Agency in the classroom

[Già pubblicato in Lantolf, J.P. & Poehner, M.E. (Eds.) (2008), Sociocultural theory and the teaching of second languages, London: Equinox.]


This chapter is about the notion of agency and its role in language learning. The main principle involved is that learning depends on the activity and the initiative of the learner, more so than on any “inputs” that are transmitted to the learner by a teacher or a textbook. This does not, of course, diminish the need for texts and teachers, since they fulfil a crucial mediating function, but it places the emphasis on action, interaction and affordances, rather than on texts themselves. Although this is nothing new if we take seriously the writings of Comenius, Vygotsky, Montessori, Dewey and many other educational thinkers over the centuries, it is good to remind ourselves of the wisdom of this fundamental pedagogical principle.

Agency can be defied in a “bare bones” way as “the socioculturally mediated capacity to act” (Ahearn 2001:112). This definition begs the question (acknowledged by Ahearn) of what kinds of sociocultural mediation might be involved or, indeed, what we mean by sociocultural mediation. In another analysis, Duranti (2004: 453) provides a working definition of agency that includes three basic properties: 1) control over one’s own behaviour; 2) producing actions that affect other entities as well as self; 3) producing actions that are the object of evaluation. Although this definition is more detailed than Ahearn’s, it does not elucidate the notion of sociocultural mediation.

According to Wertsch, Tulviste and Hagstrom (1993: 336), Western theories about agency assume that agency is a property of the individual. In contrast, these same authors argue that agency, from a sociocultural perspective (following Vygotsky) is “intermental” as well as “intramental (ibid.: 337). More plainly put, agency is not simply an individual character trait or activity, but a contextually enacted way of being in the world.

It follows that agency is always a social event that does not take place in a void or in an empty wilderness. Even when an unsolicited individual act is agentive, it is socially interpreted (as well as often socially motivated). Lantolf and Thome (2006: 143) argue that agency ‘is about more than voluntary control over behavior.’ They explain that agency also ‘entails the ability to assign relevance and significance to things and events.’ From a sociocultural perspective, agency is shaped by our historical and cultural trajectories, hence Ahearn’s notion that agency is socioculturally mediated. Lantolf and Thorne also point out that agency can be exercised by individuals as well as by communities (ibid). In a classroom, for example, learners can act individually as well as in groups, or indeed as a whole class (as when a whole class negotiates workload or deadline issues with the teacher). Thus, students can speak from an ‘I’ as well as from a `we’ perspective.

The aim of this chapter is to examine the notion of agency and its dynamics in language classrooms. A better understanding of agency can help us find ways of creating learning environments favorable to its emergence and development. A related aim is to note some difficulties and possibilities in investigating agency from the classroom perspective, through reflective teaching, action research and classroom interaction analysis. In the process we will examine some notions that have been researched before and that may seem to be related to agency, such as control, initiative, autonomy and motivation.

Locating agency: when and where?

In this section I will present some extracts from diverse classrooms to examine some of the problems and possibilities involved in the analytical search for agency. It will become apparent that locating agency is by no means an easy or straightforward matter and, like other complex, multifaceted constructs (motivation comes to mind), we know that it is important but it may be difficult to observe and analyze in natural instances of its occurrence.

We will look at six extracts that I place roughly in an ascending order of agency, in a largely intuitive way. When I discuss each one in turn, some of the problems and ambiguities will be brought out and some of the difficulties of assigning agency in actual classroom activity will become apparent. [The extracts are on the accompanying slides n. 14 – 19] [Click here for comments on the slides]

As mentioned, I placed these extracts in a suggested ascending order of agency, so that (1) displays the least and (6) displays the greatest degree of agency. However, I am only doing so to initiate a discussion about what agency in the classroom is really about and to invite the reader to think about the various ingredients that may indicate more or less agency. It may well be, for instance, that some readers disagree with the ordering and consider the relative placement of one or more extracts rather arbitrary or even wrong. By analyzing the diverse reasons why one might disagree about the relative level(s) of agency displayed by learners, we may hope to gain a clearer idea of what the construct means and what its value in practical classroom terms might be. We are thus engaged in an exploratory exercise, aimed at an improved understanding of practice.

I will first give a straightforward (but not necessarily complete or accurate) gloss for each of the extracts, with an indication of sorts as to why they are placed in this order. The reader may provisionally go along at this point, or note doubts or disagreements. After the six glosses, we will attempt to work toward a deeper understanding.

Extract (1) takes place in an EFL class in Thailand. On the surface it clearly fails to engage the agency of the students. The teacher attempts to get a discussion going about moving to a foreign country but, in spite of repeated IRF-style questions, only manages to extract one contribution from a student, a single-word “Yes”.

Extract (2) takes place in a project-based multimedia classroom where FSL learners are learning to produce simple websites. Here the teacher is giving simple instructions that are carried out by the learner. One might say that agency is more in evidence here than in (1), since the learner is carrying out actions by moving her mouse and cursor on the screen, clicking and replacing one thing with another thing (mediated by the teacher’s instructions), whereas in (1) the learners were simply not responding overtly at all, with one small exception.

Extract (3) (from a secondary school CLIL class in the Netherlands) is a straightforward example of an IRF exchange, where the learner gives a response to a teacher’s question and this response is positively evaluated. This can be seen as another step up in agency, since the learner now has to formulate and articulate a thought. It is reasonable to argue that other learners may also be formulating and articulating thoughts, even if they do not express them verbally and that these inner-speech articulations are equally valid linguistic responses (Lantolf 2007). However, in this instance the learner has to employ some additional level of initiative, since the teacher does not nominate him specifically, but rather makes a general solicit (van Lier 1988) to which this learner chooses to respond publicly.

Extract (5) L1’s higher sense of agency consists of the fact the she volunteers to teach her neighbor how to make links on a web page and she proceeds to lead her fellow student step by step through the process. This is an instructional initiative that requires a strong sense of autonomy. However, we can also look at the event as one that exhibits a strong level of joint (collaborative) agency, since not only does LI offer to teach, but also L2 requests to be instructed.
In this sense the event manifests a quality that transcends the individual learner and that characterizes the joint activity of collaborative teaching/learning. Thus, we can see agency as not only an expression of individual volition, but also as a feature that can characterize a collaborative, co-constructed enterprise.

Finally, Extract (6) shows high levels of agency in a number of learners during a spirited discussion about the nature of language. Here learners are directly debating with one another, without having been prompted by the teacher and they are contributing to the debate because they feel they have  something to say and they have a strong opinion about it. As in extract (5), we can attribute agency to individual learners as well as to the entire speech cent. Indeed, we might want to postulate two types of agency: individual and collaborative, where the second could be argued to be of a higher level in terms of classroom quality, since it would draw together the creative energies and symbolic capacities of a larger number of learners. Especially in extract (6), one can argue that the entire class is energized by the spirited discussion about animal communication and language (see Lantolf 2007 about the enhanced symbolic resources that become available in dialogical activity).

Summing up, the 5 extracts can be categorized in terms of agency in the following manner:

(1) Learners are unresponsive or minimally responsive
(2) Learners carry out instructions given by the teacher
(3) Learners volunteer answers to teachers’ questions
(5) Learners volunteer to assist or instruct other learners and create a collaborative agency event
(6) Learners voluntarily enter into a debate with one another and create a collaborative agency event.

I suggest that it is indeed the case that, on the surface at least (in terms of overt, observable behavior), we have here an ascending scale or continuum of agency. Notice that I put the notion of ‘volition’ in (3)-(6) but not in (1) and (2). One might argue, of course that in (1) learners are purposely unresponsive, or that they are physically unresponsive, but mentally active and so on and that in (2) the learner merely ‘goes through the motions’ but does not in effect have a clue what she is doing or why.

There are many other observations one can make about any and all of the extracts. Nevertheless, I submit that the order given has a certain degree of intuitive (if not incontrovertible) plausibility. There are other things a casual observer might note, subjectively, upon observing these six events; the observer might say something like (for extract 1), ‘The learners were passive.’ Let’s list a representative adjective for each one, realizing that behind each adjective there may lie a host of assumptions based on perceived learner behaviors and presumed learner attributes (whether they be of the state or trait variety): 1) passive (2) obedient (3) participatory (4) inquisitive (5) autonomous (6) committed

This brief look at classroom interaction from the perspective of agency has raised a number of questions about the attributes of agency as well as some of the ways in which it can be observably manifested in the classroom by individual learners and groups of learners (as in extracts (5) and (6)). We will now relate this initial analysis to various learner characteristics and properties of the learning environment that have been discussed in the applied linguistics and general educational literature.

Problematizing agency: active, passive and other dichotomies

Victor Borge, the late Danish pianist and comedian, said that he had learned to speak Japanese by putting a tape recorder under his pillow and playing tapes while he slept. As a result he noted that he was quite fluent in Japanese, but unfortunately he could only speak it when he was fast asleep.

On the basis of the classroom extracts looked at above, we can say that agency is situated in a particular context and that it is something that learners do, rather than something that learners possess, i.e., it is behavior rather than property. This is in line with sociocultural perspectives such as those of Lantolf and Thorne (2006) and Wertsch et al. (1993), as noted above. Ahearn’s (2001: 111) definition of agency as ‘the socioculturally mediated ability to act’ also goes in this direction, if ability is not equated with competence (as an individual possession), but rather is seen as action potential, mediated by social, interactional, cultural, institutional and other contextual factors.

Agency can he related to issues such as volition, intentionality, initiative, intrinsic motivation and autonomy, all of which have been extensively studied in educational research (a good overview, using autonomy as the umbrella term, is Benson 2001). It is my view that these terms in practice refer to very similar phenomena and if we use agency as in umbrella term, they may all fall under that umbrella, indeed they may be regarded as synonyms – differing shades of meaning, connotation and pedigree notwithstanding. If agency is the family name, then all the other terms mentioned are family members, sharing strong resemblances (following Wittgenstein 1958).

There is always a strong tendency to look at complex constructs in terms of dichotomies, or binary oppositions. The most obvious one that comes to mind in this case is active-passive. Learners can be active or passive, the former being conducive and the latter being detrimental to learning. The problem that this binary view raises is that there may be many ways and degrees of being active or passive. In addition, we cannot rule out the possibility that some forms of being active may not appear to be all that propitious in terms of learning (see e.g., Allwright’s Igor,’ who participated eagerly and frequently, yet did not seem to make progress at all; Allwright 1980). Conversely, some forms of what we might perceive as passivity may not be quite as devoid of learning potential as casual observers might assume. Verbosity and overt participation have been unquestioningly associated with positive learning potential and the quiet student has often been ignored and regarded as a weaker student, even though no evidence exists to support such a view (see Ohta 2001).

There are other dichotomies in our field and they may be too numerous to mention. However, a few of them must be looked at in order to put the construct of agency on a more solid footing. Table 6.1 summarizes some of the major dichotomies and we will briefly examine how they influence the language learning field.

The comparative table allows us to get a clearer idea about what we may mean by agency in the classroom. It allows us to relate the definitions that we started out with to terms that are already familiar from classroom research and applied linguistics. I suggest that on the basis of this analysis we can propose three core features of agency which are broadly compatible with the definitions of Ahearn (2001); Duranti (2000; Lantolf and Thome (2006) and Wertsch et al (1993):
1) Agency involves initiative or self-regulation by the learner (or group)
2) Agency is interdependent, that is, it mediates and is mediated by the sociocultural context
3) Agency includes an awareness of the responsibility for one’s own actions vis-à-vis the environment, including affected others.

With this ground clearing in mind, we will now look at some central classroom issues that relate to agency. These issues, while not being in any way exhaustive, will bring us closer to establishing an empirical construct that can facilitate research and inform curriculum design and pedagogy. They include initiative, perception, identity and contingency. This will then be followed by recommendations for an agency-based (or action-based) approach to language education.

Initiative in interaction

Earl Stevick (1980: 19) defines initiative as ‘choices about who says what, to whom and when.’ Learners’ initiative does not necessarily conflict with a teacher’s control in terms of structuring the work and managing the classroom. Rather, a teacher’s control in this sense in fact enhances the possibilities for initiative to emerge.

For practical purposes I will use initiative as essentially synonymous with autonomy, self-regulation and self-determination (Deci and Ryan 1985; Williams and Burden 1997). I will, however, place it in an interactional and dialogical context, that is, a context of interaction among learners in the classroom. The question asked here is therefore, how is initiative demonstrated and enacted in classroom interaction?

The idea of initiative can be related to long-established questions in language teaching, such as responsive teaching (Bowers and Flinders 1990), autonomy-supporting versus controlling teaching behavior (Deci 1995), active participation and so on. We may also relate it to the crucial defining feature of handover/takeover in scaffolding (van Lier 2004), which ties scaffolding to the learner departing from a set script or ritual and introducing something new. What we are talking about is thus a learner who makes some effort, however small and seemingly insignificant, to be original, say something new and different, set off in an unpredicted direction. Within sociocultural theory and Bruner’s original notion of scaffolding, this departure, let’s call it initiative-taking, is a pedagogical moment, a teaching opportunity and a learning promise.How do we empirically capture this notion of initiative and tie it to agency? An obvious way to do so is by studying learner contributions in their context, i.e. in classroom interaction. How can we tell – unambiguously – from a transcript whether one contribution exhibits more initiative than another one?

To exemplify a measure that tries to get at both amount and type of initiative, I want briefly to revisit a coding and quantifying scheme I developed in my classroom interaction study (van Lier 1988) In this study, I allocated initiative in four areas:

  1. Topic work
  2. Selection to speak
  3. Allocation of speaker(s) or activity
  4. Sequencing the talk or activity.

Each of these categories was further subdivided into turn types, so that instances of initiative could be noted and tabulated. Thus, a variety of calculations can be carried out to show levels of participation and initiative for individual learners, groups, or types of activities or discussions (see also Kinginger 1994).

This level of analysis addresses aspects of the voluntary initiation of verbal behavior in socialinteractive contexts. The advantage of this procedure is that evidence of initiative (as one constituent aspect of agency) is located in the interaction itself (following conversational-analytical conventions), so that prior theoretical biases or agendas do not read into the data subjective notions that may or may not in fact be valid. However, there is more to agency than overt interactive behavior.

As mentioned in the introduction, Lantolf and Thorne (2006: 143) claim that agency also includes ‘the ability to assign relevance and significance to things and events.’ Similarly, in their model of ‘multi-factor’ interaction, Novick and Sutton (1997) include (in addition to choice of task and choice of speaker, which are similar to the focus of my scheme described above), also choice of outcome, i.e., a purposeful pursuit of a particular goal. In addition, Duranti’s model (1993) includes consideration of the possible effect of activity on others and oneself, as well as the realization that one’s actions are one’s own responsibility and thus will be subject to evaluation (including approval as
well as possible censure, criticism and so forth).

It is clear that this second level of analysis of agency that takes it beyond the interactional analysis of initiative will require a different research methodology, since evidence of such things as desired outcome, knowledge of potential impact and the critical dimensions of one’s actions may not be unambiguously locatable in the interaction (following ethnomethodological principles of context-free analysis), but may depend on features of the socio-cultural-historical context that are not immediately visible in the interaction but nevertheless influence, steer, or even determine it in subtle ways (Ratner 2006).

The difficulty for the researcher is to get at these influences, assignations of significance and awareness of consequences, without bringing to the data our own presuppositions and arguments, without clear evidence that they are actually there. We must always be cognizant of the fact that conversation analysis was developed precisely to avoid reading into data more than is actually given in the data. Harvey Sacks used to say, ‘If it’s a phenomenon, it must be in the interactíon.’ This is a prudent bit of advice and we are well-advised to take care not to read into the interactional data more than is strictly warranted by those data. Likewise, Aristotle already admonished the researcher to ‘let the data lead the way.’ But therein lies the rub: How can we address the other important factors mentioned without making unwarranted assumptions from the interactional data?

To address this problem I propose to add to the analysis of interaction a second level of analysis, namely analysis of narrative, introspective and dialogical data. Ethnographic and sociolinguistic interviewing can add an important level of depth to interactional analysis (Silverman 2001; Potter and Wetherell 1987).

In sum, then, the notion of initiative can be analyzed through interactional data, but in order to get at issues of content, purpose and higher levels of awareness (of possibilities, dangers and consequences) further methodological tools are needed in order to triangulate the complexities of the multiplex notion of agency.