Like most of us, I've heard many times about how many words people from snowy cultures have for snow. I've always found it implausible, but recently I found a beautiful account of this urban myth in Geoffrey K. Pullum's book The Great Eskimo Vocabulary Hoax (University of Chicago Press, 1991; and of course available from Amazon) -- I highly recommend this book. Since it also includes some relatively hard facts, and since I really enjoyed reading it, I thought it worth putting up this page about it.
Pullum cites a detailed article on the phenomenon, that I have not yet obtained: Laura Martin, 1986. "Eskimo Words for Snow: A case study in the genesis and decay of an anthropological example." American Anthropologist 88(2), pp. 418-423. She traces the myth to Franz Boas' Introduction to The Handbook of North American Indians (at least 15 volumes, sadly out of print and going for $75+ per volume used). In that introduction he says merely (as described by Pullum on p. 162):
just as English uses derived terms for a variety of forms of water (liquid, lake, river, brook, rain, dew, wave, foam) that might be formed by derivational morphology [suffixes and other such stuff] from a single root meaning 'water' in some other language, so Eskimo uses the apparently distinct roots aput 'snow on the ground', gana 'falling snow', piqsirpoq 'drifting snow', and qimuqsuq 'a snow drift'.
From there Martin and Pullum trace the steady growth of exaggeration, starting with Whorf's much quoted 1940 article that grows the word-count magically to 7, and eventually reaching sources claiming totals as high as 400. Just a few weeks ago I sat across a table at dinner from someone claiming to actually know "all 57 Eskimo words for snow" -- but I had neither a native speaker nor Pullum's article at hand.
As Pullum deftly points out, the traditional claim is "Eskimos have N words for snow" (for growing N) -- and every part of that claim is problematic:
There is no single language "Eskimo", just as there is no single language "Indian". And, like "Indian", "Eskimo" is not a very good name: it lumps together two major cultural groups, the Inuit and Aleut, and ignores major differences (including huge language variation) within each group.
How do we count "words"? As a computational linguist (for you co-nerds, technically that would be "corpus linguist", and my dissertation was on Hidden Markov Models for part-of-speech disambiguation in American English and Koine Greek), I am innately suspicious of any claim about enumerations of "words". What is a word, anyway? There is no single definition (see below).
What qualifies a word as being "for snow"? Surely it cannot mean the word must have exactly the same range of applications as English "snow", or Eskimo surely has no words for snow at all. Does Eskimo have a word for the grey mist in a poor TV picture? Just how wide or narrow do we draw the boundaries, and how do we ensure we're drawing them the same in the languages being compared?
Just deciding what we mean by "word" is subtle. For example:
Purveyors of the myth seem bound to include as many faintly snow-related phenomena as possible when deciding which Eskimo words count, but to require perfect synonymy to "snow" when deciding about English (thus leaving us but one "word for snow").
One example of this conveniently inconsistent definition is from a list of 20 ostensible Eskimo snow-words that came to Pullum: igluqsaq. This is a compound word meaning 'house-building material' (note the familiar 'iglu' at the beginning): it can mean plywood or brick just as well as snow. If we pulled the same trick with English, we would start counting words like "etiology" (a slip on snow can cause injury), "projectile" (you can throw snowballs), "food" (you can eat snow as long as it's white), and so on. Hmmm.
Pullum cites several sources on how many words certain Inuit dialects actually have for snow. The two main ones are:
The Dictionary of the West Greenlandic Eskimo Language (C. W. Schultz-Lorentzen, Copenhagan: Reitzels, 1927) gives just two words: qanik for snowflakes in the air, and aput for snow on the ground.
The Yup'ik Eskimo Dictionary (Steven A. Jacobson, Fairbanks: University of Alaska, 1984) has, according to Pullum's colleague Anthony Woodbury, about 24 if you're very generous. By "very generous", I mean including words for "stuff for sinking habitually into", "blizzard", "avalanche", and so on.
So 24 seems to be the outer limit that could be defended, at least for Yup'ik. Unless there are speakers somewhere who make a living by coining new snow-words and selling them.... No one seems to have checked on that possibility.
It's only fair to see how many snow-words we can find in English. I didn't even poke around the OED yet, but even if we skip the inflected forms (snows, snowed, snowing, snowy, snowness, snew, snewn,... well, we weren't going to count those anyway), there is still a veritable hail of terms:
berg, cornice, crevasse, floe, frost, glacier, hail, hardpack, hoarfrost, ice, iceball, icecap, iceberg, ice field, icicle, powder, rime, snow, slush, sleet, snowball, snowcap
Not to mention a blizzard of words for the parts of snow and for snow as a weather condition:
avalanche, blizzard, dusting, flurry, ice crystal, ice storm
And the list snowballs as we notice compound words related to snow (excluding snow-related objects like snowboards and snowshovels):
snowball, snowbank, snowcapped, snowdrift, snowfall, snowflake, snowlike, snowman, snowstorm
And let us not forget the storm of words that are spelled with a space in them: phrases that have (arguably, of course) become lexical items through frequent and distinctive use:
freezing rain, new-fallen snow, yellow snow, glare ice, purple wax snow (and a host of others skiers can cite)
I can't imagine I've listed anywhere near all the good candidates, but already that's 40. Which, by the way, is several more than the generous estimate for Inuit.
If we were as generous as some are for Eskimo, beyond those we'd add etiology, construction material, food, weapon, toy, floor, projectile, sculpture, refrigerator, obstruction (reaching double the total for Inuit), and probably dozens more. And we haven't even considered any synonyms for "snow job" (a quick look in a thesaurus reveals over 100), "snow" on your TV, and being "snowed" under.
It is the conclusions drawn from this bizarre claim that puzzle me most. One seemingly popular conclusion takes Whorf (of the "Sapir-Whorf hypothesis") farther than perhaps even Whorf would go, and suggests that such differences in vocabulary make cross-language communication fundamentally impossible, or even make a notion of any underlying reality (accessible or not) impossible. None of this of course follows. As Pullum again points out, anyone concerned about a particular subject has a proportionately detailed vocabulary for it. There is little surprising or interesting about this, and it doesn't seem to have much affect on communication, realism, or anything else beyond signalling that you are (or aren't) an expert on a precise topic:
I have precise words for things others would merely lump together and call "hypertext links".
In reading a book on knife sharpening, I just discovered formal terminology for bits of rock, carefully distinguished by diameters from 10 inches down to 0.00015 or so. They're also described here.
Wine tasters have another set of terms, far more detailed than I can use competently.
So if some language(s) did have many words for snow, it should be no more interesting than these other everyday cases. But if you think about it, people who live in ever-snowful lands may perhaps care no more about fine variations of snow, than we in warmer climes care about fine variations of grass or pavement: anything so constant disappears into the background and becomes less interesting.
A similar common idea is that Greek is far better equipped to express "love" than English. We poor benighted Anglophones have (alas!) but a single word, while those deeply philosophical Greeks of old were blessed with four. And of course, they always used each one in precisely the same way, and never let them overlap (oh yeah? Guess which term the Septuagint uses of Amnon's feelings for Tamar in 2 Samuel 13:1-4).
Again, we can find more words in English if we try to cover the same range as the 4 Greek words cover: adoration, affection, ardor, amourousness, attachment, caring, concern, cherishing, compassion, devotion, enamorement, fancy, favor, fondness, liking, love, lust, passion, tenderness -- to name but a few (19 to be precise).
Perhaps the myth arises from our need to count, to draw boundaries, to categorize; to pretend things are simpler and clearer than they are. Oh well; that's my small contribution to de-bunking a linguistic legend.
Back to home page of Steve DeRose or The Bible Technologies Group. or The Bible Technologies Group Working Groups. Or, contact me via email (fix the punctuation).