Construction grammar emerged in the late eighties as a new approach to grammatical analysis, and has since then attracted a growing body of linguists applying and further developing this model of grammar. Construction grammar has now become a well-established framework with its own peer-reviewed journals (Constructions and Constructions and Frames), a dedicated book series (Constructional Approaches to Language) and an international biennial conference. However, until recently, an authoritative introduction to the framework had been lacking. The Oxford Handbook of Construction Grammar aims to fill this gap, with twenty-seven chapters addressing theoretical and empirical issues in construction grammar written by leading linguists in the framework.

The handbook begins with an introductory chapter by the editors, Thomas Hoffmann and Graeme Trousdale. They insightfully relate the central notion of ‘construction’ in construction grammar to the Saussurean sign, i.e. an arbitrary and conventional pairing of form and meaning. Construction grammar, in their words, explores ‘the idea that arbitrary form–meaning pairings might not only be a useful concept for describing words or morphemes but that perhaps all levels of grammatical description involve such conventionalized form–meaning pairings’ (1). This programmatic statement captures well the common message of the contributions in the handbook, while also allowing for a substantial breadth in theoretical and methodological approaches. The following chapters in the handbook are organized around five central topics: principles and methods, constructionist approaches, constructicon, acquisition and cognition, and language variation and change.

Adele Goldberg opens Part I, with an overview of shared underlying assumptions in construction grammar. She identifies four major tenets, shared by all constructionist approaches, which set them apart from mainstream generative grammar: phrasal and clausal phenomena are analyzed as constructions, the focus is on surface form, constructions are organized in networks, and functional and cognitive motivations for constructions are emphasized. A final tenet, not equally recognized by all construction grammarians, is the usage-based nature of linguistic structure. It is true; these main principles recur in many chapters in the handbook, despite the (sometimes surprisingly) wide diversity in theoretical and methodological approaches to these shared principles. Goldberg's chapter is followed by three chapters highlighting some fundamental controversies in construction grammar, concerning the limits of grammar and the relationship between linguistic competence and performance.

Paul Kay defends a rather restricted notion of grammar in his contribution, stating that ‘a grammar should contain the strictly linguistic information required to produce and understand all possible utterances of a language and no more’ (32, emphasis in the original). This essentially generative perspective on grammar excludes partially productive generalizations over lexical items, which he calls ‘patterns of coining’ (33). Such patterns are considered to be part of meta-grammar, ‘a compendium of useful statements about the grammar’ (33–34, emphasis in the original). Kay explicitly couches his strict division between grammatical constructions on the one hand and patterns of coining on the other in the broad generative tradition based on Chomsky's idealization of linguistic competence. His generative approach is quite different from the usage-based perspective defended by Joan Bybee in the subsequent chapter.

Bybee points out that, while the term ‘usage-based’ was first coined by Ronald Langacker in the late eighties, the perspective ultimately goes back to earlier functional-cognitive work by Greenberg, Givón, Li, Thompson, Hopper and many others. The central premise of usage-based grammar is that linguistic structure is formed by the repetition of linguistic patterns in use. Bybee, in particular, presents an exemplar-based model of grammar, where ‘the general categories and units of grammar can emerge from experience that is recorded in memory’ (52). This emergent and gradual approach to grammar strongly contrasts with the fundamental division between grammar and meta-grammar suggested by Kay.

The third chapter, also taking up controversial issues in construction grammar, comes from Ray Jackendoff. His point of departure is his framework of the Parallel Architecture, a model of grammar sharing some (but by no means all) of its basic tenets with construction grammar. Jackendoff provides an insightful discussion of the relationship between grammar and lexicon, mainly with reference to Kay's chapter, and arrives somewhere in between the positions of Kay and Bybee. However, it remains rather unclear exactly why Jackendoff was asked to participate in this discussion, given his status as an outsider to construction grammar.

The final chapter of Part I moves on to the empirical foundations of construction grammar. Stefan Gries points out that, while the pioneering constructionist studies of the late nineteen-eighties largely relied on introspection, data from experiments and corpora have become increasingly popular since the late nineties. His chapter provides an instructive overview of the analytical tools that have been developed and used in the construction grammar framework since then.

Part II of the handbook gives an overview of the different constructionist frameworks that have been developed in the last twenty-five years. Charles Fillmore begins by reconstructing the basic principles of Construction Grammar, as it originally emerged in the late eighties at Berkeley, on the basis of his own recollections and notes. Berkeley Construction Grammar (BCG), as the framework is known today, developed as a reaction to mainstream generative linguistics seeking simple and universal generalizations of human language. In contrast, BCG searched for a way to incorporate noncanonical and idiomatic patterns in linguistic description, arriving at constructions as an analytical tool for describing both peripheral and core structures in language. Fillmore introduces some formal conventions of BCG, such as the use of the attribute–value matrix for the representation of construction structure, and illustrates a number of formal operations that can be applied to these matrices. Laura Michaelis, in the next chapter, goes on to introduce Sign-Based Construction Grammar (SBCG), which is a recent synthesis of Berkeley Construction Grammar and Head-driven Phrase Structure Grammar, aimed at combining the broad empirical coverage of the former with the formal explicitness of the latter. Both BCG and SBCG can be characterized as highly formal and generative models of grammar.

The following two chapters in Part II present a computational approach to construction grammar. Luc Steels introduces Fluid Construction Grammar (FCG), a formalism originally developed in the context of artificial intelligence research. It may serve as an open (fluid) instrument for construction grammarians to formulate their intuitions in a precise way and to test their models for language parsing, production and learning. Benjamin Bergen & Nancy Chang then present an alternative formalism which grew out of their neurocognitive study of language. They are particularly interested in what cognitive and neural mechanisms people engage with while using human language. One of their central claims is that ‘meaningful words and constructions serve as interfaces that activate and constrain modality-rich embodied simulations’ (173). The importance of embodied simulations in the brain is reflected in the name of their framework, i.e. Embodied Construction Grammar (ECG). At first sight, the frameworks of FCG and ECG seem to be closely related to the formalist approaches of BCG and SBCG discussed above. However, it is important to stress that FCG and ECG are essentially usage-based, focusing on how constructions are used in parsing, production and learning, whereas BCG and SBCG represent generative constructionist models of grammar.

The last three chapters of Part II introduce usage-based frameworks that largely eschew highly formalized descriptions. They all take a cognitive perspective on grammar, assuming linguistic structure to be grounded in domain-general cognitive processes such as association, automatization, schematization and categorization. Cristiano Broccias presents the framework of Cognitive Grammar (CG) developed by Ronald Langacker since the late eighties. Constructions are defined in CG as symbolic structures pairing a semantic pole (meaning) with a phonological pole (form). This definition diverges from other constructionist approaches, as syntactic structure is not considered to be part of the form pole of grammatical constructions, but rather is thought to reside in the symbolic link between the form and meaning pole. William Croft, in turn, summarizes the principles of his framework of Radical Construction Grammar (RCG), launched in 2001. Dispensing with universal categories, RCG argues instead that categories are defined by the particular constructions they appear in. Moreover, RCG does not assume any syntactic structure apart from the part–whole structure of constructions and grammatical roles. Hans Boas closes Part II with a presentation of Cognitive Construction Grammar (CCG), based on Adele Goldberg's 1995 monograph. CCG is best known for its work on argument structure, challenging one of the core areas of mainstream generative grammar.

Parts I and II give a rich overview of both established and upcoming construction grammar frameworks. The increasing array of diverging theoretical premises and formalisms, however, calls for some systematic comparison. There exist a number of contrastive literature reviews (Croft & Cruse Reference Croft and Cruse2004; Langacker Reference Langacker, Ruiz de, Ibáñez and Cervel2005, Reference Langacker2009; Goldberg Reference Goldberg2006), but the rapid developments in the field mean that these texts are already obsolete, at least with respect to coverage. Chang, de Beule & Micelli (Reference Chang, De Beule, Micelli and Steels2012) and van Trijp (Reference van Trijp2013) partly fill this gap by comparing the newcomers in the field in pairs, specifically, Embodied CG versus Fluid CG, and Fluid CG versus Sign-Based CG. Nevertheless, there remains a pressing need for a more comprehensive comparison of all constructionist approaches that have emerged in the last twenty-five years. The absence of such a contrastive chapter is, in my opinion, a regrettable lacuna in the handbook.

Another issue that also receives too little attention, to my mind, is the relationship between construction grammar and other theories of grammar, both historically and at present. One recurrent theme of many chapters in the handbook is that construction grammar emerged as a reaction to ‘mainstream’ or ‘traditional’ generative grammar. However, both construction grammar and generative grammar are by no means monolithic theories (any more), so that this contrast is often too simplistic and sometimes even misleading. Consider, for instance, the controversy around the limits of grammar as addressed in the chapters by Kay and Bybee. I have pointed out that Kay (representing Berkeley and Sign-Based CG) aligns himself with the broad generative tradition, whereas Bybee (representing usage-based approaches) ties in with the broad functional-cognitive tradition (in North America, one should add). Such fundamental oppositions are, in my opinion, too often swept under the carpet, leading to inconsistent statements such as ‘Constructionist Grammar offers a psychologically plausible, generative theory of human language’ (3). I believe that the handbook would have benefited from an additional chapter situating the diverging constructionist approaches against a broader theoretical and historical background.

Part III moves on to constructionist analyses of linguistic phenomena at various levels of schematicity and complexity, or, as the editors put it, from morphemes to clauses and beyond. Geert Booij presents arguments for a constructionist approach to morphology, stressing the holistic properties of morphological constructions and the hierarchical nature of the lexicon with different layers of abstraction. Stefanie Wulff surveys the constructionist research into idiomatic expressions, initiated by the pioneers of Berkeley CG in the late eighties, and currently taking a more usage-based perspective, exploring how the degree of semantic and syntactic irregularity can be measured quantitatively. Anatol Stefanowitsch introduces collostructional analysis, which has become a well-established quantitative measure for assessing collocational preferences in schematic constructions. Thomas Hoffmann demonstrates how constructionist approaches deal with abstract constructions at the clause level (such as interrogative or relative clauses), and with filler–gap phenomena. Finally, Jaakko Leino illustrates how information structure has attracted much interest in construction grammar. He points out that information structure is needed to explain why speakers choose a given expression type rather than some other alternative. Part III as a whole demonstrates that construction grammar has moved beyond its original research focus on idiomatic expressions and argument structure. As some of this research is still in its infancy, it can be expected that these chapters will also have a programmatic function for future research.

Part IV surveys constructionist research on language acquisition and cognition. Holger Diessel shows how a usage-based model of first language acquisition, crucially influenced by the assumptions of construction grammar, provides an appealing alternative to the long-standing nativist account developed in mainstream generative grammar. The central idea of this usage-based model is that children acquire language ‘in a piecemeal bottom–up fashion based on general learning mechanisms that are also relevant for the acquisition of knowledge in other cognitive domains’ (351). Nick Ellis elaborates on how this inductive emergent model of learning can be extended to second language acquisition, albeit with the twist that second language learners are often affected by transfer from their native tongue. Giulia Bencini presents a psycholinguistic model of language production, only turning to constructions, however, in a final (very brief) section. The direct relevance of this chapter to mainstream constructionist theory building and analysis remains rather obscure to me. Finally, Friedemann Pulvermüller, Bert Cappelle & Yury Shtyrov explore the pertinent question of whether the basic tenets of construction grammar are compatible with current brain theories of language. Brain research indicates that it is correct to treat language as an extension of other (older) neurocognitive features and to treat syntax and semantic as interwoven. However, the authors also warn against a total abolition of the lexicon–syntax distinction, as words and syntactic sequences of words appear to trigger different responses in the brain. The chapters in Part IV, taken together, demonstrate how constructionist, and in particular, usage-based approaches to language are supported by (and in their turn support) recent insights from psycholinguistics and neuroscience.

Finally, Part V turns to the study of language variation and change from a constructionist perspective. Mirjam Fried shows that construction grammar is increasingly used to model language change, often in close association with grammaticalization research. The chapter draws our attention to the gradual nature of constructional change, the micro-steps involved in these changes and the importance of context. Jóhanna Barðdal demonstrates how the concept of a construction, as a pairing of meaning and form, allows us to reconstruct syntactic structures in a similar vein as lexical, phonological and morphological structures. Martin Hilpert provides an overview of the growing body of research using methods of historical corpus linguistics to uncover patterns of constructional change. On a theoretical level, he makes a fundamental distinction between general patterns of language change and construction-specific processes. Jan-Ola Östman & Graeme Trousdale investigate how synchronic variation may be modeled and constrained in construction grammar, focusing in particular on dialect and discourse variation. Finally, Willem Hollmann explores the question of how sociolinguistics and construction grammar can benefit from each other, drawing on the recent plea for social-cognitive linguistics by William Croft and recent work in the framework of Cognitive Sociolinguistics. He concludes that cognitive linguists increasingly acknowledge the need to take on board issues related to social structure in their linguistic analysis.

In the introduction, the editors claim that their handbook is the most authoritative and comprehensive overview of construction grammar presently available. On the whole, I agree with this claim. The preceding summary of the handbook should have given the reader an impression of its theoretical and empirical breadth. This is not to say, however, that there is no room for some objections at a more detailed level. I have already raised questions about the relevance of some chapters in this handbook. I have also pointed out the (unfortunate) absence of a chapter presenting a contrastive overview of available constructionist approaches and one discussing the relation of construction grammar with other theories of grammar. I should also mention here that some chapters turn out to be quite a tough read (in particular Michaelis, Steels and Pulvermüller et al. were rather a challenge to me) due to the formalist nature of the framework presented, or as a result of interdisciplinary references to computational linguistics or brain science. Nevertheless, the handbook as a whole provides, in my opinion, both an excellent introduction to construction grammar for novices to the framework, and a comprehensive literature survey for more seasoned construction grammarians.