Construction Grammar has many features of a theory of language use. However, a shortcoming of Construction Grammar has been that pragmatics and the communicative context have been neglected both synchronically and diachronically. With this in mind, Elizabeth C. Traugott undertook to write Discourse Structuring Markers in English as a contribution to a new field of historical constructionalist pragmatics incorporating pragmatics and the context into historical linguistics.

The book is organized in three parts: ‘Foundations’ (chapters 1–5), ‘Case studies’ (chapters 6–10) and ‘Open issues for historical Construction Grammar’ (chapters 11–14). Chapter 1 (pp. 1–18) is the introduction where the goals of writing the book are outlined. The author's primary goal is to ‘infuse more pragmatics into constructional thinking’ (p. 3). She wants to do this by studying the history of Discourse Structuring Markers (DSMs) such as after all, but, by the way, in addition, instead and now. Another goal is to discuss and refine aspects of Diachronic Construction Grammar in the light of the empirical analysis of DSMs. The data for the analysis are drawn from different historical and present-day corpora. However, the methodology is mainly qualitative, implying that the history of DSMs is ‘outlined in broad strokes only’ (p. 14).

Chapter 2 (pp. 21–32) surveys some assumptions concerning Construction Grammar, especially in the model of Construction Grammar articulated in cognitive linguistics, e.g. by Langacker (Reference Langacker1987) and Goldberg (e.g. Goldberg Reference Goldberg2006). Among the tenets of Cognitive Grammar are that grammar consists of constructions [Form]<–>[Meaning] pairings. Constructions with phonological form such as by the way are called substantive. They can be conceptualized as micro-constructions that are abstractions away from the utterance (‘constructs’) and stored in a constructional lexicon (‘a construction’). A distinction is also made between substantive constructions and schemas which may be partly filled (p. 25). Construction Grammar also recognizes ‘assemblies of constructions or discourse uses’ and the role of constructions in the constructionalization process: ‘When speakers produce an utterance, constructions stored in the constructicon are assembled’ (p. 24). Croft's (Reference Croft2001) constructional model of the link between form and function, which will be used for several of the historical analyses in the book, is presented. The model consists of a form and a meaning component with a symbolic link between the two parts (p. 26). A distinction is made between semantic properties having to do with referential properties and pragmatic properties involving procedural meaning. Procedurals are assumed to point ‘hearers to particular – more or less – schematic frames of interpretation for the utterances hosting such expressions’ (Hansen Reference Hansen and Schmid2012: 595). In keeping with recent thinking in Construction Grammar, it is assumed that there is no absolute divide between semantics and pragmatics. The traditional view of this division assumes that semantics is [+truth-conditional] and [+conventional]. However, on a view that there is overlap between semantics and pragmatics, some expressions ‘can be characterized as both non-truth-conditional and truth-conditional, both conventionalized and non-conventionalized’ (p. 30).

Chapter 3 (pp. 33–58) is a key chapter where constructions are considered from a historical perspective and aspects of Diachronic Construction Grammar are outlined. Several important theoretical issues concerning language change are discussed. Specifically, Traugott adopts the perspective that what changes is usage and not grammar. Moreover, change is not the same as innovation but involves conventionalization of innovations. It is also argued that morphosyntactic changes are mostly gradual, i.e. there is always variation for some time (p. 36). The mechanisms underlying change are said to be neoanalysis (aka reanalysis), analogy, borrowing and frequency. On the other hand, pragmatic inferencing and (inter)subjectification are not considered mechanisms of change but are ‘default concomitants of constructionalization as procedurals’ in diachronic constructionalist terms and in connection with the rise of DSMs (p. 38). The kinds of contexts where a new utterance meaning occurs are argued to be replicated contexts (p. 41).

Importantly, the distinction between constructionalization and constructional change originally proposed by Traugott & Trousdale (Reference Traugott and Trousdale2013) is updated and redefined responding to some criticisms made of the distinction (p. 46). Constructionalization is now ‘the establishment of a new symbolic link between form and meaning which has been replicated across a network of language users, and which involves an addition to the constructicon’ (p. 49). Over time new types of expressions may give rise to ‘micro-constructions’, i.e. ‘form–meaning pairings that generalize over accumulated constructs’ (p. 48). As will be shown in the empirical discussion of DSMs, the distinction between two types of change is helpful in identifying when and how a construction is conventionalized (becomes part of a subset of language users’ knowledge of language). Constructionalizations are now seen as the outcome of small changes referred to as pre-constructionalization changes (p. 50). These changes are tendencies and can be exemplified by some loss of compositionality of a construction undergoing change. There are also post-constructionalization changes that can be described as tendencies, for example changes in frequency (p. 51).

Constructional changes are largely contextual changes. The constructions that are relevant are ‘linguistic, relevant and conventional’ (p. 55). According to Traugott, the types of context that are important are the surrounding context (co-text) (p. 55) and discourse contexts consisting of ‘assemblies of discourse uses or constructions’ (p. 56). For scholars who are not familiar with Construction Grammar, chapters 2 and 3 serve as an excellent introduction to Cognitive Linguistics and Construction Grammar and, in particular, to cognitivist views of change.

Chapter 4 (pp. 59–84) is an introduction to what is usually meant by Pragmatic Markers, including Discourse Markers. A distinction is made between lexical connectors that are contentful and DSMs that are only partly contentful. DSMs are further characterized as either monofunctional (referred to as 1DSM) or as multifunctional (referred to as Discourse Marker, DM) depending on their position on a cline of pragmaticality ranging from largely contentful to primarily pragmatic meanings. A model of a construction (based on Croft Reference Croft2001) is presented showing the default properties of DSMs (p. 65). Syntactically, they are Conjuncts, and the discourse-functional property is ‘connectivity stance’ (such as elaboration or contrast). The pragmatic properties are subjectivity and intersubjectivity.

DSMs are Connectors signalling a connective relationship between discourse segments. The kinds of DSMs discussed are Elaborators, Contrastives, Digressives and Markers of Return to a prior topic. DSMs are assumed to arise historically from Circumstance adverbials. A Discourse Structuring Marker Trajectory is proposed outlining the developments from Circumstance adverbial to a Conjunct adverbial to DSM (p. 73). The account assumes that the development of Conjuncts out of Circumstance adverbials is a case of constructionalization as both meaning and form change. The development of a DM, on the other hand, is a constructional change since it involves an increase of functions. A preliminary case study of the changes of the inferential marker after all is presented using Croft's (Reference Croft2001) constructional model. In this chapter the distinction between lexical connectors, DSM and DM is particularly interesting.

Chapter 5 (pp. 85–100) compares three proposals about the rise of DMs, namely grammaticalization, pragmaticalization and grammaticalization accompanying cooptation. The development of DMs is consistent with some processes of grammaticalization such as attrition and bondedness. However, there are assumed to be ‘too many aspects of the developments of DSMs that are not consistent with grammaticalization as it is generally understood for an account in terms of that theory to be adequate’ (p. 90). Proponents of the pragmaticalization hypothesis generally treat pragmaticalization as a process distinct from grammaticalization. According to Traugott (and many other scholars), the choice between grammaticalization and pragmaticalization ‘seems to hinge not on the process itself but on what is encompassed by “grammar”’ (p. 91).

The hypothesis of ‘cooptation to thetical grammar’ has been proposed more recently by Heine et al. (e.g. Heine et al. Reference Heine, Kaltenböck, Kuteva and Long2021). Heine et al. (Reference Heine, Kaltenböck, Kuteva and Long2021) propose that a broad conception of grammar is needed in the framework of Discourse Grammar that distinguishes between sentence grammar and thetical grammar. Sentence grammar is restricted to the syntax of internal constituent structure. Thetical grammar concerns ‘external’ phenomena, including DSMs. Formulaic theticals, such as DMs, are regarded as the result of ‘cooptation’ from sentence grammar whereby a chunk of sentence grammar is deployed for use on the metatextual level thereby turning into a thetical. According to Traugott, the problem with the thetical hypothesis is that there is no evidence of instantaneous cooptation. Specifically formulaic theticals take time to arise and be conventionalized. Another argument is that initial position is precisely the kind of context in which changes can be expected to occur gradually. The rise and change of DMs has given rise to a great deal of controversy and the debate is likely to continue. The account given in grammaticalization theory has, for instance, been called into question by many researchers. It is clear that we need much more empirical research on the changes of many different types of pragmatic markers.

Part II investigates elaborative markers, contrastive markers, markers of ‘digressive’ topic shift and markers of Return to a prior topic. Chapters 6–9 provide evidence in support of the Diachronic Construction Grammar position.

The focus of chapter 6 (pp. 103–20) is on how markers of elaboration come into existence and on the methodological issue of how to determine the status of an expression as a Circumstance adverbial or Conjunct on the form side or as a multifunctional DM on the meaning side. It is shown that an elaboration subschema existed already in Old English and that the subschema in Standard English now licenses a large number of expressions ranging from lexical Connectors like in addition to monofunctional DSMs like further and multifunctional DMs like and. Specifically, it is shown how the constructionalization phase of the DSM Trajectory Hypothesis (CircAdv>Conjunct) is borne out in the history of all expressions having an elaborative meaning, but that only also underwent the further development into a multifunctional DM through constructional changes (p. 119).

The topic of chapter 7 (pp. 121–38) is the contrastive Connectors but, instead, all the same and their histories. But is a multifunctional DM in contemporary use, unlike instead and all the same, which are monofunctional DSMs. But illustrates a transition from a preposition to a Conjunct, which is an example of constructionalization since it involved a change of form while the further shift to a multifunctional DM is a constructional change. The development of all the same draws attention to the fact that the origin of the Connector could be an adjective phrase. All three contrastive markers support the hypothesis that contentful phrases can be reinterpreted as DSMs. The development of instead into a Conjunct is said to illustrate the pattern in English of preposition>adverbial.

Chapter 8 (pp. 139–54) deals with the development of the digressive markers by the way, by the by, incidentally and parenthetically. As a Circumstance adverbial by the way may be used dynamically in a context called ‘talking en route’. An important factor accounting for the development of the DSM use would appear to be a new metaphorical use where way is understood as a textual journey not literally as ‘road’ and by the way means ‘in the course of discussion, ‘in passing’ (p. 141). In many cases speakers used by the way meaning ‘in passing’ to refer to something as ‘falsely casual’. Over time, by the way referring to something as casual also came to acquire meanings as a mitigating hedge (p. 144). By the by is mainly a variant of by the way. It is shown that incidentally (but not parenthetically) can be used both to introduce a new topic and as a hedge and is therefore a multifunctional DM but only marginally so. Some formulae such as I almost forgot, I just remembered are argued to be lexical connectors that have little to do with digression (p. 153).

Chapter 9 (pp. 155–64) elaborates on the idea of lexical Connectors characterized by syntactic structure and compositional meaning. The lexical expressions discussed mark Return to a topic, as in to return to x point, back to x point and back to x topic. The examples of to return to x point show that it is adverbial, contentful and compositional and that it has not conventionalized as a Connector. On the other hand, back to x point is coming to be used in exploratory ways with different pronouns that could lead to a gradual shift from lexical, contentful meanings to a Conjunct with Discourse Structuring meaning over the next few decades (p. 160).

Another topic that has been discussed recently is what constraints there are on the linear ordering of Pragmatic Markers. The focus in chapter 10 (pp. 165–88) is on the distinction between combinations of independent DMs and combinations that have been fused resulting in a new construction. The question is how combinations with also, the combination now then and the rise of the combination oh, by the way have risen. Conclusions from the study of combinations with also are that only certain combinations are possible and that combinations that come to be chunked are understood differently from their parts (e.g. so also). The study of the topic-orientation unit now then shows that ‘when combinations of individual micro-constructions come to be used as a unit the resulting meaning is not the same as that of the micro-constructions used separately’ (p. 179). The change is an example of constructionalization because form has changed and frequency of use has increased. Oh, by the way is shown to have two separate uses. In the first use oh, by the way appears in a context where a hedge can be intended. In the second use, it is a unit DM with a mocking function. The study of the development of DMs is a neglected area in diachronic linguistics. The data investigated by Traugott suggest that we also need studies of the combinations of DSMs and DMs and their frequencies.

The chapters in part III discuss ‘three open theoretical issues’ for a historical constructionalist perspective on pragmatics. Chapter 11 (pp. 191–202) discusses subjectification and intersubjectification in the context of constructionalization and the rise of DMs. The claim is that subjectification and intersubjectification are not factors that can lead to change but they are conventionalized aspects of ‘communicative function’ which should be considered to be among the meaning properties of a construction. It is hypothesized that when a speaker uses a Circumstance adverbial as a Conjunct or a DSM it will by default be used in ways which are more subjective and intersubjective. By default, DSMs will also be used in a way that is more textual than their source because they are connectors that link two discourse segments (D1 and D2). A third process is textualization, defined as ‘increase in the degree to which SP/Ws pay overt attention to text-creation and invite AD/R to interpret textual relationships’ (p. 197). It serves a separate communicative function and is not positioned in a directional trajectory (as had been suggested by Traugott in earlier work). By hypothesis, textualization, subjectification and intersubjectification occur simultaneously when a new Conjunct to DSM is innovated. A generalization can then be made to the effect that ‘as innovated uses came to be conventionalized, the textuality and (inter)subjectivity come to be communicative functional properties of the constructions with which they are associated’ (p. 198). The model is illustrated by the development of Digressives (section 11.4.2). From the perspective of constructionalization, the development of DSMs is argued to be a case of textualization, weak subjectification and intersubjectification. According to Traugott, these occur along with constructionalization and may be followed in individual cases by further subjectification and intersubjectification (p. 200).

Chapter 12 (pp. 203–24) discusses clausal positions of DMs. The starting point is that only multifunctional DMs are likely to have different pragmatic and discourse effects in different positions. The question is now whether the slot in which a DM may be used relative to a host clause is a construction. Case studies of the two DMs after all and by the way were carried out with focus on their clausal positions and functional differences in the different positions. In order to examine the correlations between the position and the meaning of the DMs examples were selected from the Corpus of Historical American English (COHA). The results of the study indicate that ‘[w]hether or not a Connector can be used with a linking meaning in several positions and what its discourse function is in those positions depends on the individual Connector’ (p. 223).

A key concept in construction grammar is that language is a network of constructions. Chapter 13 (pp. 225–38) is concerned with the question of how such networks should be conceptualized. Two different types of synchronic networks are introduced: ‘vertical’ inheritance networks (describing inheritance relationships between constructions) and ‘horizontal’ networks (describing relationships between constructions that are similar). This is followed by a discussion of the role of networks and change. The links in the vertical network can, for instance, be strengthened as new members of the taxonomy are added. Constructionalization can also imply that one link is broken and a new one is established. An attempt is made to show that we can represent the role of the discourse context in changes involving the uses of by the way in terms of network connections and the discourse context.

Chapter 14 (pp. 239–44) summarizes the main points of the book and suggests some areas for further research. Since the field is still a new one and there are several challenges for the future, one such issue mentioned by Traugott has to do with how well the constructional model can be used for languages other than English.

This book is an important and thought-provoking contribution to historical constructional grammar combining the discussion of theoretical issues with the empirical analysis of a specific type of discourse markers. It is convincingly shown that pragmatics and discourse factors need to be incorporated into constructional models accounting for the rise of monofunctional DSMs (and multifunctional DMs) in English. Specifically, the network formalism makes it possible to describe the pragmatic and discourse-functional constraints on the changes undergone by the DSMs in a systematic way. The view of language change in this book is supported by the analysis of a number of case studies of different types of DSMs. These studies are particularly revealing because they are based on corpora and corpus-linguistic methods which are likely to have a key role in future developments in constructionalist historical pragmatics.