As its title implies, this is a hugely ambitious book, and to me a hugely convincing one. Building upon core ideas culled from the usage-based paradigm, Schmid develops a comprehensive, multifaceted model of how language works. Language is seen as a persistently dynamic complex-adaptive system, a continuous feedback process driven by the interaction of the three principles listed in the book's subtitle. Usage relates to speakers' communicative activities; it shapes conventionalization, the shared social knowledge on linguistic habits and principles which may be called a communal grammar; and in turn it is the product of and fueled by entrenchment, the cognitive process which shapes and strengthens these habits and principles in individuals' minds. The interaction between these components keeps rolling continuously, and feedback loops may magnify effects, and therefore, the author suggests, language is best viewed as a permanently evolving, self-organizing complex system. To drive this message home, Schmid brings in a remarkably wide range of linguistic subdisciplines and observations. While its basic claims and the model suggested relate to human language in general, examples and applications are adduced from English exclusively (p. 11).

After an ‘Introduction’, the book is divided into four parts. The first three zoom in on the three major processes identified in the subtitle and discuss how they work in detail. Part IV then offers a synopsis and proposes the author's overall model.

The first chapter, ‘Introduction’, actually offers a coherent but concise survey of the entire theoretical framework in a nutshell. It begins with a spot-on and lucid but concise discussion and definition of the interaction between communicative activities, social conventions and individual knowledge of these conventions as core constituents of language, reframed as usage, conventionalization and entrenchment, respectively, with the goal of providing ‘a unified dynamic explanation of linguistic structure, variation, and change … in a programmatic spirit’ (p. 3). Interconnections are ingeniously visualized as a ‘Tinguely’ machine with three interrelated wheels, the motions of which illustrate the interactive feedback processes and also slightly variant realizations which keep driving the system. Associated assumptions on the nature of a linguistic system largely comply with the core ideas of the cognitive-functional, usage-based approach to language.

Part I circumscribes various components and properties of usage from a variety of perspectives. Central to employing language are contextually bound usage events which, in addition to physical activities, realize communicative goals via appropriate utterance types (chapter 2), which are commonly associated with specific linguistic forms and license concrete utterances. All of these components have the potential to become socially conventionalized and individually entrenched, thus becoming records of their own usage history. Utterance types may become indexical in speech communities, and they operate on different levels of linguistic concreteness or abstractness – the notion is clearly reminiscent of that of constructions in Construction Grammar (Hoffmann & Trousdale Reference Hoffmann and Trousdale2013). Utterance types, as defined here, are similar to but not the same as form–meaning pairings, structures, or rules, but rather ‘understood as probabilistic regularities showing onomasiological, semasiological, syntactic, cotextual, and contextual variation’ (p. 28). Modifying the notion of ‘interpersonal functions’ suggested by M. A. K. Halliday, the author defines the process of shared meaning-making in context with respect to a given utterance as ‘co-semiosis’ (chapter 3), carefully weighing evidence on its constituent components and connecting it with pragmatic processes such as speech acts, achieving perlocutions, or stance-taking. It is shown that a speaker's behavior typically encompasses a substantial proportion of interpersonal, hearer-directed behavior, since the need for successful communication calls for accommodation to the hearer's state and needs. The processing and knowledge of language forms, it is argued, is achieved by myriads of neurons in the brain firing in a coordinated fashion, and thus storing, strengthening and activating associations in cognitive processing (chapter 4). Relative frequencies and transition probabilities contribute to the activation relationship between long-term and working memory, but essentially the ‘archive’ metaphor of storing and retrieving lexical knowledge is replaced by a model of association and activation. Section 4.5 offers an extensive, to my mind brilliant account and exemplification of how syntactic parsing works in detail, and thus employs and at the same time builds pattern knowledge in processing. The parsing process is understood as ‘syntagmatic arcs’ being opened by forms (mainly function words) encountered up to any point of an utterance. Subsequently, step-by-step sequences of probabilistic expectations of words and patterns to follow build up, based on earlier tokens of utterances activated together with their established syntagmatic experience-based associations, conceptual projections and pragmatic associations. Parsing is thus seen as an accumulative dynamic and incremental rather than rule-governed process. Possible selections of future items in a processing sequence also involve paradigmatic choices among available options (with varying probabilities) and are steered by pragmatic expectations and symbolic associations or ‘conceptual projections’, relationships traditionally labelled ‘frames’ or ‘scripts’. And finally, various forces that affect usage and the unfolding of cognitive and social processes in speech are listed and discussed (chapter 5), including production circumstances, communicative efficiency, salience, politeness, solidarity and power. All of these are continuously conventionalized and entrenched, and also modulated via self-feedback loops.

Part II details the components of ‘conventionalization’, an ongoing feedback cycle through which members of a community establish and sustain shared knowledge and expectations of utterance types, which become implicit and explicit norms produced by co-semiosis, co-adaptation and social diffusion. As the author repeatedly argues, conventionalization affects and operates on the basis of six components of conformity profiles: onomasiological (knowing how communicative goals or meanings translate into specific forms), semasiological (interpreting forms in the same way), syntagmatic (combining forms in habitualized ways, i.e. sharing probabilistic pattern completion expectations), cotextual (as to properties of typical text types or styles), contextual (identifying typical settings and situations of linguistic forms) and social (indexical of groups, strata and other social dimensions). These conventionalizations are multidimensional, gradient and continuously in flux, ‘collective habits’ (p. 87) which license specific utterances in given communicative situations. At the same time, they allow for more or less salient innovations, altered replications of established utterance types which are only partially licensed. Chapter 7 describes multiple facets of conventionalization in general; the two subsequent chapters zoom in on its two main subprocesses, usualization (chapter 8) and diffusion (chapter 9). ‘Usualization operates over recurrent co-semiotic activities … and continually sustains and adapts conventionalized utterance types’ (p. 124), including the integration of innovations on various levels. It contributes to variation through building in analogy effects and social indexicality (which explain and are associated with variability), and also to change and linguistic persistence. Diffusion describes the principles behind social conformity in and across communities, especially the spread and integration of innovations in social space. It continuously interacts with usualization and the dimensions of conformity, affecting and modifying these shared habits of linguistic behavior and interpretation through feedback loops. The chapter aligns traditional concepts of sociolinguistics with the framework proposed in the book in a convincing fashion. Diffusion is covered in its spatial (e.g. wave, cascade, or globalization), social (S-curve, types of adopters) and stylistic (e.g. colloquialization) dimensions, always emphasizing the interaction between these dimensions and other features of conventionalization.

Part III moves the focus from collective shared habits to how linguistic knowledge is established in the individual via entrenchment. It starts out by elucidating the nature of (largely statistical) learning, interactions between components of memory, and automatization. The cognitive processes underlying it are defined as ‘activation, association, routinization, and schematization’ (p. 205). The notion of ‘attractors’, familiar from chaos theory (which is not mentioned), is integrated, as frequently recurring words and patterns which are activated effortlessly and are being employed or approximated. Repetition and frequency support entrenchment but are not the prime predictors; they interact with a variety of other factors, including saliency, limits of the working memory, iconicity, self-priming, support through embodiment and more. At the heart of the process stands routinization, detailed in chapters 12 through 14 with respect to syntagmatic, symbolic and pragmatic associations, respectively, which it evokes and strengthens. It is seen as the strengthening of such cognitive association network patterns through active repetition, and as inviting generalization and schematization through the associative recognition of similarities. As to syntagmatic associations and expectations, various examples of increasingly chunked sequences in corpora illustrate how the process of constructions getting increasingly fixed is facilitated in practice. A core question, asked at various levels, is whether entities are cognitively processed analytically or holistically – the answers vary and often come down to the fact that it depends on lexical idiosyncrasies, as in the case of collocations. The insight that entrenched associations vary individually is convincingly driven home by the case study of variable lexical preferences in the that's Adj construction (pp. 245–58). Symbolic associations are shown to be either semasiological, choosing an adequate interpretation of an item (e.g. in instances of polysemy or homonymy, activating a prototypical or context-specific meaning), or onomasiological, i.e. selecting the most suitable form to express an idea (which leads to base-level prototype effects – so we normally refer to an animal near us as a dog rather than a Pekinese or a mammal). Interestingly enough, full recognition is also given to the build-up of context-dependent pragmatic associations, based on recurrent properties of and similarities between situations we tend to find ourselves in. It is argued that and convincingly shown how various communalities of situational context (such as participants and settings, communicative intentions, invited inferences or activity types) tend to manifest themselves and get entrenched as typical grammatical realizations (such as deixis, reference, tense, information structure, intonation, formulaic illocution signals, inferencing mechanisms, or register awareness). Chapter 15, a summary of part III, shows how four types of associations (paradigmatic, syntagmatic, symbolic, pragmatic) cooperate and compete in the process of routinization, and at the end a single, ingenious and highly suggestive cross-tabulation (table 15.1, pp. 291–2) connects specific linguistic phenomena, from phonemes through lexical types to grammatical phenomena such as pronouns or word order, with the kinds of cognitive associations which centrally help to establish and maintain them.

Part IV then presents a synthesis which pulls all earlier strings together, labeled the ‘Entrenchment-and-Conventionalization’ (EC) model. The framework is explicitly posited to constitute a ‘dynamic complex-adaptive system’, thus associating it with an emerging line of thinking which is established in the (social) sciences and growing in linguistics (Kretzschmar Reference Kretzschmar2015; Schneider Reference Schneider, Mauranen and Vetchinnikova2020). Chapter 16 presents a maximally concise summary of the components, processes and forces of the model, on just four pages and in four tables. The following three chapters survey the effects and predictions of the model on language dynamics, specifically on persistence and variability (both dealt with rather concisely) as well as change (looking into triggers, processes and forces, and proposing a novel classification of nine ‘modular’ realization types of language change). In conclusion, chapter 20 highlights what is distinctive about the EC model as compared to existing functionalist and usage-based lines of thinking, and projects future applications and further developments. It concludes with a concise (but nevertheless complex) definition of ‘the linguistic system’ as ‘a multidimensional dynamic contingency space populated by multidimensionally competing co-semiotic potentialities afforded by the interaction of speakers' usage activities and social and cognitive processes under the influence of a wide range of forces’ (p. 348, italics in original).

Throughout the book it is the author's strategy to break complex concepts and frameworks down into their component parts and to describe their properties and variant realizations. This is very helpful since the conceptual frames strung together in this volume are incredibly rich and multifaceted, and on the whole rather abstract. The presentation of material tends to be immensely dense. For example, in a ‘survey of communicative goals’ a wide range of thoughts and frameworks from functional linguistics, discourse analysis and pragmatics, linked with text types and specific grammatical structures, are covered and combined in a single one-page table (p. 18). This book is not an easy read. It is full of precisely phrased theoretical discussions and explications, repeatedly with data-based exemplifications (often drawn from the British National Corpus), some of which are rather complex in themselves. However, it is also richly seasoned with many small-scale illustrative, even entertaining examples (e.g. the same idea phrased with superordinate-level, base-level or subordinate-level terms, p. 264; or a linguist's and a journalist's varying routinized pragmatic associations with the phrase the role of the media, p. 281). One really valuable feature, then, is how the reader is guided throughout the volume – the introduction and chapter beginnings offer previews of what is coming up, and most chapters have clear and concise summaries that string things together and repeat core insights.

Not surprisingly, given the rich diversity of topics and approaches covered in this book, there are plenty of spots and sections where a reader will be tempted to ask follow-up questions, possibly challenge some assumptions or ask for further details. Here are a few examples.

• • Section 2.3 (and figure 2.1) survey various ‘utterance types’ of various levels of concreteness/abstractness – and I immediately wondered what their relationship to the notion of constructions in Construction Grammar is, since there are obvious similarities and possible differences. The answer is given soon (in section 2.3.4), though of course it may invite further questions. The author calls Construction Grammar ‘a major source of inspiration’ but decides to steer clear of it because, in his view, it conflates social conventionalization and individual mental storage, and may suggest too direct one-to-one mappings of form and meaning (pp. 27–8). Another related issue, not addressed, is how ‘utterance types’ relate to conventional levels of language organization, since they span the entire range from sounds through lexemes and phraseological units to abstract syntactic patterns. The relationship between the framework advocated here and the (by now) established notion of ‘constructions’ deserves closer reconsideration. Schmid suggests that rather than expecting ‘ready-made’ constructions, right-pointing ‘syntagmatic signposts’ (such as conjunctions, determiners or wh-words) open up expectations for specific ‘syntagmatic arcs’ (e.g. clauses or noun phrases) to unfold incrementally (p. 252).

• • One remarkable property of the framework developed here is how it seamlessly combines structural with social aspects of usage. For example, of the ‘six dimensions of conventionality … : the semasiological, onomasiological, and syntagmatic dimensions are mainly involved in the co-semiotic activities of sense-making, while the cotextual, contextual and community-related dimension provide the situational, communicative, and social background’ (p. 123). Section 7.6, on innovation, and section 7.7, on forces affecting conventionalization, integrate all major concepts of sociolinguistics (such as class, networks, prestige, mobility, multilingualism, contact and more), and at the same time throw a fresh light on them.

• • The text features a large number of abstract noun derivations – some established, some freshly coined. For example, the four subprocesses of usualization are said to be symbolization, paradigmaticalization, syntagmaticalization and contextualization (p. 124), and similar formatives prevail throughout. All notions are extensively explained, demanding as they may be. They compress complex processes and constellations into holistic concepts – this feature is indicative of the density of the text.

• • In various respects the dynamic perspective adopted throughout throws a fresh light on innovations in language – where they come from and how they become established (‘usualized’; e.g. section 8.2 and chapter 19).

• • An interesting point is the claim that abstract patterns cannot be borrowed as such – specific lexically tied utterances are borrowed, and abstractions of patterns emerge from there via analogy (pp. 135–6).

• • It is suggested that Grice's ‘cooperative principle’ and his maxims and implicatures could be understood as ‘routinized pragmatic associations’ (section 14.3) – an interesting idea that deserves closer scrutiny.

• • After decades of investigations of language change in linguistics, Schmid's proposal of nine basic types (‘modules’) of change, illustrated by many examples and helpfully summarized in a two-page table (19.1), introduces a fresh perspective, worth pursuing and testing. While the classification details may be open to discussion and refinement, the basic approach of combining sources, characters and systemic roles of innovations and the nature and speed of their diffusion in cognitive and sociopragmatic space and in individuals is innovative and powerful.

• • Clearly, there is an enormous amount of food for thought, topics for further theoretical inquiry and practical applications, built into this entire text, and perhaps chapter 20 in particular. For instance, what exactly does it mean to ‘claim that entrenchment operates in the associative network rather than nodes or units such as constructions or schemas’ (pp. 342–3), and how is this to be tested empirically? While this book does attempt to present a comprehensive and coherent framework of thought, in many ways it can also be seen as a starting point for further developments in linguistics.

An impressive breadth of knowledge is woven into this text, from all kinds of linguistic subdisciplines to cognitive processing and neurobiology. Obviously, Schmid's book and synopsis build upon decades of research in cognitive and usage-based linguistics, Construction Grammar, language variation and change, and many more approaches, and integrates much personal research, experience and rethinking. To my mind, the most useful previous résumé of this line of thinking was Bybee (Reference Bybee2010), also a marvelous book; but the present one goes substantially beyond in its scope, comprehensiveness, depth and versatility, in particular by pinpointing, detailing and connecting the productive processes that shape individual and communal ‘grammar’ and thus a ‘linguistic system’, including its variability and change. The cognitive and usage-based paradigm is substantially enriched by integrating psychological foundations as well as the pragmatic and social dimensions of language more fundamentally and systematically than before. This is a ground-breaking magnum opus and a cornerstone of future linguistic thinking and theorizing.