The book is a collection of articles on sluicing in a wide array of languages. Sluicing is the phenomenon illustrated in (1a), where the second conjunct involves a fronted wh-element and deletion of the material to the right of the wh-element. The sentence in (1b) is the non-elliptical counterpart to (1a):
- (1)
a. She saw someone yesterday, but I don't know who.
b. She saw someone yesterday, but I don't know who she saw.
The general focus of the papers is on the syntax of sluicing and sluicing-like structures in the different languages discussed. The emerging picture is one in which sluicing is not a uniform syntactic process across languages, but a semantic interpretation arising through different syntactic configurations (e.g. the paper by Frederick Hoyt & Alexandra Teodorescu, p. 85). What structure a particular language employs depends on other properties of the language, such as whether or not it exhibits overt wh-movement. All papers use English as the point of reference. This is perhaps not surprising given that the structure was first discussed in relation to English in Ross’ Reference Ross, Binnik, Davidson, Green and Morgan1969 paper ‘Guess who?’ (reprinted as Chapter 2 in this volume) and that sluicing has been more thoroughly investigated in English than in other languages.
The book comprises 11 chapters including an introductory chapter, the reprint of Ross (Reference Ross, Binnik, Davidson, Green and Morgan1969), and nine original papers. While it is unclear what the selecting criteria have been, the collection of papers represents an interesting array of languages, although there is a somewhat unmotivated focus on Japanese. The introductory chapter, written by the editors, Jason Merchant and Andrew Simpson, offers a brief general introduction to the topic of sluicing and the theoretical issues involved and, in addition, a detailed summary of each of the chapters in the book.
The inclusion of papers on non-Germanic languages is a very important contribution to the research field on sluicing. Unlike English and Germanic languages in general, Serbo-Croatian and Romanian, for instance, allow multiple wh-movement, and Japanese, Chinese, Malagasy, Bangla, Hindi and Turkish are wh-in-situ languages. These properties make a simple application of the sluicing analysis used for English non-trivial.
The book is of interest to anyone concerned with the empirical facts of sluicing: What do sluicing and sluicing-like constructions look like cross-linguistically, and what properties do these constructions have? The book also makes an important contribution to the theoretical discussion of two fundamental issues that theories of sluicing need to deal with, namely ellipsis and unbounded dependencies. The analyses proposed in the articles are all within the generative framework. An important point is made by Tanmoy Bhattacharya and Andrew Simpson, who observe that there is great variation in how speakers judge sluicing constructions involving islands in Hindi (p. 215). As pointed out by these authors, careful control of the data is necessary. Several of the papers look into island repair in sluicing, discussing when it takes place and fails to do so, as well as possible reasons for this. The more general question of the status of islands is not really discussed, however.
Chapter 2 is, as mentioned, a reprint of John Robert Ross's legendary Reference Ross, Binnik, Davidson, Green and Morgan1969 paper ‘Guess who?’, in which sluicing constructions were first discussed and analyzed. In his paper, Ross argues extensively that sluicing structures in English involve question formation – i.e. wh-movement to the front of the embedded clause – followed by deletion of the material following the wh-word. Evidence for this type of analysis comes, for instance, from case marking and number agreement.
In Chapter 3, ‘How do you sluice when there is more than one CP?’, Jeroen van Craenenbroeck argues for a split-CP structure, with wh-phrases appearing at different heights depending on their complexity. More precisely, simple wh-phrases are taken to be moved from an IP-internal position to a lower CP-layer, and further on to a higher CP-layer. Complex wh-phrases, in contrast, are argued to be base-generated in the highest CP-layer. In line with this, there are several logical syntactic configurations underlying sluicing structures. Van Craenenbroeck looks at dialectal data involving preposition pied-piping in sluices with inverse order between wh-word and preposition, so-called Swiping, as in (2a), as well as sluicing where the wh-word is followed by a demonstrative pronoun, so-called Spading, as in (2b):
(2)
The author shows that the otherwise unexplained difference in distribution of simple and complex wh-elements in these constructions falls out naturally from the split-CP analysis. A quick look at Swedish, however, raises some questions. Swedish too has a kind of Spading construction, but in this language both simple and complex wh-expressions are well formed, contrary to the expectation:
(3)
What accounts for the difference between Spading in Dutch and Swedish is a matter for further investigation. It thus remains to be seen if wh-elements have different syntactic distribution in these languages, or if Spading should be analyzed in different ways in Dutch and Swedish.
Chapter 4, ‘Two cases of violation repair under sluicing’, written by Sandra Stjepanović, discusses an interesting property of sluicing constructions. As has been known since Ross (Reference Ross, Binnik, Davidson, Green and Morgan1969), sentences with island violations tend to improve significantly in the presence of sluicing. Stjepanović discusses two cases where sluicing seems to repair structures in Serbo-Croatian. Serbo-Croatian does not permit preposition stranding. Yet in sluicing contexts, prepositions need not appear together with their complement wh-word but can be dropped. Although these data seem to indicate that the language allows preposition (P) stranding, it can be shown that this is not the case in non-sluicing contexts. Instead the author proposes that a mechanism of P drop applies at PF and that this mechanism is fed by sluicing. The other case of repair discussed by Stjepanović has to do with case. In Serbo-Croatian, certain numerals assign genitive case to their complements and these phrases are therefore disallowed as object of inherently case-marking verbs. This restriction, however, does not operate in sluicing constructions. Stjepanović analyzes this as an instance where an operation that is otherwise obligatory does not to take place under sluicing. The data patterns discussed are interesting and the repair effects obtained under sluicing raise a number of more general questions regarding the nature of syntactic constraints and repair mechanisms.
In Chapter 5, ‘How many kinds of sluicing, and why? Single and multiple sluicing in Romanian, English, and Japanese’, Frederick Hoyt & Alexandra Teodorescu look at sluicing in Romanian in the light of sluicing in English and Japanese. Romanian and Japanese, unlike English, both have multiple sluicing and allow for non–wh-remnants as well as overt complementizers in the sluice. These properties would suggest that Romanian and Japanese have the same type of structure in sluices, namely a cleft construction in which material has been deleted. As discussed by the authors, however, Romanian patterns with English rather than Japanese when it comes to island repair. The authors argue that sluicing in Romanian is like sluicing in English after all and that the similarities between the Romanian and Japanese constructions follow from independent properties of the languages. The authors also conclude that sluicing is not a uniform syntactic process across languages but that different ‘elliptical’ structures can give rise to sluicing readings.
The discussion of sluicing in Japanese continues in Chapters 6 and 7. In Chapter 6, ‘Case morphology and island repair’, Masanori Nakamura discusses differences between English and Japanese regarding island repair in sluicing contexts. In English islands are generally repaired under sluicing, while that is not the case in Japanese. In the latter, island repair is restricted to contexts where a PP has been extracted, and thus not cases where an NP has been extracted. Nakamura argues that this finds an explanation in the way case is assigned. According to the author, case is assigned in a Morphological Component which is fed by PF. While PPs assign case phrase-internally, NPs get case from outside. Crucially, the moved NP is dependent on its base position to get case, and if the base position is deleted, no case can be assigned. Sluicing thus can't repair islands involving NPs, since in those contexts the case transmission from base position to moved position cannot take place. Crucially, the argumentation hinges on the claim that case assignment in English is a syntactic phenomenon while it is a post-syntactic phenomenon in Japanese.
Chapter 7, ‘Island-sensitivity in Japanese sluicing and some implications’, by Teruhiko Fukaya, is also concerned with island-sensitivity and case marking in Japanese sluicing structures. Fukaya notes that sluices without islands in certain cases have two readings differing in quantificational scope but that the corresponding sluices with islands and case-marked remnants only have one of the readings. These represent two types of sluicing structures in Japanese. The author notes that one type of sluicing in English has properties reminiscent of the Japanese structure and that English too, therefore, has two types of sluicing, differing as to whether they exhibit island repair.
The English-type analysis of sluicing involves wh-movement to Spec,CP followed by deletion of the material below it. The expectation is therefore that languages that do not exhibit wh-movement should not have sluicing, or should have a different type of syntax for sluicing structures. This is indeed the case for Japanese, as discussed in Chapters 5–7 in the book. In Chapter 8, ‘Sluicing without wh-movement in Malagasy’, Ileana Paul & Eric Potsdam discuss the syntax of sluicing in Malagasy. Malagasy is a VOS language with predicate fronting. The authors argue convincingly that sluicing in this language involves predicate fronting of the wh-phrase (rather than wh-movement), followed by deletion. Like Hoyt & Teodorescu, the authors conclude that sluicing is not a uniform syntactic process but that sluicing-like structures can arise through different syntactic processes.
In Chapter 9, ‘Sluicing in Indo-Aryan: An investigation of Bangla and Hindi’, Tanmoy Bhattacharya & Andrew Simpson argue that Bangla and Hindi have wh-movement, although it has been argued that these languages are wh-in-situ languages. Sluicing in these languages is of the English type, involving wh-movement and deletion. Although the languages are related and share a number of properteis, Bangla and Hindi are subject to different restrictions on sluicing. For instance, the languages differ as to whether they allow overt complementizers in sluices; Hindi requiring them, and Bangla disfavoring them. The authors propose that this difference reflects a difference in landing-site of the moved elements (see also van Craenenbroeck's article). Furthermore, Bangla has island repair in sluicing. For Hindi data involving islands, on the other hand, judgments differ widely. The authors discuss both possible explanations for the variation between the languages with regard to island repair and the speaker variation observed for the Hindi data. As they point out, the latter reveals an important methodological issue calling for careful control of the data.
Chapter 10, ‘Sluicing in Mandarin Chinese: An instance of pseudo-sluicing’, by Perng Wang Adams & Satoshi Tomioka, discusses sluicing in yet another wh-in-situ language. The authors argue that sluicing in Chinese is substantially different from sluicing in English and certain other languages in that it does not involve deletion. More precisely, Chinese exhibits pseudo-sluicing: a phonologically silent pronominal subject, a copula and the wh-word. The authors show that the properties of Chinese sluicing structures fall out naturally from a pseudo-sluicing analysis but not from analyses involving, for instance, focus-movement and deletion, or a reduced cleft structure. They also discuss two apparent challenges to the pseudo-sluicing approach: sloppy readings and multiple sluicing. Sloppy readings are available with adjunct wh-phrases (although not with argument wh-phrases) and such readings are often taken as evidence against a pseudo-sluicing analysis because pronouns should not be able to yield sloppy readings. The authors show however that this claim is incorrect. Instead they explain the observed pattern (i.e. that sloppy readings are more easily obtained with adjunct wh-remnants than with argument wh-remnants) as a difference in semantic type between empty pronouns that refer to individuals (type e) and to events or propositions. The latter would more easily yield sloppy readings. Regarding multiple sluicing, the authors argue that that too can be explained on the pseudo-sluicing analysis. They show that these can be analyzed as instances of conjoined pseudo-sluiced sentences.
The last chapter of the book, Chapter 11, ‘Sluicing in Turkish’, by Atakan Ince, argues in favor of a movement+deletion analysis of Turkish sluicing structures. Although Turkish is a wh-in-situ language, the author shows that movement analysis is empirically motivated, while ellipsis or cleft analyses are not. While Turkish does not have wh-movement in non-sluicing contexts, there are indications that there is wh-movement in sluicing. Case-matching is one such indication (among others discussed): the wh-phrase has the same case as its antecedent. As in English, this can be explained if the wh-phrase is assigned case in its base position and moves to the clause-peripheral position after that. Since Turkish does not have overt wh-movement, some feature other than a wh-feature must be responsible for movement of the wh-phrase. The author proposes that the relevant feature is a focus feature. The wh-phrase carries focus intonation. The contrast with non-sluicing contexts, where there is no wh-movement, would then be that no focus features are present there.
The articles in the book are well written: the theoretical questions are stated clearly and in most cases the analyses argued for are well supported by data. There only editing problems observed are on pages 202–203, where there is an inconsistency in the use of parentheses and */? to mark ungrammaticality in examples (45)–(46) and (48). In the same chapter, there seems to be some confusion in the glossing of ‘ke’ and ‘ki’ (‘what’/‘who’) on pages 207–210 (main text and examples).
