Whose “I” Is Speaking in a Documentary Pitch?

Before putting a first-person sentence in a documentary pitch, decide who supplied its words and who is responsible for its claim. An attributed quotation, a filmmaker's paraphrase, and a writer-authored voice can all be useful. They cannot be exchanged without changing what the reader is being asked to trust.

The immediate task is not to remove first person. It is to preserve the speaker, the basis of their knowledge, and the uncertainty that belongs to the account. A forceful sentence becomes misleading when its clarity depends on concealing any of those things.

The comparison below uses wholly fictional testimony written for this article. There are no real interviews, participants, recordings, permissions, or performed voices behind it. The source labels are line references for an invented exercise, not production timecodes.

Give yourself a source before making an extract

Imagine a documentary pitch concerning an interruption to a building's water supply. The example has two speakers, each with a limited account.

Fictional source A

A1: “I think the pump stopped before dark.”

A2: “I did not see who switched it off.”

A3: “I was in the upstairs room when I noticed the taps were dry.”

Fictional source B

B1: “I arrived after dark.”

B2: “Someone at the gate told me maintenance had been called.”

B3: “I did not make that call, and I did not see anyone make it.”

Neither speaker establishes a complete sequence of events. A offers a tentative time and describes noticing dry taps, but does not identify who switched off the pump or describe seeing the shutdown itself. B describes an arrival and something they were told, not a call they witnessed. The accounts do not establish who acted, whether the report about maintenance was correct, or when the call was supposedly made.

Even A's wording carries an assumption worth preserving as theirs: “who switched it off” suggests an action, but the quoted passage is not independent evidence that someone deliberately stopped the pump. A pitch should not resolve that question merely by adopting the speaker's phrasing as its own narration.

The Oral History Association's Core Principles describe remembered accounts as situated in changing personal and social contexts.[^1] That is a reason to attend to what an account is and how it was produced, not a reason to dismiss it. For this exercise, begin with the narrower discipline of not making either speaker know more than their words supply.

Route one: use a quotation and retain its boundaries

An attributed extract can keep A's particular phrasing:

Speaker A — fictional testimony for this example: “I think the pump stopped before dark. I did not see who switched it off.”

“I think” preserves uncertain timing. The next sentence limits what A can say about who was responsible. It does not establish everything A saw or did not see. Keep that boundary: the extract supports neither an added eyewitness claim nor a new assertion that A saw nothing.

The extract does not need to reproduce every surrounding line to remain useful. Here A3 may be supplied in adjacent prose when the basis of the recollection matters: A says they noticed dry taps from an upstairs room. That context identifies the observation this passage supplies. It does not fill in every unanswered question about what A may have witnessed elsewhere.

Keep the functions of quotation and surrounding prose distinct. The quotation preserves the speaker's wording; the prose identifies the relevant context without quietly adding another sentence inside the quotation marks. An attribution is not a license to rewrite.

A name under a pull quote would identify the purported speaker but would not repair an altered claim. Nor would decorative quotation marks tell a reader whether several separate answers had been joined together. Those are editorial decisions that need a source-linked account of their own.

The OHA's Statement on Ethics asks users of oral sources to preserve narrators' meanings when editing or excerpting and avoid changing them through loss of context.[^2] This is ethical guidance for oral-history use, not a universal legal rule for documentary quotation. Applied here as an editorial check, it directs attention to what the shortened version now makes A appear to assert.

Route two: paraphrase openly

A paraphrase puts the wording in the writer's hands. It can reduce repetition or bring the limits of two accounts into one readable passage:

A thinks the pump stopped before dark and says they did not see who switched it off; they describe noticing that the taps were dry while upstairs. B arrived after dark and was told at the gate that maintenance had been called. B neither made nor witnessed the call. These accounts leave the cause and exact sequence unresolved.

This version does not pretend that A or B spoke the final sentence. The writer is comparing the limits of their accounts. That is an additional editorial operation, and the paragraph should remain visibly written in that voice.

Paraphrase is not automatically safer than quotation. Consider the apparently economical replacement:

A places the shutdown before dark. B confirms that maintenance was called afterward.

The wording has strengthened both accounts. “Places” drops the tentativeness of A1. “Confirms” turns B's reported information into corroboration. “Afterward” attaches a chronology that neither source establishes: B arrived after dark, but did not say when the reported call occurred.

Repairing the paragraph does not require a thicket of caveats. “A thinks” and “B was told” are short, precise descriptions of the available knowledge. They are doing more work than a general disclaimer elsewhere in the deck saying that recollections may differ.

Use paraphrase when the recipient needs a comparison or contextual explanation more than the exact wording. Keep quotation when the wording itself matters to the proposed film. The choice is about what the passage needs to carry, not a hierarchy in which direct speech is always more authentic and paraphrase always more convenient.

Audit the stronger “I” the edit has invented

Here is a deliberately misleading construction, not a usable quotation:

“I saw the pump switched off before dark, then called maintenance.”

It crosses several boundaries at once. A's tentative time becomes certainty. “I saw” supplies an observation absent from A's account. B's report about someone else's information becomes an action performed by the apparent speaker. “Then” invents a sequence. Neither source supplied this unified witness.

Labeling the sentence “A and B” would not fix it. There is still a single speaking “I” who appears to have done and seen things that neither person reports. Attribution must fit the sentence's actual construction, not merely list the people whose material was nearby during the edit.

Less conspicuous changes deserve the same attention. Joining B1 and a shortened B2 as “I arrived after dark. Maintenance had been called” removes the person at the gate and changes how directly B appears to know about the call. The events mentioned are similar; the basis of knowledge is not.

An ellipsis would not restore the lost distinction. It can show that words were omitted, but a reader still needs enough of the remaining context to understand the claim. Nor can a source note repair a sentence that says the opposite of the source in the main body.

Compare each proposed extract with its surrounding material. Track changes to who speaks, who acts, what is witnessed, what is reported, how certain the statement is, and when things supposedly happen. These are the places where an apparently cleaner sentence can become a different account.

Route three: disclose an authored voice as part of the form

A documentary may propose a performed, assembled, or otherwise authored voice. The pitch should explain that construction where the reader encounters it, rather than letting the passage pass as an individual participant's testimony.

For the fictional material, one option is a writer-authored montage for two performed voices:

Illustrative authored montage, derived from the fictional A/B passages above. These are rewritten lines for two performed voices, not interview quotations.

Voice 1, drawing on A: I can place it only roughly—before dark, I think. I noticed the dry taps from upstairs. I did not see who switched it off.

Voice 2, drawing on B: I got there after dark. At the gate, I was told maintenance had been called. I did not make that call or see it happen.

Filmmaker's connecting narration: The two accounts leave a gap: neither tells us who stopped the pump or when the reported call was made.

The wording has changed, and the label says so. The montage keeps the accounts separate while making their relationship audible as a proposed form. The connecting narration belongs to the filmmaker, not to either speaker.

A single openly composite voice is another option, but it needs a different pronoun strategy. It should not acquire the experience of one continuous witness just because one performer delivers the lines. For example:

Illustrative writer-authored composite voice. One proposed performer would carry this text; its “I” is a declared storytelling construction, not a participant or eyewitness.

I carry two incomplete accounts. In one, A thinks the pump stopped before dark and says they did not see who switched it off. In the other, B arrives after dark and is told that maintenance has been called. Neither account gives me a witnessed call or an identified person stopping the pump.

Here the “I” belongs to the declared organizing voice. A and B remain attributed in the third person. That choice gives up the intimacy of an apparently individual recollection. In return, it does not counterfeit a person who witnessed everything.

These constructions are not recommendations for every documentary. A composite voice may introduce a layer of mediation the project does not need. Two attributed extracts may be clearer. The pitch should explain the reason for construction and the cost of it, not use the word “composite” as a blanket exemption from accurate representation.

Disclosure also does not establish permission or prove that the underlying accounts are true. Those are separate questions. In an actual project, describe which source material exists and which recordings or performances are only proposed; do not let a polished sample make an unavailable participant voice look secured.

Keep the speaker, performer, and writer distinct

Three identities can occupy one short passage: the person whose account supplies material, the performer who may deliver the text, and the filmmaker who selects or writes it. Sometimes they are the same person. Sometimes they are not.

A pitch that says “A tells us” when the intended sound is an actor reading a rewrite hides a consequential choice. So does a beautifully typeset first-person passage whose construction is disclosed only in an appendix the reader may not associate with it.

For a proposed performed passage, name the mode close to the text: an attributed reading, a rewritten montage, a declared composite, or filmmaker narration. Then state the source relationship in ordinary language. Do not force the reader to infer authorship from italics, a portrait, or a change of typeface.

Keep a fuller map behind the public-facing passage. For this fictional example, it can be small:

For real material, the map needs locators that return a collaborator to the usable source and its immediate context. Keep access and privacy appropriate to the project; the pitch itself does not need to circulate a full private source archive. This article's line labels provide that check only for its fictional text.

Choose the passage the evidence can support

The revised pitch passage could keep the testimony and the filmmaker's comparison separate:

Fictional worked example — attributed quotation and filmmaker paraphrase

A offers a tentative memory: “I think the pump stopped before dark. I did not see who switched it off.” B arrived after dark and says someone at the gate told them maintenance had been called. The proposed sequence keeps those accounts distinct, placing recollection beside reported information without creating a witness to either the shutdown or the call. The unresolved interval remains part of what the film investigates, not something the pitch settles in advance.

That passage is less conclusive than the fabricated first-person account. It is also more exact about what the proposed film has to work with. Its uncertainty produces a genuine question instead of disguising missing knowledge as a revelation.

Choose sentence by sentence. An exact phrase may warrant quotation; a comparison may need paraphrase; a formal idea may call for an openly authored voice. When a necessary qualification cannot survive the chosen form, change the form or omit the sentence. The “I” should not become most persuasive at the point where its speaker disappears.

Sources

[^1]: Oral History Association, OHA Core Principles, “What is Oral History?”, especially principle 3 on recollection and changing personal/social context. Used as methodological context, not as proof or disproof of any account.

[^2]: Oral History Association, OHA Statement on Ethics, “Using Oral Sources,” particularly the guidance on edited or excerpted analyses remaining true to narrators' words and meanings, and conclusions grounded in cited evidence. Ethical guidance is not presented as legal clearance or a universal documentary rule. All A/B testimony, edits, montages, and source mappings in this article are original fictional examples.

Frequently asked questions

What does it mean to decide whose “I” is speaking before using first person in a documentary pitch?

It means identifying who supplied the words and who is responsible for the claim. An attributed quotation, a filmmaker's paraphrase, and a writer-authored voice can all be useful, but they cannot be swapped without changing what the reader is asked to trust.

Why is “I saw the pump switched off before dark, then called maintenance” not a usable quotation from the fictional A/B material?

It crosses several boundaries: A's tentative time becomes certainty; “I saw” adds an observation A does not report; B's report about someone else's information becomes an action by the apparent speaker; and “then” invents a sequence. Neither source supplied this unified witness. Labeling it “A and B” would not fix the single speaking “I.”

Can a paraphrase make an account more misleading than a quotation?

Yes. The example “A places the shutdown before dark. B confirms that maintenance was called afterward” strengthens both accounts: “places” drops A's tentativeness, “confirms” turns B's reported information into corroboration, and “afterward” supplies a chronology neither source establishes. Short repairs such as “A thinks” and “B was told” describe the available knowledge more precisely.

If a pitch uses a composite or performed voice, does disclosure settle the source questions?

No. The pitch should explain that construction where the reader encounters it, and a declared composite voice must not acquire the experience of one continuous witness. Disclosure also does not establish permission or prove the underlying accounts are true; those are separate questions. In an actual project, describe which source material exists and which recordings or performances are only proposed.

What should a source map for real documentary material contain?

For real material, the map needs locators that return a collaborator to the usable source and its immediate context. Keep access and privacy appropriate to the project; the pitch itself need not circulate a full private source archive. The article's line labels provide that check only for its fictional text.

Further reading

• Pitch a Verbatim Film Without Treating Recorded Words as a Finished Scene — Choose this after deciding a passage will be performed or rewritten; it names the source unit and performance method so wording does not imply a finished recording.
• Build a Film Research Dossier Another Writer Can Work From — Go here when the pitch's source map must become a working dossier another person can use to trace each claim to its record and your interpretation.