Daniel Altshuler & Robert Truswell, Coordination and the syntax-discourse interface (Oxford Surveys in Syntax and Morphology 11). Oxford: Oxford University Press, 2022. Pp. xii + 315.

In the last half century, much research has focused on the nature of coordination, particularly on its extraction patterns, collectively dubbed as the Coordinate Structure Constraint (CSC) by Ross (1967). The first part of the CSC, known as the Conjunct Constraint, bans the extraction of conjuncts (*Who did you see Robin and _?), and the second part, known as the Element Constraint, blocks extraction from conjuncts (*Who did you see Robin and a picture of _?). The latter is circumvented when extraction is Across-the-Board (Who did you buy a picture of _and a book about _?) or when the the order of conjuncts has certain asymmetrical interpretations, such as narration (e.g. Here’s the whiskey which I went to the store and bought _) or violated expectation (e.g. How much can you drink_and still stay sober]?). Prior research on the CSC focuses on the syntax or semantic-pragmatic components, not on their interaction. The strength of this monograph is precisely the attention that both domains earn. The conclusion (276) is that neither syntax nor discourse is exclusively responsible for explaining the observable extraction patterns; rather, both are needed to fully understand how coordination interacts with the grammar of extraction. This book offers a deep examination of the role that discourse restrictions play in explaining how extraction works in coordination, covering an extraordinary amount of data. Apart from minor issues discussed below, it is well-argued. Chapter 2 surveys definitions of the object of study and coordination, and it concludes that they defy a cross-linguistic characterization, as there are no necessary and sufficient morphosyntactic properties in the languages of the world. Another challenge for a definition of coordination that the authors mention is ‘nonconstituent’ coordination phenomena, such as Right-Node Raising (RNR) and Gapping, but the problem with this claim is that these phenomena are not strictly related to coordination:

As such, RNR should be accounted for via a more general mechanism, having nothing to do with coordination. Indeed, for some authors, RNR is a general phenomenon that omits phonological units under morphemic identity (Shiraïshi et al. 2019). Gapping is similarly not restricted to coordinate structures, as the attested examples from Park (2019: 30–38) in (2) illustrate:

Chapter 3 surveys the CSC in English and in several other languages, discussing counterexamples to the Element Constraint and to the Conjunct Constraint. Some of these apparent counterexamples are admittedly difficult to evaluate, since they could be comitatives rather than ‘true’ coordination, and/or instances of Stripping (Hankamer & Sag 1976). Competing syntactic accounts of coordination are also discussed in Chapter 3, albeit with a focus on Combinatorial Categorial Grammar and Minimalism. The book offers marginal discussion of Head-Driven Phrase Structure Grammar, except for a problem in footnote 11 (31), which is not problematic at all (Abeillé 2006). As in Chapter 2, the conclusions drawn at the end of Chapter 3 are modest: the relationship between symmetrical and asymmetrical patterns of extraction in coordinate structures is not clear. Either the unmarked pattern is symmetrical ATB extraction, or ATB extraction has no special status in the syntax (although it might have one in semantics). To resolve this impasse, Chapter 4 concentrates on syntax and Chapter 5 on discourse-pragmatics. One observation made in Chapter 3 (60) – and used later in Chapter 5 (148) – is that the CSC is operative on quantifier scope (Ruys 1993) and perhaps emerges from a parallelism constraint that is triggered by the presence of symmetric coordination (Fox 2000, Kehler 2002). But although inverse scope readings in coordination are disfavoured, they are impossible (Chaves 2007: §3.6). In (3a), the second conjunct can outscope the first, and in (3b), it can outscope the entire proposition:

This weakens the argument that the first conjunct must be structurally higher than the second and challenges the notion that quantifier scope is due to the same mechanism responsible for extraction. Regardless, the authors correctly note various other challenges to extant definitions of parallelism in Chapter 5 and Chapter 6 and propose a refinement.

Chapter 4 surveys syntactic theories that attempt to explain the extraction patterns observed in coordination structures and argues that they fail. The case can be made stronger, given constructions where the extracted phrases exhibit syncretism, as in (4). Although the authors follow Williams (1978) in dismissing these constructions as ungrammatical (104), various others have argued that such constructions are possible (Levine et al. 2001):

Such a phenomenon casts doubts on sideward movement analyses and multidominance since the extracted constituents are patently not mutually consistent. Further challenges come from dependencies in which two different wh-phrases combine to form a plurality (Munn 1998, Gawron & Kehler 2004, Vicente 2016), as in (5):

Chapter 5 focuses on Lakoff (1986) and on the various types of discourse-structural relations adopted by Kehler (2002) to account for extraction from coordinate structures. A key contribution of this chapter is a precise formal definition of the occasion, violated expectation, result, and parallel discourse relations. In the process, it is shown that occasion and violated expectation discourse relations allow extraction from initial or final conjuncts, contrary to what previous accounts would predict. It is also argued that result is a special case of occasion, which poses a problem for Lakoff’s account since the two relations give rise to distinct extraction patterns in asymmetric coordination. In particular, it is claimed that result environments block extraction from the non-initial conjunct (166-167). But the examples in (6), which are reworded versions of the original examples (61b) and (65b), are acceptable.

Perhaps the examples in (6) do not involve result, but rather consequence (264,ft.83), which the authors illustrate with Which base runner was Doc following his coach’s instructions and, as a result, keeping_close to first? However, the distinction between the two relations is not clear enough to confidently decide. More data are needed to shed light on the subtle contrasts discussed in this volume, ideally in the form of controlled acceptability experiments.

Chapter 6, the final chapter, is a veritable tour de force which guides the reader through a daunting labyrinth of observations and often conflicting factors. The chapter begins by noting the great potential for approaches based on pragmatics in explaining various island constraints – which I note is in line with growing experimental evidence that pragmatics plays a key role in many islands (Abeillé et al. 2020, Vincent et al. 2022, Lu et al. 2021) – while at the same time noting challenges for purely discourse-based approaches. For example, the contrast in (7) is supposed to show that extraction of a referential NP is only possible from the initial conjunct in Type A scenarios in Lakoff’s terms (occasion or background), even though such scenarios allow extraction from any conjunct (255, 276). However, native speakers with whom I consulted rate (7b) higher than (7a). Again, more robust judgement sources are necessary to understand such variation:

Kehler’s (2002) idea that coordination extraction patterns are determined by discourse relations, mediated by information structure, is refined and expanded using Segmented Discourse Representation Theory (Asher & Lascarides 2003). In the process, the authors show how discourse continuations contribute to discourse topics and their constraints on common themes. A new formal repertoire of discourse relations, which offers new insights about extraction in coordination, is proposed. The book makes many important and valuable contributions and stands as a new and rich resource for future generations. I recommend it to anyone interested in coordination, discourse structure, or how syntax and discourse interact to give rise to rich and complex phenomena.