Geoffrey Nunberg
From Natural Language and Linguistic Theory, 13, 1

 The world has had a decade by now to disabuse itself of  the notion that the Eskimos have a multitude of words for snow. Those readers who missed Laura Martin's original expose in the American Anthropologist (Martin 1986) could hardly have missed seeing Geoff Pullum's lucubrations on the subject, which appeared first in these very pages (Pullum 1989), and which were reprinted in Lingua Franca and as the title piece of Pullum's collection of NLLT columns and quoted and discussed at length in Steve Pinker's best-selling book The Language Instinct. The iconoclasts have done everything to publicize the truth, that is, short of making the story a running gag on the Tonight Show.

But has the world paid any heed? Amazingly, it has. To check this, I did a search not long ago in a Dialog database of around sixty American newspapers for passages containing the words Eskimo(s), word(s), and snow within a few words of each other. It turned up 190 citations, all but one of them dealing with the fabled lexical promiscuity.1 A fair number of these were repetitions of syndicated stories, but this didn't seem to me a reason for throwing them out -- after all, if you started ruling out repetitions here, where would you stop? But while I can't give you any meaningful quantitative assurances, it's clear that the myth is showing signs of melting under the light of Martin's scholarship and the heat of Pullum's invective.  A number of articles explicitly mentioned one of the linguists, and quite a few others made it clear that the story was not to be credited. One writer described it as "a popular myth," another as "an old canard... to delight Time essayists"; a third apologized for repeating in an earlier story

The writer didn't say which of the three linguists involved was the source of his information; but then one linguistics professor had better be as good as another for purposes of settling arguments about Eskimo.

  Actually it shouldn't be surprising that journalists have picked up on the revisionist story; after all, it's as much of an item going out as coming in. (Everything is a story, as a magazine editor once remarked to me: there's isn't much in this world that isn't either a novelty or a trend.)  And while it's true that the large majority of writers still cite the factoid uncritically, for us linguists the story is irrevocably blown. We may be willing to stretch the truth now and again in the service of a didactic point, but this one seems to have turned into a professional point of honor. It's one thing to have to admit your ignorance of Schultz-Lorentzen's 1927 Dictionary of the West Greenlandic Eskimo Language, and quite another to have to say that you never bother to look at "Topic... Comment." The trouble is that unlike the journalists, who are happy to go either way with this, we came off a lot better when we could repeat the story than we do having to deny it. In the first case we offered ourselves as the interpreters of alien modes of thought, in the second only as pedantic party-poops. (Dr Johnson once defined pedantry as "the unseasonable ostentation of learning," and never was learning more unseasonably ostended than here).

 Oh well, you say, the languages of the world are teeming with specialized vocabularies. But satisfactory substitutes are hard to come by. One article I ran across mentions 160 Bedouin words for "camel" and 500 Italian words for types of pasta. But whatever the truth of these figures, neither example does much to
excite my sense of wonder: they seem to involve not alien ways of seeing the familiar world but unremarkable ways of seeing worlds that are rather different from the one I encounter in my rambles around my own San Francisco neighborhood (particularly as regards the camels). And even when some other language uses several words to describe some more familiar thing that we English-speakers describe with one, the differences usually don't imply any interesting conclusions about perception. The French, for example, lack a current general word for "berry," and can only talk about blueberries, raspberries, strawberries, and so on.2 Yet no one I've mentioned the example to seems to find it remarkable that even lacking such a word, the French have managed to come up with the idea of mixing all those fruits together in a compote. In moments of frustration I have made up examples out of whole cloth: I once claimed that the Anglo-Saxons had fourteen different words for "disembowel," figuring that Martin, Pullum, et al. would be so up to their elbows in the "snow" business that they would never get around to calling me on the figure. This one, though, excited all too much curiosity, and when people pressed me for details I found myself entangled in a web of gory invention.

 It leads you to wonder, why is the Eskimo story so apt?, and for that matter, what is it apt for? One thing it has going for it, certainly, is Eskimos. As Martin and Pullum note, it seems of a part with all the other exotica ("polysynthetic perversity" as Pullum puts it) that we attribute to Eskimos -- the people who rub noses, cast their old people out in the cold, eat raw blubber, and all the rest of it. Pullum describes these stories "racist" and "xenophobic," but this seems a little off the mark to me, at least in the sense we would apply those words, say, to the stereotype that an American anglo might invoke to justify a disparaging attitude towards Mexican immigrants. I don't doubt that non-Eskimos in places like Alaska or Greenland have such stereotypes about Eskimos (though I doubt whether nose-rubbing plays an important part in them). But for people who live in temperate climes and have never seen an Eskimo outside of a fifth-grade geography text the tales have more of the flavor of the precolonial accounts of fabulous races that you find in travelers' tales running from Herodotus down to the eighteenth century -- Othello's "anthropophagi and men whose heads/Do grow beneath their shoulders."  This isn't to say that the stories are politically innocent, but like all such tales they are meant to be read allegorically rather than literally. (Though you never know. Tony Woodbury tells me that some Eskimos really do rub noses, and that in fact there is a Proto-Inuit base *kunik-, glossed as 'sniff or touch noses' which also means 'kiss' in all the attested daughter dialects.)

 But the Eskimo snow words story has another thing going for it, which is  snow itself. In the first place, snow is stuff that most of the speakers of European languages know as a temporary phenomenon, whereas for the Eskimos (or so we're told) it is a permanent condition of existence -- of course they would regard it differently from the way we do. And when you think about it, there aren't many other things that have this property -- what material experience do we speakers of Standard Average European languages have in common , after all, except for our climate, and what more perceptually salient index of climate is there than snow? What's still more important, though, is the way snow alters the familiar world. It whites out differences; it effaces the forms and colors of objects and the boundaries between them. Ground and sky, spruces and junipers, tow trucks and bus shelters -- everything seems to blend, as you come out into that first winter snowstorm, into a featureless and uniform landscape. What better figure could there be for the way language works when it abstracts away from particulars?

 This can be the point of departure for various morals, of course. The absence in Eskimo of a general word for snow can be offered as evidence for the inability of the primitive mind to reason abstractly, as Mario Pei did when he classed the Eskimo example among "those very numerous cases where a less developed language makes fine distinctions essential to its speakers and their activities but not to us" (Pei 1965, p. 119). Whorf too concluded that the Eskimos were incapable of conceptualizing the general category -- "To an Eskimo," he said with unexplained assurance (Whorf 1940; in Carroll 1956, 216), "this all inclusive word [snow] would be almost unthinkable" -- though he was interested less in deprecating the Eskimo mentality than in exoticizing it. ("How utterly unlike our way of thinking!," he said elsewhere of the Apache equivalent of "It is a dripping spring" that he rendered as "whiteness moves downward" -- a gloss that as Pinker (1994, p. 61) notes contributes in no small part to the impression of utter unlikeness that Whorf was eager to establish.)

 As it happens, though, most of the people who picked up on the story are really not interested in making a point about the Eskimos or the primitive mind as such (out of those 189 newspaper citations of the myth, only a handful are from stories about Eskimos or related topics, and only two of those cite it uncritically). And far from taking the example as evidence of the incommensurability of alien modes of thought, they tend to assume that the putative multiplicity of Eskimo words for snow is a perfectly reasonable state of affairs given the attention that Eskimos pay to the stuff, which should serve us all as a lesson in the dangers of generality. It may be, as Jespersen said, that language "is only attending to its proper business when it comprehends a multitude of like things under the same appellation...." (Jespersen 1964 (1946), p. 79) But people seem to have a persistent suspicion that many of the "likenesses" that we perceive among things that bear a single appellation are illusory, and may owe something to the uniform covering that language lays over them. It occurs to us that with experience we might see the world more differentiated than we do, the way a snowy landscape resolves itself when we stand out in it long enough to acquire a mind of winter.

 In one way or another, that is always the point of the story that people tell about the Eskimo words for snow:

In other articles, writers contrast the number of Eskimo words for snow with the poverty of the English vocabulary for talking about the varieties of heat, fog, fatigue, interoperability, rape, road surfaces, silence, stubborn children, good women, laughter, taste sensations, malnutrition, and, not surprisingly, snow itself ("an icy freezing blast of sleetstorm that would force the Eskimos to think up a 33rd word for 'snow,'" was one of my favorite entries, from the Philadelphia Inquirer).  Or, in what amounts to pretty much the same thing, writers use the story in the course of talking about the richness of some specialized vocabulary -- college students' words for throwing up, corporate euphemisms for firing people, Ben and Jerry's ice-cream flavors, the Gambino crime family's words for doing away with people, and, again not surprisingly, snow words, this time in the lexicons of skiers and resort operators.3

 It is a convenient little fable for making a point about English, and even the discovery that it isn't true turns out to be no real impediment to its usefulness:

In retelling the story, though, these writers consistently leave out the feature that was most important to Boas, Whorf, and the other linguists and anthropologists who have cited it: the absence in Eskimo of a single general word for snow. That would spoil the analogy, because the English examples usually do involve a general term or concept --"racism," "ennui," "silence," or whatever -- and go on to say that it doesn't do justice to the complexities of experience. The problem isn't that we have general words for racism or silence, but that we have no more specific hyponyms than these. Of course sometimes a highly differentiated vocabulary betrays a familiarity with a variety of experience that we would just as soon not be acquainted with in intimate detail -- drugs, ways of doing away with people, and other practices that a civilized people ought not to dignify with lexical recognition. (As James put it in The Golden Bowl: "'When I speak worse, you see, I speak French,' he said, intimating that there were discriminations, doubtless of the invidious kind, for which that language was the most apt.") But that's pretty much the most that people have to say on behalf of generality, that it occasionally serves to keeps us in ignorance of unsavory realities.4

 Saussure said that in language there are only differences, and the public adds: not nearly enough of them. There are hundreds of prescriptive rules about the importance of preserving valuable distinctions (between imply and infer, comprise and compose, shall and will, blatant and flagrant, and so on) but you are hard put to find strictures in defense of valuable conflations. And of course it's widely maintained that the best measure of the superiority of a language is the number of distinctions a language allows its speakers to make, as measured by the size of its vocabulary. Indeed, while we're on the subject of vocabulary hoaxes, I should mention a recent book by language columnist Richard Lederer that puts the English vocabulary at "as much as two million" against estimates of 185,000 for German, 130,000 for Russian, and under 100,000 for the poor French  (Lederer 1991, 23-27).5 He explains the differential, following conventional lore, by pointing to the early hospitality of English to French and Latin borrowings, which has left the language rich in synonym sets like ask, question, and interrogate --  not to mention all those words we welcomed later from Welsh, Bengali, Ojibwa, Tagalog, and so on. ("Books like Roget's Thesaurus are foreign to speakers of most foreign languages," Lederer adds; "given the scope and structure of their vocabularies, they have little need of them.") And for all that people like Pei have maintained that  "less developed" languages are characterized by the presence of numerous "fine distinctions," they seem to have had no affect on the popular wisdom, which is as unshaken now as in Bloomfield's time in its conviction that the mark of the primitive is a vocabulary of a few hundred words.

 Even we linguists have been known to appeal to these presuppositions about the value of distinctions. We defend the expressive adequacy of a nonstandard variety like African American Vernacular English, for example, by observing that it provides for a semantic distinction between two forms of the copula that is unavailable in the standard variety. But granted that the conclusion is unimpeachable, why should this form of argument be so telling? What if AAVE boasted a French- or German-type aspectual system that offered no distinction between progressive and simple present; would that count as an argument for its expressive poverty?

  Why this emphasis on distinctions? Those metaphors of lexical "poverty" and "richness" suggest some of the answer. Our prescriptive dogmas were shaped, after all, by Enlightenment aesthetic theories that made "discrimination" a synonym of "taste," and by the way, a mark of worldly accomplishment. As the rhetorician George Campbell put it in 1776:

It is hard to imagine a more un-Whorfian point of view than this one, where the vocabulary is merely an inventory of independently accumulated experience (or for accumulation tout court: how fortunate for English to have had the opportunity to take in all those words from Bengali, Ojibwa, Tagalog, and the rest).6 But I think it comes much closer to explaining the popular understanding of significance  of the Eskimo snow-words story than the lessons that Whorf tried to draw from it.

 But there's no denying that there's also a psychological barrier at work here. It isn't that the Eskimos would find our all-encompassing word for snow "unthinkable," as Whorf claimed -- some shrewd empiricist among them has doubtless noticed the reliable correlation between flurries of qanik and  mounting piles of aput at the door -- but  you suspect that they might find it hard to see what the fuss was about. It's like the way we take the word precipitation when we first learn about it. We can see why meteorologists would want to have a word to express the concept, but we aren't tempted ourselves to climb up a level in the hierarchy of weather kinds, kicking over the ontological ladder behind us. Precipitation has its uses, but it doesn't tell us how to dress when we leave home in the morning .

  All of this leaves people with a certain blindness to the virtues of generality and the perils of overdifferentiation. It has occurred to me, for example, that English could do with a superordinate term that enabled us to refer to the territory divided between force and violence without having to consider the official status of the actors. But there's no popular fable I can invoke here that does the work of the Eskimo example in reverse. Examples aren't hard to come by, of course (is English the only language on earth that has no general term covering grapes and raisins?), the only problem being that most of these have the same failing that precipitation does: people see the generalization but not any its larger significance.

 Still, sometimes you can make the point obliquely. Take, for example, the limited vocabulary that Italian makes available for the visual aesthetic. Where English has twenty or more distinct adjectives that apply wholly to physical beauty --  beautiful, handsome, cute, pretty, gorgeous, lovely, comely, attractive, fair, ravishing, stunning, good looking, fly -- colloquial Italian has little more than the lone word bello aided by a handful of diminutive suffixes, and even that word is not restricted to physical beauty (it means "nice" when applied to a house and "good" when applied to a book or film).7 According to the principles we invoke to evaluate lexical adequacy, Italian is a pretty sad case, bereft of the capacity to express the fine gradations and nuances of beauty that are available to us anglophones. And Italians in consequence are very likely deficient in the perception of beauty: the visual aesthetic must seem to them as monochromatic and undifferentiated as a snowy landscape does to us. But then you look at their sports jackets and at our sports jackets, and you sense that we may be missing something.


Campbell, George: 1963 (1776), The Philosophy of Rhetoric, ed. Lloyd F. Bitzer, Southern Illinois University Press., Carbondale.

Carroll, John B., ed.: 1956, Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf, MIT Press, Cambridge, Mass.

Jespersen, Otto: 1964 (1946), Mankind, Nation and Individual from a Linguistic Point of View, , Indiana University Press, Bloomington, Ind.

Lederer, Richard: 1991, The Miracle of Language, Simon and Schuster, New York.

Martin, Laura: 1986,  '"Eskimo words for snow": a case study in the genesis and decay of an anthropological example,' American Anthropologist  89, 2 (June), 443-444.

Pei, Mario: 1965, The Story of Language, Lippincott, Philadelphia.

Pinker, Steven: 1994, The Language Instinct: The New Science of Language and Mind, London, Allen Lane.

Pullum, Geoffrey K.: 1989,  'The great Eskimo vocabulary hoax,' Natural Language and Linguistic Theory  7, 2, 275-281.

Whorf, Benjamin Lee: 1940, 'Science and linguistics,' Technology Review (MIT) 42, 6 (April), 229-231, 247-248. REprinted in Carroll, ed., 207-219.


Xerox Palo Alto Research Center
3333 Coyote Hill Road
Palo Alto, CA 94304


1 The one exception was from a story in the Anchorage Daily News about a movie being shot on location in Alaska in which a man working as an "Eskimo stunt double" is told to "pull the snow hook at the word 'action'".

2 French does have an old-fashioned word baie that refers to berries in the broader sense of the word (i.e., as the fleshy fruit of a plant), as well as an expression fruits rouges that is sometimes used to refer to berries on the plate.

3 The trope is also used by way of introducing the lexical extravagance of some other language -- for example the number of Hebrew words for settlements, Greek words for love, or Russian words for mud.

4 Occasionally, it's true, people invoke the trope in the course of complaining about euphemisms or needless synonyms:

This gets the point of the story wrong, of course, since it implies that the Eskimo words are all synonyms for the same notion. I suppose the story could eventually mutate to the point of becoming no more than a cautionary tale about lexical superfluity, though it's hard to square with the rest of the Eskimo fable "the Eskimos, those masters of euphemism and indirection."

5 This expressive richness is one of the "essential reasons" for the success of English as a world language, Lederer says, along with its simple grammar and its orthographic "compactness," though to be sure the economic and political growth first of England and then of the United States "has, of course, contributed to [its] phenonenal spread."

6  Campbell's picture of the relation of the accumulation of vocabulary and material goods recalls the passage in Gulliver's Travels (Book III, Chapter 5) that describes the various projects of the Grand Acadamy of Lagado:

7 To be  honest, Italian does have a few other adjectives (like attraente, "attractive" and leggiadro, "pretty, graceful") that are pretty much restricted to the domain of physical beauty, and a number of others which mean things like "dear," "pleasant," "wonderful," and "magnificent," and which like their English equivalents can be used to convey appreciation of beauty in a more general way. It's a little misleading, too, to suggest that the sense of bello survives unaltered in all its derivatives. The fact remains though that Italian has fewer evaluative adjectives than English that apply only to physical appearance, and inasmuch as fables tend to sink under the weight of footnotes, we might just let it go at that.