How to Record Oral History in a Second Language

Why Language Matters in Oral History

There is a version of your grandmother's story that only exists in the language she first told it in. The idiom she uses when she describes her mother. The particular word in Yoruba or Garifuna that has no English equivalent — that holds an entire relationship, an entire way of seeing, compressed into a single sound. The rhythm of how she moves between languages mid-sentence, which itself tells you something about where the story lives in her and where it doesn't. Translation catches the content. It cannot catch all of that.

This is the specific problem this post addresses: the interviewer primarily speaks English. The elder primarily speaks — or is most comfortable in, or most alive in — Spanish, Haitian Creole, Gullah, Yoruba, Garifuna, Cherokee, or another language entirely. Or she moves between two of them, switching mid-thought in a way that feels natural to her and slightly disorienting to you. These are not unusual situations for African American, Afro-Caribbean, Afro-Indigenous, and recent immigrant families. They are common ones. And they are not a reason to delay the interview.

What follows is a practical guide for exactly that situation — how to prepare, how to record, what to do afterward, and how to preserve both the original and the translation so future researchers (and future family members) have what they need.

Record in the Language the Elder Speaks Most Comfortably

This is the cardinal rule, and it overrides everything else: always prioritize the speaker's comfort language over the interviewer's comprehension. If your grandmother is more fluent, more expressive, more herself in Haitian Creole than in English — the interview happens in Haitian Creole. Full stop.

Even if you don't fully understand what she's saying — record it. A recording you can't fully parse now can be translated later. A recording you never made cannot. The incomprehension is temporary. The story is irreplaceable. Every oral history researcher who has worked across languages learns this lesson the same way: the recording you almost didn't make because you weren't sure you understood it is often the most important one you have.

If the elder code-switches — moving between Spanish and English, between Gullah and standard American English, between two registers of the same language — let them. Do not redirect. Do not ask them to slow down or stay in one language. The code-switching is part of the story. It tells you which parts of the memory live in which language, which feelings have words in English and which ones don't. Interrupting it is interrupting the story itself.

Note the language or languages in the recording file name and in your interview record sheet — so the recording is findable and identifiable by someone who comes to it later, whether that's a family member, a translator, or a researcher. The Oral History Documentation Guide includes interview record sheets with fields for exactly this — speaker name, language(s), date, topic, and storage location.

Preparing the Interview When You're Not Fully Fluent

Preparation matters more in a multilingual interview than in a monolingual one — not because you need to be fluent, but because showing up having made the effort changes the dynamic in the room. An elder who sees you arrive with questions written in their language — even imperfectly — knows you take the conversation seriously. That is worth more than perfect grammar.

Write your interview questions in both languages before the session. Use DeepL, Google Translate, or a bilingual family member to check your phrasing — not for perfection, but for respect. You are not trying to perform fluency. You are trying to signal that you came prepared, that this conversation is worth your attention, and that you understand the elder will be most themselves in their own language. A rough translation offered with sincerity opens more doors than polished English delivered from a distance. See the 50-question oral history question bank for starting points you can adapt and translate.

Learn a handful of follow-up prompts in the elder's language — enough to keep the conversation going even when you're not following every word. Even imperfect pronunciation opens doors. The phrases that matter most:

• “Can you tell me more about that?”
• “What do you remember about...?”
• “Who else was there?”
• “What happened next?”

These four prompts, in the elder's language, will carry you through more of the interview than you expect. You don't need to understand everything to ask someone to keep going.

Consider bringing a co-interviewer who is bilingual — not to translate live (that interrupts flow), but to catch things you miss and ask follow-up questions in the elder's language that you couldn't. Brief the co-interviewer clearly before you begin: their job is to listen, not to interpret in real time. Live interpretation during an oral history interview breaks the story into pieces. It slows the speaker, creates self-consciousness, and fragments the memory in ways that are hard to reassemble later. The co-interviewer holds the context, asks the follow-ups, and notes what was said — all without stopping the recording to explain it to you. You can debrief afterward.

Recording Tips for Multilingual Sessions

The technical setup for a multilingual session is the same as any oral history session — with a few additional considerations that matter more when the audio will eventually need to be transcribed and translated by someone who wasn't in the room.

Label the file clearly. Use the format: [Name] — [Language(s)] — [Date] — [Topic]. If the session involved code-switching, note both languages: “Grandma Rosa — Spanish/English — 2026-07-02 — Migration north.” This information is what makes the file findable and usable by someone who encounters it years from now without your context.

Record at the highest quality your device allows. 44.1kHz, 16-bit minimum if your app or recorder offers the option. Translation from degraded audio is harder than translation from clean audio — and in some cases impossible when a word is ambiguous. The guide to recording high-quality audio on your phone covers settings, microphone placement, and environment in detail.

Sit closer than you normally would. You want to capture inflection and hesitation clearly — not just the words, but the texture of how they're delivered. In a language you're not fully fluent in, the emotional register carries information that the words alone won't. A transcriber or translator working from your recording can hear that the tone shifted. They cannot hear it if the speaker was six feet away from a phone on a coffee table.

Don't rush the pauses. In many languages and many oral traditions, a long pause before an answer carries meaning — it is not absence but preparation, sometimes prayer, sometimes the effort of translating an experience into words. Wait. The discomfort of silence is yours to carry. It is not the speaker's problem to fill.

If you're recording video: keep the camera on the speaker's face. Lip movement and facial expression add intelligibility during transcription, especially for languages with sounds that are hard to distinguish in audio alone. A translator working from video will catch things that would be lost in audio only.

What to Do With a Recording You Can't Fully Understand

You made the recording. You got home, listened back, and realized you understand about sixty percent of it. This is not a failure. This is the situation you planned for. Here is what to do with it.

Don't let it sit unlabeled. Even if you can't yet work with the content, label the file with the speaker's name, approximate date, language, and topic — whatever you know. An unlabeled audio file from an elder who is no longer living is, for all practical purposes, lost. A labeled file with partial information is findable. Label it now, before you forget what you knew when you made it.

Transcription services that handle non-English audio. Several widely available services work well for common languages: Otter.ai handles English and Spanish. Rev.com handles multiple languages with human transcription available. Sonix supports 50+ languages, including many with smaller speaker populations, and offers AI-assisted transcription at reasonable cost. For lesser-resourced languages — indigenous languages, creoles, and regional dialects — these services may not be adequate. In those cases, look to university language departments, diaspora community organizations, and oral history nonprofits, many of which offer translation and transcription support for exactly these materials. See also the full guide to transcribing oral history interviews for a complete walkthrough of tools and workflow.

Community translation is often the best option. For languages and dialects that professional services handle poorly, community is the resource. Genealogy Facebook groups for specific diasporas — Haitian Creole speakers, Garifuna communities, Gullah-Geechee heritage groups — often have members willing to help with short excerpts, especially when they understand the oral history purpose. University departments — African Studies, Native American Studies, Caribbean Studies, Linguistics — sometimes connect researchers with graduate students or faculty who can assist. Diaspora cultural organizations, churches, and community centers are also worth approaching directly. Lead with the context: you are trying to preserve a family elder's story, not produce a commercial product.

A partial transcript is better than no transcript. A rough translation is better than silence. The recording is the irreplaceable thing. Everything else — transcription, translation, annotation — can be improved over time. The recording cannot be unmade once the opportunity to make it has passed.

Preserving the Original, Not Just the Translation

When you do have a translation, it is important to be clear about what it is and what it is not. The translated version is the access copy. It is the document that lets English-speaking family members read the story, that researchers can work with, that you can quote in a family history document. It is not the archive.

The original recording is the archive. Always keep both. Never delete or overwrite the original to make room for the translation, never replace the audio file with a transcript, never treat the written version as a substitute for the recording. The original, in the original language, is the primary source. The translation is a tool for accessing it.

If you add a transcript, label it clearly: [English translation — translator name — date]. Note who did the translation and when, and, if possible, what their relationship to the language is (native speaker, fluent non-native speaker, academic specialist). Future researchers working with your archive will need to evaluate the translation — to know whether to treat it as authoritative or as a working draft that might benefit from review. Give them what they need to make that judgment.

This is especially important for indigenous languages, creoles, and dialects that exist on a spectrum or carry community-specific meanings. A translation of Gullah made by someone who grew up in the community carries different authority than one made by someone who studied it academically. Both are valuable. Both should be labeled accurately.

The story your grandmother told in Creole, or in Gullah, or in a language you only partly speak — that version is the true record. The translation helps you read it. Don't confuse the two.