RESULTS AND ANALYSIS OF THE INVESTIGATION 4.1 INTRODUCTION
4.2 DATA GENERATING PROCESS
My original plan was to use a group of at least 30 students from the 2004 EFL programme at Stellenbosch University in action research focused on language learning strategies. I wanted to select the group based on their English language proficiency test results. The test results would have been used to select a group of students for my research: 10 who performed above average on the test, 10 who performed average on the test and 10 who performed below average on the test.
Upon their arrival in Stellenbosch, all non-English international students are required to take an English proficiency test administered by the University. Based on the results obtained in this English test, students are strongly advised (and sometimes compelled by their departments or sponsoring bodies) to participate in the EFL programme at a certain level (there are 6 levels). I wanted to use the results on this test to select a group of students for my research, but unfortunately this was not a standardised English proficiency test.
The first standardised English proficiency test administered by the EFL programme at Stellenbosch University in 2004, was the TOEFL (paper-based) at the end of July 2004 and only thirteen students (of which eight were Gabonese students), who had participated in the EFL programme since January 2004, took the test. All these students had completed at least level 5 (Upper-intermediate) of the EFL programme at Stellenbosch University. I decided to select 9 students from this group, namely three (3) high-performance learners, three (3) average-performance learners and three (3) low-performance learners. The nine
students would have joined the study voluntarily and would have self-reported their language learning strategy use by completing the Strategy Inventory for Language Learning (SILL) (Oxford, 1990). Unfortunately the students performed rather poorly on the TOEFL test, with not one even scoring the minimum of 550 required for admission to postgraduate studies. Test scores ranged from 410 to 530. Of these thirteen students only seven volunteered to participate in my research by completing the Strategy Inventory for Language Learning (SILL) and volunteering for an interview with me to discuss the learning strategies they use to learn English.
This situation limited my intended research even further in the following ways:
• The sample who took the TOEFL test was very small (thirteen). I had initially hoped to select thirty students from a group who had taken a standardised English proficiency test;
• When I realised that I would not have a sample of 30 students and that I was not able to conduct action research, I hoped to find 3 learners each with above average TOEFL scores, average TOEFL scores and below average TOEFL scores, but I could not divide the group into above-average, average, and below average since there were no test-takers who performed above average;
• I had also hoped for a diverse group, but of the seven volunteer participants, five were Gabonese students, one was German and one was Korean.
Since I had no participants who achieved above-average scores on the TOEFL test, I was concerned that they might not have fully understood the questions that were asked in the SILL since these were in English. A week after the initial 7 participants had completed the SILL, I had individual interviews with them to verify their responses on the SILL. I came across research done by Eric Mahlobo (1999) for a DEd degree at UNISA where he also used the SILL and had the same concern that the participants might not have understood the questions in the SILL, because they were in English. He then drew up a questionnaire (in English) to test whether his participants understood certain key concepts which appeared in the SILL. I wanted to use his questionnaire, but then learnt from the EFL teacher I interviewed that she had already done the SILL questionnaire with all the participants I had in this study in a class situation where students could ask for clarity on questions they did not understand. The students were thus familiar with the SILL when I asked them to complete it. I did, however, have the Mahlobo questionnaire translated into French and a French interpreter did interviews (based on the translated questionnaire) with three Gabonese participants to establish whether they had originally understood the questions in the English SILL. From their responses, it appears that those students had understood the questions in the SILL.
To validate the participants’ responses on the SILL, I asked an EFL teacher in August 2004 if she would be prepared to answer some questions on the learning behaviour of each of the participants in my study, based on her observations during her class contact with them. The teacher was interested in my investigation and decided to introduce language learning strategy instruction in the EFL programme. This was, however, not integrated into existing classes, but offered as a practical, optional class in an afternoon session. The teacher was so busy that the interview was continually deferred. Since I felt that she might be more willing to participate if I did so, I modified the SILL slightly so that she could complete a SILL-like questionnaire for each participant in her own time. However, this did not happen. Ethical considerations meant that I could not press her. Researcher must at all times consider whether their actions are ethically acceptable: I had to bear in mind that participants must be allowed to participate in a voluntary way, free from any coercion. Ethics also dictate that research should not cause harm to participants and I was not sure whether forcing the interview issue might cause the teacher emotional stress. It was in October 2005 that I eventually had an interview with the teacher, which was taped on audio-cassette. I did as much as possible to put my interviewee at ease. I conducted the interview in Afrikaans as she requested. I also structured the interview to an extent by giving the teacher questions and guidelines on a sheet of paper beforehand in an attempt to elicit responses to questions related to my investigation (refer to Addendum A). From informal discussion with other teachers in the EFL programme, I sensed that they were eager to divulge information about their students in informal discussions, but felt a little intimidated by formal, audio-taped interviews. I therefore had various informal discussions with them on their perceptions of the English learning process in their EFL classes.
I felt that the data I had generated from the 7 EFL participants and their teachers was not sufficient to give me rich description; I needed to generate more data. It would have been ideal to obtain more data from the same EFL programme, but unfortunately the International Office EFL programme was stopped at the end of 2004. As of the beginning of 2005 the Stellenbosch University Language Centre took responsibility for English language assistance to all Stellenbosch University students who required such assistance (including international students). The Language Centre programme offers mainly English support courses which are not as intensive (full-time) as what was previously on offer by the International Office. Instead of approaching students in the Language Centre’s English programme, I generated data from a group of 22 Gabonese students who were participating in an intensive EFL teacher education programme in the Faculty of Education in 2006. I liked the idea of generating more data from Gabonese people in an English programme, because I had 5 Gabonese students in my initial group of participants and I could see a pattern of learning behaviour unfolding: I could test and supplement what I had learnt with more Gabonese
participants. This additional data generated with the assistance of the group of 22 Gabonese education students has indeed provided a more complete picture of how Gabonese students approach the learning of English.
The data I have generated is not exactly what I had originally intended to generate. I originally wanted to direct the primary focus of this research on the relationship between English language proficiency and reported LLS use on the SILL, but as my data generation process unfolded, I learnt that it would be more useful to move away from the initial more quantitative approach to a more qualitative approach which would provide a more complete picture of the English learning (EFL) situation at Stellenbosch University. The qualitative approach provided information that would have been ignored by a predominantly quantitative approach which focused merely on whether a relationship between language proficiency and reported LLS use existed or not. I learnt a great deal about factors that underlie and influence the choice of LLS use in the language learning process – knowledge I would not have gained had I kept with my initial research intention.
As Terre Blanche and Kelly (in Terre Blanche & Durrheim 1999: 130), calling on Clifford Geertz’s definition, explain: a ‘thick description’ is a thorough account of the characteristics, processes, transactions and contexts that constitute the phenomenon in authentic language, as well as an account of the researcher’s role in constructing this description. I attempt to provide a thick description of the insights gained in this investigation in the section that follows.