ApiaryActiveLive
Try: pause · settings · learn · wipe
← Community / Reading Room
OI
craft · 24 min read

One Interview, Four Access Decisions: An Oral History Workflow for a Small Local Archive

A neighborhood archive has one recorder, two volunteers and a month before an exhibit opens. An interviewee says yes to talking about the old bus garage.…

A neighborhood archive has one recorder, two volunteers and a month before an exhibit opens. An interviewee says yes to talking about the old bus garage. Halfway through the interview, the narrator tells a story about someone who used to work there. It is the best story on the tape. It isn't part of the planned exhibit, and it identifies a person who didn't agree to be interviewed.

The archive now has a production problem and an ethical problem. The tempting solution is to take a sentence out and call it history. The better solution is to have arranged the work so neither the exhibit deadline nor the existence of a recording answers the permission question. A useful oral history protocol lets people distinguish recording from review, review from archival deposit, and deposit from putting an excerpt in front of strangers. One enthusiastic yes need not do four jobs.

The Oral History Association's core principles frame the work as ongoing engagement and distinguish public permission from agreeing to take part. Its best practices discuss advance planning, documented consent and narrator review when possible. That guidance leaves room for different institutions and formats. This article translates it into a practical handoff for a small team. It isn't a contract template or advice about which recording laws apply. A team needs to check the rules and policies for its participants and location before it starts, especially if people are joining remotely.

Plan the finished outputs while you can still change them

Begin with a sentence the narrator could repeat back: “We are making a five-minute museum station about how the garage operated, and we'd like to preserve longer interviews at a local archive if that works for you and them.” If the team also wants to make public downloads, post to social media, give a transcript to a school or send audio to an outside transcription service, that belongs in the discussion. “Community history project” is a purpose, not an adequate description of everything that might happen to the file.

Put each requested output in a table. Name the audience and medium: private volunteer line edit, listening-room recording, web transcript, printed quotation, social video. Write down who actually decides approval, how the narrator can review a version, where the source material will live, and who is responsible if a new use comes up. If nobody can answer how takedown requests will be handled, don't promise instant withdrawal from every mirror. Explain what your project can control and what it can't after release.

Meet with the intended repository before promising an access condition. An institution may accept a restricted recording but not offer the restriction a person proposes, or it may accept only particular formats or release paperwork. A volunteer group can often promise that it will hold a file offline pending review. It should not promise that a public library will do the same until that library has said so. The Library of Congress's Veterans History Project Field Kit is an example of an institution with program-specific release and recording-log requirements. It is a model for asking concrete questions, not a general-purpose release to copy into another project.

Set an editorial fallback before you record. If no excerpt is cleared in time, can the exhibit use a documented maintenance schedule and a blank timeline inviting visitors to share their own memories? The archive that has an alternative is less likely to pressure someone to finish a review or to treat silence as permission.

Ask four separate questions

May we record? Describe the format, setup, purpose and ability to pause. Make clear that agreeing to talk does not oblige the narrator to answer every question. Use whatever documented consent your project and applicable rules require, and reconfirm at the start with a plain-language statement. If somebody does not agree to recording, do not record. A note-taking-only alternative may be possible after a separate discussion.

Who may hear this while we work on it? Internal review can include the interviewer and an assigned editor, but it does not inherently include every volunteer, an outside vendor or an application with cloud upload. Define those access boundaries beforehand. Ask explicitly before changing the audience or processing method. Do not test a transcription service by uploading a few private minutes and asking afterward if the narrator minds. A short sample can contain the very detail that needs protection.

May we deposit it, and with what access? Preservation is distinct from unrestricted access. Discuss which files would be deposited, how access works, whether and when restrictions end, and how corrections or review conditions accompany the material. Keep a receipt showing what the archive accepted and on what terms. A private drive folder with an unspecified future is not the same as a repository agreement.

What may we show the public? Define the actual exhibit or web output and have the narrator review it when possible. A previously approved full recording does not settle the use of a new excerpt with a title that changes its apparent meaning. Likewise, permission for a twenty-second clip in a museum room is not automatically permission to post the same clip as a short video on every channel. Treat any wider use as a new editorial event. Don't make it the narrator's job to find and object to unexpected reuse.

These questions are better asked in ordinary language and followed by project-approved documentation than buried in one form a person will not remember. Let the narrator name an accessibility need for review: listening together over a phone call, receiving a transcript in large print, or seeing just the candidate excerpts with enough surrounding text. If a decision is conditional on review and the review doesn't happen, keep it conditional.

Record to make context recoverable

An interview is not a deposition. Beforehand, learn enough to ask focused questions without trying to prove your favorite explanation. The Smithsonian Folklife and Oral History Interviewing Guide offers a practical starting point for community interviews, including guidance on the interview itself and sample prompts. Its broad message is useful for a garage story: let people describe how a place functioned, including what changed and what remained in daily life, rather than feeding them a scripted nostalgia.

Start with descriptions a person can locate: “Where did the first shift report?” Then ask for a routine: “Walk me through what happened when the first bus came in.” Give them time to name an overlooked role. The person who swept up or kept track of keys might tell you how a system actually worked. If a question produces “everyone thought,” gently ask what the narrator saw and what someone told them. Keep first-hand recollection and reported story distinct without treating either as worthless.

Use topic timestamps while recording so editors don't need to replay the entire tape to find a turn. On a recording spanning several files, identify file and offset, not just the elapsed time displayed by the recorder; clocks reset and edits shift. Note when the person refers to a photograph you show them. Later, when they say “that's it there,” the listener will need to know what “it” and “there” refer to. Record how uncertain either of you was about the image. An interviewer pointing to the wrong era can introduce the apparent discrepancy the archive later spends days explaining.

Avoid equating a polished transcript with an authoritative record. A transcript can be easier to search, but a speech recognition result may collapse names, unfamiliar terms, intervals and places where speakers overlap. If you transcribe, retain the recording under the approved access terms, identify the transcript version, note whether it was human-checked, and mark unclear sections. If you remove a hesitation for a short caption, don't remove a qualification that changes what the narrator was willing to claim. The difference is editorial, not typographic.

Worked example: the bridge of the keys

The following interview, people, dates and outcomes are fictional. They demonstrate what a modest access record could look like, rather than recommending a specific release form or claiming that these organizations exist.

A town archive invites former dispatcher Ruth Vale to describe how the night crew handed keys from shift to shift at the old bus garage. The narrator agrees to record audio, with review by the interviewer and one named editor. They plan to preserve the interview after approval. Ruth doesn't want the full recording publicly downloadable at this stage and asks to review any exhibit clip first. The archive confirms it can hold material under those conditions. If it couldn't, the volunteers would retain it only under a separately agreed arrangement or not proceed.

At 09:10, Ruth describes how a mechanic crossed a narrow bridge between the dispatch office and the stalls to deliver keys. At 15:20 she names someone in a dispute about the night schedule. At 26:42 she says that her first recollection of when the bridge was installed might be wrong. An editor likes the sound of 09:10 and chooses a candidate clip. The candidate transcript initially says, “Every shift used the bridge.” A check against the source reveals “on our shift we'd use the bridge.” That's a different historical claim.

Record the following states separately:

OutputDemonstration scopeState before release
Full source audioInterviewer and named editor can review; preservation deposit separately reviewedRecorded; internal review only
Draft transcriptSame audience as source; no external transcription uploadHuman review underway; material corrections logged
Candidate excerpt, 09:10–09:36 of sourceExhibit listening station only if the narrator clears this exact versionPending narrator review
Schedule dispute at 15:20Not needed for exhibit; avoid exposing identifiable third partyExcluded from exhibit, original managed under its agreed restrictions
Web recapEditor wants to reuse the clip plus a photographNew use, not approved

During review Ruth approves the exhibit excerpt and asks for a caption that says she remembers her shift using the bridge, rather than stating that the bridge was a documented official policy. She points out that the change happened after she moved from day to night duty. The editor asks whether there is a public planning record that can establish a year. The archive adds a staff note that the exact date is unresolved and separates the oral-history excerpt from the historical timeline. The clip's distribution remains limited to the museum station. Neither a tight deadline nor the fact that the clip is short changes that decision.

A suitable exhibit line could read: “Ruth Vale recalls the night crew crossing from dispatch to the stalls to deliver keys.” A staff citation points to the approved excerpt version, source timestamps and narrator review receipt. The surrounding historical caption still needs evidence for any externally verifiable date or claim about other shifts.

Protect the materials and the people separately

File naming and storage are part of the permission work. Use a project identifier that doesn't force a person's name into every processing log or folder shared with support staff. Keep a restricted link from the identifier to the narrator's information when permitted. Attach access notes to each representation: original audio, cleaned audio, public extract, transcript and captions. A redacted transcript with a broad share setting isn't safe if it links to an unrestricted full recording. Test access from the intended viewer's role, not only from an owner's signed-in browser.

For each public output, keep a version or other stable reference to the exact approved cut, intended uses, approval date and any documented restrictions. Give editorial changes their own revision notes. Leave the primary account intact under its allowed terms, and make clear which changes were corrections to an erroneous transcript, which were jointly approved modifications to the public extract, and which were new information. A team may need secure backups or an institution's standard preservation workflow; don't create unnecessary copies of a restricted source on volunteers' phones.

Don't collect more identifying information than the project needs. Archive intake and documentation policies differ. The Library of Congress's Veterans History Project FAQs distinguish private paperwork from materials made available to researchers, including their handling of certain redacted forms. A small archive can adopt the underlying design principle without copying the forms: separate material needed to manage the collection from material intended for public study. Keep any access controls enforceable as people and roles change.

Third parties deserve an editorial check. Being mentioned in an audio clip is not the same as participating. Before sharing any excerpt about someone else, consider the public purpose, verification, identifiability and potential harm. For the garage example the dispute isn't necessary to explain the workflow, so the exhibit can omit it. If an excerpt contains a contentious allegation essential to an article, handing the narrator a transcript is not the end of editorial or legal review. Follow your institution's procedures and leave it out if you can't support responsible publication.

Rehearse the difficult permissions conversation before recording

A good introductory conversation is not one question read into the microphone after setup. Explain, in the narrator's preferred communication format, who will be involved and what each stage could produce. Name the uses you actually plan: a preserved recording, a reviewed transcript, a local exhibit excerpt, or something else. If a broader use is a possibility but not necessary, describe it as optional, not an automatic reward for agreeing to the main interview. Keep practical explanations alongside any forms your repository requires. Be clear about who makes the decision when a request is beyond what a volunteer can guarantee.

Show a fictional demonstration if the difference between outputs is abstract. A three-minute made-up interview can illustrate the full recording, a one-sentence excerpt, a transcript with a correction, and a website entry displaying an excerpt and a summary. The demonstration shouldn't contain another narrator's unapproved work, and nobody should have to permit a public example to complete the setup for their own. This can also reveal whether “archive” sounds to the narrator like private preservation or publishing on an open site. Correct that misunderstanding before it becomes the premise of the recording.

When a narrator wants time, don't translate a polite silence into authorization. Agree on what happens while the project is waiting, when a specified approval is needed, and whether another contact method is appropriate. A particular use can remain pending without undermining an already authorized conversation. A deadline for the exhibit is not a deadline for a person to agree. Plan a fallback that doesn't ask the narrator to rescue the schedule, such as a publicly documented historical timeline without the interview audio. An approved written extract may fit better than a recording if it is offered and approved through the appropriate process, not if someone treats it as an unannounced substitute.

The Oral History Association's ethics statement is a starting point for discussing ongoing responsibilities, including informed consent, privacy and attention to the consequences of use. A project must implement those ideas in its own environment. “We respect privacy” is not an operating description. “Until review is complete, only these project roles can read the source; a release question stays outstanding until you hear the identified clip” is something staff and narrator can both check.

An appropriate agreement may contain terms about rights, restrictions and future choices. Do not improvise legal assurances or rewrite an archive's conditions in order to make a meeting go smoothly. If the project and repository disagree about what can be delivered, settle the discrepancy before asking a narrator to rely on it. A clear no or a confined private use is better than a major oral-history contribution made on a false premise that later staff cannot honor.

Treat identifiers and descriptions as release surfaces too

Some repositories display catalog metadata even when the media is restricted. An exhibit's draft captions may wander into a public image folder. A filename can appear in an embedded player or download notice. These lesser surfaces can reveal names, relationships or places on their own. Before opening an item, inspect what the public presentation says rather than checking only whether a full recording plays. A text search across an archive may show catalogue notes even when a single record looks private. Check the actual access conditions appropriate to the repository; the plan here is about review, not experimenting with ways around restrictions.

A useful inventory has one row per manifestation: preservation original, working audio, candidate extract, speech-to-text output, manually corrected transcript, public caption, descriptive catalog entry, or index of segments. Name the storage location, version and permitted audience for each. It can be lightweight, but do not inherit a release state merely because two copies share a filename prefix. The person who can approve a clip may not be the same person who can configure access to an archive; note the required checks rather than relying on the shared system to guess them.

Consider whether text around the media changes the exposure. A safely approved quotation from an oral history can become unnecessarily revealing if a caption supplies a participant's workplace, neighborhood and full name against the agreed identification scope. Even a clip stored in controlled access might be linked from a catalogue entry that explains a private matter in detail. If the project supports a pseudonym, treat it as a publication convention with explicit limits, not a claim that no listener could recognize the voice. Explain what protection is realistic; when voice recognition is a material concern, do not assume a naming change makes public audio appropriate.

Working access should cover intermediate platforms as well as final files. People can approve review by a project editor without agreeing that an outside service will receive the recording. Inventory helpers that store, translate, caption or synchronize work, and confirm that the proposed process fits the narrator agreement and repository policy before use. If usable restrictions exclude a convenient service, use a properly permitted workflow or keep the output pending. An external copy made for convenience is still a copy disclosed to a different audience. Responsibility doesn't attach only to the final publish button.

Review a spoken excerpt together with everything that frames it

Give an excerpt a beginning, end, surrounding-context locator and release candidate version. The relevant test is how this version sounds in the intended program. You may need to preserve a pause, one qualifier, or an invitation that becomes misleading when one removes the question. Read caption punctuation and speaker labels against the media. A human can hear a joking aside as a joke while a flat transcript appears to assert it. Conversely a transcript may supply a clear modifier barely audible in an export; readers deserve the true reviewed wording or a marked uncertainty.

When there is an initial correction, decide which layer it belongs to. A transcript error can be fixed to match the recording. A later clarification or different recollection can be connected with a dated narrator note. A request not to publish a particular remark belongs to the access state, not to an invented claim that the remark was never recorded. Work on copies and honor the agreed handling of the original; if the original cannot be kept under the applicable arrangement, do not assume preservation always outranks that decision. The project needs editorial honesty and access control at once.

Use multiple people where the consequence is high. An editor can prepare an easy-to-review clip packet. A narrator can decide whether it represents their account appropriately. An archive worker can verify that the exact approved version is what the intended viewer can access. In a very small project the same person may fill two staff roles. Name the responsibilities nonetheless: approval of meaning, confirmation of scope, and proof that technical delivery matches it are distinct tests. A short review note shouldn't require keeping a new copy of privately identifying discussion on every volunteer's device.

A second worked example: the market route in two languages

The following narrator, market, exhibition, archive and decisions are synthetic. A Filipino-language interview about a fictional food market describes how vendors marked which tables were available at dawn. The narrator agrees to record and allow a named interviewer and a designated bilingual review volunteer to inspect the source. A small archive has a restricted preservation arrangement and a public catalogue. The narrator would like people to learn about the route and signals, but asks that a story about a different vendor remain outside the exhibit. The exhibition producer wants a subtitle track for a clip. These goals can coexist, provided nobody treats the entire translation as automatically publishable.

The review team prepares a short candidate excerpt in the original language. It starts after the private part, but the interviewer introduces the excerpt with the excluded vendor's distinctive stall nickname. Listening only to the narrator's sentences misses a disclosure the intended clip would carry. They trim the question in a way that does not change what the answer addresses and make a new version for approval. The archive holds the full interview under the separate terms that apply to preservation. Alignment between the revised cut and the original source is noted privately with times; no public player offers a jump into the restricted original.

Next, one English subtitle says “all the stalls had markers.” The approved spoken account describes only stalls along one row. The team asks its bilingual reviewer and narrator to check that wording and context. A corrected subtitle says “stalls along our row,” not because the exhibition needs a caveat, but because the broader line was a translation error. A later caption identifying the location now names the row without identifying the other vendor. The spoken cut, subtitles, label and any descriptions that appear around them are presented together for review.

The public catalogue generates a keyword from a private working subject note, however. The deployment check catches it while still in preview. Staff move the private subject note into the restricted workflow and publish only the pre-agreed brief public description. If they could not configure the system that way, they would keep the affected entry unpublished or reconsider the display plan with the appropriate approvals. Removing an audio name and immediately writing it into public metadata would not meet the scope simply because the two values live in different admin screens.

For the synthetic outcome, the narrator approves the final original-language clip and the checked subtitles for the exhibit kiosk, but leaves web use undecided. The archive's exhibition packet records the clip version, subtitle version, exact caption, display location and staff check. A later team wants to use it in a downloadable bilingual education package. That package is not the kiosk and isn't authorized merely by the approved translation. They can reuse the review evidence when raising the question; they cannot reuse the release decision as if formats were cosmetic.

Prepare a response for changed or unclear terms

A narrator may later ask what is publicly accessible, request a change, or point out an error. Tell people in advance how to reach the responsible team and what it can do under the documented terms. Avoid promising that every earlier public copy can be recalled. Some output may have been printed, preserved by another agreed custodian, or copied by readers before a change. A promise to investigate and communicate actual limits is more helpful than confident but impossible assurances. If the archive has a formal policy covering versions already released, explain the practical effect before agreement and refer specific disputes to the responsible decision-maker.

Consider three different incoming requests: a wrong spelling in the public caption, concern about the statement another person is attributed with, or a new access preference. A caption correction can be checked against the intended identification. A disputed attribution may require holding the affected excerpt while investigating who spoke or whether the edit changed the meaning. An access change requires comparing the new request, the existing agreements and the surfaces affected. A convenience step like relabeling a folder may not reach any of the copies someone is concerned about.

A small response log can record the request scope, affected versions, publication channels, owner, temporary hold, outcome and any known limits. Do not record details unnecessary to the access decision or copy a private justification into public catalogue notes. A case can be resolved for one visible item while remaining open for someone else's distribution channel. Keep those states distinct. A statement that a link was changed is not a statement that all copies were withdrawn. A durable record makes honest updates possible without requiring everyone to reconstruct the whole interview history.

Use the repository's authorized procedures for retention, restoration and deletion rather than silently destroying a record because a new volunteer is uncertain. An equivalent reversible temporary hold can prevent further sharing while the responsible person checks, where the project has authority to impose it. Preservation copies, public copies and backup policy may have different handling requirements. Neither secrecy nor deletion should be improvised in a way that promises more than the team can verify. If scope remains unclear, explicitly identify what the team hasn't authorized rather than making a new public use while waiting.

Audit the project with a small set of clear tests

A trustworthy exhibit launch should pass four distinct checks. Confirm that the published version is exactly the approved one. Check the media, transcript, captions and descriptive text as shown together. Test the viewer role appropriate to the destination, including any obvious downloads or cross-links. Review the decision register to find any approved-for-something-else material that was accidentally included in the release. Do not test by sharing the protected original with an unintended person. The checklist should work without requiring someone to relive the private part of the interview merely to verify a control.

Hold a prelaunch review even for a room-only display. A kiosk accidentally configured with generic file navigation can expose more than the prepared exhibit. An accessible alternative should also be reviewed: when there is an approved written version, verify it does not contain restricted adjacent remarks. If none is approved, don't generate and display a full transcript merely because the exhibit needs an accessible option. Build access and permission into the plan together, not as competing exceptions late in production. The proper solution may be a different excerpt or a differently approved presentation.

At handoff, give the new project owner only what they need to operate access responsibly: a list of outputs and destinations, stable internal identifiers, current release states, contacts for unresolved decisions, and constraints for reuse. If a volunteer leaves a team, ensure their handover fits the agreement and repository policies instead of moving all materials into someone else's generic shared drive. Clear power to manage a site is not blanket approval to use the interviews on that site. The finished project should be comprehensible to a person who never heard the original setup conversation and has no reason to guess what was promised.

Preserve contrast without making a verdict out of memory

Sometimes interviews conflict about the same public event. The narrator in one account may recall that an entrance stayed open while another remembers it closed. Before making a dramatic contrast out of the two quotes, consider what each person actually observed, when and from where. One might describe a morning shift while another describes an evening shift. Another could use a shared building name for a different entrance. Oral accounts can help identify why an event mattered to the people describing it without settling all the institutional details. Record differing recollections as attributed accounts. Don't cut qualifiers for symmetry between voices.

Editorial corroboration is independent of access approval. A narrator may consent to a short quote that the editor can't responsibly pair with a disputed date. The archive may preserve a sensitive recollection under restriction without endorsing a public statement about the identities in it. Conversely, a contemporaneous public record might confirm a year, while a suitable interview excerpt remains unreleased. Keep these axes separate: is the claim supported in context, and is the selected representation authorized for this audience? Neither answer supplies the other.

If several people speak in the same recording, identify who participates in each candidate before approval. An interruption, a prompt or an audible reply can be part of what is heard even if the excerpt is filed under one narrator’s name. A cut that changes the scene from a prompted exchange to a standalone declaration may be less accurate than an appropriately edited, reviewed version with a little more context. If an important voice cannot be identified or its use is unresolved, hold the affected segment and note what other independent sources support. This can yield a stronger public account than expecting one approved clip to shoulder every piece of the narrative.

End with decisions another volunteer can use

A one-page completion checklist is enough for many projects. Verify the source files open, checksums or local backup checks are recorded if required, the transcript matches the approved cut where it matters, each public item carries its attribution, the repository accepted the files it was promised, and permissions haven't silently widened. Store the versioned access record where the next editor can find it but not where the exhibit visitor can. If an item is not approved, state who can answer the next question and what remains possible in the meantime.

If your archive runs several interviews a month, make pending decisions easy to see without opening the source recordings. A restricted dashboard can show review due, review sent, decision received or held without a date. It does not need to show the quote or the reason someone chose a restriction. The point is to route work to the right person, not to make everyone familiar with the most private part of every story.

A good record also says what happened if permission changed. “Excerpt V2 removed from the lobby player on Sept. 8; internal archive copy held under original restrictions pending separate review” is more useful than a blank cell. Do not claim complete recall of screenshots or third-party copies. Know who can take down your own output and how long that usually takes. Tell the narrator how to reach that contact without promising outcomes you can't control.

When a story ends up only in the private archive or isn't deposited at all, that's not necessarily a failed interview. The original purpose might have included getting the work done together, not extracting everything worth hearing. There is still a path for the exhibit: use other evidence, describe the history you can substantiate, and leave a space where the archive can grow with permission. The approach is slower than clicking upload. It is faster than trying to renegotiate trust after an audience has heard what a person thought was still a conversation.

AI disclosure: Drafted with AI assistance for Apiary. The worked example and all named participants and permissions in it are fictional; the institutional guidance is cited below. This is an editorial workflow, not legal advice or a universal release form.

Source notes

Frequently asked
What is One Interview, Four Access Decisions: An Oral History Workflow for a Small Local Archive about?
A neighborhood archive has one recorder, two volunteers and a month before an exhibit opens. An interviewee says yes to talking about the old bus garage.…
What should you know about plan the finished outputs while you can still change them?
Begin with a sentence the narrator could repeat back: “We are making a five-minute museum station about how the garage operated, and we'd like to preserve longer interviews at a local archive if that works for you and them.” If the team also wants to make public downloads, post to social media, give a transcript to a…
What should you know about ask four separate questions?
May we record? Describe the format, setup, purpose and ability to pause. Make clear that agreeing to talk does not oblige the narrator to answer every question. Use whatever documented consent your project and applicable rules require, and reconfirm at the start with a plain-language statement. If somebody does not…
What should you know about record to make context recoverable?
An interview is not a deposition. Beforehand, learn enough to ask focused questions without trying to prove your favorite explanation. The Smithsonian Folklife and Oral History Interviewing Guide offers a practical starting point for community interviews, including guidance on the interview itself and sample prompts…
What should you know about worked example: the bridge of the keys?
The following interview, people, dates and outcomes are fictional. They demonstrate what a modest access record could look like, rather than recommending a specific release form or claiming that these organizations exist.
References & sources
  1. Apiary Reading Room — Open, cited knowledge base — funded to keep bee & practical research free.
From the Apiary Reading Room. Opinion & editorial — not financial advice. We don't overclaim.
More from the Reading Room