Short answer
They ask many speakers to list colour words and to name colour samples, then check each word against Berlin and Kay's four criteria: a single unanalysable word, not a kind of another colour, not tied to a few objects, and salient to everyone. Salience is the criterion that can be measured, so it does most of the work, and the others are weighed against it case by case. That is why the verdict is not always yes or no: of the 121 terms in the Colourwise cross-language lexicon, 95 are treated as basic by the studies cited, 5 are basic in some regions only, 6 are contested or emerging, and 9 were tested and rejected.
Berlin and Kay's definition has four compulsory parts. The word must be monolexemic, meaning its sense cannot be worked out from its parts, which rules out 'light blue' and 'bluish'. Its range must not be included in another colour word's, which rules out crimson as a kind of red. It must not be restricted to a narrow class of things, as blond is. And it must be psychologically salient: near the start of the lists people produce, stable in what it refers to, and present in every speaker's vocabulary. The first three are judgements a linguist makes about a word. Only the fourth describes behaviour, and it is the one fieldwork can put numbers on. Supplementary criteria are held in reserve for doubtful cases, among them that a recent loanword is suspect and that a morphologically complex word counts against itself.
The field method devised by Davies and Corbett, used for Russian, English, Estonian, Finnish, Hungarian and Czech among others, has two parts. In the list task each person names every colour word they can think of, and the order is recorded. In the naming task they name 65 colour tiles one at a time. From the lists come two figures per word, how many people gave it and how early; Sutrop's cognitive salience index combines them. From the tiles come how often a word was used and whether it was the majority name for any tile. A word must clear thresholds on these measures. In Budapest in 2002, 39 of 40 people listed piros and it had the highest salience index of any word; vörös, the other Hungarian red, was listed by 12, ranked nineteenth and cleared none of the five thresholds.
A study of Spanish used written lists instead: 201 students had two minutes to write every one-word colour name they could recall, and a word counted as basic for a country if it appeared in more than half of that country's lists.
A word can pass on behaviour and fail on form, or the reverse. The Hungarian words for pink and orange are transparent compounds, 'rose-colour' and 'orange-yellow', and the Finnish word for pink is built on the word for red; all three were accepted as basic on salience and agreement. Mandarin fěnhóng is 'light red' by its form, yet in naming it overlapped red no more than English pink does. Turkish lacivert is salient and consistently used, and still fails, because 85.5% of the students asked said it is a kind of mavi. Hungarian bordó was listed by more people than vörös and was set aside as a loanword first found in 1897 that cleared one threshold of five. A compound for dark brown was the majority name for three tiles and was dropped for being a compound. Each of these is a decision about which criterion outranks which, made by the researchers and stated in their papers.
A second approach avoids deciding in advance. Participants name each of 330 Munsell chips with a single word; a cluster analysis then groups naming patterns, and a statistic chooses how many groups the data support. For 41 students in Taipei this gave eleven categories, and the eleven words that labelled them were each used by at least 93% of participants, with a sharp drop to the next word, used by ten people. The method measures consensus directly and treats a word as basic if a category forms around it. Its limits are its own: asking for single words understates vocabularies that work through modifiers, as the Vietnamese studies argue, and a sample of students in one city finds that city's usage, as the dominance of the loanword for coffee as the Taipei word for brown shows.
Counts move with the sample, the place, the date and the method. Mandarin has been given six basic terms, eight, nine and eleven by different studies, with brown and orange the unsettled categories. Japanese was given eleven from ten observers in 1987 and a candidate twelfth from 57 in 2017. Spanish has eleven in Spain and Mexico and twelve in Uruguay. Finnish has ten, because no one word for purple met the criteria in 2005 and 2006, and the authors expect that to change. Udmurt has ten in the south and eight in the north. Czech has eleven by a field study and twelve by a corpus study that the field researcher argues cannot apply an interview method to text. A bare number for a language, without the study behind it, hides all of this.
| Language | Term | Status | Who says so, and on what grounds |
|---|---|---|---|
| Italian | celeste | Contested | A third basic blue for Tuscan speakers according to Del Viva and colleagues (2022, 31 participants); the same method gave two basic blues in Verona and in Alghero. |
| Turkish | lacivert | Contested | Özgen and Davies (1998) found it meets the salience and consensus tests for a twelfth basic term but fails the test that a basic term must not be included in another; Rätsep (2012) calls it a candidate, not a settled basic term. |
| Japanese | 水 (mizu) | Emerging | Kuriki and colleagues (2017): used by 98% of 57 informants and proposed as a twelfth basic term. Uchikawa and Boynton (1987) did not find it basic. |
| Japanese | 紺 (kon) | Not basic | Kuriki and colleagues (2017). |
| Mandarin Chinese | 青 (qīng) | Not basic | Absent from the nine basic terms of Gao and Sutrop (2014) and from the eleven basic categories of Hsieh and colleagues (2020). |
| Mandarin Chinese | 棕 (zōng) | Not basic | Hsieh and colleagues (2020): used by 10 of 41 participants, below the eleven basic terms. Earlier studies of other samples had reported it as the basic brown. |
| Mandarin Chinese | 橙 (chéng) | Not basic | Hsieh and colleagues (2020): used by one or two of 41 participants. An earlier study (Lu 1997, as cited there) had reported it as the basic term for orange. |
| Vietnamese | xanh lá cây | Not basic | A multi-word name, so it fails the single-word criterion; Alvarado and Jameson (2002) argue that single-word tasks therefore understate Vietnamese colour naming. |
| Vietnamese | xanh nước biển | Not basic | A multi-word name; see xanh lá cây for the argument about single-word criteria. |
| Hungarian | vörös | Contested | Berlin and Kay (1969) supposed Hungarian had two basic reds; Uusküla and Sutrop (2007) and Benczes and Tóth-Czifra (2014) both conclude vörös is not basic. |
| Hungarian | bordó | Not basic | Uusküla and Sutrop (2007): the twelfth most salient word in listing, but it cleared only one of five thresholds and is a recent loan, so it is excluded from the basic terms. |
| Spanish | marrón | Basic in some regions | Lillo and colleagues (2018): basic in Castilian and Uruguayan Spanish; absent from the Mexican lists, where café takes its place. |
| Spanish | café | Basic in some regions | Lillo and colleagues (2018): basic in Mexican Spanish only; no Castilian or Uruguayan participant listed it as a colour. |
| Spanish | morado | Basic in some regions | Lillo and colleagues (2018): basic in Castilian and Mexican Spanish; listed by 2 of 57 Uruguayan participants. |
| Spanish | violeta | Basic in some regions | Lillo and colleagues (2018): basic in Uruguayan Spanish; below the threshold in Castilian and Mexican Spanish, where it was still listed by about two in five. |
| Spanish | celeste | Basic in some regions | Lillo and colleagues (2018): the twelfth basic term of Uruguayan Spanish, listed by 46 of 57 participants there, by 1 of 47 in Spain and 9 of 97 in Mexico. |
| Czech | rudá | Not basic | Uusküla (2008): cleared none of the study's thresholds. Schmiedtová and Schmiedtová had argued from corpus data that it is a twelfth basic term. |
| Finnish | violetti | Not basic | Uusküla, Hollman and Sutrop (2012): Finnish has no basic term for purple, because violetti, liila, lila and sinipunainen all express it; the authors predict violetti will fill the place. |
| Tsimane' | shandyes | Contested | Listed as the most common term for green samples by Gibson and colleagues (2017), who nonetheless report low consensus; no cited study applies the full basic-term criteria. |
| Tsimane' | yụshñus | Contested | Listed as the most common term for blue samples by Gibson and colleagues (2017), with low consensus. |
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.
FactModerate evidence
Berlin and Kay's four compulsory criteria for a basic colour term are that it is monolexemic, is not a hyponym of another colour term, is not restricted to a narrow class of objects, and is salient for informants.
Caveat: The criteria as restated in a later research paper; the 1969 book was not read.
Source: Basic color categories in Mandarin Chinese revealed by cluster analysis (Journal of Vision 20(12): 6, 2020); Basic Color Terms: Their Universality and Evolution (Berlin and Kay, 1969)
FactModerate evidence· Budapest, Hungary· October 2002
In a field study of 40 Hungarian speakers in Budapest, piros was listed by 39 and had the highest cognitive salience index (0.272), while vörös was listed by 12, had an index of 0.019 and cleared none of the study's five thresholds for a basic term.
Caveat: Forty informants in one city; the authors call the study preliminary.
FactModerate evidence· Taipei, Taiwan
Chips named fěnhóng (pink) by one Mandarin-speaking participant were named hóng (red) by another in a median 31% of resampled comparisons, against 32% for English pink and red and 43% for Japanese pink and aka.
Caveat: Forty-one university students; the comparison figures are taken from two earlier studies.
FactModerate evidence· Spain, Mexico and Uruguay
Elicited lists from 201 students gave eleven basic colour terms for Castilian and Mexican Spanish and twelve for Uruguayan Spanish, where celeste appeared in 46 of 57 lists.
Caveat: University students at one university in each country.
FactModerate evidence· 2005–2006
A field study of 68 Finnish speakers found ten basic colour terms and no basic term for purple, which several words compete to express.
Caveat: As summarised in a later comparative paper by the same author; the original 2007 article could not be read.
Colourwise analysisLimited evidence
Of 121 recorded terms, 95 are treated as basic by the cited studies, 5 as basic in some regions only, 5 as contested, 1 as emerging and 9 as not basic; 6 come from historical dictionaries that do not apply the criteria.
Based on: Count of lexicon records by the basic-term status each record reports from its cited sources.
Caveat: The share of basic terms reflects which inventories were added in full; it is not a property of languages.
Source: Florence “blues” are clothed in triple basic terms (i-Perception 13(5), 2022); Diatopic variation in the referential meaning of the “Italian blues” (in Progress in Colour Studies, pp. 83–106, 2018); Turkish color terms: tests of Berlin and Kay's theory of color universals and linguistic relativity (Linguistics 36(5): 919 onwards, 1998); Emphasis on the blue: Turkish basic colour terms (Rätsep; Studia uralo-altaica 49: 447–457, 2012); Güncel Türkçe Sözlük (entries mavi, lacivert); The modern Japanese color lexicon (Journal of Vision 17(3): 1, 2017)
Colourwise interpretationModerate evidence
'Basic' is better read as a graded finding about how a community uses a word than as a property the word has or lacks.
Based on: The same word is basic in one country and not another (Spanish celeste), compounds and loans are accepted on salience in some studies and rejected in others, and counts for one language differ with sample and date.
Reviewed 6 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.