Chapter Number of Basic Colour Categories

by Paul Kay and Luisa Maffi

1. The World Color Survey

The following four maps (132A, 133A, 134A, 135A) show the distribution of colour terms in some of the world's languages, based on the World Color Survey. This project, a collaboration of researchers at the University of California, Berkeley, and the Summer Institute of Linguistics (SIL), collected colour-naming data from 110 languages being studied by SIL linguist-translators in the late 1970s. For further details on methods and results of the World Color Survey see Kay et al. (1991), Kay and Berlin (1997), Kay et al. (1997), Hardin and Maffi (1997), Kay and Maffi (1999).

2. Basic colour term concepts

In recent models of the universals and evolution theory of basic colour terms, stemming originally from Berlin and Kay (1969), three kinds of universal colour categories are recognized (Kay and McDaniel 1978, Kay et al. 1991, Kay et al. 1997, Kay and Maffi 1999): primary colour categories, composite (disjunctive) colour categories, and derived (intersective) colour categories.

First, there are the six primary colours recognized in the widely accepted theory of colour opponency (Hering 1964 [1920], Jameson and Hurvich 1968; Hurvich 1982):

  • Black

  • White

  • Red

  • Yellow

  • Green

  • Blue

These are the two pairs of fundamental opponent chromatic colours <Red, Green> and <Yellow, Blue> and the single pair of fundamental achromatic colours <Black, White>. Black and white are not opposed in the same sense as the chromatic pairs. For example, we can see a combination of black and white (gray) but there is no sensation consisting of the combination of red and green or of blue and yellow. (That blue and yellow pigments can be mixed to produce a green pigment is not relevant; there are no bluish yellow or yellowish blue sensations.)

We will refer to the six colours black, white, red, yellow, green and blue as primaries.

Secondly, there are composite categories consisting of fuzzy unions of the primaries. (Colour categories are modeled as fuzzy sets, that is, sets whose membership is a matter of more or less. For non-technical discussion of fuzzy sets, fuzzy intersection, fuzzy union, and related concepts in relation to colour categories, see Kay and McDaniel (1978); for technical discussion of fuzzy sets and fuzzy logic, see Zadeh (1996).)

By far the most widespread composite in the ethnographic present is green-or-blue or “grue” (see Map 134A). Over half the languages in the World Color Survey contain a grue term. Other major composite terms documented in the World Color Survey are, using “/” for fuzzy union, Black/Green/Blue, White/Red/Yellow, Black/Blue, Red/Yellow, Yellow/Green/Blue, and Yellow/Green. (There is one, somewhat unclear, instance of a probable White/Yellow term in Waorani, an isolate of Ecuador.)

There are also what we might call minor composites. For example, in systems that lack terms for purple and brown, these colours are often included in the black term. Similarly, in systems restricted to the primaries, non-primary shades such as brown, purple, pink, orange and gray are characteristically either split between two adjacent primaries or covered by a single adjacent primary.

The third type of widespread colour category are the derived (intersective) categories, which are experienced as mixtures of the primaries. These are based formally on fuzzy intersections of primaries. Gray is experienced as a mixture of Black and White, pink of White and Red, orange of Yellow and Red, purple of Blue and Red, and brown (arguably) of Yellow and Black.

3. An evolutionary sequence of basic colour systems

Figure 1 presents both a typology and an evolutionary sequence of basic colour systems covering 101 of the 110 World Color Survey languages. (The colour lexicons of the remaining nine languages are composed mostly of terms denoting categories appearing in Figure 1. Limits of space preclude further discussion here; see Kay and Maffi 1999.)

Basic colour lexicon types are shown in square brackets and evolutionary stages are represented by columns, labeled with Roman numerals. For example, stage I contains a single type, consisting of two terms, one a composite of White, Red and Yellow and the other a composite of Black, Green, and Blue. (There are no stage I languages in the World Color Survey sample. Stage I is shown here for theoretical completeness.)

Arrows indicate observed evolutionary transitions between types. Arabic numerals within brackets indicate the number of World Color Survey languages conforming to the corresponding colour lexicon type. Arabic numerals placed between basic colour lexicon types indicate the number of World Color Survey languages transitional between the two types.

Derived categories are not considered in the typology presented in Figure 1. Usually, derived categories appear only after stage V, where each primary has received a separate term, but there are many exceptions. Creating a type for every resulting combination would lead to pointless complexity.  Composite terms are characteristic of the earliest stages of development. Evolutionary development consists essentially in a succession of increasingly fine partitions of the perceptual colour space. The vast majority of World Color Survey languages, 83%, belong to a type or transitional type along the single evolutionary trajectory I > II > III Green/Blue > IV Green/Blue > V.

Figure 1. A Typological and Evolutionary Scheme Covering Most World Color Survey Languages (Source: Kay and Maffi 1999, Figure 3.)

4. The colour term maps

Four colour term maps are presented. They tabulate the data from the 110 World Color Survey languages and comparable data from nine additional languages (Berlin and Kay 1969; Xu 1994). The World Color Survey sample hardly overlaps the standard sample of the World Atlas of Language Structures and care should be exercised in comparing the colour maps with other maps in this atlas.

The four colour term maps address four major types of questions a reader might have regarding the distribution of a basic colour term system in the world:

  1. What is the basic evolutionary stage? That is, how many non-derived (i.e., either composite or primary) terms does the system contain? This amounts to assigning each language to a column in Figure 1.

  2. How many basic colour terms (including derived terms) does the system contain?

  3. How are the “cool” primaries green and blue treated?

  4. How are the “warm” primaries red and yellow treated?

4.1. Map 132A: Number of Non-Derived Basic Colour Categories

Because there is sometimes synonymy in basic colour terms, it is more revealing to tabulate the number of categories that are encoded by basic colour terms than to count the number of basic colour terms per se. The number of non-derived basic colour categories encoded in basic color terms is a feature with self-explanatory integer values, with the exception that we define values 2, 4, and 6 to represent colour lexicons judged to be intermediate between two types which encode the adjacent whole numbers of non-derived basic colour categories.

__values_132A__
Values of Map 132A. Number of Non-Derived Basic Colour Categories

The most frequent type in this sample of mainly unwritten contemporary languages is one with five non-derived basic color categories encoded. Languages with fewer than five non-derived basic color categories encoded as separate terms tend strongly to occur in the tropics.

4.2. Map 133A: Number of Basic Colour Categories

This feature is similar to the preceding except that in this case (1) all basic colour categories, including derived categories, are counted and (2) ranges of numbers of categories form the values of the feature.

Values of Map 133A. Number of Basic Colour Categories
 ValueRepresentation
3, between 3 and 4, or 4 categories 20
Between 4 and 5, 5, or between 5 and 6 categories 26
6 or between 6 and 7 categories 34
7 or between 7 and 8 categories 14
8 or between 8 and 9 categories 6
9, between 9 and 10, or 10 categories 8
More than 10 categories 11
Total:119

As on the previous map, a tendency may be observed for the number of basic colour categories encoded to vary inversely with distance from the equator.

4.3. Map 134A: Green and Blue

Given the typology of Figure 1, and the fact (not noted there) that Green and Yellow can be associated, there are six ways green and blue can be encoded.

  1. Green and blue each gets a basic color term (as in English, for example).

  2. A green-or-blue category get a basic color term (“grue”).

  3. A black-or-green-or-blue category gets a basic color term.

  4. A black-or-blue category gets a basic color term and green gets a separate basic color term.

  5. A yellow-or-green-or-blue category gets a basic color term.

  6. There is a term for yellow-or-green and a term for blue.

In addition, we need a seventh value, not applicable, for languages in which either the green sensation or the blue sensation (or both) fail to be included in any basic colour terms.

__values_134A__
Values of Map 134A. Green and Blue

The most frequent way of encoding green and blue in this sample of languages is in a grue term.

4.4. Map 135A: Red and Yellow

Given the typology of Figure 1, and (as before) the fact that Green and Yellow can be associated, there are four ways red and yellow can be encoded – as well as a value “not applicable”.

  1. Red and yellow each gets a basic colour term.

  2. Red-or-yellow is a basic colour term.

  3. Red gets a separate basic colour term and yellow-or-green-or-blue is encoded as a basic colour term.

  4. Red is a separate basic colour term and yellow-or-green is encoded as a basic colour term.

__values_135A__
Values of Map 135A. Red and Yellow

Red and yellow are much less frequently encoded in a single composite than green and blue. In fact, no language has to our knowledge been reported – in the World Color Survey or elsewhere – that gives separate basic colour terms to green and blue while retaining a red-or-yellow composite.