Abstract

In languages like English, disjunctive questions have an alternative (ALT) reading and a yes/no (YN) reading. The two readings behave differently; there are environments in which one reading disappears and the other one survives. In this article, examining novel Croatian data, I investigate the environments where the YN reading disappears. Such environments suggest that the two readings of disjunctive questions differ in the size of the disjuncts: ALT readings arise when the disjoint constituents are bigger than the TP, while YN readings arise from disjunctions of phrases as big as the surface string suggests and not bigger than the TP.

1 Introduction

Disjunctive questions (DQs) are questions that contain a disjunction phrase, as in (1).

(1)

The string in (1) is ambiguous. It may be interpreted as having the yes/no (YN) reading in (2a) or as having the alternative (ALT) reading in (2b). The two readings are disambiguated by radically different intonation patterns (Bartels 1999, Pruitt 2008, Pruitt and Roelofsen 2013, Roelofsen and Pruitt 2011, Roelofsen and van Gool 2010).

(2)

graphic

Considerable work has been done on deriving the two readings of a DQ (Biezma and Rawlins 2012, Han and Romero 2004a,b, Larson 1985, McCawley 1988, Pruitt and Roelofsen 2013, Roelofsen and Pruitt 2011, Roelofsen and van Gool 2010). In this article, I contribute to this body of work by investigating DQs in Croatian. I focus on the environments, to the best of my knowledge unnoticed so far, in which a DQ has the ALT reading, but not the YN reading.1 I argue that the two readings of a DQ involve representations that differ in the size of the disjuncts: in YN readings, the disjoint constituents cannot include the Foc(us)0 head, which in effect reduces them to a size not bigger than the TP; but for an ALT reading to arise, the disjuncts must crucially be larger than the TP (Roelofsen and Pruitt 2011).

The analysis is in line with Han and Romero’s (2004a,b) proposal that ALT readings involve big disjuncts, with ellipsis of material in the second disjunct. In Han and Romero 2004a, the authors note that DQs with preposed negation lose the ALT reading, as illustrated by the contrast in (3).

(3)

graphic

Han and Romero (2004a) argue that this contrast follows from their (2004b) analysis of DQs, according to which ALT readings are derived from clausal (vP or TP) disjunction with parts of the second disjunct deleted (in line with Schwarz’s (1999) analysis of either/or constructions). Drawing on Larson’s (1985) analysis of disjunction, the authors also propose that ALT readings contain a wh-operator (null Q in matrix ALT DQs, whether in embedded ones), which moves from the disjunction site to the CP layer of the clause, marking the scope of the disjunction. Thus, the LF structure responsible for the ALT reading of (1) looks like (4).

Han and Romero (2004a) propose that preposed negation introduces into the structure the operator VERUM, which places “extra focus on polarity-related items such as auxiliary verbs or negation” ( p. 181). A DQ with preposed negation loses the ALT reading because all the possible representations of such a DQ are ill-formed. In the case where VERUM (introduced by the preposed negation) is present in both disjuncts, as in (5b), it is deleted in the second conjunct, violating the Focus Deletion Constraint (FDC) in (6), which states that focused elements cannot be deleted.

(5)

graphic

(6)

  • Focus Deletion Constraint

  • Focus-marked constituents at LF (or their phonological locus) cannot delete at Spell-Out.

  • (Han and Romero 2004a:199)

The FDC is not violated if VERUM is external to disjunction, as in (7). However, in that case, the wh-operator Q would have to move across it, which leads to intervention effects for the wh-chain (Kim 2002).

(7)

graphic

Finally, an FDC violation can be avoided if VERUM is present only in the first disjunct, so that it neither blocks the movement of Q nor is deleted at PF. What is wrong with this possibility, represented in (8), is that it violates the Focus Condition in (9), which requires that the two disjuncts be semantically parallel to one another.

(8)

graphic

(9)

Han and Romero’s (2004a) account of the loss of the ALT reading with preposed negation is only tenable if this reading involves deletion in the second disjunct, which in turn requires the disjuncts to be “big” (vPs or TPs). The analysis proposed here follows Han and Romero’s account in maintaining that on the ALT reading a DQ contains big disjuncts, but departs from it in arguing that “big” necessarily means “bigger than the TP.” Han and Romero do not explicitly investigate YN readings of DQs, but they do point out analyses of YN readings that are compatible with their account of ALT readings. On one analysis, YN readings involve a disjunction of clauses of opposite polarity, with deletion of the entire second disjunct together with the disjunction or, as in (10a). On the other analysis, illustrated in (10b), no clausal disjunction is present and Q does not associate with or. Evidence from Croatian DQs discussed here supports the latter proposal.

(10)

graphic

The article is organized as follows. In section 2, I introduce Croatian DQs and show the cases in which YN readings disappear. In section 3, I present my analysis of the observed phenomena. In section 4, I present additional support for the analysis, and section 5 is the conclusion.

2 Disjunctive Questions in Croatian

YN questions in Croatian are formed by the particle li, which is a second-position clitic that follows the first prosodic word in its clause (Franks and King 2000, Rivero 1993). There are two different strategies for forming YN questions in the language, depending on what element serves as the host for li (Alexander 2006):

  • the Comp strategy: li follows the complementizer da, as in (11a); and

  • the inversion strategy: li follows the tensed verb of the clause, as in (11b).

(11)

graphic

The clitic li is standardly assumed to be the question particle and has often been claimed to occupy the C0 position (e.g., Bošković 2001, Progovac 1996, Rivero 1993, Schütze 1994, Stjepanović 1999). The inversion strategy of YN question formation has been analyzed (Franks 1999, Franks and King 2000, Rivero 1993) as involving the movement of the tensed verb to C0 in order to host the clitic li, as shown in (12).

(12)

graphic

In the remainder of the article, I adopt this analysis, except for the claim that li occupies C0 position. Instead, I propose that it occupies a lower functional head, probably Foc0 (section 3.2).

The two strategies for forming YN questions in Croatian (the Comp strategy and the inversion strategy) are both attested in DQs, as shown in (13). Moreover, the DQs in (13a) and (13b) do not differ in the availability of the YN and ALT readings: both are available in both examples.

(13)

graphic

DQs formed by the Comp strategy (COMPS-DQs) and those formed by the inversion strategy (INV-DQs) behave differently, however, when the disjuncts involve a tensed verb. Such COMP-DQs have both readings, as in (14), but such INV-DQs lack the YN reading, as in (15).

(14)

graphic

(15)

graphic

I argue that the disappearance of the YN reading in INV-DQs in (15) is explained if this reading arises when the disjuncts are “small,” that is, as big as they appear on the surface and not as big as to include the clitic li. The verb movement out of the first disjunct only (necessary for the phonological support of li) is disallowed because it violates the Coordinate Structure Constraint (CSC), as illustrated in (16).

(16)

graphic

The ALT reading survives because it involves disjuncts that are “big” (bigger than the TP) and involve deletion of material in the second disjunct (Han and Romero 2004a,b). On this reading, each disjunct contains li and the tensed verb moves disjunct-internally, so no CSC violation results, as shown in (17).

(17)

graphic

I present the details of the analysis in the next section.

3 Analysis: Size Matters

3.1 The Structure of a DQ with the YN Reading

I propose that the INV-DQs in (15) do not have the YN reading because the disjuncts each contain a tensed verb, and it is the tensed verb that phonologically hosts li. However, the verb moves only from the first disjunct, in violation of the CSC, as shown in (18).

(18)

graphic

The COMP-DQs in (14a–b), where li is phonologically supported by the externally merged complementizer da, retain the YN reading because they do not involve verb movement, so the CSC is not violated. If the YN reading in (15a–b) is excluded because of a CSC violation, then in the representation that underlies this reading, li is external to the disjunction; that is, the disjuncts are too small to contain the position occupied by li. This is confirmed by the fact that the INV-DQ in (13b), repeated in (19), which contains disjoint object DPs, retains the YN reading even though the verb moves to li. This suggests that in (13b)/(19), the verb movement does not violate the CSC, that is, does not proceed out of the disjunction phrase. Thus, no contrast obtains between COMP-DQs and INV-DQs when the constituent that moves to li is external to the disjunction.

(19)

  • graphic

  • lovesiLI Jan ti Hana.ACC or Dora.ACC

  • ‘Does Jan love either Hana or Dora?’

Taken together, these considerations argue for the claim that, on the YN reading, the disjuncts in a DQ are “small,” that is, not bigger than they appear on the surface.

3.2 The Structure of a DQ with the ALT Reading

In this section, I examine the structure of the INV-DQs in (15), which yields the (attested) ALT reading. The fact that this reading survives, even though the YN reading does not, suggests that the derivation of the ALT reading in an INV-DQ has a way of avoiding a CSC violation, that is, that the movement of the verb in the first disjunct to li is syntactically legitimate. This argues for disjuncts that are big enough to include the clitic li, so that the movement of the first verb to li remains disjunct-internal. Consequently, the CSC is not violated. This analysis of (15a), for example, illustrated in (20), thus necessarily involves ellipsis in the second disjunct, argued for by Han and Romero (2004a,b).

(20)

  • graphic

  • readsiLI Vid ti newspaper or LI Vid watches news

  • ‘Is Vid reading the newspaper or is he watching the news?’

Note, however, that on the ALT reading of (15a–b), the disjuncts must crucially be bigger than the TP (cf. Han and Romero 2004a,b). Assuming, quite plausibly, that li occupies a position higher than the TP, if the ALT reading of these examples involved disjoint TPs, we would expect them to lack both the YN and the ALT readings (i.e., to be ill-formed) since the derivation of the ALT reading would also incur a CSC violation, as (21) shows.

(21)

graphic

We thus need an analysis on which disjuncts in the ALT reading of a DQ are bigger than the TP.2 The simplest possibility to explore is that ALT readings involve disjoint CPs. Such a proposal is compatible with the analysis of YN questions, illustrated in (12), on which li occupies C0 and the tensed verb moves to this position to provide phonological support for the clitic. It is also compatible with English examples like (22a–b), which have the ALT reading and have been argued to involve disjoint CPs.3

(22)

  • a.

    Did Sally bring wine or did she bring juice?

    (Roelofsen and Pruitt 2011:6)

  • b.

    Did John call Grandma or did Paul visit Grandpa?

    Croatian, however, behaves differently from English in that the second disjunct cannot be an interrogative clause, as shown in (23), regardless of the interrogation strategy used.4 The illformedness of (23a–b) suggests that the disjuncts in a DQ are not full CPs.

(23)

  • a.

    *Da li Jan piše knjigu ili da li Vid radi na scenariju?

    that LI Jan writes book or that LI Vid works on script

    ‘Is Jan writing a book or Vid working on a script?’

  • b.

    *Piše li Jan knjigu ili radi li Vid na scenariju?

    writes LI Jan book or works LI Vid on script

Interestingly, a disjunction of full CPs is allowed if embedded under an imperative ‘tell me’, as shown in (24). The examples in (24), however, have neither the YN nor the ALT reading. Instead, they instruct the hearer to answer either one disjunct or the other.

(24)

  • a.

    Reci mi da li nam Eva pjeva ili da li nam Ana pleše.

    tell me that LIUS.DAT.CL Eva sings or that LIUS.DAT.CL Ana dances

    ‘Tell me either whether Eva is singing for us or whether Ana is dancing for us.’

  • b.

    Reci mi pjeva li nam Eva ili pleše li nam Ana.

    tell me sings LIUS.DAT.CL Eva or dances LIUS.DAT.CL Ana

    ‘Tell me either whether Eva is singing for us or whether Ana is dancing for us.’

In order for an embedded DQ to be interpreted either as a YN question or as an ALT question, the second disjunct again cannot contain any overt sign of interrogation: li, da, and/or an inverted verb must all be absent. Such DQs, illustrated in (25), lose the YN reading when disjuncts involve the tensed verb that moves to li, just like the matrix DQs.

(25)

graphic

Now, how are we to derive the difference in meaning between the embedded DQs in (24a–b), which seem to involve CP-disjunction (as evidenced by the presence of the complementizer/inverted verb in both disjuncts), and the ALT interpretation of the embedded DQs in (25a–b), which I argue also involve a disjunction of constituents bigger than the TP? It is plausible to derive this difference from different scope relations that hold between the disjunction and the interrogative feature ( present on C0) in the two sets of examples. The reading of (24a–b), where the disjuncts are CPs, is intuitively compatible with an analysis on which both C0s are in the scope of the disjunction, yielding a disjunction of interrogatives (asking the addressee to answer either one of the two questions). The ALT reading observed in (25a–b), and in all the matrix DQs we have considered, obtains if the disjunction is in the scope of the single interrogative C0, instructing the hearer to answer a single or-question. This implies that on the ALT reading, theconstituents disjoined by ili ‘or’ are not CPs, although they are big enough for each to contain the clitic li. Consequently, the clitic li in Croatian does not seem to occupy the position of the interrogative C0 head. We need an analysis on which li appears in a position lower than C0, but higher than the TP.

One possibility is to follow Holmberg (2013), who argues that a direct question always contains a free variable, which moves to Spec,FocP and is the focus of the question. In wh-questions, the variable is a wh-phrase, while in YN questions, it is polarity. The question also contains “an illocutionary force feature Q, which encodes a request to the addressee to provide a value for the variable such that the resulting proposition is true” (Holmberg 2013:8). Questions thus have the structure in (26).

(26)

graphic

This view is compatible with Rizzi’s (1997, 2001) proposal that the left periphery of a clause contains multiple functional heads that dominate the TP, as in (27).

(27) Force Top* Foc Top* Fin IP

In what follows, I will assume that DQs have the same basic structure as simple YN questions, and I will use the label C0 to refer to the position that is called Force0 in Rizzi’s system and Q in Holmberg’s system. I will further maintain that the interrogative feature (which provides the question force of the utterance) resides on C0, while li occupies the position of a lower head, perhaps Foc0. An ALT DQ, then, involves a disjunction of focus phrases, embedded under an interrogative C0 and the polarity phrase, as in (28).5

(28)

graphic

On this analysis, the movement of the tensed verb in INV-DQs (and simple YN questions formed by the inversion strategy) targets the Foc0 head, where it adjoins to li (thus, the verb movement, although syntactic in nature, seems to be motivated by phonological considerations, namely, the clitichood of li). In COMP-DQs (and simple YN questions derived by the Comp strategy), li is supported by the complimentizer da in C0.6

Some support for the analysis on which li lexicalizes Foc0 comes from the fact that in a YN question, a constituent other than the verb may raise to li, in which case the relevant constituent is interpreted as narrowly focused. This is shown in (29).

(29)

  • VIDA li je Jan vidio?

  • Vid.ACCLIAUX Jan.NOM seen

  • ‘Was it Vid that Jan saw?’

The analysis is also compatible with the fact that li may appear in noninterrogative contexts—for example, in exclamations, as illustrated in (30).

(30)

  • Lijepe li djevojke!

  • beautiful.GENLI girl.GEN

  • ‘What a beautiful girl!’

If I am correct in claiming that on the ALT reading, the disjuncts are FocPs, headed by li, and do not include the interrogative head, we would like to know what excludes examples (31a–b), that is, why the li in the second disjunct must be deleted even when nothing else seems to be.

(31)

  • a.

    *Piše li Jan knjigu ili li Vid radi na scenariju?

    writes LI Jan book or LI Vid works on script

  • b.

    *Da li Jan piše knjigu ili li Vid radi na scenariju?

    that LI Jan writes book or LI Vid works on script

One possible solution is that li cannot immediately follow ili ‘or’ because this adjacency triggers some kind of haplology.7 The two end up adjacent because nothing can appear between them: the complementizer da is excluded since the syntactic position that hosts it (C0) is not part of either of the disjuncts. Verb movement to li (Foc0) is disallowed because the disjunction ili ‘or’ can itself host clitics, as we have seen in (25), where it phonologically hosts the pronominal clitic nam ‘to us’. Thus, we end up with the presumably illegal sequence ili li, which is destroyed by the obligatory deletion of li.8

If the present analysis is on the right track, the disjuncts in ALT DQs, although they are underlyingly always bigger than the TP, are not CPs (at least in Croatian), but FocPs. Such a structure, containing a single C0 head that takes scope over the disjunction, correctly derives the meaning of ALT DQs: they instruct the hearer to answer the single or-question, rather than instructing him or her to answer one of the disjoint questions. This is compatible with Krifka’s (2009) observation that speech acts (structurally encoded as CPs) can easily be conjoined, but resist disjunction (Krifka takes Rizzi’s ForceP to be the category interpreted as a speech act). The fact that speech acts can be conjoined but not disjoint, argues Krifka, derives the fact that “universal quantification into speech acts is possible” ( p. 5). This is because universal quantifiers are generalized conjunctions and as such can take scope over speech acts. Thus, the wh-question in (32a), with the universal quantifier every guest, has a pair-list reading, while the one in (33a), in which the quantifier is most guests, does not ((32)–(33) are adapted from Krifka 2009:5).

(32)

  • a.

    What did every guest bring?

  • b.

    For every guest x: What is y such that x brought y?

(33)

  • a.

    What did most guests bring?

  • b.

    #For most guests x: What is y such that x brought y?

The analysis of ALT DQs argued for here is also compatible with Pesetsky’s (2000:64) observation that “a clause interpreted as a question may not request anything less than a full answer.” Pesetsky notes that certain expressions (almost every NP, few NP, only NP), which at LF normally can take scope wider than on the surface, cannot do so in the context of a wh-question. This is illustrated in (34) for almost everyone.

(34)

  • Which newspaper did almost everyone write to ___ about this book?

  • [cannot express the following request for information: “Give me an almost complete list of people paired with the newspapers they wrote to about this book.”]

  • ( Pesetsky 2000:64)

Pesetsky also discusses intervention effects in English multiple questions with D-linked wh-phrases and points out, as noted by É. Kiss (1986) and Hornstein (1995), that even though Dlinking of wh-phrases obviates Superiority violations, as shown in (35a), the effect reemerges if a scope-bearing element intervenes between the C0 and the wh-phrase in situ, as in (35b). although both (35a) and (35b) contain negation, in (35a) it does not intervene between C0 and the in-situ wh-phrase which person, while in (35b) it does, ruling the sentence out on a Superiority violation.

(35)

  • a.

    Which book did which person not read _____?

  • b.

    *Which book didn’t which person read _____?

  • ( Pesetsky 2000:60)

If the scope-bearing element is a quantifier such as everyone, as in (36), the example is grammatical, presumably because the quantifier undergoes Quantifier Raising (QR) and at LF does not intervene between the C0 and the in-situ wh-phrase which newspaper.

(36)

  • Which book did everyone write to which newspaper about _____?

  • ( Pesetsky 2000:63)

However, if the quantifier is such that if it underwent QR, the question would require a nonexhaustive answer (e.g., almost everyone in (37)), then QR is impossible and consequently, the intervention effect reappears.

(37)

  • ??Which book did almost everyone write to which newspaper about ___?

  • ( Pesetsky 2000:64)

If Pesetsky is correct—that is, if a question always requires a full answer—it is not surprising that a matrix DQ cannot have the interpretation that instructs the hearer to answer either of the disjuncts. This reading (or the structure that might underlie it) does not exist since it asks for a nonexhaustive answer and therefore is incompatible with the fact that a DQ is a question. Accordingly, disjuncts in a matrix DQ cannot be CPs.9 A DQ embedded under tell me, however, may contain disjoint CPs, as in (24), because the matrix clause is not a question but a command, so the demand that it makes on the hearer may, in fact, request an answer to only one (either one) of the two disjoint embedded questions.10

To sum up, the analysis proposed in this section explains the fact that in COMP-DQs, both readings are systematically available, while in INV-DQs, the YN reading is available as long as neither of the disjuncts involves the verb that raises to li.

4 Additional Support for the Analysis

The analysis I have proposed for the fact that some INV-DQs lack the YN reading predicts that such DQs will retain the YN reading as long as the movement of the verb that hosts li does not violate the CSC. As we will see, this is exactly the case.

4.1 Nonfinite Verbs

Example (38) shows an INV-DQ in which the disjuncts are the size of a VP, but contain participial rather than tensed verbs. Such a DQ is ambiguous between the YN and the ALT readings. This is expected since the participial verbs do not leave their respective disjuncts; instead, li is supported by the disjunction-external auxiliary jesi.

(38)

  • graphic

  • AUX.2SGLI [written.PART paper] or [conduct.PART study]

  • ‘Have you written a paper or conducted a study?’

Similarly, in (39) the disjuncts are at least VPs (probably TPs with two controlled PRO subjects), but the YN reading is present since the infinitival verbs ići ‘go’ and gledati ‘watch’ do not move out of their disjuncts. The verb that supports the clitic li is the finite matrix verb želi ‘wants’.

(39)

  • graphic

  • wants LI Jan go.INF home.LOC or watch.INF movie.ACC

  • ‘Does Jan want to go home or watch a movie?’

4.2 Across-the-Board Movement of the Tensed Verb

The movement of the tensed verb to li avoids violating the CSC if it proceeds in an across-theboard (ATB) fashion from both disjuncts. Since ATB movement is only available to elements that are identical in both disjuncts, the analysis predicts that when the disjuncts are VPs (or TPs) but contain identical verbs, an INV-DQ should have both the YN and the ALT readings. As example (40) shows, this prediction is borne out.

(40)

  • graphic

  • sell.3SGLI Jan car.ACC or Hana apartment.ACC

  • ‘Is Jan selling the car or (is) Hana selling the apartment?’

4.3 COMP-DQs without the YN Reading

In this section, we will see that the disjuncts in a DQ with the YN reading are indeed not bigger than the TP. The relevant examples involve COMP-DQs in which li is supported by the complementizer da ‘that’ (so the CSC does not come into play) but that nevertheless lack the YN reading. I argue that this is due to the fact that the structure of the disjuncts in such DQs is necessarily bigger than the TP.

Consider the contrast in (41). Both DQs in (41) are COMP-DQs and both involve clausal disjuncts, as shown by the presence of a different subject in each. The only difference between them is that in (41a) the second disjunct does not contain the auxiliary clitic je, while in (41b) it does. Yet only the DQ in (41a) has the YN reading.

(41)

graphic

I argue that (41b) lacks the YN reading because the presence of the auxiliary in the second disjunct forces the disjuncts to be bigger than the TP; that is, it forces the structure of the ALT reading. This can be traced to the properties of auxiliaries in Croatian.

In this language, auxiliaries are second-position clitics, similar to li. In clauses that contain both li and an auxiliary clitic, both clitics appear in the second position of the clause, in a clitic cluster.11 When the clause is a YN question derived through the Comp strategy, both clitics follow the complementizer da and precede the subject, as in (42). It is thus reasonable to assume that (at least in interrogative clauses) second-position clitics occupy a position in the C layer of the clause (e.g., Bošković 2001, Progovac 1996, Rivero 1993, Schütze 1994, Stjepanović 1999).

(42)

  • Da li je Ivan ustao?

  • that LIAUX.3SG Ivan got.up

  • ‘Did Ivan get up?’

Given that the second disjunct in (41b) contains the auxiliary clitic je, which presumably occupies a position higher than the TP, it follows that the disjuncts in (41b) are bigger than TPs, as illustrated in (43). This structure, however, only maps onto the ALT reading and the DQ therefore lacks the YN reading.

(43)

  • Da [&P [li jei [TP Iva ti osvojila zlato]]

  • that LIAUXi Iva ti won gold

  • [ ili [li jej [TP Ana tj ispala iz natjecanja]]]]?

  • or LIAUXj Ana tj dropped from competition

The COMP-DQ (41a), in which the auxiliary is absent from the second disjunct, has the YN reading because it can receive an analysis on which disjuncts are no bigger than TPs. On this view, the auxiliaries from both disjuncts undergo ATB movement out of the disjunction phrase to form a clitic cluster with the single li, positioned higher than the disjunction, as in (44).

(44)

  • [CP Da li jei [&P [TP Iva ti osvojila zlato]] [ ili [TP Ana ti ispala iz

  • that LIAUXi Iva ti won gold or Ana ti dropped from

  • natjecanja]]]?

  • competition

Additional support for the claim that it is the (TP-external) position of the auxiliary clitic je, rather than its mere presence, that leads to the absence of the YN reading in (41b) comes from DQs that contain a negated version of the auxiliary. When the auxiliary is negated, it ceases to be a clitic and no longer needs to occupy the second position in the clause; instead, it occupies its canonical position, presumably T0. Crucially, a COMP-DQ with a negated auxiliary in both disjuncts retains the YN reading, as in (45).12

(45)

  • graphic

  • that LI Iva NEG.AUX put.on socks or Ana NEG.AUX put.on shoes

  • ‘Did Iva not put her socks on or (did) Ana not put her shoes on?’

The contrast between the presence of the YN reading in (41a) and its absence in (41b) thus supports the analysis on which the YN reading cannot have a structure in which the disjuncts are bigger than the TP.

4.4 Disjoint Subjects

Another piece of evidence in favor of the proposed analysis comes from DQs with disjoint singular subjects. Somewhat surprisingly, in Croatian, DQs with disjoint singular subjects are not ambiguous, regardless of the interrogation strategy used. The two readings are disambiguated by the number morphology on the verb. If the verb is plural, the DQ has only the YN reading, whereas if the verb is singular, the DQ has only the ALT reading. This is shown in (46) for COMP-DQs and in (47) for INV-DQs.

(46)

graphic

(47)

graphic

Singular disjoint subjects in a Croatian declarative clause can also appear with either singular or plural agreement on the verb, as shown in (48). The two sentences differ in having the inclusive or reading (48a) or the exclusive or reading (48b).13

(48)

graphic

Given the agreement contrasts in (48), coupled with their interpretation, it is plausible to assume that (48a) involves a disjunction of DPs, as in (49), while (48b), as well as (i) in footnote 13, involves a disjunction of TPs, as in (50).14

(49)

  • [&P [DP Jan] [&′ ili [DP Vid]]] idu u školu.

  • Jan or Vid go.3PL in school

(50)

  • [&P [TP (Ili) Jan ide u školu] [&′ ili [TP Vid ide u školu]]].

  • (or) Jan go.3SG in school or Vid go.3SG in school

Why it is that singular disjoint subjects in Croatian allow plural agreement remains mysterious (see Eggert 2002, Jennings 1994, Morgan 1985, Peterson 1986 for relevant discussion).15 However, regardless of the answer to this question, the covariance between the number marking on the verb and the interpretation of a DQ with singular disjoint subjects, observed in (46) and (47), is expected on the proposed analysis. If the YN reading of a DQ requires small disjuncts, as I have been arguing, we predict that on this reading, when the subject disjunction phrase involves singular DPs, the verb shows obligatory plural agreement, as in the declarative sentence in (48a). In INV-DQs, illustrated in (47a), this reading persists because the disjunction phrase contains only subject DPs, so the movement of the verb to li does not violate the CSC. On the ALT reading of a DQ, which requires clausal disjuncts, obligatory singular agreement on the verb is expected, since in the biclausal structure, each subject agrees with the verb in its own clause.

Not surprisingly, the ambiguity between the two readings reappears if the disjuncts within the subject phrase are plural. Such DQs obligatorily appear with a plural verb and are ambiguous between the YN and the ALT readings.

(51)

graphic

4.5 Multiple ALT DQs

In this section, we will consider a DQ that contains multiple disjunctions: for example, a disjunction of subjects and a disjunction of predicates, as sketched in (52).

(52) Subject: (A or B); predicate: (C or D)

Let us consider the logically possible interpretations that such a DQ might have. If each disjunction gives rise to two possible readings (YN and ALT), there are four different readings, given in (53), that are in principle available to such a multiple DQ.

(53)

graphic

The reading in (53d) is, however, not attested. Let us inspect the DQ in (54), which involves disjoint singular subjects, {Dan, Vid}, and disjoint PPs, {u kino ‘to the cinema’, u kazalište ‘to the theater’}. I have chosen to mark the verb in (54) with singular agreement because, as discussed above, the presence of a singular verb excludes the YN reading with respect to the disjuncts in the subject phrase; that is, it eliminates the readings in (53a–b). However, of the two remaining readings (53c–d), spelled out in (55), only (55a) (= (53c)) survives.

(54)

  • Da li Dan ili Vid ide u kino ili u kazalište?

  • that LI Dan or Vid go.3SG in cinema or in theater

  • ‘Is Dan or Vid going to the cinema or to the theater?’

(55)

  • a.

    Is it Dan or is it Vid who is going to the cinema or the theater?

  • b.

    *Is it Dan or is it Vid who is going to the cinema or is it Dan or is it Vid who is going to the theater?

This is predicted if the ALT reading arises when each disjunct is clausal (with the deletion in one of the disjuncts). The fact that the verb in (54) is singular indicates that the subject disjunction is clausal; that is, there is a deletion in the first disjunct, as in (56).

(56)

  • [Is Dan 〈going [PP to the cinema or to the theater]〉]or

  • [is Vid going [PP to the cinema or to the theater]]

The VP (deleted in the first disjunct, overt in the second) contains a “small” disjunction of the PPs, and this yields the reading in (55a). For the reading in (55b) to arise, however, in addition to the disjunction of clauses in (56), the DQ would also have to contain the disjoint clauses in (57).

(57) [Is SUBJECT going to the cinema] or [is SUBJECT going to the theater]?

The clausal disjunction in (57) is possible if each disjunct contains a “small” DP disjunction in the subject phrase (overt in the first disjunct, deleted in the second). But this is incompatible with the fact that subject disjunction is also clausal, as required by the singular verb. The reading in (55b) seems to require either that the disjunction in (56) somehow be embedded in each disjunct in (57), or vice versa. So this configuration effectively requires that a clause be embedded in itself, which is impossible.

The analysis of ALT readings by Han and Romero (2004a,b) makes the same prediction since they also argue that ALT readings involve big disjuncts (TPs or vPs). On the other hand, it is difficult to see how (55b) is blocked on analyses on which ALT readings could be derived from small disjuncts (as suggested by the reviewer cited in footnote 2). For example, Larson (1985) proposes that ALT DQs are derived from disjuncts of any size by the movement of the scopal indicator (SI), which originates at the left edge of the disjunction phrase and moves to Spec,CP, marking the scope of the disjunction. The SI is whether in embedded ALT questions and O (the null indicator) in matrix ones. Effectively, the SI behaves like a wh-phrase. Given that Croatian is a multiple wh-fronting language with more than one landing site available for wh-phrases, it is unclear what blocks movement of multiple SIs on this analysis. Beck and Kim (2006) argue against the analysis on which disjunction phrases are wh-phrases by showing, among other things, that multiple ALT readings, like the one in (55b), are impossible. On the other hand, they argue, on the basis of intervention effects in ALT DQs, that the disjuncts “cannot be too large” ( p. 204). The ALT readings arise because in addition to the ordinary semantic value, the disjunction contributes the focus semantic value, which is an alternative set that contains the two ordinary meanings of the disjuncts. The focus semantic value propagates all the way to the TP level, at which point the question operator lifts it to the level of ordinary semantics. As far as I can tell, on this analysis, the TP in (54) should be able to receive the focus semantic value in (58), which incorrectly derives the reading in (55b).

(58) {λw. Dan is going to the cinema in w, λw. Dan is going to the theater in w, λw. Vid is going to the cinema in w, λw. Vid is going to the theater in w}

In this section, we have seen that DQs with multiple disjunctions support the analysis on which ALT readings involve big disjuncts and YN readings small disjuncts. In the next section, I discuss DQs with preposed negation.

4.6 DQs with Preposed Negation

Han and Romero (2004a) show that the ALT reading of a DQ disappears with preposed negation, as in (59).

Recall from section 1 that on Han and Romero’s (2004a) analysis, such DQs lose the ALT reading crucially because the disjuncts are big (vPs or TPs) and parts of the second disjunct are deleted. In Croatian, preposing the negation is the standard way of forming negative polar questions, shown in (60) (but see (45) and the discussion in footnote 12). The negative marker ne/ni surfaces as an affix on the tensed verb at the front of the question. Thus, negated DQs in Croatian are always INV-DQs. As expected, such DQs have only the YN reading, as illustrated in (60).

(60)

  • graphic

  • NEG drinks LI Jan coffee or tea

  • ‘Doesn’t Jan drink coffee or tea?’

If the analysis proposed here is correct, it appears that (61) cannot be the source of (60).

(61)

  • *[ne pije li Jan kavu] ili [li Jan ne pije čaj]

  • NEG drinks LI Jan coffee or LI Jan NEG drinks tea

This structure can be excluded in several ways. We could follow Han and Romero and assume that besides big disjuncts, the ALT readings necessarily involve the movement of the wh-operator Q from the left edge of the disjunction to the Spec,CP. The unavailability of (61) follows automatically, since the operator VERUM, introduced by the preposed negation, would either be deleted in the second disjunct (violating the FDC), as in (62a), or be present in the first disjunct only (violating the Focus Condition), as in (62b). Finally, VERUM could be external to the disjunction, in which case it would interfere with the movement of Q to Spec,CP, as in (62c).

(62)

graphic

It is, however, not clear that ALT DQs in Croatian involve wh-movement. For example, the DQ in (63) is well-formed on both readings, although the disjunction is embedded in the complex NP, which is otherwise an island for wh-movement.

(63)

  • Vjeruješ li u tvrdnju da je Petar dao otkaz ili otišao u mirovinu?

  • believe.2SGLI in claim that AUX Petar given resignation or gone in retirement

  • ‘Do you believe the claim that Petar resigned or retired?’

If no operator movement is involved in the derivation of ALT readings, then the representation in (61) should actually be licit because it could involve the analysis in (62c), modulo the Q-movement. However, given that the ALT reading is absent, it must be that something excludes (61). I believe that the impossibility of (61) is related to the fact that a negated YN question in Croatian cannot contain the strong negative polarity item ni ‘neither’, as shown in (64), and always conveys a positive expectation on the part of the speaker.

(64)

  • Ne dolazi li i / *ni Petar?

  • NEG comes LI and(also) / *neither Petar

  • ‘Isn’t Petar coming too/*either?’

Holmberg (2013) shows that a subset of English speakers judge English YN questions with contracted negation (n’t) in the same way. He proposes that for those speakers, the n’t is externally merged in the C domain, more precisely in Pol0, the head that dominates the FocP in (65), repeated from (26). The TP thus does not contain negation, which excludes the negative polarity item and suggests that the speaker believes that the positive answer is true.16

(65)

graphic

I would like to suggest that the negative affix in Croatian negated YN questions is also externally merged in PolP and that the verb from the first disjunct raises to Pol0 to support it. Since the disjuncts in the ALT reading are FocPs, this movement violates the CSC, just as the comparable movement of the tensed verb to li violates it in INV-DQs with VP or TP disjuncts. In fact, if we negate an INV-DQ with VP or TP disjuncts, which only has the ALT reading to begin with, it becomes ill-formed since neither of the two readings is able to survive. This is confirmed by (66b), which is ungrammatical because the ALT reading—the only reading available in the INV-DQ with disjoint VPs, shown in (66a)—is “destroyed” by the preposed negation.

(66)

graphic

4.7 Intervention Effect in DQs

The proposed analysis also fits well with Beck and Kim’s (2006) observation that the ALT reading of a DQ disappears if the disjunction phrase is preceded by a focus-sensitive operator, as shown in (67).

Again, the present analysis predicts that in INV-DQs that lack the YN reading (because of a CSC violation), adding a focus-sensitive operator before the disjunction should lead to ungrammaticality. This prediction is also borne out, as shown by (68b). The ill-formedness of (68b) can be explained by the violation of the FDC, which prevents the deletion of focused material, as shown in (69).

(68)

graphic

(69)

  • *[Gleda li samo Vid televiziju] ili [li samo Vid sluša radio]?

  • watches LI only Vid television or LI only Vid listens radio

5 Conclusion

In this article, I have argued that the two readings of a DQ (ALT and YN) have very different syntactic sources. On the YN reading, the disjuncts are as big as their surface string suggests and cannot exceed the TP. By contrast, on the ALT reading, the disjuncts are always bigger than the TP, but are not as big as the CP. For concreteness, I proposed that they are FocPs and that li is the lexicalization of Foc0 (the Focus head), but nothing hinges on this particular choice: any other functional head between T0 and C0 would do. There is obligatory deletion of material in the second disjunct, and the deleted material minimally includes the clitic li (for unknown reasons, perhaps because of haplology).

The proposal argues against the analysis of ALT readings by Beck and Kim (2006), who (on the basis of intervention effects, discussed in section 4.7) argue that on this reading, the position of the intervener “puts a roof on the size of the disjuncts” ( p. 204).

Finally, the data from Croatian I discussed suggest that there is something very syntactic about the differences between the YN and the ALT readings of a DQ, and that any semantic explanation of these differences must be supplemented by different syntactic representations of the two readings.

Notes

I would like to thank the editor and three anonymous LI reviewers for their constructive criticisms of the previous versions of the article; their comments led to significant improvements. An earlier version of this work was presented at SinFonIJA 5 in Vienna, in September 2012; many thanks to the audience for their helpful comments and questions. For help with English judgments, I am indebted to Seth Cable, Sarah Hulsey, and Omer Preminger. Discussions of disjunctive questions with Boban Arsenijević and David Pesetsky always helped me see more clearly the problems I was interested in. Responsibility for any errors in proposing solutions to them lies solely with me.

1 Environments in which a DQ retains the YN reading but loses the ALT reading are discussed in Beck and Kim 2006, Han and Romero 2004a,b, and Larson 1985.

2 A reviewer asks whether big disjuncts are required in all DQs with ALT readings, suggesting that the ALT readings in DQs like (19), in which the tensed verb that moves to li is external to the disjunction, could also be captured by an analysis that posits small disjuncts. The reviewer suggests that the Croatian data discussed here (and also the English data in (22)) call for an asymmetric analysis of DQs, on which YN readings require small disjuncts, but ALT readings are compatible with both big and small disjuncts. In section 4.5, where I discuss multiple ALT DQs, we will see that the asymmetric account of Croatian DQs suggested by the reviewer cannot be maintained.

3 Example (22b) is due to a reviewer, who judges it as having only the ALT reading. My informants, however, seem to be able to access the YN reading of this example as well. If the example indeed has the YN reading, then DQs in English have different properties than DQs in Croatian. I leave for further research exactly how these differences should be captured. In footnote 5, I briefly discuss a possible structure for English DQs with the ALT reading.

4 I omit comparable examples in which the two disjuncts are derived through different interrogation strategies since they could also be excluded because of lack of parallelism.

5 If (28) is the correct representation for Croatian DQs with the ALT reading, then the question arises whether the same structure underlies the English DQs in (22), where the presence of a fronted auxiliary in each disjunct suggests a disjunction of CPs. If these examples have the structure in (28), then the auxiliaries do not in fact occupy C0; rather, they occupy Foc0. One piece of evidence that this might be so comes from (i), where each disjunct contains a fronted negation.

(i)

graphic

None of my informants could access the YN reading of this example, and all of them found it severely degraded on the ALT reading. These judgments indicate that in English DQs the disjuncts cannot be as big as to include Pol0 (the position of the negation in (i)) either on the YN or on the ALT reading. Recall also from footnote 3 that the examples in (22) were judged to be ambiguous between the YN and the ALT readings, suggesting that in English, disjunction of FocPs may give rise to both ALT and YN readings.

6 The fact that an interrogative C0 is lexicalized by a declarative complementizer is not problematic: Aboh and Pfau (2011), for example, discuss the cooccurrence of the declarative complementizer and a question particle in embedded interrogatives in Gbe and Bantu languages, where the former shows up clause-initially, the latter clause-finally.

7 I would like to thank Susi Wurmbrand for helpful discussion of this issue.

8 It is unclear to me at this point whether this requirement is phonological or semantic.

9 A reviewer points out that besides exhaustive answers, wh-questions admit mention-some answers, suggesting that the data discussed by both Krifka (2009) and Pesetsky (2000) is more likely to be explained by Krifka’s analysis, which crucially relies on universal quantifiers being generalized conjunctions. As far as I can tell, the fact that DQs do not have the reading on which the hearer is instructed to answer either one of the disjoint questions is equally compatible with both Krifka’s and Pesetsky’s approaches.

10 As a reviewer points out, if speech acts resist disjunction in matrix contexts, then they should resist disjunction in embedded contexts as well. This seems to be falsified by the examples in (24), where the presence of the complementizer da indicates the presence of C0 in both embedded disjuncts, which in turn suggests a disjunction of speech acts. While I share the reviewer’s concern, it seems to me that embedded disjoint interrogatives require a different treatment than matrix ones: either we need to say that such CPs are not speech acts and thus can be disjoined, or we need to say that at the embedded level, speech acts can, in fact, be disjoined. See Krifka 2014 and McCloskey 2006 for a discussion of embedding question speech acts.

11 In a clitic cluster, clitics appear in the following order:

(i) LI < AUX (except for 3rd person singular auxiliary je) < DAT(pron) < ACC(pron) < je

12 In Croatian, examples like (45) are possible, but rather marginal. A DQ with negation usually involves either the inversion strategy, as in (i), or the interrogative particle zar, as in (ii). In both cases, the DQ has only the YN reading (Han and Romero 2004a) and carries the presupposition that the positive proposition is true.

  • (i)

    graphic

  • (ii)

    graphic

13 Singular agreement is the only option for the ili . . . ili ‘either . . . or’ construction, as in (i).

  • (i)

    Ili Jan ili Vid ide / *idu u školu.

    or Jan or Vid go.3SG / *go.3PL in school

    ‘Either Jan or Vid goes to school.’

14 A reviewer suggests that the data could also be derived by a purely semantic account that relies on a semantic approach to number marking, defended by Sauerland, Anderssen, and Yatsushiro (2005). On this approach, the plural is regarded as unmarked and is used whenever the speaker cannot commit himself or herself to the singular, which conveys cardinality “one.” The inclusive or reading allows for both disjuncts to have the property denoted by the VP, so the singular morphology on the verb is blocked and the unmarked plural appears. On the exclusive or reading, the singular marking on the verb is then expected for semantic reasons even if the disjuncts are not bigger than DPs. The reviewer points out that the semantic account outlined above is preferred, given that on Schwarz’s (1999) analysis of either/or constructions in English, the disjunction involves deletion only if either (the first ili in (50)) appears displaced from the first disjunct. Since in our examples this is not the case, it might be an indication that we are in fact not dealing with a clausal disjunction, in which case the syntactic account proposed in the text fails.

I agree with the reviewer that the purely semantic analysis of the verb’s number marking with disjoint subjects accounts for the observed facts, and it may well be that it is the correct account. However, the syntactic analysis that I have proposed is not implausible either, since either (ili) behaves quite differently in Croatian than it does in English. The distribution of Croatian ili differs from that of English either, and unlike the English either . . . or construction, the ili . . . ili construction in Croatian never allows narrow scope for the disjunction (relative to other scope-bearing elements in the clause). Thus, the fact that the first ili in (50) does not appear displaced from the first disjunct cannot be taken as evidence that disjuncts are small, as the nondisplacement of either can in English. Consequently, the data are compatible with both the syntactic account presented in the text and the semantic account proposed by the reviewer.

15 Plural agreement on the verb with singular disjoint subject DPs is reported as possible in English (Morgan 1985, Peterson 1986), in Modern Greek (Kazana and Flouraki 2009), and in certain contexts in Russian (Ivlieva 2012) and Mi’gmaq (Bale 2012).

16Sailor (2013) presents experimental evidence that English speakers do not accept strong negative polarity items, such as punctual until- and for-phrases, in polar questions with high negation (*Didn’t John finish the assignment until yesterday?), suggesting that such questions do not allow the “inner” negation reading (cf. Ladd 1981).

References

References
Aboh,
Enoch O.
, and
Roland
Pfau
.
2011
.
What’s a wh-word got to do with it?
In
Mapping the left periphery: The cartography of syntactic structures
,
vol. 5
, ed. by
Paola
Benincà
and
Nicola
Munaro
,
91
124
.
Oxford
:
Oxford University Press
.
Alexander,
Ronelle
.
2006
.
Bosnian Croatian Serbian: A grammar with sociolinguistic commentary
.
Madison
:
University of Wisconsin Press
.
Bale,
Alan
.
2012
.
Agreement without AGREE: Disjunction in Mi’gmaq
.
Poster presented at the 43rd North East Linguistic Society (NELS), New York, 19–21 October
.
Bartels,
Christine
.
1999
.
The intonation of English statements and questions: A compositional interpretation
.
New York
:
Garland
.
Beck,
Sigrid
, and
Shin-Sook
Kim
.
2006
.
Intervention effects in alternative questions
.
Journal of Comparative Germanic Linguistics
9
:
165
208
.
Biezma,
María
, and
Kyle
Rawlins
.
2012
.
Responding to alternative and polar questions
.
Linguistics and Philosophy
35
:
361
406
.
Bošković,
Željko
.
2001
.
On the nature of the syntax-phonology interface: Cliticization and related phenomena
.
Amsterdam
:
Elsevier
.
Eggert,
Randall
.
2002
.
Disconcordance: The syntax, semantics, and pragmatics of or-agreement
.
Doctoral dissertation, University of Chicago, Chicago, IL
.
Franks,
Steven
.
1999
.
Clitics in Slavic
.
Paper presented at the Comparative Slavic Morphosyntax Workshop, Bloomington, IN, 5–7 June
.
Franks,
Steven
, and
Tracy Holloway
King
.
2000
.
A handbook of Slavic clitics
.
Oxford
:
Oxford University Press
.
Han,
Chung-Hye
, and
Maribel
Romero
.
2004a
.
Disjunction, focus, and scope
.
Linguistic Inquiry
35
:
179
217
.
Han,
Chung-Hye
, and
Maribel
Romero
.
2004b
.
The syntax of whether/Q . . . or questions: Ellipsis combined with movement
.
Natural Language and Linguistic Theory
22
:
527
564
.
Holmberg,
Anders
.
2013
.
The syntax of negative questions and their answers
. In
Proceedings of GLOW in Asia IX 2012: The Main Session
, ed. by
Nobu
Goto
,
Koichi
Otaki
,
Atsushi
Sato
, and
Kensuke
Takita
,
1
18
.
Tsu, Japan
:
Mie University
.
Hornstein,
Norbert
.
1995
.
Logical Form: From GB to Minimalism
.
Cambridge, MA
:
Blackwell
.
Ivlieva,
Natalia
.
2012
.
Implicatures in agreement
.
In Coyote papers. Working papers in linguistics 20: Proceedings of the Poster Session of the West Coast Conference on Formal Linguistics (WCCFL), held April, 2011, ed. by Jaehoon Choi, E. Alan Hogue, Jeffrey Punske, Deniz Tat, Jessamyn Schertz, and Alex Trueman, 28–36. Retrieved on 27 May 2013 from http://arizona.openrepository.com/arizona/bitstream/10150/253454/5/Ivlieva%2c%20Natalia%20Implicatures%20in%20Agreement%2028-36.pdf
.
Jennings,
Ray E
.
1994
.
The genealogy of disjunction
.
New York
:
Oxford University Press
.
Kazana,
Despina
, and
Maria
Flouraki
.
2009
.
Constraining disjunctive agreement in Modern Greek
. In
Proceedings of LFG09
, ed. by
Miriam
Butt
and
Tracy Holloway
King
,
282
296
.
Stanford, CA
:
CSLI Publications
.
Kim,
Shin-Sook
.
2002
.
Intervention effects are focus effects
. In
Japanese/Korean Linguistics 10
, ed. by
Noriko
Akatsuka
and
Susan
Strauss
,
615
628
.
Stanford, CA
:
CSLI Publications
.
É. Kiss,
Katalin
.
1986
.
Against the LF-movement of WH-phrases
.
Ms., Hungarian Academy of Sciences, Budapest
.
Krifka,
Manfred
.
2009
.
Embedding speech acts
.
Paper presented at Recursion: Structural Complexity in Language and Cognition, Amherst, MA, 26–28 May
.
Krifka,
Manfred
.
2014
.
Embedding illocutionary acts
. In
Recursion: Complexity in cognition
, ed. by
Tom
Roeper
and
Peggy
Speas
,
59
87
.
New York
:
Springer
.
Ladd,
Robert
.
1981
.
A first look at the semantics and pragmatics of negative questions and tag questions
. In
Papers from the 17th Regional Meeting of the Chicago Linguistic Society
, ed. by
Roberta A.
Hendrick
,
Carrie S.
Masek
, and
Mary Frances
Miller
,
164
171
.
Chicago
:
University of Chicago, Chicago Linguistic Society
.
Larson,
Richard
.
1985
.
On the syntax of disjunction scope
.
Natural Language and Linguistic Theory
3
:
217
264
.
McCawley,
James D
.
1988
.
The syntactic phenomena of English
.
Chicago
:
University of Chicago Press
.
McCloskey,
James
.
2006
.
Questions and questioning in a local English
. In
Crosslinguistic research in syntax and semantics: Negation, tense, and clausal architecture
, ed. by
Raffaella
Zanuttini
,
Héctor
Campos
,
Elena
Herburger
, and
Paul H.
Portner
,
87
126
.
Washington, DC
:
Georgetown University Press
.
Morgan,
Jerry
.
1985
.
Some problems of determination in English number agreement
. In
Proceedings of the First Eastern States Conference on Linguistics
, ed. by
Gloria
Alvarez
,
Belinda
Brodie
, and
Terry
McCoy
,
69
78
.
Columbus
:
Ohio State University, Department of Linguistics
.
Pesetsky,
David
.
2000
.
Phrasal movement and its kin
.
Cambridge, MA
:
MIT Press
.
Peterson,
Peter
.
1986
.
Establishing verb agreement in disjunctively conjoined subjects: Strategies vs. principles
.
Australian Journal of Linguistics
6
:
231
249
.
Progovac,
Ljiljana
.
1996
.
Clitics in Serbian/Croatian: Comp as the second position
. In
Approaching second
, ed. by
Aaron L.
Halpern
and
Arnold M.
Zwicky
,
411
428
.
Stanford, CA
:
CSLI Publications
.
Pruitt,
Kathryn
.
2008
.
Mapping prosody to interpretation in alternative questions
.
Poster presented at the 21st CUNY Conference on Human Sentence Processing, University of North Carolina, Chapel Hill, 13–15 March
.
Pruitt,
Kathryn
, and
Floris
Roelofsen
.
2013
.
The interpretation of prosody in disjunctive questions
.
Linguistic Inquiry
44
:
632
650
.
Rivero,
María-Luisa
.
1993
.
Bulgarian and Serbo-Croatian yes-no questions: V0-Raising to -li vs. -li hopping
.
Linguistic Inquiry
24
:
567
575
.
Rizzi,
Luigi
.
1997
.
The fine structure of the left periphery
. In
Elements of grammar: Handbook in generative syntax
, ed. by
Liliane
Haegeman
,
281
337
.
Dordrecht
:
Kluwer
.
Rizzi,
Luigi
.
2001
.
On the position Int(errogative) in the left periphery of the clause
. In
Current studies in Italian syntax: Essays offered to Lorenzo Renzi
, ed. by
Guglielmo
Cinque
and
Giampaolo
Salvi
,
287
296
.
Amsterdam
:
Elsevier
.
Roelofsen,
Floris
, and
Kathryn
Pruitt
.
2011
.
Disjunctive questions: Prosody, syntax, and semantics
.
Paper presented at a seminar at the Georg August Universität Göttingen, April 2011. Available at http://www.illc.uva.nl/inquisitive-semantics
.
Roelofsen,
Floris
, and
Sam van
Gool
.
2010
.
Disjunctive questions, intonation and highlighting
. In
Logic, language, and meaning: Selected papers from the 17th Amsterdam Colloquium
, ed. by
Maria
Aloni
,
Harald
Bastiaanse
,
Tikitu de
Jager
, and
Katrin
Schulz
,
384
394
.
Berlin
:
Springer
.
Sailor,
Craig
.
2013
.
Questionable negation
.
Paper presented at the 87th annual meeting of the Linguistic Society of America, Boston, MA, 3–6 January
.
Sauerland,
Uli
,
Jan
Anderssen
, and
Kazuko
Yatsushiro
.
2005
.
The plural is semantically unmarked
. In
Linguistic evidence
, ed. by
Stephan
Kepser
and
Marga
Reis
,
413
434
.
Berlin
:
Mouton de Gruyter
.
Schütze,
Carson
.
1994
.
Serbo-Croatian second position clitic placement and the phonology-syntax interface
. In
Papers on phonology and morphology
, ed. by
Andrew
Carnie
and
Heidi
Harley
with
Tony
Bures
,
373
473
.
MIT Working Papers in Linguistics 21
.
Cambridge, MA
:
MIT, MIT Working Papers in Linguistics
.
Schwarz,
Bernhard
.
1999
.
On the syntax of either . . . or
.
Natural Language and Linguistic Theory
17
:
339
370
.
Stjepanović,
Sandra
.
1999
.
What do second position cliticization, scrambling, and multiple wh-fronting have in common?
Doctoral dissertation, University of Connecticut, Storrs
.