The Count That Snowballed

Everyone “knows” the Eskimos have fifty words for snow — or a hundred, or four hundred. The number is famous, repeated, and untraceable. Here is where it actually came from, and some of the figures it was later given; then the harder truth the debunkers miss: because these languages are polysynthetic, the real answer to “how many words?” isn’t a number at all.

In 1911 the anthropologist Franz Boas mentioned, in passing, that the Eskimo language has a few separate roots for snow where English uses one word plus modifiers. It was one example in a grammar handbook’s introduction — an illustration of how languages package ideas, no more a boast than noting that English has separate words for rain, fog, dew and ice. Then the example left home. Step through what happened to it.

I · The number that grew while the evidence shrank

Only Boas’s 4 rests on data. Whorf’s quote is real, but the five or more snow words it implies already had none. Everything taller is a number in search of a citation. The linguist Laura Martin traced this in 1986 and Geoffrey Pullum made it famous in 1991 as “The Great Eskimo Vocabulary Hoax”: once popular sources got hold of the example, numbers were “just made up as the writer thinks appropriate for the readership” (in 1984 alone Martin found nine, a hundred and two hundred; in 1988 the Times said “about four dozen”). The New York Times printed “100 types of snow” in a 1984 editorial and credited the figure to Whorf, who never gave it; no grammar lay behind it.

So far, so much like every other “the real version is actually…” legend. But here the smug correction — “actually they have no more words than English, it’s all a hoax” — overshoots in the other direction. To see why, you have to try to count.

II · Now try to count them yourself

That is the whole point, and it is not a dodge. In Inuit and Yup’ik languages, what looks like a “word” is built up from a root by stacking suffixes — so a single root spawns hundreds of fully grammatical word-forms, and a phrase English needs a sentence for can be one long word. Counting “words” is like asking how many numbers there are between 0 and 1. Boas’s real point was the modest, true one: every language carves the world finely where its speakers need the distinctions — English skiers and glaciologists have powder, hardpack, firn, névé, graupel, rime and sleet — and Inuit hunters, living on it, carve snow and ice finer still. That part was never the hoax. The hoax was the number.