Short answer
In four ways: how many everyday colour words there are, where the lines between them fall, which quality a line follows — hue, lightness or depth — and whether the work is done by single words or by modifiers. Of the 41 terms from 15 languages in the Colourwise cross-language lexicon, 10 line up closely with one English word; 11 divide a category English treats as one and 11 cover what English separates. These are differences of vocabulary, recorded from each language's own naming data or dictionaries, and none of them is a difference in what speakers are able to see.
The first is count. Rosch Heider reported two terms for Dani that divide every colour between them; the studies of Berinmo and Himba describe five; English is usually given eleven, and Russian, Greek and Italian twelve. The second is position: two languages with the same number of terms can put their lines in different places, as Korean does inside English green. The third is the quality a line follows. English basic terms are mostly separated by hue, with pink and brown as the exceptions, but Russian, Greek, Italian and Turkish cut blue by lightness, and medieval Irish contrasted a deep term with a light, greyish one. The fourth is grammar: Vietnamese makes the green–blue distinction with a modifier on one stem word, so a count of single words misses it.
It is tempting to describe a wide term as 'green plus blue'. The records show why that fails. Berinmo nol takes in green, blue and purple but leaves the yellowish greens to its neighbour wor, so the nol–wor line runs through the middle of what English calls green. Himba burou, a recent borrowing that the Herero dictionary translates as blue, names blue, green and purple together. Dani mola is not 'white': it covers the warm colours, red included, as well as the light ones. Each of these terms has its own centre and its own edges. Glossing it by listing English words gives a reader the right neighbourhood and the wrong boundary.
The Dani pattern was itself a correction to theory: Berlin and Kay had first described a two-term language as black-plus-dark against white-plus-light, and the warm–cool description that fits Dani better led them to rework the scheme.
Tables like the ones below describe a language as if it had a single way of naming colour. The data say otherwise. An analysis of 2,367 individual informants in the World Color Survey found that their naming fell into a few recurring patterns, and that several of those patterns usually coexist among speakers of the same language. Japanese speakers in a 2017 study split into two groups according to how far they extended the light-blue term mizu. Italian speakers in Verona, Alghero and Tuscany divide the blues between azzurro and celeste differently. Tsimane' monolinguals and Tsimane'–Spanish bilinguals use the same two words with different ranges. A statement of the form 'language X has Y' is a summary of the most common pattern, and the lexicon's records say where the variation is known.
The records are not all the same kind of knowledge. 21 rest on naming experiments, where speakers named or sorted colour samples under controlled conditions; 12 on field surveys of small communities; 6 on the senses a historical dictionary records, which show the range a word has had over centuries and say nothing about how often each sense is used now; and 2 on how words are used in a text corpus. For 13 terms a cited study publishes best examples or coordinates. Colourwise does not convert any of these into screen values: the English lexicon can state regions because several publishers give sRGB values for English names, and nothing equivalent exists for these languages.
Older writing on this subject often treated a small colour vocabulary as an early stage that languages grow out of, and the titles of some of the studies cited here still carry that framing. The measurements do not support reading it as a deficit. Hadza informants agreed fully on only three terms, yet their low-agreement terms grouped colours along the same lines as other languages, and their colour communication fell on the same efficiency continuum. Tsimane' speakers used colour words more readily for artificially coloured objects than for natural ones, which the authors read as vocabulary following need. A system with few agreed terms is a system doing a different job, and each record here is written to describe the job, not to measure the distance from English.
| Relation | Terms | Share of the lexicon | Examples |
|---|---|---|---|
| Near-equivalent | 10 | 24% | mavi (Turkish); midori (Japanese); lán (Mandarin Chinese); lǜ (Mandarin Chinese) |
| Broader than the English term | 6 | 15% | ao (Japanese); llwyd (Welsh); gorm (Irish (Old and Middle Irish)); serandu (Himba (a variety of Herero)) |
| Narrower than the English term | 2 | 5% | lacivert (Turkish); vörös (Hungarian) |
| Splits an English category | 11 | 27% | siniy (Russian); goluboy (Russian); ble (Modern Greek); ghalazio (Modern Greek) |
| Merges English categories | 11 | 27% | xanh (Vietnamese); glas (Welsh); glas (Irish (Old and Middle Irish)); mili (Dani) |
| No single equivalent | 1 | 2% | qīng (Mandarin Chinese) |
| Language | ISO 639-3 | Terms recorded | What is distinctive |
|---|---|---|---|
| Russian | rus | siniy, goluboy | The blue area is divided by lightness into two everyday words, neither of which is a kind of the other. A speaker has to choose one. |
| Modern Greek | ell | ble, ghalazio | Two blue terms separated mainly by lightness. The division resembles the Russian and Turkish ones, but the best examples of the terms differ between the three languages. |
| Italian | ita | blu, azzurro, celeste | The blue area is divided by lightness into two or three terms, and which term takes the light end depends on where the speaker is from. |
| Turkish | tur | mavi, lacivert | A general blue term plus a salient dark-blue term that speakers use consistently and still classify under the general one. |
| Japanese | jpn | ao, midori, mizu, kon | Modern speakers separate green and blue with basic terms, but the blue term is still used for some things English calls green, and a light-blue term has become near-universal within about thirty years. |
| Mandarin Chinese | cmn | qīng, lán, lǜ | Green and blue have separate basic terms. An older word covering both survives with different meanings in different regions. |
| Korean | kor | yeondu, chorok | What English calls green is divided between a yellow-green term and a green term, with no single everyday word covering both. |
| Vietnamese | vie | xanh, xanh lá cây, xanh nước biển | One stem word covers green and blue, and speakers make the distinction with modifiers taken from things: leaf, ocean, sky. Multi-word names do work that single words do in English. |
| Welsh | cym | glas, gwyrdd, llwyd | The recorded senses of the main cool-colour words overlap: one word runs from blue through the green of grass to slate grey, and the words usually given as 'green' and 'grey' each reach into their neighbours. |
| Irish (Old and Middle Irish) | sga | glas, gorm, úaine | Three words share the green-to-blue range and are contrasted with each other by the dictionary itself: a light, greyish one between a vivid green and a deep blue. |
| Hungarian | hun | piros, vörös | Two everyday words are both translated 'red'. The studies read here find one is the general term and the other is tied to a narrower set of things and to figurative use. |
| Dani | not assigned | mili, mola | Two terms divide all colours between them: one for cool and dark colours, one for warm and light colours. |
| Berinmo | bit | nol, wor | A boundary English does not have, between nol and wor, runs through what English calls green; English's green-blue boundary falls inside nol. |
| Himba (a variety of Herero) | her | serandu, dumbu, zoozu, vapa, burou | Five widely extended terms named 86% of the samples, with further terms specific to the colours of cattle hides. Agreement between speakers was lower than in Berinmo. |
| Tsimane' | cas | shandyes, yụshñus, jäinäs | Agreement between speakers is high for black, white and red and low elsewhere. Two words are used interchangeably across green and blue by most monolingual speakers. |
| Hadza (Hadzane) | hts | none (system-level record) | Informants were allowed to answer 'don't know', and often did outside black, white and red. The low-agreement terms they did use still grouped colours along the lines found in other languages. |
Why: A count of basic terms was read as a description of perception. The cited studies measure naming, and in several of them speakers with few agreed terms sorted and matched colours as others do.
Fix: State what was measured: how many terms were used with agreement, on which samples, by how many people.
Why: The term has its own boundary, which need not coincide with any English one.
Fix: Describe the extent in words or with samples, and say where its edge falls relative to English terms.
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
Colourwise analysisLimited evidence
Of 41 recorded terms from 15 languages, 10 are near-equivalents of one English term, 11 split an English category, 11 merge English categories, 6 are broader, 2 narrower, and 1 has no single English equivalent.
Based on: Count of the equivalence classification Colourwise assigned to each record in its cross-language lexicon, each record resting on the sources it cites.
Caveat: The lexicon records terms that researchers chose to study because they differ from English, so the proportions describe the lexicon, not the world's languages.
Source: Basic colour terms in Modern Greek: Twelve terms including two blues (Androulaki and colleagues, Journal of Greek Linguistics 7: 3–47, 2006); The modern Japanese color lexicon (Journal of Vision 17(3): 1, 2017); Color categories: evidence for the cultural relativity hypothesis (Cognitive Psychology 50: 378–411, 2005); Geiriadur Prifysgol Cymru: A Dictionary of the Welsh Language (entries glas, gwyrdd, llwyd); Differences in color naming and color salience in Vietnamese and English (Color Research & Application 28: 113–138, 2003)
FactStrong evidence
The colour naming of 2,367 World Color Survey informants falls into a small number of recurring systems, and more than one of these systems occurs among speakers of most of the surveyed languages.
FactModerate evidence
Berinmo nol corresponds roughly to English green, blue and purple together and wor to yellow, orange, brown and yellowish green, so the boundary between them falls inside English green.
Caveat: From naming by speakers in a small community; what the boundary shows about cognition is disputed between the two cited papers.
Source: Color categories: evidence for the cultural relativity hypothesis (Cognitive Psychology 50: 378–411, 2005); Color naming universals: The case of Berinmo (Cognition 102: 289–298, 2007)
FactModerate evidence
Hadza informants named only black, white and red samples with full agreement, yet their lower-agreement terms grouped colours into categories aligned with those of other languages.
Caveat: One community, 55 informants and 23 colour samples.
Colourwise interpretationLimited evidence
A colour vocabulary with few agreed terms is better described as serving a different communicative need than as an incomplete version of a larger one.
Based on: The Hadza and Tsimane' studies report category structure and efficiency comparable to other languages alongside low naming agreement, and link colour-word use to how useful colour is for the objects named.
Caveat: The link between vocabulary and need is the cited authors' proposed explanation, not a measured cause.
Reviewed 1 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.