6. Methods
6.3 Phase 2: Qualitative component
6.3.3 Transcription
As noted by Brinkmann and Kvale, transcription is not straightforward and simple, but “an interpretative process” (2015, p. 203). In the interview setting, a lot of information is transferred between interviewer and interviewee in the form of body language – elements that are lost when the oral interview is turned into written form. In addition, oral language is a different narrative mode than written language; the two represent different “language games and […] cultures” (Brinkmann & Kvale, 2015, p. 204). Pauses, fillers, and sounds that cannot be distinguished as clearly articulated words are extremely common, and in the context of the oral conversation they make sense to the participants. When read in a transcript, though, they could create the impression of indecisive and ineloquent conversation partners. The many different Norwegian dialects that sound very different, and in some cases employ different words for the same concept, is another issue worth considering, as writing out the participants’ dialects could endanger the teachers’ anonymity, since the dialect linked to other information could make them identifiable. My main concern when transcribing the teacher interviews was therefore to balance the verbatim oral (that is unique for each teacher) with a standard written style that veiled the teachers’ identities and made their responses readily accessible to readers.
When I transcribed the teacher interviews, I followed the standardized spelling and grammar rules of Norwegian bokmål for the interviews conducted in Norwegian, and
American English for the interviews conducted in English. This had the greatest consequences for the Norwegian-language interviews: the respondents spoke using several different dialects, some of which were closer to Norwegian nynorsk than bokmål. Some used other words for concepts than those that are commonly used in bokmål, which means that I not only altered the spelling and grammar of their utterances, but also in some cases the vocabulary. In the English-language interviews, I followed the same principle in order to protect anonymity. In the chapters that follow, extracts from interviews conducted in Norwegian have been translated into English. This means that the translated extracts have gone through another round of interpretation, which could, potentially, bring them further away from the teachers’ original meanings as expressed through their oral dialects. In order to make sure that the translations were as close to the original utterances as possible, I listened to the audio recordings after translating, and revised the translations one more time.
This type of transcription focused on the content of what the teachers said, not the linguistic form of their utterances. I put the non-verbal sounds that the teachers made in brackets: for instance [laughs], [sighs], [yawns], [incomprehensible sound]. Additionally, I used (…) to denote long pauses. Whenever the teacher and I would talk simultaneously, the transcript always included what the teacher said. What I said was included in the instances in which I completed the phrase I had begun; in most of the cases, I would pull back and let the teachers finish what they had begun talking about, and in these cases I did not record my own, brief utterances. As each interview lasted between one and three hours, one interview could result in more than 30 pages of transcribed text. For these reasons, I decided not to include stressed words, perceived mood, tone, volume, and speed of voice in the transcription (Cohen et al., 2011). In order to navigate the texts easily when analyzing them, I needed the meaning of the teachers’ utterances to be clearly conveyed in writing. I believe that the complexity level at which I placed the technicalities of my transcription allowed me to do that.
6.3.4 Reliability, validity, and analysis of data
In qualitative studies, reliability centers around “the craftsmanship and credibility of the researcher” (Brinkmann & Kvale, 2015, p. 283). The results must be deemed trustworthy by the reader, and this can only be conveyed through broad descriptions of all parts of the study in question. In the interview situation, the biggest potential threat to reliability is the behavior of the researcher. The questions in qualitative, semi- structured interviews are not set in stone even though an interview guide is followed, and the interviewer is open to changing the structure and content of each interview along the way. Therefore, there is a risk that the interviews might become too different, and be difficult to analyze and compare. However, as has been discussed, researchers need to maintain flexibility in order to discuss the unique experiences of each interviewee; this is particularly important when examining teacher beliefs, as it is the “individual, subjectively true, value-laden mental constructs” (Skott, 2015, p. 19) that we wish to explore. In order to balance these concerns, Brinkmann and Kvale suggest operating with two levels in the interview guides: the broad, thematic research questions that are explored on one level, and the specific questions that probe into these issues in a more detailed manner on the other level (2015, p. 158). This allows the researcher to make sure all the broad, thematic concerns are addressed in all interviews, whereas the specific questions that are asked might differ according to each individual situation. This balancing act can be challenging and hard to plan for as the researcher does not know how the interviews will turn out in advance, and this is where the interviewer’s craftsmanship becomes important: a good interviewer should be knowledgeable, structuring, clear, gentle, sensitive, open, steering, critical, remembering, and interpreting (Brinkmann & Kvale, 2015, pp. 194-195).
In this study, each teacher was interviewed twice, which allowed for the asking of follow-up questions and probing deeper into issues found to be important in the preliminary analyses conducted between interviews 1 and 2. Leading questions were avoided by asking questions about the main issues in as neutral a manner as possible. However, the teachers were sometimes asked leading questions towards the middle and end of the explorations of the main issues in order to see whether the interviewees were
consistent in their answers, and in order to verify my interpretations of their responses along the way. Brinkmann and Kvale point to a limited use of leading questions as being a useful way of checking the reliability of the answers provided by the interviewees (2015, p. 200). It is important to note that although the power balance was not as lopsided in these interviews as it would have been if the interviewees were not adults, my role as a researcher representing an institution of higher education could provide me with a position of authority in the eyes of the teachers. However, my impression is that the teachers overall were not afraid of expressing their opinions; in the second interview in particular, some of them disagreed with my interpretations and voiced clear criticisms regarding the novel they had read.
Validity in qualitative studies is mainly concerned with internal validity, namely whether “the explanation of a particular event, issue or set of data which a piece of research provides can actually be sustained by the data” (Cohen et al., 2011, p. 183), and especially whether there are any possible alternative explanations for the interpretations presented in the study. As is the case with reliability in qualitative interviews, validity is also to a great extent dependent on the interviewer’s craftsmanship. In this study, triangulation of data, interviewing each teacher twice, and member validation have strengthened the validity of the research.
Triangulation of data was especially important in the first interview, as the teachers volunteered year plans and information about the resources they used in their English lessons in addition to partaking in research interviews. These texts provided an additional source of information that could tell me more about the literature choices and the context in which literature was taught in the teachers’ classrooms, than only the teachers’ own oral accounts of their choices and practice. Another important feature of this study that strengthened its validity was, as discussed earlier, the fact that each teacher was interviewed twice. The third key component of validity checking in the interviews was member validation (Brinkmann & Kvale, 2015, p. 290). All eight teachers received drafts of the teacher profiles that were to become part of the dissertation, to ensure that they had not been misrepresented. The teachers responded positively to the drafts, and approved the versions that are found in chapter 7.
Coding was a central part of the analysis of the qualitative data, both when working with the teachers’ year plans and teaching resources, and when working with the interview transcripts. The coding of year plans and teaching resources, mostly textbooks, entailed highlighting the literary texts present and denoting the genres they belonged to. The lists of works that emerged from this process were used as I prepared for the interviews, as well as in the subsequent analyses. The coding of the interviews was more complex. Following the first round of interviews, the following codes were used to organize the transcripts thematically:
Background information Specific literary texts Literary genres
Student groups: vocational and general studies Textbooks and other teaching resources
Criteria when choosing texts
Purpose(s) of literature in subject English
Core curriculum, subject curriculum, competence aims, and examinations Literary canon
Literary texts’ geographical origin and/or setting Literary quality
Violence, profanity, and sex in literary texts
Preliminary coding was conducted before the second round of interviews. In the analyses of the second interviews, I used the following codes:
Follow-up questions to interview 1
Dystopian novel: teacher’s views as a reader Dystopian novel: teacher’s views as a teacher
Dystopian novel: teacher’s responses to reviews, teaching guide, and/or other resources
The follow-up questions to interview 1 were coded according to the interview 1 categories described above.
When all the interviews had been conducted, transcripts from both rounds of interviews were coded thoroughly. In some cases, the same section of an interview transcript could be linked to two or more codes. In these cases, sections were coded with all the relevant categories. When all 16 interviews had been coded, I found that the categories I had used had produced manageable sections of text for each teacher’s responses, and the coded material based on the categories described above was used in order to get an overview of the teachers’ beliefs about the various issues. The coded material was used for the analyses that led to the results presented and discussed in chapters 7-12. As mentioned in section 6.2.4, abduction was the overarching explanatory model in use, which means that I moved back and forth between the different types of data, theories, and previous research in order to analyze and explain my findings.