Monday, 4 July 2016

Some thoughts on the benefits of using an independent professional to provide transcription for qualitative research

Once a recording of an interview (or focus group or meeting or event or happening) has been made, the transcription process can on one level be seen as a translation (Slembrouck, 2007) or transformation of verbal utterances to text (Duranti, 2007). This though is too simple a perspective on the process.

A discussion of research into transcribing qualitative research interviews might sound a bit like a snake eating its own tail.[i] Methodologically speaking, however, how transcripts are produced for qualitative-research purposes is an important consideration in the planning of the wider process, lying as it does between the devising and conducting of the interviews and the analysis of the data.

For starters, the difference between even the most detailed, annotated strict-verbatim transcript produced by a skilled transcriptionist adhering to a detailed brief from an experienced qualitative researcher can only ever be the equivalent of a series of snap shots compared to a film of the same event.

This is not a criticism of transcription or a claim that transcripts are not ‘good enough’ to produce valid qualitative data; it’s just a case of fish and fowl in that a transcript can never be anything other than a text-based record of something else. And even if we embrace all the technology at our disposal and provide embedded audio and/or video links to selected elements for illustration, or even the complete data set for veracity, and publish research in electronic formats,[ii] right here, right now, academic standards require a text-based publication standard.

So if we can’t have Comtean confidence in the direct relationship between an audio recording of a research event and the subsequent transcript upon which findings are based, how can this be negotiated through methodology and how can researchers be pragmatic when it comes to pragmatics?

The starting point for answering these questions is maybe defining what the transcribed outcomes will be ‘for’.

If a transcript is for linguistic study, then the transcription process itself is surely part of the researcher’s analysis (Bucholtz & Du Bois, 2006; ten Have, 2007) in making distinctions and decisions about interpretation and representation of speech?

When it comes to qualitative research however, employing a ‘third party’ to transcribe the initial data[iii] not only has benefits in terms of time and priorities in a study’s schedule or in a marketing exercise, but involving an intermediary could be a positive decision with regards to the quality and authority of subsequent research outcomes and findings.  

For instance, knowing what you are looking for makes you more likely to find it? Selection in transcription (and perhaps obfuscation too) can be an unconscious thing; like we read what we thought we’d written,[iv] perhaps we hear what we think we would like to hear. For a researcher who through necessity is using a necessarily unideal source (in a strict scientific positivist sense) by relying on transcripts from audio data as their source material for research, knowing what you are looking for could introduce subconscious and unintended bias, but bias nonetheless, if transcribing your audio data yourself.

If you decide that your methodology should include taking steps to explicitly avoid such potential bias (and potentially gaining you some useful distance from your initial data-collection interviews before you launch into your analysis) by using a third party, you are then perhaps presented with issues to do with consistency and quality if you decide to outsource the process beyond yourself and your department.[v]

For instance, when you submit your audio to an agency, the work is allocated and distributed to a range of transcriptionists with varying levels of experience and skill, each with their individual styles.[vi] While a good agency will have a thorough proofreading process in place, this however is in most cases a proofread not a check on the relationship between the audio and the text. So while this goes to the issue of quality in terms of the agency’s reputation with regards to what a good transcript looks like, it doesn’t address the issue of quality in relation to the relationship between audio and text or consistency between transcripts in a particular ‘batch’ since multiple transcriptionists’ work may be checked by multiple proofreaders.

Using an independent professional transcriptionist can help you overcome these issues.

By asking one transcriber to work with you to produce your transcripts for a study, you can overcome many of the issues associated with not only consistency but also this methodological decision can increase your confidence regarding the transcribed data in relation to the audio recording. Through adopting a briefing/feedback/quality dialogue with one person, you can feel reassured that your transcribed data is consistent to an agreed standard and can be ‘read’ in a consistent way so insights gained from within one transcript can be related to those gleaned between transcripts.[vii]

The only real drawback potentially of using an independent professional transcriptionist as an integral feature of your methodology and process as a member of your research team is one of timing. This is really only an issue of planning. A transcriptionist can usually transcribe 80 minutes of audio a day. If presented with 30 one-hour interviews as one batch, they will need nearly a month to complete the task! If you are conducting two interviews a week, engaging a transcriptionist at the start of the process, feeding the audio through to them as and when research is conducted, your data collection is achieved in a timely way in relation to the research event and the production of source material for analysis.

Independent professional transcribers can be found through a Google search, or you could get in touch with Sound Words. If I can’t do the work, I can put you in touch (for no fee to yourself) with other independent, skilled, experienced transcriptionists (for no fee paid by them).

Transcription is more than translation, transformation and certainly more than ‘just’ typing. By incorporating reflection on the process into the design of research methodology and by using an independent person, this not only allows researchers to confront the issues raised by some commentators into the role transcription plays or the effect of transcription on qualitative-research data and subsequent outcomes but also raises the bar not only in terms of expected quality of transcripts as reliable data sources but also boosts the profile of the professionalism good transcriptionists bring to the work they do and places appropriate value on their work and skills.[viii]

[i] Although doing research into research might seem a bit inward looking, some researchers have looked at the role transcription plays in the research process in the context of methodology. See SA Tilley & Powick, KD, Distanced Data: transcribing other people’s research tapes, Canadian Journal of Education 27 (2002) pp 291-310; JC Lapadat & Lindsay, AC, Transcription in Research and Practice: from standardization of technique to interpretive positionings, Qualitative Inquiry 5:64 (1991); DG Oliver, Serovich, JM & Mason TL, Constraints and Opportunities with Interview Transcription: towards reflection in qualitative research, (2005) 84(2), pp 1273—1289. Others have considered in detail the role of transcription within their wider research projects such as CSG Witcher, Negotiating Transcription as a Relative Insider: implications for rigor, International Journal of Qualitative Methods, 9(2), (2010).

[ii] Obviously such an approach raises horrendous issues around participant anonymity and confidentiality. Also, audio quality is an issue in some research settings, aside from the methodological and ethical considerations. This is where a skilled and experienced transcriptionist can be the bridge between seemingly unintelligible audio and useful important qualitative data for subsequent analysis.

[iii] That said, “Some accounts of transcription take research recordings to be data (Coates & Thornborrow, 1999; Mondada, 2007), whereas others view transcripts as data (see Johnson, 2000; Ochs, 1979).” (Davidson, 2009) In this discussion, I’m using the word ‘data’ to refer to what is captured by a qualitative-research approach, which can be sometimes augmented by information about pragmatics, and analysed above and beyond the ‘stuff’ or ‘content’ of the outcome of an interview for instance. Data in this sense exists in both the audio content and the transcribed content.

[iv] Dr Tom Stafford from Sheffield University has analysed why it is we can’t spot our own typos because we read what we think we’ve written not what is actually on the page.  

[v] I had the pleasure of editing a PhD which was based on a series of transcripts produced by an agency. I was hugely surprised by the differences in the way various ‘styles’ were adopted in the different transcripts, for instance from average sentence length to use of punctuation.

[vi] We don’t speak in sentences. When transcribing verbal utterances, the transcriber makes his/her own decision as to where breaks can be made and represents this through use of punctuation. The mark of a good transcriptionist is the ability to make lengthy seemingly non-segmented utterances intelligible through having a keen sense of how speech is verbally co-ordinated and then they represent this through standard punctuation. Each transcriber’s take of how this can be achieved varies, for instance in terms of what makes for a ‘good’ sentence length in a context, how parts of an utterance are connected in more or less relative ways through content and theme, thus determining whether a new sentence can be begun or if a particular co-ordinator (such as ‘and’, ‘or’, ‘but’, ‘then’) should remain as just that in order to strike a balance between the integrity of a speaker’s proposition and the intelligibility of the transcript. While the words on the page themselves may not vary (although transcribers have varying levels of guess tolerance or query tolerance or may be more or less open to note taking and querying aspects of a person’s speech – I myself am more likely to offer choices between potential words ‘heard’ than make a guess and I provide notes to flag any queries in sense and meaning) between one transcriber’s work and the next, how this structuring and representation of phrases inevitably will differ.

[vii] Having such a relationship can also mean that you are working with a team member and colleague during what can be quite an isolating experience. Please see my testimonials for implicit evidence of this!

[viii] Oliver and Serovich state: “While often seen as a behind-the-scenes task, we suggest that transcription is a powerful act of representation.” Looking ‘behind the scenes’ can be a useful enabling thing if researchers wish to ‘incorporate reflection into their research design by interrogating their transcription decisions and the possible impact these decisions may have on participants and research outcomes.” (P.1273)  I would argue that the ethics of a study could be extended to the means of production of the data source you will use for a qualitative-research study. The person an agency uses to transcribe your qualitative-research audio will not be an employee but a freelance, self-employed sole-trader. They are therefore not covered by any employment law in relation to how the agency deals with them or treats them. They can be paid as little as 40p per audio minute with an industry average of 60p. A good transcriber can usually manage between 10 and 15 minutes of audio per hour of their time. This means they are earning less than or just about the minimum wage. This is just for the typing. Most agencies will then expect the transcriber to proofread and check the transcript (as the transcriber would want to a) for their own sense of having done a good job and b) so the agency continues to engage them). Their initial hourly rate therefore drops substantially with this final treatment of the transcript. This situation could potentially affect decisions made regarding the ethics of a research study. For instance research into the impact of austerity based on transcription of audio of interviews with those affected by such government policy which were based on transcripts produced by someone earning less than the minimum-wage hourly rate throws up one or two ethical contradictions! In addition, while the person who is responsible for producing your data source is being paid not very much, you are being charged for the fees the agency takes. This can be as much as 60p per audio minute (if you are being charged £1.00 per audio minute and the transcriber is being paid 40p). On top of this, you may have VAT to pay to an agency whereas a sole-trader is unlikely to earn enough to be VAT registered. If you are responsible for the judicious use of research grants or client money, such an awareness of the economies behind agency transcription might be useful in terms of justifying spending. Obviously time scales, practicality and the nature of your data may mean that an agency is needed and many people rely on them for work and employment in both the short and long term. 

No comments:

Post a Comment