4. Research Methodology
4.4 Data Collection Instruments and Procedures for data collection and analysis
4.4.2 Semi-structured interviews
In teacher cognition research, teachers should be given the opportunity to explore and reflect on their own personal beliefs (Kagan, 1990). Interviews can serve as a very useful tool to elicit in-depth and detailed information and insights about participants’ beliefs, thoughts and knowledge (Bell, 2010; Cohen et al., 2011; Denscombe, 2014). However, it has been
practices, and their responses may not be genuine but rather carefully structured (Kagan, 1990). Therefore, it is essential to employ strategies that would help teachers remain comfortable and encouraged to express their thoughts and beliefs. One way to accomplish this state of moderate openness can be through semi-structured interviews. In semi-structured interviews, a set of questions are prepared for the interviewee, while at the same time
additional questions might be asked during interviews to clarify and/or further elaborate on certain issues.
4.4.2.1 Semi-structured interview design.
This study employed semi-structured interviews to generate the participants’ own descriptive accounts of their beliefs and practices regarding English grammar assessment to help answer RQ1 and RQ2. Semi-structured interviews were also employed to illuminate the link between the beliefs of the teachers and their current practices while assessing English grammar (RQ3).
My initial plan had in fact involved a structured interview. However, as I piloted my initially structured interview (see Appendices E and F), a number of issues turned out to be
problematic. Particularly, the interview questions were very restrictive in the sense that interviewees’ responses were fixed and specific, almost a few words per question. This problem was highly evident with the close-ended questions. The participants in the pilot study were just satisfied with ‘Yes/No’ answers and showed no interest in elaborating on their responses.
The semi-structured interview therefore was deemed more suitable for my research. I felt that a fully unstructured interview could be difficult to handle well for a novice researcher such as myself. It ran the risk of wandering off target and not producing optimally relevant
information. The semi-structured format however seemed more suitable, since it consisted of flexible questions that provide a basic structure but allow the interviewer to organise a conversation and steer it properly, so it does not elicit one-dimensional answers from the interviewee as in fully structured interviews. A semi-structured interview is open and allows new ideas to be brought up in its course as a result of what an interviewee says.
The design of the semi-structured interview followed the following principles:
• Questions in the interview were adapted and based on interviews conducted in the reviewed studies (Karim, 2015; Mansory, 2016; Mussawy, 2009; Saad et al., 2013) (see Appendix G).
• The interview questions were broad and open-ended to allow the interviewees latitude in constructing answers.
• Any questions of the interview could be modified wherever required, which would allow for more relevant questions to be asked and the interviewees to clarify their responses.
• The wording, the structure and the order of some of the questions were changed based on the pilot study.
• Questions were clear, simple and short so as not to confuse the interviewees. • Questions were designed within a time frame that suited the participants – not less
than 30 minutes and no more than one hour – since long interviews might have led the respondents to experience fatigue, making them unwilling to continue (Robson, 2011).
• Interview questions were developed and grouped into four themes: 1) understanding what grammar assessment means, 2) identifying the purposes of English Grammar Assessment (EGA), 3) elaborating on the relevant factors and 4) discussing their
roles in constructing EGA. Organising the questions into themes facilitated the coding and the analysis phases (Table 8).
Table 8.
Themes for the Interview Guide and Questions
Themes Categories Questions
Beliefs Type of EGA 1, 2
Purpose 3, 7
Role of EFL teachers in constructing EGA 8
Factors 3, 7
Practice
Type of EGA 1, 4
Purpose 5
Role of EFL teachers in constructing EGA 6
Factors 4
As the above table shows, there are 8 questions which target the core themes of the present study: beliefs and practices. Five questions were used to elicit the participants’ underlying beliefs while four questions focused on the participants’ practices. The questions in the semi- structured interviews were interconnected to allow the identification of the relationship between the participants’ beliefs and practices.
4.4.2.2 Semi-structured interview data collection procedures.
In the present study, face-to-face semi-structured interviews with EFL English grammar teachers were conducted. English was used as the medium of communication in the interviews because all the participants spoke and understood the language very well,
including the interviewer. Moreover, using English from the start made the process of transcription easier, since there was no need for translation from Arabic to English.
All the interviews lasted for thirty minutes or more and were recorded using a digital voice recorder (see Appendix H). The use of a voice recorder assisted in the production of highly detailed and accurate transcripts, since it provided the opportunity to examine the recordings as many times as required (Silverman, 2000). Also, the use of a voice recorder proved to be easier in retrieving information and analysing the findings of the study.
Approximately 20 hours of interview data were collected. Interviews with the female participants took place in their offices (in their educational facilities) during office hours. Male participants were interviewed in the Executives Hotel17 in the hotel lobby, due to the religious and cultural aspects that regulate meeting with male strangers in public places, and with a chaperone; my husband was with me and would keep an acceptable distance, which allowed for private conversations without complete seclusion.
4.4.2.3 The interview participants’ demographics.
32 EFL teachers participated in the interviews. These included both females (N = 26) and males (N = 6) and were representative of the same population which was sampled for the questionnaires (4.5.1.3). All the participants were teaching English grammar courses at the time of interviewing. They had, on average, been teaching for 12 years, the most experienced having taught for 25 years while the novices had only one year to three years of experience. With regard to participants’ qualifications, most of the participants were PhD holders (N = 18) in the field of applied linguistics, theoretical linguistics, education and sociolinguistics,
while the remaining (N = 14) had a masters’ degrees. The majority of the participants were Saudis. The profile of the sample is displayed in Table 8.
Table 9.
Interview Participants’ Demographics
Institutions
Qualification Average years of Experience Origin Masters PhD English Teaching English Grammar teaching Saudis Non- Saudis A 7 11 10 6.5 17 2 B Ø 1 20 10 1 Ø C 6 3 12 6 8 Ø D 1 3 18.5 8 3 1 Total number (% of sample) 14 (44%) 18 (56%) 29 (91%) 3 (9%) Overall Mean 12 7
The table above sums up the interview participants’ demographics. It is essential to mention here that there is no definite way to ensure that the teachers participating in the interviews took the questionnaire, since the questionnaire was anonymous. Also, personal information about age was not provided, because in Saudi Arabia most people, especially in a voluntary interview, are not comfortable talking about these aspects.
4.4.2.4 Semi-structured interview data analysis procedure.
As Merriam (1998) states, semi-structured interviews are typically analysed qualitatively. Qualitative analysis involves continual reflection and interpretation of the data obtained in order to generate sufficient information that would be tailored to answer specific research questions (Creswell, 2003). Data from the interviews were therefore subject to content
analysis. ‘Content analysis is the process of organising information into categories related to the central questions of the research’ (Bowen, 2009). In this respect, Cohen et al. (2007) suggest that content analysis involves not only coding and creating meaningful categories but also comparing and making links among data from different sources.
The analysis process unfolded in three phases: transcription/reading, coding and
categorization. First, the audio-recorded material from each interview was transferred from the voice recorder storage unit to my personal laptop in preparations for transcription. Each interview was then imported to Dragon, the speech-to-text software (see Appendix I). Dragon allowed the transformation of voice into text within minutes and facilitated the transcription of digitalised audio-recorded files. The texts were then exported to word documents. Once the transcription of each audio material was completed, the audio along with its generated text were uploaded to oTranscribe, which is a free web app that allows one to bring both text and audio material together. Thus, one does not need to shift back and forth between the word document and the audio player. Another advantage of using oTranscribe is that it offers interactive timestamps to adapt the audio speed to one’s convenience. All the transcriptions generated by Dragon were reviewed and edited in oTranscribe. Both the anonymity and confidentiality of the data collected through the interviews were guaranteed by giving each participant a number (e.g. 004) and associated initials for pseudonym (e.g. RSh) and by deleting any possible identifiable details immediately after transcription (see Appendix J). Figure 18 illustrates how the transcription process is presented in the oTranscribe template.
Figure 18. oTranscribe template and layout.
Soon after each transcription, the transcript was carefully checked for (verbatim) accuracy against the original digital recording. Furthermore, the accuracy of transcripts was checked and verified by one interviewee, who even requested to self-review the transcript of her interview. Checking and editing the transcripts of the interviews gave me the opportunity to familiarise myself with the data and mentally begin the coding process. As Braun and Clarke (2006) state, ‘It is vital that you immerse yourself in the data to the extent that you are
familiar with the depth and breadth of the content’ (p. 87). Therefore, a systematic reading of the transcripts was conducted, and some initial thoughts and interesting points were noted before I engaged with the formal coding and the initial ideas. The process of reading was instrumental in facilitating the coding phase of the analysis.
With the first phase of interview data analysis completed, the second phase, the coding, commenced. As I mentioned earlier, the transcribed data were read multiple times;
afterwards, a preliminary list of codes based on the topics under investigation (RQs) was generated through the use of MAXQDA 2018, a software that facilitates coding frequency searches, word frequency and text searches and keyword searches. MAXQDA proved to be highly useful in allocating codes within and across the transcribed data. This programme also helped in finalising codes as well as generating specific categories and themes (Figure 19).
Figure 19. General overview of the coding system in MAXQDA 2018.
After quotes were coded, the third phase of data analysis began. The codes were entered under several different categories, which were grouped under three major themes: EFL teachers’ beliefs, EFL teachers’ practices and relevant factors (Figure 20).
The large amount of data collected in this study necessitated the recurrent reading of the transcripts, along with listening to audio recordings of the data and coding and recoding the transcripts. Braun and Clarke (2006) asserted that ‘the need for re-coding from the data set is to be expected as coding is an ongoing organic process’ (p. 21). In summary, data analysis began with the coding of the data, progressed to emerging categories and finally streamed into the themes more closely related to the research questions.