This sub-section explains how the theoretical frameworks were taken into account to design the interview questions, what lessons were learned from the pilot study, and how NVivo was used to code the data collected.
3.2.1. Operationalising the theoretical frameworks into interview questions
Kvale (1996, pp. 94-97) highlights the importance of reviewing the literature relevant to the research study, and understanding the theoretical framework(s) used in order to thematise the interview questions. As recommended, a thorough analysis of the role of the court interpreter, the operationalisation of role-space and Actor-Network Theory (and more specifically Translation) was undertaken before establishing the themes and pointer questions used in the interviews. The thematising process was deeply anchored within the works of Llewellyn-Jones and Lee (2014) and Callon (1986) in order to design questions for role-space and Translation, respectively.
With regard to role-space, questions were divided into three main categories, each representing one role-space axis. Based on R. G. Lee and Llewellyn-Jones
(2011)’s list of criteria discussed in Chapter 2, pointers on Presentation of Self
49 The principle of sufficiency refers to the representativeness of the sample in the study, in
contrast with the population surveyed. The saturation of information indicates the point by which no new information is discovered during the interviews.
116 were thematised around the introduction/being sworn-in process, impartiality, trust, and how readily participants felt that they were seen as the court interpreter by the other court actors. With regard to Participant Alignment, themes evolved around body language, tone and register, cultural differences, feedback and back-
channelling, the possibility to hear and see the other participants, or to intervene. Finally, pointers dealing with seeking clarification or repetition, managing dyadic exchanges, and overlapping speeches were used to gather data on Interaction Management. This approach then enabled the researcher to assess the
participants’ interviews on a low to high continuum (for Presentation of Self and Interaction Management), and their Participant Alignment (based on their position compared to the axis’ point of origin), in accordance with Llewellyn-Jones and Lee (2014)’s study. More specifically, their role-space model for the court interpreting (Llewellyn-Jones & Lee, 2014, pp. 74-79) was used to benchmark for low
presentation of self, an equal alignment, and a quite high interaction management, as discussed in Chapter 2.
A similar approach was adopted for Translation. The themes were divided into four categories, each one representing one of the Translation phases (see Chapter 3). In order to gather information on the problematisation phase, participants were asked to list the people present during the hearing, and to define their role as a court interpreter and the extent to which they felt that they were an essential link in court. For the interessement phase, they were asked how the other participants perceived their role, and if they were using any strategies to reconcile any role perception differences. Then, they were asked whether they had had to
renegotiate their role later on in the interaction (enrolment phase), and if they had had to renegotiate their role from one court hearing to another (mobilisation phase). Such an approach allowed the researcher to create the participants’ networks, and to assess their Translation process.
3.2.2. Pilot study
Regarding the importance of conducting a pilot study, Seidman stated:
I urge all interviewing researchers to build into their proposal a pilot venture in which they try out their interviewing design with a small number of
participants. They will learn whether their research structure is appropriate for the study they envision. (Seidman, 2006, p. 32)
117 Following such recommendations, a pilot study was conducted with three
practising court interpreters from 13/11/2013 to 29/11/2013. This led the
researcher to question the number of VCI court assignments required in order to take part in the study, and the robustness of the interview pointers.
First, it was expected that all the participants in the study would have interpreted at least five times via VCI in a Magistrates’ or Crown Court. The number of required VCI assignments was decided in a rather subjective manner, but it was believed that such a number would generate a breadth of experience.
Furthermore, the participants would have enough experience to reflect on their role perception in a non-anecdotal manner. However, during the pilot study, it came to light that gaining access to interpreters with such a background would be difficult. Despite combining forty-seven years of experience between them,
Participants 1, 2 and 3 reported that they had only interpreted in VCI 2, 4, and 2 times, respectively, in the court settings under investigation. It also transpired that these participants had not had recent experience of interpreting in VCI A.
Nevertheless, their recollection of the events was still very vivid. Therefore, it was decided not to include any timeframe in terms of their experience.
Furthermore, the interview pointers were first designed to align with Callon (1986)’s methodology, as they focused mainly on the four Translation phases. However, when analysing the data, it became apparent that the themes discussed during the interviews did not generate sufficient data in order to design fully-
functional role-space models for all the participants, and Participant 2 and Participant 3 had to be later contacted by email to provide further information regarding their presentation of self.
Therefore, taking into account that the study’s aim was not to compare participants’ experience, but to focus on a more in-depth analysis of their individual perception, the participant requirement was amended, and only one assignment conducted in VCI A or VCI B, regardless of the timeframe, in a Magistrates’ or Crown Court was deemed necessary for this study. Moreover,
118 further interview pointers and themes50 were designed in order to align with the three-axes themes covered by Llewellyn-Jones and Lee (2014).
3.2.3. Using NVivo for coding
As argued by Basit (2003, p. 145), “electronic methods of coding data are
increasingly being used by innovative researchers”. This has led to an increasing number of publications highlighting the advantages and the pitfalls of using Computer Assisted Qualitative Data Analysis Software (CAQDAS). Arguably, NVivo is one of the leading examples of CAQDAS, assisting researchers in analysing their qualitative data. Various scholars (such as Basit, 2003; Bergin, 2011; Bong, 2002; Robert & Wilson, 2002; Siccama & Penna, 2008; Weitzman, 1999; Welsh, 2002) argue that the use of such a piece of software enables the researcher:
• to be more consistent by using only one analysis system; hence improving data analysis reliability (in terms of consistency and reproducibility) and validity (i.e. the extent to which the use of NVivo can measure what it is intended to do);
• to be more transparent with regard to the analysis process; • to store and organise all the data in one location;
• to handle a rather large amount of data.
Eighteen participants were interviewed in this doctoral thesis, which generated 12.53 hours of interview recording. Once transcribed, the data corpus was
105,104 words. Given this rather large amount of data, it was decided that NVivo would be used to code the interviews. Using NVivo was also a means to improve further validity and reliability, and make the process more transparent. However, the above scholars also warned that there would be drawbacks when using such a piece of software, namely:
• learning how to use the software can be time consuming;
• researchers may be reluctant to change their coding categories once they are established;
• researchers may feel more distant from their data;
119 • researchers may take a quantitative approach to analysing qualitative data. In order to avoid such pitfalls, various steps were taken. First, the University of Salford offered a two-day training session on the use of NVivo. This was particularly useful to learn in a rather short period of time the various functions relevant to this research project. Parents, child nodes and free nodes were used so that data could be organised, coded, recoded, and decoded easily. It was also decided not to use the quantitative functions in NVivo, but to focus principally on the coding facilities to avoid taking a quantitative approach. Finally, based on the researcher’s personal experience and ICT knowledge, it was not likely that using NVivo would make the researcher more distant from his data.
Furthermore, coding was also carried out in accordance with the thematising process described in 3.2.1 above, which according to Welman et al. (2005, p. 214) makes the coding approach descriptive. Parent nodes were used as main themes (e.g. Presentation of Self or Problematisation) under which each interview pointer was classified as a child node. As Bergin (2011, p. 8) argues, free nodes are a “stand-alone node that had no clear logical connection with other nodes – it does not easily fit into a hierarchical structure”. In this study, free nodes were also used to code data that did not fall within role-space or Translation, but some themes emerged whilst coding such as the use of simultaneous versus consecutive interpreting, the interpreter’s feelings about court interpreting or VCI, or the
reasons why VCI is used in court. All in all, 7 parent nodes, 44 child nodes, and 16 free nodes were created51, which enabled 1,137 reference segments to be
categorised in the eighteen interviews.
Finally, coding through NVivo was coupled with the use of geometry freeware, namely Geogebra, in order to create the three-dimensional role-space models that represent the participants’ perceptions of their role.