Color Naming Across Languages

Why languages carve the spectrum differently, what the Russian blues result really showed, and how the Himba story got mangled on its way to the internet.

Short answer: languages differ in how many basic color terms they have and where the boundaries fall, and those differences are neither arbitrary nor decisive. Large cross-language surveys find strong statistical regularities in where the terms land, which is the universalist half of the story. Careful experiments find that your language's categories can speed up or slow down how quickly you tell two colors apart, which is the relativist half. Both halves are supported by evidence. The version you meet online, in which speakers of some languages literally cannot see a color, is not.

The proposed sequence, and what happened to it

The modern discussion starts with the claim that basic color terms enter a language in a constrained order rather than arbitrarily: dark and light first, then red, then green or yellow, then blue, then brown, then the remainder. The proposal was made in 1969 and it was startling, because the previous consensus held that color vocabulary was a free cultural choice.

The sequence did not survive unchanged. By 1999 the original authors had substantially revised it in the light of new data, replacing a rigid ladder with a more flexible account in which languages partition a space of composite categories in a limited number of ways. That revision is the honest state of the field: the strong universal claim was weakened, and something regular remained.

What remained was tested against a much larger dataset. The World Color Survey collected naming data from speakers of many unwritten languages, and analyses of it found that color naming across those languages clusters far more tightly than chance would allow, and that the best examples of each category recur across unrelated languages. A separate analysis found that the naming patterns fall into a small number of recurring motifs rather than spreading evenly.

Why the regularities might exist

Two explanations are on the table and they are not exclusive.

The first is perceptual. The visual system is not neutral about the space it encodes. Some regions are more distinctive than others, which is the same non-uniformity that makes just-noticeable differences vary across the space, so a category boundary in a low-discriminability region is a cheaper boundary to have.

The second is communicative. A 2017 study across many languages found that warm colors, the yellows and reds, are communicated more efficiently than cool ones almost everywhere, and connected this to the statistics of the world: objects tend to be warm-colored and backgrounds tend to be cool. On that account a color vocabulary is shaped by what its speakers need to talk about, which explains both the regularities and the variation without appealing to either pure biology or pure culture.

The Russian blues, carefully

Russian has no single basic word covering the English range of "blue". It has two, one for the lighter range and one for the darker, and the split is obligatory in the way that the English split between "blue" and "green" is obligatory.

A 2007 study used this. Russian and English speakers had to say which of two blue squares matched a third. Russian speakers were faster when the two candidates fell on opposite sides of their language's boundary than when both were on the same side, and English speakers showed no such pattern. Crucially, the advantage disappeared when participants performed a verbal task at the same time, and survived a spatial one, which points at language rather than at some general attentional effect.

The result is real and it is small, and it has been probed since. A 2020 study revisited the same question with different methods and did not find the boundary reflected in processing speed in the way the earlier work reported. That is how this field works, and the responsible summary is that a category effect exists and its size and conditions are still being pinned down.

The Himba story, and how it went wrong

Work with Berinmo speakers in Papua New Guinea, who use five basic color terms, and later with Himba speakers in Namibia, found that color vocabulary affected recognition memory and similarity judgements, and that categorical perception effects differed between languages with similar numbers of terms. The papers are careful and their claims are about naming, memory and categorisation.

What circulates online is a different animal: a demonstration in which Himba participants supposedly cannot pick a blue square out of a ring of green ones, offered as proof that they cannot perceive blue. That framing does not match the published record. What the research supports is that speakers are slower or less accurate at tasks that lean on categorisation and memory when their categories do not line up with the stimulus set, which is a claim about cognitive processing, not about the retina. Nobody in this literature has claimed a difference in the eye.

The distinction matters beyond this example, because it is the same distinction that separates color vision differences, which are about the photoreceptors, from category effects, which are about what happens after.

Where language gets its foot in the door

The most influential synthesis of this debate concluded that language affects color perception in the right visual field more than the left, which fits the fact that the left hemisphere, which processes the right visual field, is where language is usually lateralised. The summary offered was that Whorf was half right, and half is exactly the point: the effect is real, it is confined, and it is nothing like the strong claim that language determines what you can see.

Two mechanisms are proposed and neither requires anything mystical. Labels give you a durable code for a color, which matters when the color has to survive a delay, a theme our color memory page takes up. And a label can bias a judgement at the moment of comparison, nudging an ambiguous case towards the category you have a word for.

What this means if you build things

Color words do not translate one to one. A single English color term can map onto two obligatory terms in another language, and interfaces that pin meaning to a color name will drift in translation.

Do not use a color name as an identifier. "Select the blue option" is a fragile instruction for reasons that have nothing to do with vision, and combined with the accessibility reasons for not using color alone, it is simply not worth it.

Expect boundary disagreement. Where teal becomes blue is not a fact you can settle by asking more people, and a design that depends on everyone agreeing has a problem.

The connection to the games

The experimental paradigm behind most of this research is delayed matching: you see a color, it goes away, and after an interval you have to pick it out or reproduce it. That gap is where language does its work, because a label is what survives the delay when the sensory trace does not.

Hueecho is that paradigm as a game. You are shown a color and asked to reproduce it from memory, and you will notice yourself reaching for words to hold onto: a slightly greenish blue, a dusty pink. Whether that helps or hurts depends on how close the target sits to your language's boundaries, which is the effect in question, running in your own head. The measurement side is on color working memory.

Keep reading

• Color memory and recall: what survives the delay, and what does not.
• Color working memory: how many colors you can hold at once.
• How many colors humans see: discrimination against naming, which are different questions.

Sources: Kay, P. and Maffi, L. (1999), Color appearance and the emergence and evolution of basic color lexicons, American Anthropologist 101(4), 743; Kay, P. and Regier, T. (2003), Resolving the question of color naming universals, PNAS 100(15), 9085; Regier, T., Kay, P. and Cook, R. S. (2005), Focal colors are universal after all, PNAS 102(23), 8386; Lindsey, D. T. and Brown, A. M. (2006), Universality of color names, PNAS 103(44), 16608; Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R. and Boroditsky, L. (2007), Russian blues reveal effects of language on color discrimination, PNAS 104(19), 7780; Martinovic, J., Paramei, G. V. and MacInnes, W. J. (2020), Russian blues reveal the limits of language influencing colour discrimination, Cognition 201, 104281; Roberson, D., Davies, I. and Davidoff, J. (2000), Color categories are not universal: replications and new evidence from a stone-age culture, Journal of Experimental Psychology: General 129(3), 369; Roberson, D., Davidoff, J., Davies, I. R. L. and Shapiro, L. R. (2005), Color categories: evidence for the cultural relativity hypothesis, Cognitive Psychology 50(4), 378; Regier, T. and Kay, P. (2009), Language, thought, and color: Whorf was half right, Trends in Cognitive Sciences 13(10), 439; Gibson, E., Futrell, R., Jara-Ettinger, J., Mahowald, K., Bergen, L., Ratnasingam, S., Gibson, M., Piantadosi, S. T. and Conway, B. R. (2017), Color naming across languages reflects color use, PNAS 114(40), 10785.