The study of color terminology is a substantial field of modern linguistics. Much of the research undertaken today seeks to refine the model presented in the seminal work on color terminology, Berlin and Kay’s 1969 Basic Color Terms: Their Universality and Evolution (hereafter B&K). Their work, the first cross-linguistic study of basic color terms conducted with modern scientific methods, demonstrates that languages encode color terms that denote the same color foci for particular categories. This finding of semantic universals in color names contradicts the extreme relativism associated with the Sapir-Whorf hypothesis that predicts arbitrary, language-specific segmentation of the color continuum. Further, they posit a particular order for the appearance of basic color terms that refer to these categories. This order is applied both synchronically, predicting which categories a language will encode given the number of basic color terms, and diachronically, predicting the order of the division of the color space into new categories as color vocabulary evolves. Subsequent studies refine the model from simple division of the color space to the dissolution of two channels, white/warm and black/cool (Hardin & Maffi: 22). The identification of black and white as the first two color terms encoded in all languages has been reconceived an early system based on brightness that switches to coding by hue as a language’s color vocabulary increases. Casson discusses evidence of this shift in the history of English (Hardin & Maffi: 224-239).

Early theories of color tended to assume that color vocabulary was limited in ancient writings and in primitive cultures because the culture members’ eyes were unable to make the numerous color distinctions evidenced in the basic color terms found in modern Indo-European languages—an idea that was disproved for primitive cultures by Post (B&K: 149). Berlin and Kay, with the assistance of a seminar class of graduate students, collected data from native speakers of twenty languages using a set of 329 Munsell color chips (figure 1). Each informant listed the basic color terms of the language used, and then mapped for each color term the focus and boundaries of the category, i.e. all members of the category as well as the best example of it.

Mapped together, the cross-linguistic data shows that although different languages encode in their vocabularies different numbers of basic color categories, a total universal inventory of exactly eleven basic color categories exists from which the eleven or fewer basic color terms of any given language are always drawn. (B&K: 2)

The terms are listed below in an order that restricts which categories may appear in a language with fewer than eleven basic color terms.

 

A language with n basic color terms must have the first n terms listed from left to right. White and black are distinguished in all languages, and purple/pink/orange/gray are the last basic terms to appear. Green and yellow present a special case: any language that has a basic term for green and/or yellow has a basic term for red, and all languages with a term for blue have terms for both green and yellow, but a language with four color terms may encode either green or yellow. For a language to evolve to encode n + 1 basic colors, it must proceed from left to right; therefore, the arrows indicate the order of appearance of the terms in the development of a given language. Berlin and Kay map this evolution to seven stages: stage I encodes white and black, and one term is added at each stage through stage VI (brown added). The last four terms are added in varying order in stage VII, where there is a tendency for languages in the stage to possess terms for all four remaining colors.

Berlin and Kay posit that as the color space is systematically mapped into a greater number of categories, the area covered by previous categories is reduced. Though it may be difficult for speakers of English to imagine how the meaning RED could be encoded within the word for black, it must be recognized that two-term systems map the entire color space, and that the precise meaning of the words is not simply that of the hues denoted by English white and black, but encodes distinctions of lightness/darkness, brightness/dullness, and perhaps shininess/dullness as well.

Rosch supplanted the idea of two-term systems that encode light colors of any hue in opposition to dark colors with the model of a two-channel system, one for white and the warm colors red and yellow with the other for black and the cool colors green and blue (Hardin & Maffi: 21). The evolution from two categories to six occurs via two stages of dissolution for each channel. Of the four divisions, the first and last are fixed, while the middle two may occur in either order, explaining the two possible orders of appearance for the terms for green and yellow. It seems fitting that green and yellow should share the ability to appear after red, given that the primary colors are red, green, and blue in additive synthesis but red, yellow, and blue in subtractive synthesis; though a specific hypothesis would be premature, it makes sense that linguistic division of a visual characteristic would correlate with optical phenomena conditioned by the properties of light.

Berlin and Kay present the intriguing case of Paliyan, the language of a "technologically marginal" group of southern India (48-50). Paliyan is a dialect of Tamil and is classified as a stage I language despite the presence of five brightness terms, the cognates of which are color words in the adjacent dialect of Plains Tamil:

Paliyan		Plains Tamil
velle	‘illuminated’	vellai	‘white’
manja	‘bright’	manja	‘yellow’
nilam	‘of medium brightness	nilam	‘blue’
sihappu	‘dark’	sivappu	‘red’
karuppu	‘dark’ or ‘in shadow’	karuppu	‘black’

 

Evidence suggests that the Paliyans may originally have spoken a two-term (black/white) language unrelated to Tamil that was supplanted by the language of their neighbors, which coded distinctions of hue as well as of brightness. Thus, the new words from Plains Tamil used to describe the "quality of reflected light" were mapped onto the dimension of brightness which was the only distinction coded in the original language of the Paliyans (50). Berlin and Kay also find a systematic discrepancy in the "premature" appearance of gray in some languages, i.e. in languages that are not yet in stage VII. Though they propose making gray a "wild card" that can appear at more than one point in the sequence, the data is easily incorporated into a model that allows new divisions of brightness even after division according to hue begins. Within Rosch’s model, once the two channels are dissolved into six elements, both brightness and hue may serve as the basis for distinguishing new color categories.

The idea that brightness is evolutionarily precedent to hue as the basis for color terms is taken up by Casson in regards to English. He finds that "Old English" [c. 600-1150] color terms that had pure brightness senses became obsolete or lost their brightness senses in the Middle English period [c. 1150-1500]" (224), while "Old English terms that had hue senses, even those whose hue senses were very minor, survived into the Modern English period and in the process experienced a shift in color emphasis from brightness to hue" (225). An example of the former is torht ‘light, luminous’ which is obsolete, and salu ‘dark, dark-colored, dusky’ which survived as sallow ‘sickly yellow, pallid’ (225-226). As color vocabulary shifted to focus on hue, secondary (i.e., non-basic) color terms began to appear. Secondary terms are coined by metonymy, the semantic transfer of the name of an object to refer to the color of the object. The earliest secondary terms appeared in late Middle English, contemporary with the rise of the textile industry; there is a smaller metonymic distance from a dyestuff to the color it produces than from some other object to the color that describes it, e.g. crimson, indigo vs. lemon, coffee. Secondary color terms may be transparent, as fawn, salmon, pearl, coral, or opaque, as buff, puce, vermilion, once the names of animals or animal products (233). Metonymy is still at work in the coining of secondary color terms today: transparent color–object relationships may be inserted unremarkably in speech, such as She tried to dye her shirt orange, but it turned out more like manila envelope. The transition from transparence to opacity of terms over time is evidenced in the names of Crayola crayons: how many kids learn what color a cornflower is by coloring with the crayon of that name, versus the number able to pick cornflower out of the box because they’ve seen a cornflower? Colors like mulberry, maize, and raw umber are retired in favor of kid-friendly modern referents like macaroni and cheese, denim, and cotton candy (http://www.crayola.com/colorcensus/history/chronology.cfm).

Most of the world’s languages do not have crayons with color names printed on them; technological advances like dyeing are the measures of cultural complexity that predict the number of basic color terms a language will encode. Linguistic and psychological categorization of color has been studied intensely since Berlin and Kay’s groundbreaking work of 1969, but much research supports the model of dissolution of large regions of the color continuum into smaller ones with basic evolutionary trends such as the precedence of brightness divisions to hue divisions, formalized in Berlin and Kay as the black and white of two-term systems, in Rosch as the dissolution of warm and cool channels, and in Casson as the diachronic semantic shift of specific English terms.