In the Odyssey, the sea is "wine-dark." In the Iliad, too, wine-dark. Homer also describes sheep as violet, iron as violet, and honey as green. The sky, in Homer's epics, is never blue. The sky is bronze. The colour blue does not appear in ancient Greek literature at all.
This fact was noticed in the nineteenth century by William Gladstone, yes, the British prime minister, who spent his retirement years studying Homer and observed the complete absence of blue as a distinct category. He also noticed that Homer's colour vocabulary was generally impoverished compared to modern language, and proposed, tentatively, that the ancient Greeks might have perceived colour differently from modern humans.
Gladstone was wrong about the mechanism but right that something interesting was happening. Understanding what makes this one of the more genuinely illuminating observations in the history of linguistics and cognitive science.
Blue's strange absence
Linguist Guy Deutscher picked up this thread and extended it across cultures. Surveying ancient texts from multiple civilisations, Sanskrit, Chinese, Japanese, Hebrew, Icelandic, he found the same pattern. Blue appears last. Black, white, and red emerge in language first, universally. Then yellow and green. Then blue. Every culture that has been studied shows blue coming last in the development of colour terminology, often much later than the other basic colours.
This follows a pattern first documented by Brent Berlin and Paul Kay in 1969, who compared colour terminology across a wide range of languages and found a remarkable cross-cultural regularity: if a language has only two colour terms, they distinguish dark and light. If three, they add red. And so on, with blue always appearing late in the sequence.
Why blue comes last
The most compelling explanation draws on the relationship between colour vocabulary and the things in the natural environment that come in those colours. Red things that matter are everywhere: blood, fire, berries. Yellow and green appear throughout edible plants and natural hazards. These are colours that carry practical significance and therefore got labelled early.
Blue, by contrast, is relatively rare in the natural world in ways that matter. The sky is blue, but it doesn't need to be distinguished from other sky, it's just sky. The sea is blue-ish, but Homer's "wine-dark" suggests the Greeks were more focused on its texture and mood than its hue. There are very few blue foods, very few blue animals that require close discrimination, very few situations where distinguishing "blue" from "green" or "violet" was survival-relevant.
You don't develop a dedicated colour category until you need one. And it turns out, for most of human history, blue wasn't a category you needed.
The experiment that proved perception follows language
The Namibian Himba people have a language with no distinct term for blue but several distinct terms for shades of green that English treats as a single category. When shown a circle of green squares with one blue square, they find the blue square slow to identify, even though to English eyes it is obvious. When shown a circle of green squares with one slightly different green square (a distinction the Himba language marks with a specific term), they identify it quickly. English speakers perform the reverse.
What you can see easily is, to a measurable degree, what your language has given you a word for. This is not a claim that you literally cannot perceive colours you don't have words for, it is a claim that categorical language affects the speed, ease, and automatic salience of perceptual discrimination. Language doesn't change the input; it changes how the input is processed.
What this implies about the rest of perception
Colour is a useful test case because the stimulus is clean and the variation is measurable. But the logic extends to anything that language categories influence, which is a great deal of human experience. The way time is grammatically structured in a language influences how speakers think about the future. The way spatial relationships are encoded influences navigation. The presence or absence of a word for a particular emotion influences how frequently and precisely that emotion is reported and perhaps experienced.
This is not linguistic determinism, the strong claim that language limits what you can think. It is something more modest and probably more true: language shapes the paths of least resistance in cognition. What is easy to say becomes easy to think. What lacks a label requires more cognitive effort to isolate and communicate. The ancient Greeks were not blind to the sky. They just hadn't decided it was interesting enough to name.
Disagree? Say so.
Genuine pushback is welcome. Personal abuse is not.
