I don’t read much popular science. That’s not a boast, as if to say that I’m above such things, but a guilty confession – I ought to read more, but am too slow a reader. That I’m missing out is being confirmed for me now as I finally get round to reading Guy Deutscher’s Through the Language Glass, which was shortlisted for the Royal Society Winton Prize last year. I knew this was a book I wanted to read, because it deals in some detail with the linguistics of colour terminology, which I looked into while writing Bright Earth. I was finally moved to get it after writing the piece below for the BBC Future site a month or so ago, and wanting to do more with this very interesting work. Whether I will be able to do that or not remains to be seen, but I’m glad it motivated me to get Deutscher’s book, because it is absolutely splendid. I remember Richard Holmes, chairing the book prize panel, questioning how helpful it really was for a book to advertise itself with Stephen Fry’s quote “Jaw-droppingly wonderful”, but His Fryness is quite correct. There’s another chapter – well, perhaps another section – that I would have added to Bright Earth, had I known some of this stuff: I wasn’t aware that Gladstone (that Gladstone) had postulated that the invention of new dyes and pigments actually stimulated the development of colour terminology itself, since it was only (he said) when people could abstract colours from their manifestations in natural objects that they figured they needed words for them. It’s not at all clear if this is true, but it is an intriguing idea, and not obviously nonsense.
The artist Derek Jarman once met a friend on London’s Oxford Street and complimented him on his beautiful yellow coat. His friend replied that he’d bought it in Tokyo, where it wasn’t considered yellow at all, but green.
We don’t always agree about colour. Your red might be my pink or orange. Vietnamese and Korean don’t differentiate blue from green – leaves and sky are both coloured xanh in Vietnam. These overlaps and omissions can seem bizarre if they’re not part of your culture, but aren’t even visible if they are.
But we shouldn’t be too surprised by them. The visible spectrum isn’t like a paint colour chart, neatly separated into blocks of distinct hue, but is a continuum in which each colour blends into the next. Why should we expect to agree on where to set the boundaries, or on which colours are the most fundamental? The yellow band, say, is as wide as the cyan band, so why is yellow considered any more ‘basic’ than cyan?
A new study by physicist Vittorio Loreto at the University of Rome ‘La Sapienza’ and his colleagues argues that this naming and hierarchical ranking of colours isn’t, after all, arbitrary. The researchers say that there is a natural hierarchy of colour terms that arises from the interplay between our innate ability to distinguish one hue from another and the complex cultural negotiation out of which language itself appears.
In essence, their argument pertains to the entire edifice of language: how it is that we come to divide the world into specific categories of object or concept that we can all, within a given culture, agree on. Somehow we arrive at a language that distinguishes ‘cup’, ‘mug’, ‘glass’, ‘bowl’ and so on, without there being well-defined and mutually exclusive ‘natural’ criteria for these terms.
But the researchers have not chosen arbitrarily to focus on colour words. These have long been a focus for linguists, since they offer an ideal multicultural example of how we construct discrete categories from a world that lacks such inherent distinctions. Why don’t we have a hundred basic colour terms like ‘red’, ‘blue’ and so on, given that we can in principle tell apart at least this many hues (think back to those paint charts)? Or why not get by with just four or five colours?
In fact, some cultures do. The Dugerm Dani people of New Guinea, for example, have only two colour words, which can best be translated as ‘black’ and ‘white’, or light and dark. A few other pre-literate cultures recognize only three colours: black, white and red. Others have only a handful more.
The curious thing is that these simplified colour schemes are not capricious. For one thing, the named colours tend to match the ‘basic’ colours of more complex chromatic lexicons: red, yellow, blue and so on. What’s more, the colours seem to ‘arrive’ in a culture’s evolving vocabulary in a universal order: first black and white, then red, then green or yellow (followed by the other of this pair), then blue... So there is no known culture that recognizes, say, just red and blue: you don’t tend to ‘get’ blue unless you already have black, white, red, yellow and (perhaps) green.
This universal hierarchy of colour names was first observed [actually Deutscher shows that this wasn’t the first observation, but a rediscovery of an idea proposed in the nineteenth century by the German philologist Lazarus Geiger] by anthropologists Brent Berlin and Paul Kay in 1969, but there has been no explanation for it. This is what Loreto and colleagues now purport to offer. They use a computer model of language evolution in which new words arise as if through a kind of ‘game’ played repeatedly between pairs of individuals in a population: one the speaker, the other the hearer. The speaker might talk about a particular object – a colour say – using a word that the hearer doesn’t already possess. Will the hearer figure out what the speaker is referring to, and if so, will she then adopt the same word herself, jettisoning her own word for that object or recognizing a new sub-category of such objects? It is out of many interactions of this sort, which may or may not act to spread a word, that the population’s shared language arises.
For colour words, this negotiation is biased by our visual perception. We don’t see all parts of the visible spectrum equally: it is easier for us to see small changes in hue (that is, in the wavelength of the light entering our eyes) in some parts than in others. Loreto and colleagues impose this so-called “just noticeable difference function” of colour perception on the inter-agent interactions in their model. That’s what makes it more likely that some bands of the spectrum will begin to emerge as more ‘deserving’ than others of their own colour word. In other words, the population of agents will agree faster on a word associated with some hues than others.
This speed at which a consensus arises about a colour word with an agreed meaning specifies the resulting hierarchy of such words. And the order in which this happens in the computer experiments – red first, then violet, green/yellow, blue, orange and then cyan – is very close to that identified by Berlin and Kay. (Black and white, which aren’t themselves spectral colours, must be assumed at the outset as the crude distinction between dark and light.) Crucially, this sequence can’t be predicted purely from the “just noticeable difference function” – that is, from the physiology of colour vision – but arises only when it is fed into the ‘naming game’.
The match isn’t perfect, however. For one thing, violet doesn’t appear in Berlin and Kay’s hierarchy. Loreto and colleagues explain its emergence in their sequence as an artificial consequence of the way reddish hues crop up at both ends of the visible spectrum. And Berlin and Kay listed brown after blue. But brown isn’t a spectral colour – it’s a kind of dark yellow/orange, and so can be considered a variant shade of orange. Whether or not you accept those explanations for the discrepancies, this model of language evolution looks set to offer a good basis for exploring factors such as cultural differences and contingencies, like those Jarman discovered, and how language gets transmitted between cultures, often mutating in the process.
Paper: V. Loreto, A. Mukherjee & F. Tria, Proc. Natl Acad. Sci. USA doi: 10.1073/pnas.1113347109.