Modes of interrogatives entail modes of sluicing

0 downloads 0 Views 151KB Size Report
Modes of interrogatives entail modes of sluicing. Evidence from Emirati Arabic. Tommi Leung. United Arab Emirates University. Recent studies of sluicing as an ...
Modes of interrogatives entail modes of sluicing Evidence from Emirati Arabic Tommi Leung

United Arab Emirates University Recent studies of sluicing as an elliptical construction are divided with respect to how the bare wh-word in the sluicing clause (i.e. wh-sluice) manifests its expected grammatical properties on the one hand, and receives its semantic interpretation on the other hand. In this paper, I investigate Emirati Arabic (ea) sluicing and conclude that ea sluicing should be analyzed as tp-deletion from an underlying wh-construction, at the level of pf. This supports the pf-deletion approach (Ross 1969, Merchant 2001) and argues against the lf-copying approach (Chung, Ladusaw and McCloskey 1995, 2006, Chung 2005) to sluicing. Moreover, I demonstrate that the ea sluicing source is predetermined by the  type of wh-construction in ea. ea allows two types of wh-constructions, namely wh-fronting and wh-clefts. Both wh-strategies, while morphosyntactically distinct, are fully attested in the formation of sluicing. This paper also claims that the typology of wh-constructions has a direct impact on the typology of sluicing. Keywords: Emirati Arabic, Sluicing, Wh-fronting, Wh-cleft, Ellipsis

1. Introduction The central discussion of this paper stems from some recent inquiries concerning the syntactic properties of sluicing, a term which was first coined by Ross (1969), and has been revived in the seminal work by Merchant (2001). The textbook example of sluicing is shown in (1), and in the coming pages I shall refer to the bare wh-word within the brackets in the second clause (i.e. sluicing clause) ‘wh-sluice’, e.g. (1):

(1) John met someone, but I don’t know [who].

 Tommi Leung

In this paper, I investigate Emirati Arabic (ea) sluicing and address the following issues: which syntactic theory is more optimal to account for the syntactic and semantic properties of the sluicing facts in ea? Two major approaches to sluicing, namely pf-deletion (Ross 1969, Merchant 2001, Toosarvandani 2008) and lfcopying (Chung et al. 1995, 2006, Chung 2005), will be discussed and compared.1 I shall demonstrate that overall, ea is one such language which supports the pf-­deletion approach to sluicing. Relevant evidence is drawn from the fact that ea licenses two sluicing sources which mirror two types of wh-constructions. I  shall show that both wh-constructions, namely wh-fronting and whclefts, are structurally independent and moreover fully attested in ea sluicing. Since the two wh-constructions do not exhibit identical behaviors, the two sluicing sources cannot be in free variation, but instead their choice depends on other factors such as the numeration of the antecedent clause, and the particular lexical choice of the wh-sluice, which will be discussed later. The pf-deletion approach also leads to the claim that the elided parts of sluicing must be derived from the two wh-constructions generated in narrow syntax, which are subsequently deleted at the level of pf. I shall show that the pf-deletion approach to ea sluicing is more descriptively adequate, whereas the lf-copying approach has its weaknesses, both conceptually and empirically. The analysis of ea sluicing by pf-deletion also gives rise to a testable claim. I propose that wh-constructions are structurally deterministic of sluicing, and moreover, the modes of wh-constructions of language L entail the modes of sluicing in L. The paper is structured as follows: Section 2 illustrates two types of wh-constructions in ea. Section 3 introduces some significant properties of ea sluicing. Section 4 discusses the issue of sluicing source and the two approaches to sluicing, and Section 5 concludes the paper. 2. Two Types of wh-constructions 2.1

Wh-fronting

ea is similar to other Arabic dialects with regard to the formation of wh-questions (e.g. Wahba 1984, Holes 1990, Watson 1993, Aoun and Choueiri 1996, 1999, Brustad 2000, Benmamoun 2000, Mohammad 2000, Choueiri 2002, Aoun and Li 2003, Aoun, Benmamoun and Choueiri 2010). For the first type which is attested 1. Earlier versions of this paper were presented in 46th and 47th Annual Meeting of the Chicago Linguistics Society (cls 46, 47), and 25th Annual Symposium on Arabic Linguistics (als 25). I would like to thank the audience and reviewers for comments and corrections. All errors are mine. Some studies, such as Craenenbroeck (2010), claim that both approaches to sluicing may be needed.



Modes of interrogatives entail modes of sluicing 

in all Arabic dialects, a fronted wh-word or wh-phrase is linked to a gap at the base position (similar to English wh-questions). Let us call this wh-fronting or the gap strategy. Wh-fronting can be productively formed by wh-arguments as well as whadjuncts. For instance:2 (2) a. [ʃu]i ʔɛʃtr-ɛt ti ? what bought-2sm ‘What did you buy?’ b. [ʔaj kitab] i ʔɛʃtr-ɛt ti? which book bought-2sm ‘Which book did you buy?’ ti ? c. [mɛno]i ʃɛf-t Who saw-2sm ‘Who did you see?’ d. [weɪn]i laag-ɛt John ? where met-2sm John ‘Where did you meet John?’ e. [mɛta]i laag-ɛt John ti ? when met-2sm John ‘When did you meet John?’ ɛl-waGɛb ti? f. [keɪf]i xəlʕlʕS-t how finished-2sm the-assignment ‘How did you finish the assignment?’ g. [leɪʃ]i mɛtʔxxɛr ti? why late ‘Why are you late?’ It has been proposed in various works (e.g. Aoun and Choueiri 1996, Brustad 2000, Mohammad 2000, Choueiri 2002, Aoun and Li 2003, Aoun et al. 2010) that Arabic wh-fronting is formed by overt wh-movement. This claim is attested in ea mainly by means of island constraints (Ross 1967). For reasons of space, I only demonstrate how strong islands such as the Complex np Island Constraint are observed in wh-fronting. (3) a. *ʃui t-ʕarf [np ɛr-rayaal [ɛlli ʔəʃtra  ti]]? what 2S-know the-man that bought ‘What do you know the man who bought?’ 2. The abbreviations are used as follows: 1-first person, 2-second person, 3-third person, du-dual; F-feminine; fut: future marker; M-masculine; pl-plural; S-singular. For reasons of space, some morphemes may not be fully listed.

 Tommi Leung

b. *mɛnoi t-ʕarf [np ɛr-rayaal [ɛlli laag-eet  ti]]? who 2S-know the-man that met-3sm ‘Who do you know the man who met?’ c. *weɪni t-ʕarf [np ɛr-rayaal [ɛlli y-sken  ti]]? where 2S-know the-man that 3sm-live ‘Where do you know the man who lives?’ d. *mɛttai t-ʕarf [np ɛr-rayaal [ɛlli raaħ l-ħəflah  ti]]? when 2S-know the-man that went the-party ‘When do you know the man who went to the party?’ e. *keɪfi sɛməʕ-t [np saalfət [ʔənna John ħəl əl-məʃkəlah  ti]]? how heard.2sm story that John solved the-problem ‘How did you hear the story that John solved the problem?’ [np l-qəssah [John j-krah Mary  ti]]? f. *leɪʃi sɛməʕ-t why heard-2sm the-story John 3sm-hate Mary ‘Why did you hear the story that John hates Mary?’ Another argument that wh-fronting involves overt wh-movement comes from Superiority effects (Chomsky 1973). This can be shown when two wh-words are situated in root (4) and embedded contexts (5) respectively. Example (4b) and (5b), which violates Superiority, are worse than (4a) and (4b), respectively: (4) a. ?meno y-ħeb meno? who 3sm-like who ‘Who likes who?’ b. *menoi meno y-ħeb  ti? who who 3sm-like ‘Who does who like?’ y-ʃtri ʃu? (5) a. ??menoi ʔqnəʕ-t ti who persuaded-2sm 3sm-buy what ‘Who did you persuade to buy what?’ meno yʃtri  ti? b. *ʃui ʔqnəʕ-t what persuaded-2sm who 3sm-buy ‘What did you persuade who to buy?’ The above facts suffice to lead to the following generalization in ea:

(6) Emirati Arabic wh-fronting observes movement constraints.



Modes of interrogatives entail modes of sluicing 

2.2 Wh-clefts Another type of wh-construction that is available in ea is more controversial in that its detailed grammatical properties may not be attested in all Arabic dialects. Let us call this wh-clefts or the resumptive strategy.3 In ea, wh-clefts are defined by (i) an obligatory complementizer ɛlli ‘that’ which is characteristic of a cleft/relative construction, and (ii) an obligatory object resumptive pronoun at the base position.4 Let us start from some basic examples in (7): ʔmɛs? (7) a. ʃui ɛlli ʔɛʃtr-ɛt-ahi what that bought-2sm-it yesterday ‘What did you buy (it) yesterday?’ fɛ ɛl-mTʕəm? b. mɛnoi ɛlli ʃɛf-t-ahi who that saw-2sm-him in the-restaurant ‘Who did you see (him) in the restaurant?’ By contrast, wh-clefts are strictly ungrammatical without a relative marker or a resumptive pronoun:5,6 (8) a. *ʃui ʔɛʃtr-ɛt-ahi? what bought-2sm-it ‘What did you buy (it)?’ b. *ʃu ɛlli ʔɛʃtr-ɛt? what that bought-2sm ‘What did you buy?’ 3. The distinction between wh-fronting and wh-clefts is reminiscent of Shlonsky’s (2002) distinction between ‘Class I interrogatives’ and ‘Class ii interrogatives’, respectively, for Palestinian Arabic (pa). pa’s Class ii interrogatives are also marked by the obligatory use of a complementizer (i.e. ʔilli ‘that’) and a resumptive pronoun. 4. In this paper, I restrict the discussion of resumptive pronouns in ea (and probably other Arabic dialects as well) to weak pronouns, i.e. direct object and indirect object pronouns. Weak pronouns are morphologically cliticized to the verbal/prepositional host. On the other hand, subject (or strong) resumptive pronouns are banned in simple relative clauses, and also in wh-clefts (instead a subject gap is used in wh-clefts). The ban on subject resumptive pronouns in Arabic dialects was discussed by Shlonsky (1992, 2002) viz-a-viz the Highest Subject Restriction (hsr). See McCloskey (1990) for the original discussion of hsr. 5. Example (8) can be subject to dialectal variations, i.e. it is not surprising that some Arabic dialects allow the co-occurrence of a wh-word and a resumptive pronoun without an intervening complementizer. 6. Again, as pointed out in footnote 4, subject wh-clefts do not observe this condition, i.e. a subject gap (instead of a subject resumptive pronoun) is used in wh-clefts. I suggest that this is due to the overall ban on subject resumptive pronouns in simple relative clause and wh-questions.

 Tommi Leung

Apparently, wh-clefts and wh-fronting are independent structures, in addition to the different use of lexical items. First, as we have shown before, wh-fronting allows the use of all types of wh-words and wh-phrases. On the other hand, wh-clefts are more restricted. One salient difference is that wh-clefts strictly ban the occurrence of wh-adjuncts such as keɪf ‘how’ and leɪʃ ‘why’, and wh-words contained within a pp such as b-ʔaj tʕriqah ‘by which way’ (9):7,8 (9) a. *[keɪf] ɛlli xəlʕlʕS-t ɛl-waGɛb ti? how that finished-2sm the-assignment ‘How did you finish the assignment?’ b. *[leɪʃ] ɛlli mɛtʔxxɛr ti? why that late ‘Why are you late?’ c. *[b-ʔaj tʕriqah] ɛlli xəlʕlʕsʕ-t əl-waaʤɪb? by-which way that finished-2sm the-assignment ‘By which way did you finish the assignment?’ Second, the structure of wh-clefts embeds a relative structure. This is evident in (10), in which the head noun can be used after the wh-word: (10) ʃu (ʃ-ʃay) ɛlli ʔɛʃtr-ɛt-ah? what the-thing that bought-2sm-it ‘What is the thing that you bought (it)?’ The wh-word in wh-clefts can also be followed by a copular pronoun hu:9 (11) ʃu hu ɛlli ʔɛʃtr-ɛt-ah? what 3sm that bought-2sm-it ‘What did you buy?’

(lit. ‘What is it that you bought (it)?’)

The copular pronoun is a defining feature of cleft structures (12) and is incompatible with wh-fronting (13): (12) hu al-kitab ɛlli John ʔakra-ah. 3sm the-book that John read-it ‘It is the book that John read (it).’ 7. The ungrammaticalityof wh-adjuncts in wh-clefts can also be explained by the fact that they are non-referential and therefore cannot be related by an object resumptive pronoun. This corresponds to earlier claims that resumptive pronouns are D-linked (Doron 1982, Sells 1984, ErteschikShir 1992, Sharvit 1999). Thanks to one reviewer for pointing out some of these references. 8. Other dialects such as Egyptian Arabic (Wahba 194) and Palestinian Arabic (Shlonsky 2002) observe a similar distinction between different types of wh-expressions in the expressions of wh-clefts. 9. C.f. Eid (1983).



Modes of interrogatives entail modes of sluicing 

(13) *[ʃu]i hu ʔɛʃtr-ɛt ti? what 3sm bought-2sm ‘What did you buy?’ Third, wh-clefts (also cleft structures as well) obligatorily include a resumptive pronoun at the base position (instead of a gap), which suggests that the structure may be derived by base-generation of the wh-word at the sentence-initial position and not necessarily by wh-movement.10 This is shown by the violation of movement constraints: (14) a.

?ʃu

ɛlli t-ʕarf ɛr-rayaal ɛlli ʔəʃtra-h? what that 2sm-know the-man that bought-it ‘What is the thing that you know the man that bought (it)?’

b. ʃu ɛlli t-ba t-ʕarf ʔɛða John ʔəʃtra-h? what that 2sm-want 2sm-know whether John bought-it ‘What is the thing you want to know whether John bought (it)?’ To schematize the two wh-constructions: (wh-fronting) (15) a. whi... ti? b. whi... ɛlli...proi? (wh-clefts) In the following section, we look at how the two wh-constructions shed light on the grammatical properties of ea sluicing. 3. Properties of Emirati Arabic sluicing ea is similar to other languages in which a bare wh-word can be used to form sluicing. I call this bare wh-word ‘wh-sluice’. The following sentences illustrate some typical examples: (16) a. ħəd jəʃrʌb al-gahwa, bəs maa ʕərf [mənu]. someone drink the-coffee but not 1S.know who ‘Someone drinks coffee, but I don’t know [who].’ b. John jəʃrʌb ʃaj, bəs maa ʕərf [ʃu]. John drink something but not 1S.know what ‘John drinks something, but I don’t know [what].’ c. John laaga Mary, bəs maa ʕərf [weɪn]. John met Mary but not 1S.know where ‘John met Mary yesterday, but I don’t know [where].’ 10. This claim, however, should not be taken as a refutation of any movement approach to resumption (e.g. Boeckx 2003).

 Tommi Leung

d. John kaan jəʃrʌb xʌmər, bəs maa ʕərf [ʔaj nooʕ]. John was drink alcohol but not 1S.know which kind ‘John was drinking alcohol, but I don’t know [which kind].’ e. John jəʃrʌb al-gahwa, bəs maa ʕərf [wɪjja mənu]. John drink the-coffee but not 1S.know with who ‘John drinks coffee, but I don’t know [with who].’ Recent theories of sluicing focus on the mechanism by which the wh-sluice receives its semantic interpretation, an issue which I address in Section 4. Regardless of the approach to sluicing, cross-linguistic studies show convincingly that the wh-sluice displays connectivity effect with its correlate in the antecedent clause. That is to say, the wh-sluice retains the syntactic and semantic properties of its antecedent correlate (whether it is overt or implicit). This is fully observed in ea. ea sluicing observes binding reconstruction (17), scope reconstruction (18), and idiom chunk reconstruction (19): (17) a.

kɪl lʊɤawii əntəqad baʕaðʕ ʃəɤl-ah, bəs all linguist criticized some work-his but maa mətʔakd-ah [ʃ-kəər]. not sure-1sf what-much ‘Every linguisti criticized some of hisi work, but I’m not sure how much (of hisi work).’

b.

John taʕjəb-ah sʕurah mɪn sʕʊwar-ah, bəs John like-3sm picture of pictures-his but maa ʕərf [ʔaj sʕurah]. not 1S.know which picture ‘Johni likes one of hisi pictures, but I don’t know which (of hisi pictures).’

(18) a.

jtkallam lʊɤah ʕarabɪjjah, hʊm jboon jwaðʕf-uun ħəd 3pl want hire someone talk language Arabic bəs maa ʔðkər [ʔaj lʊɤah]. but not 1S.know which language ‘They want to hire someone who speaks an Arabic language, but I don’t remember [which language].’ (which language>want)

b. əl-kɪl jħəb ħəd, bəs maa ʕərf [mənu]. the-all like someone but not 1S.know who ‘Everyone likes someone, but I don’t know [who]. (who>everyone, everyone > who)



Modes of interrogatives entail modes of sluicing 

(19) a.

John ðʕərab ʕasʕfuur-een, bəs maa ʕərf [ʔaj ʕasʕfuur-een]. John hit bird-du but not 1S.know which bird-du ‘John solved two problems, but I don’t know [which two problems].’ (lit. ‘John hit two birds, but I don’t know which two birds.’)

b.

John ħərag wajh ħəd, bəs maa ʕərf [wajh mənu]. John burned face someone but not 1S.know face who ‘John insulted someone, but I don’t know [who].’ (lit. ‘John burned someone’s face, but I don’t know whose face).

Another well-known property of sluicing is island insensitivity (Ross 1969). While the connectivity effect reveals a formal link between the antecedent clause and the wh-sluice, the bare wh-sluice does not appear to observe island constraints established in the antecedent clause. In (20), the antecedent correlate ħəd ‘someone’ is contained within the Complex np Island constraint, yet the wh-sluice appears to ignore it: (20) ʔana səmaʕt [np əl-ɪʃaaʕah [ɪnna Mary laaga-t I heard the-rumor that Mary met-3sf ħəd]], bəs maa ʕərf [mənu]. someone but not 1S.know who ‘I heard the rumor that Mary met someone, but I don’t know [who].’ The third property of sluicing originates in the seminal work by Merchant (2001) and is subject to more controversy (given that the first two properties are theory-independent, whereas the third one assumes a particular model of syntax in which the division of narrow syntax and interface conditions is clear-cut). Merchant claims that while the wh-sluice completely ignores island constraints since island constraints are pf/interface conditions as recent syntactic theory assumes, it obeys robust constraints defined by narrow syntax. Starting from Chomsky (1995) and advocated by most syntacticians, there is a fundamental difference between constraints as defined by narrow syntax and those defined by interface conditions. Narrow syntax concerns exclusively the issue of formal feature checking/licensing and structure-building mechanism (i.e. a computational algorithm of syntax), whereas interface conditions are external to the central computational system. Instead the latter are subject to other biological or cognitive constraints. Their differences have a deterministic consequence on the effect of repair strategies (including sluicing). Since repair strategies (e.g. ellipsis) are effective merely at the interface level, all interface constraints (e.g. island constraints) can be repaired via such strategies. On the other hand, violations at the level of narrow syntax are too robust to repair. Among many other syntactic conditions, Merchant claims that the condition on preposition stranding is robustly defined by narrow syntax, and moreover proposes the Preposition Stranding Generalization (psg) (Merchant 2001) for sluicing:

 Tommi Leung

(21) A language L will allow preposition stranding under sluicing iff L allows preposition stranding under regular wh-movement. That is to say, P-stranding as a result of wh-movement is viewed as not a ‘speaking’ or ‘hearing’ issue, but a computational one. Merchant argues for the descriptive adequacy of the psg by listing more than twenty languages.11 In this regard, Arabic is well known as a non-P-stranding language. For all instances of wh-movement (e.g. wh-fronting), if the wh-word is contained within a pp, the whole pp has to be pied-piped (22b). On the other hand, wh-fronting cannot strand a preposition, e.g. (22a): (22) a. *[ʔaj mʊkaan] laag-eet John fi? which place met-2sm John at ‘Which place did you meet John at?’ b. [f-ʔaj mʊkaan] laag-eet John ? at-which place met-2sm John ‘At which place did you meet John?’ Given the ban on P-stranding in ea, it is crucial to note that in (23), the wh-sluice ʔaj sʕadiq ‘which friend’ can either be bare, or be preceded by a preposition wijja ‘with’. The use of a bare wh-sluice suggests that it can be the result of wh-movement which stranded a preposition, followed by tp-deletion at pf (schematized in (24)): (23)

John ʃərab gahwa [pp wijja sʕadiq], bəs maa John drank coffee with friend but not ʕərf [(wijja) ʔaj sʕadiq]. 1S.know with which friend ‘John drank coffee with a friend, but I don’t know [(with) which friend].’

(24) ... bəs maa ʕərf [cp ʔaj sʕadiqi...[pp wijja ti]]. Note, however, that the structural description of the antecedent clause has a strong impact on whether the preposition can be stranded at the sluicing clause an important observation, which is missed by the statement of the psg. (23) is a case of Merger in which the antecedent correlate (i.e. sʕadiq ‘friend’) is overt. In the case of Sprouting (e.g. the antecedent correlate is not overt), the preposition in the sluicing clause is obligatory. For instance, (25) shows that the preposition mɪn ‘from’ is 11. There are, however, individual differences with respect to the judgment of the psg. My informants do not fully agree with Merchant’s judgment of Greek, or provide alternative explanations for the ungrammaticality of some examples that are not consistent with the psg (e.g. Moroccan Arabic).



Modes of interrogatives entail modes of sluicing 

obligatory since the wh-sluice, which is a wh-pp, does not have a pp-correlate at the antecedent clause:12 (25) John ʃtəra ktaab, bəs maa ʕarf [*(mɪn) wɛɪn]. John bought book but not 1S.know from where ‘John bought a book, but I don’t know [*(from) where].’ As a result, it is important to keep in mind that whether the psg is violable depends crucially (though not exclusively) on the numeration of the antecedent clause, a point to which I will return shortly. 4. Determining the sluicing source As mentioned above, the central issue surrounding sluicing is the source of the wh-sluice. A descriptively adequate theory has to properly address the three properties of sluicing mentioned in Section 3 (i.e. connectivity, island insensitivity, and the psg). Two competing approaches to sluicing, i.e. pf-deletion and lf-copying, are introduced here. As the name suggests, the pf-deletion approach (Ross 1969, Merchant 2001) claims that sluicing is the result of ellipsis at the level of pf. Accordingly, the bare wh-sluice is derived from wh-movement to Spec-cp, followed by tp-deletion at pf. Under this approach, the semantic interpretation of the whsluice is determined at the point of Spell-out (which gives rise to the lf representation for free): (26) Someone called, but I don’t know [cp who [tp ti called]] (spell-out) → Someone called, but I don’t know [cp who [tp ti called]] (pf) On the other hand, the lf-copying approach (Chung et al. 1995, 2006, Chung 2005) argues that no ellipsis whatsoever occurs at the level of pf. That is to say, syntax generates nothing more than a bare wh-sluice. At the level of lf, an interpretive rule is postulated which copies particular elements from the antecedent clause to the sluicing clause (example drawn from Merchant 2001):13 (27) Someone called, but I don’t know [cp who [e]] (spell-out) → Someone called, but I don’t know [cp who [tp someone called]] (by lf-copying) → ... [cp whox [tp someonex called]] (lf-interpretation) Logical form: λp[∃x.person (x, w0)∧p(w0) ∧ p = λw.call(x,w)] 12. The term ‘Merger’ and ‘Sprouting’ originates from Chung et al. (1995, 2006). 13. Such a copying mechanism is also conditioned by the mapping rule between numeration and spell-out. For some recent developments along this line, please refer to Chung (2005). See also the following discussion.

 Tommi Leung

The two approaches differ crucially with regard to how well formdeness is evaluated. For the pf-deletion approach, it is about whether the grammatical constraints involved are stated at the level of narrow syntax or at pf, whereas for the lf-copying rule, structural descriptions would be overall grammatical if a proper semantic interpretation (in the form of variable binding such as the logical form in (27)) can be obtained. At the outset, both approaches are descriptively adequate in that the three abovementioned properties of sluicing can be fully addressed. The connectivity effect can be accounted for by establishing a formal link between the antecedent clause and the sluicing clause, and the two approaches differ merely in where to define such a formal link (i.e. narrow syntax vs. pf). For island insensitivity of sluicing, the pfdeletion approach states that islands are interface conditions that can be repaired by ellipsis, whereas the lf-copying approach says nothing about island repair since no movement operation has ever occurred (i.e. they postulate that the wh-sluice is not the result of wh-movement)! For the psg, the pf-deletion approach suggests that P-stranding is a robust syntactic condition which cannot be repaired at pf. On the other hand, since the lf-copying approach completely dispenses with movement operations, something needs to be said which defines the psg as an interpretive condition on the wh-word contained within a pp (which leads to variable binding violations, for instance).While Merchant claims that the psg is a big challenge to the lf-copying approach, I consider his opposition inconclusive given that the descriptive adequacy of the psg remains questionable, and moreover there is a possibility in which the psg can be stated as an interface condition (contra Merchant). Such option has been recently considered and typologically supported, and it is therefore expected that the psg can be parametricized as either a syntactic or an interface condition, an issue that I will put aside in this paper.14,15 Indeed, given that ea allows two types of wh-constructions, the issue with regard to the sluicing source becomes a challenging one. Several levels of inquiries 14. See Leung (2014) for further arguments for the parametric theory of the PSG in Emirati Arabic sluicing, and Sato’s (2011) analysis of Indonesian sluicing and his proposal of parametrization of the psg. 15. Merchant (2001) claims that the psg comes as a surprise to the lf-copying approach since variable binding at lf should be insensitive to movement constraints (which locates P-stranding). One possible suggestion given by the lf-copying approach, which was pointed out by Bayer (1996) and is still under controversy, is to postulate that P-stranding is an lf constraint. Combining this with lf-copying means that in the process of lf-copying, the indefinite in the antecedent clause (e.g. someone called) functions as a generalized quantifier and moves to Spectp as expected, whereas lf-conditions such as chain uniformity rules out psg violation. Consider the following English translation of non-P-stranding languages such as Greek:

(i) a. John spoke [pp with someone], but I don’t know [with who [e]].



b. *John spoke [pp with someone], but I don’t know [who [e]].



Modes of interrogatives entail modes of sluicing 

have to be considered. First, are the two wh-constructions fully observed in sluicing, or is it the case that one construction is always preferred in the formation of sluicing? Second, assuming that both wh-constructions can be used in sluicing (which I shall argue to be the case), does it provide stronger support to the pf-deletion or the lfcopying approach? How is each approach compatible with the two sluicing sources? Third, linguists note that some languages present a pseudosluicing structure, i.e. a structure that surfaces like sluicing, yet their underlying structure does not involve wh-movement out of a full-fledged wh-­construction (Erteshik-Shir 1977, Merchant 2001). The pseudosluicing structure usually contains a cleft structure with a copula as a pivot, schematized as follows:16 (28) ... but I don’t know [cp who [ip it is]]. Merchant provides extensive discussions of the grammatical distinctions between sluicing and pseudosluicing, and his underlying motivation is that any language/ sentence which does not conform to the psg actually stems from a pseudosluicing  source. Since the psg is only effective in sluicing, those counterexamples would become irrelevant. I deal with the third issue first since the sluicing-vspseudosluicing discussion is merely tangential in ea. To begin with, recall that wh-clefts allow the use of an optional copular pronoun after the wh-word. Example (11) is repeated here: (29) ʃu hu ɛlli ʔɛʃtr-ɛt-ah ? what 3sm that bought-2sm-it ‘What did you buy?’ (lit. ‘What is it that you bought (it)?’) Such an optional copular pronoun can also be found after the wh-sluice: (30) John ʃərʌb ʃaj, bəs maa ʕarf [ʃuu (hu)]. John drank something but not 1S.know what 3sm ‘John drank something, but I don’t know [what (it is)].’ The antecedent pp moves to Spec-tp and leaves a pp trace at lf. The pp trace is only typecompatible with the variable under the sluicing clause (ii-a). On the other hand, (ii-b) which corresponds to (i-b) violates chain uniformity, i.e.:

(ii) a. [pp with who], [John spoke [pp ti]]



b. *[dp who], [John spoke [pp ti]]

16. See Fukaya and Hoji (1999), Hiraiwa and Ishihara (2002), Takahashi (2004) and Fukaya (2003, 2007) for discussions of Japanese pseudosluicing, and Rodrigues, Nevins and Vicente (2009) for Spanish pseudosluicing. The distinction between sluicing and pseudosluicing is crucial to other languages since Merchant claims that the psg is only effective in sluicing but not in pseudosluicing. As a result, any support for or refutation against the psg from the analysis of pseudosluicing languages would be powerless.

 Tommi Leung

However (30) should not be taken as an example of pseudosluicing with nothing but a bare wh-word and a copula (c.f. Japanese; also see the schema (28)). Alternatively, (30) would be considered as a case of ellipsis of either wh-fronting or wh-clefts. It is mainly because the use of copular pronouns in sluicing follows exactly the same restrictions as in normal wh-constructions. To name a few, while wh-adjuncts can be used as a wh-sluice, they cannot be followed by a copular pronoun:17 (31) John laaga Mary, bəs maa ʕarf [wɛɪn/məta (*hu)]. John met Mary but not 1S.know where/when 3sm ‘John met Mary, but I don’t know [where/when (*it is)].’ This completely mirrors wh-constructions, i.e. the copular pronoun cannot be a pivot for wh-adjuncts: (32) a. [weɪn] (*hu) laag-ɛt John? Where 3sm met-2sm John ‘Where did you meet John?’ b. [məta] (*hu) laag-ɛt John? when 3sm met-2sm John ‘When did you meet John?’ 17. A reviewer questions if the ungrammaticality of (31) is due to tense mismatch between the antecedent clause (which is in the past) and the use of hu as a present tense copular pronoun: (i) Ahmed hu teacher. Ahmed 3sm teacher ‘Ahmed is a teacher.’  or past and future tense, the auxiliary kaan ‘was’ and baykuun ‘will be’ should be used. On the F other hand, hu is strictly banned in these contexts: (ii) Ahmed kaan teacher. Ahmed be.past.3sm teacher ‘Ahmed was a teacher.’ (ii) Ahmed ba-ykuun ustaadh. Ahmed fut-be.3sm teacher ‘Ahmed will be a teacher.’ Still, the use of the past tense auxiliary is incompatible with wh-adjuncts: (iv) *John gaabl Mary, bəs maa ʕarf [məta kaan] John met Mary but not 1S.know when be.past.3sm ‘John met Mary, but I don’t know when it was.’  n the other hand, example (33) also shows that the copular pronoun hu is compatible with O other tenses. This suffices to show that (31) is not a problem of tense mismatch, but instead a clash between wh-adjuncts and copular pronouns.



Modes of interrogatives entail modes of sluicing 

Moreover the copular pronoun is gender-sensitive depending on the intended referent of the wh-word. If the gender of the wh-word is presupposed, the copular pronoun should be gender-informative, followed by the use of a resumptive pronoun of the respective gender: ʔms? (33) a. [ʃuui] hu ɛlli ʃtər-et-ah what 3sm that bought-2sm-him yesterday ‘What did you buy (it) yesterday?’ (unmarked, or if an entity of masculine gender is presupposed) ʔms? b. [ʃuui] hi ɛlli ʃtər-et-ha what 3sf that bought-2sm-her yesterday ‘What did you buy (it) yesterday?’ (if an entity of feminine gender is presupposed) Gender-sensitivity is also observed in sluicing: (34) John laaga bint, bəs maa ʕarf [mənu (hi/*hu)]. John met girl but not 1S.know who 3sf/3sm ‘John met a girl, but I don’t know who (she is/*he is).’ This suffices to show that the wh-sluice with a copular pronoun in (34) is the result of ellipsis from a full wh-construction (in this case we mean wh-clefts) instead of base-generation. On the other hand, the pseudosluicing approach appears to lack the level of descriptive adequacy displayed by the ellipsis approach. Unless we are faced with more compelling empirical support, the pseudosluicing approach to ea should be discarded.18,19 I have demonstrated that the two wh-constructions can be fully observed in sluicing, giving further support to the pf-deletion approach in which the wh-sluice derives from full-fledged wh-questions. Certainly we should notice that examples such as (35) may not be helpful in determining the sluicing source, since the whsluice mənu ‘who’ is compatible with either wh-fronting or wh-clefts:

18. The claim against pseudosluicing in ea is independent of whether the pf-deletion approach or the lf-copying approach should be adopted here. In this paper, ‘ellipsis’ is a term for all fragmented constructions with a formal link with other full-fledged underlying structures. The issue of how the underlying structure is retrieved from the fragments is a theory-internal issue though. 19. The lack of pseudosluicing in ea also entails that Merchant’s distinctions between sluicing and pseudosluicing in a single language (e.g. English) is not useful in the discussion of ea. In this paper we focus mainly on the distinction between wh-fronting and wh-clefts, both of which are full-fledged wh-constructions.

 Tommi Leung

(35) ħəd jəʃrʌb al-qahwa, bəs maa ʕərf [mənu]. someone drink the-coffee but not 1S.know who ‘Someone drinks coffee, but I don’t know who.’ That is to say, the two following schemas for (35) are equally plausible: (wh-fronting) (36) a. ... bəs maa ʕərf [cp mənui [ti jə-ʃrʌb al-qahwa]] b. ... bəs maa ʕərf [cp mənui [ti ɛlli jə-ʃrʌb al-qahwa]] (wh-clefts) However we notice that there are clear distinctions between wh-fronting and wh-clefts, and we assume that those distinctions can be preserved in ea sluicing. For instance wh-adjuncts and wh-words contained within a pp, which can only be used in wh-fronting, can also be found in ea sluicing, implying that the wh-sluice must be the result of ellipsis from wh-fronting, not wh-clefts, i.e.: (37) a. John laaga Mary, bəs maa ʕarf [wɛɪn [John laaga Mary ti]]. John met Mary but not 1S.know where John met Mary ‘John met Mary, but I don’t know [where]’ b. John ʃərab gahwa [wɪjja ħəd], bəs maa ʕarf John drank coffee with someone but not 1S.know [[wɪjja mənu] John ʃərab gahwa ti]]. with who John drank coffee ‘John drank coffee with someone, but I don’t know [with who].’ In addition, as we mentioned before, the numeration of the antecedent clause has an impact on the numeration of the sluicing source (e.g. (25)). In the case of Sprouting which contains an implicit antecedent correlate, wh-fronting is the only possible sluicing source. This is evident in (38), in which only the bare mənu ‘who’ can be used. On the other hand, the copular pronoun which is only compatible with wh-clefts is banned in the absence of an overt antecedent correlate: [mənu (*hu)]. (38) John laaga [dp], bəs maa ʕarf John met but not 1S.know who 3sm ‘John met, but I don’t know [who (*he is)].’ A similar conclusion is reached if the antecedent correlate is contained within a pp which is implicit. As shown in (39), the wh-sluice cannot strand a preposition, i.e. the wh-sluice must be accompanied by a preposition: [*(mɪn) wɛɪn]. (39) John ʃtəra ktaab [pp], bəs maa ʕarf John bought book but not 1S.know from where ‘John bought a book, but I don’t know *(from) where.’



Modes of interrogatives entail modes of sluicing 

Again, this distinction is parallel to the distinction between wh-fronting and wh-clefts, i.e. wh-fronting allows the use of a wh-word contained within a pp (40a), whereas wh-clefts do not (40b, c): (40) a. [mɪn wɛɪn]i John ʃtəra al-ktaab ti? from where John bought the-book ‘[From where] did John read the book?’ b. *[mɪn wɛɪn]i ɛlli John ʃtəra al-ktaab ti? from where that John bought the-book c. *[mɪn wɛɪn]i hu ɛlli John ʃtəra al-ktaab ti? from where 3sm that John bought the-book This strongly suggests that wh-fronting is the only possible sluicing source if the antecedent clause contains an implicit antecedent correlate. On the other hand, there exists at least one case in which only wh-clefts can be used. The expression of exhaustivity is more compatible with wh-clefts than wh-fronting. This can be shown by the use of exhaustive marker əl-kɪl ‘all’. The following shows that exhaustivity cannot be used in wh-fronting: (41) a. [mənu əl-kɪl] ɛlli ʃəf-t-hum? who the-all that saw-2sm-3pl ‘Who all did you see (them)?’ b. *[mənu əl-kɪl] ʃəf-t? who the-all saw-2sm ‘Who all did you see?’ Now consider the antecedent clause in (42), which states that a large number of people attended the party. The wh-sluice requests an exhaustive list of party participants: (42) naas wajɪd jəw əl-ħaflah, bəs maa ʕarf [menu əl-kɪl]. people lot came the-party but not 1S.know who the-all ‘A lot of people came to the party, but I don’t know [who all].’ To stress the relation between the wh-sluice in (42) and wh-clefts, the third person plural copular hʊm can be used: (43) naas wajɪd jəw əl-ħaflah, bəs maa ʕarf [menu hʊm əl-kɪl]. people lot came the-party but not 1S.know who 3pl the-all ‘A lot of people came to the party, but I don’t know [who all].’ To summarize the distinction between wh-fronting and wh-clefts in ea sluicing, the following generalization can be stated:

 Tommi Leung

(44) a. Both wh-fronting and wh-clefts can constitute an underlying source for Emirati Arabic sluicing. b. Wh-fronting is the only sluicing source if (i) the wh-sluice is an adjunct; (ii) the antecedent clause contains an implicit antecedent correlate. c. Wh-clefts are the only sluicing source if the wh-sluice expresses semantic exhaustivity (e.g. əl-kɪl ‘all’). As a corollary, we can safely claim the following: (45) a. Emirati Arabic allows two sluicing sources which mirror two types of wh-constructions. b. The sluicing source can be determined by (i) the particular lexical choice of the wh-sluice (e.g. wh-arguments vs. wh-adjuncts, dp vs. pp), and (ii) the numeration of the antecedent clause (e.g. Merger vs. Sprouting). c. Emirati Arabic supports the pf-deletion approach to sluicing. To schematize the pf-deletion approach to ea sluicing: (46) a. ... but I don’t know [cp whati [tp ...ti]]. b. ... but I don’t know [cp whati [ɛlli [tp ...proi]]].

(wh-fronting) (wh-clefts)

By contrast, there are various reasons to question the descriptive adequacy of the lf-copying approach, at least in the context of ea sluicing. The major problem surrounding the lf-copying approach stems from the origin of the sluicing source. While the exact statement of such an approach has recently undergone a number of revision (e.g. Chung 2005), the fundamental spirit remains intact in the sense that the underlying sluicing source is predetermined by the numeration of the antecedent clause. This, at first glance, fits perfectly into ea sluicing since I have shown that the choice between wh-fronting and wh-clefts is heavily influenced by the numeration of the antecedent clause. However we have also concluded that the numeration of the antecedent clause is not a sufficient condition for the choice of the sluicing source. On the other hand, the lf-copying approach stresses that the numeration of the antecedent clause exclusively determines the choice of the wh-sluice. Consider the following Sprouting example in English that is used as supporting evidence for the lf-copying approach (Chung 2005): (47) a. She’s reading a novel, but I don’t know *(by) who. b. She’s eating a pizza, but I don’t know *(from) which restaurant. In both examples, the preposition in the sluicing clause is obligatory and it establishes a formal link with the antecedent correlate (which is implicit). Chung



Modes of interrogatives entail modes of sluicing 

conceptualizes such a formal link by means of a set-theoretic relation between the numeration of the antecedent and the sluicing clause: (Chung 2005: 83, her (29)) (48) Every lexical item in the numeration of the sluice that ends up (only) in the elided ip must be identical to an item in the numeration of the antecedent cp. Chung’s claim is that all members within the numeration must be spelled out at pf (i.e. no pf-deletion). In (47a), the numeration of the antecedent clause is the set {she, is, reading, a, novel}. The numeration of the sluicing clause contains {by, who}, and therefore the elided tp in the sluicing clause contains {she, is, reading, a novel} which is identical to the numeration of the antecedent clause, hence it obeys (48). If the preposition ‘by’ is elided in (47a), the elided tp of the sluicing clause becomes {she, is, reading, a novel, by}. Since ‘by’ is not part of numeration of the antecedent clause, (48) is violated. On the other hand, (49) is grammatical with or without the preposition ‘by’ in the sluicing clause. This is easily accountable since the numeration of the antecedent clause is {she, is, reading, a, novel, by, someone}. Whether the preposition ‘by’ is numerated in the sluicing clause does not violate the condition in (48): (49) She’s reading a novel by someone, but I don’t know (by) who. However, the fact that wh-clefts can be a sluicing source (e.g. (42)) presents an unsolved puzzle to the lf-copying approach. The question is, how can wh-clefts be drawn from the numeration of the antecedent clause. Imagine that the wh-sluice is derived from wh-clefts (c.f. 46b): (50) ... I don’t know [whoi [Cɛlli [... proi]]]. According to (48), the elided tp of the sluicing clause contains the complementizer ɛlli and a resumptive pronoun. Neither can be part of the numeration of the antecedent clause. That is to say, it is theoretically impossible to establish a formal (i.e. syntactic or lexical) link between the antecedent clause and the sluicing clause in the case of wh-clefts. There is nothing in the antecedent clause from which one can copy to the sluicing clause at lf. As a remedy, one needs to somehow postulate that the sluicing source in ea must be numerated afresh (i.e. after the point of spell-out of the antecedent clause). To conclude, while the numeration of the antecedent clause has a significant impact on the sluicing source, it is by no means the only condition to be met when the numeration of the sluicing source is chosen. The above consideration can be considered as a potential challenge to the lf-copying approach. On the other hand, the pf-deletion approach does not face the same problem. According to the pf-deletion approach, the numeration of the sluicing clause is generated immediately after the antecedent clause is spelled out. In principle, the numeration of the antecedent clause and the sluicing clause do not need to be identical to each other.

 Tommi Leung

5. Conclusion I conclude this paper by stating some claims about sluicing in Emirati Arabic (ea). First, ea allows two major types of wh-constructions, i.e. wh-fronting (i.e. gap strategy) and wh-clefts (i.e. resumptive strategy). The two wh-constructions differ not only in terms of the lexical choice, but also the structural descriptions and behaviors of various processes (e.g. island constraints). Given the two wh-constructions, I propose that ea allows two sluicing sources. The two sluicing sources mirror the two wh-constructions perfectly. This supports the pf-deletion approach to sluicing (Ross 1969, Merchant 2001) which states that sluicing is the result of tp-deletion of a full-fledged wh-construction at the level of pf.20 On the other hand, the lf-copying approach faces a problem of the choice of the sluicing source. Despite the fact that the pf-deletion approach to sluicing gives rise to other problematic issues such as the falsification of the Preposition Stranding Generalization (Merchant 2001), its overall descriptive power should still be regarded. Moreover given the special behavior of wh-constructions in ea, the following typological hypothesis can be proposed: (51) The typology of wh-constructions entails the typology of sluicing. Hypothesis (51) can be a statement within a single language or between languages. If language L has n types of wh-constructions, (51) predicts that n types of sluicing should be fully observed. On the other hand, the wh-sluice may not display properties that are unattested in their unelided counterpart. Across languages, (51) can be understood to mean that if any two languages differ significantly in the structure of wh-constructions, they should also do so in the morphosyntactic properties of sluicing. Further work should not only be limited to the debate between the pf-deletion and the lf-copying approach to sluicing, but also extend to the issue about which of the grammatical properties of sluicing mirror those of wh-constructions. The facts from Emirati Arabic demonstrate convincingly that the grammatical properties of the two wh-constructions are fully preserved in the formation of two types of sluicing.

20. Merchant’s original thesis about sluicing is that the wh-sluice has undergone wh-movement followed by tp-deletion. We therefore depart from his thesis slightly by including the wh-cleft as another type of wh-construction without wh-movement. This, however, does not have any conflict with the central idea that sluicing is analyzed as pf-deletion of a full-fledged wh-construction.



Modes of interrogatives entail modes of sluicing 

References Aoun, J., Benmamoun, E., and Choueiri, L. (2010). The syntax of Arabic. Cambridge: Cambridge University Press. Aoun, J., and Choueiri, L. (1996). Resumption and last resort. Unpublished manuscript, University of Southern California, Los Angeles. Aoun, J., and Choueiri, L. (1999). Modes of interrogation. In E. Benmamoun (Ed.), Perspectives on Arabic linguistics XII (pp. 7–26). Amsterdam: John Benjamins. Aoun, J., and Li, A. (2003). Essays on the representational and derivational nature of grammar: The diversity of wh-constructions. Cambridge, Mass.: MIT Press. Bayer, Y. (1996). Directionality and logical form. Kluwer: Dordrecht. Benmamoun, E. (2000). The feature structure of functional categories. Oxford: Oxford University Press. Boeckx, C. (2003). Islands and chains. Amsterdam: John Benjamins. Brustad, K. E. (2000). The syntax of spoken Arabic. Washington, D.C: Georgetown University Press. Chomsky, N. (1973). Conditions on transformations. In S. R. Anderson and P. Kiparsky (Eds.), A festschrift for Morris Halle (pp. 232–286). New York: Holt, Rinehart and Winston. Chomsky, N. (1995). The minimalist program. Cambridge, Mass: MIT Press. Choueiri, L. (2002). Re-visiting relatives: issues in the syntax of resumptive restrictive relatives. Doctoral dissertation, University of Southern California, Los Angeles. Chung, S. (2005). Sluicing and the lexicon: the point of no return. Proceedings of the 31st meeting of the Berkeley Linguistics Society, University of California, Berkeley, CA. Chung, S., Ladusaw, W., and McCloskey, J. (1995). Sluicing and logical form. Natural Language Semantics, 3, 239–282. Chung, S., Ladusaw, W., and McCloskey, J. (2006). Sluicing revisited. Paper presented at the annual meeting of Linguistic Society of America. Albuquerque. Craenenbroeck, J van. (2010). The syntax of ellipsis: evidence from Dutch dialects. Oxford: Oxford University Press. Doron, E. (1982). On the syntax and semantics of resumptive pronouns. Texas Linguistics Forum, 19, 1–48. Eid, M. (1983). The copula functions of pronouns. Lingua, 59, 197–207. Erteshik-Shir, N. (1977). On the nature of island constraints. Bloomington, IN: Indiana University Linguistics Club. Erteschik-Shir, N. (1992). Resumptive pronouns in islands. In H. Goodluck and M. Rochemont (Eds.), Island constraints (pp. 89–108). Kluwer Academic Publishers, Dordrecht. Fukaya, T. (2003). Island (in)sensitivity in Japanese sluicing and stripping. Paper presented at 22nd West Coast Conference on Formal Linguistics (WCCFL-22). San Diego, CA. Fukaya, T. (2007). Stripping and sluicing in Japanese and some implications. Doctoral dissertation, University of Southern California, Los Angeles. Fukaya, T., and Hoji, H. (1999). Stripping and sluicing in Japanese and some implications. Proceedings of the 18th West Coast Conference on Formal Linguistics (WCCFL-18) 145–158. Cascadilla Press: Somerville, Mass. Hiraiwa, K., and Ishihara, S. (2002). Missing links: cleft, sluicing, and ‘no da’ construction in Japanese. In T. Ionin, H. Ko, and A. Nevins (Eds.), MIT working papers in linguistics 43 (pp. 35–54). MIT, Cambridge, Mass.

 Tommi Leung Holes, C. (1990). Gulf Arabic. London: Routledge. Leung, T. (2014). Preposition stranding generalization and conditions on sluicing: evidence from Emirati Arabic. Linguistic Inquiry, 45. McCloskey, J. (1990). Resumptive pronouns, A’-binding and levels of representation in Irish. In R. Hendrick (Ed.), The syntax and the modern Celtic Languages (pp. 199–248). San Diego: Academic Press. Merchant, J. (1998). Pseudosluicing: elliptical clefts in English and Japanese. In A. Alexiadou et al. (Eds.), ZAS working papers in linguistics (volume 10) (pp. 88–112). Berlin: Zentrum fur Allgemeine Sprachwissenschaft. Merchant, J. (2001). The syntax of silence. Oxford: Oxford University Press. Mohammad, M. (2000). Word order, agreement and pronominalization in Standard and Palestinean Arabic. Amsterdam: John Benjamins. Rodrigues, C., Nevins, A. I., and Vicente, L. (2009). Cleaving the interactions between sluicing and P-stranding. In D. Torck, and W. L. Itzels (Eds.), Romance languages and linguistic theory 2006 (pp. 75–198). Amsterdam: John Benjamins. Ross, J. R. (1967). Constraints on variables in syntax. Doctoral dissertation, MIT, Cambridge, MA. Ross, J. R. (1969). Guess who? In R. Binnick, A. Davison, G. Green, and J. Morgan (Eds.), Proceedings of 5th Chicago Linguistic Society (CLS 5) (pp. 252–286). Chicago, IL. Sato, Y. (2011). P-stranding under sluicing and repair by ellipsis: why is Indonesian (not) special? Journal of East Asian Linguistics, 20, 339–382. Sells, P. (1984). Syntax and semantics of resumptive pronouns. Doctoral dissertation, University of Massachusetts, Amherst. Sharvit, Y. (1999). Resumptive pronouns in relative clauses.  Natural Language and Linguistic Theory, 17, 587–612. Shlonsky, U. (1992). Resumptive pronouns as a last resort. Linguistic Inquiry, 23, 443–468. Shlonsky, U. (2002). Constituent questions in Palestinian Arabic. In J. Ouhalla, and U. Shlonsky (Eds.), Themes in Arabic and Hebrew syntax (pp. 137–155). Dordrecht: Kluwer Academic Publishers. Takahashi, D. (1994). Sluicing in Japanese. Journal of East Asian Linguistics, 3, 263–300. Toosarvandani, M. (2008). Wh-movement and the syntax of sluicing. Journal of Linguistics, 44, 677–722. Wahba, W. (1984). Wh-constructions in Egyptian Arabic. Doctoral dissertation, University of Illinois at Urbana Champaign, Urbana. Watson, J. (1993). A syntax of Sanʕani Arabic. Wiesbaden: Otto Harrassowitz.