Quote a Documentary Participant Without Changing What They Answered

An interview produces a line you want. On tape it came in answer to a narrow question, with a condition attached. In the pitch, the same words sit under a headline about something else, and they read better than they did in the room: cleaner, more certain, more usable. Nothing was altered. The quote is verbatim.

Verbatim is not the same as faithful. A quotation is faithful when it still answers the question it was answering, at the level of certainty the person actually offered. Change the question underneath it — through narration, a headline, an unattributed summary, or two remarks joined into one breath — and you can keep every syllable and still misrepresent the speaker.

So the check happens before you decide the line is good, and it is mostly mechanical: go back to the exchange, test each cut for the claim it changes, choose between a contextual quotation and a transparent summary, and read the finished use rather than the isolated sentence.

Return to the recording and recover the question

A transcript tells you what was said. It doesn't always tell you what the sentence was doing. Locate the surrounding exchange — the question immediately before, and the one before that if the answer ran long — then pin down four things:

• The referent. What do "it," "there," "that," "the program" point to? Pronouns survive cuts far better than their antecedents do.
• The occasion. "Since we started the Thursday group" is not the same statement as "last year."
• The question actually put. Not the question you wish had been put.
• The qualification. Every "if," "unless," "might," "usually," "some weeks," "not really."

Keep a findable source position — recording number and time range, or transcript page and line numbers. Something as plain as "recording 2, 00:31:10–00:31:24" is enough. The point isn't ceremony; it's that the excerpt and its original can be laid side by side later by someone who isn't relying on your memory of the interview.

Where the meaning turns on uncertainty, compare the transcript against the recording. Transcripts regularize speech and lose tone, and a line that reads as a commitment may have been delivered as a shrug. That's not a reason to distrust transcripts in general, since most trims won't turn on a pitch contour. It's a reason to check the particular passage where the certainty is doing the work.

Test every cut for a changed claim

Two tests, in order.

First: what did the cut remove? Some removals are housekeeping. False starts, "you know," a word repeated twice — trimming these rarely changes what was claimed, and a five-second answer that becomes three is usually still the same answer. Other removals are load-bearing:

• A condition — "if the evenings come back."
• A hedge — "might," "probably," "I think."
• A time frame — "back then," "since the new manager took over."
• A scope limit — "some of us," "not for long."
• An attribution — "my sister says," which becomes the speaker's own view once it's gone.
• A hesitation in the place where certainty drops. Trimming an "um" is not the same as trimming the pause inside an answer that goes "yes—well, maybe," because the second cut promotes a maybe to a yes.

Second: what do the new surroundings say the question was? Read the excerpt next to the headline above it, the narration that introduces it, the caption, and whatever is on screen. Context isn't decoration; it assigns the implied question. Set "I might do one session" under a headline about demand for the daytime program and it becomes a statement about the daytime program, whether or not the speaker has ever been near it.

This second test is where verbatim language offers no protection at all. A stitched quote can be word-perfect and still assert a continuity or a certainty the recording never contained. Joining two remarks into one sentence implies they were said one after the other; an ellipsis marks a gap inside a thought, not a bridge between two thoughts. And two separate hedged remarks placed side by side can harden into a plan.

Compare contextual quotation with transparent summary

When an excerpt can't carry its own question, there are three honest instruments.

Restore the context. Put the question back, or introduce the answer with a short framing clause: "Asked whether she'd return if the evening hours came back…" This costs words and can slow a pitch down. It's the cheapest fix when it works.

Summarize in your own voice, attributed as a summary. "The participant said she might attend one session but wouldn't commit to a weekly schedule." These are not her words, and the sentence shouldn't pretend otherwise — no quotation marks, no borrowing her syntax to borrow her authority. Being visibly your language is the virtue here, not the compromise.

Choose a different passage, or don't use one. If the accurate version won't support the claim the story needs, the claim is the thing to revisit. A quote that has to be mis-set in order to fit is usually a sign that the section wants a different source or a smaller assertion.

What you don't do is put your own summary inside quotation marks, or splice separate answers into apparently continuous speech. Either move converts your edit into testimony.

Check the finished use, not only the isolated sentence

The unit of review is not the sentence. It's the exchange, the selected passage, and everything that will assign the passage its meaning: headline, lead-in narration, caption, image, lower third. Put all of it on one page and read it the way a viewer will, top to bottom, once, without the recording in hand.

Then write down two things: what the edit changed, and who still has to check the accuracy. Not "approved" as a status, but the specific outstanding question — "no one has confirmed the evening hours were ever cut," or "this pullquote sits under a daytime headline and the participant was asked about evenings." Naming the change saves the next reader from rediscovering it.

Two different questions get confused here, and they shouldn't. Editorial accuracy is whether the excerpt still means what it meant. Permission to publish is whether you may use the material at all. An accurate excerpt you aren't authorized to use is a rights problem, not a quotation. A signed release doesn't make a mis-set quote accurate. The second doesn't cure the first, and the first doesn't grant the second.

For the accuracy half, the Oral History Association's best-practices page (checked September 8, 2026, in its "Access & Use" section) asks people working with recorded interviews to preserve a narrator's perspective, contextualize excerpts, avoid manipulative use, and identify the fuller interview to users. That's oral-history practice, not a broadcaster's release policy or a legal entitlement, and it isn't a rulebook for a pitch deck. It does name the instinct this article is after: an excerpt shouldn't be made to mean more, or something other, than the exchange it came from.

The same answer, three edits

The exchange below is invented, which makes the arithmetic of each edit checkable. No participant testimony is used here.

The setting: a community center that once ran evening hours and stopped. The interview was recorded in connection with those evening hours, at recording 2, 00:31:10–00:31:24, transcript page 9, lines 4–7. The question and answer run:

Q: If we brought the evening hours back, would you come back? A: I might do one session, if the evenings come back. I couldn't do it every week, though.

Now suppose the pitch isn't about evenings at all. It's for the daytime drop-in program that currently runs, and the producer wants evidence of demand. That's the pressure: a quote about evenings, a story about weekday mornings.

Edit 1 — the pullquote. Narration: "Demand is building for the daytime drop-in program." Pullquote: "I might do one session." Image: the daytime room at 10 a.m., busy.

What changed: the answer now appears to answer a question about the daytime program. The condition is gone, the weekly limit is gone, and nothing in the excerpt says the speaker attends the daytime program — the exchange doesn't establish that. "Might" survives, but it is now a maybe about a different offer. Hedged vocabulary doesn't protect a quote whose referent has moved.

Edit 2 — the contextual quote. Lead-in: "We asked whether she'd come back if the evening hours resumed." Quote: "I might do one session, if the evenings come back. I couldn't do it every week, though."

Both sentences are contiguous in the recording, so nothing has been stitched. The condition and the limit are intact. The cost: it's a quote about evenings, so it supports an evening-hours story and not a daytime one. If the section is about evenings, this is the version to use.

Edit 3 — the attributed summary. "Asked whether she'd return if the evening hours came back, the participant said she might attend one session but wouldn't commit to a weekly schedule."

Shorter than the quote, still carrying the referent, the condition, and the limit. What it gives up is her words, which is the trade. If the pitch depends on hearing a participant speak, this version may read flat, and that's a legitimate reason to prefer Edit 2 wherever the subject actually fits.

What no version does is supply evidence of demand for the daytime program. If the pitch needs that, the options are the ones you actually have — sign-in figures, staff observation, a different interviewee asked about the daytime schedule — or a smaller claim. Omission is an editorial decision, not a failure. Dropping the line and writing "we don't yet have participant comment on the daytime schedule" is more useful to a commissioner than a quote that appears to promise weekly attendance at something the speaker never mentioned.

Keep the two versions side by side

The misleading version is worth keeping next to the corrected one, at least until everyone who signs off on the pitch has seen exactly what moved:

"I might do one session." — under the headline "Demand is building for the daytime drop-in program."

We asked whether she'd come back if the evening hours resumed: "I might do one session, if the evenings come back. I couldn't do it every week, though."

Thirteen words removed, and one question quietly reassigned. That's the whole distance between a faithful quote and a misleading one — same speaker, same recording, same "might."

The test isn't whether the words are exact. It's whether the excerpt still answers the original question at the original level of certainty, and whether a viewer encountering it cold would come away with the exchange that actually happened. If the answer is no, and no honest version fits the slot, leaving it out is a defensible edit — the kind you can explain to a participant, a commissioner, or a lawyer, each of whom may be asking a different question about the same twelve seconds of tape.

Frequently asked questions

Why isn't a verbatim quote automatically faithful?

A quotation is faithful when it still answers the question it was answering, at the level of certainty the person offered. Changing the question underneath through narration, a headline, an unattributed summary, or joining remarks can keep every syllable while misrepresenting the speaker.

What should you recover from the recording before using a quote?

Locate the surrounding exchange—the question immediately before, and the one before that if the answer ran long. Pin down the referent, the occasion, the question actually put, and any qualification such as if, unless, might, usually, some weeks, or not really. Keep a findable source position, such as recording number and time range or transcript page and line.

Which cuts are load-bearing, and how can new surroundings change a quote?

Load-bearing removals include a condition, hedge, time frame, scope limit, attribution, or hesitation where certainty drops. Trimming an um is not the same as trimming the pause inside 'yes—well, maybe,' because the second cut promotes a maybe to a yes. Also read the excerpt next to its headline, narration, caption, and image; those surroundings assign the implied question, and verbatim language offers no protection if the referent has moved.

What are the honest options when an excerpt cannot carry its original question?

Restore the context by putting the question back or adding a short framing clause. Or summarize in your own voice, attributed as a summary, without quotation marks or borrowed syntax. Or choose a different passage, or leave it out. If the accurate version will not support the claim the story needs, revisit the claim. Do not put your own summary inside quotation marks or splice separate answers into apparently continuous speech.

What is the difference between editorial accuracy and permission to publish?

Editorial accuracy is whether the excerpt still means what it meant. Permission to publish is whether you may use the material at all. An accurate excerpt you are not authorized to use is a rights problem, not a quotation problem. A signed release does not make a mis-set quote accurate; the second does not cure the first, and the first does not grant the second.

Further reading

• What Are the Interviews Doing in Your Documentary? — Start here if you have not decided what the participant’s account is for; it asks whether the film needs a witness, researcher, participant, or remembered voice before quotation.
• Whose “I” Is Speaking in a Documentary Pitch? — Read this before writing a first-person line into the deck, especially when you cannot tell whether the words belong to the participant, a paraphrase, or the production.
• Pitch a Verbatim Film Without Treating Recorded Words as a Finished Scene — Read this if the pitch promises verbatim performance or lip-sync, because it separates the source unit from the staged scene and flags omissions that change what viewers infer.