Data collection methods

This section describes the data collection. First, I discuss the piloting of the collection methods. I then turn to the recording of the study events, and describe the importance of using naturally-occurring interactional data. The last subsection moves on to the process of conducting the interviews.

Pilot phase

Before embarking on the actual data collection, the collection methods were piloted. A small group of students studying medicine were recorded during three interrelated group

70

meetings and interviewed for their experiences and views of studying in English.⁴⁰ The study event, however, was not a good case for the wider study, because all the three students were L1 speakers of Spanish, and English was not a lingua franca between the students, but only between the students and the Finnish-speaking teacher.

The pilot study event being a course in medicine, ethical issues of data collection were highlighted. The course teacher and the students were asked for consent beforehand, and this practice was used in the actual data collection as well. But especially collecting participants’ written texts dealing with actual patient cases, as well as gaining access to the students’ interaction with their patients became complicated because of patient confidentiality. These problems, however, did not persist for the final data collected from other fields of study.

The interviews were done to test interviewing as a technique as well as to try out interview questions. It became clear that even though the SELF project had already been introduced to the students, it was important to introduce the project again before each interview. The open-ended format of the interviews proved to be practical, but certain question formulations in the interview guide were problematic in that they were too abstract or complicated. These questions were either revised or left out from subsequent interviews. For instance, in example 4.1, the interviewer’s question appears to remain on a too abstract level – it would have made more sense to link the question to a specific encounter (e.g. the recorded study event) or to first inquire about the interviewee’s daily activities and relate the question to them. These kinds of concretisations of interview questions were developed throughout the process of doing the interviews (see Conducting the interviews below).

(4.1)⁴¹

IR: what about er have you ever been in a situation where cultural differences would have created misunderstandings,

IE: (w-) n- no i cannot @remember@ but maybe some (yeah) someone say some sentence or and you cannot understand because it’s not the same humour for example (in spanish) [(and in another country and)] <IR> [mhm okay mhm]

</IR> i mean you say something and (this guy is) sorry i cannot understand but no i cannot say any example (xx)

(pilot interview, IE: L1 Spanish)

40 I conducted the pilot phase in collaboration with Pirjo Surakka-Cooper.

41 For transcription conventions, see appendix A.

71

The interviewer’s question in example 4.1 also includes the word misunderstandings, which has a negative connotation, and it narrows down the possible reasons for misunderstandings to cultural differences. A more informative way would have been to give the interviewee a chance to talk about her experiences of intercultural encounters on a more general level first and only then introduce the possibility of misunderstandings.

Listening through the interviews also forced me to focus on my own expressions more.

For instance, I tended to repeat what about (then), also found in the beginning of example 4.1. Awareness of such habits helped reduce them in subsequent interviews.

Recording of the study events

The study events in the final data took place as part of the normal course of events in the study programmes, and were thus not arranged for the purposes of this study. Naturally-occurring study events were chosen in order to see what happens in situations where participants use ELF for real-life purposes and where what the participants do has consequences to them. In experimental conditions, it is possible to control external elements (e.g. background noise) and social variables (e.g. linguistic background of the participants), and even to control the interactional situation itself in order to elicit speech that includes phenomena that interest the researcher. However, interaction in experimental settings tends not to have social or interpersonal consequences for the participants, which means that it may not matter for the participants whether and how well they finish a task set by the researcher. Also, the topics are typically set by the researcher, rather than the participants themselves. This means that elicited data does not tell us what actually happens in real life when people come together to communicate in order to achieve specific outcomes that matter to them more than to the researcher. In real-life events, speakers are first and foremost participants in the events, not informants for linguistic study.

I did most of the recordings myself, but also received help from SELF project assistants. The person(s) recording the study events were present during the events, but did not participate in them. The recordings were done with a small Olympus audio recorder that could be placed relatively unnoticed on a side table or a window sill of the recording room, but which also meant that to ease processing of the data, someone had to be present to take notes of the succession of the speakers as well as other notes that could help interpret the events later. This was done as non-participant observation. Only if directly addressed by the participants did I or the assistant take part in the interaction,

72

and then too, the purpose was to keep the intervention to the minimum so as not to affect the course of the interaction too much.

Attending the recording sessions was not only done for the mechanical purpose of helping the processing of the data, but I also wanted to familiarise myself with the groups and the study event practices personally. This is typically done in ethnographic studies to gain an overall image of the practices and to be able to later zoom in and focus the data collection as well as the research itself (Blommaert & Jie 2010). For instance, the student group I recorded for this study was part of a course which I also partly recorded, but the course lectures did not end up as data. The lectures included only little discussion, which means that the material was less suitable for the purposes of this study. The ethnographically-informed data collection (frequent face-to-face contact with the participants) also made it easier to convince the students and the teachers to take part in the research interviews (even if this still remained somewhat of a challenge).

I asked for the participants’ consent (see appendix B for sample consent forms) in the beginning of the first events, which means that the participants were aware of the recordings taking place. I identified myself as a researcher of English and a representative of the SELF project dealing with English as a lingua franca and studying in international settings (the same goes for the project assistants). It is clear that this language expert background may have prompted the participants to pay more attention to their language, but this did not change their reasons for attending the study events. In addition, recording several sessions from the same speakers surreptitiously would have been impractical and ethically problematic, and asking for consent only afterwards would have risked the whole process had the participants felt cheated and denied consent. In relation to everyday conversational data, though, it has been argued that only surreptitious recordings provide totally natural speech (e.g. Warren 2006). This is plausible especially considering the beginnings of conversations that would have to be interrupted by the asking of consent.

Moreover, in conversations, participants are free to choose the topics to be discussed, and the presence of a recorder may well affect the choice of topics. However, while the participants in this study occasionally refer to the recording or the presence of the researcher(s) in the room, which shows their occasional awareness of the recording, this does not change their need to pursue the aims of the study event. It might then be argued that the kind of specialised discourse analysed in this study is not as easily influenced by the recording process as conversational data. Moreover, the successive recordings are likely to have reduced the effects of the recording process on the discourse (see Smit 2010: 94–95).

73 Conducting the interviews

Some of the interviews were done by me alone, and some I did in collaboration with a SELF project assistant. The interviews for the student group work and the teacher-led course I conducted with Pirjo Surakka-Cooper so that we were both present and active in the interview situations. We also collaborated in doing the speech event recordings, which means that the interviewees knew us beforehand. In these interviews, both interviewers posed questions, although either one or the other tended to take the lead. The rest of the interviews I conducted on my own; some of the sessions were, however, attended by Anni Holopainen who had done related study event recordings and who took an interest in interviews as a research method.

Most of the interviews were conducted with one interviewee, but I also tried pair and group interviews to see whether they would make the situations more informal and discussion-like. An interview with two Finnish students from the student group work turned out to support the assumption as the students fuelled each other’s responses.

Interviews with students from the teacher-led course further suggested that interviewing more than one student at a time may be beneficial if the students have trouble expressing themselves in the language that the interview is conducted, since the students can rely on each other for translations and clarifications. However, pair and group interviews do not necessarily give the interviewees equitable representation. To ensure that all participants had equal opportunities to express themselves, I then conducted the interviews with the members of the guided group individually.

The interviews were conducted mainly in English, but Finnish was used with those interviewees whose L1 was Finnish and who thus shared the L1 of the interviewer(s).

This reflects the practice of the study communities: those who shared an L1 (primarily) used their L1 when communicating with each other, and otherwise the choice was English or some other shared language.

All the interviews conducted for this study are semi-structured thematic interviews designed around the themes of studying and working in an ELF setting (see appendix C for the interview guides). The interviews were audio recorded, and after each interview, I wrote down notes about the interview situation, including perceptions of the interaction (e.g. what was it like to conduct the interview, how did the interviewee(s) seem to react to the interview). The notes provide brief accounts of each interview as a speech event, and are thus important in interpreting individual responses as parts of the interviews as a whole (see Briggs 1986: 104).

74

The interviews followed pre-prepared guiding questions, a so-called interview guide (see Dörnyei 2007), but the format was open-ended, which means that I (or my co-interviewer) could adapt the interview to the situation, reacting to the interviewee’s comments. As suggested by Dörnyei (2007: 137), the interview guides were used to ensure that the topic was properly covered, and they also included suggestions for appropriate question wordings and offered useful probe questions and comments to bear in mind.

The aim was also to keep the interviews as informal as possible so as to ensure that the interviewees could explain their experiences and perceptions freely. For the same reason, I did not seek to be “neutral” in the sense that I would restrict from reacting to the interviewees’ responses, but rather tried to encourage the interviewees to share their experiences freely giving them carry-on and reinforcement feedback typical of normal spoken communication (for feedback in interviews, see Dörnyei 2007: 142). It is thus acknowledged that an interview is a communicative situation where both the interviewee’s and the interviewer’s contributions influence the course of the interaction.

I developed the interview guides during the data collection. This is in accordance with ethnographically-informed data-driven research where the collected material steers the research towards certain directions and where fieldwork is not only data collection, but also a learning process for the researcher (Blommaert & Jie 2010). For instance, my experiences of interviews with participants of the student group and the teacher-led course suggested that it was hard for the participants to respond to some of the questions because they remained on a too abstract level. As a solution to better ground the interview questions to the interviewees’ daily lives, I asked the interviewees from the guided group to fill in two clock faces representing the hours of the day (see appendix D), and mark in the clock faces what they do during a typical day, with whom and what language(s) they use in the activities (see Satchwell 2005, Mäntylä, Pietikäinen & Dufva 2009). The responses were then used to inquire more about the interviewees’ use of English. Further, questions directly related to the discussion group meetings that the interviewees attended were done more systematically than in the earlier interviews. These changes appeared to make it easier for the participants to talk about their experiences.

I now turn to the methods of analysis.