Sophie Repp's new book, Negation in gapping, has an overly narrow title, priming the reader to expect only a narrow look at gapping and its interaction with negation, a relevant and challenging phenomenon. However, in addition to a lucidly argued novel analysis of gapping, the work also contains thoughtful, in-depth treatments of the syntax and semantics of polarity, denials, and negation.

Gapping, widely addressed in the ellipsis literature, persistently resists a unified analysis with other, better-understood types of ellipsis, such as sluicing. Gapping, which occurs only in coordinated matrix clauses, targets, under identity with an antecedent in the first conjunct (C1), the finite verb/auxiliary and, optionally, additional material (together, the gap) in the second conjunct (C2). C2 contains at least two remnants, each consisting of phonologically represented material that contrasts with a correlate in C1. The gap's complement can be a remnant, which initially suggests that gapping can – atypically for ellipsis – target a non-constituent. Many treatments of gapping, including Repp's, argue that this non-constituency is only apparent, as is the ‘ellipsis’ itself (if narrowly defined as ‘deletion at P(honological)F(orm)’).

A further oddity of gapping – and the main empirical puzzle Repp investigates – involves its interaction with negation, which yields three different scopal readings, and which Repp argues implicates a non-unified analysis of gapping.

1. (1)

   1. (a) Distributed scope ((¬A)∧(¬B)) (42, ex. (2.2a)) Max didn't read the book and Martha the magazine. =[Max didn't read the book] and [Martha didn't read the magazine]

   2. (b) Narrow scope ((¬A)∧B) (2, ex. (1.1c)) Pete wasn't called by Vanessa but John by Jessie. =[Pete wasn't called by Vanessa] but [John was called by Jessie]

   3. (c) Wide scope (¬(A∧B)) (171, ex. (4.52)) Kim didn't play bingo and Sandy sit at home all night. =It is not the case that [Kim played bingo and Sandy sat at home all night]

Repp seeks to fill a gap (in another sense) in the literature: no previous proposal successfully accounts for all three readings. Of the three major families of gapping analyses – small conjunct coordination accounts, deletion and copying accounts, and three-dimensional/sharing accounts – Repp's analysis is most closely aligned with small conjunct coordination accounts (e.g. Johnson 1996/Reference Johnson2003, Winkler Reference Winkler2005), which are primarily concerned with accounting for wide scope. Repp proposes that the derivation of wide scope readings indeed involves ‘small’ conjuncts, but departs from earlier analyses in identifying these as Topic Phrases (TopPs in Rizzi's Reference Rizzi and Haegeman1997 left-peripheral architecture), and in claiming that distributed/narrow scope readings instead involve large conjuncts, viz. Force Phrases. To yield narrow scope, C2 contains a null ‘positive’ (polarity) morpheme, which contrasts with the overt negative morpheme in C1; under distributed scope, both conjuncts contain negation.

Repp's discussion of wide scope leads her into a close investigation of negation, and she argues, in a further departure from other small conjunct coordination analyses, that two kinds of negation are seen in gapping. Propositional negation is found with distributive scope and contrastive narrow scope. Illocutionary negation, or negation at the level of the speech act, is operative with wide scope and corrective narrow scope: both are denials, ‘a speech act that objects to a previous utterance’ (151). Propositional and illocutionary negation are syntactically distinct, the key to Repp's demonstration that wide scope involves TopPs: illocutionary negation takes a very high position, scoping over both conjuncts. This is further supported by wide scope intonation, which is a single intonational phrase.

Repp further proposes that all gapping involves an ‘L(ogical)F(orm)-Copying’ mechanism (Chung, Ladusaw & McCloskey 1995), which she interprets as sideward movement (Nunes Reference Nunes2004) from C1 to C2, following transfer of C1 to PF. It applies only to those elements that are required to create a convergent derivation in C2, which is otherwise constructed from an ‘impoverished’ numeration containing only material needed to build the remnants. Adjuncts, which (in some languages) include negation, do not copy into C2, a claim I return to below. To capture the information structure of gapping, Repp assumes that both focus movement, of the remnants, and topic movement, of the remainder of the clause, obtain. Lastly, a semantic/pragmatic principle, the principle of balanced contrast (PBC), must be satisfied for gapping to succeed: ‘both conjuncts must make the same kind of contribution to a common discourse topic’ (83).

My chief criticisms of Repp's very thoughtful analysis concern aspects of the LF Copying mechanism she proposes. In line with standard conceptions of sideward movement, she posits that C1 and C2 are built separately, from separate numerations, but in parallel; C1 and C2 are combined into a single syntactic object in the final stage of the derivation, when each merges with the matrix conjunction. By hypothesis, only what is required for a convergent derivation is copied from C1 into C2, with the consequence that adjuncts are not copied. This seeks to account for differential behavior with respect to distributed scope in English and German: while English speakers usually find (1a) acceptable, German speakers find its analogue (2) ill-formed.

1. (2) ??MaxhatdasBuchnichtgelesen,undMarthadieZeitschrift.MaxhasthebooknotreadandMarthathemagazine‘Max didn't read the book and Martha the magazine.’(42, ex. (2.2b))

Obtaining a distributed scope reading in German requires overt negation in C2, which is not possible in English. To capture this, Repp treats English negation as a head, which therefore undergoes sideward movement, and German negation as an adjunct – specifically, an adverb adjoined to vP – which is not copied. (Repp further argues German negation is, in contrast, copied when the gap contains a negative polarity item, because negation c-commanding the gap is in this case necessary for a convergent derivation.) As Repp acknowledges, it is impossible to unequivocally demonstrate that English negation is a head, given that it famously has adjunct-like properties. To maintain her proposed distinction, she suggests, following Zeiljstra (Reference Zeiljstra2004), that English negation has ‘moved away from the adverbial [properties it] used to have in earlier stages of English and now ha[s] acquired head properties’ (54f.). Given that this distinction is a crucial part of her analysis, this virtually unfalsifiable suggestion is disconcerting.

Another potential pitfall lies in Repp's claim that only elements required for a convergent derivation copy from C1 into C2. As these elements, which interleave with the remnants, are disjoint – recall that gapping requires at least two remnants – sideward movement applies piecemeal: the categories V, v, T, and so on copy individually. Ideally, sideward movement is an unconstrained operation, so any number of applications of it, to any syntactic object, should be able to take place, with the resulting derivation being evaluated for convergence. Sideward movement applying piecemeal, according to the needs of the derivation of C2 (e.g. a feature unchecked at a given point), permits C2 to be built according to standard Minimalist assumptions. However, this appears incompatible with the claim that only the elements required to create a convergent derivation can undergo sideward movement: there is arguably no principled way to block copying of adjuncts, which Repp seeks to do, since the resulting derivation would converge either way. She suggests (78) that an economy principle disallows copying adjuncts, but it is not clear to me how, or whether, a standard notion of economy could accomplish this. Recall too that this block is not absolute: the German negative adverb must be able to copy in some distributed scope contexts. Although adjuncts are known to be unusual – as Repp notes, only adjuncts can merge post-cyclically – it is not obvious how their unusual properties could be capitalized upon here. Repp seems to acknowledge this, comparing negation with locative and temporal Prepositional Phrases in C1, which are not ‘reliably interpreted’ in C2 in gapping (71): ‘[I]t seems that the elided negative marker in the distributed scope cases in German behaves just like other adjuncts in gapping, that is its presence can be somewhat undecided. This generally renders the outcome unacceptable’. Repp suggests that a process of ‘accommodation’ is available for other adjuncts, although it is a mystery why negation is immune. Repp notes that in cases like (2) above, C2 ‘dangles between a positive and a negative interpretation’ (78); it is intriguing that this is the result, rather than a straightforward narrow focus reading.

Another question worth exploring involves the licensing of gapping. Although gapping is cross-linguistically not uncommon – Repp draws on gapping data from Dutch, Polish, Russian, Slovak, Japanese, and Korean, in addition to German and English – not all languages allow it. Indonesian, for one, does not, raising the question of what property is shared by this diverse set of languages but absent from Indonesian. Under Repp's analysis of gapping, there are a number of potential candidates. For example, is sideward movement somehow blocked in Indonesian? Or is the PBC somehow inoperative? At first blush, however, none seems especially realistic. The components of her analysis are – as is desired – quite general, and independently motivated, suggesting that a closer investigation of the licensing of gapping is warranted.

My final comment is empirical. This work makes its largest empirical contribution with the narrow scope data, offering the first in-depth examination of intonation patterns associated with narrow scope (as Winkler Reference Winkler2005 did for wide scope intonation), and introducing some novel data involving narrow scope in auxiliary gapping, illustrated in (3).

1. (3) Context: Asking John and Pete to help in the kitchen is useless. They always mess things up. Don't ask them tomorrow to help with the cake.

   John will not find any of the ingredients

   and Pete smash the cake bowl.(109, ex. (3.47))

As a native English speaker, along with several others consulted, I find the narrow scope data to be extremely murky. Neither (1b) nor (3) is, to me, well-formed. Of concern is whether the data's fuzziness undermines Repp's proposals about the syntax of polarity. Repp proposes the PBC primarily to account for narrow scope, noting that the presence of a focus particle – only, even – in C2 is crucial to obtaining narrow scope readings in German. (In English this does not for me result in any improvement.) She argues that but (as in (1b)) associates with contrastive topics, and assumes that contrastive topics have respective foci, which she proposes to be negative and positive polarity, i.e. polarity morphemes are instances of Foc. This is an intriguing proposal, and worth exploring with a much broader, and more stable, range of data.

The book is divided into six chapters. Chapter 1, ‘Introduction’, provides an introduction to gapping, with particular attention to English and German, and an overview of previous analyses of gapping. Chapter 2, ‘The syntax of clausal negation: The distributed scope readings in main verb gapping’, analyzes distributed scope (as exemplified in (1a) above), addresses the difference in behavior between English and German, and compares clausal negation in these languages. Chapter 3, ‘The right kind of contrast: Narrow scope readings’, addresses narrow scope (exemplified in (1b)). Chapter 4, ‘Negation and the speech act’, analyzes wide scope readings (see (1c)) as denials. Chapter 5 looks closely at the role of ‘Finiteness in gapping’, which Repp takes to be crucial in ‘anchoring’ the proposition expressed by the gapped clause to the world and its antecedent, and which again yields differences between English and German. Chapter 6, ‘Summary’, summarizes and concludes the book.

Negation in gapping is overall a very successful book. Repp's methodical deconstruction of every element of her analysis is refreshing and informative. The book is well-organized, and the prose very clear; with it, Repp makes significant empirical and theoretical contributions to the bodies of literature on negation, gapping, and ellipsis in general.