Comment by jedberg | Hacker News Reader

Comment by jedberg | Hacker News Reader

This title isn't a great title and the article doesn't explain the concept very well.

What this is trying to say is that human languages all over the world come up with color names in a similar path, but they stop at different points.

It's all very fascinating, and interestingly is the argument some use against strong AI -- that is they believe that to truly understand language you must have a human or human-like body that can perceive the world in the same way as a human, because these color words develop based on the physical properties of your eye and brain. For example, many people that are blind from birth have a hard time understanding the concept of color.

A few of you mentioned that Russian has two words for red. Russian is actually at the end of the spectrum (pun intended) -- that is they have the maximum number of color words.

English is close but we don't have separate words for blue and light blue.

You can see this as English speaking children learn words -- they will call "pink" "red" until they learn pink, and will almost always learn "red" before "pink".

Some cultures just stop with "warm" and "cold" (which are red orange yellow vs green blue purple).

Ah the joys of being a CogSci major -- we had entire weeks of classes devoted to this.

>English is close but we don't have separate words for blue and light blue.

Azure. Celeste. Cerulean. Cyan. Teal. Turqouise. aqua. Aquamarine. Lapis.

None of those are light blue :-)

Any of these light blue?

http://www.sherwin-williams.com/homeowners/color/find-and-ex...

Yes, but see my other comment about why they don't count (they aren't basic color terms).

Azure would be a subset of light blue in Russian.

None of those are the same as the russian color, but more importantly, those aren't considered basic color terms[0]

Basic color terms need to be:

- monolexemic ("green", but not "light green" or "forest green"),

- high-frequency, and

- agreed upon by speakers of that language.[7]

All of those words you listed aren't high frequency.

[0] https://en.wikipedia.org/wiki/Color_term#Basic_color_terms

That definition seems arbitrary and debatable.

For instance, what is high frequency?

And what if say blind speakers of the language disagree?

> For instance, what is high frequency?

The 300 most frequent words in a representative corpus of text.

> And what if say blind speakers of the language disagree?

Blind speakers represent a small portion of all speakers, so even if every one of them disagrees, the majority of speakers will still agree.

The 300 most frequent words in a representative corpus of text.

That doesn't seem to be a very good definition. As we subdivide colors into greater specificity, it'll naturally occur that each subdivision is referenced fewer times. So this definition intrinsically precludes a discovery of finer distinctions.

IMHO, the whole idea that language shapes thought and perception (see TFA reference to Whorf) is overblown. I've seen this refuted where one person made the claim that while Nepalese(?) has a term for "mountain peak", English has no such word. The reply was: "of course it has a term for this: 'mountain peak'". As expert language users, we have no trouble synthesizing new concepts by using our language, which after all is far more than just a list of words, to frame our thoughts.

I once was sitting near an argument about whether an object was blue or purple, and they decided to drag me into the argument. Not wanting to take sides, I said "how about periwinkle?", and they both seemed satisfied with that answer.

> it'll naturally occur that each subdivision is referenced fewer times.

That's true but their relative frequency will not change. Ie. if your only words are warm and cold you may see each of those 6% of the time, but if your have red, orange, yellow, green, blue, and purple then you will see each one 2% of the time.

But in both cases all the other color related words in the language would be significantly less than 2%.

To test this I went to google and did searches for the basic 11 colors, and a couple of others, noting the number of hits found.

The big 11:

red 12,110,000,000

orange 7,040,000,000

yellow 6,370,000,000

green 13,850,000,000

blue 15,100,000,000

purple 10,160,000,000

black 13,390,000,000

white 12,330,000,000

gray 14,950,000,000

brown 10,160,000,000

pink 4,700,000,000

... and some test cases

maroon 152,000,000

teal 7,320,000,000

navy 870,000,000

cyan 110,000,000

I think the frequency explanation has trouble, since "teal" has more hits than "orange", "yellow", and (by a lot) "pink".

Google is not a corpus of text.

Usually you remove proper nouns when doing a word frequency analysis.

Pink, Maroon and Navy are all proper nouns -- Singer/songwriter, band (Maroon 5), and branch of the military.

Also, a Google search only gets you documents that have that word at least once, but removes duplicates. So maybe the page with red has it listed 7,000 times but you wouldn't know.

Trying using all the text in Project Gutenberg[0], removing the proper nouns, and then run the analysis again.

Linguists have done this for you already -- and the results were that your top list are in the top 300 and the bottom are not.

[0]https://www.gutenberg.org

Why not 325? Or 350? Or 400?

Because past 300 it very dependent on the specific corpus, so linguists have settled on 300 since just about any corpus gives the same top 300.

I don't really know how to explain it, but trust me as a native Russian speaker: in Russian, the felt difference between "light blue" (a more accurate translation might be "azure" or "sky blue", as that is considered the archetypical manifestation of it) and "dark blue" is the same as between "blue" and "green" in English. Now, as a native English speaker, do you feel that "blue" vs "green" is the same difference as "blue" vs "azure"? I'd wager that even if you can't quite verbalize it, you can feel it.

I suppose one of the clearest ways to demonstrate is to note that it's perfectly all right to describe something as a 'greenish blue' but you never hear someone talking about an 'azurish green'.

Mandatory XKCD link here

https://blog.xkcd.com/2010/05/03/color-survey-results/

This isn't just a comic - Randall Munroe ran a color naming survey on internet users. Native language was not collected. I assume native English speakers were over-represented but not universal.

Interestingly there was a slight gender difference in naming among respondents (see the second table). Is that perceptual or a cultural gender-specific difference in how often men or women talk about color?

> Is that perceptual or a cultural gender-specific difference

It's perceptual. Women have a higher cone to rod ratio, making them see colors easier while men see movement better. Also women are more likely to be tetracromats (four different cones instead of three).

There are some unproven evolutionary reasons that this may be the case -- men needed better movement detection to hunt while women needed better color vision to detect bad food when gathering.

Also colour blindness is sexist (1 in 12 men vs 1 in 200 women has colour vision deficiency).

There are some anecdotes that colourblind sharpshooters find it easier to see through camouflage as they're more sensitive to shapes (i.e. because for them a lot of things are "naturally camouflaged" they have more experience detecting shapes) but I'm not sure how factual they are.

I'd say the 1 in 12 vs 1 in 200 difference alone might account for a significant difference when looking at a uniform sample of both sexes.

Some cultures just stop with "warm" and "cold"

This is where the article (and maybe the underlying studies?) seems to go off the rails:

asked him or her to name each one’s “basic color term”—the simplest, broadest word that described its shade

If I (a native English speaker) were asked in those terms to describe the coloration of an object, I think I'd conclude that I would need to limit my answers to "warm" and "cold". These are certainly color concepts in English, and indeed, as a photography enthusiast I work with them all the time when I edit my photos in Lightroom. So I can't see why they've then broadened the terms out to the more specific red, orange, etc., as more general terms are available.

The article seems to make an overture toward the possibility of thinking in colors even outside of the hue, saturation, and value model that westerners at least seem to think in - which could be very interesting. But the question above also seems to preclude such a discovery. As I read it, "shade" seems to be synonymous with "hue", thus demanding that I follow our familiar model. But if a different word were substituted for "shade", like the generic term "color", I think I would have answered that the most generic words available in English are "light" and "dark". (That, of course, is basically the V in HSV, and not a different model, but still provides another set of possible English terms of broader foundational breadth than "red", "orange", etc.

The article is extremely verbose. The diagram on page 1 of this document sums up the idea:

http://www.ucl.ac.uk/medical-education/reprints/1997-Percept...

Reply on news.ycombinator.com