Abstract

In this article, we argue for the existence of covert pied-piping in wh-questions through a previously unnoticed pattern of intervention effects in Superiority-obeying English multiple wh-questions. We show that the preference of covert pied-piping, unlike that of overt pied-piping, is for movement of larger constituents. We argue that this discrepancy stems from conflicting requirements of PF and LF: overt pied-piping feeds both LF and PF, but covert pied-piping feeds LF only. The study of covert pied-piping thus reveals the true preference of LF and narrow syntax with regard to pied-piping: larger pied-piping constituents are preferred over smaller ones. This preference can be overridden by certain PF constraints that apply to overt pied-piping.

1 Introduction

Much recent work on syntactic and semantic theory has debated the status of covert movement in grammar. At the same time, much empirical work has been done to better describe and explain the pied-piping observed with overt wh-movement. In this article, we ask the question, Does covert wh-movement also trigger pied-piping? We argue that the existence of covert pied-piping can be detected through the use of focus intervention effects (Sauerland and Heck 2003, Beck 2006, Beck and Kim 2006, Grohmann 2006, Cable 2007, Tomioka 2007, Mayr 2010, Kotek 2014a). We present a novel pattern of focus intervention effects in Superiority-obeying English multiple wh-questions, which we argue is evidence for the existence of covert pied-piping.

Adopting the theory that focus intervention effects occur when the interpretation of a wh-phrase is interrupted by certain focus-sensitive interveners (Beck 2006), we show that the logic of intervention can be used to diagnose areas in the derivation in which Rooth-Hamblin focus alternatives are computed. We then use this diagnostic to investigate the status of the ( phonologically) in-situ wh-phrase in Superiority-obeying English wh-questions such as those in (1). In such questions, it has been argued that the lower wh covertly moves to C by LF (Karttunen 1977, Huang 1982, Hagstrom 1998, Pesetsky 2000, Beck 2006, Cable 2007, 2010).

(1) No causes intervention effect in Superiority-obeying multiple wh-question

  • a.

    Which student read a book from which library?

  • b.

    *Which student read no book from which library?

Building on work by Sauerland and Heck (2003), Cable (2007) shows that focus intervention effects occur inside overtly pied-piped constituents. We argue that the ungrammaticality of (1b) is the result of a focus intervention effect inside a covertly pied-piped constituent. That is, the surface in-situ wh-word which covertly moves at LF and pied-pipes a constituent containing the intervener no. The contrast in (1) and other examples presented in section 3 is neatly predicted by the semantics of pied-piping constituents proposed by Cable (2007), together with Beck’s (2006) system of focus intervention.

Furthermore, we demonstrate that the constituent pied-piped with wh corresponds to the largest among the corresponding overt pied-piping possibilities for a parallel single wh-question. This, we argue, reveals the true nature of pied-piping: unlike overt pied-piping, which is subject to restrictions from both LF and PF, covert pied-piping is guided only by LF. The investigation of covert pied-piping, then, shows the true preference of LF and narrow syntax with regard to pied-piping: LF chooses the largest possible constituent as its preferred candidate for movement. In overt pied-piping, however, this choice can be overridden by certain PF constraints (Heck 2008, 2009).

The article is structured as follows. In section 2, we provide background on Beck’s (2006) theory of focus intervention effects in wh-questions and on its extension to overt pied-piping constructions (Cable 2010). We present the use of focus intervention effects as a diagnostic for covert movement vs. Rooth-Hamblin focus alternative computation. In section 3, we show that, assuming that intervention correlates with focus alternative computation, the data motivate the existence of covert pied-piping. We show that covert wh-movement prefers to pied-pipe larger constituents, in contrast to overt wh-movement, which tends to prefer pied-piping smaller constituents. In section 4, we argue that the different preferences for overt vs. covert pied-piping result from conflicting constraints on the PF and LF interfaces: the true preference of LF and narrow syntax is for larger pied-piping constituents, but this can be overridden by PF constraints, often resulting in smaller overt pied-piping. Section 5 is the conclusion.

2 Background: Intervention Effects in Overt Pied-Piping

In this section, we present Beck’s (2006) theory of focus intervention effects for English questions and its extension to intervention inside pied-piping constituents (Cable 2007, 2010). We will motivate the use of intervention effects as a diagnostic tool for regions in the derivation of a question in which covert movement has occurred and regions in which Rooth-Hamblin alternatives are computed. We will then use this diagnostic to address the question, Does covert pied-piping exist and, if so, how large is it? Answering that question will be the main goal of section 3.

2.1 Beck’s (2006) Theory of Focus Intervention Effects

Pesetsky (2000) observes a correlation between superiority (e.g., Kuno and Robinson 1972, Chomsky 1973) and intervention effects in English multiple wh-questions with D-linked wh-phrases. Motivating examples are given in (2). Examples (2a–b) show that English generally allows both Superiority-obeying questions and Superiority-violating questions with D-linked wh-phrases. Examples (2c–d) show that when certain interveners—here, negation—are introduced into the questions, only the Superiority-obeying structure is grammatical. The Superiority-violating question is ungrammatical.1

(2)

  • D-linked questions can violate Superiority but Superiority-violating questions are sensitive to intervention

  • graphic

Other operators that give rise to an intervention effect include only, very few, never, and no one, as shown in (3)–(6) from Pesetsky 2000:61.2

(3) Intervention effect with only only affects Superiority-violating question

  • a

    Which girl did only Mary introduce ___ to which boy?

  • b

    *Which boy did only Mary introduce which girl to ___?

(4) Intervention effect with very few only affects Superiority-violating question

  • a

    Which picture did very few children want to show ___ to which teacher?

  • b

    *Which teacher did very few children want to show which picture to ___?

(5) Intervention effect with never only affects Superiority-violating question

  • a

    Which student did he never claim ___ would talk about which topic?

  • b

    *Which topic did he never claim which student would talk about ___?

(6) Intervention effect with no one only affects Superiority-violating question

  • a

    Which book did no one give to which student?

  • b

    *Which student did no one give which book to ___?

To explain the relation between superiority and intervention effects, Pesetsky (2000) argues that Superiority-obeying questions and Superiority-violating questions are derived from different structures. Of particular importance is the location of the ( phonologically) in-situ wh-phrase: Pesetsky proposes that the in-situ wh-phrase covertly moves to C at LF in a Superiority-obeying question, but remains in its base-generated position in a Superiority-violating question. The resulting structures are given in (7a–b).

(7)

  • LF representations of Superiority-obeying and Superiority-violating questions

  • graphic

When the ( phonologically) in-situ wh-phrase does not undergo covert movement (wh1 in (7b)), it is interpreted via Rooth-Hamblin alternative computation (e.g., Hamblin 1973, Kratzer and Shimoyama 2002).3 These two strategies for establishing a relation between the interrogative C and wh and assigning wh a semantics are schematized in (8a) and (8b), respectively. Both strategies have been independently proposed in the literature for the analysis of ( phonologically) in-situ wh-phrases and are widely used in current research on the syntax and semantics of multiple wh-questions.

(8)

  • Interpretation of wh using covert movement and using Rooth-Hamblin alternatives

  • graphic

Rooth-Hamblin alternatives are a parallel mode of semantic interpretation, where a focus-semantic value can be computed compositionally for each syntactic node, in parallel to its ordinary semantic value (Hamblin 1973, Rooth 1985, 1992). This computation has been argued to supply operators such as focus operators and question complementizers with a relevant set of alternative denotations or propositions. Consider the LF representation for the wh-in-situ pseudo-English question Alex likeswho? in (9). Focus-semantic values—also referred to as alternatives—are given for each node.

(9)

  • Toy LF of question interpretation through Rooth-Hamblin alternative computation

  • graphic

In (9), the wh-phrase who has a focus-semantic value corresponding to relevant individuals in its domain—here, the animate individuals Bobby, Chris, and Dana.4 These alternatives compose pointwise at each nonterminal node,5 with the result that the complement of the interrogative C has a set of propositions as its focus-semantic value.6 The interrogative C then computes the question denotation using these alternatives in its complement, so that these alternative propositions correspond to possible (weak) answers to the question. In this way, the focus-semantic value provided by the in-situ wh-phrase is interpreted by the interrogative C. This yields the appropriate question semantics without establishing a syntactically local relationship between the wh-phrase and C.

Beck (2006) argues that the Rooth-Hamblin alternative computation strategy of interpreting wh-phrases is sensitive to focus intervention effects.

(10)

  • Description of focus intervention effect7

  • When a focus-sensitive operator occurs between an LF in-situ wh-word and its associated complementizer, the operator disrupts the projection of the wh-word’s alternatives. As a result, the wh-word cannot be interpreted by C, and the derivation crashes.

This illicit configuration is schematized in (11b). In essence, the problem is one of Minimality. The alternatives projected from the wh-word must be interpreted by the associated complementizer. However, focus-sensitive operators also interpret alternatives in their scope (Rooth 1985, 1992). When a focus-sensitive operator (represented as intervener) intervenes between the in-situ wh-word and the interpreting complementizer, the alternatives projected by the wh-word will be interpreted by the intervener instead of by C, thus disrupting the question interpretation.8

(11)

  • Focus alternative computation, but not covert movement, is sensitive to intervention effects

  • graphic

The covert movement strategy of interpreting wh-phrases, on the other hand, is immune from intervention effects: intervention only affects wh-phrases that project focus alternatives but not traces of wh-movement (11a).

Beck’s (2006) schema for intervention effects between an in-situ wh-phrase and C is summarized in (12). Given this schema for intervention effects, we predict such effects to arise in Superiority-violating questions when an intervener is introduced between the in-situ wh-word and C. We expect not to find intervention effects for Superiority-obeying questions, since all wh-phrases in such questions move to C by LF. These structures are illustrated in (13). This is consistent with the findings reported by Pesetsky (2000).

(12)

  • Intervention schema with in-situ wh-phrases (Beck 2006)

  • LF: *[C . . . intervener . . . wh]

(13)

  • Interaction of superiority and interveners at LF (cf. (7))

  • graphic

Following this logic, focus intervention effects can diagnose whether or not covert wh-movement has occurred in the derivation of a question: the presence of an intervention effect shows that a (phonologically) in-situ wh-phrase is interpreted through Rooth-Hamblin alternative computation, and the lack of an intervention effect shows that the wh-phrase has covertly moved above the scope of the intervener.9

2.2 Intervention Effects in Pied-Piping Constituents and Cable’s Theory of Pied-Piping

Cable (2007, 2010) shows that intervention effects also occur in Superiority-obeying English questions inside pied-piping constituents (14). Cable’s work builds on a study by Sauerland and Heck (2003), who show a similar effect in German (15).

(14) Intervention effect in English pied-piping (based on Cable 2007:262)

  • a.

    ?[A picture of which president] does Jim own _____?

  • b.

    *[No pictures of which president] does Jim own _____?

  • c.

    *[Few pictures of which president] does Jim own _____?

  • d.

    *[Only PICTURES of which president] does Jim own _____?

(15)

  • Intervention effect in German pied-piping (Sauerland and Heck 2003:349)

  • Fritz möchte wissen [[ein/*keinwie schnelles Motorrad] du fahren darfst].

  • Fritz wants know a/*no how fast motorbike you drive may

  • ‘Fritz wants to know how fast a/*no motorbike you are allowed to drive.’

Cable (2007, 2010) argues that the presence of intervention effects inside pied-piping constituents is explained if wh-words inside such constituents are interpreted via Rooth-Hamblin alternative computation between wh and the edge of the pied-piping. Cable’s theory is designed to explain the presence and size of pied-piping in wh-questions in different languages. In this theory, wh-movement is recast as QP-movement: in the construction of a wh-question, a Q-particle (silent in English but overt in some languages) is adjoined to a particular wh-containing constituent. The resulting QP is targeted by the interrogative C for Agree and Attract operations. Movement of a QP containing more than just the wh-word leads to what has previously been described as pied-piping.

The size of pied-piping is then determined by the position to which Q adjoins. In (16a), Q adjoins to the wh-containing DP a picture of which president; in (16b), Q adjoins to the PP of which president; in (16c), Q adjoins to the which-phrase DP which president. The positions to which Q can adjoin are restricted in some languages by a local Agree operation between wh and In English, Q cannot adjoin to constituents larger than the option in (16a).10

(16) Different sizes of pied-piping correspond to different positions of Q-adjunction Base structure: C Jim owns (Q) a picture (Q) of (Q) which president

  • a.

    ?[QP Q A picture of which president] does Jim own______?

  • b.

    [QP Q Of which president] does Jim own a picture______?

  • c.

    [QP Q Which president] does Jim own a picture of______?

In the derivation of a question like (16), then, two steps occur: first, QP moves to the specifier of the interrogative complementizer, and second, inside QP, the wh-word itself is interpreted via Rooth-Hamblin alternative computation between wh and Q.

(17)

graphic

Building on Beck 2006, Cable argues that this alternative computation inside QP is sensitive to intervention effects, as in the schema in (18). That is, if a focus-sensitive operator occurs between wh and Q, it will interrupt the projection of alternatives from wh to Q and lead to the same problem that is caused if an intervener occurs between an LF in-situ wh and C (the Beck schema, (12)).11

(18) Intervention schema for wh-pied-piping (Cable 2007; cf. Sauerland and Heck 2003)

*[QP Q ... intervener ... wh ... ]i ... ti ...

Cable argues that the configuration in (18) explains the ungrammaticality of examples (14b–d): an intervener—no, few, or only—occurs between wh and the edge of the pied-piping constituent, where Q is merged, and blocks the transmission of Rooth-Hamblin alternatives between wh and Q.

Cable offers further evidence to show that (18) is the relevant structural configuration for these contrasts—namely, that an intervener between the wh and the edge of the pied-piping (Q) is the source of ungrammaticality in (14b–d). For example, the intervention effect in (14b) can be avoided by pied-piping a smaller constituent that does not include the potential intervener (19). In addition, placing an intervener inside QP but below the wh-word does not trigger an intervention effect (20).

(19) No intervention effect if intervener is not inside pied-piping constituent

  • a.

    *[QP Q No pictures of which president] does Jim own____? ( = (14b))

  • b.

    [QP Q Of which president] does Jim own no pictures____?

  • c.

    [QP Q Which president] does Jim own no pictures of____?

(20)

  • No intervention effect if intervener is inside pied-piping constituent but below wh

  • [Which picture containing no presidents] does Jim own____?

In this article, we investigate the syntax and semantics of multiple wh-questions. Following the work presented in section 2.1, we assume that the ( phonologically) in-situ wh-phrase in a Superiority-obeying question moves to C by LF. The question that we set out to answer here is this: Does covert wh-movement pied-pipe material along with the wh-word? And, if so, how much? We will use the presence or absence of intervention effects inside potentially covertly pied-piped constituents as a diagnostic for regions where covert movement has occurred and regions where Rooth-Hamblin alternatives are computed. We elaborate on this logic in section 3, where we present a novel pattern of intervention effects in English Superiority-obeying questions.

3 Diagnosing Covert Pied-Piping

In this section, we will argue that the pattern of focus intervention effects in English multiple wh-questions shows that covert pied-piping exists, and furthermore that covert pied-piping of larger constituents is preferred over that of smaller constituents. Our main focus will be on questions such as (21), where we assume that the ( phonologically) in-situ wh-phrase moves to C by LF.12

(21)

graphic

Given that in Superiority-obeying questions the in-situ wh-phrase covertly moves to C by LF, we are interested in knowing whether covert wh-movement pied-pipes material along with the wh-word—and, if so, how much. At least three different sizes of covertly pied-piped constituents could be available in (22), corresponding to the possible sizes of overt pied-piping in examples such as (16).13

(22)

  • Options for covert pied-piping at LF, based on options for overt pied-piping

  • Which student read a book from which library?

  • graphic

The movement of the lower QPs in (22) is covert, and thus all three of the LF representations in (22) correspond to the same PF string: Which student read a book from which library? Unlike with overt pied-piping, the choice of covert pied-piping size is not detectable in the resulting linearization.

We instead propose the use of focus intervention effects as a diagnostic for the existence and size of covert pied-piping. The different sizes of covert pied-piping in (22)—corresponding to different QP sizes—predict different regions that should be sensitive to intervention effects.

The logic of this diagnostic is as follows. As we showed in section 2.2, in cases of overt pied-piping, the region inside the pied-piping constituent (QP) and above the wh-word is intervention-sensitive; that is, the insertion of a focus-sensitive operator in this region leads to ungrammaticality ((23a); cf. (18)). In Cable’s (2007, 2010) theory for the semantic interpretation of pied-piping constituents, wh-phrases are interpreted by their Q-particle through Rooth-Hamblin alternative computation, thus explaining why the region between the wh and Q is intervention-sensitive under Beck’s (2006) logic of intervention. Following Cable, we assume that covertly moved QPs are also interpreted using this same semantic mechanism. Therefore, if covert pied-piping does occur, we predict a small region near and above the surface-in-situ wh-word to be intervention-sensitive (23b).

(23)

graphic

The different covert pied-piping options in (22) predict different regions to be intervention-sensitive, represented by the gray shading in (24). (In (24) and subsequent examples, the lower QP is represented in situ, to reflect the linearization at PF.) If the largest QP is chosen, (24a), the entire region between a and wh is intervention-sensitive. If the smaller QP in (24b) is chosen, only the preposition from is inside the intervention-sensitive region. If the smallest QP is chosen, (24c), there is little or no intervention-sensitive region between Q and wh.

(24)

  • Different covert pied-piping options predict different intervention-sensitive regions

  • graphic

In what follows, we will use the size of the intervention-sensitive region as a diagnostic for the presence of covert pied-piping and its size. We use the interveners no, only, and very few, placed inside regions suspected of being intervention-sensitive. If the presence of an intervener in our examples causes an intervention effect, the conclusion will be that the intervener occurs inside an intervention-sensitive region. We will show that the intervention-sensitive regions we find occur inside covertly pied-piped constituents—that is, between wh and Q (schema in (18))—and not between an in-situ wh and C (schema in (12)). Since it has been argued that intervention effects only affect the pair-list reading of a question (Beck 1996, Pesetsky 2000, Kotek 2014a), we present each example in a context designed to bring out the pair-list reading of the multiple wh-question and satisfy its presuppositions (Dayal 2002).14

We begin with our baseline example (21), repeated in (25) with a context supporting its pair-list interpretation. The question is reported to be grammatical and felicitous in this context.

(25) Baseline: Multiple wh-question with pair-list reading

  • a.

    Context: Over the break, every student read a book from a local library and submitted a book report. Each report gave the title of the book and which library it was borrowed from. You have read all the book reports. So tell me, what I want to know is:

  • b.

    Which student read a book from which library?

Next, we consider an example in which the determiner a is replaced with the determiner no (26), which we showed to be an intervener in cases of overt pied-piping in section 2.2. Again, we include a context designed to favor the pair-list reading of the question and satisfy its presuppositions. If no occurs inside an intervention-sensitive region, we expect this manipulation to lead to an intervention effect. The result, as reported by native speakers of English, is that the question is ungrammatical despite appearing in a context that should make it felicitous. That is, we observe an intervention effect.15

(26) Intervention effect caused by presence of no in question

  • a.

    Context: Over the break, the students were assigned to go read one book each from every library in the area and submit a book report. No student completed the entire assignment; every student went to all but one of the libraries. You have read all the book reports. So tell me, what I want to know is:

  • b.

    *Which student read no book from which library?

The contrast between (26) and its baseline (25) tells us that the intervener no is in an intervention-sensitive region.16 Recall that different sizes of covert pied-piping predict different intervention-sensitive regions. In particular, no is in an intervention-sensitive region if the largest pied-piping option was chosen: no book from which library (27a). If, however, a smaller pied-piping constituent is targeted for movement—from which library (27b) or which library (27c)—we expect no to be outside the intervention-sensitive region and therefore not cause an intervention effect. Therefore, the intervention contrast observed in (26) is predicted by pied-piping of the largest option as in (27a) but not the smaller options in (27b–c).

(27)

  • Different covert pied-piping options predict different intervention-sensitive regions; only (27a) predicts (26)

  • Which student read no book from which library?

  • graphic

Note that this effect is not a general effect attributable to the presence of a negative quantifier in the question. Example (28), which includes the quantifier less than three, is judged by speakers to be felicitous and grammatical in the context.17

(28) No intervention effect with less than three

  • a.

    Context: Over the break, the students were assigned to go read three books each from every library in the area and submit a book report. No student completed the entire assignment; every student had one particular library from which they failed to read three books. You have read all the book reports. So tell me, what I want to know is:

  • b.

    Which student read less than three books from which library?

Furthermore, an intervention effect can be observed not only with no but also with only and very few, as shown in (29)–(30) (though the judgment with very few appears to be less sharp than the judgments with no and only).18

(29) Intervention effect caused by presence of only in question

  • a.

    Context: At the flea market, a number of collectors are selling pictures and autographs of past presidents. For most presidents, they have successfully sold both pictures and autographs, but according to the records, every collector has one president for which they did not sell any autographs. You have read through all the records. So tell me, what I want to know is:

  • b.

    *Which collector sold only PICTURES of which president?

(30) Intervention effect caused by presence of very few in question

  • a.

    Context: We at Big Kahuna Burger are testing three new toppings for burgers: cranberries, jicama, and natto. As a pilot, these toppings were offered at several branches around the world for one week only. At each branch, only two toppings sold thousands while the other sold about a hundred. Culinary tastes vary across the world, so there was no clear overall winner. You have looked at all the sales records. So tell me, what I want to know is:

  • b.

    ??Which branch sold very few burgers with which topping?

With these results in mind, the main contrasts that we aim to explain are the minimal pairs in (31)–(32). We argue that these contrasts are the result of intervention effects, caused by placing the focus-sensitive interveners no and only between wh and the edge of a covertly pied-piped constituent that is at least as large as the DP headed by no and only. We acknowledge that the judgments behind the contrasts we present here are subtle. This is in part due to the complexity of even our baseline questions, as well as the relatively complex contexts that accompany these questions, in order to ensure that they are interpreted with the relevant (pair-list) reading. Despite this, we have found these contrasts to be robust across a variety of elicitation and verification sessions.

(31)

  • Minimal pair: No causes intervention effect in multiple wh-question

  • graphic

(32)

  • Minimal pair: Only causes intervention effect in multiple wh-question

  • graphic

Note that the ungrammaticality of (31b) is not due to a general negative island effect between the in-situ wh-word and C (e.g., Ross 1984, Rizzi 1990, Szabolcsi and Zwarts 1993, Beck 1995, Rullmann 1995, Kuno and Takami 1997, Fox and Hackl 2006). It has been observed that certain wh-phrases are unable to move across intervening negative elements. For example, in (33), movement of the manner wh-word how is ungrammatical over both sentential negation and the negative determiner no.

(33) Negative islands affecting how-movement (Szabolcsi and Zwarts 1993:239)

  • a.

    *How didn’t you think that I behaved______?

  • b.

    *How did no one think that I behaved______?

If the ungrammaticality of (31b) were due to a negative island, we would expect that any negation between the in-situ wh-phrase and C would trigger a similar negative island effect. This is not the case. In example (34), sentential negation (shown to trigger negative islands in (33a)) does not trigger ungrammaticality, in contrast to (31b). Therefore, the ungrammaticality of (31b) cannot be due to a negative island effect.

(34)

  • Sentential negation does not cause negative island effect (cf. (31b))

  • Which student didn’t read a book from which library?

Similarly, the contrasts in (31)–(32) cannot be due to a general Beck (2006)–style focus intervention effect between the in-situ wh-phrase and C. As we showed in section 2.1, high interveners in Superiority-obeying questions—including other negative interveners such as no one and never—do not cause intervention effects. As a result, it cannot be the case that the surface in-situ wh in these examples is interpreted in situ at LF, projecting alternatives all the way up to C. If that were the case, we would predict the examples with higher interveners in (35) to be ungrammatical, contrary to fact. Interpreting these wh-phrases in situ at LF would not be able to explain the contrast between the grammatical (35a–b) on the one hand and the ungrammatical (31b) and (32b) on the other hand.19

(35)

  • No intervention effect with higher intervener

  • graphic

Additionally, no intervention effect is observed when a focus-sensitive operator occurs inside the potentially pied-piped region but below the intervener, as in (36).

(36) No intervention effect with intervener below wh

  • a.

    Which student read which book containing no princesses?

  • b.

    Which student read which book discussing only princesses?

Finally, no intervention effect arises when no heads a coargument to the wh-phrase in a double object construction such as (37).20 As in (26), the negative head no occurs above the in-situ wh-word. However, unlike in (26), intervention does not occur in this structure.

(37) No intervention effect with no in coargument

  • a.

    Context: Professors White and Black are co-teaching a syntax class. Over the break, each student in the class was expected to write three book reports. Each book report had to be submitted to both professors for grading. Each student wrote three book reports, but oddly enough, each student submitted their three book reports to only one of the two professors. You have looked at the grading book. So tell me, what I want to know is:

  • b.

    Which student submitted no book report to which professor?

The important difference between structures (26) and (37) lies in the pied-piping options available to them, and in particular that no book report to which professor in (37) is not a possible QP and therefore not a possible target for covert pied-piping. The cases of intervention observed in examples such as (26), then, are precisely those where the intervener is contained within a constituent that is a pied-piping candidate for the wh-phrase.

Combining all our findings, we observe intervention effects in questions in which the intervener occurs in a small region above the wh as in (31b) and (32b), but not in questions that contain a high intervener such as (35) and (37). Therefore, we know that the interveners in (31b) and (32b) are inside intervention-sensitive regions but the ones in (35) and (37) are not. Consequently, it cannot be the case that the wh in the Superiority-obeying questions that we are examining remains in situ and is interpreted via Rooth-Hamblin alternative computation between its base-generated position and C. Instead, the observed pattern of intervention effects is consistent with the presence of covert pied-piping: an intervention effect occurs only when the intervener is placed inside QP and above wh, as in (38). This is predicted if the derivation of Superiority-obeying questions involves covert movement of the lower wh to C at LF, with pied-piping of a constituent inside of which Rooth-Hamblin alternatives are computed between wh and Q.

(38)

  • No causes intervention effect because it is in intervention-sensitive region inside QP

  • graphic

Cable’s (2010) theory predicts that VPs and larger constituents cannot form QPs in English. Thus, interveners at the VP level and higher will not be in an intervention-sensitive region, explaining the lack of intervention effects by higher interveners as in (39a). Similarly, since intervention only affects the region between the wh-word and Q, this approach also explains the lack of an intervention effect in (39b).21

(39)

  • No intervention effect when intervener is outside intervention-sensitive region

  • graphic

Finally, the evidence from intervention presented here allows us to further pinpoint the size of covert pied-piping. Assuming that the options for covert pied-piping parallel the options for overt pied-piping in a singular wh-question, we imagine the covert pied-piping options in (27), repeated here. Note that only (27a) predicts the intervention effect that we have observed. If (27b–c) were possible candidates for covert pied-piping, we would predict no intervention effect in (38b). Hence, we must conclude both that covert pied-piping exists and that it necessarily chooses the largest option among those available for overt pied-piping.

(27)

  • Different covert pied-piping options predict different intervention-sensitive regions Which student read no book from which library?

  • graphic

To summarize the findings of this section, we have presented evidence for the existence of covert pied-piping from the pattern of intervention effects in Superiority-obeying English multiple wh-questions. We showed that the pattern is predicted only if we assume a local region of focus alternative computation: an intervention effect occurs only in a small region above the surface in-situ wh-phrase. This pattern does not support a theory in which these surface in-situ wh-phrases are generally sensitive to negative islands; nor does it support a theory in which such wh-phrases are interpreted in situ and project Rooth-Hamblin alternatives from their base-generated position to C. Instead, this pattern is predicted by Cable’s (2010) theory of pied-piping. This theory predicts both the size of pied-piping that should be possible, and that inside the covertly pied-piped constituent we should find a region that is sensitive to intervention effects, under Beck’s (2006) theory of focus intervention effects. However, unlike overt pied-piping, covert pied-piping must choose the largest possible candidate for movement and smaller constituents are not viable candidates. In the next section, we discuss the reasons for this discrepancy between the sizes of overt and covert pied-piping.

4 Pied-Piping and the Interfaces

In this section, we address the differing sizes of pied-piping in overt vs. covert movement. As is well-known, overt pied-piping can target constituents of varying sizes, as shown in (40). Furthermore, we have shown in (19)—repeated in (41)—that targeting a smaller constituent for pied-piping can rescue a question from an intervention effect.

(40) Options for overt pied-piping

  • a.

    [QP Q A picture of which president] does Jim own_______?

  • b.

    [QP Q Of which president] does Jim own a picture_______?

  • c.

    [QP Q Which president] does Jim own a picture of_______?

(41) No intervention if intervener is not inside pied-piping constituent

  • a.

    *[QP Q No pictures of which president] does Jim own_______?

  • b.

    [QP Q Of which president] does Jim own no pictures_______?

  • c.

    [QP Q Which president] does Jim own no pictures of_______?

In contrast to this state of affairs in overt pied-piping, to correctly predict the pattern of intervention effects we observed in covert pied-piping it must be the case that only the largest QP available in (40), observed in (40a), is targeted for movement in the case of covert pied-piping. If smaller constituents could also be targeted for movement, we would predict no intervention effects at all in the Superiority-obeying questions surveyed above. That is, in a multiple wh-question like (42), only (42a) is a valid QP that can be attracted by C; the QPs in (42b–c) cannot be available targets for movement. (Here, ○ indicates a derivation that the grammar considers, and × indicates those that the grammar does not consider.)

(42)

  • Only largest covert pied-piping correctly predicts pattern observed in section 3 Which student read no book from which library?

  • graphic

This discrepancy is particularly puzzling since most speakers report a preference for the smaller overt pied-piping options in (40b–c) over (40a). Thus, the least preferred among the overt pied-piping options is the only candidate for covert pied-piping, (42a). Moreover, as we have shown, the upper bound on the possible size of QP is consistent across overt and covert pied-piping: in both cases, QPs formed by merging Q with VP or with a larger structure are ruled out. Consequently, we suggest that possible QP sizes are the same across overt and covert pied-piping.

The observed discrepancy in the pattern of intervention effects results from the different constraints that overt and covert pied-piping must satisfy. While overt pied-piping is subject to both LF and PF constraints, covert pied-piping is subject to LF constraints only. That is, overt pied-piping feeds both LF and PF and therefore must satisfy constraints at both interfaces. Covert pied-piping only feeds LF and thus reveals the true preference with regard to pied-piping: LF prefers to pied-pipe as large a constituent as possible, but in overt movement this preference can be overridden by the needs of PF.

In particular, several researchers have noticed that overtly moved wh-phrases prefer to be near the left edge of the clause (see, e.g., Horvath 2007, Heck 2008, 2009, Richards 2010, Cable 2012, 2013). Richards (2010) proposes that this is a linearly oriented PF interface constraint. He defines this constraint as a requirement on the prosody of wh-questions, (43), and shows that it can be satisfied in different languages in different ways—either through overt movement or through manipulation of prosodic phrase boundaries in wh-in-situ languages.

(43)

  • Prosodic constraint on wh-questions (Richards 2010:151)

  • Given a wh-phrase α and a complementizer C where α takes scope, α and C must be separated by as few Minor Phrase boundaries as possible, for some level of Minor Phrasing.

In English, the interpreting complementizer is always to the left and wh-movement is invoked to satisfy (43). Therefore, the result of this prosodic constraint is a requirement for leftness, as in (44).

(44)

  • The leftness preference of wh-phrases: A PF constraint

  • Wh-phrases prefer to be closer to the left edge of the clause.

We argue that the effects of this constraint are observable in English overt pied-piping. For example, consider the options for overt pied-piping in (40), repeated here. While all three pied-piping options are strictly speaking grammatical, many speakers report a preference for (40b–c) over (40a). This preference correlates with the fact that the wh-word is significantly farther away from the clause edge in (40a) than in (40b–c).

The items in (40) make clear that all theories of pied-piping must allow for some measure of optionality with regard to satisfying Richards’s (2010) prosodic constraint (our leftness constraint). In Cable’s (2010) theory, the syntax allows several different options for the merger site of the Q morpheme, as we have shown above. This grammatical mechanism must then interact in some way with the prosodic constraint to predict that some positions for Q are preferred over others.

Further examples help clarify that it is linear distance from the left edge of the clause that matters here, not the size of the pied-piping constituent or the depth of embedding of the wh-word within. Consider the examples in (45). Although the syntactic content and the semantic content of the two examples are very similar, speakers report a difference in their acceptability. Example (45a), which contains a large pied-piping constituent with a deeply embedded wh, is judged by speakers to be perfectly acceptable. In sharp contrast, (45b) is ungrammatical. This contrast is attributed to the fact that the wh is near the left edge of the pied-piping constituent in (45a) but not in (45b) (data from Cable 2012:823).

(45) Large pied-piping is grammatical if wh is at edge of moved constituent

  • a.

    [QP Q [[Whose brother]’s friend]’s father] did you see?

  • b.

    *[QP Q The father of [[whose brother]’s friend]] did you see?

Furthermore, even larger pied-piping in which the wh is even more deeply embedded, as in (46), remains grammatical, so long as the wh-word is near the left edge of the pied-piping constituent. Thus, PF does not restrict large pied-piping per se, but only pied-piping in which wh is not near the edge of the moved constituent.

(46) Pied-piping remains grammatical even with very deep embedding of wh in large QP

  • a.

    [QP Q [[[Whose brother]’s friend]’s father]’s boss] did you see?

  • b.

    [QP Q [[[[Whose brother]’s friend]’s father]’s boss]’s secretary] did you see?

This general preference of wh-phrases to be at the left edge (44) also results in corollaries such as Heck’s (2008, 2009) Edge Generalization.22

(47)

  • Edge Generalization (Heck 2008:88, 2009:89)

  • If α pied-pipes β (and movement of α to the edge of β is grammatically possible), then α must be at the edge of β.

Heck’s Edge Generalization explains cases of secondary wh-movement as in (48a). Here, a QP containing wh is pied-piped to the edge of a question, and following this movement wh must move to the edge of the pied-piping constituent (48b). The general possibility of such movement for degree heads is illustrated by (48c) (data from Cable 2012:821).

(48) Secondary wh-movement of degree wh predicted by Edge Generalization

  • a.

    [[How big] a car] did Bill buy?

  • b.

    *[A [how big] car] did Bill buy?

  • c.

    Bill would never buy [[that big] a car].

Again, this requirement that wh appear near the edge of QP appears to be a PF requirement, not one of structure-building or interpretation. Example (49a) is interpretable despite the fact that secondary wh-movement has not taken place and in fact cannot take place (49b). The grammaticality of (49a) reflects the more general unavailability of this form of PP-internal movement, as shown in (49c) (data from Cable 2012:822). From this point of view, then, secondary wh-movement is motivated solely by PF, not by LF.

(49) Secondary wh-movement blocked for possessive wh but question remains interpretable

  • a.

    [In [whose honor]] was this made?

  • b.

    *[[Whose honor] in] was this made?

  • c.

    *This was made [[Dave’s honor] in].

To summarize, there is a PF preference for having wh near the edge of an overtly pied-piped constituent. This is not an absolute requirement: larger pied-piping as in (40a) is accepted by most speakers, even if it is found to be less well-formed than smaller pied-piping options (40b–c). Furthermore, as a general rule the PF preference is not for smaller pied-piping, but for realizing the wh-word as far left as possible. This was demonstrated by the large possessive pied-piping in (45)–(46).

We propose that this PF constraint in (44) is the source of differences in size between pied-piping for overt movements and pied-piping for covert movements. By hypothesis, covert movement only affects the LF interface, and therefore the choice of covertly moved QP size will not affect the PF linearization.23

(50) Choice of covertly moved QP size does not affect PF linearizationWhich student read a book from which library?

  • a.

    [QP Q Which student] read [QP Q a book from which library]?

  • b.

    [QP Q Which student] read a book [QP Q from which library]?

  • c.

    [QP Q Which student] read a book from [QP Q which library]?

The results of section 3 thus help disentangle the preferences of PF and LF with regard to pied-piping size. The observed intervention pattern supports the conclusion that only the largest pied-piping is possible for covert movement. Hence, we learn that the preference of LF and narrow syntax is for larger pied-piping. This preference can be overridden by the PF constraints that govern overt pied-piping, such as (44), resulting in the complex patterns of pied-piping that we have pointed out in this section.

Formally, this preference for larger covert pied-piping could be thought of as being derived from a top-down application of the constraint Attract Closest (Relativized Minimality: Rizzi 1990; Minimal Link Condition: Chomsky 1995, 2000; also Shortest Move: Chomsky 1993). That is, C probes from above for positions where Q could grammatically be merged to form a QP and attracts the first Q-compatible constituent that it finds.24 In Cable’s (2010) system, these possible positions are constrained as shown in (50), ruling out large pied-piping of VPs or larger material. Because of the top-down nature of probing (Chomsky 2000), this would yield a preference of narrow syntax for larger QPs.

This kind of logic, dictating that attraction operations must apply to a larger constituent that may itself contain a possible second target of movement, is prevalent in the linguistic literature. For example, this is how Chomsky (1964) and Ross (1967) define the A-over-A Principle, requiring that operations that may in principle apply either to a larger target or to a smaller target contained within the larger target apply only to the larger target. We propose that in cases of overt movement, this operation is in competition with constraints on the prosody of wh-questions, as in Richards 2010. Since these constraints only affect movement that has consequences for the PF branch—that is, for overt movement—they are inactive for the purpose of deriving covert pied-piping. In the latter case, we predict that only Attract Closest is at play, and therefore that only the largest pied-piping option is available in the grammar.25

The end result of this discussion is somewhat different from the proposal in Cable 2010. There, all QPs derived from the grammatical merger of a Q-particle are predicted to be equally good. The grammaticality pattern observed here shows that only the largest pied-piping option is considered for pied-piping by LF and narrow syntax. This choice can be overridden by the PF constraint preferring that wh be closer to the left edge of the clause, thus making smaller pied-piping options available for overt pied-piping only.

5 Conclusion

In this article, we addressed the question of whether covert movement triggers pied-piping. Unlike pied-piping triggered by overt movement, covert pied-piping is not reflected in surface word order. Instead, we used the unique syntax/semantics of pied-piping constituents to diagnose the existence and size of covert pied-piping. We argued that covert movement does trigger pied-piping and that the preference of covert pied-piping, unlike overt pied-piping, is for movement of larger constituents.

The core data discussed here exhibit a previously unobserved pattern of focus intervention effects in English multiple wh-questions. We showed that a small region above the surface in-situ wh-word in Superiority-obeying multiple wh-questions is intervention-sensitive—that is, the insertion of a focus-sensitive operator in this region causes the disappearance of the multiple wh-question’s pair-list interpretation.

In this article, we adopted Beck’s (2006) theory of focus intervention effects that affect regions of Rooth-Hamblin alternative computation. Following Cable (2007), the region between a wh-word and the edge of its pied-piping constituent is a region of alternative computation and therefore intervention-sensitive. Thus, the existence of intervention effects in these small regions above the surface in-situ wh-word constitutes evidence for a pied-piping constituent around the wh-word. Alternative explanations were also discussed and shown to not derive the observed contrasts. The contrasts observed here therefore form an argument for the existence of covert pied-piping.

Furthermore, the diagnostic employed here helps determine the size of covert pied-piping. Whereas in overt movement, different-sized constituents containing the wh-phrase can be chosen for pied-piping, we showed that in covert movement, only the largest pied-piping constituent is considered by the grammar. We explained this contrast in terms of the different interface constraints that are fed by overt and covert movement: overt movement feeds both PF and LF, while covert movement feeds only LF. We concluded that the true preference of narrow syntax and LF is for pied-piping larger constituents. In overt movement, however, certain surface-oriented PF constraints were shown to override this choice of pied-piping.

The observations presented here also constitute a further argument for Beck’s (2006) theory of focus intervention effects. We extended the empirical scope of the phenomenon of ‘‘focus intervention’’ in English—building on previous work on Superiority-obeying and Superiority-violating questions ( Pesetsky 2000, Beck 2006) and intervention in overt pied-piping (Cable 2007)—and showed that these novel data are best accounted for in Beck’s theory. At its core, this amounts to further empirical support for Beck’s thesis that the alternatives utilized for focus interpretation (Rooth 1985, 1992) and alternatives utilized for wh-question interpretation (Hamblin 1973) are the same formal objects and that they interact in nontrivial ways.

Notes

For helpful discussion, we thank Danny Fox, Isaac Gould, Martin Hackl, Irene Heim, Ivona Kučerova, Sasha Podobryaev, Norvin Richards, Maziar Toosarvandani, Coppe van Urk, and especially David Pesetsky, as well as audiences at NELS 43 and the 2013 annual meeting of the LSA. We would also like to thank Jay Keyser and the anonymous LI reviewers for their constructive criticisms and suggestions.

1Pesetsky (2000) reports that in cases of intervention in multiple wh-questions such as (2d), many speakers report that the question is ungrammatical, while some others report that the question’s single-pair reading is maintained (ia) but its pair-list reading is lost (ib). For this reason, the multiple wh-questions that are crucial to our argumentation (section 3) are presented with explicit contexts that make their pair-list readings appropriate. See Dayal 2002 for more discussion of the semantics of the readings and Pesetsky 2000 for a discussion of the judgments.

(i) Single-pair and pair-list readings of Which boy didn’t read which book?

  • a. John didn’t read Robinson Crusoe.

  • b. John didn’t read Robinson Crusoe, Bill didn’t read Moby-Dick, and Fred didn’t read Don Quixote.

2 The characterization of the set of interveners has been a source of debate. Beck (2006) and Beck and Kim (2006) have identified a number of focus-sensitive operators, including only, also, even, and negation, as a crosslinguistically relatively stable set of interveners. Beck’s (2006) theory of intervention is discussed and adopted here.

Büring (1996a) argues that sentential negation is not a focus-sensitive operator, possibly raising a problem for Beck’s (2006) theory. Büring (1996a) argues that apparent focus-sensitive effects of the interpretation of negation in declaratives can be reduced to sentences with different prosody being congruent to different questions under discussion. (A similar argument is made in Beaver and Clark 2008:chap. 3.) However, these arguments are based solely on the contribution of negation in declaratives. A full investigation of the potential effects of focus on negation in interrogatives is outside the scope of this article. Here, we simply follow Pesetsky (2000) and Beck (2006), who show that sentential negation in English patterns together with the other items as an intervener.

Other authors have observed that the set of interveners can vary across languages and have proposed correspondingly different theories for intervention itself. See the discussion of analyses of interveners as antitopic items (Grohmann 2006, Tomioka 2007) in footnote 11, and Mayr’s (2014) analysis of interveners as nonadditive items in footnote 17.

3 Here and throughout, we use straight arrows to indicate movement and curly arrows to indicate areas in which Rooth-Hamblin alternatives are computed. These curly arrows are used here as a notational convenience only. Dashed arrows indicate covert movement.

4 The semantics of what and who in Beck 2006 are given in (i) and (ii).

  • (i) Semantics of what

    Ordinary semantics: 〚what〛 = undefined

    Focus-semantics: 〚whatf = {xs,e : x 1 human}

  • (ii) Semantics of who

    Ordinary semantics: 〚who〛 = undefined

    Focus-semantics: 〚whof = {x(〈s,e : x 1 human}

5 In this article, we will not present computations of Rooth-Hamblin alternatives beyond the toy example in (9). Regarding the technical details of Rooth-Hamblin (focus) alternative computation, see Hamblin 1973, Rooth 1985, 1992, and subsequent work.

6 The semantic denotations here must be interpreted intensionally. To simplify the presentation, world variables are not illustrated here.

7 It is worth noting that for intervention effects to occur as Beck (2006) theorizes in structures such as (11b), the alternatives used by focus-sensitive operators (Rooth 1985, 1992) and the alternatives used for question interpretation (Hamblin 1973) must be the same formal objects. This equivalence of focus alternatives and wh-interpreting alternatives is therefore an important theoretical claim of Beck’s (2006) system. This formal equivalence reflects the long-observed similarities between wh and focus realization crosslinguistically, including in syntax and prosody (Horvath 1986, Culicover 1991, Simpson 2000, Arregi 2002, Szendrői 2003, Simpson and Bhattacharya 2003, Truckenbrodt 2013, and many others). However, see for example Mayr 2014 for arguments against this approach.

8 Note that there may also be another problem when interpreting configuration (11b). Standard focus-sensitive operators utilize both the focus-semantic value and the ordinary semantic value for their interpretation (Rooth 1985, 1992), but the ordinary semantic value for the complement of the intervener in (11b) will be undefined (see footnote 4). See Beck 2006 for more technical discussion.

9 See also Kotek 2014b and Erlewine and Kotek 2014 on the use of focus intervention effects to diagnose regions of Rooth-Hamblin alternative computation, in particular in wh-questions with islands and association-with-focus constructions, respectively.

10 Cable argues that English is a limited pied-piping language, defined as follows:

(i) Limited pied-piping language (based on Cable 2010:147)

A language where a wh-word cannot be dominated in a pied-piped phrase by either an island or a lexical category Cable (2010:190ff.) does recognize and discuss the availability of pied-piping with larger constituents, such as (16a), in English.

11 Several alternatives to Beck’s (2006) theory of intervention effects have been proposed in the recent literature. Although such theories exist, they have been developed for other languages and have not been applied to the English data. Furthermore, we show that applying these theories to our data presents problems that we believe are unsolvable. In this footnote, we discuss two theories that identify interveners as antitopic items—items that cannot be topicalized or backgrounded. See footnote 17 for another alternative approach, proposed by Mayr (2014).

Grohmann (2006) proposes a theory of intervention effects in German under which an intervention effect is caused when an antitopic item occurs between two wh-phrases in the CP periphery. We note that this theory does not clearly extend to English, as discussed by Grohmann himself (see Grohmann 2006:13, 24), and that furthermore the theory predicts no intervention effects inside pied-piped constituents, because an intervention effect is caused when interveners move to a topic position in the CP periphery, but not when they move inside DPs.

Tomioka (2007) proposes a prosody-based account of intervention effects in Japanese and Korean single wh-questions, according to which an intervention effect arises when an antitopic item occurs to the left of a wh-word in a question, with these interveners including items such as negative polarity items and disjunctions. This theory does not clearly extend to English, because it crucially relies on prosodic properties of Japanese and Korean questions, which English does not share. More worrisome, we believe that Tomioka’s (2007) proposal makes incorrect predictions for Japanese questions with configurations such as the ones discussed here. Examples (ia–b) are judged by speakers to be similarly (un)acceptable. In (ia), the intervener dareka ‘someone’ occurs above the wh-word nani ‘what’, which is predicted by Tomioka to cause an intervention effect. Native speakers report that (ia) may be slightly degraded, but not clearly ungrammatical like the baseline Japanese intervention examples provided in Tomioka 2007. Example (ib) displays an order that is predicted by Tomioka to be acceptable, because the wh-word occurs before the intervener, but speakers do not judge this sentence to be better than its counterpart in (ia). Consequently, we are unsure how to extend Tomioka’s (2007) theory to our test cases.

(i) Data with wh-word and intervener inside single DP, unexplained by Tomioka’s (2007) proposal

a. ?[Dareka-no [nan-no shashin]]-o mi-ta no?

someone-GEN what-GEN picture-ACC see-PASTQUESTION

‘[Someone’s [picture of what]] did you see?’

b. ?[Dare-no [nanika-no shashin]]-o mi-ta no?

who-GEN something-GEN picture-ACC see-PASTQUESTION

‘[Whose [picture of something]] did you see?’

12 Although we present this as an assumption to facilitate the discussion, below we will show that the pattern of intervention effects that we observe further supports the conclusion that the lower wh could not be interpreted in situ in English Superiority-obeying questions.

13 See discussion in section 2.2 and Cable 2010 for arguments that Q cannot be merged with VP and larger constituents in English.

14 Note that Reinhart (1998) develops a theory that allows for the interpretation of a multiple wh-question without requiring movement of the ( phonologically) in-situ wh-phrase. However, this theory can only derive a single-pair reading of the question and not a pair-list reading (see Dayal 2002 for discussion). Crucially, the examples in this section are given in a pair-list context and thus cannot be interpreted using Reinhart’s choice-function mechanism.

15 We note that Cable (2007) reports examples similar to our (26) to be grammatical; compare his example (81) on page 139. However, Cable does not report whether the question can be interpreted as having a pair-list reading or only as having a single-pair reading. As we noted earlier, the single-pair reading of the question is indeed predicted to be possible, accounting for the judgment reported by Cable.

16 Note that, for some speakers, the determiner no in object position is generally slightly degraded. Nonetheless, native speakers report a strong contrast between the example in (25) and the example in (26) with the relevant reading, beyond what may be expected from simply having no in object position. Furthermore, this contrast is not limited to no; additional examples will show that other items in this position can also act as interveners. Therefore, (26) constitutes one in a series of intervention effects in this configuration, rather than a confounding outlier.

17Mayr (2014) offers an alternative characterization for intervention effects, proposing that intervention occurs when a nonadditive element has a wh-phrase in its scope. We believe that Mayr’s theory may, in principle, predict intervention effects inside DPs. However, his theory predicts that all nonadditive quantifiers, including less than n, should act as interveners. Data provided by Mayr suggest that this is indeed the case in German, but the native speakers we have consulted find a clear contrast between example (26) with no and example (28) with less than three in English. This finding thus constitutes a challenge to the generality of Mayr’s theory and its extension to English.

18 Much work on proportional determiners such as few argues that these determiners have three truth-conditionally distinct readings, including the cardinal reading, the proportional reading, and the so-called reverse-proportional reading. The computation of two of these readings—the proportional and the reverse-proportional—has been argued to involve focus, but the cardinal reading of few NP can be derived through a simple comparison of the cardinality of the NP with a contextual standard. See Büring 1996b, Herburger 2000, and Kotek, Sudo, and Hackl 2012 for discussion of the readings and their derivations. Example (30) in the context we give is designed to be a felicitous use of very few only under the focus-sensitive proportional reading. However, it is possible that interference from a cardinal reading is making the judgment more difficult than in the other cases we consider.

19 Also recall that, for the pair-list readings we target here, the surface in-situ wh-phrase cannot be interpreted in situ through a choice-function mechanism such as Reinhart’s (1998). See also footnote 14.

20 We thank an anonymous reviewer for bringing examples similar to (37) to our attention.

21 An anonymous reviewer asks whether we predict intervention effects to occur inside these larger pied-piped constituents in languages that allow for overt clausal pied-piping. We believe that the answer is no, because secondary wh-movement may target the edge of the pied-piped clause (see, e.g., Romanian data in Ratiu 2005). (Snejana Iovtcheva ( pers. comm.) notes that clausal pied-piping with secondary wh-movement also occurs overtly in Bulgarian.) Furthermore, Kotek (2014b) argues that similar partial wh-movement to the edge of a clause occurs covertly in the interpretation of in-situ wh-words inside clausal islands. Kotek (2014b) notes that in such questions, an intervention effect occurs when the intervener is above the island, but not when it is inside it. This is explained if the wh can move covertly to the edge of the island, above any interveners inside it, but it cannot escape the island, thus remaining sensitive to intervention effects above the island. See also Kotek 2014a for additional evidence of partial wh-movement to projections of propositional type.

22 Both Heck (2008, 2009) and Cable (2007, 2010) have developed syntactic accounts of the Edge Generalization, which explain it through interactions in the narrow syntax. For Heck, the Edge Generalization follows from Attract Closest, while Cable attributes it to the Phase Impenetrability Condition and the need for Q to agree with wh. It is beyond the scope of this article to decide which approach to the Edge Generalization is correct.

23 But note that overt secondary wh-movement does still apply to PF in-situ wh-phrases.

(i) Secondary wh-movement in PF in-situ wh-phrases

a. I know [which student bought [[how big] a car]].

b. *I know [which student bought [a [how big] car]].

This contrast shows that surface in-situ wh-words are still subject to the general leftness preference (44). The choice of QP size in (50), on the other hand, simply cannot affect the PF output, owing to the general unavailability of multiple wh-fronting in English.

24 In this variant, then, it may be the case that QP is not at all constructed, but instead that what matters is Q-compatibility, in Cable’s (2010) system. Another option is that Q is countercyclically merged to produce a QP for C to attract. Alternatively, under a left-to-right approach to structure building as in Phillips 1996, 2003, such a merger of Q would in fact not be countercyclic. Because Q is silent in English, these two variants are difficult to tell apart and we will not attempt to distinguish between them here.

25 One could also imagine another proposal similar in spirit, which involves a transderivational application of Attract Closest. That is, one could imagine the grammar considering the three derivations in (50) in parallel, differing only in the position where Q is merged. The derivation with the largest QP (50a) would optimally satisfy Attract Closest–type constraints for the probing of QP by C. See Reinhart 2006 for more on transderivational constraints and motivations for them.

References

Arregi,
Karlos
.
2002
.
Focus on Basque movements
.
Doctoral dissertation, MIT, Cambridge, MA
.
Beaver,
David Ian
, and
Brady
Clark
.
2008
.
Sense and sensitivity: How focus determines meaning
.
Oxford
:
Wiley-Blackwell
.
Beck,
Sigrid
.
1995
. Negative islands and reconstruction. In
On extraction and extraposition in German
, ed. by
Uli
Lutz
and
Jürgen
Pafel
,
121
143
.
Amsterdam
:
John Benjamins
.
Beck,
Sigrid
.
1996
.
Quantified structures as barriers for LF movement
.
Natural Language Semantics
4
:
1
56
.
Beck,
Sigrid
.
2006
.
Intervention effects follow from focus interpretation
.
Natural Language Semantics
14
:
1
56
.
Beck,
Sigrid
, and
Shin-Sook
Kim
.
2006
.
Intervention effects in alternative questions
.
Journal of Comparative Germanic Linguistics
9
:
165
208
.
Büring,
Daniel
.
1996a
. Drinking, accents, and negation. In
Proceedings of Workshop on Focus
, ed. by
Elena
Benedicto
,
Maribel
Romero
, and
Satoshi
Tomioka
,
37
50
.
University of Massachusetts Occasional Papers in Linguistics 21
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Büring,
Daniel
.
1996b
. A weak theory of strong readings. In
Proceedings of SALT VI
, ed. by
Teresa
Galloway
and
Justin
Spence
,
17
34
. Available at journals.linguisticsociety.org/proceedings/index.php/SALT/issue/view/103.
Cable,
Seth
.
2007
.
The grammar of Q
.
Doctoral dissertation, MIT, Cambridge, MA
.
Cable,
Seth
.
2010
.
The grammar of Q: Q-particles, wh-movement, and pied-piping
.
Oxford
:
Oxford University Press
.
Cable,
Seth
.
2012
.
Pied-piping: Introducing two recent approaches
.
Language and Linguistics Compass
6
:
816
832
.
Cable,
Seth
.
2013
.
Pied-piping: Comparing two recent approaches
.
Language and Linguistics Compass
7
:
123
140
.
Chomsky,
Noam
.
1964
. The logical basis of linguistic theory. In
Proceedings of the 9th International Congress of Linguists
, ed. by
Horace G.
Lunt
,
914
978
.
The Hague
:
Mouton
.
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
.
1993
. A minimalist program for linguistic theory. In
The view from Building 20
, ed. by
Kenneth
Hale
and
Samuel Jay
Keyser
,
1
52
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
1995
.
The Minimalist Program
.
Cambridge, MA
:
MIT Press
.
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
.
Culicover,
Peter W
.
1991
. Topicalization, inversion, and complementizers in English. In
Going Romance and beyond
, ed. by
Denis
Delfitto
,
Martin
Everaert
,
Arnold
Evers
, and
Frits
Stuurman
,
1
43
.
Utrecht
:
Utrecht University
.
Dayal,
Veneeta
.
2002
.
Single-pair versus multiple-pair answers: Wh-in-situ and scope
.
Linguistic Inquiry
33
:
512
520
.
Erlewine,
Michael Yoshitaka
, and
Hadas
Kotek
.
2014
. Intervention in focus pied-piping. In
NELS 43
, ed. by
Hsin-Lun
Huang
,
Ethan
Poole
, and
Amanda
Rysling
,
1
:
117
130
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Fox,
Danny
, and
Martin
Hackl
.
2006
.
The universal density of measurement
.
Linguistics and Philosophy
29
:
537
586
.
Grohmann,
Kleanthes K
.
2006
. Top issues in questions: Topics—topicalization—topicalizability. In
Wh-movement: Moving on
, ed. by
Lisa Lai-Shen
Cheng
and
Norbert
Corver
,
249
288
.
Cambridge, MA
:
MIT Press
.
Hagstrom,
Paul
.
1998
.
Decomposing questions
.
Doctoral dissertation, MIT, Cambridge, MA
.
Hamblin,
Charles
.
1973
.
Questions in Montague English
.
Foundations of Language
10
:
41
53
.
Heck,
Fabian
.
2008
.
On pied-piping: Wh-movement and beyond
.
Berlin
:
Mouton de Gruyter
.
Heck,
Fabian
.
2009
.
On certain properties of pied-piping
.
Linguistic Inquiry
40
:
75
111
.
Herburger,
Elena
.
2000
.
What counts: Focus and quantification
.
Cambridge, MA
:
MIT Press
.
Horvath,
Julia
.
1986
.
FOCUS in the theory of grammar and the syntax of Hungarian
.
Dordrecht
:
Foris
.
Horvath,
Julia
.
2007
. Separating ‘‘focus movement’’ from focus. In
Phrasal and clausal architecture
, ed. by
Simin
Karimi
,
Vida
Samiian
, and
Wendy K.
Wilkins
,
108
145
.
Amsterdam
:
John Benjamins
.
Huang,
C.-T. James
.
1982
.
Move wh in a language without wh movement
.
The Linguistic Review
1
:
369
416
.
Karttunen,
Lauri
.
1977
.
Syntax and semantics of questions
.
Linguistics and Philosophy
1
:
3
44
.
Kotek,
Hadas
.
2014a
.
Composing questions
.
Doctoral dissertation, MIT, Cambridge, MA
.
Kotek,
Hadas
.
2014b
. Intervention out of islands. In
NELS 44
, ed. by
Jyoti
Iyer
and
Leland
Kusmer
,
1
:
234
246
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Kotek,
Hadas
,
Yasutada,
Sudo
, and
Martin
Hackl
.
2012
.
Many readings of most
.
Handout, 48th Regional Meeting of the Chicago Linguistic Society, Chicago, IL
.
Kratzer,
Angelika
, and
Junko
Shimoyama
.
2002
. Indeterminate pronouns: The view from Japanese. In
The Proceedings of the Third Tokyo Conference on Psycholinguistics (TCP 2002)
, ed. by
Yukio
Otsu
,
1
25
.
Tokyo
:
Hituzi Syobo
.
Kuno,
Susumu
, and
Jane J
.
Robinson
.
1972
.
Multiple wh questions
.
Linguistic Inquiry
3
:
463
487
.
Kuno,
Susumu
, and
Kenichi
Takami
.
1997
.
Remarks on negative islands
.
Linguistic Inquiry
28
:
553
576
.
Mayr,
Clemens
.
2010
.
The role of alternatives and strength in grammar
.
Doctoral dissertation, Harvard University, Cambridge, MA
.
Mayr,
Clemens
.
2014
.
Intervention effects and additivity
.
Journal of Semantics
31
:
513
554
.
Pesetsky,
David
.
2000
.
Phrasal movement and its kin
.
Cambridge, MA
:
MIT Press
.
Phillips,
Colin
.
1996
. Order and structure.
Doctoral dissertation, MIT, Cambridge, MA
.
Phillips,
Colin
.
2003
.
Linear order and constituency
.
Linguistic Inquiry
34
:
37
90
.
Ratiu,
Dafina
.
2005
.
Questions multiples en Roumain
.
Master’s thesis, Université de Nantes
.
Reinhart,
Tanya
.
1998
.
Wh-in-situ in the framework of the Minimalist Program
.
Natural Language Semantics
6
:
29
56
.
Reinhart,
Tanya
.
2006
.
Interface strategies: Optimal and costly computations
.
Cambridge, MA
:
MIT Press
.
Richards,
Norvin
.
2010
.
Uttering trees
.
Cambridge, MA
:
MIT Press
.
Rizzi,
Luigi
.
1990
.
Relativized Minimality
.
Cambridge, MA
:
MIT Press
.
Rooth,
Mats
.
1985
.
Association with focus
.
Doctoral dissertation, University of Massachusetts, Amherst
.
Rooth,
Mats
.
1992
.
A theory of focus interpretation
.
Natural Language Semantics
1
:
75
116
.
Ross,
John Robert
.
1967
.
Constraints on variables in syntax
.
Doctoral dissertation, MIT, Cambridge, MA
.
Ross,
John Robert
.
1984
. Inner islands. In
Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society
, ed. by
Claudia
Brugman
et al.,
258
265
.
Berkeley
:
University of California, Berkeley Linguistics Society
.
Rullmann,
Hotze
.
1995
.
Maximality in the semantics of wh-constructions
.
Doctoral dissertation, University of Massachusetts, Amherst
.
Sauerland,
Uli
, and
Fabian
Heck
.
2003
. LF-intervention effects in pied-piping. In
NELS 33
, ed. by
Makoto
Kadowaki
and
Shigeto
Kawahara
,
347
366
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Simpson,
Andrew
.
2000
.
Wh-movement and the theory of feature-checking
.
Amsterdam
:
John Benjamins
.
Simpson,
Andrew
, and
Tanmoy
Bhattacharya
.
2003
.
Obligatory overt wh-movement in a wh-in-situ language
.
Linguistic Inquiry
34
:
127
142
.
Szabolcsi,
Anna
, and
Frans
Zwarts
.
1993
.
Weak islands and an algebraic semantics for scope taking
.
Natural Language Semantics
1
:
235
284
.
Szendrői,
Kriszta
.
2003
.
A stress-based approach to the syntax of Hungarian focus
.
The Linguistic Review
20
:
37
78
.
Tomioka,
Satoshi
.
2007
.
Pragmatics of LF intervention effects: Japanese and Korean interrogatives
.
Journal of Pragmatics
39
:
1570
1590
.
Truckenbrodt,
Hubert
.
2013
.
An analysis of prosodic F-effects in interrogatives: Prosody, syntax, and semantics
.
Lingua
124
:
131
175
.