Abstract
Berlin and Kay (1969) hypothesized that the naming of color across world languages is universally constrained. In previous work (PNAS, 2006, www.pnas.org/cgi/reprint/0607708103v1), we confirmed this hypothesis by examining the color naming patterns of the World Color Survey (WCS; www.icsi.berkeley.edu/wcs/data.html), a large database of color names in 110 unwritten world languages elicited using 330 Munsell samples from a standard color chart. In that study, we examined the 14336 purely chromatic color naming patterns using unsupervised k-means cluster analysis, and we found that the color naming patterns clustered into a set of universal chromatic color categories. Those universal categories glossed readily to single or composite color terms from English.
We now extend our analysis to the 7,830 achromatic WCS color naming patterns that included one or more of the ten achromatic samples, which English speakers call black, white, or gray. As the k-means parameter, K, which specifies the number of categories, varied between 2 and 10, we found that the achromatic categories that appeared glossed readily to composites of English color name categories: (1) Chromatic colors with either low or high Munsell value (lightness) were more likely than colors with middle lightness to be included in the set of achromatic color naming patterns. (2) “Warm” chromatic colors were associated with mid to light gray. (3) “Cool” chromatic colors were associated with dark gray or black. (4) Brown was associated with gray. (5) Brown-or-purple was associated with gray and black.
These results, together with our prior results for chromatic color naming patterns, are consistent with the view that color categorization is guided universally by a dimensionalization of color space that is similar to that seen in English speakers. This was true, regardless of the number of distinct words in a WCS informant's color lexicon.