3.2.1.1 Methods of Data Collection. Following an exploratory sequential design (Creswell, 2015), Study 1 was conducted to identify themes for RQs 1 and 2 that would inform the development of the questionnaire to be administered in Study 2. In Study 1, data were collected with the help of one-to-one interviews. As Kvale (1996) formulates, the one-to-one interview is a “professional conversation” with the aim of obtaining “descriptions of the life world of the interviewee with respect to interpreting the meaning of the described phenomena” (pp. 5–6). As such, it provided an appropriate method to explore participants’ perspectives of the concepts of DI and TEDI.

3.2.1.2 The Development, Piloting and Validation of the Interview Schedule. The interview schedule was developed based on the theoretical and empirical studies conducted in the field and following the guidelines described by Dörnyei (2007), Prescott (2011), and Richards (2003). It was written in the participants’ and the researcher’s mother tongue, Hungarian, and followed a semi-structured format with broad grand tour questions and possible follow-up questions. This type of schedule was deemed suitable for the purposes of the study as it invited the participants to elaborate on the issues in an exploratory fashion, without limiting the depth and breadth of the respondent’s story with ready-made response categories (Dörnyei, 2007).

 

The pilot interviews took place in the spring of 2022 and were recorded with the participants’ consent. After my first interview with Bella, following Richards’ (2003) recommendations for evaluating interview schedule, I re-listened to the recording to evaluate the data’s depth and breadth. Generally, the responses met my criteria, but I noticed a redundancy issue: the same topic of difficulties in DI was discussed twice, first broadly (“What problems and difficulties can arise when differentiating in the EFL class?”) and then specifically related to the teacher’s experiences (“What difficulties do you encounter in differentiating?”). This overlap led to repetitive answers, as Bella, quite understandably, had already discussed her experiences when answering the first question. Therefore, I decided to remove the general question from the first part and introduced it only in the second part. Based on my impressions of the subsequent interviews, this change made the discussion more natural and generated responses that were more relevant to the research questions.

3.2.1.3 Participants and Data Collection Procedures. Study 1 was based on one-to-one semi-structured interviews with 9 Hungarian primary and secondary school EFL teachers. The main goal of this study was to explore the beliefs and self-reported practices of teachers coming from different backgrounds, therefore, I used maximum variation sampling to find participants who have different perspectives with regard to the focus of the investigation (Palys, 2008). As Creswell (2015) points out, this sampling procedure is applied by first identifying the characteristics in terms of which we would like to find participants and then approaching participants who display different dimensions of these characteristics. The characteristics for which I sought to ensure as great variety as possible were the teaching context (primary and secondary school), the years of teaching experience and the geographical location. For this purpose, I contacted my teacher colleagues in various cities and towns in Hungary and invited them to participate or recommend potential candidates. As mentioned earlier, the interviews made with Bella and Gréta were also included in the main study. The participants’ background data are shown in Table 5.

 

 

Nine teachers, seven females and two males, agreed to participate in the study. They represent different age groups and have a teaching experience ranging from 1 to 33 years. One of the participants, Gréta, worked in public education for 11 years but had also taught previously at various language schools for 15 years, which means her total teaching experience is approximately 26 years. It is important to note, however, that the majority of the participants (six out of nine) had no more than 12 years of teaching experience, which may have shaped their perspectives on DI and TEDI and thus had to be considered during data analysis, as well. Regarding the subjects taught, all of the participants have a degree in teaching EFL, and three of them are teachers of another subject as well (Hungarian, Hungarian as a FL, German as a FL). At the time of the interviews, five participants were teaching at schools located in the capital, Budapest, and three of them were working in the countryside (Pécs, Veszprém and Vác).

3.2.1.4 Methods of Data Analysis. The data were analysed with template analysis (TA), a type of thematic analysis, in line with the guidelines proposed by King (2012). In TA, an initial template of themes is constructed based on the literature and on the coding of a subset of the data, which is then applied, modified, and re-applied through an iterative process. As such, TA is positioned in the middle ground between top-down and bottom-up styles of qualitative analysis, i.e., it allows for the generation of a priori themes with the flexibility to redefine or discard any of these themes and to generate new ones as the analytic process unfolds (King, 2012). This approach was deemed suitable for the purposes of the present study as it facilitated the identification of evidence in the data for pre-defined patterns by handling themes as ‘inputs’, but also allowed room for new themes to be developed as analytic ‘outputs’ (Braun & Clarke, 2022). Altogether five overarching a priori themes were identified, which corresponded to the research questions of the study, and each had sub-themes drawn from the consulted theoretical and empirical literature. The a priori themes and sub-themes are shown in Table 6.

 

 

Data were analysed using qualitative data analysis software ATLAS.ti 8.4.3. The first transcript was given close reading and any section that addressed ideas relevant to the research questions were coded. Next, the codes were clustered into groups, which were checked against the a priori themes, and were either assigned to these themes or were labelled under new themes. This preliminary template was then applied and refined in a recursive manner during the coding of the rest of the transcripts. The emerging structure was displayed in a “mind-map” (King, 2012, p. 433) to indicate the hierarchical and lateral relationships between the themes. The thematic map and the first draft of the results and discussion section of the paper were subjected to peer debriefing, upon which some thematic clusters were further refined in light of the feedback received from a trusted and knowledgeable colleague (Leech & Onwuegbuzie, 2008). One month after the initial coding, I applied the code-recode strategy (Saldaña, 2013), i.e., I revisited my dataset to assess the reliability of the coding. This confirmed that the emerging template provided a suitable framework for organising and analysing the participants’ responses. Altogether 7 main themes, 18 sub-themes and 42 codes were assigned to a total of 206 interview segments (for a sample of a coded interview extract, see Appendix D). Following the practice of Fekete (2023) in documenting the coding of interview data, I provide a numerical overview of the themes, sub-themes and codes of the study in a table format (see Table 7). The final thematic map is available in Appendix E.