Particles serve up a complex cocktail of morphosyntactic puzzles, opening up a window on a range of broader questions while at the same time leading the researcher out into a thicket of seemingly intractable paradoxes. Particles are independent words, yet straddle the boundary of morphology and syntax by occurring in positions that similar syntactic constituents usually cannot appear in. Particles may diachronically develop into subparts of complex verbs, but the distribution of those affixes suggests that they maintain a syntactic presence outside the heads they are hosted by. Particles and the verbal affixes they morph into are typically in complementary distribution with other secondary predicates, suggesting that they are secondary predicates themselves, yet they occasionally combine with secondary predicates, even with other particles. These particle puzzles have been studied for decades. Still their complexity continues to confound us. A fresh look that is both closely comparative and diachronic, shedding light at the same time on the synchronic distribution of particles in more than just a single language and on the historical development of these elements, is most welcome. The book under review here aims to present just such a fresh look.

The volume, whose six substantive chapters weigh in at approximately 200 well-written and easy-to-read pages (plus a brief introduction and four pages of general conclusion), was put together by a quinquevirate from the Netherlands. The unusually large number of authors is due to the fact that this work synthesises two doctoral theses (Blom Reference Blom2005 and Elenbaas Reference Elenbaas2007, both looking at synchrony as well as diachrony, with the former focused on Dutch complex predicates and the latter on English verb–particle constructions) and the other fruits (including an international workshop on preverbs in 2001, in Nijmegen) of a nationally funded research project on the diachrony of complex predicates in the West Germanic languages run by Booij and Van Kemenade, and with Los as a key participant. As the dates mentioned in the previous sentence indicate, the book has been fairly long in the making.

The back-cover blurb whets the reader's appetite by asking many time-honoured questions about particles. The first question raised in the blurb frames the book's interest in the division of labour between morphology and syntax: ‘Is a particle verb like look up one word or two?’ The authors answer that it is two words – but there is something peculiar about the second word: ‘Particles are words that are optionally projecting. If particles do not project a phrase, they have the status of X⁰ (bare head); if they project a phrase (XP), they can be modified and topicalized’ (53).Footnote ¹ Here lies the core of the proposal for verb–particle constructions: there are three structural patterns for particle verbs, given here in (1) (=example (30) on page 68).

1. (1)

   1. (a) [_V′ XP V⁰] where XP=PP, AdvP, AP or NP

   2. (b) [_V′ X⁰ V⁰] where X⁰=P, Adv, A or N

   3. (c) [_V⁰ X⁰ V⁰] where X⁰=P, Adv, A or N

For the idea that particles are optionally projecting heads, the authors acknowledge Neeleman (Reference Neeleman1994), Zeller (Reference Zeller2001, Reference Zeller2002) and Toivonen (Reference Toivonen2002, Reference Toivonen2003). While its predecessors' proposals are firmly embedded in well-defined theories of syntax, this book does not present a syntactic theory which could deliver the taxonomy in (1) and its associated properties on principled grounds (bare phrase structure cannot procure (1); standard X-bar-theoretic approaches cannot countenance (1b) alongside (1a) and (1c)). To the extent that the book contains syntactic analysis at all, this is almost entirely presented in the form of prose descriptions: the number of explicit syntactic representations is minimal. The structures in (1) are repeatedly referred to in the text as ‘templates’, ostensibly part of ‘a hierarchy in which more specific templates can be seen as instantiations of the more general templates that dominate them and from which they inherit all the properties that they both share’ (87) – the authors' allegiance to construction grammar is apparent here (and also in the expression ‘phrasal constructional schema’ offered in the conclusion (211) as a clarification of the term ‘template’).

The authors clarify that (1a) ‘represents a regularly productive syntactic option’ (i.e., complementation of a secondary predicate; in Chapter 5 they embrace a Larsonian/Hale&Keyser-style approach to secondary predication), that (1b) ‘represents the option of optional projection, in which the particle cannot be topicalized or modified’ (68), and that (1c) ‘represents optional incorporation, a prerequisite for incorporation of the particle in the verb cluster’ (68). When V takes a non-projecting X as its complement, we have two options for labelling the resulting object: as V′, as in (1b), or as V⁰, as in (1c). Only in the latter case can the non-projecting X show up inside the verbal cluster. But even then, apparently, it is not forced to incorporate: ‘[b]y assigning a V⁰-node to [separable complex verbs], we represent their phrasal nature, and hence their syntactic separability’ (67).

If it is true that (1c) is the only way to get the particle incorporated into the verbal cluster, the prediction is that incorporated particles must always be placed to the immediate left of the verb that selects them. This is false: all of the variants of Dutch dat Jan Marie (op) had (op) willen (op) kunnen (op) bellen ‘that Jan had wanted to be able to call Marie up’ are grammatical. Barring a treatment of verb clusters as complex V⁰s or mysterious rules allowing particles to migrate through the verbal cluster outside the V⁰ that they are base-hosted by (neither of which is supportable), the authors' approach to particle incorporation is descriptively inadequate.

Nominalisations like fallout/startup and downpour/upstart aside, present-day English resists particles as subparts of 0-level units. So it probably does not employ (1c) anymore – for reasons that remain obscure, and to which this book has nothing to add.Footnote ² On the authors' analysis, present-day English does heavily exploit (1b). Here, the particle is forced to morphologically merge with the predicate head, delivering the V–Prt–Object order. The alternative V–Object–Prt order is derived when the particle projects a syntactic phrase, as in (1a). Since (1a) and (1b) are both made available by the grammar, English can oscillate between the two linear orders.

This oscillation is explicitly reined in by another key ingredient of the book: the idea that particles are heads by default, and that the language user/learner will have them project a full phrase only if there is ‘robust evidence’ (54) for this – by Elenbaas' (2007) Structural Economy Principle, ‘a lexical head does not project, unless it is required to do so by syntactic factors’ (54, 129; my emphasis). Intriguingly, however, Chapter 5 takes syntax entirely out of the picture for the choice between (1a) and (1b): the discussion capitalises on the role of information structure (closely following Dehé's Reference Dehé2002 work on particle constructions). The claim is that ‘particles in focus position project a phrase, but do not do so in non-focus position’ (129). But there is, in fact, no correlation between focus and the placement and phrasal projection of a particle. A particle can be contrastively focused when it clearly does not project, as in Dutch dat Jan wildoorlopen, nietteruglopen ‘that Jan wants on-walk not back-walk, i.e., that Jan wants to walk on, not back’ (where the inclusion of the particle in the verbal cluster is, for the authors, unequivocal evidence for non-projection). Conversely, a particle that is modified and hence clearly projects does not have to be focal. The following question–answer pair is perfectly felicitous: (A) Which of these laws did the Supreme Court strike/vote right/straight down? (B) It struck/votedthisone right/straight down. Here speaker A's question contains the premodified particle down, and speaker B repeats this (now clearly topical) modifier+particle combination in the answer, placing it to the right of the object, which is the focus of B's utterance. Similarly, the object can be contrastively focused in verb–particle constructions in which the particle is modified: The Court struckthislaw right/straight down, notthatone. If such utterances were information-structurally infelicitous, this would heartily corroborate the close connection between information structure and particle placement/projection that the authors postulate. The fact that they are fine undermines a purely information-structural approach to particle placement.

The authors resort to pragmatics in other contexts as well, perhaps most eye-catchingly in their discussion in Chapter 4 (102, 109, 110) of the fact that particle verbs like toe-spreken ‘to(prt)-speak’ used to be ditransitive in Middle Dutch but no longer support two objects in the present-day language. They say that the direct object (something like ‘words’) became pragmatically redundant, and that the pragmatic redundancy of the direct object led to its demise. Not only is the use of inchoative ‘become’ (in ‘became pragmatically redundant’) inappropriate (the object ‘words’ of ‘speak’ has always been ‘redundant’, it did not suddenly become redundant); more importantly, ‘pragmatic redundancy’ cannot explain syntactic distribution. Pleonastics are redundant par excellence yet do and often must show up. And even with ‘speak’-type verbs, ‘pragmatically redundant’ objects must be present under particular syntactic conditions, which may vary from language to language – thus, while the combination of speak and the particle out in English does not force a direct object (John spoke out against discrimination), the combination of Dutch spreken and uit does (either in the form of a fake reflexive, as in Jan sprak *(zich) uit tegen discriminatie ‘Jan spoke refl out against discrimination’, or in the form of a ‘pragmatically redundant’ noun phrase headed by woord, as in Jan sprak *(de woorden) uit ‘Jan spoke the words out, i.e., Jan pronounced the words’). Pragmatics is no panacea, and it most certainly cannot be a substitute for syntax.

Information structure is also claimed to answer one of the primordial questions about English verb–particle constructions, raised in the book's blurb: Why can we say He looked it up but not He looked up it? Since it is by its very nature non-focal, the particle has to take the focus position, which forces it to project, which in turn forces it to follow the object. But in Swedish, which is systematic in its use of the V–Prt–Object order, this linear order survives perfectly well with non-focal pronouns. If there were such a dramatic information-structural penalty on V–Prt–Pronoun strings as to make them ill-formed, Swedish would be expected to proscribe such sequences. Passivisation of the verb–particle construction would handily manoeuvre the object out of the way; but Swedish does not force its verb–particle constructions with pronominal objects to be passive. Besides, when Swedish does passivise its verb–particle constructions, the particle is preferably (in fact, for most speakers obligatorily) incorporated into the participle (Skräpet måste bli utkastat ‘The scrap had to be out-thrown’; Svenonius Reference Svenonius1996: 20). Thus, passivisation is not for the benefit of focus on the particle; and the inherent non-focality of pronouns like it is not the ‘trigger’ for V–Pronoun–Prt order, unless information structure in Swedish is strikingly different from that in English and the other Scandinavian languages.

The authors would have been well advised to stay closer to their own formulation of the Structural Economy Principle, and to give syntax rather than information structure the central role in adjudicating the distribution of (1a) and (1b). That syntax rules here is particularly clear from the particle placement and modification properties of constructions which besides a verb and a particle also contain an additional secondary predicate associated with the theme. These come in a variety of surface forms, depending on the category of the additional secondary predicate. Examples include make John out (to be) a liar, bring Bill up a Catholic, put a book down on the shelf, and the ditransitives send a schedule out to stockholders and send the stockholders out a schedule. Kayne's (Reference Kayne, Guéron, Obenauer and Pollock1985) seminal paper (which gets just a single token reference in the book, on page 36) put these constructions on the map. Among the puzzles that these constructions present are the fact that the particle must be placed to the right of the theme in make (*out) John (out) a liar but is flexibly placed when to be is added (make (out) John (out) to be a liar), and the fact that modification of the particle is possible in send a schedule (right) out to the stockholders but not in send the stockholders (*right) out a schedule. These puzzles are syntactic in nature; it is difficult to see how information structure could solve them. Their significance for the syntax of verb–particle constructions can hardly be overestimated. These constructions should give us pause before agreeing with the authors that the placement of the particle in the linear string in English ‘does not seem to be dictated by any syntactic considerations’ (5).

In their analysis of the diachronic development of Germanic particles, the authors claim that particles were ‘phrasal secondary predicates in origin’, and underwent ‘a process of grammaticalization’ that ‘resulted in optional phrasal projection, indicating lexicalization of the [verb–particle construction]’ (139). Nothing in the data reviewed by the authors suggests, however, that particles in early Germanic were uniquely phrasal and in due time grammaticalised into optionally projecting heads: indeed, by their own admission, ‘particles adjacent to the verb may already have been optionally projecting’ (150) even in the earliest historical records. Unless structural economy or default non-projection is itself taken to be the result of historical change, the logic of the Structural Economy Principle would have it that particles exhibited a steady state of preference for non-projection throughout the history of Germanic.

Grammaticalisation is arguably involved, in the case of Dutch (and German), in the development of (i) erstwhile particles into verbal prefixes (be-, ver-, ont-/ent-), and (ii) transitive adpositions into the initial members of inseparable complex verbs (e.g., Dutch door- of doorbréken ‘(lit.) through-break, breach’ as distinct from the separable particle verb dóórbreken; similarly for onder-, over-, voor-, achter-, om- and aan-). The discussion in Chapter 7 makes the important point that (i) and (ii) are historically different evolutions, but that both involve a process of grammaticalisation from the head of a phrase to a non-projecting head.

Though it invokes grammaticalisation in various contexts, I have not discovered much novel insight into the workings of morphosyntactic change in this book. Interestingly, the book's main title is identical with Olga Fischer's (Reference Fischer2007) Morphosyntactic change: Functional and formal perspectives. Fischer (another Dutch historical linguist; Van Kemenade and Fischer were two of the co-authors of the valuable CUP book The syntax of early English) presents a critical evaluation of grammaticalisation theory and casts her net wide in her empirical discussion of morphosyntactic change. The present book, by contrast, is empirically much narrower in scope. It presupposes grammaticalisation theory and uses it as a tool, not advancing in any significant way our fundamental understanding of the theory of morphosyntactic change.

One of the major conclusions drawn in the book is that verb–particle constructions, while compositional and formed productively, can have conventionalised properties extraneous to verb and particle – ‘idiomaticity and transparency, or rather, conventionality and compositionality, are not mutually exclusive’ (211). The authors accommodate this with their ‘phrasal lexical templates’. Though Distributed Morphology does not feature in the book, this ‘lexicalisation’ of phrasal units is best thought of against the background of DM's postsyntactic Encyclopedia, which lists the idiosyncratic semantic properties of idioms, whether lexical, phrasal, or transformational (to have no leg to stand on is a syntactic construct that exists as an idiom ‘have no valid argument’ only thanks to the application of infinitival relativisation: to have to stand on no leg, to the extent that it is meaningful at all, is not an idiom; feeding the output of a syntactic transformation back into the lexicon would be cumbersome).

Let me return to the book's blurb once more, and review the last questions that it raises. ‘How did such hybrid words arise and how do they function?’, and ‘How can we make sense of them in modern theories of language structure?’. By giving syntactic analysis such short shrift, the authors leave the latter unanswered. To its credit, however, the book does provide plenty of discussion on the question of how particle verbs arise and function. As a case study of the historical evolution of particle verbs, the book offers much useful material and perspective to students of particles, the interface between morphology, syntax and semantics, and diachronic morphosyntax. In that capacity, this book has much to recommend it.