Critical evaluation of CA

The very bedrock of CA is its pointed focus on how sequences of actions are socially organised through turn-taking in interaction (Hutchby and Wooffitt, 2011). With Goffman, CA argued that structural organisation underlies all institutional interactions. This idea underlies all the studies reviewed above. The important undertaking of CA is, thus, to unveil these structures. As a result, CA has demonstrated how people accomplish tasks and construct roles and meaning through the complex organisation of speech, but I argue that it overlooks important aspect of what happens in interaction. In this section, the underlying principles of CA are critiqued in order to qualify the method’s objectives. The critique is divided into three interrelated sections concerning: (a) the sociological agenda and member’s perspective; (b) inductive ‘unmotivated looking’ and ‘why that now’ and (c) from one reductionism to another: CA and cognition.

2.2.3.1 The sociological agenda and member’s perspective

CA has widely been criticised for being unable to respond to the sociological agenda (Hutchby and Wooffitt, 2011). In particular Critical Discourse Analysis has criticised CA for being unwilling to link micro particularities with macro levels of sociological variables (Fairclough, 1995a; 1995b). Detailed focus on particularities in interaction, as for instance turn constructions, complicates the relation between micro-analytical findings and general sociological claims related to gender, class, etc. (Nielsen and Nelson, 2005). Furthermore, another repeated critique is CA’s lack of contextualisation of utterances that appear in wider social practices (Hutchby, 1999). CA finds it problematic to start with or to base itself on pre-established, sociologically contextual variables. CA disassociates itself from sociological claims about institutional characteristics as hierarchical power relations (Hutchby and Wooffitt, 2011). These critiques point lead directly to another CA stance concerning membership categorisation (Sacks, 1972). Hutchby (1999) defends CA by referring to how Schegloff (1998) eloquently defines CA’s analytical imperative as a concern for how categories matter for members rather than analysts (Hutchby, 1999:86).

CA’s agenda, thus, does not immediately connect with the sociological agenda and:

this quickly makes it problematic to talk in terms of sociological variables such as power, ideology, and so forth, because the typical sociologist’s question is turned around: Rather than asking how social power affects the nature of the discourse, or whatever facet of social life is under scrutiny, we need to ask how, and indeed whether, the participants in a social scene show themselves to be oriented to power as a relevant phenomenon in the ongoing course of their interaction. (Hutchby, 1999:86)

Hutchby (1999) argues that when conventional macro-sociologists deal with a social theory encompassing invisible concepts of power, gender, class etc., CA is able to make such concepts visible by identifying the structures of interaction to which the members orient, and the consequences of this orientation in terms of showing how participants overtly categorise what they do (Hutchby, 1999:92).

Whereas CA makes a strong argument for denying that action is dictated by discourses arising outside local conversation, it is reluctant to deal with a more nuanced analytical perspective. Because CA confines its perspective to dealing with participants’ first-person perspectives, the method ignores other important questions such as: “is this consultation good?” At best, the answer will rely on whether the treatment seems good from the local perspectives of the participants. If local, internal orientation contrasts with outside observations, CA ignores this inconsistency. According to Sacks, investigations should be concerned with what is observable, and this tends to be found in verbal utterances. For instance, Pilnick and Zayts (2012) define their locus of interest on the basis of observation:

“We have focused our analysis particularly on consultations where these social and economic circumstances are interactionally visible, and the consequences that this presence has for how consultations unfold and how decisions are made and accepted or challenged”

(Pilnick and Zayts, 2012:278). The analytic intra-logical perspective only allows for observations to be interpreted in a macro-sociological perspective when they can be explicitly tied to verbal utterances. When non-verbal utterances are dealt with in the reviewed studies, there is a tendency to treat them as interactional epiphenomena that only add information in the co-construction of meaning. Thus, the main focus is not on bodily dynamics per se, but rather on how verbal utterances are modulated by other multimodal resources. As a result, real-time inter-bodily dynamics and dynamic coaction are overlooked because actions are interpreted through the same socio-normative lens as verbal utterances. In contrast, my analyses demonstrate that relatively often participants themselves are unaware of the impact situated interactions have for their perceptions of their own roles.

2.2.3.2 Inductive approach and ‘why that now’?

While CA claims that the inductive question “why that now” (Heritage and Clayman, 2010:14) serves as a guiding principle in its analysis, the review reveals the fact that analyses are far more often driven by pre-defined assumptions about how conversations work, as I will demonstrate below. Over time the inductive apparatus has consolidated a deductive, category-dense framework that guides the analytic gaze toward certain aspects in conversations. Inductively inferred generalisations gain status as a theory used to deduce explanations. The question ‘why that now’ was not addressed in the reviewed studies.

Rather, based on sequentiality, interest falls on turns in interactions and how such turns relate to a pre-defined problem. In a recent article, Pilnick and Zayts (2012) state:

A basic assumption of this perspective is that social interaction is structurally organized, and the focus of analysis is to uncover the socially organized features of talk in context. In analysing these data, we proceeded as follows. For each consultation, the opening sequence was analysed sequentially. More specifically, a display of a first medically relevant concern in a particular turn by the patient allowed us to locate the strategy, used by the doctor in immediately preceding talk that had ‘occasioned’ it. We also looked at the doctor’s talk immediately after the patient’s display of a concern in order to ascertain that our understanding

of a particular sequence corresponded to participants’ own understandings. (Pilnick and Zayts, 2012:243)

This statement directs analytical attention toward the mechanical structure of conversation.

At the same time, it blindfolds the analyst, whose gaze is primed for spotting turns. If one is looking for turns, one finds turns. To give a few examples from the reviewed papers:

“The analysis itself was carried out on a turn-by-turn basis. The principle behind our analysis was to examine how turns were taken with regard to other participants’ speech and what sequential implications each turn had for the next” (Kettunen et al. 2001:403). Webb (2009) adds: “The key phenomena analysed are the structure, wording and timing of the doctor’s questions; the structure and wording of patients’ answers, as well as the topics the patients choose to address and the order in which they produce them; and the doctor’s subsequent utterances after the patients’ responses” (Webb, 2009:858); and: “When we examined nurses' and patients' speech word by word, we discovered four participation frames that produced taciturnity: in the hands of professionals, compliant, guilty, and polite” (Kettunen et al, 2001:399). CA’s focus on turns has been related to production of speech. Its history of dealing with verbal turns ad litteram becomes its biggest obstacle.

When interactions are transcribed verbatim, there is a risk that non-verbal actions are reduced to simple meaning transporters, and their dynamical characteristics are replaced by symbolic values. Thus, in the transcripts, only ‘meaningful’ actions are annotated (head nods, gaze orientation and gestures as pointing toward something in the same way as a verbal deictic marker). Remarkably, such actions are assigned the same rules as those valid for analysing verbal utterances. At worst, the non-verbal actions are completely ignored and it is hypothesised that this is due to the sociological explanatory framework that often works for verbal utterances but might be inadequate in the study of inter-bodily dynamics.

Although CA defines talk-in-interaction as its unit of analysis, its close-knit methodology primes the attention of the analyst to identify words in sequences of turn-taking. In such cases, dynamics extending beyond turn-taking will not be registered. In fact, CA methodology has, over time, accumulated a theory of conversation.

2.2.3.3 From one reductionism to another: CA and cognition

From Goffman, CA took the idea that structural organisation underlies all institutional interactions: “I assume that the proper study of interaction is not the individual and his psychology, but rather the syntactical relations among the acts of different persons mutually present to one another” (Goffman, 1967:2). As mentioned in the introduction to CA, and as Steffensen (2015) emphasises, CA became an important replacement of formal, generative and structural linguistic analysis of texts through its argument against language as an individual and internal system controlled by the individual’s neural circuits within the brain. The CA position argued against the localisability of language within individuals, and rationalised that such a reductionist view was incommensurable with the idea that talk was a sociological phenomenon ordered by norms and rules across contexts. However,

according to Steffensen (2015), a closer look at CA’s attempt to go beyond a reductionist view of language reveals that CA merely exchanges one reductionist view for another:

Indeed, they tend to replace one reductionist model of language with another model that is no less reductionist. Thus, given that “The disciplinary motivation for such work is sociological”

(Sacks et al., 1974:698), the conversation analytical model replaces the bio-reductionism of Chomskyan linguistics with a socio-reductionism that ignores the biological and ecological constitution of language. Evidently, it is fully legitimate to invest one’s research interests in the micro-sociology of conversations, but it is illegitimate to assume that such a sociological perspective represents a better or fitter approximation to the complex reality that we call language (or conversation). (Steffensen, 2015:110ff)

CA’s socio-logic replaced an intra-logic and changed the analytical point of departure, but it remained reductionist in its methodology. Reducing all interactional phenomena to social behaviour is a fallacy. Some activities in interaction play out too fast to be defined as social dialogue (Steffensen, 2013; Pedersen, 2012): thus, the explanatory frame should be qualified and developed further. For instance, gestures are not only relevant when they add local meaning to a conversation. In fact, meaning is often generated post festum, enabled by non-verbal dynamics in interactions.

Moving beyond this reductionism requires an ecological approach to interaction. In institutional settings, goals and tasks need to be accomplished. The achievement of a goal involves perception, action, reasoning, decision making, and as such, interaction is not purely social behaviour, but also bio-cognitive. Whether or not cognition, as an underlying basis, has a place in CA is a central question that needs to be addressed. Offhand, CA researchers do not participate in discussions about how cognition matters for managing interaction. Garfinkel’s pun “there is nothing in the head except brains” (Garfinkel, 1963:190) refers to the belief that cognition is a mental state that is unobservable in natural interaction, and so not of interest for conversation analysts. Or, as Hutchby and Wooffitt, (2011: 220) emphasise: “Conversation analysts reject the determinism of cognition on methodological grounds, arguing that talk-in-interaction is an independent domain of activity, the properties of which are not dependent on psychological (or sociological) variables” (Hutchby and Wooffitt, 2011: 220). Or as Ataki (2012) argues: “There is in any case a profound difference between individuals’ undemonstrated inner feelings, hopes and intentions […] and the visible ‘participants’ concerns’ which are available for public consumption. […] CA’s interests are in what is publicly transacted, not what is privately thought or felt” (Antaki, 2012:494, 497). Mistakenly, CA holds the view that cognition is clearly segregated from interaction activities due to its mental internalism. Only when internal cognition is represented externally through talk-in-interaction can it, according to CA, be treated as an analytical object, because the cognitive dimension of thinking transforms into a social phenomenon that members manage in interaction:

cognitive and psychological phenomena are separate from social behaviour. They may be disclosed in social contexts, in that we can report our thoughts, verbalize our memories of events and articulate attitudes and beliefs; ultimately, though, they are not private phenomena.

Finally, it seems uncontentious to assume that cognitive and psychological phenomena determine public social behaviour. (Hutchby and Wooffitt, 2011:217)

While CA’s objections against interpretations of private thoughts, hidden mental processes and the importation of general discourses are appreciable, its understanding of the ontological status of cognition is flawed. Recent embodied and distributed theories about cognition break with conventional intra-cranial models of cognition, and in the cognitive sciences this tendency has dramatic consequences: the analytical attention moves away from brain-bound processes toward interactions between people in an environment (Hutchins, 1995a; 2014; Clark, 2008; Hollan et al., 2000). What people do together, and how they solve problems and find solutions, is not just a question of how they talk-in-interaction; it is just as much a question of distributed thinking. In that sense, cognitive processes or thinking are likewise observable activities that can be studied in detail. CA has adopted a model of cognition as an invisible phenomenon. However, when new models and perspectives, such as the ones provided by DC, show that cognition is not purely invisible, CA needs to address these insights and adapt its methods accordingly.

It seems rather paradoxical that the clear-cut distinction between linguistic and cognitive processes is less rigid than assumed by CA researchers. In fact, CA and DC work with the same basic material, but despite the fact that the two positions share a unit of analysis, they remain incommensurable on a theoretical and practical level. Finally, very few scholars attempt to bridge the gap between cognitive and linguistic endeavours. As an exception, a new group of scholars operate beyond a reductionist framework. In numerous studies, Goodwin (1994; 2000; 2003; 2007) Streeck (2009) and Heath (2002) amongst others, link situated cognition to how members orient to shared, material artefacts in interaction and how embodiments affect problem-solving activities. For instance, by opening up for dealing with multiple timescales in interaction, Goodwin (1994; 2002) shows how nature over time is being transformed into culture. In other words, micro-sociology is explained in relation to how predecessors have sown the seed of local action (Goodwin, 1994; 2002), and not just in relation to how local utterances emerge. Orthodox CA, as well as the reviewed studies, however, does not match this cognitive and non-local turn in interaction studies.

2.3 Cognition as distributed, embedded, ecological, extended, embodied and situated