Did you know that one uncertain word can change the meaning of a whole interview quote? In a local-history recording, ‘we did visit’ and ‘we didn’t visit’ make opposite claims. If the audio is muffled and a student quietly selects the word that sounds more natural, the finished transcript may look professional while reporting something that was never established.
The immediate answer is to mark uncertain speech as uncertain, preserve the original recording and identify where another listener can check it. A transcript is a written representation of a recording, not an opportunity to improve a speaker’s story. When a word is unclear, replay only as permitted, use a consistent visible uncertainty marker and a time location, and never present a guess as a confirmed quotation.
This guide is for pupils making oral-history interviews, heritage projects, podcasts, research assignments or recorded class discussions. The same principle helps teachers distinguish a transcription problem from a listening-comprehension error.
Quick answer: a trustworthy unclear-word protocol
- Keep the source: retain the original audio in its authorised secure location.
- Identify the speaker: distinguish who is speaking from what the speaker is saying.
- Replay the uncertain span: compare the word with its surrounding sentence, but do not fill it in from context alone.
- Use a visible marker: for example [inaudible, 02:14], when speech cannot be recovered reliably.
- Separate a tentative reading: if your project allows a best guess, label it plainly as uncertain and do not quote it as certain.
- Mark a timestamp: the listener should be able to locate the passage without searching the whole file.
- Review and document correction: if a later listener confirms the word, update the same transcript with a dated change note.
First diagnose: was it not heard, or not understood?
These are not identical problems. Inaudible means the sound cannot be distinguished reliably—for instance, a bus passes a microphone. Unfamiliar but audible means the speech can be heard, yet the transcriber may not recognise a place name or technical term. In the second case, a directory, spelling supplied by the speaker or other reliable source can help, while the transcript must still disclose significant editorial corrections.
Give a student a fictional audio sentence: ‘We met at the [unclear] gate after lunch.’ Ask what they can safely claim. The accurate first claim is that the interviewee reported meeting at a gate after lunch, but the gate’s identifier is unresolved. ‘Main gate’ and ‘east gate’ may both fit the grammar, and grammar does not settle audio evidence.
A useful diagnostic is to ask the learner what they heard, what they infer, and what an independent listener could verify. If all three answers are identical despite poor sound, the learner may be concealing uncertainty rather than resolving it.
Worked example: an interview note with a missing word
Invented teaching example, not a real interview or student outcome: imagine a 3-minute recording for a school heritage project. At 02:14, a neighbour says, ‘The community hall reopened in [unclear] after repairs.’ A passing vehicle covers the year. The speaker resumes: ‘I remember the dance that weekend.’ The student’s notes elsewhere mention 2018, but those notes were written after the interview.
The tempting but unreliable transcript
The student writes: ‘The community hall reopened in 2018 after repairs.’ That sentence is fluent, but it attributes an exact date to a recording that does not establish the date. It may also confuse the year of a later dance with the reopening. Once placed in quotation marks, the invented certainty becomes especially misleading.
A transparent working transcript
A suitable project-level notation is: ‘The community hall reopened in [inaudible, 02:14] after repairs.’ In a separate researcher’s note, the student can write: ‘The year may relate to a later note about 2018; the recording alone does not confirm this.’ This keeps heard speech separate from a hypothesis.
If a second authorised listen later confirms the speaker said ‘2016’, the transcript can be corrected to ‘2016’, with a record such as ‘02:14 reviewed against original audio on 5 November 2026; corrected former inaudible marker’. The date here is only part of the invented demonstration, not a claim that such a recording exists.
Hand-check the location and evidence trail
Suppose the complete file is 3 minutes 20 seconds long. Its duration is 3 × 60 + 20 = 200 seconds. The uncertain passage at 02:14 begins 2 × 60 + 14 = 134 seconds from the beginning. It sits 200 − 134 = 66 seconds before the end. This calculation helps a reviewer find it, but it does not tell us which word was spoken.
The timestamp refers to the version of the file being transcribed. If a teacher clips out the introduction, later timestamps move. Record the filename and version, or use timestamps tied to the preserved original; otherwise the uncertainty marker cannot be independently checked.
What established oral-history practice teaches
The Library of Congress Veterans History Project’s transcription guidance recommends identifying inaudible speech, preserving the connection to recordings and making uncertain words visible. Its approach distinguishes the value of an accessible transcript from the richer evidential value of the recording.
The Smithsonian Archives of American Art’s Oral History Program style guide similarly provides conventions for editorial brackets, timestamps and inaudible passages. Institutions may use different symbols. In a classroom, the important thing is a declared, consistent convention that does not silently convert speculation into quoted words.
Never claim that your invented bracket marker is the single universal rule. A school or archive may prescribe its own transcription standard. If so, follow that standard while protecting the original content and making uncertainty visible.
Four mistakes that make a transcript look more certain than it is
- Guessing from the story: a familiar date or place may be plausible but is not automatically audible.
- Using an ordinary ellipsis for unclear audio: many guides use ellipses to show omission, not unintelligibility; define any convention before using it.
- Removing an uncertain word without a note: deletion can change grammar, emphasis or evidence.
- Replacing the quote with a polished paraphrase inside quotation marks: a summary of meaning is not a verbatim quotation.
Guided, paired and independent practice
Guided practice: classify the gap
A teacher supplies three invented descriptions, not recordings: A: a surname is clearly pronounced but the spelling is unknown; B: a vehicle masks two seconds of speech; C: a speaker stops halfway through a sentence. Students should classify A as audible with spelling to verify, B as inaudible, and C as an incomplete utterance. Explain why each needs a different note.
Paired practice: audit the transcript
One student marks three doubtful words in a fictional practice transcript with timestamps. A partner reads the provided evidence notes and highlights any word that was silently supplied without audio support. The goal is not to maximise completed words; it is to separate confirmed and unconfirmed ones correctly.
Independent practice: write a quotation you can defend
Given the fabricated line ‘We started work in [unclear, 00:47] and completed it by July,’ write a one-sentence summary. A defensible response is ‘The interviewee said work had begun by some unspecified point and was completed by July’; the recording as described does not establish a specific start date. If the year of July is also absent, do not add it.
Advanced extension: compare an edited transcript with the original audio and create a two-column change log: the original marker, and the documented reason for any correction. This is especially useful when two reviewers disagree about a proper noun.
Practical steps for the learner, parent and teacher
Learner: retain one clearly named original file, keep a versioned transcript, add a timestamp to each unresolved span and write no invented speech inside quotation marks. After a day, try to verify one doubtful passage independently instead of relying on memory.
Teacher: assess the distinction between confirmed and uncertain words, not only how many blank spots remain. A helpful checklist is: correct speaker, accessible timestamp, preserved quote, visible uncertainty and source file identified. Use anonymised demonstration recordings with appropriate permissions.
Parent: help the child practise with a short consensually recorded conversation, if everyone agrees to recording and sharing. Ask which words are actually clear and why a contextually likely word may still be unconfirmed. Respect privacy: never upload or distribute someone else’s voice without authorisation.
Observable progress without invented performance claims
A simple classroom record might use 10 prepared transcript decisions consisting of six clear words and four intentionally unclear intervals. The assessment asks whether the pupil correctly marks each as supported or uncertain. A learner moving from seven correct decisions to nine on comparable fresh items has demonstrated a possible improvement in that task. It does not establish that every real interview will be accurate.
Keep uncertainty reporting separate from spelling, punctuation and writing style. If the learner is still confusing audio evidence with contextual inference, give another short practice rather than a longer transcript. If they can defend their decisions on fresh material, increase the complexity.
Frequently asked questions
Can I put a possible word in brackets?
Only when the project’s policy permits it and the uncertainty is unmistakable. A tentative rendering must never be treated as a proven verbatim word. When in doubt, mark the passage as unclear and add a separate note.
Should I remove pauses and fillers?
Follow the transcript’s stated purpose and the relevant guide. Readable transcripts and strictly verbatim records may make different choices. Do not silently revise meaningful hesitation or change a statement’s factual content.
Does a transcript prove the interviewee’s historical claim?
No. A transcript records what was said. Whether the statement is historically accurate can require separate documentary evidence. Even a perfectly audible date may be a mistaken recollection.
What if the original sound is lost?
State that it is unavailable rather than claiming the transcript was independently verified. Preserve available provenance and label limitations. Do not reconstruct exact quotation marks from memory.
The larger educational aim
Careful transcription teaches a powerful general habit: maintain the boundary between observation and interpretation. This also underpins the Core Aim guide to interview recordings versus memory notes and checking shortened historical quotations against sources.
Explore the Education Hub, the Core Aim learning hub and the diagnosis-and-independent-checking learning reference for wider transferable study habits.
The best transcript is not the one with the fewest brackets. It is the one where another reader can tell what was actually heard, what remained uncertain and what evidence could resolve the question.