2 Theoretical framework and literature review
Stage 1: Transcripts and digitisation
The first stage of data analysis was to digitise my data for easier analysis. This involved the making of transcripts of discourse and the capturing of hand-written data into digital form (fieldnotes, learners’ classwork and test scripts). I created folders on my computer for each day that I went to Success High. I transcribed fieldnotes from my notebook into Word files for each day. I labelled photographs of the environment of the school or learners’ work and stored these in folders and transcribed them into spreadsheets for ease of comparison. I downloaded audio and video files nightly from the cameras and recorders and stored these in folders. Then I made transcripts from the audio and video recordings. These were made with the help of four bilingual isiXhosa-English transcribers over the course of the project. I worked with each to communicate the requirements of my analysis and demonstrate the conventions I preferred. The value of these working relationships cannot be overstated. Each
transcriber, but Babalwa Molate in particular, did important analytical work through making the transcripts. I also consulted Babalwa many times after she had completed the transcripts to listen to extracts with me and help me settle on a version. Fellow scholars in my department were also enlisted to help with this re-listening work. In total the five transcribers produced 18 transcripts of interactional data representing roughly an hour of discourse each. The transcripts were made predominantly from the video camera placed at the back of the classroom and the solitary camera in the study group as well as the audio recorder placed near one of the learners, Mthobeli, in the class lessons. The other recordings were used to check the transcriptions or to add details that the main recordings did not pick up. Three transcripts were made from the three audio recordings of the interviews which were conducted in English.
I will describe the structure of two levels of transcript which were created during my study. A first level transcript was broad, covered all the interactional data and assisted in drawing up codes for analysis. The second was more detailed and only utilised once extracts were selected for microethnographic analysis and for use in this dissertation. An example of a first level transcript appears as Extract 3.4.
Extract 3.4: SG1, Level 1, completed 100616 (for transcription conventions, see Addendum 1)
Time/Activity type
Actor/Action Speech Gloss
Khethiwe What’s the question? What’s the aim (indistinct)
Robyn What is the question? What is the question? Nandipha What’s the question?
Robyn We want to know what are these made of?
The first column served as a tracking device for me to identify parts of the transcript for analysis by time and activity type. Time was recorded in minutes and seconds as it related to the camera recording, with Mthobeli’s audio time markers appearing in brackets. I recorded shifts in activity type (Lemke, 1990) in this column as well. These entries formed the beginning of Stage 2 coding. I consider the writing of the transcripts as part of the data analysis following Ochs (1979) who asserted that transcription is theory-making. For example, the column
headings I used to focus my own and other transcribers’ attention displayed my theory of meaning-making as multimodal. I created a column entitled ‘Actor/Action’ drawing on multimodal discourse analysis (Iedema, 2003; Norris, 2016). Cognisant of a horizontal hierarchy of print which privileges text on the left, I placed action before speech in order to correct my natural inclination to view the verbal as most important. I also used the nomenclature ‘speech’ and ‘gloss’ to avoid naming languages and to enable the writing of different versions in the last column to assist my own and my audience’s understanding where necessary. For example, an utterance (speech) in a variety which draws on features of English and isiXhosa is re-versioned (glossed) into standard English, not translated.
These first level transcripts along with the raw data were then used to identify moments for what Li Wei describes as ‘moment analysis’ (Li Wei, 2011) which:
aims to capture what appears to be spur-of-the-moment actions that are semiotically highly significant to the actors and their subsequent actions, what prompted such actions and the consequences of such moments including the reactions by other people (2011, p. 1222)
The transcripts of these moments of high semiotic significance were then worked up into multimodal transcripts using the other video and audio recordings to increase the detail of the discourse. At this point further nuance was added to the spoken discourse in the form of indicators or tone, overlap etc. These transcripts were then subjected to microethnographic analyses in Stage 2 and some were selected for inclusion in this thesis.
Here is the same piece of data as Extract 3.4 transcribed as a Level 2 transcript: Extract 3.5: SG1, Level 2, ongoing
Turn Actor/Action Speech Gloss
1 Khethiwe
Raises eyebrows when looking at Siphosethu
What’s the question
What’s the aim //of the investigation
2 Robyn //What is the question
What is the question
3 Khethiwe the aim
4 Nandipha What is the question
‘Time/Activity type’ gave way to ‘Turn’ as I prepared extracts during the write-up phase as I attended to the needs of an audience. ‘Turn’ does not always align with the conversation analysis definition as at times I include a back-channel or minimal response as a separate ‘turn’. It merely serves the practical purpose of making it easier to refer to in the analysis. Next, I present the transcription of the action during the turn, including the name of the actor, again privileging the actional over the linguistic mode.
This Level 2 transcript can be described as the beginnings of a microethnographic analysis (Bloome et al, 2005; Kramsch & Whiteside, 2008; Zuengler & Mori, 2002; Erikson, 1996; Martin-Jones, 1995). This kind of analysis is a feature of linguistic ethnography and indexes the narrowed and focused nature which sets linguistic ethnography apart from a general ethnography. During microethnographic analysis more turns have been identified and actions modes added. This level 2 transcript is never complete as details may be added during further analysis for different purposes. Indeed, the interactional data that is used in the findings chapters does not conform to a set transcription structure, but rather the details that are relevant to the argument are transcribed.
The transcription of the interview data also involved two levels, with the second being employed in the write-up. The first was a prose transcription of the utterances of each speaker, while the second was a re-writing of each utterance into a verse form enabling the reading of phrases with short breaks at the end of each line as they would have been spoken (Extract 3.6).
Extract 3.6: Poetic transcription, Principal Interview Principal: I am Afrikaans
as my home language is Afrikaans um so
but I can make myself understood in English
6 Robyn
Learners start to open books and prepare to write
We want to know what’s what are these made of