The instrument

The interview protocol

Version 1 ran wave 1 (25 to 35 minutes as designed; 11 to 33 minutes as run). Version 2 is the short form for wave 2, from September 2026: 12 to 15 minutes, built for school days, with every comparable element kept so that wave 2 pools with wave 1. The full script, the consent practice, the written release, and the codebook for the rating data are published here so that a reader can judge the data by the instrument that produced it, and so that a trained interviewer can run a session that meets the archive's standard.

Then, still recording: today's date, the place, the interviewer's name, and the contributor's chosen attribution. The contributor's real name is not spoken on the recording if they chose a code.

Kept from version 1 and asked only after the wrap, so a short session never loses a fixed element to a reserve one.

• Pink as a second deep-dive color (the pink-economy thread).
• Has what a color means to you changed at different stages of your life? (added after wave 1, suggested by contributor CA-01)
• Is there a color you refuse outright? What does it bring to mind? (added after wave 1, suggested by contributor CA-01)
• Name a film you grew up with. How does it use color? (the curator's own question)
• In your culture's eyes, what color is money? Death? Love? Anger?
• Have you noticed ads or brands in your country using a color differently than elsewhere?
• Have you ever changed what you wore, bought, gave, or did because of a color? Tell me about that time.
• Do you have photos (festivals, weddings, packaging, streets) where these colors show up, that could be credited to you?

• Silence is your friend. Count to five before rescuing them; the best material comes after the pause.
• Can you give me a specific example? is the most valuable follow-up in anthropology. Use it constantly.
• Chase stories, not opinions. Tell me about a time beats What do you think about, every time.
• Write down exact words in the original language and script. Paraphrase kills primary data.
• Never correct or debate, even if they contradict your reading. Contradiction is a finding; record it.

• A quiet room and a phone voice memo are enough. Name the file by the contributor's code, never by their name.
• Photograph the completed card before the contributor leaves.
• Within 24 hours: upload the recording and the card photo to the project's Recording and Rating Card folders, and add the row to the private log (code, background, language, date, attribution choice, photos promised).
• Within 48 hours: pull the quotes, the idiom, the taboo, and the generational note into the contributor record while the voice is still in your ear.

Version 1 ran five parts: a warm-up (background, languages, favorite color); a quick-fire round for all six colors in randomized order; a deep dive on three colors, red and pink always plus the color with the strongest reaction, each with five questions (important moments, prohibitions, an idiom with literal translation, the grandparents, a time the color changed a decision) and then three or four culture-wide questions chosen live; the rating card; and a wrap (what was missed, photos, the next contributor, attribution). After wave 1 it gained two questions suggested from inside an interview, on life-stage change and on the refused color, and one custom question from the curator's research interest, on color in a film the contributor grew up with.

Consent is given on the recording before every session. This written form is sent afterward and confirms, in the contributor's hand, how they are credited and what is published. Until it is returned, the contributor appears by code. It follows the Oral History Association's practice on consent, attribution, and withdrawal; see the standards page.

The rating data ship as a flat file, ratings.csv, with the full fieldwork layer in fieldwork.json. Every variable is defined below with the question wording that produced it.

A missing value is an empty cell. It means the digit could not be read with confidence from the card, or the contributor declined that cell; the flag column says which. One wave 1 form displayed the like scale with its poles reversed; that contributor's q1 values are normalized (8 minus raw) and the raw values are preserved on her record.

From wave 2, a trained interviewer may run version 2 and submit the session for review. The curator publishes it as an interview record if it meets the protocol. The path is:

1. Run the consent script on the recording, word for word, and log the contributor's attribution choice.
2. Follow version 2 exactly. Add reserve questions only after the wrap.
3. Use the canonical card and photograph it.
4. Submit the recording, the card photo, and the log row to the curator within 24 hours. Name files by code.
5. The curator reviews the recording against the protocol, assigns the code, transcribes and translates the quotes, enters the card, and publishes the record. A session that skipped the consent script is not published.
6. The interviewer is credited on the contributors list by first name and initial with the role Investigation. The contributor's attribution choice governs everything else.

Interviews are the one part of the collection that requires a person in a room with another person. Opening the protocol to trained interviewers is how the archive grows past one curator's network without lowering its standard; the review step is what keeps the standard.