This book brings together a wide range of issues bearing on sluicing from a cross-linguistic perspective.Footnote ¹ The languages discussed in this work include those with overt wh-movement, such as English (Chapters 2 and 5), Dutch (Chapter 3), Serbo-Croatian (Chapter 4), and Romanian (Chapter 5), and those without, such as Japanese (Chapters 5–7), Malagasy (Chapter 8), Bangla, Hindi (Chapter 9), Chinese (Chapter 10), and Turkish (Chapter 11). The analyses proposed can be divided into two groups. The first group of analyses crucially depends on movement of the wh-remnant, the focus element, or the predicate, which is followed by deletion. The second group of analyses emphasizes the lack of overt wh-movement in some languages and entertains an approach which treats sluicing as clefts or similar constructions. However, even in this group of analyses, deletion is often invoked. In a sense, therefore, most of the analyses in this book employ ellipsis to derive sluicing structures.

An important generalization one obtains from the studies contained in this book is that the wh-construction of a language does not correlate directly with the derivation of sluicing. Instead, as Merchant (Reference Merchant2001) and Frederick Hoyt & Alexandra Teodorescu (Chapter 5) point out, sluicing typically follows from the syntactic mechanism independently available in the language. A corollary that follows is that similarity in surface patterns of sluicing sentences could only be apparent. This makes an insightful typology of sluicing constructions particularly difficult.

The diversity and non-uniformity of sluicing, within and across languages, can be seen in the following examples. The wh-in-situ parameter could lead one to the presupposition that movement plays no role in the derivation of sluices in languages without overt wh-movement. Indeed, Hoyt & Teodorescu (Chapter 5) argue that Japanese sluicing is derived from a reduced cleft structure involving focus movement. Without recourse to the cleft structure, Masanori Nakamura (Chapter 6) shows that the wh-remnant in Japanese sluices is derived by overt focus movement. Teruhiko Fukaya (Chapter 7) claims that Japanese sluicing comes in two types in terms of case marking: case-marked wh-remnants exhibit movement properties, while non-case-marked wh-remnants are derived from a kind of copula structure (pseudo-sluicing) involving no movement.

In fact, a variety of approaches is proposed to derive sluices in wh-in-situ languages dissociated from the wh-in-situ parameter. Ileana Paul & Eric Potsdam (Chapter 8) propose that sluicing in Malagasy is generated by fronting of the wh-predicate, followed by deletion of the TP. Likewise, Tanmoy Bhattacharya & Andrew Simpson (Chapter 9) argue that sluices in Bangla and Hindi are generated by disguised overt wh-movement and deletion. Atakan Ince (Chapter 11) argues that Turkish sluicing involves movement of the wh-remnant, motivated by the focus feature rather than the wh-feature. However, languages with overt wh-movement, i.e. Dutch, Serbo-Croatian, Romanian, and English (Chapters 3–5), are all analyzed as employing overt wh-movement to derive the sluices. Thus, the studies contained in this book clearly show that the wh-in-situ parameter does not play a decisive role in the derivation of sluicing. Some wh-in-situ languages still adopt (disguised) overt wh-movement and deletion to derive sluicing. Only Teruhiko Fukaya's analysis of Japanese sluicing with non-case-marked wh-remnants (Chapter 7) and Perng Wang Adams and Satoshi Tomioka's analysis of Chinese sluicing (Chapter 10) – both of which involve an implicit pro subject, a copula, and the base-generated wh-remnant without ellipsis throughout – are directly associated with the wh-in-situ parameter of the languages. All this also points to the somewhat unfortunate fact that cross-linguistic studies on sluicing do not seem to shed much light on the possible effects of any micro- or macro-parameters of grammar in the formation of this particular construction.

An important issue concerning sluicing is deletion or ellipsis. A number of proposals have been given in the literature: deletion is said to apply in narrow syntax (Baltin Reference Baltin2012), in PF (Merchant Reference Merchant2001, Fox & Lasnik Reference Fox and Lasnik2003), or in the post-PF component (Sandra Stjepanović in Chapter 4). In this book, deletion is discussed along with considerations associated with preposition stranding (or P-stranding), Case, and other questions.

In Serbo-Croatian (Chapter 4, see also Stjepanović Reference Stjepanović2008), P-stranding is not allowed in regular wh-questions. However, sluicing in the language optionally permits preposition omission. This violates Merchant's (Reference Merchant2001: 92) P-stranding generalization, which states that a language allows P-stranding under sluicing iff it also allows P-stranding under wh-movement. In her chapter, Stjepanović argues that the loss of P is not due to P-stranding; instead, the preposition is dropped in a post-syntactic component. Thus, preposition omission in sluicing in Serbo-Croatian is not a violation of the P-stranding generalization. It results from ‘phonetic loss’ in the post-PF component, not TP deletion in PF.

Nakamura (Chapter 6) shows that in Japanese, island violations arising from focus movement of case-marked NPs cannot be repaired, whereas those arising from movement of argument PPs can. Nakamura argues that this has to do with the difference of Case assignment between NP and PP: NP needs Case, but PP does not. Nakamura assumes that deletion repairs island violations. As a result, the fact that island-violating focus movement of case-marked NPs cannot be repaired has nothing to do with the repair effect of deletion in PF; it arises from the blocking effect of islands on Case transfer after Spell-Out in Morphology under the Cyclic Spell-Out Model. This makes the problem of the repairing effects in sluicing more complicated than one thinks. (We will come back to the repair effects later.)

Atakan Ince (Chapter 11) proposes that deletion resolves the nominative–genitive Case mismatch in Turkish sluicing. In Turkish, the subject wh-remnant bears the nominative Case, but its correlate in the embedded clause takes the genitive Case. To account for this mismatch, Ince first assumes that Turkish is an overt V-raising language. In the process of V-raising under the Multiple Spell-Out Model, the T-v-V amalgam checks the nominative Case of the subject in SpecTP prior to TP ellipsis. However, it cannot further agree with C and check the Case of the wh-remnant, which then gets the genitive Case in SpecCP.

The above remarks reveal that the specific timing of deletion is crucial for sluicing in some proposals: Stjepanović considers P-drop in Serbo-Croatian sluicing as an operation in the post-PF component; Nakamura assumes that in Japanese sluicing, deletion applies in PF; and Ince proposes that in Turkish sluicing, deletion follows each step of cyclic Spell-Out and therefore is syntactic (and this makes his theory fairly similar to that of Baltin Reference Baltin2012). There seems to be no agreement on the question at which specific grammatical level deletion applies specifically for the purpose of deriving sluicing. This makes a unified theory for sluicing even more difficult and challenging.

Repair effects in sluicing also diverge from language to language. One often comes across discussion on how deletion saves a structure; however, sometimes the repair is not due to deletion, but to factors other than deletion. Sometimes what appears to be a repair effect superficially turns out not to be a repair effect at all.

Merchant (Reference Merchant2010) classifies repair effects of sluicing into three types: the bleeding type, the feeding type, and the miscellanea. Repair effects need not involve syntactic islands. Stjepanović shows that the P-drop in Serbo-Croatian sluicing, a post-PF phenomenon, is a feeding operation. Besides, the violation of the Inverse Inherent Case Filter (IICF), a constraint requiring that inherent Case be morphologically realized, is also repairable in sluicing by deleting the undischarged morphological feature. Such repair is a bleeding operation. Ince argues that phase-by-phase deletion can salvage the non-parallel genitive–nominative case mismatch in Turkish sluicing, which is a feeding type of repair effect. None of these involves a syntactic island.

Furthermore, island repair effects vary with different types of sluices. Fukaya argues that sluicing with a case-marked wh-remnant in Japanese appears to involve movement and respect island effects, whereas sluicing with a non-case-marked wh-remnant does not involve movement and shows no island-sensitivity. This contrast indicates that deletion is not always directly associated with island violations and their repairs; some other factors may be at play. The nature of the remnant matters because a specific type of remnant may indicate a special type of construction. In point of fact, according to Fukaya's analysis, the Japanese case in which there is an apparent island violation does not really involve a syntactic dependency between the wh-remnant and a trace. In general, though, deletion can ameliorate a locality violation caused by wh-movement or focus-movement. More examples can be found in Bangla (Bhattacharya & Simpson's chapter), a wh-in-situ language, where sluicing exhibits island repair effects with wh-movement and ellipsis.

A priori, sluicing without wh-movement will not trigger any island violation. Thus island repair in pseudo-sluicing is only apparent. Adams & Tomioka propose that in Chinese, a wh-in-situ language, no movement or deletion is involved in sluicing, so nothing needs to be repaired. The superficial repair effect can be attributed to the construal of the subject pro in the sluice clause, akin to Fukaya's account of island-insensitivity of Japanese sluicing with non-case-marked wh-remnants (see above). This is why ‘redemptive effects’ seem so easy to obtain in the language (see also Wei Reference Wei2004, Reference Wei2011a).

In fact, what we call the ‘repair effects’ in sluicing come as a heterogeneous set. The effects can be achieved by genuine deletion, or simply by the construal of a pro subject. Some relevant phenomena, in particular restrictions on possible and/or impossible repairs, are not well studied and remain vague. For example, questions about the types of island that can be repaired have not received due discussion cross-linguistically (see e.g. Merchant Reference Merchant2001 and his other works on the topic). Likewise, questions about what kind of non-island phenomena could be repaired by ellipsis have not received much attention either. There appears to be further work to be done on the theoretical basis of sluicing, such as permissible and impermissible sluicing phenomena.

A short remark on the role of the copula is in order. As first observed by Kizu (Reference Kizu, Singer, Eggert and Anderson1997), the copula can appear in the sluicing constructions of wh-in-situ languages, such as Japanese, Chinese, Turkish, Hindi, and Bangla. One might think that the appearance of the copula implies that the sluice results either from a reduced cleft (see Kizu Reference Kizu, Singer, Eggert and Anderson1997 and Hoyt & Teodorescu's chapter) or from the identificational, specificational or emphatic type of pseudo-sluicing (see Wei Reference Wei2004 and Adams & Tomioka's chapter). However, this claim is not always true, for two reasons. First, sluicing in Japanese, Turkish, Bangla, and Hindi has been argued to derive from overt movement and deletion rather than from cleft-based constructions and pseudo-sluicing structures. What is more, there are wh-in-situ languages that lack an overt copula in sluicing, for instance some Austronesian languages such as Malagasy (Paul & Potsdam's chapter) and Amis (Wei Reference Wei2011b). Thus, the presence or absence of the copula in sluicing, just like the wh-in-situ parameter, has no direct bearing on the derivation of sluicing. In their chapter, Bhattacharya & Simpson hold a similar view.

In conclusion, many intriguing questions are raised and analyzed in this book on sluicing, but the diversity of the grammatical mechanisms involved and the richness of the relevant issues reveal that it is premature to claim any typological classification of sluicing with syntactic significance. Also, a deeper understanding of the relevant theoretical questions is still wanting. This partially arises from a fact mentioned in the beginning of this review and recognized by the editors of the volume: that the derivation of sluicing typically follows independent language-specific syntactic properties but such properties show great cross-linguistic differences.