Method and scope

Methodology

This collection makes one kind of claim and avoids another. It claims that a given meaning is documented in a cited source. It does not claim that a meaning is the only one a color holds, or that any culture is uniform. The records are built to keep evidence and interpretation visibly separate.

1. Collect. For each color, published research and reliable reference sources were gathered from cross-cultural psychology, color science, linguistics, history, and economics.
2. Structure. Each record is organized into the same fields: core meanings, cross-cultural variation across at least three cultures, a historical note, the evidence type, a survey-data slot, and a source list. Each record also carries a Dublin Core metadata block.
3. Cite. Every variation claim names its source inline. Every source is listed with a resolving identifier (a DOI, a stable URL, or a full pullable reference) and checked to resolve.
4. Separate evidence from interpretation. Quantitative claims (for example, color-emotion ratings across 30 nations) are sourced to peer-reviewed studies. Cultural-background facts (for example, red envelopes at Lunar New Year) may use a reference source, but never carry a fabricated number.

This build is transparent about how it was made. The structured collection, the sourcing, and the rendering can be assembled from published material. The part that requires original fieldwork, the project's own interviews and rating survey, is supplied only by fieldwork actually done: wave 1 of The Color Testimony Project (eight recorded interviews and seven rating cards, summer 2026) is the primary data now shown on the records, with its instrument and cleaning notes documented.

The fieldwork layer is built as an oral-history archive with a dataset attached, on the model of the Densho Digital Repository (persistent per-interview identifiers and a fixed citation form), PARADISEC (collection, item, and file levels with ISO 639-3 language tags), and ELAR (graded access for sensitive recordings), and it is measured against the World Color Survey and the cross-national semantic-differential studies as the disciplinary precedents for a color instrument. The interviews follow the Oral History Association's Principles and Best Practices on consent, attribution, and withdrawal; every record is expressed in Dublin Core; the rating data ship with a codebook. The standards page lists each precedent, what the archive copies from it, and an honest status for every standard, including the two that are not yet met: a DOI and a faculty advisor.

The published protocol page carries the full interview script in both versions, the consent script, the written release, and the codebook, so a reader can judge the data by the instrument that produced it. The theory page sets out the six positions in the anthropology of color and what each predicts for this data, so the records can be read from any of them.

The collection's records are available as structured data, so the metadata can be reused or checked independently of the rendered pages.

• records.json, the full record set and the source list with resolving identifiers.
• entries.csv, the six color records as a flat Dublin Core table.
• sources.json, the master citation list.
• fieldwork.json, the wave 1 fieldwork layer: contributors by code, first associations, ratings, and cleaning flags.
• ratings.csv, the rating-card data in long format, one row per contributor and color, with a wave column so later waves pool. The codebook is on the protocol page.

• Six colors is a deliberate core, not a complete palette. White, purple, orange, and brown are documented in the literature and are candidates for future records.
• The cross-national studies lean on WEIRD and East Asian samples; coverage of African, Indigenous, and South Asian contexts is thinner, which the records note where relevant (for example, Davis et al. 2021 is valuable precisely because it samples small-scale societies).
• The Berlin and Kay (1969) evolutionary sequence is foundational but contested; the blue record flags this rather than presenting it as settled.
• The fieldwork layer is one wave deep: eight interviews, all with contributors of Chinese heritage across mainland Chinese, Canadian, and American contexts. It is real primary data, and it is honestly small; the records treat the wave 1 numbers as descriptive only. Wave 2, under way from September 2026 with the short-form protocol, targets contributors beyond the Chinese diaspora; no wave 2 record is published yet.
• Written confirmation of attribution is still being collected from the wave 1 contributors, which is why every contributor appears by code and no recording is published. The archive has no DOI and no faculty advisor yet; both are listed as open on the standards page rather than implied.