Abstract

This article shows that stripping, the elision of declarative TPs, is possible not only in coordinate structures, but also in embedded clauses—however, only when the complementizer is absent. This Embedded Stripping Generalization is not predicted by earlier accounts of stripping, but it falls out from a certain combination of independently available assumptions. Specifically, I propose a zero Spell-Out view of ellipsis in a dynamic (or contextual) phasehood approach, which, together with the lack of a CP layer in that-less embedded clauses, derives this generalization in languages like English. I then briefly consider stripping in other languages and suggest that the analysis also has the flexibility to accommodate crosslinguistic differences in the distribution of stripping.

1 A Restriction on Stripping

The term stripping was coined by Hankamer and Sag (1976:409): “Stripping is a rule that deletes everything in a clause under identity with corresponding parts of a preceding clause except for one constituent (and sometimes a clause-initial adverb or negative).” Constructions such as (1a–e) illustrate the phenomenon (see also Lobeck 1995, Merchant 2003). Merchant (2003) provides several arguments against a structure involving movement of the remnant (the nonelided stranded material) from a conjoined phrase. Following these observations, I assume that stripping involves a structure with clausal conjunction (to be specified below) and clausal ellipsis.1

(1)

Stripping is claimed to be restricted to coordinate structures and impossible in subordinate clauses, as in (2a–c). Corresponding VP-ellipsis constructions, on the other hand, are possible in embedded contexts, as in (2d–e).

(2)

  • a.

    *Jane loves to study rocks, and John says that geography too.

    (Lobeck 1995:27, (72b))

  • b.

    *Abby wanted to take Dutch, because Ben.

    (Merchant 2003:3, (20))

  • c.

    *Abby claimed Ben would ask her out, but she didn’t think that Bill (too).

    (Merchant 2003:4, (21))

  • d.

    Abby wanted to take Dutch because Ben does.

  • e.

    Abby claimed Ben would ask her out, but she didn’t think that Bill would.

Two proposals have been made to account for this restriction (I will discuss a third approach in section 3 after I present my analysis). Following Johnson (2009), who discusses a similar restriction on gapping, stripping could involve vP-coordination as in (3a) (= (1c–e)), with the shared VP undergoing ellipsis or across-the-board movement. Since conjunction is one clause up in (3b) (= (2c)), a low vP-coordination structure is not possible, and therefore, the modal would and the lowest VP cannot be shared by the two conjuncts (i.e., across-the-board movement of the VP above the conjunction is not possible). Only a VP-ellipsis configuration in which the embedded modal is present in the second conjunct can be derived.

(3)

graphic

A different approach, following Merchant (2003), would be to encode the stripping restriction as part of the lexical properties of the feature-licensing ellipsis, as in (4). Merchant suggests that the head selecting the TP in a stripping context is equipped with an ellipsis feature (Estripping) that licenses its complement to be elided. Estripping involves a strong focus feature that attracts a focus element to its specifier. Furthermore, Estripping involves an uninterpretable feature uConj that needs to be checked by a higher conjunction. The restriction that stripping is only possible in conjunctions thus comes from the feature licensing ellipsis.

(4)

graphic

While these two approaches capture the basic stripping data in (1) and (2), they make the wrong prediction for embedding structures without complementizers. As shown in (5), there is a sharp contrast between Merchant’s (2003) and Lobeck’s (1995) embedding examples, repeated as (5a) and (5c), respectively, in which the complementizer is present, and the corresponding examples in (5b) and (5d) where the complementizer is omitted. Further examples are given in (6) and (7). The examples in (6) are from a search conducted on the Corpus of Contemporary American English (COCA) and have been confirmed with native speakers. The contexts of these examples clearly point to embedding structures rather than quotes, since you is interpreted indexically. The same is true in (7), where this year refers to 2013. According to native speakers, both (6) and (7) are ungrammatical if that is added.

(5)

  • a.

    *Abby claimed (that) Ben would ask her out, but she didn’t think that Bill (too).

  • b.

    Abby claimed (that) Ben would ask her out, but she didn’t think Bill (too).2

  • c.

    *Jane loves to study rocks, and John says that geography too.

  • d.

    Jane loves to study rocks, and John says geography too.

(6)

  • a.

    [W]hen we asked her . . . who her favorite new country star is, she said you.

  • b.

    When I get asked who’s the biggest diva on the set, I say you.

(7)

  • a.

    First, they thought it would be done last year, then they thought (*that) THIS year.

  • b.

    First, they predicted there would be driverless cars in 2000, then they predicted (*that) THIS year.

The difference between (5a,c) and (5b,d) leads to the following generalization:

(8)

  • Embedded Stripping Generalization

  • Stripping of embedded clauses is only possible when the embedded clause lacks a CP.

It is hard to see how the Embedded Stripping Generalization can be derived in Johnson’s (2009) or Merchant’s (2003) approaches. As in (3b) (= (5a)), conjunction one clause up from the clause with the stripped material in (5b) would prohibit low vP-coordination and sharing of the embedded VP and IP. The presence or absence of a complementizer would not make any difference here. Similarly, in Merchant’s approach the EStripping feature would need to be licensed by conjunction, which is too far away in (5a), and it would presumably also be too far away in (5b).

In the next section, I introduce an alternative approach to ellipsis, which, together with a particular view of phasehood, provides a unified account of the Embedded Stripping Generalization and the facts discussed so far.

2 Accounting for the Embedded Stripping Generalization

2.1 Dynamic Phasehood Approach

Since the account of ellipsis to be presented is set in a dynamic approach to phasehood determination, I will first lay out and summarize some of the arguments for this approach. Broadly, phases are defined in two ways: statically as specific syntactic constructs (typically vP and CP) or dynamically as domains that are “complete” by some linguistic measure. Although Chomsky (1998, 2000, 2001) settles on a static definition, he also considers a dynamic one.

(9)

The main issue for the view that the phasal projections are vP and CP is that the static definition in (9a) does not single out only these two projections. Therefore, one question has been why only vP and CP are typically considered to be phases and not, for instance, TP as well. In relative clause CPs, it is particularly difficult to see in what way CP is propositional but TP is not—semantically, exactly the opposite conclusion would hold. Thus, as Grano and Lasnik (2015:7) note, “[T]his leaves us with a ‘list problem’: the set of phase types has to be stipulated rather than following from something more general.” While the split of a clause into two cycles has been widely adopted, the definition in (9a) and specifically the claim that phases are (statically) vP and CP have been questioned and other approaches have been pursued. Two theoretical research areas have provided insights into how phases are determined: (a) the phasal status of syntactic configurations that lack vP and CP, and (b) the question of exactly where the boundaries of the two cycles are in cases where the vP and CP domains involve a richer structure.

As for the first question, a converging view in many works on phases is that the domain that includes the verb, all of its associated arguments, and possibly other selected entities constitutes a phase. As a result, unaccusative and passive vPs/VPs qualify as phases, despite the absence of an agentive (strong) vP, since they are complete thematic domains of the verb (similar to Grimshaw’s (1991) notion of extended projection). This view is advocated by, among others, Legate (2003), Sauerland (2003), Bobaljik and Wurmbrand (2005), Wurmbrand (2015), and Wurmbrand and Shimamura (to appear) on the basis of evidence from scope, binding, Case assignment, agreement, and restructuring. I only replicate one argument from Sauerland (2003) here. As (10a) shows, raising constructions allow a moved subject to bind a variable in the matrix experiencer and, at the same time, be interpreted in the scope of matrix negation. Assuming that negation does not undergo movement (see, e.g., Sauerland 2003 for evidence), the not >every interpretation requires the subject to occupy an LF position lower than matrix negation but higher than the matrix experiencer. If raising constructions have the phase structure outlined in (10b), this state of affairs is exactly as expected. Movement across a phase boundary must pass through the phase edge, and any position occupied in the course of the derivation is a possible LF position (all lower positions, though available as reconstruction sites in principle, would not be options in this case since the quantified subject could then not bind into the matrix experiencer).

(10)

graphic

The view presented above leads to the dynamic definition of phases in (11) (see also Bošković 2014 for a similar view and further arguments for (11a) for DPs and other syntactic categories).

(11)

  • a.

    The highest projection of a cyclic domain constitutes a phase.

  • b.

    The cyclic domains of a clause are

    • i.

      the extended thematic domain of V and

    • ii.

      the combined T and C domains.

The definition of the lower cyclic domain in (11bi) has been refined in works that consider a more complex temporal and aspectual domain, and a crucial question is where the phasal boundary lies in expanded tense, aspect, and v domains. A concurring conclusion reached in several works is that different types of aspect belong to different syntactic cycles—while perfect groups with the tense/C cycle, progressive is part of the lower clausal cycle, that is, part of the “vP” phase. For instance, Harwood (2015) shows that several phenomena, among them existential constructions, idiomatic constructions, VP-fronting, and VP-ellipsis, treat the progressive, but not the perfect, as part of the lower clausal domain. To just give one example, fronting must include the progressive auxiliary, as in (12), but must exclude the perfect auxiliary, as in (13) (the same distribution holds in specificational pseudoclefts). Assuming that fronting targets phases, this distribution then indicates that the lower clausal phase includes the progressive but not the perfect.

(12) If Darth Vader says that Han Solo was being frozen in carbonite, then . . .

(13) If Luke says he would have fought hard, then . . .

Turning finally to the higher clausal domain, the traditional CP phase, the dynamic definition of phases in (11) has the result that not only CPs but also CP-less clauses with a TP domain function as phases (for detailed arguments, see Wurmbrand 2013, 2014, Wurmbrand and Haddad 2016). Under this view, raising constructions such as (14a) include three phase boundaries that need to be crossed during subject movement to matrix Spec,TP, thus yielding the four subject positions in (14b). In (10), we have already seen some evidence for the stopover in position ➁ . Movement to position ➂ has been argued for at length in several works, with evidence coming from binding, floating quantifiers, and scope (see, e.g., Chomsky 1973, Grohmann, Drury, and Castillo 2000, Pesetsky and Torrego 2007, Castillo, Drury, and Grohmann 2009); for specific evidence, see these works.

Here, I would like to summarize a different piece of evidence for the four subject positions in a raising configuration as predicted by the dynamic phasehood approach in (14b).

(14)

graphic

In Wurmbrand and Haddad 2016, we show that the four positions in (14b) can be occupied by the subject overtly in a language like Standard Arabic. The examples corresponding to (14c) are given in (15). Positions ➀, ➁, and ➂ are reflected straightforwardly in the word order in (15ac). Pinning down the subject to position ➂ is more complicated since Arabic is a VSO language and since, owing to the high position of the verb, word order does not directly distinguish between positions ➁ and ➂. However, as shown in (15d), in certain constructions, the subject must occur with accusative, rather than nominative. In Wurmbrand and Haddad 2016, we propose that case is determined at Spell-Out, and if the subject is spelled out together with the matrix VP, it receives accusative case. Since the subject in position ➂ is at the edge of the lower TP phase, it is not spelled out with the embedded clause; rather, it is spelled out together with the matrix VP once the matrix vP phase is completed. If the subject occupies position ➁ , on the other hand, it is spelled out together with the matrix TP, in which case it is realized with nominative case. Thus, the case difference between (15b) and (15d) is a reflex of the different positions—➁ and ➂ —in (14c).

(15)

graphic

With this short summary of a dynamic phasehood approach, I now return to the main focus of this article, the Embedded Stripping Generalization.

2.2 Ellipsis as Zero Spell-Out

Recently, several works have proposed that ellipsis is a form of Spell-Out: phase heads trigger Spell-Out of their complements, and ellipsis is the option of not realizing a Spell-Out domain (SOD) at PF (see Gengel 2006, 2009, Van Craenenbroeck 2010b, Gallego 2010, Rouveret 2012, Bošković 2014). Thus, elided constituents are unpronounced SODs. I will refer to this approach as the zero Spell-Out approach.

Following Merchant’s (2003) basic structure, I assume that the remnant of stripping occupies a focus projection (see also Ortega-Santos, Yoshida, and Nakao 2014, Yoshida, Nakao, and Ortega-Santos 2015). A zero Spell-Out approach to ellipsis, in conjunction with the dynamic phasehood approach laid out in the previous section, allows us to derive the Embedded Stripping Generalization. Let us first consider the structure of the examples in (1) and (2), illustrated in (16). In a main clause stripping context such as (16ab), the root projection of the stripping clause is the focus phrase hosting the stripping remnant. Being the top projection of the higher clausal domain, the FocP is a phase and the TP therefore a SOD. Leaving that SOD unpronounced is thus correctly predicted to be possible. In the embedded context in (16cd), on the other hand, the top projection of the embedded clause is the CP, making the FocP the SOD (see below). Since the TP is not a SOD, stripping is correctly predicted to be impossible.

(16)

graphic

This approach predicts that the complement of a phase head C should be elidable, which is indeed the case in standard sluicing constructions (see Merchant 2001, Van Craenenbroeck 2010b, among many others). However, while being a SOD is a necessary condition for ellipsis, it is, of course, not a sufficient condition. In addition to involving the appropriate structural configuration, ellipsis is also subject to a parallelism/identity requirement between the elided and antecedent constituents (see, among many others, Sag 1976, Fiengo and May 1994, Johnson 2001, Merchant 2001, 2008, to appear). In the present context, this will exclude ellipsis of FocP (a SOD, but one without a parallel antecedent) in cases such as (16d) (cf. *Abby speaks passable Dutch, becauseBen speaks passable Dutch).3

Although the three accounts discussed so far make the same predictions for the cases in (1) and (2), they differ regarding the ones in (5)–(7). Both Johnson’s (2009) low coordination account and Merchant’s (2003) EStripping feature account prohibit stripping in contexts where the elided XP is in an embedded clause, even if the higher clause is part of a conjunction. The zero Spell-Out account, on the other hand, predicts stripping to be possible in embedded clauses, as long as there is no CP. This is, I argue, exactly what we find in embedded clauses lacking that.

The structure of that-less complements is controversial, and whether such clauses involve a CP-less structure or an embedded CP with an empty complementizer is still under debate.4 While I will not be able to resolve this debate, I believe that the properties of stripping may shed some interesting new light on it. Recall that the crucial difference between (16ab) and (16cd) is the presence of the CP. If there is no CP above FocP, the TP is a SOD and can thus be elided. In contrast, if a CP is present above FocP, the TP is not a SOD and hence cannot be elided. If, following the CP-less view, that-less embedded clauses lack their top projection, we are in a position to account for the contrasts in (5)–(7). The relevant parts of the structures are given in (17). In (17a) (= (5a)), think combines with a CP, which, as in (16cd), prevents the TP from eliding since it is not a SOD. The same is true for (5c) and the versions with that in (6) and (7). In (17b) (= (5b)), on the other hand, think combines with a CP-less embedded clause, which makes the FocP the top projection and hence a phase. As in (16ab), the TP is a SOD, and zero Spell-Out of the TP is thus correctly predicted to be possible in (5b,d), (6), and (7).5

(17)

  • a.

    *. . . but she didn’t think [CP=phase that [FocP=SOD Bill *[TPwould ask her out]]]

  • b.

    . . . but she didn’t think [FocP=phase Bill ✓[TP=SODwould ask her out]]

The current account also extends to fragment answers (thanks to Jason Merchant for drawing my attention to these cases). Assuming a clausal ellipsis account of fragment answers (see Merchant 2004), the utterance in (18b) involves a FocP with the remnant PP in its specifier. If this FocP is the top projection, deletion of the complement TP is possible, exactly as in (16ab)/(17b). However, if a complementizer (i.e., a CP) is added, FocP ceases to be a phase and TP will not be a SOD anymore, thus not allowing TP-deletion.

(18)

As a final point, certain limitations of the current account should be noted. The zero Spell-Out approach only demands that a stripped constituent be a SOD, and the dynamic phasehood approach provides the algorithm for calculating whether a particular syntactic configuration is a SOD. For embedded contexts, this establishes that TP-stripping is only possible when the embedded clause lacks a CP. Whether that-drop is possible or not, however, is an independent property that is not covered by the account proposed here. In other words, the zero Spell-Out approach makes stripping contingent on the availability of that-drop, but it does not regulate when CP-less embedded clauses are possible. As proposed by Pesetsky and Torrego (2004, 2006), for instance, clausal complements of nouns typically do not license that-drop (but see Wurmbrand 2014 and references therein for a more nuanced description of this property). It is then expected that stripping is not possible in embedded clauses combining with nouns. Adjectives, on the other hand, can occur with that-less complements and stripping is acceptable, at least in certain contexts.

(19)

  • a.

    Bill was afraid the storm will be destructive.

    ( Pesetsky and Torrego 2004:502, (15b))

  • b.

    Mary is afraid Bill will be arrested.

  • c.

    Mary knows Tom will be arrested, and she’s afraid Bill too.

Finally, in addition to restrictions and speaker variation regarding the availability of that-drop, stripping seems to generally be a marked phenomenon and one restricted to only a subset of contexts that allow that-less embedded clauses. As pointed out by Jason Merchant ( pers. comm.), embedded stripping works best with predicates of reporting (say, claim, think, predict, suggest), but is less readily available with others (e.g., learn, determine, confirm). The account offered here only provides the upper bounds of when stripping is possible; it does not distinguish between different types of embedding structures. I leave open here whether such differences should be treated as additional restrictions on the formal licensing of stripping, as semantic properties (e.g., by perhaps relating stripping to quotes), or as an extragrammatical phenomenon.6

3 An Alternative Account?

In this section, I discuss a potential alternative account of the Embedded Stripping Generalization and provide reasons for not adopting it. In languages involving wh-movement to Spec,CP, the so-called Sluicing-Comp Generalization holds: in those languages, complementizers or elements moved to C cannot cooccur with sluicing (see Lobeck 1995, Lasnik 1999, Merchant 2001; and see section 5 for an example of a language where this generalization is not in effect). Baltin (2010) proposes that the Sluicing-Comp Generalization, illustrated in (20ab),7 can be derived from a split-CP structure—specifically, the structure in (20c) where the wh-phrase is in Spec,FocP and moved T or that is in the lower Fin(ite) head.

(20)

  • a.

    They discussed a certain model, but they didn’t know which model (*that).

    (Baltin 2010:331, (2))

  • b.

    A: He visited somebody.

    B: Oh, really. Who did he visit?/Who (*did)?

    (Baltin 2010:331, (3))

  • c.

    [ForceP [TopP [FocP who/which model [FinP that/did]]]]

Assuming that sluicing targets FinP, it follows that moved T elements and complementizers must always delete. This account could be extended to stripping in the following way. Examples such as (5cd) (repeated as (21a)) would have the structure in (21b) (in what follows, the elided part is marked by shading). If stripping, like sluicing, targets FinP, the complementizer that would have to be deleted; in other words, the Embedded Stripping Generalization would boil down to a subcase of the Sluicing-Comp Generalization.8

(21)

  • a.

    Jane loves to study rocks, and John says (*that) geography too.

  • b.

    says [FocP geography [FinP that [TP Jane loves to study t]]]

Although this account is attractive, as it would relate the obligatory omission of that in stripping contexts to similar properties found in sluicing, it faces several questions when considered for English (see below for other languages). First, as Rizzi (1997) and Haegeman (2000a,b) argue, in embedded questions the wh-XP is not in Spec,FocP, but in Spec,ForceP, since embedded interrogatives are selected by the matrix verb and the highest CP projection must therefore carry the wh-feature, which then attracts the wh-XP to its specifier. Furthermore, embedded wh-XPs can cooccur with lower focus elements, such as the negative inversion cases given in (22ac) or the topic constructions in (22de).9

(22)

  • a.

    Lee wonders ?whether/why under no circumstances at all would Robin volunteer.

    (Culicover 1991, Haegeman 2000a:136–137, (28))

  • b.

    ?Lee wonders who under no circumstances would Robin help.

  • c.

    ?Lee wonders what under no circumstances at all would Chris ever buy.

  • d.

    I was wondering for which jobs, during the vacation, I should go into the office.

    (Haegeman 2000a:135, (25))

  • e.

    No one could ever explain to me how in just fifteen minutes two children could make such a mess.

    (Culicover 1996:460, (47b))

Proponents of a split-CP structure therefore assume that the structures of embedded questions are as in (23) rather than as in (20c).

(23)

  • a.

    [ForcePwh-XP [FocP under no circumstances {would} [FinP {would}]]]

  • b.

    [ForcePwh-XP [TopP during the vacation [FinP ]]]

Second, in contrast to what Baltin (2010) suggests, Haegeman (2000a,b) assumes that T-to-C movement in (22ac) targets Foc rather than Fin, since this movement is triggered by the Neg Criterion, which requires a Spec-head relation between the focused XP and the Foc head. Thus, under Haegeman’s account, the Sluicing/Stripping-Comp Generalization (the prohibition against stranding C elements) could not be derived by the assumption that sluicing targets FinP.

Finally, as laid out in (21b), the extension of Baltin’s sluicing account to stripping would entail that complementizers are lower than focused or topicalized XPs in embedded clauses. For English, this is clearly not the case, as shown in (24ac). For that reason, Rizzi (1997) and Haegeman (2000a,b) argue that the complementizer in embedded declaratives is in Force (see (24d)), rather than Fin.

(24)

  • a.

    John thinks that geography, Jane loves to study.

  • b.

    *John thinks geography that Jane loves to study.

  • c.

    I stress that, during the vacation, on no account will I go into the office.

    (Haegeman 2000a:135, (26a))

  • d.

    stress [ForceP that [TopP during the vacation [FocP on no account will [FinP ]]]]

To maintain the structure in (21b) thus appears to be quite unmotivated for English. Note that the cooccurrence of that and the inverted auxiliary in (24c) also makes unlikely an account according to which that corresponds to T that is moved into the CP domain. Rather, the most straightforward account seems to be that that is base-generated in the highest CP projection (e.g., ForceP in a split structure). However, the question then arises why that must be missing in stripping cases, which is exactly what the zero Spell-Out analysis proposed here accounts for.

A Baltin-style split-CP structure may, however, be a promising approach for stripping in verb-second languages such as German. As shown in (25a), stripping is possible in German, but only when the verb (traditionally assumed to be in C) is also deleted. Under a traditional CP structure, this is surprising. However, if the moved XP and the finite verb are in different projections, as suggested by Baltin and shown in (25b), the outcome is as desired: TopP or FocP is the top projection of the clause, hence a phase, and its complement FinP, which contains the finite verb, is the SOD that undergoes zero Spell-Out. Furthermore, German, like English, allows embedded stripping—however, as predicted, only when the complementizer is not present (see (25c)).10

(25)

  • a.

    Leo spricht Deutsch, und Viktor (*spricht) auch.

    Leo speaks German and Viktor (*speaks) also

    ‘Leo speaks German and Viktor too.’

  • b.

    and [TopP/FocP=phase Viktor [FinP=SOD speaks [TP tSubj tT German tV]]]

  • c.

    Leo spricht Englisch, und Kai behauptet, (*dass) Lina auch.

    Leo speaks English and Kai claims (*that) Lina also

    ‘Leo speaks English and Kai claims (*that) Lina, too.’

4 Additional Material in Stripping Contexts

In this section, I turn to constructions in which the stripping remnant includes the focused phrase plus additional material. If such a complex remnant entails two separate projections, and only the higher projection constitutes a phase, stripping of the third projection would be predicted to be impossible in the zero Spell-Out approach. I will show that the stripping analysis proposed here successfully deals with several cases that at first sight appear to challenge it.

4.1 Why-Stripping and Sluice Stripping

There are two constructions that may lead to the conclusion that the elided part is not a SOD: why-stripping as in (26a) and sluice stripping as in (26b). Why-stripping involves the wh-phrases why or how come plus a focused XP (natto in (26a)), and sluice stripping (also sometimes called wh-stripping) involves a wh-phrase (which can be of any type) plus another XP (the PP about phonology in (26b)). If, as depicted in (26c), the why-phrase and the other stranded remnant are in different CP-type projections, the elided part will not correspond to a SOD and zero Spell-Out should be impossible. However, building on the accounts by Nevins (2008), Ortega-Santos, Yoshida, and Nakao (2014), and Yoshida, Nakao, and Ortega-Santos (2015), I propose that these two types of stripping only involve one C projection and adjunction of one of the remnants to that CP. The structures for the two types of stripping are slightly different, so I will discuss them in turn.

(26)

A property that sets sluice stripping apart from other types of stripping and sluicing (with the exception of why-sluicing) is clause-boundedness. Sluice stripping is possible in a simple clause, as in (27a), but not when long-distance sluicing applies, as in (27b). Why-stripping, in contrast, can span across a clause boundary, as in (27c).

(27)

  • a.

    A phonetician talked about syntax, but I don’t know who about semantics.

  • b.

    *No phoneticiani thought that a syntactician talked about hisi paper, but I wish I could remember who no phoneticiani thought that twho talked about hisi presentation.

    (Yoshida, Nakao, and Ortega-Santos 2015:345, (72))

  • c.

    No linguisti thinks that a student should talk about hisi supervisor’s paper, but I don’t understand why no linguisti thinks that a student should talk (about) HISi SUPERVISOR’S PAPER (not his presentation).

    (Yoshida, Nakao, and Ortega-Santos 2015:345, (73))

Following Nevins (2008), Ortega-Santos, Yoshida, and Nakao (2014), and Yoshida, Nakao, and Ortega-Santos (2015), the strict clause-boundedness of sluice stripping is accounted for if this form of stripping involves regular wh-movement to Spec,CP plus rightward movement of the non-wh-remnant. The structure of (26b) is thus as in (28). Since there is only one CP, the TP is the complement of the phase head, hence a SOD, and zero Spell-Out is correctly predicted to be possible.

(28)

graphic

Turning to the second construction, Ortega-Santos, Yoshida, and Nakao (2014) and Yoshida, Nakao, and Ortega-Santos (2015) provide an in-depth analysis of why-stripping in English and Spanish. They show that only one of the remnants in why-stripping, the non-wh-XP, involves movement to the CP domain, whereas the wh-element (why) is base-generated in the CP. Evidence for movement of the non-wh-remnant comes from Case connectivity effects, binding connectivity effects, selectional properties, the distribution of P-stranding, implicit correlates, and voice-matching requirements. To just give one example, (29) illustrates binding connectivity. If whystripping involves clausal ellipsis (indicated by the shading in (29c)) and movement of the nonwh- remnant, it is correctly predicted that (29a), but not (29b), involves a Condition C violation.

(29)

The claim that in why-stripping the wh-element is base-generated and not moved is motivated by the lack of strict locality and scope properties. As shown in (30a), why-sluicing is impossible across a finite clause (Merchant 2001). Whatever the reason for this restriction, the important observation is that why-stripping in (30b) is not subject to this constraint.

(30)

Furthermore, while why-questions such as (31a) show scope ambiguity between the wh-phrase and a universal quantifier (in particular, (31a) can receive a pair-list answer), why-stripping does not permit an interpretation of the wh-phrase within the scope of a quantifier—a pair-list answer is infelicitous in (31b). Both of these properties follow straightforwardly if there is no whmovement in why-stripping.

(31)

graphic

The last crucial property of why-stripping is that this construction is restricted to certain whelements: why and how come. As shown in (32), other wh-phrases do not allow this form of stripping.

(32)

  • a.

    A: John was eating natto. B: Why NATTO?

  • b.

    A: John was eating natto. B: How come NATTO?

  • c.

    A: Someone was eating natto. B: *How/*When/*Where NATTO?

  • d.

    A: Something made John eat natto. B: *What NATTO?

    (Yoshida, Nakao, and Ortega-Santos 2015:326, (6))

The zero Spell-Out account proposed here predicts that stripping is impossible in cases where there are two (or more) XPs above the TP. I assume that true wh-movement targets CP and that there is a Focus projection below CP. Thus, examples such as (32cd) have the structure in (33). Since the TP is not a SOD, ellipsis (i.e., zero Spell-Out) is not possible.11

(33)

graphic

Why then is why different? The special behavior of why has been noted in several works on wh-movement, and a common (well-motivated) assumption is that why can be base-generated in CP (see, among many others, Lasnik and Saito 1984, 1992, Rizzi 1990, Collins 1991, Bromberger 1992, Ko 2005). Following Bromberger (1992), Yoshida, Nakao, and Ortega-Santos (2015) note that why-interrogatives have a unique focus association property. Answers to why-questions differ depending on which constituent receives stress in the why-question, since the focus association triggers different alternatives. Putting the three main observations (base-generation of why, movement of the focus phrase to the left periphery, and obligatory focus association) together, Yoshida, Nakao, and Ortega-Santos (2015) adopt a split-CP domain and propose the structure in (34). The wh-phrase is base-generated in the highest projection (Int(errogative)P or ForceP) and requires focus association with a lower XP. When stripping applies, this association must be made visible for reasons of recoverability; thus, overt focus movement must apply in stripping contexts.

(34)

The specific structure in (34) will not be compatible with the zero Spell-Out approach in this article, since TP would not qualify as a SOD. However, the slightly modified alternative structure in (35) is compatible with it. The main difference is that in (35) there is no separate projection for the wh-element; instead, why is adjoined to the FocP.

(35) [FocP why [FocP NATTO [TP=SOD John was eating NATTO]]]

The main reason for the assumption that why occupies a separate projection from the FocP, according to Yoshida, Nakao, and Ortega-Santos (2015), is that why-stripping constructions can be interpreted as interrogatives (they can also be exclamatives). A clause-typing projection such as IntP or ForceP is thus assumed to mark a clause as an interrogative. However, as we show in Bobaljik and Wurmbrand 2015, following approaches such as Ginzburg and Sag’s (2000), an interrogative semantics does not require a syntactic interrogative structure. In various wh-movement languages, certain wh-in-situ constructions, such as those in (36), can be interpreted as genuine (nonecho) questions; however, the syntactic context they appear in is clearly declarative.

(36)

In Bobaljik and Wurmbrand 2015, we propose that the interrogative force of these constructions comes from the combination of a wh-element (equipped with an interpretable wh-feature) and focus semantics. These are exactly the pieces that Yoshida, Nakao, and Ortega-Santos (2015) argue are present in why-stripping. Thus, the structure in (35) covers all the properties of why-stripping, including their potential as interrogatives, and it is also compatible with the zero Spell-Out view, thereby allowing a uniform account of regular stripping, sluice stripping, and why-stripping.

4.2 Negation

A last set of facts that bear on the analysis proposed here are cases in which negation is stranded under stripping in addition to the focused element, as in examples like (37ab). In such cases, negation is typically obligatory with the connectors but and although, as illustrated in (37c).

(37)

I assume that in the above contexts, but and although12 are coordinators that require ( possibly via a conventional implicature; see Grice 1975) a contrast between the two conjuncts (see Toosarvandani 2013, 2014), which can be instantiated via reversed polarity in the two conjuncts (cf. (38)).13

(38)

  • a.

    Abby speaks Dutch, but/although Ben doesn’t/*does.

  • b.

    Abby doesn’t speak Dutch, but/although Ben does/*doesn’t.

Polarity differences are possible in VP-ellipsis contexts (see (38a)/(39a)), since the TP in the second conjunct is overt (only the VP is elided) and can thus be different from the TP in the first conjunct. In stripping contexts, on the other hand, parallelism requires that the two TPs be identical. Since, as shown in (39b), the elided TP (the shaded part) cannot simultaneously involve the same polarity value (as required by parallelism) and a different polarity value ( polarity reversal triggered by but, although), stripping in (37c)/(39b) is impossible. The only context that allows both requirements to be met is the one where the TP matches in polarity, but the remnant of ellipsis involves polarity reversal. This is the case in (37ab)/(39c), where the focused phrase involves constituent negation.

(39)

graphic

Merchant (2003) leaves open whether examples such as (37ab) involve constituent negation or a NegP projected above the FocP. Unless NegP constitutes a phase by itself, the zero Spell-Out approach is only compatible with the former. As Winfried Lechner notes (pers. comm.), the addition of either supports this claim (see Klima 1964). While either can be added to VP-ellipsis cases such as (40a), which clearly involve sentential negation, adding either to a stripping context with a negative remnant is impossible, as illustrated in (40bc).14

(40)

  • a.

    ?Sue doesn’t speak Dutch, but Abby does. Ben doesn’t either.

  • b.

    Sue doesn’t speak Dutch, but Abby does, (al)though not Ben.

  • c.

    ?*Sue doesn’t speak Dutch, but Abby does, (al)though not Ben either.

5 Extensions and Conclusion

In this final section, I provide a preliminary suggestion for extending the current proposal to stripping in languages where the Embedded Stripping Generalization does not hold, as well as a possible extension of the current approach to gapping.

Looking beyond English and German, we find that stripping is allowed in constructions with a complementizer in certain languages, as illustrated in (41a) for Hungarian and (41b) for Spanish.15 An interesting generalization about which languages allow these constructions is provided by Van Craenenbroeck and Lipták (2006, 2008): such stripping is only possible in languages in which wh-movement targets FocP, rather than CP. Van Craenenbroeck and Lipták propose that in those languages, Foc is equipped with an operator feature, which, like C in English-type languages, licenses ellipsis of its complement.

(41)

  • a.

    János meghívott valakit és azt hiszem, hogy BÉLÁT.

    János invited someone.ACC and that.ACC think that Béla.ACC

    ‘János invited someone and I think it was Béla whom he invited.’

    (Van Craenenbroeck and Lipták 2006:260, (26a))

  • b.

    Me dijeron que si llueve (que) se quedan aquí, y que si nieva (que) también.

    CL said that if rains (that) CL stay here and that if snows (that) too

    ‘They told me that they are going to stay here if it rains or snows.’

    (Villa-García 2012a:210, (16))

The approach proposed here also has a way of implementing this generalization: the head that attracts a wh-phrase (i.e., the head equipped with the operator property) functions as a phase head. In English, this coincides with the top projection of the CP domain. However, in wh-focus languages, this is Foc. As shown in (42), if FocP is a phase, it then follows that its complement, TP, can be elided, despite the presence of a higher complementizer.

(42)

graphic

While the proposed structural difference is still mostly descriptive, the possible advantage of this approach is that Van Craenenbroeck and Lipták’s generalization can potentially be derived. By tying both properties—the operator status of Foc and ellipsis—to phasehood, a connection between the licensing ellipsis and having the operator property is established, which may turn out to be less arbitrary than the assignment of an ellipsis feature. This claim will, of course, need to be backed up by showing that English-type languages and wh-focus languages behave differently regarding locality or other phasehood properties. But some initial evidence for the phasal status of FocP is provided by the distribution of complementizers in Spanish. As Villa-García (2012a,b) shows, and as illustrated in (41b), Spanish allows multiple occurrences of the complementizer que. Interestingly, Villa-García (2012b) argues that medial que, such as the que associated with a TopP or FocP, creates island effects. This is expected if FocP constitutes a phase, exactly as proposed here.

The second extension concerns gapping. The restrictions noted for stripping in English are very similar to restrictions known for gapping. Since the distribution of gapping is considerably more complex, I can again only provide a preliminary suggestion for how to extend the current account of stripping to similar facts found with gapping. Like stripping, gapping appears to be restricted to conjunctions and to be impossible in embedded contexts (see, e.g., Koutsoudas 1971, Sag 1976, Hankamer 1979, Wilder 1994, Williams 1997, Johnson 2001, 2009). This is illustrated in (43).16 If gapping can be analyzed as involving movement of multiple XPs to a TP-external focus projection, the difference between (43a) and (43be) can be derived in the same way as suggested for analogous examples with stripping. This would then lead us to expect that omission of a complementizer as in (43f) should also show an ameliorating effect in gapping. While most speakers consulted find a contrast between (43e) and (43f ), judgments for gapping are in general significantly less clear and stable, and further empirical studies are needed to see whether a unification of stripping and gapping is possible.

(43)

  • a.

    Some have served mussels to Sue and others have served swordfish.

    (Johnson 2009:289, (1))

  • b.

    *Some had eaten mussels because others had eaten shrimp.

    (Johnson 2009:293, (13b))

  • c.

    *Some will eat mussels because others will eat shrimp.

  • d.

    *Some had eaten mussels and she claims that others had eaten shrimp.

    (Johnson 2009:293, (15b))

  • e.

    *Some will eat mussels and she claims that others will eat shrimp.

  • f.

    %Some will eat mussels and she claims others will eat shrimp.

To conclude, I have proposed that a zero Spell-Out approach to ellipsis, together with a dynamic phasehood view, provides a straightforward account of the Embedded Stripping Generalization—the restriction that stripping can only apply in embedded clauses when the complementizer is missing. While puzzling for other ellipsis approaches, this distribution follows from a zero Spell-Out approach combined with the view that embedded root clauses are CP-less. Furthermore, following the proposals by Nevins (2008), Ortega-Santos, Yoshida, and Nakao (2014), and Yoshida, Nakao, and Ortega-Santos (2015), I have shown that why-stripping and sluice stripping are compatible with the zero Spell-Out approach to ellipsis, and that the distribution of negation in stripping remnants can be derived from a parallelism requirement and a constituent negation structure. More speculatively, I have suggested that the generalization proposed by Van Craenenbroeck and Lipták (2006, 2008)—namely, that that-stripping is only possible in languages in which wh-movement targets FocP rather than CP—can be derived if a head equipped with the operator property in the CP always constitutes a phase head ( possibly in addition to the top projection of the CP domain). If the current account can be maintained after considering the stripping and phasehood properties of more languages, a general consequence of the system would be that stripping can be used as a phase detector in the CP domain.

Finally, the zero Spell-Out approach also extends to other types of ellipsis: sluicing (CP phase, TP SOD), NP-ellipsis (DP phase, NP SOD), pseudogapping (FocP phase, vP SOD; see footnote 16), and VP-ellipsis (vP phase, VP SOD). While the account proposed here covers the basic cases of these constructions, the distribution of the different types of ellipsis is far more complex, and a detailed account of the structural and phasal properties of these constructions will need to consider each type of ellipsis separately.

Notes

This article originated as a squib and I thank all the reviewers and editors for their useful comments, which eventually led to the current version. I am also grateful for comments and feedback from Jonathan Bobaljik, Željko Bošković, Kyle Johnson, Sabine Laszakovits, Winfried Lechner, Jason Merchant, Troy Messick, Wilfried Öller, David Pesetsky, Luis Vicente, Julio Villa-García, Masaya Yoshida, and the University of Connecticut students and auditors of my spring seminars.

1 Thanks to Jonathan Bobaljik, Jon Gajewski, Jason Merchant, Troy Messick, and Peter Smith for providing English examples.

2 Example (5b) also shows that a parenthetical analysis of the higher clause (she didn’t think) cannot be maintained. In section 4, I suggest that the coordinator but imposes a contrast requirement, which can be met by reversed polarity in the two conjuncts or a contextual contrast arising through the implicature of too. In the absence of too, negation is thus required in one of the conjuncts. With the right intonation and with stress on BILL, (5b) can be used without too, which shows that the contrast requirement of but is met by negation in the second conjunct (cf. the impossible *Abby claimed Ben would ask her out, but Bill). Thus, the embedded clause must be an integral part of the sentence and cannot be parenthetical.

3 Following Merchant (2003), I assume that two conjuncts with parallel traces, which are bound by different antecedents outside the conjuncts, qualify as identical for the purpose of ellipsis. That is, the TP [TP tAbbyspeaks passable Dutch] in (1ce)/(16ab) is an appropriate antecedent for the elided TP [TP tBenspeaks passable Dutch].

4 See, among others, Pesetsky and Torrego 2001, 2004, 2007 and Bošković and Lasnik 2003, for the empty complementizer view; and Hegarty 1991, Webelhuth 1992, Doherty 1993, 1997, 2000, Svenonius 1994, Bošković 1997, Franks 2005, and Wurmbrand 2014, for a CP-less approach.

5 Unlike in accounts along the lines of Merchant’s (2003) or Johnson’s (2009), stripping is not dependent on a conjunction in the current approach. This is motivated by the possibility of embedded stripping where no conjunction is present. A reviewer notes that this may be too unrestrictive since but seems to be necessary in cases such as (i); omitting but as in (ii) is impossible. However, the contrast between (i) and (ii) does not necessarily point to the conclusion that stripping needs to be licensed by a conjunction. As shown in (iii), which differs from (ii) only in the polarity of the second sentence, stripping is possible when negation is omitted, despite the absence of any conjunction. Thus, the difference between (ii) and (iii) seems to indicate that a polarity change strongly favors an overt marking—for example, with a connecting but ( possibly for pragmatic reasons). However, this property is not a necessary factor in the formal licensing of stripping.

  • (i)

    Abby claimed that Ben would ask her out but Mary didn’t think Bill too.

  • (ii)

    *Abby claimed that Ben would ask her out. Mary didn’t think Bill too.

  • (iii)

    Abby claimed that Ben would ask her out. Mary thinks Bill too.

6 Jason Merchant ( pers. comm.) suggests that perhaps judgments are influenced by the relative proportion of CP vs. DP complements found with different verbs: verbs that bias for DP complements (see, e.g., Gahl and Garnsey 2004, Gahl, Jurafsky, and Roland 2004) are less likely to allow stripping (which requires a clausal complement). If this suggestion is on the right track, the account proposed here can be maintained as is: complements that allow an embedded that-less clause allow stripping in principle, but other extragrammatical (e.g., frequency) factors are responsible for the bias in the distribution.

7 Note that the Sluicing-Comp Generalization holds even in languages in which an embedded interrogative can cooccur with a filled C (as a reviewer points out, (20a) is not the best example to illustrate the generalization, since complementizers can never cooccur with a wh-phrase in modern English).

8 I thank a reviewer for emphasizing this point.

9 Embedded wh-constructions with topics are generally accepted. Embedded wh-constructions with lower foci and inversion are marked for most speakers; however, for many they are not impossible. (The speakers I consulted showed various degrees of acceptance, (22a) with why being the best liked (but see section 4) and (22c) the least liked.) Culicover (1996:456, (37a–b)) also gives relevant examples with relative clauses: Terry is the person to whom only books like these would I give; Terry is the person for whom on my vacation not even a postage stamp did I remember to buy.

10 A disadvantage of this approach noted by a reviewer is that it may seem like a conspiracy that the phase head that triggers zero Spell-Out is always empty. To avoid this issue, an alternative to the split-CP approach in (25b) would be to manipulate the way Spell-Out operates. A specific mechanism that has been proposed by Ott (2011) and Arano (2014a,b) is to define SODs in such a way that, at least in certain configurations, the phase head spells out together with its complement. In other words, if sluicing and stripping could be defined as configurations in which C′ counts as the SOD, rather than just TP, the Sluicing/Stripping-Comp Generalization could be derived. In the works cited, the empirical domains for such a Spell-Out extension, as well as the motivation for it, are different. However, if a general and uniform algorithm could be developed for when a phase head is spelled out with its complement and when it is grouped with the higher SOD, this approach would provide a very attractive account for the Sluicing/Stripping-Comp Generalization.

11 An alternative account may be to assume that a focus XP (but not a topic XP) blocks wh-movement across it. However, as we saw in section 3, Culicover (1991, 1996) and Haegeman (2000a) provide cases (which have been confirmed by native speakers I consulted) where wh-movement is compatible across lower foci. Furthermore, wh-movement across topics is possible; thus, if there is an intervention effect, it would need to be restricted to inversion cases. The zero Spell-Out approach proposed here excludes stripping in wh-contexts independently of whether wh-movement across a focus XP is possible or not, thus allowing wh > Foc inversion cases (for speakers who allow those constructions) but still accounting for the stripping restriction in (32).

12 Like because, although (see Huddleston and Pullum 2002:735–736) can be used in different ways. In one of its functions, although is a subordinator, which I assume is not what we find in cases like (37). Like adverbial clauses introduced by because (expressing causation), concessive although can also trigger embedded root phenomena in the clause it combines with (see Wegener 1993, Heycock 2006). In German, obwohl ‘although’ can combine with a verbsecond clause, and in English, too, although can introduce a root question (e.g., Some people want me to learn Greek, although why should I?; Agbayani and Zoerner 2004:205). For that reason, Agbayani and Zoerner (2004) assume that although is a CP-adjoined adverb. Huddleston and Pullum (2002:1321) further note that although can link finite VPs (They both remembered Jane, though rarely spoke of her), in which case the second “conjunct” cannot occur in initial position (*They both, though rarely spoke of Jane, remembered her), and they state that such cases “might be regarded as a marginal coordinative construction” (p. 1321).

I leave open here whether although is a coordinator syntactically or an adverbial adjoined to the highest projection of the clause it combines with. Under either view, the XP combining with although (FocP in (37b)) counts as a phase, and hence stripping of its complement will be allowed.

13 The contrast requirement can also be met contextually (without negation in either conjunct) if the context supports an ‘only’ interpretation of the first conjunct, which can then be contrasted with too in the second conjunct (e.g., Who speaks Dutch? Abby speaks Dutch. Oh, but Ben speaks Dutch, too). For the stripping cases considered here, these contexts are not relevant.

14 A further interesting construction pointed out by Masaya Yoshida (pers. comm.) and a reviewer involves elliptical if-statements. Stripping appears to be possible in “if not” contexts: for example, John will invite someone to the party,but if not Mary, I’d be surprised. Negation seems to play an important role in this construction since it is often obligatory, which may point to a tight connection between if and not, at least in certain cases (cf. *If now, when else should we eatfugu?, *If Ralph, who else should we invite?; Jason Merchant, pers. comm.). Moreover, the if-clause can combine with an interrogative root clause (e.g., If not now, (then) when should we eat fugu?), which may indicate a higher-level structure (above CP).

I tentatively suggest that (at least some of) these constructions involve, not stripping, but a concealed copula/cleft construction, similar to Van Craenenbroeck’s (2010a) pseudosluicing. This is supported by cases that do not appear to involve ellipsis but are readily accounted for under a “pseudostripping” analysis (e.g., We need a new representative.If not John, then who will we get to do this?). However, as Masaya Yoshida notes (pers. comm.), a concealed cleft structure may not always be available—for instance, in sprouting contexts. I have to leave a full analysis of if-stripping for further research.

15 I thank Luis Vicente for drawing my attention to Spanish. Morgan (1973:749n5) states that that-stripping is possible in Spanish and Albanian; furthermore, Van Craenenbroeck and Lipta´k (2006, 2008) note that, besides Hungarian, that-stripping is possible in Basque, Polish, Russian, and Hebrew.

16 All examples in (43) become grammatical with pseudogapping, which in the current approach would involve focus movement to a low focus projection in the lower clausal cycle. The (low) FocP closes off the extended vP phase, making the vP a SOD and hence elidable (see Merchant 2007, 2008, Bošković 2014).

References

References
Agbayani,
Brian
, and
Ed
Zoerner
.
2004
.
Gapping, pseudogapping and sideward movement
.
Studia Linguistica
58
:
185
211
.
Arano,
Akihiko
.
2014a
.
Agreement and spell-out of “intermediate projections.”
Ms., University of Connecticut, Storrs
.
Arano,
Akihiko
.
2014b
.
Spell-out of intermediate projections
. In
Papers from the 14th annual conference of the English Linguistic Society of Japan (JELS 31)
,
3
9
.
Baltin,
Mark
.
2010
.
The nonreality of doubly filled Comps
.
Linguistic Inquiry
41
:
331
335
.
Bobaljik,
Jonathan David
, and
Susi
Wurmbrand
.
2005
.
The domain of agreement
.
Natural Language and Linguistic Theory
23
:
809
865
.
Bobaljik,
Jonathan David
, and
Susi
Wurmbrand
.
2015
. Questions with declarative syntax tell us what about selection? In
50 years later: Reflections on Chomsky’s Aspects
, ed. by
Ángel J.
Gallego
and
Dennis
Ott
,
13
32
.
MIT Working Papers in Linguistics 77
.
Cambridge, MA
:
MIT, MIT Working Papers in Linguistics
.
Bošković,
Željko
.
1997
.
The syntax of nonfinite complementation: An economy approach
.
Cambridge, MA
:
MIT Press
.
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
.
Bošković,
Željko
, and
Howard
Lasnik
.
2003
.
On the distribution of null complementizers
.
Linguistic Inquiry
34
:
527
546
.
Bromberger,
Sylvain
.
1992
.
On what we know we don’t know: Explanation, theory, linguistics, and how questions shape them
.
Chicago
:
University of Chicago Press
.
Castillo,
Juan Carlos
,
John E.
Drury
, and
Kleanthes K.
Grohmann
.
2009
.
Merge over Move and the Extended Projection Principle: MOM and the EPP revisited
.
Iberia
1
:
53
114
.
Chomsky,
Noam
.
1973
. Conditions on transformations. In
A festschrift for Morris Halle
, ed. by
Stephen R.
Anderson
and
Paul
Kiparsky
,
232
286
.
New York
:
Holt, Rinehart and Winston
.
Chomsky,
Noam
.
1998
. Minimalist inquiries: The framework.
MIT Occasional Papers in Linguistics 15
.
Cambridge, MA
:
MIT, MIT Working Papers in Linguistics
.
Chomsky,
Noam
.
2000
. Minimalist inquiries: The framework. In
Step by step: Essays on Minimalist syntax in honor of Howard Lasnik
, ed. by
Roger
Martin
,
David
Michaels
, and
Juan
Uriagereka
,
89
155
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
2001
. Derivation by phase. In
Ken Hale: A life in language
, ed. by
Michael
Kenstowicz
,
1
52
.
Cambridge, MA
:
MIT Press
.
Collins,
Chris
.
1991
. Why and how come. In
More papers on wh-movement
, ed. by
Lisa
Cheng
and
Hamida
Demirdache
,
31
45
.
MIT Working Papers in Linguistics 15
.
Cambridge, MA
:
MIT, MIT Working Papers in Linguistics
.
Craenenbroeck,
Jeroen van
.
2010a
.
Invisible Last Resort: A note on clefts as the underlying source for sluicing
.
Lingua
120
:
1714
1726
.
Craenenbroeck,
Jeroen van
.
2010b
.
The syntax of ellipsis: Evidence from Dutch dialects
.
New York
:
Oxford University Press
.
Craenenbroeck,
Jeroen van
, and
Anikó
Lipták
.
2006
.
The cross-linguistic syntax of sluicing: Evidence from Hungarian relatives
.
Syntax
9
:
248
274
.
Craenenbroeck,
Jeroen van
, and
Anikó
Lipták
.
2008
. On the interaction between verb movement and ellipsis: New evidence from Hungarian. In
Proceedings of the 26th West Coast Conference on Formal Linguis- tics
, ed. by
Charles B.
Chang
and
Hannah J.
Haynie
,
138
146
.
Somerville, MA
:
Cascadilla
.
Culicover,
Peter W
.
1991
. Topicalization, inversion and complementizers in English. In
OTS working pa- pers: Going Romance and beyond
, ed. by
Denis
Delfitto
,
Martin
Everaert
,
Arnold
Evers
, and
Frits
Stuurman
,
1
45
.
Utrecht
:
Utrecht University
.
Culicover,
Peter W
.
1996
.
On distinguishing A¯ -movements
.
Linguistic Inquiry
27
:
445
463
.
Doherty,
Cathal
.
1993
.
Clauses without that: The case for bare sentential complementation in English
.
Doctoral dissertation, University of California, Santa Cruz
.
Doherty,
Cathal
.
1997
.
Clauses without complementizers: Finite IP-complementation in English
.
The Linguis- tic Review
14
:
197
220
.
Doherty,
Cathal
.
2000
.
Clauses without ‘that’: The case for bare sentential complementation in English
.
New York
:
Garland
.
Fiengo,
Robert
, and
Robert
May
.
1994
.
Indices and identity
.
Cambridge, MA
:
MIT Press
.
Franks,
Steven
.
2005
. What is that? In
Indiana University working papers in linguistics 5.2
, ed. by
Yoshihisa
Kitagawa
and
Dorian
Roehrs
,
33
62
. Available at https://www.indiana.edu/~iulcwp/wp/article/view/05-10A.
Gahl,
Susanne
, and
Susan M.
Garnsey
.
2004
.
Knowledge of grammar, knowledge of usage: Syntactic proba- bilities affect pronunciation variation
.
Language
80
:
748
775
.
Gahl,
Susanne
,
Daniel
Jurafsky
, and
Douglas
Roland
.
2004
.
Verb subcategorization frequencies: American English corpus data, methodological studies, and cross-corpus comparisons
.
Behavior Research Meth- ods, Instruments, and Computers
36
:
432
443
.
Gallego,
Ángel
.
2010
.
Phase theory
.
Amsterdam
:
John Benjamins
.
Gengel,
Kirsten
.
2006
. Phases and ellipsis. In
NELS 37
, ed. by
Emily
Elfner
and
Martin
Walkow
,
233
246
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Gengel,
Kirsten
.
2009
.
Phases and ellipsis
.
Linguistic Analysis
35
:
21
42
.
Ginzburg,
Jonathan
, and
Ivan A.
Sag
.
2000
.
Interrogative investigations: The form, meaning, and use of English interrogatives
.
Stanford, CA
:
CSLI Publications
.
Grano,
Thomas
, and
Howard
Lasnik
.
2015
.
How to neutralize a finite clause boundary: Phase theory and the grammar of bound pronouns
.
Ms., Indiana University, Bloomington, and University of Maryland, College Park
.
Grice,
H. P
.
1975
. Logic and conversation. In
Speech acts
, ed. by
Peter
Cole
and
Jerry L.
Morgan
,
41
58
.
New York
:
Academic Press
.
Grimshaw,
Jane
.
1991
.
Extended projection
.
Ms., Brandeis University, Waltham, MA
.
Grohmann,
Kleanthes K.
,
John E.
Drury
, and
Juan Carlos
Castillo
.
2000
. No more EPP. In
Proceedings of the 19th West Coast Conference on Formal Linguistics
, ed. by
Roger
Billerey
and
Brook Danielle
Lillehaugen
,
153
166
.
Somerville, MA
:
Cascadilla
.
Haegeman,
Liliane
.
2000a
.
Inversion, non-adjacent inversion, and adjuncts to CP
.
Transactions of the Philo- logical Society
98
:
121
160
.
Haegeman,
Liliane
.
2000b
. Negative preposing, negative inversion, and the split CP. In
Negation and polar- ity: Syntactic and semantic perspectives
, ed. by
Laurence
Horn
and
Yasuhiko
Kato
,
21
61
.
Oxford
:
Oxford University Press
.
Hankamer,
Jorge
.
1979
.
Deletion in coordinate structures
.
New York
:
Garland
.
Hankamer,
Jorge
, and
Ivan
Sag
.
1976
.
Deep and surface anaphora
.
Linguistic Inquiry
7
:
391
426
.
Harwood,
William
.
2015
.
Being progressive is just a phase: Celebrating the uniqueness of progressive aspect under a phase-based analysis
.
Natural Language and Linguistic Theory
33
:
523
573
.
Hegarty,
Michael
.
1991
.
Adjunct extraction and chain configurations
.
Doctoral dissertation, MIT, Cambridge, MA
.
Heycock,
Caroline
.
2006
. Embedded root phenomena. In
The Blackwell companion to syntax
, ed. by
Martin
Everaert
and
Henk
van Riemsdijk
,
2
:
174
209
.
Oxford
:
Blackwell
.
Huddleston,
Rodney
, and
Geoffrey K.
Pullum
.
2002
.
The Cambridge grammar of the English language
.
Cambridge
:
Cambridge University Press
.
Johnson,
Kyle
.
2001
. What VP ellipsis can do, and what it can’t, but not why. In
The handbook of contemporary syntactic theory
, ed. by
Mark
Baltin
and
Chris
Collins
,
439
479
.
Oxford
:
Blackwell
.
Johnson,
Kyle
.
2009
.
Gapping is not (VP-) ellipsis
.
Linguistic Inquiry
40
:
289
328
.
Klima,
Edward
.
1964
. Negation in English. In
The structure of language: Readings in the philosophy of language
, ed. by
Jerry A.
Fodor
and
Jerrold J.
Katz
,
246
323
.
Englewood Cliffs, NJ
:
Prentice-Hall
.
Ko,
Heejeong
.
2005
.
Syntax of why-in-situ: Merge into [Spec,CP] in the overt syntax
.
Natural Language and Linguistic Theory
23
:
867
916
.
Koutsoudas,
Andreas
.
1971
.
Gapping, conjunction reduction, and coordinate deletion
.
Foundations of Lan- guage
7
:
337
386
.
Lasnik,
Howard
.
1999
.
On feature strength: Three Minimalist approaches to overt movement
.
Linguistic Inquiry
30
:
157
196
.
Lasnik,
Howard
, and
Mamoru
Saito
.
1984
.
On the nature of proper government
.
Linguistic Inquiry
15
:
235
289
.
Lasnik,
Howard
, and
Mamoru
Saito
.
1992
.
Move α
.
Cambridge, MA
:
MIT Press
.
Legate,
Julie Anne
.
2003
.
Some interface properties of the phase
.
Linguistic Inquiry
34
:
506
516
.
Lobeck,
Anne
.
1995
.
Ellipsis: Functional heads, licensing, and identification
.
New York
:
Oxford University Press
.
Merchant,
Jason
.
2001
.
The syntax of silence: Sluicing, islands, and the theory of ellipsis
.
Oxford
:
Oxford University Press
.
Merchant,
Jason
.
2003
.
Remarks on stripping
.
Ms., University of Chicago, Chicago, IL
.
Merchant,
Jason
.
2004
.
Fragments and ellipsis
.
Linguistics and Philosophy
27
:
661
738
.
Merchant,
Jason
.
2007
.
Voice and ellipsis
.
Ms., University of Chicago, Chicago, IL
.
Merchant,
Jason
.
2008
.
An asymmetry in voice mismatches in VP-ellipsis and pseudogapping
.
Linguistic Inquiry
39
:
169
179
.
Merchant,
Jason
.
To appear
. Ellipsis: A survey of analytical approaches. In
Handbook of ellipsis
, ed. by
Jeroen van
Craenenbroeck
and
Tanja
Temmerman
.
Oxford
:
Oxford University Press
.
Morgan,
Jerry L
.
1973
. Sentence fragments and the notion “sentence.” In
Issues in linguistics: Papers in honor of Henry and René Kahane
, ed. by
Braj
Kachru
,
Robert
Lees
,
Yakov
Malkiel
,
Angelina
Pietrangeli
, and
Sol
Saporta
,
719
751
.
Champaign
:
University of Illinois Press
.
Nevins,
Andrew
.
2008
.
Sluicing ≠ stripping: Evidence from P-stranding
.
Paper presented at the 3rd Annual Moscow Student Conference on Linguistics, Moscow
.
Ortega-Santos,
Iván
,
Masaya
Yoshida
, and
Chizuru
Nakao
.
2014
.
On ellipsis structures involving a wh- remnant and a non-wh-remnant simultaneously
.
Lingua
138
:
55
85
.
Ott,
Dennis
.
2011
.
A note on free relative clauses in the theory of phases
.
Linguistic Inquiry
42
:
183
192
.
Pesetsky,
David
, and
Esther
Torrego
.
2001
. T-to-C movement: Causes and consequences. In
Ken Hale: A life in language
, ed. by
Michael
Kenstowicz
,
355
426
.
Cambridge, MA
:
MIT Press
.
Pesetsky,
David
, and
Esther
Torrego
.
2004
. Tense, case, and the nature of syntactic categories. In
The syntax of time
, ed. by
Jacqueline
Guéron
and
Jacqueline
Lecarme
,
495
537
.
Cambridge, MA
:
MIT Press
.
Pesetsky,
David
, and
Esther
Torrego
.
2006
. Probes, goals and syntactic categories. In
Proceedings of the 7th annual Tokyo Conference on Psycholinguistics
, ed. by
Yukio
Otsu
,
25
60
.
Tokyo
:
Hituzi Syobo
.
Pesetsky,
David
, and
Esther
Torrego
.
2007
. The syntax of valuation and the interpretability of features. In
Phrasal and clausal architecture
, ed. by
Simin
Karimi
,
Vida
Samiian
, and
Wendy
Wilkins
,
262
294
.
Amsterdam
:
John Benjamins
.
Rizzi,
Luigi
.
1990
.
Relativized Minimality
.
Cambridge, MA
:
MIT Press
.
Rizzi,
Luigi
.
1997
. The fine structure of the left periphery. In
Elements of grammar: Handbook of generative syntax
, ed. by
Liliane
Haegeman
,
281
337
.
Dordrecht
:
Kluwer
.
Rouveret,
Alain
.
2012
.
VP ellipsis, phases and the syntax of morphology
.
Natural Language and Linguistic Theory
30
:
897
963
.
Sag,
Ivan A
.
1976
.
Deletion and Logical Form
.
Doctoral dissertation, MIT, Cambridge, MA
.
Sauerland,
Uli
.
2003
.
Intermediate adjunction with A-movement
.
Linguistic Inquiry
34
:
308
314
.
Svenonius,
Peter
.
1994
.
Dependent nexus: Subordinate predication structures in English and the Scandinavian languages
.
Doctoral dissertation, University of California, Santa Cruz
.
Toosarvandani,
Maziar
.
2013
.
Corrective but coordinates clauses not always but sometimes
.
Natural Language and Linguistic Theory
31
:
827
863
.
Toosarvandani,
Maziar
.
2014
.
Contrast and the structure of discourse
.
Semantics and Pragmatics
7
:
1
57
.
Villa-García,
Julio
.
2012a
. Characterizing medial and low complementizers in Spanish: Recomplementation que and “jussive/optative” que. In
Current formal aspects of Spanish syntax and semantics
, ed. by
Melvin
González-Rivera
and
Sandro
Sessarego
,
198
228
.
Cambridge
:
Cambridge Scholars Publishing
.
Villa-García,
Julio
.
2012b
.
Recomplementation and locality of movement in Spanish
.
Probus
24
:
257
314
.
Webelhuth,
Gert
.
1992
.
Principles and parameters of syntactic saturation
.
Oxford
:
Oxford University Press
.
Wegener,
Heide
.
1993
.
Weil—das hat schon seinen Grund: Zur Verbstellung in Kausalsätzen mit weil im gegenwärtigen Deutsch
.
Deutsche Sprache
21
:
289
305
.
Wilder,
Chris
.
1994
. Coordination, ATB, and ellipsis. In
Minimalism and Kayne’s asymmetry hypothesis
, ed. by
C.
Jan-Wouter Zwart
,
291
331
.
Groninger Arbeiten zur germanistischen Linguistik 37
.
Groningen
:
University of Groningen
.
Williams,
Edwin
.
1997
.
Blocking and anaphora
.
Linguistic Inquiry
28
:
577
628
.
Wurmbrand,
Susi
.
2013
. QR and selection: Covert evidence for phasehood. In
NELS 42
, ed. by
Stefan
Keine
and
Shayne
Sloggett
,
2
:
277
290
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Wurmbrand,
Susi
.
2014
. The Merge Condition: A syntactic approach to selection. In
Minimalism and beyond: Radicalizing the interfaces
, ed. by
Peter
Kosta
,
Lilia
Schürcks
,
Steven
Franks
, and
Teodora
Radev- Bork
,
139
177
.
Amsterdam
:
John Benjamins
.
Wurmbrand,
Susi
.
2015
. Complex predicate formation via voice incorporation. In
Approaches to complex predicates
, ed. by
Léa
Nash
and
Pollet
Samvelian
,
248
290
.
Leiden
:
Brill
.
Wurmbrand,
Susi
, and
Youssef A.
Haddad
.
2016
. Cyclic spell-out derived agreement in Arabic raising constructions. In
Perspectives on Arabic Linguistics XXVIII
, ed. by
Youssef A.
Haddad
and
Eric
Potsdam
,
193
226
.
Amsterdam
:
John Benjamins
.
Wurmbrand,
Susi
, and
Koji
Shimamura
.
To appear
. The features of the voice domain: Actives, passives, and restructuring. In
The verbal domain
, ed. by
Roberta
D’Alessandro
,
Irene
Franco
, and
Ángel
Gallego
.
Oxford
:
Oxford University Press
.
Yoshida,
Masaya
,
Chizuru
Nakao
, and
Iván
Ortega-Santos
.
2015
.
The syntax of why-stripping
.
Natural Language and Linguistic Theory
33
:
323
270
.