A description of the data and the instruments used for data collection

The data was collected during the winter term, over one unit of work lasting two and half months in total. The following data collection instruments were used:

Audio-recordings included recordings of ten pair interactions on four selected classroom tasks which were part of the 8th _{and 9}th _{grade syllabus. The number of tasks} carried out differed slightly across pairs, ranging between 10 to 12 tasks per pair. However, only four to six interactions per pair (total 52 interactions) were selected for the qualitative analysis, and four interactions per pair for the quantitative analysis (Chapter 5). Although some recordings were made by me during regular English lessons, the majority of them were made by students themselves during the study time lessons. It has to be noted that some data is missing due to students’ illnesses or technical failure to record interactions properly. This data addressed RQ1 and RQ2.

Although adding video recording could have served to incorporate the learners’ paralinguistic expressions such as gestures and facial expressions during their interaction into the analysis (see also Watanabe & Swain 2007, p.127), video recording could not have been conducted because of the inappropriateness of the seating order of the students, who face the wall when they work, and therefore the space for placing the video-camera was insufficient. In order to increase ecological validity, the tasks which

took place in the middle and towards the end of the whole period were chosen for analysis. Another reason for this step was that students will by then have had a chance to become used to the use of microphones/digital recorders (Philp et al., 2010, p. 264), which were provided to the students for the whole period of the unit of work.

Artefact collection includes student’s pieces of writing, students’ notes and classroom achievement tests which were conducted individually at the end of the unit of work. One of the tests contained one task which was identical to the task which students had been required to complete jointly.

Interviews (see interview questions in appendix A) were conducted mainly within the first two days after the last task had been completed. However, due to curricular reasons and learners’ illnesses three interviews were conducted in pairs, and two were conducted at the beginning of the second week after the last task. The aim of the interviews was to understand participants’ feelings and perceptions of their interactions with an older/younger classmate over the whole period. Learning about learners’ perceptions is especially important in this context because it is likely that perceptions of partner’s age or/and proficiency will affect behaviour on tasks at hand, and as such impact on learning (Watanabe & Swain, 2007). As such, interviews can reveal some salient features of the interactions, which cannot be inferred from audio-recordings only. For the sake of learning about learners’ perceptions of their collaborative work (RQ3), interviews with all participants were conducted. Interviews were audio recorded using individual microphones/digital recorders and transcribed using a transcription software f4. Conducting a stimulated recall interview one day after a chosen lesson with selected participants had also been considered. Stimulated recall is a type of introspective method, which is used to stimulate recall of participants’ thoughts when they were engaged in the activities (Gass & Mackey, 2000). This method could have been very useful in order to understand participants’ behaviour and their thinking processes during their interaction which may not have been detectable from audio-recordings and their transcripts (see alsoWatanabe & Swain 2007). However, there were two reasons for not using this method. The first was the lack of time and the organizational difficulties as the researcher is also the teacher of the classes, and because learners had subsequent lessons. The second was that the main purpose of conducting the interviews was not to gain understanding of their immediate perceptions and feelings of work on one task at

hand but to learn about general perceptions and feelings of their interactive work over a longer period of time.

With regards to the interviews, I would like to align with the constructionist stance which treats interviews as “collaborative or interactional events in which the interviewer or moderator plays an important, participative role” (Edley & Litosseliti, 2011, p.155). In other words, I do not regard an interview as “as mechanism by which one party (i.e. the interviewer) extracts vital information from another (i.e. the interviewee)” but as a social interaction on or reciprocal or two-way exchanges” (p.157). However, I am aware that one of the main weaknesses of the interviews is that the participants attempt to present themselves in a positive light which may hinder finding the necessary facts (p.157). Moreover, participants may supply “what they imagine is the ‘right’ or ‘sought after’ response” (p.163). It follows that it is dangerous to presume that this is what they really think (p.163). I opted for a semi-structured interview given that while I wished to ensure some level of order and the wording to be used during each interview, I also strived for a more “free-flowing and indeterminate process” (p.158). Interviews were piloted twice. While piloting, I became aware that I tended to impose predetermined questions on my students, which seemed to have hindered the flow of talk as well as my neutrality as an interviewer. Therefore, for the main study, I attempted to preserve neutralness of our talk, and make it seem as if it was a conversation, rather than eliciting some scientific data using a set of prescripted questions. I tried to remain neutral during the interview process, to avoid opinions, to ask simple and open questions and to make my students feel as comfortable as possible. However, because participants were my students, it also became obvious that some tended to give what Edley and Litosseliti (2011) call ‘standardized’ answers based on what they thought was ‘a right’, ‘expected’ answer in order to please their teacher. For example, when asked about their perceived benefits of working with an elder or younger partner, some tended to give answers based on what they had heard from other teachers or the headmaster when discussing these issues. It follows that I do not claim that absolute neutrality was achieved, or that my identity as a teacher together with my behaviour and questions had no bearing or impact on what the participants said. Despite their limitations, interviews provided some important insights into students’ perceptions and feelings, and were also helpful in order to understand what was actually happening during interactions. Finally, the transcriptions of the interviews were compared with the audio recordings of the

interactions. As such, combining audio recordings and interviews achieved method triangulation and content validity.