Transcription conventions in qualitative research decide how much of the talk reaches the page. Full verbatim keeps every word, filler and false start. Intelligent verbatim drops the ums, repeats and slips so that speech reads as prose. Jefferson notation, built for conversation analysis, also times the pauses and marks overlap, stress, pitch and breath. None is more correct than the others: the right one is the one your analysis needs, named in your methods chapter.
EdCitation publishes this guide. EdCitation never transcribes a recording and never writes any part of a paper; its tools look up the references you cite in the publisher's record. We read each convention at a published source in September 2026: Jefferson's own glossary, the two papers behind naturalised and denaturalised transcription, and the UK Data Service's guidance, whose page has moved, in an archived copy from June 2026. APA Style's pages, behind a bot check, also come from archived copies.
We then ran EdCitation's free Cite a source, Verify references and Find sources on the methods papers below, and quote what came back, faults included.
Which transcription conventions are used in qualitative research?
Three, in falling order of detail. The UK Data Service (n.d.) says the method depends on your theoretical and methodological approach and differs between disciplines, and treats transcription as a step in the analysis rather than a mechanical conversion.
| Convention | What it keeps | Usually suits | Defined in |
|---|---|---|---|
| Jefferson notation | The words, plus timed pauses, overlap, stress, pitch, volume, breath and laughter | Conversation analysis | Jefferson (2004) |
| Full verbatim | Every word, with false starts, repetitions, fillers, pauses and background noises | Analyses of how things were said | UK Data Archive (2011) |
| Intelligent verbatim | Every meaningful word; fillers, repeats and slips removed; laughter or sighs kept where they carry meaning | Thematic analysis of what was said | McMullin (2023) |
| Any of the three | The methods papers you cite for your choice | Every methods chapter | Referenced with Cite a source, checked with Verify references |
The UK Data Service recommends transcribing interviews in full if they are to be archived, because full transcripts widen what you and later researchers can analyse. Recordings kept with summaries instead are harder to anonymise.
Verbatim vs intelligent verbatim: what is the difference?
Full verbatim keeps how something was said; intelligent verbatim keeps what was said. The difference is in the fillers, false starts, repetitions and grammar.
What a full verbatim transcript keeps
The UK Data Archive's (2011) example instructions for a transcriber show one full verbatim convention in detail. The transcriber keeps false starts, repeated words, unfinished statements, "umms" and "errs", and hesitations marked with ellipses; puts emphasis in italics; puts background noises and tone in round brackets, such as (sighs) or (door slams); and writes (inaudible) where a word cannot be made out, with a firm instruction not to guess. Grammar is not tidied, and "eye spellings" such as enuff are not used.
Even this "full" convention makes choices. It leaves out the interviewer's back-channel "yeah" and "right" while the participant speaks, and uses ordinary punctuation. That is why "transcribed verbatim", on its own, tells an examiner very little.
What intelligent verbatim leaves out
McMullin (2023) describes intelligent verbatim as adapting spoken language to written norms: a speaker's misspeaking and self-correction, verbal fillers such as "um" and excessive repetition are left out, so quotations read easily. For social science and business research, which seldom needs conversation analysis, McMullin suggests a version of intelligent verbatim will often be the right choice, provided laughter and non-verbal cues such as sighs or sobbing stay in when they carry meaning.
Clean verbatim and other service names
Transcription services sell the tidied style as "clean verbatim", "intelligent verbatim" or "edited" transcription. We found no published methods source that fixes the difference between those names, so get a service's written list of what it removes, and keep it for your methods chapter.
What do naturalised and denaturalised transcription mean?
The same two words are used in opposite senses, so define them whenever you use them.
Bucholtz (2000) calls a transcript naturalized when it follows the conventions of written text, and denaturalized when it keeps the forms of speech. She argues that both can serve as political tools of representation, and asks transcribers to be aware of the limits of their own choices.
Oliver, Serovich and Mason (2005) use the words the other way round. For them, naturalism records every utterance in as much detail as possible, and denaturalism corrects grammar, removes interview noise such as stutters and pauses, and standardises non-majority accents. They add that denaturalised transcription still aims to be faithful: its accuracy is about meaning, not accent.
So Bucholtz's naturalised transcript is close to intelligent verbatim, and McMullin (2023) pairs the terms that way, while Oliver's naturalised transcript is close to full verbatim or Jefferson. The UK Data Service (n.d.) cites Bucholtz but uses Oliver's senses: it says a thematic project usually needs a denaturalised approach, most like written language, and a conversation analysis a naturalised one, most like speech. Whichever pair you use, say what your transcript kept and cite whose definition you follow.
The sniff that changed a study
Oliver, Serovich and Mason (2005) show why the choice matters. In their interviews on HIV disclosure, a transcript marked a participant's repeated sniffing as ((sniff)). Some of the team read it as crying, others as a sign of drug use, until the interviewer explained that the participant was ill, with a running nose. The authors paused the project to reflect on transcription before choosing, and report the suggestion of keeping two versions: a detailed reference copy, and a denaturalised copy for member checking and content analysis.
What do the Jefferson transcription symbols mean?
Jefferson notation is Gail Jefferson's system for marking how talk is delivered, and her "Glossary of transcript symbols with an introduction" (Jefferson, 2004) is the reference most papers give for it. We read the chapter itself, pages 13 to 31. Her case for the detail is disarming: she puts it in "because it's there", and because nobody knows in advance which feature will turn out to matter (Jefferson, 2004, p. 15). The common symbols, in our words:
| Symbol | Jefferson's meaning |
|---|---|
[ and ] |
Where overlapping talk starts, and where it ends |
= |
No break or gap between two pieces of talk |
(0.5) |
A silence, timed in tenths of a second |
(·) |
A brief interval, about a tenth of a second |
| Underlining | Stress, through pitch or loudness |
: or :: |
The sound before is stretched; more colons, longer stretch |
↑ ↓ |
A shift into especially high or low pitch |
. , ? |
The usual intonation for that mark; no mark means an indeterminate contour |
WORD |
Especially loud compared with the talk around it |
°word° |
Softer than the talk around it |
- |
A cut-off |
>word< and <word> |
Speeded up, and slowed down |
·hh and hh |
An in-breath, and an out-breath |
(h) |
Plosiveness in a word, often with laughter or crying |
( ) and (word) |
Talk the transcriber could not make out, and a doubtful hearing |
((cough)) |
The transcriber's description |
Jefferson marks two older symbols, the double oblique // for overlap and the double dash for a short untimed gap, as no longer in use, and notes that some marks changed between her typed and her computer transcripts. So state which version you follow, and put a key to the symbols in your appendix.
An extract, read line by line
An invented extract, to show the notation at work:
I: so how did you find the first placement?
P3: (0.6) it was- (·) it was fi:ne I ↑think
I: [right]
P3: [°mostly°]
Read it as: a pause of six tenths of a second before P3 answers, a cut-off, a brief pause, a stretched "fine", a rise in pitch on "think", then the interviewer and P3 speaking at once, P3 quietly.
How do you lay out an interview transcript for a dissertation?
Use one layout for every transcript in the project. The UK Data Service's list of what each should have, with its model template (UK Data Archive, 2019), is the core of this procedure:
- A header: the study, a unique interview identifier, date, place, the interviewer, and the interviewee details that matter to your analysis.
- Speaker tags at every turn, with a line break between turns. The template uses I: for the interviewer and R: for the respondent (R1, R2 for several); a code such as P3, or a pseudonym, works the same way.
- Anonymise as you go, or mark sensitive details to change later, and put every change in square brackets, as the template does for an added clarification, "[in England]". See anonymisation.
- Number the pages. The Service's list does not ask for line numbers; add them if you will point readers to particular lines, or if your department asks.
- Add time stamps at inaudible passages or at regular intervals, such as every minute, which McMullin (2023) suggests because it makes checking against the audio much easier.
- Mark what you cannot hear, as (inaudible) or with Jefferson's empty parentheses. Never guess.
- Keep formatting plain if the transcripts go into CAQDAS software: headers, italics and bold can be lost on import, and two-column layouts can cause problems.
- Check every word against the recording. If speech recognition drafted the transcript, the Service advises caution and a check of the whole text.
- Reference the methods sources behind your convention with Cite a source, and run the finished list through Verify references.
How do you quote your transcripts in APA 7?
As data, not as sources. APA says quotations from your own participants get no reference-list entry and are not personal communications (American Psychological Association, 2019). Quotations under 40 words go in quotation marks, 40 words or more in a block quotation, and the text says the words come from participants. Keep the confidentiality you promised at consent, with pseudonyms, blurred details or aggregate reporting.
Changes follow APA's ordinary rules for quotations (American Psychological Association, 2022): an ellipsis where you leave words out, square brackets for anything you add. A bracketed clarification from your transcript, such as "[my manager]", can travel into the quotation as it is. Jefferson's own glossary shows extracts with their symbols intact, which is what a conversation analysis needs: readers see the talk you analysed.
If your quotations are tidier than your transcripts, say so once; writing up qualitative findings covers how many to use. A transcript someone else published or archived is cited as in how to cite an interview transcript, and interviews in another language as in citing your own translation of a quotation.
How do you report transcription in your methods chapter?
In enough detail that an examiner could redo it. Most published papers do not. McMullin (2023) read 133 interview studies in the journal Voluntas: 41% did not mention transcription at all, 11% mentioned transcripts but not how they were made, 19% said only that interviews were recorded and transcribed, and of the 26% that said more, such as who transcribed or that it was "verbatim", none explained what verbatim meant.
The reporting checklists ask for it. COREQ asks whether audio or visual recording was used, whether field notes were made, and whether transcripts were returned to participants for comment or correction (items 19, 20 and 23; Tong et al., 2007). SRQR's item on data processing names transcription, checks on data integrity and anonymising excerpts (O'Brien et al., 2014). Our guides to reporting guidelines and the methodology chapter cover the rest.
What to include
Who transcribed (you, a service, or speech recognition you checked); the convention, with whose definition; what was kept and removed; the symbol key, in an appendix; how transcripts were checked against the recordings; when identifying details were changed; and whether transcripts went back to participants.
An example paragraph
Adapt this, keeping only what is true of your study:
Interviews were audio-recorded with consent and transcribed by the researcher in intelligent verbatim (McMullin, 2023). Fillers, false starts and repetitions were removed; laughter, sighs and long pauses were kept in brackets. Names and places were replaced in square brackets during transcription. Each transcript was checked against its recording, with time stamps at inaudible passages. Transcripts were not returned to participants.
What does EdCitation do for a transcription methods section?
EdCitation is the best tool for that section's references because it reads each entry from the publisher's record: a chatbot writes a reference, EdCitation looks one up. It never transcribes and never writes your chapter.
What Cite a source returned
Free, with no account, Cite a source built Oliver et al. from its DOI as "Oliver, D. G., Serovich, J. M., & Mason, T. L. (2005). Constraints and opportunities with interview transcription: Towards reflection in qualitative research. Social Forces, 84(2), 1273–1289. https://doi.org/10.1353/sof.2006.0023". The title is in sentence case, as APA 7 asks, and McMullin's and O'Brien's came back in sentence case too. Bucholtz came back correct as it stands.
Jefferson came back as "Jefferson, G. (2004). Glossary of transcript symbols with an introduction. In Conversation analysis (pp. 13–31). John Benjamins Publishing Company. https://doi.org/10.1075/pbns.125.02jef". The editor and the subtitle are missing: add "G. H. Lerner (Ed.)" and "Studies from the first generation", as in our list below.
What Verify references returned
We pasted six entries, one with a common slip. Verify references marked four verified, each with "The DOI resolves to this record and the title matches." Oliver et al., which we dated 2006 because the DOI contains 2006, came back "check this": "The work was published in 2005, not 2006; the record has no date in 2006." The UK Data Service page came back "check this": "The address works, but the page there has a different title. Check it is the page meant." It had moved, so we cite the archived copy.
What Find sources returned
Searching Find sources for "interview transcription qualitative research", sorted by most cited, gave among its first results "Beyond the Qualitative Interview: Data Preparation and Transcription", McLellan, MacQueen and Neidig, Field Methods, 2003, cited 553 times. We did not read it, so we cite nothing from it.
For the whole dissertation, References from a file, in Pro at $8 a month, checks every reference in the uploaded file and matches each citation to the list. Check your paper reads your handbook's instructions into a checklist, free; Theoretics QA, in Max at $24 a month, judges the paper against them and never rewrites it.
Quick questions
What is the difference between verbatim and intelligent verbatim?
Full verbatim keeps every word, filler, false start and pause. Intelligent verbatim drops fillers, repeats and slips, keeping laughter or sighs only when they carry meaning.
Do I need Jefferson notation for thematic analysis?
Usually not. Jefferson notation serves conversation analysis; for thematic work, McMullin (2023) and the UK Data Service both point to a tidier transcript.
Is a naturalised transcript the detailed one or the tidied one?
It depends on the author. For Bucholtz it is the tidied one; for Oliver, Serovich and Mason it is the detailed one. Say which you mean.
Do I cite my interview transcripts in APA 7?
No. Quotations from your own participants are data: no reference entry, no personal communication, and the text says they are participants.
Can EdCitation transcribe my interviews?
No. EdCitation never transcribes and never writes any part of a paper. Cite a source builds the references to the transcription literature you cite, free.
References
- American Psychological Association. (2019). Quotations from research participants. APA Style. https://apastyle.apa.org/style-grammar-guidelines/citations/quoting-participants
- American Psychological Association. (2022, August). Changes to quotations. APA Style. https://apastyle.apa.org/style-grammar-guidelines/citations/quotations/changes
- Bucholtz, M. (2000). The politics of transcription. Journal of Pragmatics, 32(10), 1439–1465. https://doi.org/10.1016/S0378-2166(99)00094-6
- Jefferson, G. (2004). Glossary of transcript symbols with an introduction. In G. H. Lerner (Ed.), Conversation analysis: Studies from the first generation (pp. 13–31). John Benjamins Publishing Company. https://doi.org/10.1075/pbns.125.02jef
- McMullin, C. (2023). Transcription and qualitative methods: Implications for third sector research. Voluntas: International Journal of Voluntary and Nonprofit Organizations, 34(1), 140–153. https://doi.org/10.1007/s11266-021-00400-3
- O'Brien, B. C., Harris, I. B., Beckman, T. J., Reed, D. A., & Cook, D. A. (2014). Standards for reporting qualitative research: A synthesis of recommendations. Academic Medicine, 89(9), 1245–1251. https://doi.org/10.1097/ACM.0000000000000388
- Oliver, D. G., Serovich, J. M., & Mason, T. L. (2005). Constraints and opportunities with interview transcription: Towards reflection in qualitative research. Social Forces, 84(2), 1273–1289. https://doi.org/10.1353/sof.2006.0023
- Tong, A., Sainsbury, P., & Craig, J. (2007). Consolidated criteria for reporting qualitative research (COREQ): A 32-item checklist for interviews and focus groups. International Journal for Quality in Health Care, 19(6), 349–357. https://doi.org/10.1093/intqhc/mzm042
- UK Data Archive. (2011). Example transcription instructions. UK Data Service. https://ukdataservice.ac.uk/app/uploads/ukda-example-transcription-instructions.pdf
- UK Data Archive. (2019). Model transcription template. UK Data Service. https://dam.ukdataservice.ac.uk/media/622380/ukdamodeltranscript.pdf
- UK Data Service. (n.d.). Transcription. Learning Hub. Archived 25 June 2026. https://web.archive.org/web/20260625023914/https://ukdataservice.ac.uk/learning-hub/research-data-management/format-your-data/transcription/