HCI & Design at UW
Published in

HCI & Design at UW

There Is No “Blue” in Korean

Different Languages Have Different Colors

This fact that Korean has no equivalent to the English word “blue” was part of what motivated me to join Younghoon Kim (first author), Gabriella Silva Gorsky, and Jeffrey Heer on a project investigating how different languages handle colors differently. I’ll explain this fact about “blue" in detail below, but first:

Please consider taking our 12 minute color perception survey before you read the rest of this post (we could always use more data, and reading this post first might influence your answers).

Collecting Color Names

In order to find out how colors vary between languages we created the color perception survey (linked above) on Lab in the Wild. As part of this survey, we ask people to tell us what languages they speak, and we later ask them to name colors in their primary language. We started by only asking people to name random “hue colors.” We used this limited set of colors first since it lets us look for interesting patterns with limited data:

The “hue colors,” that is, the brightest, most saturated colors that can be displayed on computer monitors (basically the rainbow colors + purple).

Once a language got enough data for comparisons with hue colors, we switched that language over to having people name random colors from all possible rgb colors (which we call “full colors”). We can then get data on browns, grays and all other colors for those languages, though we need a lot more color names to do valuable comparisons.

Note: Other projects have collected some color names before ours.

The Color “Blue”

Researchers (and people who know certain languages) already knew that some languages have two separate color names for what in English is one word: “blue.” We can see this in our data by comparing the hue colors for English, Korean and Russian:

Stacked graph showing the divisions of the “blue” hue colors in English, Korean, and Russian. The area above each color swatch square what color names are used for it. The color name areas sizes are proportional to how often the names are used and they areas are colored with the average color given to that color name.

As you can see in the above, the “blue” area in English is used almost to the left end of the graph where the greens are, while in Korean and Russian, the dark blues (“파랑” and “синий”) only extend part way to green, and there is a significant light blue color (“하늘” and “ голубой”) which extends the rest of the way to green.

We can use the full color data we collected to provide another view of the difference between English “blue” and the two Korean blues, this time also including colors that are not full brightness and saturation (e.g., blueish grays):

The color ranges of the English word “blue” and the Korean words “파랑” and “하늘”. We used self-organizing maps based off the full color data to find a representative grid of colors.

Differences in how languages handle blue colors have been shown to influence perception. For example, one study found that Russian speakers were relatively faster at picking out color differences along the синий/голубой (light blue/dark blue) boundary, while English speakers didn’t show this improvement.

English and Korean Divisions of Colors

We collected enough color names to compare the full color range in English and Korean. Doing this let’s us find more differences than just “blue”:

Map of the color divisions in English and Korean. Colors are binned in CIELAB color space. Squares are scaled by the largest percentage of a name given to that square (e.g, “63% of people called this square ‘blue’”). To color the squares, we found that largest percentage name, then used the average color of that name. We found 10 color regions in English, and 16 in Korean.

We can see that where English had one word “green,” Korean had two words: “연두” (light green) and “초록” (darker green):

The color ranges of the English word “green” and the Korean words “연두” and “초록”. We used self-organizing maps based off the “full color” data to find a representative grid of colors.

What Does This Mean for Translations?

We can use our data to suggest color translations, choosing names that have the most similar range of colors. We can compare this with what online translators suggest, often finding our suggestions are different.

Sometimes our data indicates that there is no translation that would give a matching color range. As we saw before, there is no word in Korean that means the same set of colors as English “blue” or “green.” Going the other direction, our 10th most common Korean word, “청록,” doesn’t have a particularly good translation, but the closest translations according to our model were “dark turquoise” (226th English color) and “teal” (10th English color).

The color ranges of the Korean word “청록” and the English words “dark turquoise” and “teal”. We used self-organizing maps based off the “full color” data to find a representative grid of colors.

When there aren’t translations that preserve color names, then there might be color areas on images or visualizations that can be easily referred to in one language, but not another.

Even if we are able to come up with color words that mean the same color range, they won’t necessarily have the same connotation. For example “연보라” is a very common color (7th most common color name when we asked Korean speakers to name random colors), but the closest English word is the slightly obscure “lavender," (22nd English color name). Within English we also find examples of connotation differences, such as “pea green” and “puke green” referring to the same color ranges but with different meanings (the same with “blood red” and “deep red”).

Explore More

  • Compare the hue colors of many languages here.
  • Compare how Korean and English divide the full color spectrum here.
  • Find color translations and color synonyms here.
  • Discover which skin colors are and which are not labeled by English speakers as “flesh,” “skin,” and “nude” (if you don’t already know).
  • Read our peer reviewed paper from EuroVis.
  • Download our data set and color naming models from our open source project.

The research paper we wrote was peer reviewed and accepted at the 2019 EuroVis Conference:

Color Names Across Languages: Salient Colors and Term Translation in Multilingual Color Naming Models. Younghoon Kim, Kyle Thayer, Gabriella Silva Gorsky, and Jeffrey Heer (2019). EuroVis.


  • Take and share our color perception survey to get us more data.
  • Help us translate the perception study directions, this blog post (with relevant modifications; just note at the top the that it is a translation of this post and let me know), and translate our visualizations pages (though those would also need technical work to set up).
  • Contribute to our open source project, improving our analysis of the raw data, creating more friendly downloadable color info files, and improving our current visualizations and creating new ones.

You can contact me at kmthayer@uw.edu or visit my website http://www.kylethayer.com.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store