Part I: Towards a phenomenologically and
6. Collecting and organising materials
6.3. Interviews
6.3.3. Transcribing, coding and translating the interviews
About 90 per cent of the interview materials recorded for this thesis have been transcribed. Using transcriptions, as well as translating them, gives rise to additional issues which one should be aware of. Full transcriptions are useful and desirable. Despite many new technical possibilities to code and tag audio files directly, transcriptions are still the best tool to look for, easily find, and organise information, not least because of the possibility of full-text searches.
I performed the coding itself and the analysis of the coded segments with the help of the software MaxQDA. The ORHELIA project’s list of topics, which was formulated before the fieldwork (see Appendix 2), served as a starting point for my own coding tree. It was significantly modified and extended through open coding grounded in the data. The main advantage of coding software is that it helps to keep materials organised by making data easily searchable and retrievable. The use of software when working with codes also allows for more flexibility in reorganising data when needed, which aids in the transparency of the research process and preventing de-contextualisation. The retrieval of coded segments through software is an excellent way of grouping together data from different sources according to a selection of topics. Manually this would never be possible as reliably and efficiently. The advantages of such grouping – besides speed and comprehensiveness – are that it enables one to double-check across all the data whether the structure of codes (and hence topics) makes sense and to compare across data and find patterns.
The main risk of coding is de-contextualisation, as coherent narratives are chopped into pieces. However, it is precisely computer-assisted coding that can prevent the data’s detachment from the contexts in the process of code analysis, as such coding makes it possible to switch between context (the interview in which a retrieved segment is embedded) and code (the sum of segments from the entire corpus of data with the same code) at any moment. Software could, however, be misused if it should encourage a focus on numbers, a classic form of de-contextualisation: Looking at
the frequency of codes, although technically possible, is not a good idea because, as I have outlined in section 2.2., Lifeworld, rare or extreme experiences can have
a crucial significance both as personal experiences shaping one’s life (“epiphanies”, Denzin 1989) and as indicators of similar, more common, but less visible patterns in other contexts.
Last but not least, when doing computer-assisted analysis, one is more likely to end up in the “coding trap” (Gibbs 2014, 286) – having too many rather descriptive codes – due to the ease of coding with software. Indeed, my codes in the software were rather descriptive, but as I did not have an overly large amount of data, this did not result in losing the big picture. I went outside the software with the broader analytic-theoretical categories such as ‘individualisation of the negative/ medicalisation of social suffering’ (the value of this category can be seen in Article 3, for example). One tool for grouping together relatively larger amounts of descriptive codes is to make a hierarchy, forming a tree of codes. There is a retrospective need for such reorganisation from time to time, and software, again, can make this task much easier.
Computer-assisted linkage of transcriptions to the corresponding audio files to make it possible to read and listen simultaneously also eliminates the main drawback of transcripts, that is, their limited possibilities to convey emotions. An important advantage of transcriptions is that excerpts can be partially transmitted as ‘raw material’ to the reader.
In scholarly literature, direct quotations from transcripts contribute to avoiding a frequent drawback of social sciences research, one which Fabian has put so succinctly: “The other’s empirical presence turns into his theoretical absence” (Fabian 1983, xi). Referring to Fabian’s critical concept of distancing, Agarwal rightfully says, “Fabian suggests that ethnographic discourse ‘rests upon personal, prolonged interaction with the other’ but ethnographic knowledge ‘construes the other in terms of distance, spatial and temporal’” (Agrawal 2002, 293). Recording and transcribing, and the use of longer excerpts from interviews brings the ‘empirical presence’ into the ‘theoretical absence’: it bridges this gap which is felt more, in my opinion, in works which rest mostly on classical field notes, where the relation between empiricism and theorisation/conceptualisation is a more intimate and less transparent personal affair of the researcher. Direct interview quotations, with all their limitations due to selection and editing, are pieces of evidence about the immediate interaction between informant and researcher. The process of theorisation and conceptualisation becomes more transparent and easier to grasp if quotations from field partners are used extensively and can ultimately make a piece of research more lively, convincing and thus appealing.
Yet, we must remain aware that evidence from transcriptions remains relative. It is a piece of narration that has emerged out of certain circumstances and that has gone through a multiple selection process: the narrator deciding what to say and how; the
transcriber deciding how to transcribe and the extent to which to ‘cleanse’ the speech of emotions and linguistic irregularities; and the researcher deciding which excerpts to select for direct quotation. An additional source of distortion when quoting transcriptions is their translation, in the present case from Russian to English. There should be a general concern about how to properly convey original cultural and contextual meanings (Roulston 2014, 301). Finding the right compromises between, on the one hand, closeness to the original and, on the other, understandability and plasticity in the target language is not an easy task (Esin, Fathi, and Squire 2014, 208). As the data are quoted in a language other than they were originally acquired in, layers are removed, and new ones added, for example in the case of metaphors or idiomatic expressions. In any case, questions about how to translate the numerous direct quotations from interviews require thoughtful treatment.