Primary data

The Color Testimony Project

The fieldwork arm of the database: recorded, semi-structured interviews about what colors mean to real people, run by the curator with a standardized protocol, paired with a five-scale rating card. This page documents wave 1: what was collected, how, and what it already shows.

Every wave 1 session ran the same five parts, so the material is comparable across contributors: a warm-up (background, languages, favorite color), a quick-fire round (first association for each of the six core colors, order randomized each session), a deep dive on three colors (red and pink always, plus the color that drew the strongest reaction), the rating card, and a wrap (what was missed, photos, the next contributor, consent confirmation). Stories are collected before numbers on purpose: numbers asked first would flatten the stories, while stories told first barely move the numbers.

The protocol also grows on contact with the contributors. After wave 1 it gained two questions suggested from inside an interview, on how color perception shifts across life stages and on the color a person refuses outright, and one custom question the curator wrote from her own research interest: name a film you grew up with, and how it uses color. Wave 2 runs a short form, version 2, that keeps every comparable element. The full script in both versions, the consent practice, the written release, and the codebook for the rating data are on the protocol page; the archives and studies this layer is modeled on, with an audit of the standards it meets and does not yet meet, are on the standards page; the public project log, with counts, releases, open items, and milestones, is on the progress page; the positions in the anthropology of color that the data can be read against are on the theory page.

After the conversation, each contributor rates all six core colors on five 1-7 scales: like to dislike, calm to energetic, traditional to modern, positive to negative, and the strength of the color's tie to a specific meaning in their culture. Cards were completed on paper during in-person sessions, or from a shared digital form on calls, with the handwritten card photographed and sent back.

Scale directions: 1 = like, calm, traditional, positive, weak tie; 7 = dislike, energetic, modern, negative, strong tie. Each color record page shows the same data with its full range, next to the quotes it sits with.

• Red carries the most meaning. Red's cultural-meaning score (6.7) is the highest in the data, and red is the color contributors argue with: luck and celebration, blood and warning, the exam-day shirt, the name never written in red ink.
• Pink is the most contested color. Its like scores split hardest of any color, covering the full 1 to 7 scale: one contributor calls it her lifelong favorite, another says flatly that it is not for her. Three separate contributors told a version of the same arc: pink imposed or adored in childhood, rejected at the age of labels, and, for some, deliberately reclaimed as an adult. This is the project's pink-economy thread caught live.
• Taboos are sharper than preferences. The crispest, most confident answers were prohibitions: no bright colors at a funeral, no name in red ink, no black and white portrait of a living person, no white towel on the head at grandmother's house, and, said unprompted by exactly one contributor, do not wear a green hat.
• The generational stories are the richest material. Harvest red pasted on straw baskets, worker blue as employment and standing, a deep blue bicycle as village status, burial clothes already bought in black, and a Tang dynasty argument that color marked rank rather than gender long before pink meant girls.
• Institutions still paint meaning today. Contributors read markets and states as fluently as rituals: the stock ticker that rises red in China and green in New York, the flag reds of two countries, phone carriers sorted by hue, and the pink blazer that costs a woman authority in a meeting.
• White is missing from the instrument and keeps arriving anyway. Mourning cloth, the robe of Guanyin, the commoner's robe, the bridal taboo: wave 1 makes the case for white as the collection's seventh record.

Wave 1 is a convenience and snowball sample of the curator's own network: all contributors are of Chinese heritage, most are young and bilingual, and the cross-cultural variation captured so far comes from diaspora contexts (mainland China, Canada, the United States) rather than from unrelated cultural traditions. Sample sizes are small; means are descriptive only. Interviews were conducted in Mandarin with English code-switching; the public layer is English only, with translations by the curator and idioms romanized, and the original-language transcripts held in the private tier. Recording lengths differ, and one session lost its rating card to a schedule. Wave 2, under way from September 2026, targets contributors outside the Chinese diaspora.

Wave 2 opened in September 2026 with the short-form protocol (version 2, 12 to 15 minutes), built so that the curator can interview contributors from many national and cultural backgrounds in the windows a school day allows. The target is twelve interviews across at least five backgrounds outside the Chinese diaspora by mid-October 2026. Every comparable element of version 1 is kept and the rating card is unchanged, so wave 2 pools with wave 1. Interviews run by trained interviewers under the protocol are accepted through the contribute page and reviewed by the curator before publication.

Status as of this update: no wave 2 record is published yet. Records are added here as each session is processed and the contributor's attribution is confirmed in writing; the roster above will grow by code, and this paragraph will be replaced by a wave 2 summary table.

Each interview is processed against the six color records: quotes are tagged to colors, ratings join the per-color panels, idioms enter romanized with a literal translation, and quotes appear on record pages cited by contributor code and date. The full transcripts and the name log stay in the private archive; what is published is the structured, consented layer.