- 1 I used the electronic version of the book for which no stable page numbers exist. I therefore cite (...)
1In 2020, Thora Tenbrink (University of Bangor) published a synthesis on Cognitive Discourse Analysis (CODA hereafter). She defines the book as “the result of twenty years of engagement with a seemingly simple idea: we use language to express our thoughts” (Preface, §11).
2While this claim and the resulting research program might not seem new, as Tenbrink confesses herself, it remains that the question could receive much more attention from the many fields tackling the link between language, discourse and cognition, in each field separately and as a joint endeavor.
3The crucial contribution of the book and of CODA is that it transforms its focal interest in the relation between “complex cognitive processes” (passim) and language into an opportunity to bring together different relevant disciplines to delimit a quite vast scientific common ground. Tenbrink succeeds in building upon a wide variety of approaches within the cognitive science of language, discourse and reasoning to present a rather novel integrated “theory neutral” methodology. CODA is indeed not intended as a theoretical endeavor. In Tenbrink’s words:
“CODA is a method designed to reveal how speakers’ thoughts and concepts are reflected in language, both explicitly and implicitly. CODA looks at linguistic data, at spoken or written language (i.e. discourse) produced in situations relevant to thinking. Closely related to traditional discourse analysis methods, CODA means doing discourse analysis with a focus on concepts and thought processes.” (Introduction, §4).
4This core definition provides us with quite a number of insights into what to expect, and not to expect from the book. Four of them seem crucial to our reading:
51. CODA is not a theory, it is, in Tenbrink’s words, theory neutral.
62. CODA is a methodology, that is, it should provide the investigator with a toolkit to collect, organize, and analyze data.
73. Discourse constitutes the cardinal (and tutelar) notion of this methodology.
84. It relies heavily on “situations relevant to thinking”, that is, task-based situations of important cognitive load involving for instance abstract reasoning or route planning.
9The present review aims at discussing these four main points, all of them clearly explained, demonstrated, illustrated throughout the book. Tenbrink successfully presents an impressive number of notions, works, experiments and studies to back up her stance. Numerous examples usefully clarify methodological or, in fact, theoretical points, taken from studies that Tenbrink often authored or co-authored. The book constitutes a rather expert synthesis of the approaches both cited and applied in CODA, but also of Tenbrink’s previous work and that of related projects or laboratories. This alone justifies saluting the book.
10The nine chapters of the book can be divided in three sections.
11Tenbrink starts by thoroughly describing the scientific context of and for CODA (Chapter 1), the core idea of language as a representation for thought (Chapter 2), the resources available to the CODA researcher (Chapter 3) from Cognitive Linguistics to Systemic Functional Grammar and Discourse Analysis.
12Chapters 4, 5 and 6 then focus on cognitive processes and notions which build up the conceptual framework of CODA. They are logically presented in detail and precisely illustrated by works in which Tenbrink took an active part, often as co-author. Chapter 4 discusses mechanisms of cognitive orientation (attention and perspective). Chapter 5 discusses the notions of cognitive depth (granularity and certainty). Chapter 6 focuses on cognitive constructiveness (inference and transformation).
13The last three chapters focus on applications of CODA and its validity for different fields of research. Chapter 7 takes an in-depth look at communication (“Using language to convey thought”); Chapter 8 details the different methodological tools and steps of CODA. It is a key chapter since it offers a very clear and practical guide on how to do CODA. Finally, chapter 9 takes a look “Beyond CODA”, that is, at the relevance and extension of the presented methodology for the study of cognition in different disciplines, and outside the lab. It includes a conclusive section (9.3).
14I will refer the reader to Holmgreen (2020)2 for a detailed summary of the book chapter by chapter. It would be quite useless to repeat here what has been successfully done elsewhere3. Nevertheless, it is still necessary to give a more precise definition of what CODA is before presenting what constitutes the major contribution of the book, from my point of view as a cognitive linguist and discourse analyst in French academia.
15CODA collects data, usually within the lab - that is, data is usually elicited but it is unscripted and as spontaneous as can be. The researcher does not elicit words or sentences or short (and timed) responses as can be done in some psychological research, but entire sequences of more or less interactional discourse. A good example of this comes from Ericsson and Simon (1993): Verbal Reports are often used in the experimental setting. For instance, participants are asked to think aloud as they perform a cognitive task (after a short training). Think-aloud protocols are thus examined as they can offer unscripted verbalizations following a constrained train of thoughts, usually in task-based situations (building something, coming up with a route for a trip, problem-solving). Developing a discursive-analytic approach is justified by the consubstantial link between the linguistic productions and the task or situation.
16Once data is collected, it is analyzed following methods from discourse analysis quite logically, starting with segmentation and coding. Here lies the first novelty of the method: whether in the experimental set-up, data gathering, or coding protocols, findings from Cognitive Linguistics (after Langacker 1987, Talmy 2000, Evans 2009, to name a few) or cognitive psychology are systematically looked at and taken into account. In other words, Tenbrink manages to identify the numerous points of consensus between different approaches of language and cognition, successfully building and mapping this crucial common ground. Space, which is Tenbrink’s area of expertise, can only be approached after careful consideration of the state of the art in several disciplines. Neurological aspects of spatial orientation and motion, psychological experience of route design and spatial language (including metaphors) all intervene at some point in the reported experiments and subsequent data analysis. Methodological integration goes beyond data analysis: Tenbrink systematically confronts her results to findings in the approaches she built her study from. Problem-solving or visual representation are two other areas benefiting from this wide-range integrative approach. Because there is no single approach which frames the entirety of the studies, it is indeed theory neutral.
17Paradoxically, this neutrality can only be understood by looking at the three main theories which inform CODA: Cognitive Linguistics (CL), Discourse Analysis (DA) and Systemic Functional Grammar (SFG). Tenbrink recognizes that each of these fields is not as homogenous as it seems but she identifies their core stance(s). For instance, CL holds that linguistic structures, semantic and syntactic, reflect conceptual structures because language is not an autonomous cognitive faculty. From DA, she isolates the core idea that linguistic productions are situated and that situational parameters must be taken into account in studying language data, even when the study aims at shedding light on cognitive rather than social mechanisms. The idea of situation, almost an operational concept, allows CODA to go beyond words and sentences, and to tackle the problem of the scale of language from lexical units to social norms Finally, SFG brings levels of analysis: interpersonal, ideational and textual. The core insight is that speakers make contextual choices: the situation filters through the network of options that speakers have (see 3.3.1, §6). Even if Tenbrink concedes that SFG has “a lot to offer to CODA from a theoretical point of view” (first lines of 3.3), the discussion never deals with modeling language for instance, or interaction. There is no theoretical agenda – that is why she can say that CODA is theory-neutral. It gathers insights from these three main linguistic theories (as well as from psychology and cognitive science) to gain insights – the word is recurring – into the link between language, thought and situation. Her goal is “to point to relevant insights within and across theories that can be drawn upon to understand speakers’ use of language in collected data in relation to cognition” (last lines of Section 3). These insights bring a repertory of linguistic features which becomes the basis of analysis (modality, perspective, presupposition, coherence markers, etc. - she offers a “register of linguistic features” at the end of the book). Most of these are indeed discussed in all three approaches.
18That is why the book is as much an introduction, as rightly subtitled, as a synthesis. For this fact alone, the book deserves our praise. It makes all these theories, approaches, methodologies and notions accessible to different types of readers. Tenbrink is aware of this but I would like to insist: from undergraduate students to confirmed researchers from neighboring disciplines or theories, CODA can provide elegant and efficient access to the vast literature systematically reviewed by the author. Here lies the pedagogical value of her book: in conscientiously building this common ground, she maps out the network of relevant notions quite clearly with crucial bibliographical elements. Her definition of key notions such as discourse or information structure are simple and efficient. She manages to remain descriptive in her theory neutral approach, creating what could become an important pedagogical tool. This dimension should not be neglected. The book is quite up-to-date and it is rare enough to see syntheses that take into account very recent studies. It offers both a very solid state of the art and a novel approach. Because it is written very clearly, with a constant mindfulness of potentially less experimented readers, CODA could prove an important didactic material for the theories and methodologies Tenbrink tackles. But can it really be theory neutral?
19It must be acknowledged that the author places herself and CODA at the intersection of fields between which the discussion is not always consensual but usually friendly. Indeed, Chomskyan and formalist approaches are (logically) excluded from the literature she cites. Cognitive linguistics is to be understood as referring to non-Chomskyan approaches, which is in fact not entirely neutral. Her research program however is entirely based on the idea that language is not an autonomous faculty. This constitutes the strongest linguistic-theoretical standpoint. It forbids her to make use of formalist approaches, if only because their agenda is quite different. In the same line of ideas, CODA being a type of discourse analysis, the notion of discourse is central. In short, CODA manages to focus on discourse, thus operating a clear selection of usage-based theories. It then assumes a systemic conception of language (langue) which further restricts the theoretical framing of CODA as a methodology to speaker-based theories where meaning is central and discourse the only actual observable phenomenon. It successfully draws on an area of language study within which we could – or should? - all work to further integrate our hypotheses and insights. She concludes herself (penultimate § of section 3) that “relevant insights within and across theories can be drawn upon” even if these approaches “are often felt incompatible”, which we argue they are not.
- 4 I would like to suggest the numerous common points CODA shares with some research in pragmatics, mo (...)
20As we mentioned above, Tenbrink’s theoretical blend allows her to identify linguistic phenomena to analyze, a few of which we cited. These features are a shared focus among the three main linguistic approaches. This relies on the idea of speakers “choosing” (more or less consciously) not only what they say but how the say it4. SFG (Halliday and Matthiessen 2014) provides CODA with the more advanced reflection on this matter, which is in fact a theoretical standpoint:
“Doing CODA means looking at linguistic choice: identifying what speakers do with language, out of the many things that they could have done, understanding the significance of what speakers choose to do, therefore, has a lot to do with understanding what else they could have done. SFG provides a framework for this kind of understanding (…) by showing what the overall system of linguistic choices looks like and what it offers to speakers.” (Tenbrink 2020: 3.3.1, §1)
21The choice is viewed as expressing the speakers’ relation to the situation and linguistic content, which is in turn highly compatible with the psychology of dialogue (e.g. Pickering and Garrod 2004). Later in the book, this idea shapes the methodological tool of “semantic repertories” which can in turn provide insights into defining analytic categories. This is one of the many examples proving that Tenbrink does not indulge in concept- or theory-dropping and actually makes use of the literature she refers to, not to build or rebuild a model of linguistic activity but to design a methodology. Some theories are excluded from the get-go or much less integrated (such as Conversation Analysis), which allows Tenbrink to delimit a very coherent theoretical field. Within that field, CODA identifies common interests and shared postulates (or maybe biases?). This constitutes an important step in the direction of theoretical integration within that field so that usage really enters the realm of CL, or cognition, the field of dialogue. Her references to the psychology of dialogue leads the reader in the same direction. In this sense she is not theory neutral, as it is in fact strictly impossible. However, CODA is indeed theory neutral if we adopt a stricter definition of theory (as model building in order to give an exhaustive account of a data set, and a prediction making tool).
22Let us also remark that CODA as a methodology bridges quantitative and qualitative approach. Corpus linguistics, understood as Big Corpus linguistics, is used in line with the systemic view of language since it provides reference corpora. It also allows to better align descriptions of linguistic utterances with statistics-based approaches such as cognitive psychology. Semantic coding in CODA builds upon psychological methodologies in which quantitative and statistical reasoning is central, essentially statistical confronting a null hypothesis to observed and formalized data. The regularly debated p<0.05 confirms this view. However, this methodology is complemented by descriptive and analytic coding categories which come from cognitive grammar, context description or pragmatics. Units of utterance might be “scored” but they are scored relative to qualitative-analytic tools. Many researchers in discursive semantics should recognize some methodological elements although applied to cognitive processes rather than social constructs.
23The limitations of CODA come with its strengths: some aspects of language that are central in some of the cited references could be more represented. Indeed, the social aspect of language is quite logically less considered. The in vitro data partly justify a modeling of language and linguistic interaction rather oblivious to sociocultural parameters. Sociolinguistics, Conversation Analysis or phenomena pertaining to speech styles are only mentioned. The wish that more could have been done is to be read as a compliment of what has already been achieved, and as a suggestion that there is room to include such approaches in further details. Notions such as idiolect, linguistic asymmetry or again interactional features could prove extremely pertinent to CODA-type dataset. For instance, interactional accounts of collaborative story-telling could inform the study of collaborative problem-solving. Tenbrink cannot however be criticized for facing the same problem most usage-based cognitive language scientists face: how can we reconcile cognitive and interactional approaches? How can cognitive linguistics really enter the realm of dialogue? She does find consensual ground by triangulating for instance: Pickering and Garrod (2004) approach to dialogue is compatible with both Cognitive linguistics and Interactional studies. This is interesting because the two psychologists only rarely cite cognitive linguistics or functional linguistics but they do refer to Schegloff and colleagues’ work. Schegloff also happens to be one of the commentators of their 2004 paper (the commentaries are to be read immediately after the article itself). However, this psychological account remains based on cognitive processes such as alignment or cognitive load. This is where Tenbrink’s claim of theoretical neutrality could perhaps be better coined theoretical coherence. This coherence is both essential to the methodological success of CODA and to the main achievement of the book as a systematic and almost exhaustive presentation of the field. But it prevents any real theoretical confrontation. Situation prevails over interaction. Again, one can hardly be attacked for being coherent. The optimistic conclusion would be that many a researcher could take CODA as a starting point to precisely confront the theoretical and methodological challenges posed by the social approaches to language. The space “beyond CODA” that Tenbrink points to herself argues in favor of extensions of the model and, I argue, extensions towards ecological and interactional aspects of linguistic productions. The more general scientific context seems to be pointing in this direction as well. Indeed, situated and distributed cognition favor approaches trying to conciliate interactive and mentalist views of language and language activity. Several endeavors within the field of linguistics have been made in this direction (Du Bois 2014’s Dialogic syntax, Zima and Brône 2015; and in sociolinguistics, Hollman 2017).
- 5 This is but one reference taken from a vast literature, of which I do not claim expertise.
24Is it all relevant then to Cognitive linguists or linguistics (both abbreviated CL hereafter)? I will argue that it is for two reasons. First of all, CODA is extremely relevant for CL who work in experimental settings, an approach which is developing in our field of study to include experimental pragmatics for instance. It is just as relevant for linguists focusing on behavior research usually with psychological inputs: research on Autism Spectrum Disorders, or dysphasia and related disorders would benefit from a methodological discursive approach which is based on cognitive postulates to conceptualize the discourse situation (cognitive load, point of view, task, etc.) and its analytical tools (annotating categories, ontologies to name these two). CODA’s focus on discourse and situation-dependent features could inform such approaches directly. It offers a set-up that can be the basis for many other explorations. For instance, high-functioning autistic speakers are all at once neuro-atypical, linguistically proficient and communicatively and/or socially impaired. Julia Parrish-Morris and Robert T. Schulz focus on the use of discourse markers by people with ASD to better doctor-patient communication or avoid diagnosis delay. Thus, in Parrish-Morris et al. (2017) the authors speak of “linguistic camouflage”5.
25Second of all, it provides precious insights into what colleagues in psychology or cognitive science know of linguistic theories. Indeed, close and detailed attention to actual language productions is the core argument in studying thought processes and cognitive tasks. Many features of language CODA draws attention upon will not seem new or unusual to linguists but the approach is extensive and offers a very clear and comprehensive review every student of linguistics should benefit from - it represents a sort of biased state of the art: what do cognitive scientists or psychologists retain from linguistic studies? I argue that it is absolutely essential for CL, and I include myself, to know what representation of (cognitive) linguistics colleagues from neighboring disciplines have. It is the basis for further discussion and collaborative work on common matters. Indeed, if CODA relies at its core on a language-related assumption, the methodology is not intended for the study of language alone or in itself. Space cognition, perception cognition and reasoning are three of the main areas informed by CODA studies. It demonstrates that studying language in use and not only the language of space proves relevant to space cognition (Section 8.6 shows how discursive markers for instance are useful in studying planification process during navigation). Metaphors, grammatical features, information structure are among the linguistic tools Tenbrink and her colleagues used in field-advancing studies on space cognition. It is worth noting that linguistics and discourse analysis are not viewed as service providers but as fields of inquiry in their own right with specific methodologies and insightful – and valued – state of the art. Linguists are sometimes at a loss to value their expertise to the wider scientific community. CODA could show us how to do just that.
26I would like to add that Tenbrink’s scientific endeavor and subsequent book are all the more welcome in the French academia since it is written very accessibly indeed and presents, again, a very clear synthesis of cognitive approaches to language and how they relate to other areas of the study language and/or cognition. CL in France largely remains the domain of English linguistics (anglicistes) and is most frequently used in English departments – even if works by Myriam Bouveret, Philippe Gréa or Dominique Legallois, to name a few, are notable exceptions. Although written in English, this book could help spread this type of linguistics outside of English departments. Tenbrink uses studies and examples from a variety of languages, including French and her native German. If some of the contents will not appear new or novel to some readers, especially of this journal, it sheds a novel light on these matters and how they bridge different fields of study based on their mutual relevance. It offers our scientific community the opportunity to become more aware of both what actually brings us together, and what can be deeply relevant to neighboring fields. The book also invites us to discover what we already know – or know how to do – in a new light. Indeed, I suggest that many of us have been doing some form of out-of-the-lab CODA without knowing it and that gaining explicit knowledge of this field could prove useful to gather, share and discuss insights. Chapter 9 suggests triangulation methodologies and expansion outside of experimental settings into corpus linguistics and natural situations. The author of these lines is part of that effort.
27This envisioned expansion is where CODA could make its most significant contribution because it would allow to recharacterize ecological corpora in terms of situations and the constraints it bears on linguistic productions (Danino 2014 for an attempt). Based on the experimental set-up and studies presented, if we were to translate CODA into French, we could say Analyse de discours cognitive or analyse cognitive de discours (Cognitive [Discourse Analysis]). Any DA limits itself to the contents of the discourse in the sense that it cannot be topic-neutral. Discourse is related to what is being discussed. It also has to deal with its analytic categories that cannot be ad hoc but cannot be completely oblivious to the discourse content. Whether cognitive or critical, or both (as in van Dijk 2008), DA must model discourse and its relations to its production modes and context. In this definition, CODA really is a thorough discourse analysis, one which focuses on cognitive contents (what type of reasoning, what conception of place or space, etc.) and cognitive constraints (limited time, particular object manipulation, etc.). Indeed, discourse analytic tools are used to gain insights into cognitive phenomena. Discourse is viewed as a cognitive tool and process.
28While this will not be disputed, I would like to suggest another possible translation in French: Analyse de discours cognitifs ([Cognitive Discourse] Analysis). Indeed, all discourses collected in the relevant studies are cognitive in nature in the sense that they are produced in a context constraining, forcing or hindering a particular cognitive ability. It can interfere with language production or favor it. In other words, the very content of discourse is cognition. Thinking-aloud protocols or reasoning explanation are particularly telling. Of course, every discourse is cognitive and each one can provide insight into cognitive activity (especially since social action is seen as a cognitive ability; e.g. Pennington 2012; Clément/Kaufman 2011 to contrast social cognition with cognitive sociology). But in CODA, discourse is the tool in which and through which cognition is studied, as a foregrounded subject matter, as a topic. This is by no means reductive: cognitive discourses are a legitimate type of discourses or interactions. Teaching situations, medical consultations, group reasoning in a variety of decision-making processes would greatly benefit from such a typology. Not to mention all the studies in psychology, language sciences or sociology, for instance, which heavily rely on elicited, constrained interactions or on interviews with a researcher. As a matter of fact, a lot of research is cognitive discourse. Studies presented by Tenbrink seem to argue in favor of such a reconsidered typology. Many of the novel and often surprising results CODA contributes to the study of language and cognition emerged from clever experimental set-ups or lesser-studied domains. Their take on the so-called tunnel experiment [Section 4.2.2] reveals that speakers, while orientating in a basic virtual tunnel blend allocentric and egocentric reference frames, in quite individual ways. Similarly, discourses of architecture and mountaineering [Ergorova et al. 2015]) offer new types of data. Two remarks must be made: 1) most examples and studies cited in the book deal with space cognition, Tenbrink’s central research object, offering a large amount of knowledge on the matter; 2) most if not all studies build on CODA’s key conceptualization of a link between situation and language production(s). CODA’s main breakthroughs came from studies in which discourse is conceived in terms of situational constraints on our cognition and what could be coined natural control parameters, as I used the phrase in my dissertation (Danino 2014). As I suggested above, situation primes over interaction, which is one parameter of it. This means that situational parameters (cognitive constraints, task at hand, level of expertise, time constraint, etc.) appear to be more defining of the resulting discourse than interactional parameters. This is not to say that interaction is second only to cognition. And it does not reach any conclusion outside of the lab. But it defends the centrality of situation, not only as a descriptive category but a pivotal concept. It directly impacts the definition of discourse which is not only viewed as a critical notion - stretches of talk and linguistic productions which reveal social and cultural positioning (hence “critical” in CDA) - but as the locus of the cognitive processing of reality. This phenomenological approach seems sometimes assumed in CODA though never explained, although themes such as attention, space or perception seem rather fitting.
29As a conclusion, I hope to have shown that CODA can constitute an excellent starting point to try and tackle some of the challenges posed to cognitive linguists in their relation with quantitative, interactional and psychological approaches of language and discourse. I believe it actively recruits the reader into one of the most challenging discussions for linguists: conciliating cognition and communication. If Tenbrink’s book is read as a preparatory reading for this engagement and as a program for expansion, then it should be of interest to any linguist, cognitivist or not, experienced or beginner. The case of discourse markers (DM hereafter) is always revealing. They received a lot of attention in various fields (Fox Tree 2010, Crible 2017 for two attempts at exhaustivity) and are regularly analyzed by Tenbrink. She considers that they indicate interpersonal positioning (attention to the co-speaker) and build textual and discursive coherence. In other words, DM are both cognitive and interactional. DM also serve a type of theory of mind in which interactional competence includes a representation of the co-speaker and of the dialogue itself (Danino, Knutsen, Col 2020), which increases the complexity of their definition and treatment. Conversation Analysis has long recognized the importance of these words or phrases in structuring sequences of talk and regulating conversational behaviors. I would like to suggest that DM still lack an integrated multidisciplinary approach. It would be interesting to see if CODA could solve the analytical puzzle posed by DMs which force each discipline to choose between a formal or a functional definition. Indeed, CODA invites the community of linguists to focus not only on the links between language and cognition, but on the relation of discourse and cognition.