Transcription for Researchers: Interviews and Focus Groups

Qualitative research runs on transcripts. Coding, quoting, thematic analysis — all of it happens on text, and the quality of the text sets a ceiling on the quality of the analysis.

Interviews are the easy case

One participant, one researcher, a room you chose. Put the recorder between you, close, and the result is usually clean.

The failure here is not audio. It is language. A participant asked to reflect on something personal will very often answer in the language they think in, not the language the interview opened in. That is good data and a transcription problem at the same time.

Set the language to match the recording rather than the study protocol. Where support is partial, the languages page says so plainly.

Focus groups are the hard case

Six people, one room, and the moments of greatest analytical value are exactly the moments they talk over each other.

Three things help materially.

Two recorders, placed apart. Not for stereo — for insurance. A group where one recorder sits nearer the quiet participants is a group you can still transcribe.

A round of names at the start. Ask each participant to say their name. It is a courtesy in the room and an anchor for speaker labels afterwards.

A facilitator who names people. "Chidinma, you were saying" does more for your transcript than any setting, because it tags the next passage in the audio itself.

More on the mechanics and the limits in speaker diarization.

Verbatim, and what it costs

Decide before you start whether you need true verbatim — every false start, repetition and filler — or clean readable text.

Verbatim matters for conversation analysis and for anything where hesitation is data. For most thematic work it is noise that makes coding slower. Machine transcripts sit closer to clean than to verbatim by default, so if you need the hesitations you will be adding them back by hand.

Keeping it defensible

Two habits protect you at write-up.

Never quote from an unreviewed transcript. Skim with the audio open and correct names, places and numbers first — those are precisely what machines get wrong and precisely what a reviewer will check.

Keep the audio. The transcript is derived data; the recording is the evidence. If a finding is ever challenged, the recording is what settles it.

Multilingual studies

Where fieldwork happens in one language and reporting in another, the translation step is usually somebody's evening. Working from a transcript rather than from audio makes that tractable, and the original stays alongside the translation rather than being replaced by it.

Starting

A free account includes transcription minutes each month with no card. Sign up and run one real interview through before committing a study to any workflow.