3. Methodology and theoretical framework
3.6 Data
3.6.2 The data
For this study, I have conducted a re-analysis of these 92 psychiatric consultations. The transcripts were provided by the original researcher in an anonymised form. Permission for
using the data was sought through the original research institution and the research team, and ethical approval for the reanalysis was obtained through the University of Sussex. My access to this data was facilitated by my late involvement with analysing the data for publications from the original study. The analysis I was involved in at that stage was a Conversation Analysis of specific sections of the data in which changes to prescriptions were decided on (Quirk et al, 2012). The analysis at that stage was conducted with the original researcher (AQ), and for this study AQ was also involved in ensuring that the use of the data did not contravene the stated objectives of the data collection and abided by the ethical requirements of the original NHS ethics approval.
The audio recordings had been typist-transcribed and were identified only by an ID code. I also had access to the original recordings and listened to these at the start of analysis to provide me with an overview of the data. Detailed analysis was conducted using the transcripts, however I was then able to return to the original recordings to check the accuracy of the transcripts. I did this particularly where sections of the transcript made little sense, or were recorded as ‘unclear’, or where small changes in the wording may have had a significant impact on the interpretation. This was particularly important where I was looking in detail at how framing was achieved (see section 4.3, page 161). In addition, I had an overview of each consultation which provided basic characteristics of the participants; gender, age and
ethnicity. I did not have access to the names or other personal details of the participants in this study.
Transcripts
Edwards (1991) identifies five properties which should be aimed for in choosing a system of transcription: minimally-theory committal; easily readable; computationally tractable;
minimally time consuming to learn and to use; and expandable for specific research use. In this study, the decisions concerning transcription have been informed by these criteria, and by the
specific aims of this study. Importantly, the data came to me already transcribed which meant that decisions made were around adapting these transcripts, rather than taking a particular approach in transforming audio to written data.
The transcripts used were originally typed up by a typist and then checked by AQ. The data was transcribed using an orthographic approach, using conventional English spelling for words, rather than attempting to capture accent and pronunciation in the transcript (modified
orthography). Audible pauses were included in the transcript, along with hesitations (er, um, etc). Views differ on the use of orthographic or modified orthographic transcription. My decision to use orthographic transcripts for this study were partly pragmatic – that is the form they were in when I accessed them – and partly theory-driven. An important criticism of modified orthography is that by capturing the nuance of dropped consonants, strong accents and mispronunciations, participants can often appear less fluent and coherent (Edwards, 2003). In this study particularly, I was conscious that the psychiatrists were likely to have higher levels of education and would be more likely to use a form of Received Pronunciation than some of the patients. By emphasising pronunciation, the transcripts themselves may contribute to an asymmetry between participants’ talk, which might influence both my analysis and the interpretations of readers when presented with extracts. A final reason for avoiding modified orthography was the recognised difficulties of being consistent in this approach. Certain types of pronunciation are more likely to be picked up in transcription than others, based in part on the researcher’s own accent and what they hear as unusual. This adds a level of bias to the data that does not contribute usefully to the analysis.
My analysis was based on these orthographic transcripts, allowing me to identify aspects of talk and conventions that were of interest to this study. When exploring these sections in more detail I returned to the original audio recordings. In doing so I was able to check for errors in
the transcript and to ensure greater consistency in the level of transcript, to ensure that pauses and hesitations were treated the same throughout.
I have not used Conversation Analysis transcription symbols in this study, since I have not been conducting a Conversation Analysis. While valuable for identifying nuances in the text, these detailed transcripts have disadvantages. First among these is the practicality of attempting this level of transcription across such a large dataset, which would be a monumental and hugely time consuming task (Silverman, 1987: 8). Second, it is suggestive of a type of analysis that I have not conducted. I am using a different approach to analysing discourse from that of pure Conversation Analysis, and my focus is on the use of framing, rather than the micro- analysis of turn-taking and utterances. Finally, transcriptions can be difficult for the reader to follow when presented using Conversation Analysis conventions. While the use of symbols can help the reader to follow the writer’s argument if it is based on this detailed analysis, where this is not the focus of the study it is likely to be a distraction, rather than an aid.
Though I have not applied conventional transcription symbols, I have used symbols to indicate structures in the text that are central to this analysis. In particular, I have used a number of marks to indicate shifts in frame. These are not conventional symbols but my own indications of frame shift patterns, which I hope will aid the reader in following the arguments made in the following chapters. These are as follows2:
╔
: Indicates participant attempt to shift frame
╪
: Indicates participant rejecting an attempted frame shift
╤
:Indicates participant realigning to frame shift
2
╦ :
Indicates an alternative reframing to the one made by the other speaker
╟ :
Indicates a frame break or out of frame activity
The significance of these is discussed in more detail in Chapter Four. It should be noted that these symbols were added during and after the analysis. They were not part of the
transcriptions initially used for analysis.
Finally, as discussed in the introduction to this thesis, transcripts have been labelled with the terms ‘doctor’ and ‘patient’. I will not repeat here my reasons for this decision. However, it is important to recognise that this choice of labels is as much a methodological decision as the use of orthographic transcription techniques, and its influence on reading of the data must be acknowledged.