Syntactic change is a well-studied topic in the field of historical linguistics. Questions such as ‘How does syntactic change happen?’, ‘How and where does it start?’, ‘Why does it happen?’, and ‘How can it be explained?’ have occupied historical syntacticians for decades. Much recent work on (English) diachronic syntax is Chomskyan in outlook, adopting the Minimalist framework (Chomsky Reference Chomsky1995). Notwithstanding the fact that it was not designed to explain language change, an ever-increasing number of historical syntacticians employ the generative framework in their search for an explanation of syntactic change (Fischer et al. Reference Fischer, van Kemenade, Koopman and van der Wurff2000, Biberauer & Roberts Reference Biberauer and Roberts2005, among others). In these works, syntactic change goes hand-in-hand with (first) language acquisition and is taken to be the result of imperfect learning, causing parameters to be set differently from one generation to the next. This approach was pioneered by Lightfoot (Reference Lightfoot1991, Reference Lightfoot1999), whose cue-based model sheds light on the role of first-language acquisition in language change, which is accounted for in terms of parameter change. Nonetheless, some important details of this approach remain rather impressionistic.

Ian Roberts's Diachronic syntax is a new addition to historical syntactic work set within the Minimalist tradition. In this book, ‘syntactic change is seen as changes in values of parameters of UG [Universal Grammar] taking place through the process of first-language acquisition’ (291). In what follows, I will give a chapter-by-chapter summary and evaluation of the book, followed by some evaluative remarks on more general aspects of its contents. For reasons of space, the summary is selective, focusing on the most important components of the proposal.

Chapter 1, ‘Comparative and historical syntax in the principles-and-parameters approach’, which is preceded by a short introduction to the entire book, is devoted to a discussion of syntactic variation and change within the Principles and Parameters framework. The chapter provides the synchronic motivation for parameters and gives a first insight into how parameters can be used diachronically to explain syntactic variation and change. Roberts discusses all the well-known parameters, including the null-subject parameter and the wh-parameter. Here I will concentrate on his discussion of the parameters for head–complement order, as this topic has received considerable attention in diachronic syntactic studies, especially those on English. Roberts convincingly shows that, despite some problems with dating some of the changes involved, parametric change can be used to explain word order changes, such as the change from object–verb (OV) to verb–object (VO) in the history of English. The notion of parameter and its role in language variation and change are presented rather sketchily in this chapter, but since it is one of the book's objectives to flesh out an account of parameters, I will refrain from comments about the conception of parameters for the moment.

In chapter 2, ‘Types of syntactic change’, it is shown how parameter theory can account for the various types of syntactic change: reanalysis, grammaticalisation, changes in argument structure, changes in complementation and word order change. Reanalysis, which Roberts (largely following Harris & Campbell's Reference Harris and Lyle1995 definition) takes to be a mechanism which alters the underlying structure, is a crucial ingredient of the analysis. Roberts proposes that reanalysis is intrinsically related to parameter change, in the sense that ‘reanalysis is usually a symptom of a change in the value of a parameter’ (123). In fact, it is shown that grammaticalisation, changes in argument structure and changes in complementation can all be treated as instances of parametric change and therefore all involve reanalysis.

In explaining reanalysis, Roberts makes use of the notion of abductive change, relating reanalysis to language acquisition. The idea is that a new grammar is abduced from the output (corpus) of the old grammar. The new grammar differs from the old grammar because the abduction process involves mismatches, leading to reanalyses. Two problems related to the notion of abductive change, the regress problem and the chicken-and-egg problem, are tackled by appealing to language contact and to other parts of the grammar. The regress problem, which addresses the origin of innovations (if innovations in the new grammar are triggered by something in the output (corpus) of the old grammar, why are innovations not already present in the old grammar?), is solved by the suggestion that language contact may be the cause of innovations in the new grammar. The chicken-and-egg problem, which describes the difficulty of establishing whether an innovation is the effect of a reanalysis or whether the innovation is the trigger for the reanalysis, is handled by positing that changes in other parts of the grammar, such as phonology, may in fact be the crucial trigger for reanalysis. As admitted by Roberts, this latter solution is rather inelegant as it simply shifts the problem to another part of grammar.

If reanalysis is ‘the structural manifestation of the change in the value of at least one parameter’ (132), then how do language learners set a parameter? Roberts answers this question by introducing the concept of P-expression (reminiscent of Lightfoot's term ‘cue’), which entails that syntactic constructions express, and thereby trigger, a certain parameter value. Reanalysis arises when the expression of a parameter by a syntactic construction is strongly P-ambiguous, which means that the construction triggers either value of the parameter. A decisive factor in the choice of parameter value is structural simplicity.

An issue not addressed by Roberts is the question as to when a P-expression is (in)sufficiently frequent in a language learner's input to trigger the parameter value expressed by the construction or, alternatively, to trigger either or neither value (P-ambiguity). The same question remained unanswered in Lightfoot (Reference Lightfoot1999), who uses the notion of robustness – the robust presence of a cue triggers a certain parameter value – without specifying it.

Roberts further shows in this chapter how Kayne's (Reference Kayne1994) antisymmetry approach can account for word order variation and change. An advantage of this approach is that it does not allow the operation of Merge to undergo any change. Only Move and Agree may vary and change, limiting the number of formal options. Thus, the loss of OV orders in English can be seen as ‘the loss of the trigger for movement’ (192).

Chapter 3, ‘Acquisition, learnability, and syntactic change’, explores in more detail the role of first-language acquisition in parameter change. This is no easy task because ‘no-one has really looked carefully for a connection between acquisition and change’ (225). Future work on this topic is therefore eagerly awaited. Roberts uses the concepts of markedness and complexity to refine the connection between parameter change and first-language acquisition. He emphasises once more the triggering role of P-ambiguity in abductive parametric change and adds that P-ambiguity gives rise to opacity. Opacity is defined in terms of markedness, which holds that a marked option or parameter value is more complex than an unmarked option. The unmarked value is the default by virtue of its structural simplicity. Complexity is described in terms of the number of formal features contained in a given structure, so that a marked option has more formal features than an unmarked option. The advantage of ‘typing’ parameter values as marked or unmarked is that direct evidence will be required to trigger a marked parameter value setting. Roberts notes that a parameter may change from an unmarked to a marked value if functional heads acquire an Extended Projection Principle (EPP) feature. This possibility expresses the ‘tension between a drive for simplicity and a drive for expressivity’ (283).

Though the discussion certainly shows how parametric change can be related to language acquisition, its use of technical devices and notions is rather dazzling. The terms used in the discussion (ambiguity, opacity, complexity, markedness) are intricately related and the difference between them is not always clear. For example, opacity and complexity are not evidently distinct, and the discussion would have been clearer if the term opacity had simply been omitted.

In chapter 4, ‘The dynamics of syntactic change’, Roberts addresses several issues concerning the progress of syntactic change. An important issue is the observed gradualness of syntactic change. At first blush, this appears to be problematic for Roberts's approach, which takes syntactic change to be a reflex of changes in parameter values, suggesting abrupt change. However, since parameters are specified in the formal features of functional categories (listed in the lexicon), changes to these features will make them seem to be ‘taking place over a long period and giving the impression of a single, large, gradual change’ (300).

Another factor giving syntactic change its gradual appearance is formal optionality. Roberts suggests that this represents ‘a gradual shift in choice of options’ (309), assuming Kroch's (Reference Kroch1989) competing grammars approach and the Constant Rate Effect, which predicts the same rate of replacement in different contexts. Roberts argues that variation may turn into change, i.e. one variant may win out over the other, as a result of social conditions. Despite the problems of the competing grammars approach pointed out by Roberts (e.g. concerning learnability), he nonetheless considers competing grammars to be a possible way of accounting for the period of variation. It is somewhat disappointing that the proposed parametric account has to resort to competing grammars for its treatment of the gradualness of syntactic change.

Chapter 5, ‘Contact, creoles, and change’, concentrates on the influence of language contact on syntactic change. Roberts persuasively shows how contact situations may unsettle the primary linguistic data (PLD), leading to weak P-ambiguity and causing the child to set parameters differently from his/her parents, which results in a new grammar that diverges from the old grammar. He then turns his attention to the well-known contact case of creoles. The development of creole languages has traditionally been thought to be different from the generation-to-generation development of non-creoles because of their origin as contact varieties, which is said to involve some kind of exceptional break in language transmission. On the basis of data from different creoles, Roberts argues that creoles do not in fact represent an extreme or special case of the mechanism of language change, but are different only because their PLD (from a pidgin) is more impoverished than usual.

The language contact cases discussed in this chapter all involve what Roberts terms ‘grammatical expansion’, which is the result of ‘weak P-ambiguity in the PLD’ (439), which causes neither of the two parameter values to be triggered. As pointed out by Roberts in the closing lines of the chapter, this contrasts with the claim made in chapter 2 that abductive reanalysis involves strong P-ambiguity. This raises ‘the question of whether the types of ambiguity naturally favour different types of change’ (439), but Roberts does not pursue this question.

Lastly, some general aspects of the book deserve to be highlighted. The book's structure is well organised and typographical errors are infrequent. Overall, the topics are presented and discussed clearly and in a logical order, although it has to be observed that at times the regular referrals to later parts of the book interrupt the flow of reading. The book uses a huge range of cross-linguistic data to illustrate the approach taken, something that this review has been unable to do justice to for reasons of space. Each chapter is concluded by an extensive ‘Further reading’ section, subdivided into sections by topic. These provide a rich source of additional literature and will be useful to researchers and (graduate) students alike. The same holds for the explanatory boxes throughout the book, which explain concepts mentioned in the main text in more detail. Equally useful is the glossary of terms included at the end of the book. Terms listed in this glossary are printed in bold in the main text, allowing easy cross-reference.

Overall, this book is an important contribution to our understanding of syntactic change (and language change in general) in terms of the Principles and Parameters approach and its successor, the Minimalist framework. It shows convincingly how generative theory can be applied to historical linguistics, in a way that is almost unprecedented in its detail and coverage. For these reasons, Diachronic syntax will serve as an incentive and inspiration for generative researchers of historical linguistics.