Article

Transcribing and Translating Research Interviews

Transcribing research interviews means turning recordings into consistent, analysable text in a style that matches your analytic method. Recordings of identifiable participants are personal data under the UK GDPR, so transcripts usually need pseudonymising with a separate anonymisation log. Interviews in other languages are normally transcribed in the original language first, then translated.

Person typing on a laptop at a white desk

Key takeaways

  1. The UK Data Service describes transcription as a critical quality step and recommends consistent conventions, standard templates and clear speaker labelling.
  2. Conversation and discourse analysis need full verbatim transcripts, while thematic and framework analysis usually work with clean verbatim.
  3. Research recordings of identifiable participants are personal data, and many contain special category data such as health or ethnic origin.
  4. Pseudonymised transcripts remain personal data for anyone who holds the key, so the key and anonymisation log must be kept separately.
  5. Interviews in other languages are best transcribed in the original language and translated in a bilingual, timecoded layout so quoted extracts can be checked.

Qualitative researchers, from doctoral students to university teams, NHS research staff and social or market research agencies, often record dozens of interviews and focus groups. Turning them into transcripts that can be analysed, shared with supervisors or deposited in an archive takes more than typing: the style must suit the analysis, participants must be protected, and interviews in other languages need a translation plan.

This guide covers choosing a transcription style for your method, handling interview recordings under the UK GDPR, anonymising transcripts, translating interviews conducted in other languages and briefing a transcription provider. For the broader picture of transcription and translation services, see our guide to transcription and translation of audio and video.

Why does transcription quality matter in qualitative research?

Transcription quality matters in qualitative research because the transcript, not the recording, is usually what gets coded, quoted and archived. Errors or inconsistencies in the transcript become errors in the findings.

The UK Data Service, which curates social science data in the UK, puts it directly in its guidance on recorded and qualitative data:

"Transcription converts recorded material into textual form and is a critical quality step."

The same guidance says high-quality transcription includes "consistent use of transcription conventions, standard templates, clear speaker labelling and transparent documentation of anonymisation edits". It also warns that "Poor sound or image quality can limit transcription accuracy and reduce the long-term value of the data", so good recording practice is part of transcription quality.

Which transcription style suits your research method?

The transcription style for a research interview should follow the analytic method, because different methods treat hesitations, pauses and overlaps as data or as noise. Decide the style before the first file is transcribed and keep it consistent across the project.

Analytic approachUsual transcription styleLevel of detail
Thematic analysis, framework analysisClean verbatimEvery meaningful word, speaker labels, notes on laughter or emotion where relevant
Interpretative phenomenological analysis (IPA), grounded theoryVerbatim or clean verbatim, by agreementExact wording kept; pauses and emotional cues often noted
Discourse analysisFull verbatimFillers, repairs, emphasis and pauses recorded
Conversation analysisFull verbatim with specialist notationTimed pauses, overlaps, intonation and breath, using a recognised notation system
Market research, consultation summariesClean verbatim or editedReadable text focused on content

If you are unsure, discuss the choice with your supervisor or principal investigator and record it in your data management plan. Our article on verbatim vs clean verbatim transcription explains the two main styles in detail.

Are research interview recordings personal data under the UK GDPR?

Research interview recordings are personal data under the UK GDPR whenever a participant can be identified, and a voice recording usually allows that. Many interviews also capture special category data, which receives extra protection.

The ICO's guidance on special category data lists categories that include "personal data revealing racial or ethnic origin", "data concerning health" and "data concerning a person’s sexual orientation", explaining that the use of such data "could create significant risks to the individual’s fundamental rights and freedoms". Health research, migration studies and studies of identity regularly touch these categories.

A transcription or translation provider normally acts as a processor for the research institution, which is the controller. The ICO's guidance on controller–processor contracts lists terms that Article 28(3) requires, including "Processing only on the documented instructions of the controller", a "Duty of confidence", "Appropriate security measures" and an obligation to "delete or return" the data at the end. Check with your institution's data protection or research governance office before sending recordings to any external service.

How should research interview transcripts be anonymised?

Research interview transcripts should be anonymised by replacing identifying details with meaningful descriptors rather than deleting them, and by recording every change in a separate log. Planning this at transcription stage saves a second pass through every file.

The UK Data Service's guidance on anonymising text data says direct identifiers "should normally be replaced by pseudonyms or removed prior to sharing", and recommends that you "Maintain an anonymisation log recording changes". It gives examples of descriptive replacements such as "[colleague 1]", "[a local company]" and "[city in Eastern England]". It also cautions that "Over-anonymisation can remove too much contextual detail, reducing interpretive value and limiting meaningful reuse."

Pseudonymised transcripts are not automatically outside data protection law. The ICO's guidance on pseudonymisation states that "Pseudonymised data is personal data in the hands of someone who holds the additional information", such as the key that links codes to real names. Keep that key and the anonymisation log separate from the transcripts and restrict access to them.

Should the transcriber or the researcher anonymise?

Either can work. Asking the transcriber to apply agreed replacements during transcription is efficient for large projects, but you then need clear rules and a full unredacted version if your ethics approval allows it. Anonymising yourself afterwards gives more control over borderline cases, such as rare job titles or unusual events, which are often the indirect identifiers that matter most.

How do you transcribe and translate interviews conducted in another language?

Interviews conducted in another language are usually transcribed in that language first and then translated, so that the analysis can always be checked against the participant's own words. Some projects translate only the extracts they quote, but that works only if a member of the team can analyse the original-language transcripts.

Points to agree with your translator:

  • Translation approach. Decide whether translations should stay close to the participant's phrasing or read as natural English. For qualitative work, closeness is usually preferred, with translator's notes explaining idioms and culturally specific terms.
  • Layout. A bilingual, timecoded layout makes it easy to check any coded segment against the original.
  • Consistency. Use the same translator, or a shared glossary, for all interviews in one language, so that key concepts are rendered the same way throughout.
  • Quoted extracts. Quotations used in theses and papers deserve a final check, because they are what readers and examiners see.

Participant information sheets and consent forms may also need translating before fieldwork. For workflow options in more detail, see foreign-language transcription and translation.

What should you send a transcription provider for a research project?

A transcription provider for a research project needs a written brief that reflects your ethics approval and data management plan. The UK Data Service publishes a set of tools and templates, including a transcription template, example transcription instructions "to help transcribers use a consistent style, layout and editing" and a transcriber confidentiality agreement. Use this checklist when you prepare the brief:

  1. Transcription style and any notation system, with a short example.
  2. Template and layout: header details, speaker codes, line numbering if needed, file naming.
  3. Anonymisation rules: what to replace, with which descriptors, and whether a log is expected.
  4. Glossary of technical terms, place names and organisations that participants mention.
  5. Language details for non-English interviews, including dialects and any mixed-language speech.
  6. Security requirements from your institution: transfer method, storage, confidentiality agreement and deletion at project end.
  7. Output format: Word, plain text or a format that imports cleanly into your qualitative analysis software.

Can you use automatic transcription for research interviews?

Automatic transcription can be used for research interviews as a first draft, but the output must be checked against the audio before analysis. The UK Data Service says automated speech recognition tools "may be used to support transcription workflows", but "outputs should always be reviewed and corrected for accuracy."

Data protection is the other consideration. Many online speech-recognition services process audio on servers you do not control, so confirm with your institution whether a given tool is approved for participant data, particularly where interviews contain special category data.

Common mistakes when transcribing research interviews

The most common mistakes in research transcription come from decisions left until after fieldwork. Watch for these:

  • Choosing a style after transcription has started. Switching from clean verbatim to verbatim halfway through makes transcripts inconsistent and hard to compare.
  • Over-anonymising. Removing every place, job and event can leave transcripts that support no analysis and cannot be reused.
  • Under-anonymising indirect identifiers. A rare occupation plus a small town can identify someone even without a name.
  • Using unapproved tools. Uploading recordings to a consumer app may breach your consent terms or ethics approval.
  • Losing the link between translation and original. Translations without timecodes or segment numbers make it hard to verify a quoted extract later.

How TranslationWorks can help with research interview transcription

TranslationWorks is a UK-based translation agency that transcribes research interviews and focus groups and translates them between English and 150+ languages, working with native-speaking linguists. Because interview recordings often contain sensitive personal data, it is relevant that TranslationWorks is ISO 17100 (translation services) certified, ISO 27001 (information security) and ISO 9001 (quality management) registered and Cyber Essentials certified. Send us your transcription conventions, anonymisation rules and language details and we will reflect them in your free quote. Find out more about our transcription services and translation services, read about data security when outsourcing translation, or request a free quote, which we send within 24 hours.

Need research interviews transcribed or translated?

Send us your conventions, anonymisation rules and languages, and we will prepare a free quote for your project.

Keep reading

More from TranslationWorks

Related reading

Get a quote Call