Clausal Ellipsis in Jibbali: The Case of Sluicing

This study investigates the syntax of sluicing in Jibbali from a generative perspective to identify its morphological and syntactic properties. It also seeks to provide an explanation for the preposition stranding sluices that seem to violate of the Preposition Stranding Generalization (PSG) posited by Merchant (2001). The study concludes that sluicing exists in Jibbali and that it results from an overt wh-movement operation plus IP ellipsis at PF. Furthermore, it is argued that Jibbali sluicing allows for two sources of clausal ellipsis, referred to herein as sluicing and pseudosluicing. Both sluicing and pseudosluicing are derived by wh-movement and TP deletion despite the fact that sluicing stems from regular wh-questions, whereas pseudosluicing derives from a copular underlying source, i.e., cleft wh-questions. With regards to preposition-less (P-less) sluices, it is argued that such sluices are instances of pseudosluicing and that they do not involve preposition stranding as wh-movement in pseudosluicing proceeds from Spec TP. The proposed analysis for both forms of clausal ellipsis rests upon two arguments: a) the analysis of pronouns in the elided clause as copular pronouns, and b) the restriction on clefting non-nominal expressions in Jibbali which mirrors the parallelism between P-less sluices and cleft wh-questions in the language.

pronounced is only the displaced wh-expression. Jibbali exhibits structures like (1) above, and they are interpreted by native speakers of the language as complete wh-questions. Example (2) is a sluicing structure from Jibbali.
'Salim bought a book, but I don't know when.' The current study attempts to provide the first investigation into the syntax of sluicing in Jibbali. It aims to identify the morpho-syntactic properties of Jibbali sluicing (JS) and propose an analysis for this elliptical structure. The data used in the study was collected from native speakers of Jibbali in the governorate of Dhofar in the south of Oman. The informants are university students majoring in English Language, Translation Studies and engineering. The data obtained from informants was grammatically judged and elicited. All sentences used in the study were transcribed, glossed and translated into English.
The study consists of six sections. The first section is an introduction to the study. It provides background information on Jibbali, the kind of clausal ellipsis dealt with in this study, the aims, methodology and organization of study. Section 2 discusses sluicing in syntactic theory and reviews previous studies on sluicing with a special reference to Semitic languages. Section 3 is concerned with wh-question formation in Jibbali. Section 4 investigates sluicing in Jibbali and attempts to determine its properties and underlying source with a special reference to the interaction between sluicing and preposition stranding. Section 5 puts forward an analysis for Jibbali sluicing constructions. Finally, section 6 concludes the study.

Previous Studies on Sluicing: A Cross-Linguistic Perspective
The term 'sluicing' was first coined by Ross (1969) and has been revived in Merchant's (2001) seminal work on sluicing. Merchant's (2001) work on sluicing provided the first comprehensive investigation of sluicing arguing that the sluiced clause contains an invisible syntactic structure of a full wh-question. Based on cross-linguistic morpho-syntactic properties of sluicing such as morphological case matching and preposition stranding, Merchant (2001) concludes that sluicing results from a wh-movement operation plus IP deletion at PF. In addition, Merchant (2001) opposes assimilating sluicing to reduced wh-clefts as both structures are distinct.
To start with, Shimoyama (1995) and Kuwabara (1996) studied sluicing in Japanese and concluded that sluicing is a form of an elliptical wh-cleft. Similarly, Mandarin Chinese has been analysed as pseudosluicing in which the elided material is considered a phonologically empty category (Adams & Tomioka, 2012). The proposed analysis for Mandarin sluicing captures some language-specific properties such as the requirement of overt correlates to refer to argument wh-phrases, the use of copulas and finally island amelioration under ellipsis.
Regarding research on sluicing in Semitic languages, sluicing has been investigated in some modern varieties of Arabic including Libyan Arabic (Algryani, 2012), Emirati Arabic (Leung, 2014a) and Saudi Arabic (Alshaalan & Abels, 2020). Algryani's (2012) study of sluicing revealed that sluicing is manifested in the Libyan variety of Arabic and that it stems from wh-movement plus IP ellipsis at PF. More importantly, it is argued that sluicing in the language does not violate the Preposition Stranding Generalization (PSG) as prepositionless (P-less) sluices derive from cleft interrogative clauses that do not involve wh-movement from a clause-internal position. The implication of this sluicing property, which has also been attested in other languages, provides evidence that there are two distinct forms of elliptical wh-interrogative clauses and that both are derived via movement and deletion. Leung (2014a) investigated Emirati Arabic sluicing constructions and found out that they result from wh-movement plus clausal deletion. The study further points out that the underlying source and typology of sluicing are predetermined by the modes of wh-interrogatives available in the language. In another work, Leung (2014b) argues that sluicing in Emirati Arabic violates Merchant's (2001) PSG in the sense that while preposition stranding (p-stranding) is strictly disallowed in the formation of wh-interrogatives, it is allowed in sluicing. In his account for P-less sluices in Emirati Arabic, Leung (2014b) argues that the p-stranding effects can be repaired at PF, i.e., repaired by ellipsis.
Alshaalan and Abels (2020) studied Saudi Arabic sluicing with a special focus on p-stranding sluices. They concluded that the ellipsis site in sluicing contains a syntactic structure and that sluicing is derived by wh-movement and clausal deletion. However, with respect to P-less sluices, they argue that they can stem from a cleft source or a resumptive wh-question. Such as a conclusion is problematic for Merchant's (2001) PSG as P-less sluices in the language can derive from a non-copular source, which constitutes a violation to the PSG.
Given that sluicing is a form of elliptical wh-questions, the next section discusses the formation of wh-questions in Jibbali as this will provide more insights into understanding the syntax of wh-movement and sluicing constructions in the language.
In regular wh-interrogatives, the wh-word surfaces clause-initially and is marked by a gap bound by the fronted wh-expression, as illustrated in (3) and (4). Regular wh-questions, which involve fronting the wh-phrase, can be formed by both argument wh-expressions and wh-adjuncts.
( It has been proposed in previous studies on wh-movement in Semitic languages (e.g., Bakir, 2017;Aoun et al., 2010;Shlonsky, 2002 among others) that wh-fronting is a result of an overt wh-movement process to a clause-initial position. Such an argument is based on a set of diagnostic tests referred to as syntactic islands, i.e., syntactic configurations out of which extracting a wh-phrase is not permissible. For instance, examples (5) and (6) are ungrammatical because they involve extracting the wh-phrase mun 'who' from a relative clause island and an adjunct clause island respectively. '*Who did you ask Ahmad if he saw?' (Bakir, 2017, p. 52) The formation of cleft wh-questions involves a clause-initial wh-phrase and a resumptive pronoun occupying the gap considered to be the position of the moved wh-phrase. The following are examples of cleft wh-questions.
'What is the thing that Suhail bought?' what thing REL. bought.3MS-it Saed 'What is it that Saed bought?' There are some morpho-syntactic features that characterize cleft wh-questions. First, as the case in Semitic languages, e.g., Jibbali (Bakir, 2017), Libyan Arabic (Algryani, 2012), Egyptian Arabic Wahba (1984), Palestinian Arabic (Shlonsky, 2002) among others, these questions are marked by a clitic resumptive pronoun in the position assumed to be the place where the wh-phrase was first generated. Second, cleft wh-questions are only compatible with nominal wh-expressions; PP wh-phrases and wh-adjunct expressions are incompatible with such wh-questions, as in shown (10) and (11) Third, unlike regular wh-questions, cleft wh-questions are insensitive to island constraints, as evidenced in (12) and (13). The grammaticality of (12) and (13) suggests that this class of wh-questions does not involve wh-movement from a clause-internal position.
'*What is the thing that Mahad saw the man who bought it?' (Bakir, 2017, p. 55) '*Who is the person that S. went before seeing?' (Bakir, 2017, p. 55) Finally, cleft wh-questions differ from regular wh-questions in that they permit pronouns to appear between the wh-word and the relativizer, as shown (14) ( 'What is the thing that Suhail bought?' The last type of wh-questions is in-situ wh-questions in which the wh-word remains in the position it was first generated in. Such wh-questions are used in Jibbali and are normally perceived as echo wh-questions. In the second type, the remnant refers to a null argument licensed by the argument structure of the verb, as in (21), whereas in the third, the wh-phrase of the sluice is an adjunct that does not refer to any expression in the antecedent clause, as in (22) 'Saed bought car, but I don't know when'.
Finally, contrastive sluices involve a contrast in the content of the sluiced wh-phrase and that of its correlate in the previous discourse, as shown in (23).
Saed bought.3MS three books but NEG-know.1S how many magazine NEG 'Saed bought three books, but I don't know how many magazines'.
It is worth noting that Jiballi sluicing clauses share some properties attested in other languages. First, it is only licensed by interrogative wh-phrases, as shown in examples (24) and (25). Second, Jiballi sluicing allows the presence of a variety of non-wh-remnants such as pronouns and the negative particle 'lo'. This is also showed in (24) and (25) respectively. In (24), the sluice contains a non-wh-remnant which is the negative particle 'lo' (Note 2), and the sluice in (25) contains an optional pronoun in addition to the negative particle 'lo'. Third, sluicing in Jibbali is island-insensitive in the sense that it allows remnants to be extracted out of island domains inside elided material (Note 3). Finally, despite the fact that Jibbali is a non-p-standing language under regular wh-movement, as illustrated in the contrast in (26) and (27), the language seems to permit preposition stranding under sluicing, as evidenced in (28). The issue of sluicing in p-stranding contexts will be discussed in detail in section 4.2. 'Mahad talked with someone, but I don't know who'.

Jibbali Sluicing: Sluicing or Pseudosluicing?
Since Jibbali possesses two main types of wh-questions in addition to in-situ wh-questions, it seems plausible to assume that Jibbali sluicing derives from either regular or cleft wh-questions. In section, I attempt to determine whether the underlying source of Jibbali sluicing (JS) is regular or cleft wh-questions. It is worth noting that sluicing in some languages has been analysed as pseudosluicing (Note 4) despite its superficial appearance as sluicing. For instance, according to Shimoyama (1995) and Kuwabara (1996), Japanese sluicing is a form of an elliptical wh-cleft. Likewise, Adams and Tomioka (2012) analyse Mandarin Chinese as pseudo-sluicing (Note 5). Another alternative analysis of sluicing put forward by Ross (1969) and developed recently by Merchant (2001), argues that sluicing derives from regular wh-questions by wh-movement and clausal deletion at PF, as illustrated in (31). Such an analysis derives its strength from cross-linguistic morpho-syntactic properties of sluicing such as morphological case-matching, p-stranding and binding effects that sluicing constructions share with their non-elliptical counterparts. Such cross-linguistic observations provide evidence for assuming a structure in the ellipsis site of sluicing and thus deriving sluicing by movement and ellipsis (Note 6).
(31) Someone called Jane, but no one knows who i called Jane t i .
Given that there are two types of wh-questions that involve overt wh-movement, Jibbali sluices are assumed to have two underlying sources. In this section, I test Jibbali sluicing in different contexts in order to determine whether what appears as sluicing can be analysed as sluicing or pseudo-sluicing, and the contexts in which both structures occur. To achieve this, I adopt some of Merchant's (2001) diagnostics which can determine the behaviour of the sluiced wh-phrase in sluicing and clefting contexts.

Diagnostic 1: Sluicing with Adjunct Wh-Phrases
Merchant (2000) observes that sluicing is different from pseudo-sluicing with respect to adjunct wh-phrases. While the former can be formed with adjunct wh-words, the latter is unattainable. This seems to be the case in Jibbali as sluicing with adjunct wh-phrases is acceptable, as shown in (32); however, cleft wh-questions with adjunct wh-phrases whether elliptical or non-elliptical are not grammatical, as evidenced in (33a)  The data in (32) and (33) indicate that sluiced clauses with adjunct wh-remnants are genuine sluices, i.e. they derive from regular wh-questions. The pseudosluicing source is not available as illustrated in (33) as wh-phrases of adverbial functions (i.e. adjuncts) are not compatible with clefts wh-questions.

Diagnostic 2: Sluicing with Implicit Arguments
Implicit arguments, as pointed out by Merchant (2001), can differentiate between sluicing and reduced wh-clefts in English. While sluicing is perfectly fine with implicit arguments, reduced wh-clefts are ungrammatical. This seems not to be the case in Jibbali as both sluicing and elliptical wh-clefts are compatible with dropped arguments, as illustrated in (34)  The conclusion that can be drawn from the above data is that there are two underlying structures, i.e. sources of derivation, for sluicing in implicit argument contexts. The possibility of spelling out the copular pronoun in the sluicing clause allows for an underlying copular source.

Diagnostic 3: Language Clefting Strategy
This test is used to determine whether there are restrictions on clefting certain grammatical categories in a language as this can help in distinguishing between sluicing and pseudosluicing. In this respect, it seems that Jibbali has restrictions on clefting non-nominal constituents, as evidenced in (36) below, which involves clefting a prepositional wh-constituent, and (37), which involves the use of a copular pronoun in a sluiced clause with a PP wh-remnant. Mahad talked.3MS with someone but NEG-know.1S with whom he NEG 'Mahad talked with someone, but I don't know with whom it was'. (Intended) The contrast in (36) and (37) confirms that elliptical wh-questions with PP-remnants can only be cases of sluicing. The pseudosluicing scenario is excluded due to the restriction on clefting PP constituents in the language.

Jibbali Sluicing in P-Stranding Contexts
In his extensive cross-linguistic study of sluicing, Merchant (2001) observes that stranding prepositions is only permissible if it is allowed under wh-movement. Based on this observation, Merchant (2001) put forward the Preposition Stranding Generalization (PSG) stated below.
A language L will allow preposition stranding under sluicing iff L allows preposition stranding under regular wh-movement. (p. 92) The PSG argues for deriving sluicing by wh-movement and clausal ellipsis at PF as such as analysis accounts straightforwardly for the parallelism between a sluice and its non-elliptical counterpart. For instance, English allows p-stranding under wh-movement, so it is possible to strand prepositions under sluicing. Thus, the P-less sluice in (38) is grammatical because it is assumed to derive from an available p-stranding wh-question, as in (39b).
(38) Mary went with someone, but I don't know who Mary went with.
(39) a. With whom did Mary go?
b. Who did Mary go with?
As mentioned earlier in section 4, it has been observed that Jibbali sluicing exhibits p-stranding effects although Jibbali is a non-p-stranding language under wh-movement, as illustrated in (40) and (41)  Recent research has shown cases of P-less sluices in non-p-stranding languages such as Portuguese (Almeida & Yoshida, 2007), Spanish (Rodrigues et al., 2009) and Arabic (Algryani, 2012) among others. These studies concluded that regardless of the apparent p-stranding effects displayed by sluicing in these languages, such cases ijel.ccsenet.org International Journal of English Linguistics Vol. 10, No. 6; of P-less sluices do not derive from regular wh-questions but rather from a copular underlying structure. This confirms that sluicing in such languages is not a counterexample to the PSG. Furthermore, the findings of these studies provide evidence for the availability of an alternative source for elliptical wh-questions (Note 7). On the other hand, other studies on p-stranding sluices such as those of Sato (2011), Wang (2007) and Stjepanović (2008) for Indonesian, Mandarin Chinese and Serbo-Croatian respectively argue that these languages constitute counterexamples to the PSG as sluicing permits p-stranding. Given that a cleft source is not available as an underlying source for P-less sluices in these languages, Wang (2007) argues that P-less sluices resort to resumption, whereas Stjepanović (2008) argues that Serbo-Croatian P-less sluices involve preposition drop at PF. Accounting for the P-less sluices in Indonesian, Sato (2011) argues while some violations can be repaired at the phonology-syntax interface, violations that occur in the syntactic computation cannot be repaired.

Analysis of Jibbali sluicing
Building on Bakir's (2017) analysis of wh-questions in Jibbali, I assume that the wh-phrase in both regular and cleft wh-questions undergo wh-movement from a clausal internal position to the CP domain. Following this line of analysis, sluicing stems from regular wh-questions, and it is derived by movement of the wh-phrase to the CP layer followed by TP deletion at PF, as schematized in (42)  As for JS under p-stranding, e.g., (43), the question is whether what appears p-stranding under sluicing is superficial or genuine p-stranding as this has significant implications on the analysis of sluicing in the language. There are two possible analyses of sluicing in p-stranding contexts in non-p-stranding languages. The first is that the p-stranding effects are apparent and what appears as sluicing is pseudo-sluicing, i.e. an elliptical wh-cleft and this form of ellipsis is distinct from sluicing despite the fact that it can be derived via movement and deletion (Vicente, 2008;Rodrigues et al., 2009;Algryani, 2012). The second possible analysis implies that sluicing permits p-stranding and thus violates Merchant's (2001) PSG (Stjepanović, 2008Sato, 2011;Leung, 2014;Alshaalan & Abels, 2020) In this section, I argue that Jibbali sluices with apparent p-stranding effects are instances of pseudo-sluicing and therefore the language cannot be considered a counter example against the PSG. This argument is built on some language-specific morpho-syntactic properties displayed by sluicing in the language.
Salim talked.3MS with someone but NEG-know.1S who he NEG 'Salim talked with someone, but I don't know who'.
First, JS under p-stranding permits the spell-out of pronouns, as shown in (43). This is also attested in cleft wh-questions, which are analysed as copular clauses, i.e. cleft wh-questions, as exemplified in (44). The analysis of such wh-questions as cleft wh-questions rests upon some evidence that shows close resemblance between the two structures. Such wh-questions, for instance, contain a relative clause, a pronominal clitic and an optional copular pronoun, as in (44). I consider the admissibility of the presence of copular pronouns in sluicing as initial evidence for the proposed pseudosluicing analysis for P-less sluices. It is worth noting that the pronouns that surface in the sluice are used to function as copulas. Jibbali, in this respect, patterns with other Semitic languages, e.g. Hebrew (Doron, 1983(Doron, , 1986 and Arabic (Eid, 1991;Ouhalla, 1999;Shlonsky, 2002), in the use of pronouns to realize copula functions. For instance, Jibbali uses pronouns to achieve copula functions in equative clauses and cleft structures, as in (45) and (46) (47a), which is mirrored in sluicing in p-stranding contexts, it is argued that P-less sluices, as in (43)  Thus, for p-stranding sluices, it is argued that such cases should be analysed as pseudo-sluicing, an elliptical cleft wh-question consisting of a DP, a copula and a headless relative clause functioning as a predicate.
Pseudosluicing results from movement of the wh-phrase from Spec TP to Spec CP followed by clausal ellipsis, as shown in (49) Given that wh-movement in pseudo-sluicing proceeds from Spec TP position, it is predicted, which is the case, that Jibbali sluicing constructions do not exhibit p-stranding effects. The deletion of the preposition with the relative clause gives rise to apparent p-stranding effects.

Conclusion
The paper has discussed the ellipsis phenomenon of sluicing in Jibbali to identify its morpho-syntactic properties.
The study concludes that sluicing in Jibbali is derived from regular wh-questions via wh-movement and clausal deletion at PF. With respect to p-stranding sluices, these are analysed as pseudo-sluices derived from copular clauses by wh-movement and deletion. As wh-movement in pseudosluicing proceeds from a clause-initial position, i.e., Spec TP, it follows that Jibbali sluicing is not a counterexample to the PSG as wh-movement does not involve p-stranding. The apparent p-stranding effects arise as a result of the deletion of the preposition with the headless relative clause leaving behind a remnant wh-phrase and an optional copular pronoun. Finally, the accounts put forward for sluicing and pseudo-sluicing explain some properties of Jibbali sluicing such as p-stranding effects, the presence of pronouns in the sluicing clause and the unavailability of pseudosluices with prepositional and adjunct wh-phrases in the language.