Review of Elizabeth C. Traugott. 2022. Discourse Structuring Markers in English. Amsterdam: John Benjamins, xviii+274 pp. ISBN 978-90-272-1091-3

Full Text

Discourse Structuring Markers in English (Traugott 2022) is about the diachrony of pragmatic markers, in particular a subtype which Elizabeth Traugott identifies with the syntactic form often referred to as ‘Conjuncts’. These connectives are derived from circumstantial adverbials (CircAdvs) and some are further recruited as discourse structuring markers (DSMs). What makes this monograph a valuable read is its attempt to tackle some thorny questions about language change. Among these tough questions are: (1) Is pragmaticization a separate phenomenon from grammaticalization? (2) If pragmatic markers emerge as a result of reanalysis (or ‘neoanalysis’), is the process gradual or instantaneous? (3) If new functional categories also emerge via analogy, what kind of language change model can insightfully capture this type of process? In trying to address these challenging questions, Traugott turns to Construction Grammar, offers a diachronic perspective to how new functional categories are constructionalized, and suggests how a network model can account for the interlocking patterns in language change.

The monograph starts off with an Introduction that provides an overview of the book, including the goals and methodology, and in particular clearly delimiting the type of pragmatic markers being analyzed to Conjuncts derived from CircAdvs. The choice of terminology appears to be deliberate. A Conjunct, as Traugott notes, can be more than just a connective, since it also often expresses the speaker’s stance in the way it conjoins discourse segments. For example, conclusive adverbial after all has over time also come to be used as a contrastive connective (e.g. I can’t imagine anyone disliking Nala. After all, who can detest a cat named after a character from The Lion King?¹).

Following the Introduction, the monograph is then divided into three major parts. In Part 1, Traugott focuses on foundational issues in language change from a cognitive linguistic and construction grammar perspective. Chapter 2 pays special attention to Goldberg’s model of construction grammar. This helps familiarize the reader with basic terms such as micro-constructions and (sub)schemas, which are used in describing functional categories and which are useful in visualizing—metaphorically—the changes in semantic, syntactic and prosodic properties of each construction as they are put to use in different discourse contexts. Special attention is also given to Croft’s construction model, which defines constructions as form-meaning pairings that are, paradoxically but wonderfully, ‘holistic yet analyzable’. Traugott explains the beauty of this view of constructions as follows. A construction with the form all the same can be used holistically with the meaning ‘in exactly the same way’, but this form-meaning pairing can also be used to mean ‘in contrast’ to express disagreement, in which case the form remains the same but the meaning changes, especially in a pragmatic way. As Traugott points out, “If a construction were truly unanalyzable, change could not occur” (p. 27). With both Goldberg’s and Croft’s models of grammar in tow, Traugott proceeded with her analysis of what happens at the semantics-pragmatics interface, which is crucial to our understanding of how pragmatic markers emerge.

In Chapter 3, Traugott focuses on the tenets that are crucial when we adopt a diachronic approach to Construction Grammar. She makes a distinction between ‘usage changes’ and ‘grammar changes’, highlighting that what changes is the way speakers use a construction and this then results in shifts in the constructional networks within the language. Traugott also distinguishes between ‘innovation’ and ‘change’, recognizing that innovation can begin with the individual (or with the group when teamwork is involved, we think) and the process could be abrupt, while change often takes time before a new functional category becomes conventionalized and entrenched within a speech community. Traugott also maintains that change often involves semantic shifts in the form of metonymic mappings and, as such, the trajectory of a construction from primarily contentful meaning to increasingly pragmatic meaning is a continuous rather than discrete trajectory. For this reason, Traugott argues that pragmaticization is but an extension of grammaticalization, and not a separate phenomenon. This semantic shift from more contentful and monofunctional meaning to more pragmatic and multifunctional meaning relies on more than just metonymic mapping. What also motivates this drift toward increasing pragmaticization is the process of (inter)subjectification, whereby there is greater orientation toward the speaker’s or/and the addressee’s epistemic and emotive stance. Consistent with the view that language change is gradient rather than abrupt, Traugott maintains that degree of (inter)subjectification is also gradient in nature. Interestingly, despite the emphasis on continuity, Traugott makes a clear distinction between ‘constructionalization’ and ‘constructional changes’ (the latter now renamed as ‘constructional shifts’ in Traugott (2023)). As defined in Traugott (2022), ‘constructionalization’ refers to the stage when a new functional category is recognized by a speech community, while ‘constructional changes’ (or ‘constructional shifts’) refer to additional uses (including pragmatic ones) that do not significantly alter the functional category. For example, the construction after all can have more than one pragmatic reading, among them a pre-clausal DM use where the subsequent discourse segment (D2) justifies the claim in the prior discourse segment (D1) (e.g., I’m not going. After all, who is she to tell me what to do?) and a post-clausal DM use with the meaning ‘in contrast to speaker’s expectation’ (e.g. She’s not coming, after all). Worth noting is that a different DM use does not change the DM category of the construction.
In other words, differences in pragmatic meaning do not arise from ‘constructionalization’; they are the result of ‘constructional shifts’. Constructionalization, on the other hand, involves category change, for example, when circumstantial adverbs (CircAdvs) are neoanalyzed as Conjuncts. In principle, constructional shifts can precede as well as follows constructionalization. This, as we will discuss later, has interesting parallels with yet another theoretical model of language change, namely, Discourse Grammar.

Chapter 4 first reviews previous studies on pragmatic markers (PMs), giving readers a glimpse of the various definitions of PMs in extant literature, and clarifying how the terms pragmatic markers (PMs) and discourse markers (DMs) are used in the monograph. In a nutshell, the term pragmatic marker is an umbrella term for a broad category of stance markers, which include discourse structuring markers (DSMs), and the latter in turn include discourse markers (DMs). This chapter then goes on to elaborate how circumstantial adverbs (CircAdvs) that are constructionalized as Conjuncts come to serve as discourse structuring markers (DSMs). Recall that the term ‘Conjunct’ refers to the syntactic form, while the term ‘Discourse Structuring Marker’ refers to the pragmatic function(s). Traugott notes that, in the case of Conjunct-DSM form-function pairings, the Conjunct form can remain the same (e.g., after all, by the way) or become fused (e.g., morever) or become phonologically reduced (e.g., eal swa ‘exactly like this’ > also), while the DSM function(s) can vary from being largely limited to expressing more contentful, monofunctional meaning (e.g., further, furthermore, moreover) to being further recruited to express highly pragmatic, multifunctional meanings (e.g., after all, by the way). Traugott adopts the term ‘1DSM’ for the more contentful and monofunctional DSMs and reserves the term ‘DM’ for the more highly pragmatic and multifunctional DSMs. Broadly speaking, DSMs = 1DSMs + DMs, with the distinction between the two subcategories being gradient rather than discrete. In other words, the DSM category comprises a wide array of Conjuncts with pragmatic functions that vary along a cline from monofunctional 1DSMs to multifunctional DMs. To further illustrate from a diachronic perspective, when a circumstantial adverb (CircAdv) such as temporal adverbial after all (this) is neoanalyzed (hence ‘constructionalized’) as conclusive adverbial after all (Conjunct), it begins to serve not only connectively (textuality) but also affectively (pragmatically) as a discourse structuring marker (DSM), since the speaker’s (inter)subjective stance is manifested in the way the speaker chooses to conjoin discourse segments D1 and D2. Some DSMs (e.g., finally) remain largely monofunctional (1DSMs), while some (e.g., after all) acquire multiple pragmatic uses as their domains of use are extended to different discourse contexts. The Conjunct after all, for instance, not only expresses connectivity but have also come to be further deployed as a contrastive DM or a justificational DM, depending on the context. Traugott refers to these extended uses beyond the constructionalization stage as ‘constructional changes’ (or ‘constructional shifts’). In short, contentful, compositional/analyzable and monofunctional after all becomes increasingly pragmatic, non-compositional/unanalyzable and multifunctional as its range of use extends along a cline from 1DSM to multiple DMs.

Chapter 5 gets exciting as it reviews alternative diachronic models of language change, with special attention to differences between Discourse Grammar (DG) and Diachronic Construction Grammar (DCG). Discourse Grammar, proposed by Bernd Heine and colleagues, posits a dual-layered grammar comprising a sentence-level grammar and a thetical-level grammar. The latter is influenced by discourse factors and operates beyond the sentence level. The transition from sentence grammar to thetical grammar is realized through ‘cooptation’, a process whereby the propositional meaning of a construction at the sentence level shifts to a pragmatic meaning at the discourse level. Structurally, the construction that contributes pragmatic meaning becomes syntactically detached from the sentence, hence its emergence as a ‘thetical’ (a category which includes not only clause-medial parentheticals, but also detached pragmatic markers found at the left and right peripheries). Initially, there were frequent references to cooptation being ‘instantaneous’, i.e. an abrupt shift. Not surprisingly, given that Diachronic Construction Grammar views the rise of pragmatic markers as a gradual process, Chapter 5 of this monograph takes issue with the characterization of cooptation as an instantaneous process.

Part II, then, is devoted to case studies that meticulously trace how various types of Conjuncts come to be deployed as discourse structuring markers (DSMs), many of which are used as multifunctional discourse markers (DMs). Chapter 6 is devoted to the development of elaboratives such as also, further(more) and moreover; Chapter 7 to contrastive markers such as but, all the same and instead; Chapter 8 to markers of ‘digressive’ topic shift such as by the way, incidentally and parenthetically; Chapter 9 to markers of ‘return to a prior topic’ such as back to point X; finally, Chapter 10 to the development of combinations of DMs such as and also, so also, now then, and Oh, by the way. The detailed diachronic analyses in these chapters in Part II pay special attention to syntactic positions. These not only include clause-initial, clause-medial and clause-final positions, but also pre-clausal and post-clausal positions. Recognition of a distinction between pre-clausal and clause-initial positions is theoretically important; likewise, recognition of a distinction between post-clausal and clause-final positions. In this respect, Diachronic Construction Grammar may share some common ground with Discourse Grammar, since there are some parallels between pre-/post-clausal DMs and theticals, particularly in terms of detachment from the clause and semantic-syntactic scope expansion. As Traugott points out, discourse structuring markers (DSMs) in post-clausal position tend to have scope over the preceding discourse, while those in pre-clausal position tend to have scope over the following discourse (2022: 62, 88, 99). On this point, then, both analytical frameworks embrace syntactic scope expansion, each in their own way going beyond traditional views which had initially characterized grammaticalization in terms of syntactic scope reduction (2022: 86).

Convergent findings can also be found in other subfields in linguistic studies. Consider, for example, prosodic studies in Sinitic languages, which are well-known for their sentence final particles (SFPs), many of which are often stacked together (not unlike combinations of DMs in English noted in Chapter 10 mentioned above). Pauses in prosodic analyses from Cantonese, for instance, reveal detachments of topics at the left periphery and detachments of sentence final mood particles at the right periphery (Wee 2024, see also Wakefield 2010). These observations are also consistent with formal syntactic accounts, which identify SFPs with broader scope syntactically (given that SFPs c-command their immediate dominating clause, TP) and pragmatically (since SFPs determine the construction type and much of the pragmatic meaning of the sentence, e.g., question, doubt, surprise, assertion, etc.) (Wee 2024: 2, see also Simpson 2014, Cheng & Tang 2022). In sum, a diachronic view that links the rise of discourse markers (DMs) with semantic and syntactic scope expansion, as espoused in Traugott (2022), is well-supported by recent research findings.

In Part III, the last section of the monograph, Traugott initiates discussion of how we might conceptualize the architecture of a dynamic construction network that can account for the rise of circumstantial adverbials (CircAdvs) as Conjuncts, which function as DSMs (inclusive of 1DSMs and DMs). She raises three issues for open discussion as we go about constructing a pragmatic model that is faithful to diachronic evidence. In Chapter 11, Traugott emphasizes the synergistic interaction of three processes: subjectification, intersubjectification, and also ‘textualization’. The latter process involves the speaker paying attention to text-creation (i.e., the process of arranging how discourse segments should be linked to each other) and inviting the addressee to interpret the intended textual relationships (Traugott 2022: 198, 202). These three processes occur simultaneously, with textualization resulting in Conjuncts, while subjectification and intersubjectification may continue in strength to develop into DSMs and DMs. All three processes—subjectification, intersubjectification, and textualization—need to be incorporated into a network model of pragmatic language change.

In Chapter 12, Traugott addresses the question of whether clausal position determines the pragmatic meaning of connectives. Her analysis for English shows that pre-clausal position is the default position for Conjuncts (e.g., and, also, so, now), and also for many DSMs, particularly those that still primarily “link back to prior discourse and forward to upcoming discourse” (p. 208), but with some DSMs also found in other positions (e.g., first of all, by the way). DMs can be found in various positions (pre-clausal, post-clausal and also clause-medial), with multifunctional DMs “likely to have different pragmatic and discourse effects in different positions” (p. 203). Because DMs can be found in various positions, Traugott concludes that clausal position does not necessarily determine the pragmatic meaning(s) of a connective. From such a statement, one could then infer that clausal position is but one of many variables in the conceptualization of a network model of pragmatic language change.

Traugott dedicates Chapter 13 to the description of a ‘network’ model to help explain the diachrony of DSMs from a Construction Grammar perspective. This network metaphor is inspired by neural network simulations but is not a computational model. It comprises various types of links between form-function pairings, each of which is a construction with “a rich bundle of features with values” (p. 226). Different features become more prominent in different contexts, with their values changing as a construction comes to be linked to different combinations of constructions at different times. This allows a network to be dynamic and malleable. Traugott focuses on two major types of linkages, namely, ‘vertical networks’ and ‘horizontal networks’. Constructions that are linked via vertical networks yield inheritance relations, with links formed between constructions “at different levels of abstraction” (p. 227), while constructions that are linked via horizontal networks yield resemblance relations, often via analogy.

At this juncture, it is worth returning to the question of whether Traugott’s emphasis on gradualness in the rise of pragmatic markers can be reconciled with Heine’s notion of abruptness or instantaneity associated with the phase referred to as ‘cooptation’. Diachronic evidence supports Traugott’s observation that new functional categories often emerge gradually, often via overlapping metonymic mappings that take some time before a newly emerging construction is conventionalized and widely adopted by a speech community. Structural evidence can be used to show that neoanalysis (= reanalysis) has taken place, affirming that a new functional category has emerged, but it cannot resolve the question of whether such change is instantaneous or gradual. To some extent, neural network simulations can help shed some light here.

Such networks have produced scatterplot graphs showing that random inputs over time result in similar inputs being attracted to a common (i.e. shared) region in mental space. Each input is made up of patterns of activations, analogous to the rich bundle of features in Construction Grammar (the latter in fact being inspired by the former). At a more abstract level, each pattern of activation can be represented as a whole unit. Although initially occurring as random and sporadic activations, over time similar patterns of activation begin to boost each other’s prominence and visibility and together they begin to form a category that is distinctive, and in this sense become strong enough to ‘break away’ from other categories, yet without breaking off links with related categories. Such breakaways are comparable to Traugott’s notion of ‘constructionalization’. In a sense, this is also roughly equivalent to Heine’s notion of cooptation, except that constructionalization is gradual while cooptation is thought to be instantaneous. Neural networks are dynamic systems and each pattern of activation is sensitive to neighboring patterns of activation, which gives these systems their malleability, as well as the potential to form other new categories or constructionalizations. What neural networks are able to additionally capture is that change happens in poly-dimensional space. It is this complex scenario of interlocking (hence context-dependent) neural-like processes that Diachronic Construction Grammar tries to highlight.

At the same time, in neural network literature, these rich patterns of activations are sometimes transformed into graphs which provide us with an abstract 2- or 3-dimensional representation of dynamic activations which in fact occur in poly-dimensional space. One such 2-dimensional graph is the sigmoid curve, which shows a relatively flat trajectory that suddenly spikes up in a steep slope before leveling off again. This sigmoid curve helps explain, at a more abstract level, how language change often goes unnoticed in the early stages of grammaticalization, then is sometimes followed by accelerated change (indicated by the steep slope in the sigmoid curve) before the process of change decelerates and plateaus off. The sudden spike is what Heine and colleagues seem to be trying to capture, and which is compatible with what they refer to as ‘cooptation’. Heine mentions an early stage of grammaticalization prior to cooptation, and a later stage of grammaticalization following cooptation. These two stages of grammaticalization are comparable to the relatively flat trajectories before and after the sudden spike in the sigmoid curve. In a sense, then, one could say that Heine and colleagues see this sudden spike as the process that flips meaning at the level of sentence grammar to meaning at the level of discourse grammar.

In light of what we know about neural network metaphors, Heine’s Discourse Grammar (DG) can be appreciated as a model of pragmatic language change at a more abstract level of analysis, while Traugott’s Diachronic Construction Grammar (DCG) models the same phenomenon at a far more fine-grained level of analysis. The former (DG) makes room for researchers to focus more on the discursive aspects of language change, especially for those working from a more socio-interactive perspective. The higher level of abstraction can also be useful for surveying typological tendencies across languages. The latter (DCG) reminds us that pragmatic strengthening is a continuous process that often goes largely unnoticed, essentially because the process is often gradual despite the appearance at times of (near-)instantaneity when a speech community recognizes the emergence of a new functional category. Given its attention to diachronic details, Diachronic Construction Grammar can further help explain the motivations that influence language-specific variation. Both theoretical models contribute to our growing understanding of the cognitive processes involved in language change; together they expand our views of language at both more neural-like and more abstract levels of analysis.

Chapter 14 of the monograph calls for further explorations into still-unresolved issues. These include a call for new studies investigating discourse structuring markers in other languages, and also diachronic studies within a constructionist framework of pragmatic markers derived from lexical sources other than Conjuncts and CircAdvs. Overall, this 274-page monograph is a valuable addition to the literature on language change and is written with meticulous attention to diachronic data. It should find a place in our library to inspire further diachronic studies that are informed by a deeper understanding of neural network processes. In terms of style, this monograph is intended for fairly advanced language scholars with some background in grammaticalization and pragmaticization studies and interested in theoretical issues based on historical evidence. But it can also be used as a resource book for undergraduate students to trace the development of specific connectives across different chapters, giving them the opportunity to hone their extensive reading and problem-solving skills and mentoring them in diachronic linguistic analysis. In terms of content, this monograph is bound to stimulate further discussions on the processes of language change.

 

_{¹ Retrieved, with thanks, on August 14, 2024 from https://www.uis.edu/learning-hub/writing-resources/handouts/learning-hub/transition-words}

About the authors

Foong Ha Yap

Chinese University of Hong Kong

Author for correspondence.
Email: foonghayap@cuhk.edu.cn
ORCID iD: 0000-0002-7831-0522

Associate Professor at the Chinese University of Hong Kong, Shenzhen. Her research focuses on the development of pragmatic markers. She has numerous publications in peer reviewed journals including Diachronica, Discourse & Society, Journal of Historical Pragmatics, Journal of Pragmatics, Text & Talk among others.

Shenzhen, China

Mikyung Ahn

Hankuk University of Foreign Studies

Email: mkahn@hufs.ac.kr
ORCID iD: 0009-0007-1464-3276

Associate Professor at Hankuk University of Foreign Studies, where she received her PhD in Linguistics in 2009. Her research interests include language change, discourse analysis, and cognitive-functional linguistics, with a focus on the development of discourse markers and their functions from both pragmatic and typological perspectives. Her work has been published in Diachronica, Journal of Pragmatics, Lingua and Text & Talk./ among others.

Seoul, Republic of Korea

References