Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  Collaborative dialogue and ‘languaging’

  Up to this point, I have drawn largely on the work of Lantolf, but I will now turn to look at the contribution made by Swain and her co-researchers. In Chapter 7, we considered Swain’s Output Hypothesis. This was framed within the cognitive paradigm. That is, output constituted data that fed into the mental mechanisms held responsible for learning. Swain (2000), however, subsequently reframed her understanding of the role of output in sociocultural terms. She now viewed it as a socially constructed cognitive tool for mediating learning and, accordingly, focused on examining the role of ‘dialogue’ and, in particular, ‘collaborative dialogue’, which she defined as ‘dialogue in which speakers are engaged in problem solving and knowledge building’ (p. 102).

  Swain and her co-researchers conducted a whole series of studies designed to examine the contribution of collaborative dialoguing to language learning. These studies typically involved asking learners to complete a challenging task in pairs, recording the conversations they took part in as they did so, and then investigating the language-related episodes (LREs) that occurred when they ran into problems. See Example eight in Chapter 7 for an example of a LRE. Swain was interested in two main questions: Were learners successful in jointly solving linguistic problems? Were these learners subsequently able to use the linguistic forms that figured in their solutions independently? In other words, Swain looked for evidence to show that collaborative dialogue in problem-solving was knowledge-building. The methodology she employed was to carry out detailed analyses of the LREs that occurred in collaborative dialogue and then investigate—sometimes by means of post-tests—whether these had contributed to the learners’ self-regulated use of the linguistic forms.

  A good example of the kind of study Swain conducted is Swain and Lapkin (2002). In this study, two Grade seven immersion learners first viewed a videotape of a mini-lesson on French reflexive verbs and watched two learners modelling writing a story. They then wrote a story collaboratively based on a picture jigsaw task. Their text was then reformulated by a native speaker after which the learners worked together to notice the differences between their text and the native speaker’s. There was also a stimulated recall interview where they were asked to comment on the features they had noticed. Finally, they were given a copy of their original story and asked to rewrite it. Swain and Lapkin examined the LREs that arose when the learners were writing the initial story and during the noticing activity. They then compared the learners’ initial and final texts, treating them as a pre-test and a post-test. 80 per cent of the changes that the learners made were correct. Some of the correct changes corresponded to the changes in the reformulated text but other changes were directly traceable to the LREs.

  In a later study, with a very similar design, Swain and Lapkin (2007) investigated the role played by ‘interaction with the self’ (p. 82) (i.e. private speech). The key difference between this and the earlier study was that a single learner performed the writing tasks independently. The focus was on the differences that this learner reported noticing between his original text and the reformulated text and how these related to the changes he made to his original text. Swain and Lapkin reported similar results to the earlier study. By interacting with himself, this learner was able to focus his attention on problematic linguistic forms, construct new understandings, and consolidate existing knowledgeNOTE 3.

  The attention to linguistic form that occurs in talk—whether social or private—when a problem arises constitutes languaging (Swain 2006). Languaging can occur when any kind of problem arises but the particular type that Swain was interested in was ‘languaging about language’. She argued that it constitutes one of the principal ways in which advanced levels of language learning can be achieved as it involves learners (1) articulating and transforming their thinking into ‘an artefactual form’ and, thereby, (2) providing a means for further reflection on this form. Languaging, then, is not just a facilitator of learning; it is where learning takes place.

  Final comments

  Sociocultural SLA has attracted increasing interest among both researchers and language educators. It is the most established of the ‘social’ theories I will consider in this chapter. It differs from the input-output model informing the research we considered in Chapter 7 by its insistence that the ‘social’ and the ‘cognitive’ are dialectally connected. It disputes the dualism inherent in cognitive SLA by claiming that social and cognitive processes are not distinct and separate, but two sides of the same coin. It also provides a basis for what Lantolf (2011) called ‘educational praxis’, a process whereby theory and the practice of teaching are mutually informing. The importance sociocultural SLA attaches to learning as a mediated external, not just a mental internal phenomenon, is of obvious relevance to language pedagogy. In fact, many of the studies by Lantolf and Swain involved instructed L2 learners. We might ask, however, to what extent sociocultural SLA truly belongs to the ‘social turn’, which—as I have already noted—it pre-dates. In some respects it clearly does. It rejects the idea of a universal order and sequence in L2 acquisition—since learning in the ZPD varies from one individual to another—and it adopts a constructivist stance to the role played by interaction in learning whereby the social context is the actual site of learning. Moreover, its methodology emphasizes the importance of detailed analyses of interactions involving learners as, for example, in Swain’s investigation of language related episodes.

  But in other respects, sociocultural SLA does not conform to the key features of the ‘social turn’ (see Table 9.1). Much of the research has focused quite narrowly on adult foreign or second learners in ‘social’ contexts, largely restricted to those created by some kind of instructional activity. There is, of course, no reason why sociocultural theory cannot be studied in the naturalistic contexts found in the real world but—by and large—this has not happened. As a result, there has been no exploration of the myriad social identities that L2 learners can and do draw on in their social encounters in the real world. The L2 learner is positioned as a ‘novice’ and the teacher/ native speaker is an ‘expert’. In its methodology, too, it is not adverse to using tests as a complement to detailed analyses to establish whether self-regulation has taken place. However, I do not intend to suggest these aspects of sociocultural theory constitute a limitation. In fact, they can be construed as signs of its main strength, namely that it is clearly both a social and a cognitive theory and therefore manifests epistemological and methodological aspects of both paradigms.

  The sociocognitive approach

  Like sociocultural SLA, the sociocognitive approach acknowledges that there is both a social and a cognitive dimension to all forms of language use and thus it follows that to understand language acquisition it is necessary to understand how these two dimensions interrelate (Batstone 2010). Thus, like sociocultural theory, the sociocognitive approach emphasizes learning-as-participation when mind and body come together in and through the interactions that occur in specific situational contexts.

  The question, of course, is exactly how the social and the cognitive dimensions interrelate. For Batstone, there are distinct cognitive and social dimensions that come together when we engage in social interaction. He acknowledges that we have prior conceptions—i.e. schematic knowledge, scripts derived from previous interactions and personal identities—which we draw on while simultaneously attending to the linguistic signs that signal how speakers are orienting to the discourse in a particular context. In this way, speakers achieve a convergence of their internal and external worlds. Learning takes place when the exigencies of a particular interaction challenge pre-existing schematic knowledge, causing modification to take place.

  Atkinson (2011c) takes a different view of the interrelationship of the social and the cognitive. He sees language as simultaneously in the world and in the learner’s head and, thus, it is neither ‘social’ nor ‘cognitive’, but both together. Thus there is no hard division between �
�inner’ and ‘outer’ processes. For Atkinson, language co-exists in cognitive and social space. For example, the simple activity of exchanging greetings involves ‘facial expression, physical orientations, affective stance, a conventionalized social scene in a social setting with a social purpose performed by social actors and the effective deployment of the social tool of language’ (2002: 531–2). Cognition, then, is not something mental but is ‘fundamentally continuous with the world’ (2011: 149). In Atkinson’s sociocognitive approach ‘neither language acquisition nor language use – nor even cognized linguistic knowledge – can be properly understood without taking into account their fundamental integration into a socially-mediated world’ (2002: 534).

  Atkinson argues for the inseparability of the social and the cognitive. A sociocognitive event is ‘profoundly integrative’. However, unlike Batstone, and also unlike sociocultural theory, Atkinson rejects a view of learning as something that begins in the social world and then ends up in the learner’s head. There is no such thing as ‘internalization’ in Atkinson’s theory. Learning is a purely external phenomenon.

  The key construct in Atkinson’s theory of L2 learning is ‘alignment’. Through interaction learners and their co-participants align adaptively to the ‘ever-changing mind-body-world environment’ in which they are jointly functioning (Atkinson, Churchill, Nishino, and Okada 2007: 171). To illustrate this, Atkinson et al. describe a series of interactions between a Japanese high school student (Ako) and her aunt as they work collaboratively on an English exercise, practising the ‘have you ever’ construction in order to demonstrate Ako’s ‘growing ability to participate meaningfully in conversations using the Have you (ever) experience construction’ (p. 182). An implication of Atkinson’s sociocognitive approach—reflected in the study of Ako—is that knowledge of an L2 is not abstract but a situation-dependent process. Learners participate in ‘situated activity systems’ that ‘support language growth’ and what they come to do and know is embedded in these systems. Learners learn by doing, so what they learn is tied to the activity in which they learned it. Interaction is central to this process as it is through interaction that alignment takes place. Alignment is ‘the necessary condition’ for acquisition to occur.

  To date, there is little research to support the claims of a sociocognitive approach. As Atkinson (2011c) admitted, it is ‘new and undeveloped’. Ideally, it requires longitudinal studies of L2 learners to show how the world-body-mind interface manifests itself in interaction at different times so that change can be identified. It also requires a methodology for examining the details of the interactions in which learners participate. Atkinson drew on conversation analysis to achieve this. This is the next approach we will consider.

  The conversation-analytic approach

  Like sociocultural theory, the conversation-analytic approach emphasizes the importance of learning as a process—i.e. as something that happens in social interaction. It also resembles sociocognitive theory methodologically through its insistence on the detailed analyses of the sequential talk and the non-verbal behaviour that accompanies it in specific interactions involving L2 learners. I will begin with a brief exposition of the conversation-analytic approach and its methodology and then examine how it has been used to investigate L2 learning.

  Theoretical principles

  Conversation analysis (CA) grew out of branch of sociology—ethnomethodology—the aim of which is to describe and explain the methods that ordinary people use to participate in everyday events and to make sense of their daily lives. Drawing on many of the theoretical premises of ethnomethodology, it developed a methodology for analysing how conversations are constructed, how participants make sense of what transpires in a conversation and—of special relevance to conversations involving L2 learners—how problems that arise are dealt with.

  A key construct in the conversation-analytic approach is interactional competence, conceived of not as a cognitive ability, but as how interaction is accomplished in different situations. Interactional competence includes how the participants in an interaction mutually co-ordinate their turns, how roles are established and role-relationships enacted, how linguistic and non-verbal resources are used to produce and interpret turns and actions, and how problems are repaired. Interactional competence is highly situated. That is, it consists of the know-how about how to behave in specific situations—for example, in service encounters; in language proficiency interviews; and in pharmacist-patient consultations. Examples of how conversational analysis was used to investigate interactional competence in these situations can be found in Hall, Hellerman, and Pekarek Doehler (2011).

  As Kasper and Wagner (2011) noted, ‘interactional competence is understood to serve a double duty as both a fundamental condition for and (as an) object of learning’ (p. 119). That is, learners need to know how to interact, but through interaction they develop the routines associated with specific interactional events. In developing these routines, L2 learners also expand the linguistic resources needed to participate in them. Thus, the conversation analytic approach is concerned with how learners learn to do talk-in-interaction and how this leads to the acquisition of conversational structures and also how specific linguistic forms—i.e. sounds, words, and grammatical structures—are used and learned. In this respect, it has a broader agenda than cognitive SLA, which by and large has only addressed the acquisition of linguistic forms.

  The methodology of conversation analysis

  Conversation analysis seeks to describe ‘how participants understand and respond to one another in their turns at talk’ (Hutchby and Wooffitt 1998: 14). Analysis is entirely bottom-up and data driven (i.e. there is no reference to a priori norms or structures). Conversation analysts make no assumptions about or reference to inner, cognitive states. They base their work on very narrowly transcribed conversations in order to examine them in minute detail. Markee (2005) listed what he considered the minimal requirements for a CA transcript. It needs to show:

  … how members hesitate, pause, or become silent during talk, how they speed up or slow down their delivery, how they modulate the volume of their speech, how they emphasize certain words or sounds through stress, and how they overlap each other’s talk. (p. 358)

  He then added that if conversations are video-recorded it will also be possible—and desirable—to include information about gestures, embodied actions, and eye-gaze behaviours.

  Two aspects of conversations have received particular attention—turn-taking and repair. Conversation analysts have identified a number of rules that underlie speaker selection and change: only one speaker speaks at a time; a speaker can select the next speaker by nominating or by performing the first part of an adjacency pair (for example, asking a question that requires an answer); a speaker can alternatively allow the next speaker to self-select; and, in everyday conversations, there is usually competition to take the next turn. However, turn-taking in classrooms—and in other institutional settings—differs in that ‘the organization of turn-taking is constrained and related to the institutional goal’ (Seedhouse 2004: 168).

  Conversational analysts define repair as the treatment of ‘trouble’—i.e. anything that the participants consider is impeding communication—and treat it as a joint production. In this respect, then ‘repair’ includes much more than the negotiation of meaning and corrective feedback, see Chapter 7 and Chapter 10. Schegloff, Jefferson, and Sacks (1977) distinguished different types of repair work in terms of who initiates the repair (the speaker or the hearer) and who carries it out. The four basic types they found were: (1) self-initiated self-repair; (2) self-initiated other repair; (3) other-initiated self-repair; and (4) other initiated-other repair.

  There is now a rich literature on the application of conversation analysis in SLA (CA-SLA for short), although much of this work is purely descriptive. As our interest here is on what CA can show about learning, I will examine research that has explored three themes: ‘doing learning’—i.e. how learne
rs and their conversational partners engage in explicit learning events—’short-term development’, and ‘long-term development’ (Markee 2000).

  Doing learning

  ‘Doing learning’ involves the ‘interactional moments where the participants’ focal concern of their interactions is learning’ (Kasper and Wagner 2011: 127). In other words, ‘learning’ becomes the topic of social practice resulting in the participants foregrounding the identity of the learner as ‘learner’, as opposed to some other identity, such as customer or patient. In such interactions, the learner’s interlocutor (another learner or a native speaker) becomes a resource from which the learner can obtain new input. ‘Doing learning’ is clearly very similar to Swain’s languaging.

  One common way of doing learning is when one of the participants experiences a vocabulary problem and repair work ensues to address the problem. Kasper and Wagner (2011) provide a detailed analysis of part of a business phone call between two sale representatives—one Danish (J) and the other English (T). J begins by suggesting they have lunch when T arrives in the afternoon, but then—believing he has used the wrong word for the mealtime—self-repairs by substituting ‘breakfast’. Further repair work follows. As the conversation continues, it becomes clear that J has a very uncertain understanding of the English names for mealtimes. Recognizing this, T now helps out by explicitly telling J the names. In this way ‘a repair sequence transforms into a lecturing/instruction sequence’ (p. 131).

  Doing learning in this way is, of course, not surprising given that—even in naturally occurring interactions such as that between J and T—occasions will arise where it becomes necessary or helpful for the more competent speaker to help out the less competent, not just by engaging in repair work, but by adopting the role of a tutor. In instructional settings, of course, this constitutes an authorized social practice.