Chapter 2: Methodology
2.4 Data collection, production and analysis
The data for this research consists of audio and video recordings of naturally occurring interactions in a range of SLT contexts. This section provides information about the process of data collection and the conventions used to produce detailed transcriptions of the recorded data in written form. In addition, it presents an overview of the processes involved in analysing the combination of recordings and transcriptions.
2.4.1 Data collection
To develop a better understanding of how SLTs and clients interact during the completion of therapy tasks, it was imperative to collect naturally occurring interactions. In a
procedure approved by ethics committees from three universities (the University of Newcastle, Charles Sturt University and the Australian National University), potential SLTs were identified through professional networks and through calls for participants made at a series of workshops focusing on interactional research in speech-language therapy professional practice5. Whenever SLTs expressed interest in participating in the research, they were individually contacted and provided with an information sheet about the research and a consent form. Those SLTs who agreed to participate in the data collection undertook to recruit clients who agreed for their regular therapy sessions to be recorded and included in the research. As many speech pathologists routinely record
aspects of therapy sessions, this process would not have been unusual for either therapists or clients. The only instructions given to SLTs who had expressed interest in participating in the research was to record, using either audio or video, examples of what they
considered ‘ordinary SLT interactions’ in their particular workplace6
. Recorded samples were then submitted to the researcher by mail, along with consent forms from both client and SLT. The therapeutic focus of each recorded session was recovered intutitively from the recordings, and verified through discussion with speech pathology colleagues.
As a result of the method of data collection, the data covers a range of therapy domains and situations. The Competency Based Occupational Standards for entry-level practice (SPA, 2011) describes six areas or domains of professional practice – speech, language, voice, stuttering, dysphagia and multi-modal communication. The recorded interactions sent to me involved SLT-client interactions in four of the six areas of professional practice: speech, language, dysphagia and multi-modal communication. The two areas of practice that are not represented in our data set are voice and stuttering, which may in part be due to the fact that these are areas of practice most commonly occurring in the private sector. The recordings were produced in a range of physical contexts including homes, schools,
community centres and clinics. They involve therapy sessions at individual and at group level, involving both children and adults as clients. Some clients had an impairment that was likely to be chronic, whereas for others there was a chance that their impairment would reduce or change over time. The majority of the recordings involved interactions in intervention sessions – focusing on speech production, on language processing and
production, on social skills and on the effective use of alternative and augmentative (AAC) resources. Two recordings involved a case history and observation/assessment of a client with a swallowing disorder.
The nature and length of recordings is set out in Table 2.2. Area of SLT practice 1:1 or group Length of recording (in minutes) 1 Speech 1:1 32
6 SLTs were not asked to provide any description of the nature of the relationship, the length of therapy or the focus of
the therapy activities. While this information could provide useful insights into the relationship, a decision was made to work from recordings alone, and not to cloud the analytical process with the SLTs perceptions of their therapy sessions.
2 Speech 1:1 25 3 Speech 1:1 43 4 Speech 1:1 18 5 Language 1:1 25 6 Language 1:1 37 7 Language 1:1 14 8 Language 1:1 25 9 Language Group 41 10 Language Group 20 11 Dysphagia 1:1 31 12 Dysphagia 1:1 43 13 AAC Group 26
Total minutes of recording 6 hours, 20 min Table 2.2 Nature and length of recordings
2.4.2 Transcription and analysis
Transcription is a pivotal aspect of qualitative research (Oliver, Serovich et al, 2005). As many researchers and theorists have noted (Hutchby & Wooffitt, 1998; ten Have, 1999; Liddicoat, 2007; Flood, Lapp & Brice-Heath, 2004), transcription is not a neutral activity. In deciding how to represent the recorded talk in written form, a researcher must make decisions about the level of detail to be included in the transcript. The first decision related to the mixture of audio and video recordings that were received from SLTs. The decision was made to transcribe all interactions using audio recordings only, for two reasons. Firstly, the majority of recordings sent to me were audio recordings. Some SLTs considered video too intrusive for their clients and/or clients only consented to sessions being audio-recorded. In addition, audio-recording is a more common practice in therapy practice than is the use of video recordings – in part becaue of the flexibility provided by smaller, more portable audio recording devices. The second reason related to the poor quality of video recordings received: some involved images of the client only, with the SLT not visible at all, and those involving multi-party interactions were not recorded with sufficient focus on the faces of the interactants to make visual details of interaction analysable.
The second decision made in the process of transcription related to the labelling of participants with reference to their institutional roles: SLT as ‘T’ and client as ‘C’. This decision was made to highlight the respective roles and to camouflage particular
individuals in the most thorough way possible, as members of two different groups of people rather than as individuals per se. It also facilitated routine attention to the ways institutional roles are enacted under the umbrella of SLT therapy.
The first round of transcription involved representing the talk in its temporal dimensions – adding overlaps, silences and sound stretches to the representation of words as spoken by the interaction participants. To achieve this initial level of detail, all recordings were transcribed from the audio using the conventions developed by Gail Jefferson (2004). The use of Wavepad software allowed for a visual representation of the sound wave, which is particularly useful in measuring silence length. The list of transcription notation symbols used can be found in Appendix A at the end of the thesis.
Presentation of data to various groups of professionals and CA researchers provided an opportunity for verification of the transcriptions. Recordings and transcription data were presented and discussed: (a) at regular data analysis sessions at universities offering PhD programs in Conversation Analysis in Australia (ANU and The University of Melbourne) as well as in the UK (Manchester University, University of York and Sheffield University); (b) at data sessions involving speech pathologists (Charles Sturt University and Curtin University); (c) at conferences (SPAA 2009, 2010; ALAA 2009; IPrA 2009); and (d) through discussions with members of my supervisory panel.
In the process of generating and refining these initial ‘basic’ CA transcripts, my attention was drawn to the nature of evaluations in SLT interactions. Having identified, from my concurrent reading, that evaluation is an important signifier of the asymmetry inherent in institutional interactions, I went through the preliminary transcripts, identified all
evaluation segments and transcribed the sequences in which these evaluation segments occurred in more detail from the audio recordings. As ten Have (2004) notes:
While the essential characteristics of the materials, i.e. records of streams of interaction, and the general purposes of study, i.e. a procedural analysis of those streams, set broad limits to what an analyst can responsibly do, it
leaves [researchers] with ample room to develop their own best fitting heuristic and argumentative procedures. (2004:54)
Identifying examples of evaluative turns was achieved through repeated search of all segments involving some kind of therapy task for positive evaluative terms. An
understanding of the patterns of evaluation in each squential position (1st, 2nd, 3rd turn) was built over time through a iterative process of reviewing data sets, then confirming features of evaluative turns through discussion and reviewing literature, before returning to the data sets to eliminate examples that did not actually involve the action of evaluation. Patterns of evaluative practice, and raw data examples were presented in various forums (data analysis groups; supervisory sessions; conferences and professional staff meetings) for further verification
Having collected and collated the evaluation sequences, I used the recommendations provided by Pomerantz & Fehr (1997) to select sequences, characterize actions within them, and then describe the linguistic means by which these actions were achieved. The process of writing up my analyses was a further level of analytical decision making, as patterns became more apparent through their description. The excerpts presented in this thesis have been chosen for their ability to best represent the patterns that were identified in the analysis process, given that it would be impossible to include all examples of evaluation in the one thesis.