Data Analysis: The Types of Pair Interaction
Excerpt 47: High interaction
4.3.5. Step 5: The Types of Pair Interaction
Based on an analysis of the features of episodes and the level of engagement, the data were analysed to describe the role relationship of each pair. Following Storch’s model of dyadic interaction, which presents four patterns of dyadic interaction characterised by various degree of equality and mutuality, the two indices, for equality and mutuality, were used to distinguish a different role relationship for each pair. As explained
regarding Storch’s model in Chapter 2, mutuality refers to the level of engagement with each partner’s contribution and equality refers to the degree of control or authority over the activity. In this study, according to the level of involvement, coded from non-
interactive or low to high interactive, the degree of mutuality was determined. For determining the level of equality, the coded data of the features of episodes were used:
how the pairs approached each activity, who initiated the episodes and who contributed to solve the problems.
Storch (2002a) suggested four patterns of dyadic interaction: collaborative,
dominant/dominant, dominant/passive and expert/novice. However, in this study, labels used in my attempt to describe the role relationships did not perfectly match Storch’s classifications. Five distinct types of interaction emerged from the process of data analysis and were labelled as collaborative, cooperative, dominant/passive,
expert/novice and expert/passive. These five types could be graphically represented by
four quadrants as shown in Figure 4.1.
High mutuality 4 Expert/Novice 1 Collaborative Low equality 3 Dominant/Passive Expert/Passive 2 Cooperative High equality Low mutuality
Figure 4.1. A model of pair interaction
Quadrant 1 represents a type of interaction labelled ‘collaborative’, in which equality and mutuality are both moderate to high. In a collaborative relationship, the pair tends to work together on most parts of performing an activity. Through negotiations or co- construction about unsolved problems, the problems can be resolved with the agreement of both learners. Furthermore, when one learner requests information or confirmation, the other provides elaborative explanation or gives feedback to the interlocutor’s suggestion with confirmation. When suggesting correction to the other,
the learner tends to elicit the partner’s self-correction, repeating the error with rising intonation or providing minimal assistance. In the process of engagement with each other, it shows mostly medium to high levels of pair interaction.
Quadrant 2 represents a type of interaction labelled ‘cooperative’, where the level of equality is moderate to high but mutuality is moderate to low. Without any superior status in each learner’s role, they tend to contribute equally to the activity. However, they do not engage with each other’s contributions very actively. Even if they perform a activity together, as Damon and Phelps (1989) explain, it seems that they take sole responsibility for one part of the problem with the activity being subdivided. Thus, in this role relationship, the learners seem to do the activity individually rather than together showing a low to medium level of interaction.
In Storch’s pattern, this quadrant is labelled as dominant/dominant in that although both learners contribute the activity, they have unwillingness to fully engage with each other and in their discourse, they do not reach an agreement but frequently show
disagreement on a problem. Unlike in Storch’s analysis, I have hardly found any
unwillingness to work together or disagreement. Rather, pair talk shown as cooperative interaction in this study reflects that learners seem to have sole responsibility to each part of the activity without providing or receiving assistance to each other or suggesting their own opinion on the activity. Therefore, the cooperative nature of interaction is labelled in quadrant 2 in relation to division of labour.
Quadrant 3 represents a moderate to low level of mutuality and equality. This type of interaction was labelled as ‘dominant/passive’. Dominant learners tend to lead the activity by initiating conversation, suggesting their own ideas and solving problems by themselves. The other learner, who seems to be passive, tends to accept their
partner’s suggestions or solutions without any challenges or contributions. Thus, there is very little negotiation between them and their talk reflects low level of interaction.
In addition, in this quadrant, an‘expert/passive’ relationship is also included, which was not presented in Storch’s frame, but indentified in Watanabe and Swain’s (2007) study. While the role of the passive partner has the same label as that in the relationship of dominant/passive, the role of the passive partner is different in the ‘expert/passive’ relationship. The expert learner, who is similar to the one in the expert/novice
relationship in quadrant 4, tries to encourage and invite the passive one’s participation but the passive peer does not actively engage in the activity and with the expert. Rather, the passive learner tends to simply respond to expert one’s request. Therefore, their level of interaction shows low to medium. This is different from the ‘dominant/passive’ relationship because the expert learner in the ‘expert/passive’ relationship invites their partner to participate in the problem-solving process rather than solving the problem alone like the dominant peer.
Quadrant 4 represents the type of pair interaction labelled ‘expert/novice’, in which mutuality is moderate to high but equality is moderate to low. In this type of interaction, although the expert learner seems to lead the activity, their partner is consistently encouraged to participate. With the expert’s assistance, the novice learner seems to actively engage in performing the activity. According to active engagement by both learners, the level of pair interaction appears to be medium to high.
It is important to note, as Storch (2002a) has acknowledged, that learners could
perform differently according to the type of activity, and multiple types of pair interaction might emerge during a single pair work. Therefore, pair interaction was coded
separately for each activity and if learners changed the nature of their interaction during a single activity performance, each type of pair interaction was coded because it was important to explore and understand the nature of pair interaction fully.
Once categories for the type of pair interaction were established, the coding of the data was checked by intercoder-reliability. A second trained coder who was not familiar with
the data was one of my colleagues. After explaining the general aim of the analysis and how to establish the descriptive categories, eight randomly selected transcripts
(representing almost 20 per cent of the pair talk data set) were given to the second coder. The reliability of the coding was calculated using simple percentage agreement, i.e. the coding of the two coders was compared. There was 87 per cent agreement, which is considered to be an acceptable level of coding reliability (Neuendorf, 2002).
4.4. Summary
In order to answer the first research question, namely how learners interact with their partner in language learning activities, this chapter provided a detailed description of how collected pair talk data were analysed and illustrated the analysis with
representative excerpts from the data. Using five steps of data analysis, the transcribed data were analysed. In the preliminary steps of data analysis, pair talk data were
segmented into types of talk and episodes. The next step was to analyse the talk for the salient features in the segmented data. Then, these features were used to examine the level of learner engagement. Based on the data analysed in these four steps, the five types of pair interaction were identified: collaborative, cooperative,