Return to Colour terms in language (idea)

Berlin and Kay

My present write-up was preceded by one explaining the basics of [Berlin and Kay]'s seminal study of [colour] terms in language. As that noder has now removed that, I need to recap this thesis.

We tend to think of colours as intrinsic properties of things: they just are red, or yellow, or purple, and anyone with normal sight would agree. We also naturally think of words as transparent [description]s of reality. It takes some linguistic or psychological [sophistication] to realize that different languages chop the world up in different ways. What is the difference between a river, a stream, and a brook? Is a different language likely to describe the same watercourses by a matching three words of its own? In fact we shouldn't expect it to. Likewise, the range of colours is divided up differently in each language. What we think of as natural and even [primary] colours (such fundamental terms as [red], [blue], [green], [yellow]) may be structured very differently for other people. But what Berlin and Kay proposed in their 1969 paper Basic Color Terms was that the way colour is divided up is not entirely arbitrary.

There are a couple of languages that have, remarkably, only two basic colour terms: an example is [Bassa] of Sierra Leone, and another is [Dani] of New Guinea. One word covers [black] and all [warm]/[dark] colours, the other [white] and [cool]/[light] colours. A couple more languages (such as [Shona] of Zimbabwe) have just three basic terms: [red] plus what's left from the above two. Numbers four and five are [green] and [yellow], in either order. If a language has six colours, then the sixth one is something covering [blue].1 If a language has more than six, the next ones it has are also ones we are familiar with in English, such as [grey], [purple]. To sum up, there really are colour regions that are in some way more fundamental: so a language won't have names for three shades of blue but no distinction between red and yellow.

Two points need mentioning: these are basic colours (not mere shade terms like [lavender] or [carmine]), and not derivative from natural objects (like [grass green] or [emerald green]) or other names ([blue-green], [reddish]). Even if a language has only a few basic terms, it can usually describe actual colours perfectly well by using compound terms.2

Now for my original write-up:

English colour terms

[Colour]s are traditionally a source of confusion in language. After Berlin and Kay we know we shouldn't expect them to [translation|translate] neatly, but almost all [dictionary|dictionaries] and grammars were written before Berlin and Kay, or in ignorance of them. If you look up a (say) English-Bulgarian dictionary you'll find words for pink, orange, purple because it's a [dictionary] and there has to be something there, but that doesn't mean that Bulgarian has words for [pink], [orange], and [purple] the way English has.

English is said to have eleven basic colour terms. [Roget's Thesaurus] lists all the colour terms they can think of under the eleven head-words yellow, orange, red, pink, purple, blue, green, brown, grey, black, white, and I think this is accurate for many (most?) native speakers, but I suspect there's a great deal more variation in practice. Some people would say violet instead of purple, saying that [purple] was a shade of [violet], not vice versa. Some people wouldn't make this many distinctions in [basic] terms. Many's the time someone's said "the blue line", "the red car", or whatever, and I've stared at it and muttered silently, "it's not blue, it's purple".

The later basic terms of English, pink, orange, and purple, are spreading to other languages (pink often as the local word for [rose]), but the speakers even of closely related languages like Danish or Dutch might feel these are [imposition]s: to them, they're no more basic than [carmine] or [lavender], but their native system is now being reshaped by awareness of English.

Historically, [Old English] had an eight-colour system: [red], [yellow], [green], [blue], [brown], [black], [grey], [white]. The others developed from particular coloured things: [orange] the fruit; [pink] the flower; and [purple] the dye from the [murex]. Looking into pre-history we can see that green is related to grass and grow, and less obviously yellow to gall and gold.

By the way, this nonsense that black and white "aren't colours" (meaning, if anything, aren't bright colours, or aren't [colourful]) has nothing to do with this language question. I know of no language that handles [spectral] and [grey-scale] colours differently.

Basic and focal

The distinction between basic terms and [shade] terms is important. If something is [ultramarine], it's blue; if it's scarlet, it's also red. But a single shade can't be both red and purple, unless it's on the very border between them.

Berlin and Kay found that the [focal] regions of colours were much better defined than the boundaries. We think of [canary] yellow, [fire engine] red, [sky] blue, [grass] green. These same shades are usually focal across cultures: they reflect the peaks in sensitivity of our eyes. It's harder to assign border colours: a particular shade of [aqua] might be blue or green or both or neither or unclassifiable, depending on the speaker. It also seems easier to classify [spectral] colours, and harder to classify [dingy] or [muted] colours like [ochre] and [beige].

Some of the canonical uses of a word may be well out on the periphery. Neither [red hair] nor [red wine] is anything like [blood red], nor is [white wine] anything like [snow white], and is not of course "white" at all, as the word is normally used. In other languages they might call it green wine or yellow wine. These are equally good, and equally conventional, since we only need to distinguish it from the kind we call red wine or black wine.

Ancient colour terms

There is a traditional puzzle about what colour words ancient cultures had (Greek, Roman, Irish). Many of their words seem to cover baffling arrays of different things, so that one term must be translated blue, green, grey, yellow, white. Traditional solutions have included questioning whether they actually saw as we did; or saying that they didn't name colours by [hue] but by some other factor such as warmth or intensity; or locating the focus of the colour at a surprising midpoint (e.g. Latin caesius partly covering blue and grey gets glossed as 'bluish-grey'); or extracting a supposed common element and coming up with lots of words that mean [bright] or [shining].

I would like to know what study has been done on these since Berlin and Kay, because a lot of the puzzles are probably explainable under one of the heads I've covered above. We have a few peculiar words for animal colours, like [dun] and [roan] and [bay], but we could equally well say "red horse". It doesn't mean we think of them as fire engine red, or that we're twisting or extending red: we just need a [conventional] term to cover a non-focal shade.

One famous instance is [Sappho]'s description, in [To a Young Girl], of the lover trembling and about to faint as khlo:rotéra de poías 'greener than grass'. You can claim that khlo:ros means 'pale' or 'pale green' or poias means some plant other than the bright denizen of today's lawns, but I don't think an isolated example is important, and you can't wring from it a thesis that the Greeks either saw or named colours strangely. (Of course, 'as green as' is a simple metaphor - it's the 'greener(?) than' that's odd.)

Sappho, by the way, also illustrates another peril, of just not getting a reference: a famous phrase from [Homer] is '[rosy-fingered] dawn', which is a pleasing and clear image. Sappho used the same word to refer to the moon. This is an odd image if you think of rhododaktylos (Sappho's brododaktylos) as 'rosy': the element rhodo- looks very like 'red'. But it's actually an unrelated word, [rose] the flower, and she meant white. She also calls [parsley] white: referring to its flowers.

Clearly glaûkos did cover the range of blue, green, and grey in Greek. But I don't know (and my references are useless at telling me, because of how old they are) whether this is truly a basic term in an ancient few-colour system, or whether one of these translations is of the "white wine" type. My dictionary tells me kyaneos is also blue, but doesn't let me judge the range of shades or whether it's basic.

I'm sure there are scholars who have now done this, and much better answers are out there, but I am raising the point as an illustration of how particularly inadequate a simple [dictionary] listing is at conveying the nuances of colour terms.

Beyond the eleven

If we accept the standard eleven as true of English - for some English-speakers anyway, then we have a lot. Only a few languages make distinctions that we don't: [Russian] has different words for [dark blue] and [light blue], and [Hungarian] has a separate word for [dark red]... as have I myself.

To end with personal observations, I have a twelfth basic colour, [crimson]. My family are [artistic], though I seem to lack the gene for telling which end of the brush to point; so I did some tests on them and found they all had extra basic colours. They all have [aqua], and one or two extra colours in the area of [beige], [ochre], or [tan]. My father, surprisingly, has [scarlet] between red and orange. I made sure to confirm this because it seemed much less natural to me than the others. (Well, I suppose I really tested their groupings, not their colour names as such. I would have to have done forced-choice experiments to ensure that their naming matches their grouping.)

The fact that I use the same word blue for a large range doesn't mean that I see them all the same, in any sense. I'd be very happy if my language had two different terms for [dark blue] and [light blue], with their own foci. To a lesser extent I feel the same about brown and green. Yet I am stuck with the basic terms I have: I can't force myself to elevate, say, [indigo] and [azure] to basic status; nor conversely can I ever think of crimson as a shade of red.

I knew someone whose visual [spectrum] was shifted a little into the [infra-red]. I would love to have tested her, though she didn't want to be turned into a psychological experiment. I wonder what basic terms [colour-blind] people have.

Remember, colour naming is not about the ability to discriminate. As far as I know, with obvious exceptions that are clearly genetically aberrant, all people in the world are equally good at discriminating. It is the [naming] that varies so much.

1. The North American language [Cree] has a separate blue but combined green-yellow, in contradiction of this generalization Berlin and Kay made.
2. The New Guinean language [Yélîdnye] is reported as, extraordinarily, not naming all colours. It has words for only black, white, and red, and descriptions for blue, green, and yellow, but these don't cover the rest.