Transcription

Anticipating and Reducing Error

The first (and dominant) discourse on transcription in qualitative research holds that one must be vigilant in anticipating and reducing sources of error in the production of verbatim transcripts. Transcriber fatigue, poor-quality recordings (background noise, use of low-quality microphones and recording equipment, poor microphone placement), and difficulty understanding accents and culturally specific turns of phrase are said to lead to errors. When transcription is contracted out to individuals who are not affiliated with the study (as sample sizes of funded research studies grow), lack of familiarity with the subject matter, with academic and/or qualitative research, or with the interview itself can pose additional challenges, particularly when a background in nonresearch settings leads transcribers to tidy up the material (e.g., as with dictated correspondence). Even under ideal conditions it is possible to mistake words for others that sound very similar, leading to interesting reversals of meaning (e.g., consultation versus confrontation).

The Interpretive Nature of Transcription

A second discourse on transcription in qualitative research has more recently emerged that highlights the essentially interpretive nature of transcription. A more reflexive stance vis-à-vis issues of transcription accuracy is advocated. It is argued that, as the raw material from which transcripts are generated, even audiorecordings cannot be verbatim records of a research interview because they do not capture many elements of interpersonal interaction, nonverbal communication, and interview context that are essential [Page 885]for the interpretation of what is said (hence, the growing use of videorecording). Elements of pace, tonality, affect (humor, sarcasm, excitement, hesitancy), and some elements of turn-taking and verbal communication that do not translate easily into text (e.g, laughter, paraphrasing or mimicking others, or the interviewer's use of uh-huh that may steer the interviewee, intentionally or unintenionally) are, in many cases, lost in translation from oral to written. Within this second, more postmodern, discourse on transcription, not only is the possibility of verbatim transcription called into question, but also its desirability, as issues of contextuality, voice, representation, authenticity, audience, and positionality are brought into consideration.

Strategies for Maximizing Transcription Quality

In practice, these two discourses are not as mutually exclusive as might initially be assumed. It is both possible and necessary to simultaneously maintain a reflexive stance that problematizes many modernist assumptions about transcription, without entirely abandoning claims to rigor (albeit inherently limited ones) made possible by availing oneself of opportunities to anticipate and prevent many of the more obvious sources of error described above. Potential strategies for maximizing transcription quality include (but are not limited to) (a) ensuring highest possible audio quality (e.g., minimizing background noise, appropriate quality and placing of microphone); (b) use of notation systems and conventions/instructions to guide transcription (on how to deal with pauses, laughter, interruptions, intonation); (c) selection and training of transcribers; (d) reviewing transcription quality (of a random or purposeful subsample; co-transcription to assess inter-staff reliability when more than one transcriber is being used); (d) interviewer and researcher involvement in transcription (particularly in pilot phase for early detection of problems, and sensitization of researchers to the interpretive work of transcription); (e) involving transcribers in research meetings and as key informants in the analysis of data; (f) member checking (having research participants review their own transcripts for accuracy—though it should be emphasized that the process and results are not straightforward, and many issues of presentation-of-self, selective recall, and self-censorship inhere in member-checking); (g) flagging ambiguous material during the interview (prompting for clarification); (h) using fieldnotes and observational data to inform interpretation during and following transcription; and (i) reporting on transcription quality when writing up qualitative research.

...