You Only See Colors You Can Name

While color is a purely visual phenomenon, the way we see color is not only a matter of our visual systems. It is well known that we are faster in telling colors apart that have different names, but do the names determine the colors or the colors the names? Recent work shows that language has a stronger influence than previously thought.

The Sapir-Whorf Hypothesis

If and how much language shapes our thought has been the subject of many debates over the years. In the 1930s, Edward Sapir and Benjamin Lee Whorf described a view that language determines our thinking: if we don’t have a word for a concept, we cannot think about it. This was a popular view for a while, but fell out of favor in the 1960s. The pendulum then swung the other way, with researchers believing that there was no connection between language and thought, and that language was a purely abstract construct.

In the last 20 years or so, a middle ground has started to develop. While it’s clear that language does not entirely determine our thinking, there is certainly an influence. The surprising thing is how deeply seated that influence can be.

Russian Blues

In their paper, Russian blues reveal effects of language on color discrimination, Jonathan Winawer, Nathan Witthoft, Michael C. Frank, Lisa Wu, Alex R. Wade, and Lera Boroditsky looked at differences in how native English and Russian speakers distinguish shades of blue.

It turns out that there is no single word for the English “blue” in Russian. The term siniy describes what most other languages know as dark blue, while goluboy is the name for lighter blues. The question is, does that difference mean that there is a difference in color perception between Russian speakers and speakers of other languages, like English?

The test Winawer and colleagues came up with is based on the well-known fact that it is easier for us to distinguish colors that have different names. When shown a reference color and two possible matching colors, we’re much faster when presented with, say, blue and orange than just two shades of orange.

The question is whether that is also true for Russian speakers and their different words for shades of blue. After all, our color names might be based on the same perceptual effects that our color perception uses to distinguish categorically different colors.

The result was that Russian speakers did indeed have an advantage over English speakers in telling siniy and goluboy apart. The authors of that paper then went on to test whether the reason was really language and not some genetic variation or similar. They had the study participants recite nonsense words (to keep their language centers busy) while performing the study, and found that under this condition, the difference went away.

It was clearly the language system interfering with a task that was presumably purely visual: distinguishing between different colors. Categories in our thinking may go much deeper than we think.

The Himba Tribe

A tribe in northern Namibia, named the Himba, have seemingly unusual names for colors. What the video embedded below (linked here for people reading this in their newsreaders) shows is that those names make it easier for them to see some color differences that most other people would find very difficult, whereas they have trouble telling colors apart that look quite different to most of us.

What the video unfortunately does not discuss is why they have these names for colors. There is a slight hint when one of the tribesmen describes several things that are “white,” like milk and water. It seems to me that their color names do not only (or primarily) describe hue, but also function of the things whose color they name. This is a very pragmatic way of using language, and is not unlike some languages whose grammatical genders are based not on sex, but on classes of things and animals that are more specific, like large vs. small animals, plants, dead things, etc.


The impact of language and higher-level concepts on visualization is the key to understanding how visualization actually works. Abstract concepts like color, shape, size, etc. seen in isolation elicit associations and embellishments that influence what we see and how we think about it.

Caroline Ziemkiewicz’s work on visual and verbal metaphors in tree visualization and the role of gravity in visualization is a case in point. Even seemingly pure and abstract depictions of data are influenced by assumptions about the world and/or the way we think about the structure of the data.

The beauty of visualization is not only its visual nature and all the complexity it brings with it, but especially the deep connections we’re only discovering as we dive deeper into it.

“Color names” taken from The Doghouse Diaries, used under Creative Commons.


  1. Anonymous says

    Meaningless statement: “It turns out that there is no word for “blue” in Russian. ”

    How about: “There is no single word in Russian that corresponds to the English word ‘blue’.”

  2. says

    I think it’s fair to say Russian lacks “blue” the same way Emglish lacks a word for… that hue of which red and pink are two examples. How do people who speak languages that do not (yet) distinguish between red and pink do, compared to English speakers who have the two words?

    I say “yet” because English only acquired the word “pink” in the seventeenth century, naming the colour after the flower. “Orange” and I think “violet” come in about the same time, named after the fruit and the flower, respectively.

    I’ve often thought we need a pithier single word in English for baby blue or sky blue, to replace the lame “blue modified by an adjective”.

  3. says

    Robert, I really appreciate how you promote insight from other disciplines to improve the field of visualization. I commend you for it!

    And yet, as a semanticist let me tell you that visualization as a field has only just begun to reach out in earnest to other disciplines. Quite frankly, the color perception research has been among the most influential data in semantics since the 70s!

    Yes, categories go much, much deeper than what you even hint at in your blog post. There is so much more work to be read about categorical perception that could inform design choices. In fact, conflating the Sapir-Whorf debate with theories of the lexicon (that’s the linguistic discipline dealing with how language entries and categories are stored in the brain) is rather unhelpful to that end.

    Speaking of color terms: color terms are not alike. Some colors are basal categories and others derivational subentries to that basal category. What constitutes a basal category is language specific and influences the way the derivational categories are perceived. It just so happens that English has 6 categories unlike other languages that have more or make do with just two (“light/dark” which is not the same as black/white, because red is then a subentry to “light”). Japanese has long before the Boroditsky paper been famous for placing blue and green in one category! And yet, when you show a nonmetrosexual English male a mauve mallow (note that the name for the color is derivational of an object prototypically sporting the hue – if you know french it’s more obvious) he is very well able to recognize it, even if he is much less able to distinguish it from other subcolors of the same category than to distinguish between categories.

    Before I rant on about how words and concepts are different entities (you can express concepts for which your language does not have a word!) I’ll just end with the conclusion: there is a great deal of interdisciplinary potential that is still untapped.

  4. Anonymous says

    The colour blind do not necessarily differ in their descriptions or names of “colours”. The “colours of the rainbow” are all there but not in the same width or depth. It is strange. My red is brown, I am told.

  5. Josef Fruehwald says

    It could only interesting to ask why the Himba language divides up the color spectrum the way it does if it would be equally interesting and informative to ask the same question about English. But then, informative to what end?

    This is where some people take the Sapir-Whorf hypothesis and depart sharply from linguistic and psychological research, and instead try to figure out how linguistic structures are actually reflections of the fundamental nature of a culture. Typically, these attempts are over-simplistic, and suspiciously re-enforce the stereotypes which pre-existed the linguistic “evidence”.

    You, in fact, did not say that the *Himba people* are pragmatically concerned with the function of things, which is good. For most people, though, the distinction between a language and the people who speak it is not a very clear one, and I’m concerned that your proposal is easily misread as a cultural or racial generalization, which would be unfounded.

Leave a Reply