90
stereotypes showed that, similarly to L1, learners perceived certain aspects of an L2 speech variety as sociolinguistic markers, with which they associated meanings (cf. Section 4.4.1).
According to learners‟ comments in the open-ended questions following the comprehension tasks, these meanings were related to the speakers‟ origin, status as a speaker of English (learner or advanced user) (cf. Andreasson, 1994) and personality traits such as being kind, brusque, hurried or easy going. However, more in-depth research was needed to shed light on the exact sociolinguistic meanings attached to speech varieties by learners.
91
processes in depth along with their individual characteristics related to accent comprehension and attitudes. The follow-up study was expected to contribute particularly to research questions 2 and 3, concentrating on underlying reasons as opposed to factual descriptions and relationships between variables. Although the quantitative data analysis may have shown correlations between variables, convergent qualitative data could corroborate such findings and also help in their interpretation. The aim of the study was to enrich the main research project with an insiders‟
perspective of Hungarian secondary school learners; therefore, I invited the learners who did the comprehension test and the attitude questionnaire in Study One to participate in a follow-up study along with three learners from other schools to share what they know about English pronunciation, what significance it has in their life and how they relate to speech varieties.
3.4.2 Selection of participants
Participants were chosen based on purposeful sampling. In the Study One, learners were asked to write pseudonyms on the questionnaire so that they could be requested to participate in the follow-up study based on their results but allowing for anonymity at the same time. Based on this, I revisited participants who achieved a relatively high or relatively low score on the most difficult speech variety in the comprehension test.
Two types of participants were selected for the follow-up study, based on their results in the comprehension test. The first type included high achievers, who reached the highest score on the task with the Egyptian Arabic accent, which proved to be the most difficult with the lowest average score. This was assumed to be indicative of high allophonic tolerance and it was
92
therefore clearly worth investigating what characteristics distinguished those learners from the others and made them more successful in understanding a potentially difficult speech variety.
The second type consisted of different achievers, that is, learners exhibiting the greatest difference between their scores in the task with the easiest (in this case RP) and the most difficult (Egyptian Arabic) accent. In order to measure this variable, first the easiest and the most difficult accents of the whole sample were identified based on the means scores of the task. After that, a new variable was created by deducting the score of the difficult (low scoring) accent from the score of the easy (high scoring) accent. In the present case, the Egyptian Arabic accent proved to be the most difficult and RP the easiest, which meant that the additional variable generated for each case was [RP_score]-[Egyptian Arabic_score]). Based on the new variable, cases were ranked in descending order so that the cases with the highest value could be selected. This way, those learners could be identified for whom a difficult or unfamiliar accent may have posed the greatest problems, having a presumably low level of allophonic tolerance.
I also included learners who did not participate in Study One, including learners from locations other than Budapest, to enrich the data. These steps were taken to ensure maximum diversity so that data could be gained from people with different knowledge, different experiences, who come from different contexts. From the three additional learners, 2 were from the countryside, studying in Székesfehérvár. It was assumed that secondary Hungarian school learners who do not share the socio-cultural context of the participants of Study One (cf. Section 3.2.2) might relate to accent variation in English differently since they might have experiences with different accents in their L1 due to the fact that they moved to Budapest from the countryside.
93
This way, altogether 8 learners participated in the study, 5 of whom took part in the previous study as well, 2 high achievers and 3 different achievers. There were 3 female and 5 male participants altogether. The in-depth study of these learners aimed to identify problem areas in comprehension as well as individual characteristics influencing their comprehension and gain an insight into the way learners relate to English pronunciations on the affective level.
The consent of the school was requested and granted for conducting research with learners. Separate interview sessions were arranged with the three participants who had not taken part in the previous study of the research project. The latter participants were aged over 18 and granted their verbal permission to participate in the study.
3.4.3 Instruments
A relatively tight research design was used compared to the standards of the qualitative paradigm, using a structured interview as opposed to having an open discussion with learners, which allowed learners to share their thoughts on the topics related to the focus of investigation.
The reason for this was that the present study can be regarded as confirmatory qualitative research (cf. Miles & Huberman, 1994) due to the fact that it was preceded by a quantitative study on the subject with a specific focus, whose results the findings of the present study could be potentially compared with. The purpose of the follow-up study was to focus on individual cases, exploring the process of comprehension along with an in-depth investigation of how learners relate to varieties of spoken English on the level of attitudes, which would have
94
warranted an open and unstructured exploratory design; however, the research design had to be tightly controlled and focused in order to relate it to the previous study of the research project.
For this purpose, a semi-structured interview was used including a think aloud protocol, a stimulated recall and a recording of the learner‟s own pronunciation (cf. Appendix D). The procedure of the interview sessions was as follows. The data collection sessions started with warm-up questions, which also served as icebreakers to create a positive atmosphere for the interview. This was followed by a think-aloud protocol when the learners listened to and commented on an audio recording of an English accent, namely the Egyptian Arabic accent, since this speech variety proved to be the most relevant from the point of view of comprehension as it was mentioned before (cf. Section 3.2.2). Then the first part of a semi-structured interview took place, after which the learners were asked to read a few lines from the script of the audio recording in order to listen to their own pronunciation before discussing the questions in the second part of the interview. The language of the data collecting session was Hungarian, the mother tongue of the learners.
The initial warm-up questions were connected to the learner‟s English studies, which could provide an insight into the importance of English for the learner and their motivation to learn the language in general. After this, I played an audio track to the learners, which was taken from the comprehension test of the Study One. The audio script can be found in Appendix E.
The sound sample featured a bilingual Egyptian Arabic speaker of English, whose pronunciation was characterized by a slightly fast pace of speech, strong syllable-timed rhythm, the reduction of certain vowels, the pronunciation of the (inter)dental fricative /θ/ and /ð/ sounds as alveolar fricative (/s/, /z/), among some other idiosyncrasies not detailed here. I selected this accent from the others because this proved to be the most difficult to learners based on average scores in the
95
comprehension test. I was hoping to gain different insights from learners who could understand it relatively well and from those who could not.
The audio recording had several purposes. It enabled the learner to comment freely on the speech variety, identifying features hindering comprehension while seeing the transcript, and they could talk about listening strategies in the form of a think aloud protocol, discussing the accent while listening to the track for a second time sentence by sentence. It also served as stimulus for the rest of the interview for questions focusing on other speakers‟ accents and, comprehensibility and attitudes.
This was followed by a semi-structured interview with questions focusing on the learner‟s exposure to English, their awareness of English speech varieties, their reaction to comprehension difficulties, the development of learners‟ own pronunciation and how they think their pronunciation was judged by others. Before the questions concerning the learner‟s own pronunciation, they were asked to read a section from the script of the audio recording so that their actual pronunciation could be compared to what they would say about their pronunciation, and it would make it possible to cross-check the effect of their reported pronunciation models on their English accent. The sessions were recorded with the prior consent of the participants.
3.4.4 Conducting the interviews
As for the interview sessions in the schools, I arranged with the teachers to visit the groups who participated in the previous study and asked the learners whom I had selected based on their test scores to identify themselves based on their pseudonyms. In case a learner was not available, the
96
pseudonym of the participant with the closest score was called upon. The interview sessions lasted for approximately 35 minutes with each participant as they had to be conducted during a school lesson. The fairly focused questions made it possible to discuss the issues related to the investigation rather exhaustively within this time frame.
The procedure of the interview sessions were as follows. First, the learners were assured that what they said would be treated confidentially. Also, they were made aware of the fact that they were not being tested and whatever they said was valuable data for the research project.
This was necessary in order to avoid inhibition and the effect of social desirability as well as to motivate the participants to provide unselfconscious answers. Each participant was asked to give their oral consent to having the session recorded. After the participants gave their verbal permission, the digital recording device recorded the whole session.
The interview began with an introductory question about the source of the participant‟s English knowledge, which was followed by listening to the speech excerpt of the Egyptian Arabic accent. First, the audio file was presented in full length, after which it was played sentence by sentence. For the second hearing, the learners were given the transcript of the excerpt and they were asked to underline the words, phrases or parts of the text whose understanding would have been problematic for them without the text. During the process, they could also comment on what speech or sound features they found difficult and why. Learners also had a chance to comment on how they coped with understanding potentially difficult parts of the audio recording. Afterwards, they could also reflect on general features of the excerpt which might have interfered with comprehension, such as general articulatory features, speech rate, rhythm or difficult vocabulary.
97
This was followed by a semi-structured series of questions centred around the constructs which proved to be relevant in the previous study, namely exposure to spoken English, phonological awareness, tolerance of ambiguity and attitude towards problematic accents. After that, the pronunciation of the learner was recorded as they read out part of the transcript, which served two purposes. Firstly, inferences could be made about the learner‟s productive phonological competence, where their pronunciation features came from (e.g., L1 or a native English variety) and whether they tried to emulate a particular native speaker accent, which could signal cultural affiliation. Secondly, reading out loud was intended to help in raising awareness and evoking feelings relating to the learner‟s own pronunciation. Thus, it served as a stimulus for the final questions focusing on intended behaviour, that is, the effort made to improve one‟s own pronunciation and the perceived significance of accent.
3.4.5 Method of analysis
The aim of data analysis was to find emerging themes in the learner‟s accounts which could complement the constructs of the previous study, to reinterpret them, if necessary and to formulate hypotheses based on the qualitative data following the tradition of Grounded Theory (Glaser & Strauss, 1967). The recordings of the data collecting sessions were transcribed verbatim, and the transcripts were inputted into MaxQDA qualitative data analysis software to facilitate coding and recoding categories as well as searching the text during the process of analysis. The data analysis was conducted based on the Constant Comparative Method (Glaser &
Strauss, 1967). The themes related to learners‟ own pronunciation were compared to the
98
recording of their pronunciation. The frequent recurrence of similar themes and the dearth of novelties emerging from the additional three interviews indicated saturation of the data.
Participants are referred to by their pseudonyms given to them for the purpose of the study, which are not identical to the pseudonyms they gave to themselves in Study One (cf.
Section 3.2.3). The high achievers were Kálmán and Mária, the different achievers were Helga, Laci and Tamás. The pseudonyms of the three additional participants were Béla, Mária and Ferenc. The interviews were carried out in Hungarian, the native language of the participants, so that they could express their thoughts without the potential limitations of their proficiency in English. Therefore, the following quotations in the following sections are English translations from the Hungarian transcript.
The following section presents the findings gained from the data, divided into sub-sections related to the foci of the study, containing the major themes and categories related to the areas. For the quotes illustrating the emerging ideas, the pseudonyms are used in order to respect the participants‟ anonymity.
99