [? ti[CP that* [? ti[IP left Mary]]]]?

In (30), the *-marked element remains in the final PF representation, inducing a violation. However, in (34), the *-marked element is deleted in PF (under Chomsky and Lasnik’s 1977 analysis of that-less clauses), which removes the locality violation. Under this analysis, which follows Bošković (2011, 2013a), the contrast between (30) and (34) receives the same treatment as the contrast in (38). [19]

There are thus several ways of analyzing (34); I leave teasing them apart for future research.

5.2 Extraposition

The current analysis also provides a new perspective on a curious property of extraposed clauses like (40) which has resisted a satisfactory account.

(40)

It is likely/appears (that) John bought a house.

While these extraposed clauses allow object extraction they disallow subject extraction, as noted in Kayne (1984), Stowell (1981), Bošković and Lasnik (2003): compare in this respect object extraction cases in (41)/(43) and subject extraction cases in (42)/(44) (the examples are taken from Bošković and Lasnik 2003: 538). [20] Adjuncts pattern with objects, as in (45)–(46) (see Bošković and Lasnik 2003).

(41)

What is it likely (that) John will read?

(42)

*Who is it likely will read the book?

(43)

Who does it appear that Mary likes?

(44)

?*Who does it appear likes Mary?

(45)

How is it likely [(that) John fixed the car t]?

(46)

How does it appear [(that) John fixed the car t]?

A number of authors have argued that extraposed clauses are not complements but VP-Specs/adjuncts (e.g. Reinhart 1980; Stowell 1981; Bošković 2002, in press). Recall now that the options appealed to above regarding (34) are restricted to clauses in the V-complement position (truncation/bare IPs being possible only in this position). [21] This means that the embedded clause in (47) must be a CP. As a result, the problem that arose in (30) also arises here: the construction is ruled out by antilocality because movement from the IP edge to the CP edge does not cross a labeled projection. [22]

(47)

*Who does it appear t likes Mary?

Traditionally, overt C and null C were assumed to differ with respect to their locality-licensing properties: thus, in the Government and Binding framework it was often assumed that null C counts as a proper governor while complementizer that does not. Capturing the obvious phonological difference between the two in a principled way in syntactic terms, which is needed under this analysis, has proven rather tricky, given that syntax should not even know about the phonological realization of particular lexical items. The issue does not arise under the current analysis. In fact, under the current analysis, in examples like (47) we are dealing with a Comp-trace effect with a null C. There is then no need to posit any difference between that and the null C with respect to syntactic locality under the current analysis; they both raise a locality problem for subject wh-movement.

5.3 The ban on short subject topicalization

The above account also captures the ban on short subject topicalization in English, i.e. the impossibility of the local topicalization option for the subject in (48).

(48)

*I think that Johni, [IP ti likes Mary].

Lasnik and Saito (1992) provide a number of arguments that local/vacuous subject topicalization is not possible. To cite only one argument here, they note that if short subject topicalization were allowed we would expect that, as in (49), John and himself can be coindexed in (50), which is not the case.

(49)

Johni thinks that himselfi Peter likes.

(50)

*Johni thinks that himselfi likes Peter.

The account of the impossibility of subject extraction from the context in (29) can be straightforwardly extended to the ban on short subject topicalization. Given TOL, when John merges with IP the resulting object cannot be labeled immediately. The head that hosts topicalization then enters the structure. When the subject moves to merge with the head in question the movement violates antilocality for the same reason it does in (29).

(51)

*I think that Johni[? ti[IP likes Mary]].

5.4 The ban on short zero subject relatives

The above account provides a new perspective on the distribution of that in relative clauses. A well-known puzzle with relative clauses is that they do not display the that-trace effect. [23]

(52)

the stone Opi that ti broke the window

From the current perspective, the key to the lack of the that-trace effect in relative clauses lies in the optionality of that in (53)–(54).

(53)

the stone that Mary threw

(54)

the stone Mary threw

Kayne (1984) (see also Bošković 1997) observes the contrast in (55) regarding the possibility of a resumptive pronoun, which would be surprising if the relative operator were located in the same position in both constructions.

(55)
a.

*The book Op I was wondering whether I would get it in the mail.

b.

The book Op that I was wondering whether I would get it in the mail.

I will therefore assume that the relative Op is not located in the same position in relative clauses with and without that. In (54), the relative operator is located in the Spec of a relative-clause dedicated projection, which I will refer to as RelP, which is obligatory in all relative clauses. The CP headed by that, on the other hand, is present optionally; it is present when that is present (with Op in its Spec under the assumption that Op must move to the edge of the relative clause). (53)–(54) then have the structures in (56)–(57) respectively. [24]

(56)

the stone [CP Opi that [RelP [Mary threw ti]]]

(57)

the stone [RelP Opi [Mary threw ti]]

This immediately explains the lack of the that-trace effect in (52). As before, subject merger with IP does not result in immediate labeling. However, in contrast to (30), movement of the subject to the Spec of that does not violate antilocality in (58). In fact, the RelP rescues the derivation from the that-trace/antilocality effect here in the same way that the CP/the intervening adverb do in (32).

(58)

the stone [CP Opi that [RelP [?ti [IPbroke the window]]]]

Evidence for this analysis is provided by the impossibility of short zero subject relatives, noted by Bresnan (1972). While that is optional with object relatives under consideration, it is obligatory with subject relatives.

(59)

John picked up the stone that broke the window.

(60)

*John picked up the stone broke the window.

This follows straightforwardly under the current analysis. In contrast to (58), where the operator moves to CP (which is a phase here), in (61), the operator can only move to SpecRelP (which is a phase here as the highest clausal projection). This movement, however, violates antilocality.

(61)

*the stone [RelP Opi [?ti [IPbroke the window]]]

Under this analysis, zero subject relatives are expected to be possible in languages where subject movement to IP is not obligatory (where that is not obligatory in the first place). Pesetsky (1982) and Bošković (1997) show that this is indeed the case, as illustrated by (62).

(62)

Chi’e faccendaOpi tocca a noi ti.

‘This is a matter (that) does not concern us.’

(15th century Italian, Rizzi 1990)

5.5 Agreement under wh-movement

5.5.1 Kaqchikel

The account of the English paradigm discussed above gets an interesting confirmation from a Kaqchikel paradigm discussed in Erlewine (2014). Erlewine observes that there is different morphology in Kaqchikel depending on whether or not subject moves to SpecIP. As in a number of other languages, a subject moving to SpecCP is not allowed to pass through SpecIP (i.e. the usual morphology that accompanies subjects that move to SpecIP cannot be present in that case), which can be captured as discussed above. Significantly, Erlewine shows that when there is a phrase between IP and CP, a wh-subject can pass through SpecIP on its way to SpecCP (i.e. the morphology that accompanies subject movement to SpecIP is then present). This is exactly what is expected. Due to the presence of this phrase, as in (32) and in contrast to (29), subject movement from IP to CP does not violate antilocality in (63). [25]

(63)

[?(CP) whi [CP [FP [?(IP) ti [IP…]]]]]

Consider the relevant Kaqchikel paradigm (all the data below are taken from Erlewine 2014; the subject is underlined). (64a) illustrates the usual morphology that accompanies subject movement to SpecIP (following Aissen 1999, Erlewine argues that SpecIP is linearized to the right in Kaqchikel). [26] This morphology is not possible under wh-movement of the subject, as in (64b). Instead, the so-called agent-focus affix (AF), given in bold letters, appears. AF is not possible with object wh-movement (64c) or long-distance subject-movement; it occurs only with the local step of subject movement to SpecCP (even in long-distance questions, see (64d)). Interestingly, just like adding an adverb above IP rescues that-trace violations in English, adding an adverb in Kaqchikel makes available the appearance of the usual verbal morphology under wh-movement, the AF morphology not being available in this case (64e). These facts can be accounted for just like the local-subject movement data from English discussed above. A subject moved to SpecIP undergoes feature sharing with I, which results in the labeling of the object created by the subject-IP merger. However, this happens only when the structure is sent to the interfaces. As a result, if the subject that merges with IP moves to merge with CP right above it, antilocality is violated (no labeled projection is crossed). The problem does not arise in (64e) since the presence of the adverb introduces additional structure as a result of which subject that merges with IP can move to merge with CP and still cross a labeled projection. [27]

(64)
a.
Iwirx-Ø-u-tëjriwäyri a Juan.
yesterdaycom-B3sg-A3sg-eatthetortillaJuan

‘Yesterday, Juan ate the tortilla.’

b.
Achike*x-Ø-u-tëj/x-Ø-tj-öriwäy?
whocom-B3sg-A3sg-eat/com-B3sg-eat-AFthetortilla

‘Who ate the tortilla?’

c.
Achikex-Ø-u-tëj/*x-Ø-tj-öri a Juan?
whatcom-B3sg-A3sg-eat/ com-B3sg-eat-AFJuan

‘What did Juan eat?’

d.
Achiken-Ø-a-b’ijrat [chin*x-oj-r-tz’ët/x-oj-tz’et-öroj]?
whoinc-B3sg-A2sg-think2sg thatcom-B1pl-A3sg-see /com-B1pl-see-AF1pl

‘Who do you think saw us?’

e.
Achikekanqtzijx-Ø-u-tëj/*x-Ø-tj-öriwäy?
whoactuallycom-B3sg-A3sg-eat/com-B3sg-eat-AFthetortilla

‘Who actually ate the tortilla?’

The data in (65) provide further confirmation. Wh-phrases and indefinites have the same form in Kaqchikel. Erlewine shows that they are both licensed by movement to projections above IP, the indefinite-licensing projection, referred to below as QP, being lower than CP. Consider then (65b). If object is the indefinite, it will move to QP, which is located right above IP. As a result, the subject can move to IP and then to CP without violating antilocality. The usual verbal morphology is then possible on this reading in (65b). On other hand, on the unavailable “who-did-someone-see” reading, the object moves to CP and the subject moves to QP. Since QP is right above IP, subject movement to IP and then to QP violates antilocality, hence the ungrammaticality of (65b) on the reading in question. As before, the AF can save the derivation in question, hence this reading is available in (65a).

(65)
a.
Achikek’ox- Ø-tz’et- ö?
whocom-B3sg-see-AF

‘Who did someone see?’

*‘Who saw someone?’

b.
Achikek’ox- Ø-utz’ ët?
whocom-B3sg-A3sg-see

*‘Who did someone see?’

‘Who saw someone?’

Now, there was an implicit assumption above that, except in the case of that-relatives and constructions with (certain) pre-IP adverbs, English has only a single CP projection above IP; in other words, CP is not always split. Kaqchikel provides a confirmation of this; CP cannot always be split (more precisely, uniformly split) in Kaqchikel either, otherwise subject movement through IP would not be only selectively available. The conclusion from the above discussion is that there is no uniform split CP field that is present either crosslinguistically or in all constructions of a single language. [28] This is in fact not a surprising conclusion, given that even superficially, the left-periphery shows quite a bit of variation crosslinguistically.

There is, however, one point that still needs to be captured regarding Kaqchikel. The usual subject morphology requires movement to SpecIP. However, when the subject fails to move to SpecIP the usual subject morphology is not simply dropped: it is replaced by AF. Furthermore, the AF morphology appears only as last resort, to save the derivation where subject movement to SpecIP fails to occur for independent reasons. Why is that the case? A proposal made in Chomsky (2015) can be productively applied here. Chomsky (2015) restates the traditional EPP effect as a labeling effect. He suggests that languages differ regarding whether or not T is strong enough to label on its own. English T is not strong enough to label on its own, hence the subject needs to move to TP to strengthen it for labeling. On the other hand, in Italian T can label on its own, hence subject movement to TP is not necessary. [29]

Kaqchikel should then be like English, requiring movement to TP to strengthen T for labeling. I suggest therefore that AF is inserted as last resort to enable T to label when subject movement to SpecTP fails to take place. [30] This is the reason why AF occurs only in these contexts. However, given the obvious difference between English and Italian in the richness of subject agreement morphology, Chomsky ties the difference in the strength of T between English and Italian to the richness of agreement morphology. The analysis just suggested treats Kaqchikel and English in the same way although, in contrast to English, Kaqchikel has rich verbal morphology. There is, however another way of looking at the English/Italian/Kaqchikel paradigms in question where the issue in question does not arise.

Let us assume that, as in Italian, the usual morphology in Kaqchikel is enough to strengthen T for labeling. However, this morphology can only be licensed if the subject is located in SpecTP. [31] As a result, if the subject does not move to SpecTP, the morphology cannot be present, hence T is then like in English: it is weak and cannot label on its own. This is what rules out constructions where the agreement morphology is simply dropped in the absence of movement to SpecTP. As already proposed above, the AF-insertion can still be considered a last-resort strategy to strengthen T for labeling. [32] This explains the complementary distribution between the usual subject morphology and AF. When the usual subject morphology is present, which is possible any time subject moves to SpecTP, T is strong enough to label, hence there is no need for AF-insertion. When such morphology is not present, T cannot label hence AF is inserted to strengthen it. AF-insertion basically turns English-style, morphologically poor T into Italian-style, morphologically rich T. This analysis thus captures both the fact that verbal morphology cannot be simply dropped when the subject fails to move to SpecTP in Kaqchikel, as well as the last-resort nature of AF. It should, however, be noted that Kaqchikel is not quite like either English or Italian in Chomsky’s analysis of these two languages: Kaqchikel T is strong enough to label on its own, as expected given its morphological richness. In this respect, it is like Italian, not like English T. However, the relevant morphology can only be licensed in Kaqchikel when subject moves to SpecTP. [33] The motivation for the presence of the subject in SpecTP is thus different in Kaqchikel and English. In English, the subject moves there for labeling reasons, while in Kaqchikel the subject needs to be present there for morphology-licensing. In this respect, AF is more like subject movement in English. The AF-insertion takes place strictly for labeling reasons, just like subject movement in English. However, while this is always required in English (except in one context, where English uses a mechanism similar to AF, see Section 5.8.), it is required in Kaqchikel only in one, well-defined context.

5.5.2 Kinande

5.5.2.1 Subject agreement

The above account of Kaqchikel can also be extended to Kinande: the usual morphology in Kinande is enough to strengthen T, enabling it to label (cf. (25a)). However, this morphology can only be licensed if the subject is located in SpecTP. [34] As a result, if the subject does not move to SpecTP, the morphology cannot be present, hence T is then like English: it is weak and cannot label on its own. This is what rules out constructions where the agreement morphology is simply dropped in the absence of movement to SpecTP (compare *Iyondi yo langara Marya with (25c)). What happens in such contexts is that what is traditionally referred to as anti-agreement (AA) morphology is inserted (cf. (25c)). I suggest that, like AF in Kaqchikel, AA is inserted as last resort to enable T to label when subject movement to SpecTP fails to take place, which means that the usual subject agreement morphology is not licensed in such cases. This is then the reason why AA occurs only in these contexts. As in the case of Kaqchikel AF, this account explains the complementary distribution between the usual subject morphology and AA in Kinande. When the usual subject morphology is present, which is possible any time the subject moves to SpecTP, T is strong enough to label, hence there is no need for AA-insertion.

There is, however, a difference between Kaqchikel and Kinande. In contrast to Kaqchikel (cf. (64d)), long-distance contexts do not require AA-morphology in Kinande (I omit the agreement between the wh-phrase and the complementizer where this information is not relevant).

(66)
[CPiyondi [C’yo[Kambalea-alengekanaya[CPng’a-kahukaebiken]]]]
whothatKambaleagr-thoughtCagr-cookyams

‘Who did Kambale think is cooking yams?’

(Schneider-Zioga 2007)

Following Schneider-Zioga (2007), I suggest that the reason for this is that what occurs in the embedded SpecTP in (66), which also licenses the usual morphology, is pro. [35] In other words, pro here functions as a resumptive, with the wh-phrase base-generated in the matrix SpecCP. [36]

(67)

[CPiyondi [C’yo [Kambale a-alengekanaya [CPng’ pro a-kahuka ebiken]]]]

That this is an option in Kinande is confirmed by the fact that such constructions behave in all relevant respects just like constructions in which an object wh-phrase is present in SpecCP and associated with a resumptive clitic: like the resumptive clitic construction, the resumptive pro construction is island-insensitive (see (68) for the object+resumptive case and (69) for the subject+pro case) and does not induce WCO effects (see (70)).

(68)
a.
ekihikyoYosefua-kabula[iyondingay’u-ka-ki-gula]
whatthatJosephagr-wonderwhoifthatanti.agr-tense-cl-buy

‘What does Joseph wonder who is buying it?’

b.
ebaruhayahiyow-asiga[isi-wu-liw-asoma-yo]
letterwhichthatyou-leftneg-you-beyou-read-cl

‘Which article did you leave before you read it?’

(Schneider-Zioga 2007)

(69)
a.
[yondikyoYosefua-kabula[ekihijngaky’proka-kalangiratj]
whothatJosephagr-wonderwhatifthatagr-sees

‘Who does Joseph wonder what (he) sees?’

b.
omukalindikyow-asiga[prokisy-a-lyagua]
womanwhothatyou-leaveneg-agr-spoke

‘Which woman did you leave before (she) spoke?’

(Schneider-Zioga 2007)

(70)
a.
iyondiiyomamawiweia-kalengekanayaatiomugalimu
whothatmotherhisagr-thinksthatteacher
a-a-mu-nzirekutsibu
agr-past-cl-lovesbest

‘Whoi does hisi mother think that the teacher loves him the best?’

b.
iyondijyomamawiweja-kalengekanayaatiproja-anzireebitabu
whothatmotherhisagr-thinksthatagr-lovesbooks

‘Whoj does hisj mother think loves books?’

(Schneider-Zioga 2007)

I suggest that the reason why the pro-as-a-resumptive strategy is not an option in (25b) (if it were, regular morphology rather than AA would occur here) is the well-known ban on local subject resumptives, i.e. the ban on a resumptive pronoun in SpecTP that is associated with a wh-phrase in the SpecCP of the same clause (see Aoun and Li 1991; Boeckx 2003: 83–91; Borer 1984; McCloskey 1991; Bošković 2009, among others). [37]

There is also an alternative explanation. More generally, as noted in note 36, all movement in Kinande is clause-bounded; Kinande uses a variety of resumptive strategies (pro in an A-position, or a null Operator in an agreeing SpecCP which is co-indexed with a higher clause Operator) to establish long-distance dependencies. We are thus simply dealing here with an independent property of Kinande and the often-noted last-resort nature of resumptivization, which is used when movement is not an option (movement is an option in Kinande in a single-clause environment, but not in long-distance contexts).

The null operator strategy is illustrated by (71).

(71)
Ekihikyowasiga[islandembereMaryaaminye[ngakyo
whatwh.agryou.leavebeforeMaryknewCwh.agr
wasoma__]]
you.read

‘What did you leave before Mary knew you had read?’

(Schneider-Zioga 2009)

Here, a null Operator is moved to the embedded SpecCP (notice that we are dealing here with an agreeing C), and coindexed with the wh-phrase in the matrix SpecCP, voiding the adjunct island effect. Schneider-Zioga (2009) shows that even in non-island examples like (72), the fronted element cannot be reconstructed into the embedded clause to license the bound variable reading, which confirms that there is no long-distance movement taking place here, in accordance with the clause-bounded nature of movement in Kinande. [38]

(72)
ekitabukiwek/*jkyongalengekanaya[CPnga.kyo[obulimukolo]j
bookhiswh.agrI.thinkC.wh.agreverystudent
akasoma _kangikangi].
readregularly

‘(It is) Hisk/*j book that I think [every student]j reads regularly.’

(Schneider-Zioga 2009)

In this respect, consider also the AA-examples in (73), involving an adjunct island. The examples in question do not involve standard agreement, hence no pro in the SpecTP of the most embedded clause (in contrast to (69b)), which has the AA-morphology in (73).

(73)
a.
*omukalindiyowasiga[islandembere___wabuga]
womanwhowh.agryou.leftbeforespoke

‘Which woman did you leave before (she) spoke?’

b.
omukalindiyowasiga[islandembereKambaleanasi[CPko
womanwhowh.agryou.leftbeforeKambaleknewthat
yo___ wabuga]]
wh.agrspoke

‘Which woman did you leave before Kambale knew that (she) spoke?’

(Schneider-Zioga 2009)

(73a) is ruled out due to a locality effect. However, (73b) is acceptable. The only difference between (73a) and (73b) is that the theta-position of the wh-phrase is separated from the adjunct boundary by an agreeing CP in (73b). (73b) can then be derived as follows: a null Operator is base-generated in the most embedded CP, and undergoes movement to the local SpecCP (without moving to SpecTP), with the AA inserted to strengthen T. (The null element in question can be the same null element as the one located in SpecTP in e.g. (66)/(67), i.e. this may simply be pro moving to SpecCP. I will refer to such pro as null operator for ease of exposition). The operator is coindexed with the wh-phrase that is base-generated in the matrix SpecCP. Superficially, the AA-morphology is not just last resort in Kinande, since nothing here prevents the option of having pro in SpecTP and regular agreement in the most embedded clause; i.e. this option is in principle available (cf. for example (69b)). But not for the numeration in question: (73b) contains an agreeing C that requires an operator (a wh-phrase or a null element co-indexed with a wh-phrase) in its SpecCP. Since the only element that can undergo movement to the embedded SpecCP is the subject, local subject movement to SpecCP is then forced here, which disallows regular subject agreement morphology. The AA-morphology then only appears not to be last resort in Kinande: while there are cases where regular subject-agreement morphology and the AA-morphology alternate, such cases have different Cs: the AA-morphology is possible only in the context of an agreeing C, which requires movement to SpecCP; furthermore, it is the only option in such cases. [39]

5.5.2.2 Object agreement

Kinande is interesting in another respect. Chomsky (2015) suggests more or less uniform treatment of T and V when it comes to labeling. Interestingly, Kinande objects (in ditransitive and ECM environments, see Bošković 2008, Schneider-Zioga 1995 and the discussion below) also obligatorily trigger agreement, which is also not possible under wh-movement. However, with object wh-movement, this agreement is simply dropped, which suggests that T and V should not be treated in the same way when it comes to labeling.

Bošković (2008) suggests an analysis of Kinande where the object in the environments in question in Kinande must undergo object shift in order to be Case-marked, agreement being a reflex of this movement. Under Bošković’s (2008) analysis, these objects can actually be case-marked as long as they c-command the verb (at one point during the derivation, see also the discussion below); movement to SpecCP also suffices for this purpose. If object shift targets SpecVP, as suggested in Chomsky (2015), an object cannot undergo object shift and then proceed with wh-movement since the next step of movement would need to target vP, violating antilocality. [40] Wh-movement via the object shift position is then not an option, hence the impossibility of agreement in wh-contexts (the agreement morphology being licensed by the element in the object-shift position, on a par with TP).

Under this analysis, T and V are not treated uniformly with respect to labeling. While T needs to be phi-strengthened for labeling, this is not the case with V (when object agreement occurs, this is a result of a case-licensing movement, see the discussion below); this is why some subject agreement morphology is always obligatory with T, even in wh-contexts where the wh-subject is prevented from moving to SpecTP for independent reasons, which is not the case with object agreement morphology. At any rate, the Kinande facts in question indicate that T and V should not be treated in the same way when it comes to labeling.

Another important point about Kinande is that objects in simple transitives in Kinande do not trigger object agreement morphology, only objects in ECM configurations do. [41] As discussed in Bošković (2008), what is important here is that objects in simple transitive constructions c-command the verb even without movement. In other words, the descriptive generalization is that accusative-marked elements must c-command the case-marking verb: they trigger object agreement (hence move to SpecVP) only if other independently required operations do not put them in a position where they c-command the verb (wh-movement or lexical insertion itself in the case of direct objects). This can be straightforwardly accommodated under the approach to case-licensing argued for in Bošković (2007, 2008), where the DP needs to c-command the verb. More importantly for our current purposes, this also confirms that object-agreement morphology should not be tied to labeling and should be treated differently from subject-agreement morphology.

The above discussion also argues against Chomsky’s position that object NP moves to SpecVP from the V-complement position, movement which would be disallowed by antilocality. [42] Kinande in fact tells us when movement to SpecVP takes place (given its overt morphological reflex). It takes place in ECM constructions like (74) (see also note 41), but not in simple transitives like (75) and not in wh ECM constructions like (76). [43]

(74)

John believes Mary to know French.

(75)

John likes Mary.

(76)

Whoi do you believe ti to know French?

The upshot of the above discussion is that Kinande is important in that the subject/object agreement facts of Kinande argue against a uniform treatment of T/V and movement to SpecTP/SpecVP of the kind proposed in Chomsky (2015).

5.6 French que-qui alternation

I now turn to the French que-qui alternation, where qui occurs under subject wh-movement. A well-known aspect of this alternation is that qui occurs only with local subject movement.

(77)
a.
*Quelleétudiantecrois-tu[t’ que[t vapartir]]?

‘Which student do you believe that is going to leave?’

b.
Quelleétudiantecrois-tu[t’ qui[t vapartir]]?

‘Which student do you believe QUI is going to leave?’

c.
Quelleétudiantecrois-tu[t’ que[Marie va aider t]]?

‘Which student do you believe that Marie is going to help?’

d.
*Quelleétudiantecrois-tu[t’ qui[Marie va aider t]]?

‘Which student do you believe QUI Marie is going to help?’

(Rizzi and Shlonsky 2007)

The distribution of qui is in this respect very similar to AF in Kaqchikel, which also only occurs with local subject questions. In light of this similarity I suggest treating the two in the same way.

Taraldsen (2001) makes a very interesting proposal to treat qui as que+il, which is further expanded on in Rizzi and Shlonsky (2007), who also argue that this il (referred to below as –il) is not quite the same as regular 3p.sg expletive il. [44] The analysis raises a number of interesting questions, I simply refer the reader to Taraldsen (2001) and Rizzi and Shlonsky (2007) for relevant discussion. What is important for our purposes is that we are dealing here with insertion of a pronominal element, which takes place if something goes wrong with local subject wh-movement (i.e. in the configuration in (29), which, as discussed above, is disallowed). Insertion of this -il thus has a last resort flavor; it can be considered to be a rescue strategy for local subject wh-movement constructions. In this respect, this -il-insertion in fact works in exactly the same way as AF-insertion in Kaqchikel. The above account of Kaqchikel can then be extended to French, with a small difference in that French T can be treated exactly like English T when it comes to labeling (given the relative poverty of French verbal morphology). As in Chomsky’s (2015) proposal regarding English, French T is too weak to label. Normally, the subject moves to merge with it to strengthen it for labeling purposes. In a local subject wh-movement context (see (29)), such movement is not possible since, as discussed above, subject movement to SpecCP then violates antilocality. [45] In other words, wh-subjects cannot pass through SpecTP on their way to SpecCP. The latter movement cannot be avoided; without it we would get a PIC violation. The former movement can be avoided if the language has another way of strengthening T for labeling (in more traditional terms, satisfying its EPP requirement). The suggestion is then that this is precisely what this -il insertion accomplishes. It is then treated in the same way as Kaqchikel AF-insertion. It is a last-resort strategy which takes place to strengthen T in the contexts where the subject is independently prevented from moving to SpecTP.

5.7 Some alternatives

In this section I note several recent alternatives to the proposed account of local subject wh-movement constructions.

Consider first Chomsky’s (2015) account of the that-trace effect in English. As discussed above, Chomsky argues that T is too weak to label in English when it merges with vP (or whatever its complement is, given split Infl; recall also that I use interchangeably terms I(P) and T(P)). The subject then must move to SpecTP for labeling purposes (under (phi) feature sharing between the subject and T). Since T is too weak to label, if the subject were to move away there could be no labeling (given that traces are ignored for the purpose of labeling). This is the source of the that-trace effect: In constructions like (3), the subject must move to SpecCP given that CP is a phase; since traces are ignored for the purpose of labeling and T itself is too weak to label, TP cannot be labeled in such constructions.

What about Who do you think left? Chomsky argues that in such examples C transfers all of its features to T, including the phasehood property. In clauses with a null complementizer, that is actually deleted: when this occurs, TP becomes a phase. Subject can now move away from SpecTP because the label that was created for the subject-TP merger can be now preserved, the crucial assumption here being that once a phase is completed, all labels are set hence movement will not “delabel” the relevant object, i.e. in the case at hand, movement of the subject will not “de-label” TP.

Under the phase-labeling approach, we may expect labeling to be set only for what is actually sent to spell-out (i.e. the interfaces) under Chomsky’s interpretation-based approach to labels, hence not for the phasal-edge area. However, Chomsky’s analysis of Who do you think left then would not work without additional assumptions (see here Chomsky 2015). Even putting this issue aside, a more important problem is that Chomsky’s (2015) analysis appears to leave a number of constructions discussed above unaccounted for. Under this analysis, subject wh-movement via SpecTP should be possible only when there is no CP. This, however, cannot be right given the data discussed above: we have seen that if there is another projection between TP and CP, subject movement via SpecTP is possible (i.e. we have seen a number of cases of this type).

Another, independent issue with Chomsky’s analysis should be noted. If all labeling is done only at the phasal level, many constructions where both relevant elements move would be left unlabeled.

Rizzi (2006, 2013) also proposes rather interesting analyses where subject cannot move to SpecCP from SpecIP. In fact, he proposes an even stricter account where subject that moves to SpecIP can no longer move at all since he considers SpecIP a Criterial Freezing Position, from which no movement is possible (see Rizzi this volume for an account of this). This analysis has the same problem as Chomsky’s analysis regarding constructions where subject wh-movement apparently does proceed via IP.

5.8 Short subject questions

The grammaticality of simple English wh-questions like (78) now becomes puzzling. Given the above discussion, such examples cannot be derived by having the wh-phrase move to SpecIP and then to SpecCP since this derivation would violate antilocality.

(78)

Who left?

There are in fact a number of proposals in the literature to derive (78) without involving wh-movement via SpecIP. Thus, a number of authors have argued that the wh-subject in such constructions in fact stays in SpecIP, an account that is fully compatible with the system argued for here.

There are, however, some arguments against this analysis (for relevant discussion, see also Agbayani 2000; An 2007; Boeckx 2003; Pesetsky and Torrego 2001; among others). Consider for example the distribution of the hell-phrases (cf. Ginzburg and Sag 2000; Pesetsky and Torrego 2001). If the data in (79)–(81) are taken to indicate that the hell can only occur with wh-phrases in SpecCP, it then follows that the subject wh-phrase in (82) must be located in SpecCP, not SpecIP.

(79)

Who bought what?

(80)

What the hell did John buy?

(81)

*Who bought what the hell?

(82)

Who the hell arrested Mary?

While the argument is rather tempting, it might not be completely conclusive, since one could try to argue that for some reason, hell-phrases are not possible with wh-phrases located within vP. There are, however, other arguments that the wh-phrase does not remain in SpecIP in subject questions. Thus, Mizuguchi (2014) observes that if that were the case, we would expect (83) to be ambiguous, on a par with (84), which is not the case. [46]

(83)
Who loves everyone?(who>everyone; *everyone>who)
(84)
Someone loves everyone.(someone>everyone; everyone>someone)

Furthermore, the West Ulster English (WUE) data in (85)–(86), noted by McCloskey (2000), provide a rather conclusive argument that subject wh-phrases do not stay in SpecIP (or even move through SpecIP).

(85)

Whoi was arrested all ti in Duke Street?

(86)

*Theyi were arrested all ti last night.

While, in contrast to Standard English, WUE allows Q-float under wh-movement, just like standard English, WUE disallows (86). The ungrammaticality of (86) indicates that a subject located in SpecIP cannot float a quantifier in the postverbal position in passives. It must then be the case that who in (85) is not located in SpecIP; if that were the case, (85) should be ungrammatical on a par with (86) since subjects located in SpecIP cannot float a quantifier in this context. This example in fact also rules out the derivation where who in (85) moves to SpecCP via SpecTP. If that were the case, the quantifier in (85) would still be floated under movement to SpecIP, which (86) indicates is not possible. In fact, we have seen above that this derivation would anyway be ruled out by antilocality, on a par with a host of similar cases. Based on the above considerations, McCloskey (2000) in fact argues that who in (85) moves to SpecCP without moving to SpecIP. The reasoning seems sound; notice also that the derivation in question conforms with antilocality. But what happens with the usual requirement for the subject to move to SpecIP in English in such constructions?

I will take the above data to provide clear motivation that local subject wh-questions in English are derived through direct movement to SpecCP and give two suggestions why the subject-movement-to-SpecIP requirement is apparently voided in this context (for alternative accounts, see McCloskey 2000 and Bošković 2004).

5.8.1 Q in T

Chomsky (2008, 2013) argues that there is C-T association through which the heads in question share features. This means that when there is a Q-feature in C, there is also a Q-feature in T. Assume that this is indeed the case. I also suggest that, like AF in Kaqchikel and –il in French, the Q-feature strengthens T so that T can label. Given that labeling is the reason for the movement of the subject to SpecTP, there is then no need for the subject to move to SpecTP in this context; the subject can move directly to SpecCP. [47] Notice that under this account, the voiding of the traditional EPP requirement in English is confined to a +wh-CP context, i.e. to constructions where TP is immediately dominated by an interrogative CP. As a result, as desired, the account does not extend to other short subject A’-movement contexts discussed above, namely short subject topicalization, short subject relativization, and the that-trace effect.

5.8.2 T-to-C

There is an alternative analysis where the wh-subject in (78) also moves directly to SpecCP, without movement through SpecTP. English matrix wh-questions involve T-to-C movement. [48] This is the case in (78) too. (Under the analysis about to be proposed, the T-to-C movement from usual questions would also take place in embedded wh-subject questions as last resort since that would be the only way to ensure proper labeling). T-to-C movement eliminates the problem that arises due to the inability of T to label on its own. Once T moves away, given that traces are ignored for the purpose of labeling, the result of the T+vP merger is labeled by vP (this would give us two vacuous vP projections; as far as I can tell, no obvious problems arise because of this; notice also that if, as several works have suggested (see e.g. Koopman 2006), the subject must be located in the Spec of a projection that has T in its head position in order to get case, this does happen under the analysis currently under consideration). [49] Notice that this analysis is again specific to +wh-CP environments, i.e. contexts where TP is immediately dominated by a +wh-CP, hence it does not extend to other constructions involving short subject movement discussed above, namely, short subject topicalization, short subject relativization, and the that-trace effect, where subject movement through SpecTP still must take place.

6 Conclusion

The article has provided a uniform account of a number of locality effects, in particular, the Subject Condition, the Adjunct Condition, Richards’s (2001) tucking in effect, and the full Comp-trace paradigm, including (in addition to the basic case and its improvement with intervening adverbs) relative and extraposed clauses, the impossibility of short subject topicalization, French que-qui alternation, and the effect of subject wh-movement on agreement in languages like Kinande. It should, however, be noted that in spite of the wide coverage, the system argued for here does not purport to offer a comprehensive theory of all of locality of movement – there are certainly cases that are not covered by the current analysis (though the reader is referred to Bošković in press for an account of a generalized version of the Complex NP Constraint that extends to all complements of all lexical heads which is also crucially based on the approach to antilocality argued for here). [50] The account proposed here is based on the (somewhat modified) labeling framework of Chomsky (2013, 2015) and a labeling-based approach to antilocality, which treats it as a derivational constraint on movement steps that essentially requires movement to cross a labeled category (another way to think of this is that antilocality is defined on phrases, as in Bošković (2005, 2014), but that unlabeled elements do not count as phrases), and also bans movement from targeting unlabeled categories. The guiding intuition of the previous works on islands (e.g. Chomsky’s 1986Barriers), and more generally locality troublemakers, is that they are in a sense strong, with something about their strength blocking movement. The current approach is different in this respect: what makes something impenetrable is that it is underdetermined (due to the lack of a label, which makes movement from (some) islands too short). In other words, the problem with the relevant contexts is not that movement out of them is too long, as has been standardly assumed, but that it is too short. [51]

While a number of other proposals were made in the course of the discussion, one proposal needs to be emphasized: the account of the phenomena discussed in the article is crucially based on the proposal that there is a difference in the timing of labeling between the basic case where a head and a phrase merge (the base step) and the case where two phrases merge, where the former takes place in the course of the derivation and the latter takes place as part of the spell-out operation. As discussed in Section 2.2., in Chomsky (2013), the label in the former case is determined rather differently from the latter case: in the former case the label is determined via minimal search (MS), the same operation as Agree Closest, a syntactic mechanism falling under minimal computation. Since the labeling of the base step is done through an essentially syntactic mechanism, I have argued that it takes place when the relevant configuration is created. On the other hand, labeling in the case of merger of two phrases occurs when the relevant structure is sent to the interfaces, given Chomsky’s assumption that unlabeled objects are uninterpretable. It should, however, be noted that the latter assumption is not obvious, i.e. it is not obvious that labels are indeed needed for interpretation. Furthermore, at least with respect to the phenomena discussed here, nothing would go wrong in the syntax itself if the result of merger of two phrases is never labeled. If labels were not to be needed for interpretation (contra Chomsky 2013; for relevant discussion, see also Chametzky 2000; Collins 2002; Seely 2006; Hornstein 2009; Hornstein and Nunes 2008), such cases then would not need to be labeled at all. If the current discussion is on the right track, the result of head-complement merger still needs to be labeled. However, if labels are not needed for interpretation, the labeling here would then need to take place for a strictly syntactic reason. It seems plausible that in this case labeling may be required by subcategorization, i.e. that satisfying subcategorization requires that the element with the requirement to take a complement project (i.e. determine the label of the resulting object), otherwise, there would be no head-complement relation here; see in this respect Chomsky (2000). [52]

There may even be a more pressing reason for immediate labeling for the head-complement case. Phases are taken to determine the points of spell-out, i.e. when the structure is sent to the interfaces. If labeling occurs strictly for interpretative reasons we would expect it to occur at this point. But a serious chicken-or-the-egg style question then arises: as discussed in Bošković (2015), phasehood determination requires labeling, i.e. phases do not really exist prior to labeling: to know whether something is a phase we need to know its label (see Bošković 2015 for evidence that unlabeled elements cannot be phases; it is shown in that work that the well-known ban on movement out of moved elements can be deduced given that unlabeled objects cannot be phases). Since phases determine the points of spell-out, without any labeling structure cannot be sent to the interfaces, which in turn is necessary for labeling under a purely interpretative approach to labeling. The problem does not arise if head-complement merger is labeled immediately since this is actually all that is needed to determine the points of spell-out.

The upshot of the above discussion is that there is need to label (in the syntax) the result of head-complement merger even independently of Chomsky’s (2013, 2015) considerations.

Funding statement: Funding: This work is based upon research supported by the NSF under Grant BCS-0920888.

Acknowledgment

For helpful comments and suggestions, I thank Troy Messick, Jairo Nunes, Patricia Schneider-Zioga, YongSuk Yoo, the Linguistic Review reviewers, the participants of my 2014 UConn seminar, and the audiences at the International Workshop in Linguistics at Dokkyo University, Tokyo, the Syntax, Relative Clauses, and Slavic Languages Workshop at University of Utrecht, CRISSP at KU Leuven, Brussels, University of Geneva, and University of Paris 8. The paper is dedicated to the memory of Yoshiyuki Shibata.

[Corrections added after online publication December 24, 2015: On page 25, example (21a) an “e” before “popita” was deleted. “Asked-3s” was aligned under “popita”. In footnote 33, example (iib) “CLsg.m” was changed to “CL3sg.m”. On Page 27, first line below example (24) “kogo” was changed to “kakvo”]

References

Abels, Klaus. 2003. Successive cyclicity, antilocality, and adposition stranding. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Agbayani, Brian. 2000. Wh-subjects in English and the vacuous movement hypothesis. Linguistic Inquiry 31. 703–713.10.1162/002438900554523Search in Google Scholar

Aissen, Judith. 1999. Markedness and subject choice in optimality theory. Natural Language and Linguistic Theory 17. 673–711.10.1023/A:1006335629372Search in Google Scholar

An, Duk-Ho. 2007. Syntax at the PF interface. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Aoun, Joseph & Yen-Hui Audrey Li. 1991. The interaction of operators. In Robert Freidin (ed.), Principles and parameters in comparative grammar, 163–181. Cambridge, MA: MIT Press.Search in Google Scholar

Belletti, Adriana. 1990. Generalized verb movement. Turin: Rosenberg and Sellier.Search in Google Scholar

Bobaljik, Jonathan. 1995. Morphosyntax: The syntax of verbal inflection. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Bobaljik, Jonathan & Dianne Jonas. 1996. Subject positions and the roles of TP. Linguistic Inquiry 27. 195–236.Search in Google Scholar

Boeckx, Cedric. 2003. Islands and chains. Amsterdam: John Benjamins.10.1075/la.63Search in Google Scholar

Boeckx, Cedric. 2005. Some notes on bounding. Ms. Cambridge, MA: Harvard University.Search in Google Scholar

Boeckx, Cedric. 2008. Bare syntax. Oxford: Blackwell.Search in Google Scholar

Borer, Hagit. 1984. Parametric syntax. Dordrecht: Foris.10.1515/9783110808506Search in Google Scholar

Bošković, Željko. 1992. Clausal selection, subjacency, and minimality. Ms. Storrs, CT: University of Connecticut.Search in Google Scholar

Bošković, Željko. 1994. D-structure, theta-criterion, and movement into theta-positions. Linguistic Analysis 24. 247–286.Search in Google Scholar

Bošković, Željko. 1997. The syntax of nonfinite complementation: An economy approach. Cambridge, MA: MIT Press.Search in Google Scholar

Bošković, Željko. 2001. On the nature of the syntax-phonology interface: Cliticization and related phenomena. Amsterdam: Elsevier Science.10.1163/9780585474250Search in Google Scholar

Bošković, Željko. 2002. A-movement and the EPP. Syntax 5. 167–218.10.1111/1467-9612.00051Search in Google Scholar

Bošković, Željko. 2004. Be careful where you float your quantifiers. Natural Language & Linguistic Theory 22. 681–742.10.1007/s11049-004-2541-zSearch in Google Scholar

Bošković, Željko. 2005. On the locality of left branch extraction and the structure of NP. Studia Linguistica 59. 1–45.10.1111/j.1467-9582.2005.00118.xSearch in Google Scholar

Bošković, Željko. 2007. On the locality and motivation of Move and Agree: An even more minimal theory. Linguistic Inquiry 38. 589–644.10.1162/ling.2007.38.4.589Search in Google Scholar

Bošković, Željko. 2008. On successive cyclic movement and the freezing effect of feature checking. In Jutta M. Hartmann, Veronika Hegedüs & Henk van Riemsdijk (eds.), Sounds of silence: Empty elements in syntax and phonology, 195–233. Amsterdam: Elsevier.Search in Google Scholar

Bošković, Željko. 2009. On relativization strategies and resumptive pronouns. In Gerhild Zybatow, Uwe Junghanns, Denisa Lenertová & Petr Biskup (eds.), Studies in Formal Slavic Phonology, Morphology, Syntax, Semantics and Information Structure: Proceedings of FDSL 7, Leipzig 2007, 79–93. Frankfurt am Main: Peter Lang.Search in Google Scholar

Bošković, Željko. 2011. Rescue by PF deletion, traces as (non)interveners, and the that-trace effect. Linguistic Inquiry 42. 1–44.10.1162/LING_a_00027Search in Google Scholar

Bošković, Željko. 2013a. Traces do not head islands: What can PF deletion rescue?. In Yoichi Miyamoto, Daiko Takahashi, Hideki Maki, Masao Ochi, Koji Sugisaki & Asako Uchibori (eds.), Deep insights, broad perspectives: Essays in honor of Mamoru Saito, 56–93. Tokyo: Kaitakusha.Search in Google Scholar

Bošković, Željko. 2013b. Phases beyond clauses. In Lilia Schürcks, Anastasia Giannakidou & Urtzi Etxeberria (eds.), The nominal structure in Slavic and beyond, 75–128. Berlin: De Gruyter Mouton.10.1515/9781614512790.75Search in Google Scholar

Bošković, Željko. 2014. Now I’m a phase, now I’m not a phase: On the variability of phases with extraction and ellipsis. Linguistic Inquiry 45. 27–89.10.1162/LING_a_00148Search in Google Scholar

Bošković, Željko. in press. From the Complex NP Constraint to Everything: On deep extractions across categories. The Linguistic Review Glow Issue.10.1515/tlr-2015-0006Search in Google Scholar

Bošković, Željko. 2015. The ban on movement out of moved elements in the phasal/labeling system. Ms. Storrs, CT: University of Connecticut.Search in Google Scholar

Bošković, Željko & Howard Lasnik. 2003. On the distribution of null complementizers. Linguistic Inquiry 34. 527–546.10.1162/002438903322520142Search in Google Scholar

Brandi, Luciana & Patricia Cordin. 1989. Two Italian dialects and the null subject parameter. In Osvaldo Jaeggli & Kenneth J. Safir (eds.), The null subject parameter, 111–142. Dordrecht: Kluwer.10.1007/978-94-009-2540-3_4Search in Google Scholar

Branigan, Philip. 2005. The phase theoretic basis for Subject-Aux Inversion. Ms. St. John’s: Memorial University of Newfoundland.Search in Google Scholar

Bresnan, Joan W. 1972. Theory of complementation in English syntax. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Brillman, Ruth & Aron Hirsch. 2014. An anti-locality account of English subject/non-subject asymmetries. Paper presented at the 50th meeting of the Chicago Linguistic Society, Chicago.Search in Google Scholar

Broekhuis, Hans. 2005. Extraction from subjects: Some remarks on Chomsky’s On phases. In Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz & Jan Koster (eds.), Organizing the grammar: Studies in honor of Henk van Riemsdjik, 59–68. Berlin: Mouton de Gruyter.10.1515/9783110892994Search in Google Scholar

Browning, Maggie. 1996. CP recursion and that-t effects. Linguistic Inquiry 27. 237–255.Search in Google Scholar

Buesa García, Carlos. 2011. The interaction between locality and the subject-gap restriction in Spanish questions. In Suzi Lima, Kevin Mullin & Brian Smith (eds.), Proceedings of the North East Linguistic Society 39, 163–176. Amherst: GLSA, University of Massachusetts.Search in Google Scholar

Campos, Héctor. 1997. On subject extraction and the anti-agreement effect in Romance. Linguistic Inquiry 28. 92–119.Search in Google Scholar

Chametzky, Robert. 2000. Phrase structure: From GB to minimalism. Malden: Blackwell.Search in Google Scholar

de Chene, Brent. 1995. Complementizer-trace effects and the ECP. Geneva Generative Papers 3(1). 1–4. Département de Linguistique Générale, Université de Genève.Search in Google Scholar

Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.10.1515/9783112316009Search in Google Scholar

Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.Search in Google Scholar

Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press.Search in Google Scholar

Chomsky, Noam. 2000. Minimalist inquires. In Roger Martin, David Michaels & Juan Uriagereka (eds.), Step by step: Essays on Minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press.Search in Google Scholar

Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press.Search in Google Scholar

Chomsky, Noam. 2008. On phases. In Robert Freidin, Carlos Peregrin & Maria Luisa Zubizarreta (eds.), Foundational issues in linguistic theory. Essays in honor of Jean-Roger Vergnaud, 133–166. Cambridge, MA: MIT Press.10.7551/mitpress/9780262062787.003.0007Search in Google Scholar

Chomsky, Noam. 2013. Problems of projection. Lingua 130. 33–49.10.1075/la.223.01choSearch in Google Scholar

Chomsky, Noam. 2015. Problems of projection: Extensions. In Elisa Di Domenico, Cornelia Hamann, Simona Matteini (eds.), Structures, strategies and beyond – studies in honour of Adriana Belletti, 3–16. Amsterdam & Philadelphia: John Benjamins.10.1075/la.223.01choSearch in Google Scholar

Chomsky, Noam & Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 11. 1–4610.1007/978-94-015-6859-3_4Search in Google Scholar

Cinque, Guglielmo. 1999. Adverbs and functional heads: A cross-linguistic perspective. Oxford: Oxford University Press.Search in Google Scholar

Collins, Chris. 2002. Eliminating labels. In Samuel David Epstein & T. Daniel Seely (eds.), Derivation and explanation in Minimalist syntax, 42–64. New York: Wiley.10.1002/9780470755662.ch3Search in Google Scholar

Collins, Chris. 2014. Merge(X.Y) = {X,Y}. Ms. New York: New York University.Search in Google Scholar

Coon, Jessica, Pedro Mateo Pedro & Omer Preminger. In press. The role of case in A-bar extraction asymmetries: Evidence from Mayan. Linguistic Variation.10.1075/lv.14.2.01cooSearch in Google Scholar

Culicover, Peter. 1992. The adverb effect: Evidence against ECP accounts of the that-t effect. In Amy J. Schafer (ed.), Proceedings of The North East Linguistics Society 23, 97–11. Amherst: GLSA, University of Massachusetts.Search in Google Scholar

Culicover, Peter & Michael Rochemont. 1992. Adjunct extraction from NP and the ECP. Linguistic Inquiry 23. 496–501.Search in Google Scholar

Deal, Amy Rose. 2014. Cyclicity and connectivity in Nez Perce relative clauses. Ms. Santa Cruz, CA: University of California Santa Cruz.Search in Google Scholar

Dikken, Marcel den. 1995. Particles. Oxford: Oxford University Press.Search in Google Scholar

Dikken, Marcel den. 2006. Relators and linkers: The syntax of predication, predicate inversion and copulas. Cambridge, MA: MIT Press.10.7551/mitpress/5873.001.0001Search in Google Scholar

Dikken, Marcel den. 2007. Phase extension: Contours of a theory of the role of head movement in phrasal extraction. Theoretical Linguistics 33. 1–41.10.1515/TL.2007.001Search in Google Scholar

Epstein, Samuel David, Hisatsugu Kitahara & T. Daniel Seely. 2014. Labeling by minimal search: Implications for successive-cyclic A-movement and the conception of the postulate “Phase”. Linguistic Inquiry 45. 463–481.10.1162/LING_a_00163Search in Google Scholar

Erlewine, Michael. 2014. Anti-locality and optimality in Kaqchikel Agent Focus. Ms. Cambridge, MA: MIT.10.1007/s11049-015-9310-zSearch in Google Scholar

Fox, Danny & Howard Lasnik. 2003. Successive-cyclic movement and island repair: The difference between sluicing and VP-ellipsis. Linguistic Inquiry 34. 143–154.10.1162/002438903763255959Search in Google Scholar

Gallego, Ángel J. & Juan Uriagereka. 2006. Conditions on sub-extraction. Ms. Barcelona and College Park: Universitat Autònoma de Barcelona and University of Maryland.Search in Google Scholar

Gallego, Ángel J. & Juan Uriagereka. 2007. Sub-extraction from subjects: A phase theory account. In José Camacho, Nydia Flores-Ferrán, Liliana Sánchez, Viviane Déprez & María José Cabrera (eds.), Romance Linguistics 2006, 149–162. Amsterdam: John Benjamins.10.1075/cilt.287.12galSearch in Google Scholar

Ginzburg, Jonathan & Ivan A. Sag. 2000. Interrogative Investigations. Stanford: CSLI Publications.Search in Google Scholar

Grohmann, Kleanthes. 2003. Prolific domains: On the antilocality of movement dependencies. Amsterdam: John Benjamins.10.1075/la.66Search in Google Scholar

Grohmann, Kleanthes. 2011. Antilocality: Too close relations in grammar. In Cedric Boeckx (ed.), The Oxford handbook of linguistic minimalism, 260–290. Oxford: Oxford University Press.Search in Google Scholar

Halle, Morris & Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In Kenneth Hale & Samuel Jay Keyser (eds.), The view from building 20: Essays in linguistics in honor of Sylvain Bromberger, 111–176. Cambridge, MA: MIT Press.Search in Google Scholar

Hiraiwa, Ken. 2005. Dimensions of symmetry in syntax: Agreement and clausal architecture. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Hoge, Kerstin. 2001. That-t effects in English and Yiddish. In Galina Alexandrova & Olga Arnaudova (eds.), The Minimalist Parameter, 233–248. Amsterdam: John Benjamins.10.1075/cilt.192.18hogSearch in Google Scholar

Holmberg, Anders & Thorbjörg Hróarsdóttir. 2003. Agreement and movement in Icelandic raising constructions. Lingua 113. 997–1019.10.1016/S0024-3841(02)00162-6Search in Google Scholar

Hornstein, Norbert. 2009. A theory of syntax: Minimal operations and universal grammar. Cambridge: Cambridge University Press.10.1017/CBO9780511575129Search in Google Scholar

Hornstein, Norbert & Jairo Nunes. 2008. Adjunction, labeling and bare phrase structure. Biolinguistics 2. 57–86.10.5964/bioling.8621Search in Google Scholar

Hornstein, Norbert & Jacek Witkoś. 2003. Yet another approach to existential constructions. In Lars-Olof Delsing, Cecilia Falk, Gunlög Josefsson & Halldór Ármann Sigurðsson (eds.), Grammar in Focus: Festschrift for Christer Platzack, Lund: Lund University, Department of Scandinavian.Search in Google Scholar

Hornstein, Norbert, Howard Lasnik & Juan Uriagereka. 2003. The dynamics of islands: Speculations on the locality of movement. Linguistic Analysis 33. 149–175.Search in Google Scholar

Huang, C.-T. James. 1982. Logical relations in Chinese and the theory of grammar. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Hunter, Tim. 2010. Relating movement and adjunction in syntax and semantics. Barcelona and College Park, MD: University of Maryland dissertation.Search in Google Scholar

Ishii, Toru. 1999. Cyclic spell-out and the that-trace effect. In Sonya Bird, Andrew Carnie, Jason Haugen & Peter Norquest (eds.), Proceedings of the West Coast Conference on Formal Linguistics 18, 220–231. Sommerville: Cascadilla Press.Search in Google Scholar

Ishii, Toru. 2004. The phase impenetrability condition, the vacuous movement hypothesis, and that-t effects. Lingua 114. 183–215.10.1016/S0024-3841(03)00045-7Search in Google Scholar

Jeong, Youngmi. 2006. The landscape of applicatives. College Park, MD: University of Maryland dissertation.Search in Google Scholar

Kandybowicz, Jason. 2006. Comp-trace effects explained away. In Donald Baumer, David Montero & Michael Scanlon (eds.), Proceedings of the 25th West Coast Conference on Formal Linguistics, 220–228. Somerville: Cascadilla Press.Search in Google Scholar

Kandybowicz, Jason. 2008. The grammar of repetition: Nupe grammar at the syntax phonology interface. Amsterdam: John Benjamins.10.1075/la.136Search in Google Scholar

Kayne, Richard. 1984. Connectedness and binary branching. Dordrecht: Foris.10.1515/9783111682228Search in Google Scholar

Kim, Sun-Woong. 2008. English C moves downward as well as upward: An extension of Bošković and Lasnik’s (2003) approach. Linguistic Inquiry 39. 295–307.10.1162/ling.2008.39.2.295Search in Google Scholar

Koopman, Hilda. 2006. Agreement configuration: In defense of the Spec head configuration. In Cedric Boeckx (ed.), Agreement systems, 159–199. Amsterdam: John Benjamins.10.1075/la.92.09kooSearch in Google Scholar

Lasnik, Howard. 1995. Verbal morphology: Syntactic structures meets the Minimalist Program. In Héctor Campos & Paula Kempchinsky (eds.), Evolution and revolution in linguistic theory: Essays in honor of Carlos Otero, 251–275. Washington, DC: Georgetown University Press.Search in Google Scholar

Lasnik, Howard. 1999. Minimalist analysis. Oxford: Blackwell.Search in Google Scholar

Lasnik, Howard. 2001. When can you save a structure by destroying it?. In Min-Joo Kim & Uri Strauss (eds.), Proceedings of the North East Linguistic Society, 31, 301–320. Amherst: GLSA, University of Massachusetts.Search in Google Scholar

Lasnik, Howard & Mamoru Saito. 1984. On the nature of proper government. Linguistic Inquiry 15. 235–289.10.1007/978-94-015-6859-3_10Search in Google Scholar

Lasnik, Howard & Mamoru Saito. 1992. Move α: Conditions on its application and output. Cambridge, MA: MIT Press.Search in Google Scholar

Law, Paul. 1993. On the base position of wh-adjuncts and extraction. Paper presented at the 67th Annual Meeting of the Linguistic Society of America, Los Angeles, California, January 1993.Search in Google Scholar

Lohndal, Terje. 2007. Subextraction and the freezing effect: A Scandinavian case study. Ms. University of Oslo.Search in Google Scholar

Lohndal, Terje. 2009. Comp-T effects: Variation in the position and features of C. Studia Linguistica 63. 204–232.10.1111/j.1467-9582.2009.01159.xSearch in Google Scholar

McCloskey, James. 1991. Resumptive pronouns, A’-binding, and levels of representation in Irish. In Randall Hendrick (ed.), Syntax and semantics 23: The syntax of the modern Celtic languages, 199–248. New York: Academic Press.10.1163/9789004373228_008Search in Google Scholar

McCloskey, James. 2000. Quantifier float and wh-movement in an Irish English. Linguistic Inquiry 31. 57–84.10.1162/002438900554299Search in Google Scholar

Merchant, Jason. 2001. The syntax of silence. Oxford: Oxford University Press.Search in Google Scholar

Mizuguchi, Manabu. 2008. Derivation, minimalism and that-trace effects. English Linguistics 25. 56–92.10.9793/elsj1984.25.56Search in Google Scholar

Mizuguchi, Manabu. 2014. Phases, labeling, and wh-movement of the subject. Paper presented at the 32nd Conference of the English Linguistic Society of Japan, Gakushuin University, Tokyo, November 8.Search in Google Scholar

Moro, Andrea. 1997. The raising of predicates: Predicative noun phrases and the theory of clause structure. Cambridge: Cambridge University Press.10.1017/CBO9780511519956Search in Google Scholar

Müller, Gereon. 2010. On Deriving CED Effects from the PIC. Linguistic Inquiry 41. 35–82.10.1162/ling.2010.41.1.35Search in Google Scholar

Mutaka, Ngessimo Mathe. n.d. Kinande: A grammar sketch. The African Anaphora Project, Rutgers University.Search in Google Scholar

Nomura, Masashi. 2005. Nominative case and AGREE(ment). Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Nunes, Jairo. in preparation. The locus of edge features and its implications for upward and sideward movement. Ms. São Paulo: Universidade de São Paulo.Search in Google Scholar

Ott, Dennis. In press. Symmetric merge and local instability: Evidence from split topics. Syntax.10.1111/synt.12027Search in Google Scholar

Perlmutter, David. 1971. Deep and surface constraints in syntax. New York: Holt, Rinehart and Winston.Search in Google Scholar

Pesetsky, David. 1981. Complementizer-trace phenomena and the nominative island condition. The Linguistic Review 1. 297–343.10.1515/tlir.1982.1.3.297Search in Google Scholar

Pesetsky, David. 1992. Zero Syntax Vol. 2. Ms. Cambridge, MA: MIT.Search in Google Scholar

Pesetsky, David & Esther Torrego. 2001. T-to-C movement: Causes and consequences. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 355–426. Cambridge, MA: MIT Press.Search in Google Scholar

Preminger, Omer. 2011. Agreement as a fallible operation. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Reinhart, Tanya. 1980. On the position of extraposed clauses. Linguistic Inquiry 11. 621–624.Search in Google Scholar

Richards, Norvin. 2001. Movement in language: Interactions and architectures. Oxford: Oxford University Press.Search in Google Scholar

Riqueros, José. 2013. Spanish nominal(ization) patterns. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Rizzi, Luigi. 1990. Relativized minimality. Cambridge, MA: MIT Press.Search in Google Scholar

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar, 281–337. Dordrecht: Kluwer.10.1007/978-94-011-5420-8_7Search in Google Scholar

Rizzi, Luigi. 2006. On the form of chains: Criterial positions and ECP effects. In Lisa Lai-Shen Cheng & Norbert Corver (eds.), Wh-movement: Moving on, 97–133. Cambridge, MA: MIT Press.Search in Google Scholar

Rizzi, Luigi. 2013. Cartography, criteria and labeling. Ms. Université de Genève.Search in Google Scholar

Rizzi, Luigi. this volume. Labelling, maximality, and the head-phrase distinction.Search in Google Scholar

Rizzi, Luigi & Ur Shlonsky. 2007. Strategies of subject extraction. In H.-M.Gärtner & Uli Sauerland (eds.), Interfaces + Recursion = Language? Chomsky’s minimalism and the view from syntax-semantics, 115–160. Berlin: Mouton de Gruyter.Search in Google Scholar

Ross, John Robert. 1969. Guess who? In Robert I. Binnick, Alice Davison, Georgia M. Green & Jerry L. Morgan (eds.), Papers from the Fifth Regional Meeting of the Chicago Linguistic Society, 252–286. Chicago: University of Chicago, Chicago Linguistic Society.Search in Google Scholar

Roussou, Anna. 2002. C, T, and the subject: That-t phenomena revisited. Lingua 112. 13–52.10.1016/S0024-3841(00)00056-5Search in Google Scholar

Rudin, Catherine. 1988. On multiple questions and multiple wh-fronting. Natural Language and Linguistic Theory 6. 445–501.10.1007/BF00134489Search in Google Scholar

Sabel, Joachim. 2000. Expletives as features. In Roger Billerey & Brook Danielle Lillehaugen (eds.), The Proceedings of the Nineteenth West Coast Conference on Formal Linguistics, 101–114. Somerville: Cascadilla Press.Search in Google Scholar

Saito, Mamoru. 2014. Case and labeling in a language without φ-feature agreement. In Anna Cardinaletti, Guglielmo Cinque & Yoshio Endo (eds.), On Peripheries: Exploring clause initial and clause final positions, 269–297. Tokyo: Hituzi Syobo Publishing.Search in Google Scholar

Saito, Mamoru & Keiko Murasugi. 1999. Subject predication within IP and DP. In Kyle Johnson & Ian Roberts (eds.), Beyond principles and parameters, 167–188. Dordrecht: Kluwer.10.1007/978-94-011-4822-1_7Search in Google Scholar

Schneider-Zioga, Patricia. 1995. Specifier/head agreement in Kinande. Cahiers Linguistiques d’Ottawa 23. 67–93.Search in Google Scholar

Schneider-Zioga, Patricia. 2007. Anti-agreement, anti-locality, and minimality. Natural Language & Linguistic Theory 25. 403–446.10.1007/s11049-006-9014-5Search in Google Scholar

Schneider-Zioga, Patricia. 2009. Wh-agreement and bounded and unbounded movement. In Jose Brucart, Anna Gavorro & Juame Sola (eds.), Merging Features, 46–59. Oxford: Oxford University Press.10.1093/acprof:oso/9780199553266.003.0003Search in Google Scholar

Seely T. Daniel. 2006. Merge, derivational c-command, and subcategorization in a label-free syntax. In Cedric Boeckx (ed.), Minimalist Essays, 182–220. Amsterdam: John Benjamins.10.1075/la.91.13seeSearch in Google Scholar

Shlonsky, Ur. 2014. A note on labeling, Berber states, and VSO order. In Sabrina Bendjaballah, Noam Faust, Mohamed Lahrouchi & Nicola Lampitelli (eds.), The Form of structure, the structure of form, 349–360. Amsterdam: John Benjamins.10.1075/lfab.12.27shlSearch in Google Scholar

Sportiche, Dominique. 2011. French relative qui. Linguistic Inquiry 42. 83–124.10.1162/LING_a_00029Search in Google Scholar

Stepanov, Arthur. 2001. Cyclic domains in syntactic theory. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Stepanov, Arthur. 2007. The end of CED? Minimalism and extraction domains. Syntax 10. 80–126.10.1111/j.1467-9612.2007.00094.xSearch in Google Scholar

Stjepanović, Sandra. 1999. What do second position cliticization, scrambling, and multiple wh-fronting have in common. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Stowell, Timothy. 1981. Origins of phrase structure. Cambridge, MA: Massachusetts Institute of Technology dissertation.Search in Google Scholar

Szczegielniak, Adam. 1999. That-t effects crosslinguistically and successive cyclic movement. In Karlos Arregi, Benjamin Bruening, Cornelia Krause & Vivian Lin (eds.), MIT Working Papers in Linguistics, vol. 33, 369–393. Cambridge: Department of Linguistics, MITWPL.Search in Google Scholar

Talić, Aida. 2014. Extraordinary complement extraction: PP-complements and inherently case-marked nominal complements. Studies in Polish Linguistics 8. 127–150.Search in Google Scholar

Taraldsen, Knut T. 2001. Subject extraction, the distribution of expletives and stylistic inversion. In Aafke Hulk & Jean-Yves Pollock (eds.), Subject inversion in Romance and the theory of universal grammar, 16–182. New York: Oxford University Press.Search in Google Scholar

Ticio, Emma. 2005. Locality and anti-locality in Spanish DPs. Syntax 8. 229–286.10.1111/j.1467-9612.2005.00080.xSearch in Google Scholar

Truswell, Robert. 2011. Events, phrases, and questions. Oxford: Oxford University Press.10.1093/acprof:oso/9780199577774.001.0001Search in Google Scholar

Uriagereka, Juan. 1988. On government. Storrs, CT: University of Connecticut dissertation.Search in Google Scholar

Uriagereka, Juan. 2012. Spell-out and the Minimalist Program. Oxford: Oxford University Press.10.1093/acprof:oso/9780199593521.001.0001Search in Google Scholar

Watanabe, Akira. 1993. Larsonian CP recursion, factive complements, and selection. In Schafer, Amy J. (ed.), Proceedings of the 23rd meeting of the North East Linguistics Society, 523–537. Amherst: GLSA, University of Massachusetts.Search in Google Scholar

Wurmbrand, Susi. 2013. Tagalog infinitives: Consequences for the theory of phases, voice marking and extraction. Ms. Storrs, CT: University of Connecticut.Search in Google Scholar

Wurmbrand, Susi. 2014. Tense and aspect in English infinitives. Linguistic Inquiry 45. 403–447.10.1162/LING_a_00161Search in Google Scholar

Yoo, YongSuk. 2015. Labeling algorithm and its consequences on movement. Paper presented at the 17th Seoul Conference on Generative Grammar, Kyung Hee University, Seoul, Korea, August 5th-8th.Search in Google Scholar

Published Online: 2015-12-24
Published in Print: 2016-2-1

©2016 by De Gruyter Mouton

Downloaded on 19.3.2024 from https://www.degruyter.com/document/doi/10.1515/tlr-2015-0013/html
Scroll to top button