AUDIO TRANSCRIPTION OF INTERVIEWS
Generally full transcription of audio recordings of interviews or focus groups is recommended. While this does take time and involves additional expense, full and thorough transcription makes the most of the benefits of recording an interview - as opposed to taking hand written notes alone. It also allows a closer and repeated examination of what has been said, and can be quickly referred to at any stage in the project, avoiding the time-consuming real-time playback of recordings. However, this should not be interpreted as reducing the value of the original audio or video recordings. These will remain an important product of the research process. Even so, it is true that transcription – particularly full transcription – significantly extends the potential for analysis and re-use of a research collection by the original researchers and by secondary users.
Some researchers choose to produce only summary transcriptions usually due to cost. Although these are not ideal, summaries do provide a very useful additional resource on top of full transcriptions.
The nature of a transcription summaries can vary enormously, even within the same project, especially if different researchers produce them. This should be avoided where possible. Traditionally transcription was done by members of the research team or a professional transcriber, now it is often contracted out. Whichever route is taken, every project should have guidelines on transcription procedures. Transcription should also be an integral part of a well-constructed research plan, rather than an afterthought.
Editing transcriptions by the original participant is not recommended as it can distort the original source. However, editing for reasons of confidentiality can be done during any anonymisation process.
Once a full transcription exists, the researcher can produce summaries, interpretations or other edited versions to meet their immediate research needs. The starting point for any edited version should only ever be full transcriptions, maintaining the integrity of the original interview.
Pseudonyms
Pseudonyms can be added to transcripts at any stage, depending on the nature of the research. It may be easiest to replace identifying information (names, place names, etc.) immediately with pseudonyms during transcription. However, for field work that may continue into a series of follow-up interviews it may be better to use real names, and convert to pseudonyms as appropriate after data collection is finished. It is also worth pointing out that some thought should be given as to what information is being gathered during an interview. Briefing a respondent before an interview begins that there is no need to use full names may reduce the need for post-interview editing. This is especially important where audio or video material is going to be used as well as transcription. Further advice on the use of pseudonyms can be found on the consent and confidentiality anonymisation pages.Document layout
Transcribed interviews should:- possess a unique identifier
- adopt a uniform layout within the research project
- make use of speaker tags clearly indicating the question/answer sequence
- carry line breaks
- be page numbered
- carry a document header giving brief details of the interview: date, place, interviewer name, interviewee details, etc.
For researchers interested in using transcriptions in tandem with CAQDAS software, the series of support documents available from Online QDA may be useful.
Language
Where material is not in English, ideally full transcriptions in the original language should still be produced. Where possible a translation should also be provided, or at least a summary of each interview in English.Transcription services
Material should preferably be transcribed within a research team, or where possible, by a service with experience of working with social science research. Recommendations from colleagues in your own area about transcription services generally work best. As stressed above, it is vital to ensure that the transcribing team is given written rules describing the procedures for transcription - from layout to dealing with inaudible phrases and so on.Transcription software
Various software packages that aim to automatically transcribe text from an audio source have come on the market over the past decade. All require a great deal of training and calibration to be able to recognise a particular voice, accent and dialect. For this reason we do not recommend automation of the transcription of recorded interviews.For further advice or specific queries, contact acquisitions@esds.ac.uk
