Researching classroom interaction and talk
David Wray and Kristiina Kumpulainen
Introduction
The social and contextual nature of human learning has received great emphasis in research on learning and instruction (Anderson, et al., 1997; Greeno, 1997). Attention has been paid to the practices, processes and conditions leading to the social construction of knowledge in different learning situations (Fisher, 1993;Lemke, 1990; Palincsar, 1986; Tuyay, Jennings, & Dixon, 1995). The focus of analysis has been extended from external factors influencing learning processes and achievements to the student’s participation in and evolving interpretations of the learning activity (Grossen, 1994; Perret-Clermont, Perret, & Bell, 1991). In the midst of these changes in emphasis, new methodological questions concerning the analysis of classroom interaction and learning have arisen. Questions to which researchers have been trying to find answers are, for example:
What qualitative differences can be found within and between interactive activities across a variety of learning contexts?
What typically happens during classroom interaction?
How can practitioners and researchers develop and strengthen their understandings about optimal learning within inherently social settings (e.g. classrooms)?
This chapter will review and critique some approaches to studying and analyzing classroom interaction, and especially the discourse which characterizes this. It is based on an account of a programme of research into classroom interaction and social learning, and uses the methodological decisions made during that programme as a set of pegs on which to hang discussion of methodological issues in this field.
By the end of this chapter, you should have:
Gained an understanding of the complexity of investigating socially shared learning practices;
Considered some methodological issues emerging from researching classroom interaction and talk;
Become aware of the need to move from an analysis of classroom talk to a more holistic analysis of classroom interaction;
Understood the use of analytical maps that involve functional, cognitive processing and social processing analyses of classroom interaction and talk.
The framework of analysis
The analysis framework described here emerged as a result of a number of studies of primary-aged learners’ interactions whilst working in peer groups on various educational tasks in Finland, Greece and the United Kingdom (Fourlas & Wray, 1990; Kumpulainen & Wray, 2002). The main goal of these studies was to investigate the nature of students’ social activity, particularly verbal interaction, in different small group work learning situations. The initial development of the method concentrated on the functions of students’ verbal interaction as a basis for investigation of students’ roles as communicators and learners in teacher-centred and peer-group centred classrooms (Fourlas & Wray, 1990). This functional analysis method was later piloted, modified and applied by Kumpulainen (1994, 1996) in a study that investigated students’ social interaction during the process of collaborative writing with a word processor. Due to its fine-grained categorisations, the functional analysis method was felt to give a structured overview of the nature and quality of students’ verbal interaction in this learning context.
Despite the potential of the analysis method, in more recent studies of peer group learning this functional analysis of verbal interaction was found to be inadequate as a means of unravelling the complexities of socially shared learning processes. Firstly, there seemed to be a need to develop a descriptive system of analysis, which took a more holistic and multidimensional perspective on interaction. Consequently, the analysis of verbal interaction alone seemed not to be sufficient for this. Secondly, it seemed important that more attention be paid to the moment-by-moment nature of interaction in order to highlight the situated processes of meaning-making and knowledge construction within peer groups. Thirdly, it seemed important to take the individual and the group as units of analysis in order to investigate the types and forms of participation within peer groups.
In the analysis method subsequently developed, the dynamics of peer group interaction were approached from three analytic dimensions.
- The first dimension of the analysis, termed functional analysis, focussed on the character and purpose of student utterances in peer group interaction. It characterised the communicative strategies used by participants in social interaction.
- The second dimension, cognitive processing, examined the ways in which students approached and processed learning tasks in their social interaction. It aimed at highlighting students’ working strategies and situated positions towards learning, knowledge and themselves as problem solvers.
- The third dimension of the analysis, social processing, focused on the nature of the social relationships that were developed during students’ social activity. This included examining the types and forms of student participation in social interaction.
Before discussing the theoretical and methodological background of these methods and highlighting the analytical framework with some empirical examples, we will firstly review some of the other analysis methods used to study peer interaction that have contributed to the present analytical approach.
Investigating collaborative interaction in peer groups
Peer group interaction has been studied quite extensively in different contexts in and out of school. The research objectives and methodological solutions have been diverse, being linked with the research goals and theoretical perspectives adopted by the researchers.
One large group of studies focusing on peer interaction from the educational perspective is located in the systematic tradition, often referred to as process-product studies of peer interaction (e.g. Joiner, Messer, Light, & Littleton, 1995; Light, et al, 1994; King, 1989; Teasley, 1995; Tudge, 1992; Webb, Troper, & Fall, 1995). In these studies, peer interaction is analysed with coding schemes, which categorise interaction into pre-defined categories. Variables such as student achievement and performance are statistically linked to the frequency of categories as identified in the data. Usually, the development of the actual interaction process or meaning making in interaction is not the prime interest, but the focus is rather on some specific features of the interaction and their relationship to student learning or achievement. Consequently, the process of interaction over a period of time is not highlighted by such studies. The situated nature of interaction, as represented by the contextual features impinging upon it, also often receives only cursory inspection. One advantage of process-product studies is that they enable the analysis of large amounts of data and use publicly verifiable criteria to make their categorisations.
Probably the best known of such category systems is the Flanders Interaction Analysis (Flanders, 1970). This system has been used extensively in classroom observation studies (Wragg, 1999; Newman, 2004). It has two main uses. Firstly, it was intended to provide evidence of the differences in teaching patterns that distinguish one teaching style from another and, secondly, it has been used to try to explain differences in learning outcomes associated with different styles of teaching.
The Flanders Interaction Analysis Categories (FIAC) consist of 10 categories of communication, seven used when the teacher is talking and two when a pupil is talking (and one when there is silence or confusion). An observer using the system makes timed observations, usually every three seconds, and categorises the behaviour which he/she observes at each point into one of ten categories. These categories are shown below in Table 1.
Table 1: Flanders' Interaction Analysis Categories (FIAC)
Teacher Talk / Indirect Influence / Accepts feeling: accepts and clarifies the feelings of the students in a non-threatening manner. Feelings may be positive or negative. Predicting and recalling feelings are included.Praises or encourages: praises or encourages student action or behaviour. Jokes that release tension, not at the expense of another individual, nodding head or saying 'uh huh?' or 'go on' are included.
Accepts or uses ideas of student: clarifying, building, or developing ideas or suggestions by a student. As teacher brings more of his own ideas into play, shift to category five.
Asks questions: asking a question about content or procedure with the intent that a student may answer.
Direct Influence / Lectures: giving facts or opinions about content or procedures; expressing his own ideas; asking rhetorical questions.
Gives directions: directions, commands, or orders with which a student is expected to comply.
Criticises or justifies authority: statements, intended to change student behaviour from non-acceptable to acceptable pattern, bawling someone out; stating why the teacher is doing what he is doing, extreme self-reference.
Student Talk / Student talk - responses: talk by students in response to teacher. Teacher initiates the contact or solicits student statement.
Student talk - initiation: talk by students which they initiate. If 'calling on' student is only to indicate who may talk next, observer must decide whether student wanted to talk. If he did, use this category.
Silence or confusion: pauses, short periods of silence and periods of confusion in which communication cannot be understood by the observer.
A very useful website explaining and exemplifying the Flanders system can be found at Nova South Eastern University Centre for Teaching and Learning (
Other research traditions have produced quite different approaches to the analysis of peer group talk and learning. Barnes and Todd (1977, 1995), for example, developed an analytic system for studying peer group talk which was “grounded” in the data, as opposed to being derived from a pre-existing network of categories. Consequently, their system did try to take account of the context in which peer talk was occurring. In their analysis Barnes and Todd were interested in the actual processes of interaction and the ways students developed and constructed knowledge without direct teacher presence. In analysing the dialogue amongst the groups of students, they considered types of talk and their impact upon the construction of meaning during group interactions. This demanded an analysis of both the social and cognitive functions of conversation. They developed a system describing speech acts, which was based on two levels.
Level one consisted of:
discourse moves (such as initiating, eliciting, extending and responding);
logical processes (such as proposing a cause, advancing evidence, negating, suggesting a method, evaluating).
Level two comprised:
social skills (such as competition and conflict, supportive behaviour);
cognitive strategies (such as setting up hypotheses, constructing new questions);
reflexivity (such as monitoring one's own speech, evaluating one's own and others' performance).
They identified ‘exploratory’ speech characteristics such as hesitation and changes of direction, tentativeness in voice intonation, assertions and questions made as hypotheses rather than direct assertions, invitations to modify or surmise, and self-monitoring and reflexivity. They went on to propose conditions for collaborative work amongst groups in classrooms, based on this empirical evidence. Further analysis (Barnes and Todd, 1995) provided descriptive examples of the “... four categories of collaborative moves: initiating, eliciting, extending and qualifying”.
Despite some limitations in the analytical system and the tools used for data collection (tape recorders only, thus losing any information from non-verbal elements of communication), Barnes and Todd’s work made an important contribution to the analysis of peer talk since it integrated ideas from discourse and conversational analysis with research on learning and instruction. Several studies have used the Barnes and Todd frameworks to enquire into classroom interaction (e.g. Edwards, 2005).
Many other methods of analysis of peer group interaction, either with distinct categories or more interpretative “modes”, have been developed in the past twenty years and to review all of them here would be impossible. One important analytic approach which needs to be discussed, however, since it has contributed greatly to our understanding of children’s talk during small group learning is that developed by Fisher (1993), Mercer (1994, 1996, 2000), and Mercer & Littleton (2007). What is interesting in this approach is that it tries to investigate how children use talk to think together, thus, it uses a group as a unit of analysis, not individual children. By taking a sociocultural approach to children’s talk, it tries to show that particular ways of talking permit certain social modes of thinking. The analytic framework was derived from analyses of children’s talk during collaborative peer group learning and it includes three distinct modes of talk which characterise different ways of thinking together. These are:
disputational mode, characterised by disagreement and individualised decision making;
cumulative mode, consisting of positive but uncritical decision making;
exploratory mode, which is seen as the most effective mode of speaking in fostering critical thinking and cognitive development (Mercer, 1996). This is characterised by constructive and critical engagements, including argumentation and hypothesis testing.
Theoretically, this analytical framework makes an important contribution to our increasing understanding of the different modes of talk and social thinking in peer group situations. One of the limitations of the method, though, can be found in the fact that the unit of analysis is the group - the method does not take into account individual students’ participation in the “social modes of thinking”. Consequently, the method does not highlight how the different types of social thinking are actually constructed within peer groups. Moreover, by concentrating mainly on students’ talk, the analysis may not always give a complete picture of the nature of knowledge construction in peer groups. Instead, a more dynamic approach to peer interaction is necessary, which focuses on the whole interactive context and its development, including non-verbal communication and the use of different tools, before we can unravel the processes and conditions for learning in peer group activity.
Towards a new analytic method
In the remainder of this chapter we will outline and discuss a descriptive system of analysis for investigating the situated dynamics of peer group interaction. Of particular importance are the mechanisms through which the social and cognitive features of peer group activity operate. The theoretical grounding of the analysis framework was informed by sociocultural and sociocognitive perspectives on interaction and learning (Cole, 1996; Resnick, Levine & Teasley, 1991; Wertsch, 1985, 1991), whereas the methodological solutions were greatly influenced by the work of Barnes and Todd (1977, 1995), Mercer (1994, 1996) as well as by interactional ethnographers (Green & Wallat, 1981; Green & Mayer, 1991; Tuyay, et al, 1995).
In this method, learning is seen to take place as a result of individuals’ active participation in the social practices of their environments. Learning is viewed as an interactional process that requires an understanding of language, and other semiotic tools, as both personal and social resources (Cole, 1996; Halliday & Hasan, 1989; Wells & Chang-Wells, 1992). Peer interaction is treated as a dynamic process in which language and other semiotic tools are used as instruments of communication and learning. Interaction is seen as a complex social phenomenon composed of non-verbal and social properties in addition to its verbal characteristics. Peer discourse itself is not treated as representing a person’s inner cognitive world, nor even as descriptive of an outer reality, but rather as a tool-in-action shaped by participants’ culturally-based definitions of the situation (Edwards, 1993; Edwards & Potter, 1992).
The application of the method involves a microanalysis of evolving peer interactions by focusing on three analytic dimensions, namely the functions of verbal interaction, cognitive processing and social processing. Whereas the functional analysis concentrates on students’ verbal language, the analyses of students’ cognitive and social processing focus on interactive dynamics as they occur across the participants. Consequently, a group is taken as a unit of analysis. The three dimensions are treated separately for analytic purposes, although it is recognised that they are closely linked together in a complex way. In actuality the dimensions cannot be separated since each element gives meaning to all others and simultaneously obtains meaning from them.
Dimension 1: Functional analysis of verbal interaction
The functional analysis of students’ verbal interaction focuses on the purposes for which verbal language is used in a given context. It investigates and highlights the communicative strategies applied by individual students whilst taking part in interaction (Halliday & Hasan, 1989). Analysis of this nature often concentrates on the illocutionary force of an utterance, that is, on its functional meaning (Austin, 1962; Edwards & Westgate, 1994). The functions for which students use their oral language are closely linked to the topic of discussion as well as to the individuals’ expectations and evolving interpretations of the situation shaped by the sociocultural context of the activity. The functions of language used in the course of interaction serve both intra- and interpersonal purposes. On the one hand, the purposes and intentions carried by means of verbal language serve an ideational, i.e. cognitive function. On the other hand, they serve an interpersonal function relating to the personal and social relationships between the interactors (Halliday & Hasan, 1989).
The identification of language functions in peer interaction takes place on the basis of implication, that is, what a speaker can imply, suggest or mean may be different to what the speaker literally says. Consequently, the functions are not identified on the basis of specific linguistic forms. Rather, they are identified in context in terms of their retrospective and prospective effects on the actual discourse both in terms of content and form. An understanding of the functions for which students use their verbal language in interaction is greatly assisted by data gathered from direct observation, video recording and student interviews. The functions of peer interaction are the minimum units analysed in the system. They are identified on an utterance basis and defined in terms of source, purpose and situated conversational meaning. An utterance is viewed as a meaningful unit of speech, that is, a message unit. The boundary between each utterance is linguistically marked by contextual cues. Given that an utterance may serve multiple functions, more than one function can be recorded for each utterance.
Examples of language functions identified in peer group interaction across learning situations are the Informative, Expositional, Reasoning, Evaluative, Interrogative, Responsive, Organisational, Judgmental (agrees/disagrees), Argumentational, Compositional, Revision, Dictation, Reading aloud, Repetition, Experiential, and Affective functions.