The Eskimo Snow Vocabulary Debate: Fallacies and Confusions

I. Introduction

The Eskimo Snow Vocabulary (ESV) debate concerns the numbers of words Eskimo languages have for snow and ice in their various forms and situations, as compared with other languages. The debate was set off a decade ago by an essay, "The Great Eskimo Vocabulary Hoax," by Geoffrey K. Pullum, professor of linguistics at the University of California Santa Cruz (Pullum 1989, 1990, 1991). Pullum there ridiculed the idea that the Eskimos had significantly more words for snow than did English, for example. He was motivated to do so, he explained, partly by a wish to correct a specific popular misconception, but much more by a wish to use this canard as a cautionary example of human gullibility, shoddy scholarship, and even of latent racism.

His essay attracted a good deal of attention, and has even, according to his friend and ally Geoffrey Nunberg, gone a long way toward correcting the specific error in question, if not the underlying faults in human nature or society that Pullum tells us are his real targets. After reviewing the history of Pullum's attempt at straightening us all out on the ESV question, he says, "But has the world paid any heed? Amazingly, it has" (Nunberg 1997), and reports that many stories in the press and periodicals have picked up Pullum's thesis and joined him in laughing at or scolding the sillies or miscreants who still languish in a state of error. It's not clear whether these journalistic sallies really represent a paying of heed by the world, as opposed to the momentary exploitation of a very minor intellectual scandal by journalists always desperate for something to talk about, but Pullum's piece does seem to have had an unusually successful career as such things go. It is taken as authoritative, for example, by Steven Pinker (Pinker 1994), who says, "no discussion of language and thought would be complete without the Great Eskimo Vocabulary Hoax," and goes on to summarize, with glee, Pullum's supposed final settling of the matter.

To the best of my knowledge, my own essay on the subject (Halpern 2000) remains the only attempt to critically analyze Pullum's thesis. Even that essay, though, as its name indicates, did not take that thesis as its main target, but simply used what I regard as Pullum's failings to illustrate and introduce my own concerns about linguistics and usage, just as Pullum had used the ESV issue to lead into the larger questions of concern to him. And both he and I have been at some pains to emphasize that the ESV, as such, is of only minor and passing interest to us, compared with the larger social and intellectual issues that we see involved. But while all parties seem to agree on the relative unimportance of the purely linguistic ESV questions, those questions keep coming up, and keep generating heat. This note is an attempt to clear away some confusions and fallacies--not all Pullum's--that continue to impede the settling of this troublesome if incidental question. I have not forced myself to stay strictly within the linguistic arena while discussing the ESV--I'm not sure it's possible to say much about it without glancing at broader issues of scholarly practice and social outlook--but I have tried to keep myself from roaming beyond the purely linguistic any further than is necessary to present a coherent argument.

II. Native and Metic Vocabularies

Some claim that the Eskimos have words for many more varieties and conditions of snow and ice than do the languages of the temperate zone; others say they have no more. (Note that "more words for snow and ice" does not mean mere synonyms for snow and ice, as a few have supposed.) The evidence usually offered consists of lexical counts, with each side in the debate claiming authority for its own list of such words gathered from native informants, or from dictionaries compiled earlier from such interviews. But besides all the difficulties regularly associated with the compilation of word lists for a language that until recently was purely a spoken language, and whose present-day written form is largely the work of non-natives, there is in this case a special problem: these lexical counts seldom if ever distinguish between words in common use and those never used by, or even known to, ordinary users of the language in question. This failure vitiates virtually all attempts made so far to settle the matter.

Those who reject the notion that Eskimos have significantly richer vocabularies, for example, often compile lists of English snow-related words in order to show that there are roughly as many such words in English as in Yup'ik or Iñupiaq; what they fail to note is that most of those English words are far from being in common use among English-speaking people. Most English speakers live their entire lives without once using such words as rime and glare (in their snow- or ice-related senses); these words are in the English language, for the majority of speakers, only in the sense that they're in English-language dictionaries. They are not, for most speakers of English, even in their passive vocabularies--most have never seen or heard such words used, and could not define them. I think that a new term is required for words that are part of the language only in so tenuous a sense, and I propose metic for that role--a term used to describe resident aliens in ancient Athens, who were allowed to live in the city, but severely restricted in their legal rights, and sharply distinguished from Athenian citizens1. It is particularly important to make such a distinction when considering English, because that language is so acquisitive, so prehensile, so ready to assimilate words from any source--the metic contributions to English from the languages of the Indian sub-continent alone have been collected in a book that runs to just short of 1,000 pages (Yule 1886)--that otherwise, we'll soon have to conclude that English subsumes every known language, including the Eskimo languages. That would settle the debate about whether Eskimo languages have more snow-related words than English, but not in a useful way.

By contrast, the many words that appear on Eskimo-language lists of snow and ice terms are very much native words, and in common use among all speakers of the language in question. The Public Information Officer for the North Slope Borough, Barrow, Alaska, has said, "…the Inupiats have more than 30 words for snow, and more than 70 words for ice. In the Arctic, the specific conditions of snow and ice are critical to hunting and survival …" (Patkotak 1994). And again, "Ice is very important here because of whaling and hunting done on the pack ice. Each word denotes ice in a very specific place or condition. Use the wrong word when you go hunting on the ice and you are apt never to be seen again" (Patkotak 1996). If we exclude the metics, and compare a list of just those snow-related words that are in common use among speakers of English with corresponding lists from Eskimo languages, it is obvious that the Eskimo snow vocabulary is many times greater than the English (and, presumably, other temperate-zone languages), just as one would expect.

II. What is a Word? Does it Matter?

The suggestion has been made that the very term word is inapplicable in discussing a polysynthetic language like the various Eskimo languages; these, it is claimed, are largely made up of free-floating segments that can be combined freely at need, and are thus quite unlike English, which for the most part is a corpus of fixed terms that have to be used as given, or not at all. This claim is a red herring. Whether the snow-related terms of the Eskimo languages are words in the same sense as the words of English is not to the point; what matters is that those Eskimo-language features--call them fixed locutions or fixed constructs, if you like, to avoid the controversial word word--are recognized and commonly used features of their languages. What the debate is about is whether there are significantly more fixed constructs or fixed locutions for snow-related phenomena in the Eskimo languages than there are in the languages of the temperate and equatorial zones.

(When the debate gets down to specifics, the comparison is usually between one of the Eskimo languages and English, but of course this is far from satisfactory: first, the Eskimo languages should be compared to all non-Eskimo languages to eliminate the possibility that, for example, Swahili might contain just as many snow-related terms; second, if just one language is to be chosen to represent the non-Eskimo languages, English is a bad choice, because of that imperialistic tendency mentioned earlier.)

Of course any urgent warning that the Eskimo languages can convey could also be conveyed in English--but it would take in general many more English words than Eskimo-language constructs; it would take the English-language speaker much longer to frame his utterance, since he would have to improvise; and he could not be certain that his improvised utterance would be understood in time--or ever--by his audience. In contrast, the Eskimo wishing to convey the same warning would be able to reach for a shorter, ready-made expression that was immediately comprehensible to his countrymen. And the Eskimos' need for such a tool is readily understandable, I think; it's rather more important for an Eskimo to be able to say quickly "That ice you're standing on is the kind that splits suddenly when the tide turns, as it's doing right now" than it is for the typical Alabaman.

As has often been noted, when excited or in trouble, we fall back on clichés--and when it comes to trouble on ice, the Eskimos have many clichés, and we do not.

III. Need a Snow-Related Term be Entirely and Uniquely Snow-Related to Count?

Another frequently heard claim is that only those words or phrases of which all parts are uniquely snow-related are to count when we're enumerating snow-related terms. According to this notion, powder as a designator of a type of snow is not to count, since it's a term originating outside the snow context, and used only figuratively of snow. That this proposed rule is wholly arbitrary and invalid may be seen by trying to apply it to words on other topics: would we deny that strawberry roan, for example, is a horse-breeding term because strawberry originates, and is used, in other contexts? Isn't donkey engine a mechanical term, even though a donkey is an animal, not a machine?

What matters for this kind of investigation is the number of distinctions the linguistic community under examination makes, within the subject area, for whose expression they have standard, recognized constructs--constructs, that is, that can be used without improvisation, and with assurance that any member of that community will understand them clearly and immediately. For these purposes there is no reason to distinguish between simple words and more elaborate verbal constructs, or between those constructs of which every part is peculiar to its subject area and those that draw on the vocabulary of other subject areas for analogies and metaphors.

IV. Does a Rich Snow Vocabulary Subject Eskimos to Sapir-Whorf Servitude?

Yet another red herring is the claim that one's position on the Eskimo snow-vocabulary question is in effect one's position on the Sapir-Whorf hypothesis (see "Sapir-Whorf" in the Reference list) about the role that language has in shaping and constraining its speakers' worldview. But this claim simply has no merit; one can hold any imaginable position on the Sapir-Whorf hypothesis, or even never have heard of that hypothesis, regardless of the number of snow-related constructs one finds in Eskimo speakers' live vocabularies. The one point of interest offered by this particular red herring is that it may explain the heat and rancor that has attended many debates on the snow-vocabulary question. If one imagines that the attribution to Eskimos of a large snow-related vocabulary necessarily, or even probably, entails supposing that Eskimos are limited to a very narrow, primitive worldview, and hence may be treated as our inferiors, even as less than human, than one might well strongly oppose such an attribution, and even fling wild charges of racism at those who find that Eskimos do have such a vocabulary. But why one should imagine any such thing is a mystery. In the minds of most observers, it is perfectly natural, and fully human, to possess and use a large vocabulary for the most prominent features in one's daily life--so much so that we would be likely to doubt the humanity of any group that did not do so.

V. Should Eskimos Be Called 'Eskimos'?

A point that is raised often enough in debates about the ESV to need putting to rest is that it is insulting to indigenous Arctic and Sub-Arctic peoples to call them Eskimos. This notion seems to rest--insofar as it can be said to rest on any clearly expressed reasons--on the facts or supposed facts that, first, the peoples concerned do not call themselves by that name, and second, the name is a Cree Indian word meaning 'eater of raw meat,' or 'eater of fish,' or even 'eater of rotten fish,' and hence an insult.

The response to the first point is that while the Eskimos do not, in their own language, call themselves by that name, that is because Eskimo is an English word, and they are speaking a different language. The Eskimos, more tolerant than their self-appointed champions, not only do not require English-speakers to use Eskimo names, but often call themselves Eskimos when speaking English. The principle that one must call each people by the name they call themselves in their own language is in effect the claim that there must be no English-language name for any foreign group. The right principle is that stated by C. S. Lewis when rebuked for using 'Scotch' to describe those Britons who live north of the Tweed; he simply pointed out that he was talking English, not Scots.

The response to the second is that no insult is intended, nor felt by the Eskimos, when they are called Eskimo (a term whose origin is unclear, but which there is no reason to regard as derogatory--see the discussion by Damas in the Introduction to the volume cited in the reference list under Woodbury). And how, one wonders, do the politically correct explain their rejection of a term coined by the Cree, another indigenous people--is that not insulting to the Cree? And is it not insulting to the Eskimos to assume that a bunch of Cheechakos can detect insult where the Eskimos themselves cannot, and to assume that the Eskimos must be protected as if they were children or mentally deficient? Hard is the path of the politically correct--and may it remain so.



Halpern, Mark (2000), “Why Linguists Are Not to Be Trusted on Language Usage,” The Vocabula Review , ( ) (Written in 1996, partially published in the Atlantic Monthly, March 1997, pp. 19-22.)

Nunberg, Geoffrey (1997), “Snowblind,” ( ) (also in Natural Language and Linguistic Theory 13, 1.)

Patkotak, Elise Sereni (1994), letter to the editor, New York (June 13, 1994), page 8. (reproduced in Halpern 2000)

----------------------- (1996) , private communication, March 26, 1996.

Pinker, Steven (1994), The Language Instinct . New York: Morrow.

Pullum, Geoffrey (1989), “The Great Eskimo Vocabulary Hoax,” Natural Language and Linguistic Theory 7 , 2; pp. 275-281.

------------------ (1990), “The Great Eskimo Vocabulary Hoax,” Lingua Franca , pp. 28-29. A two-page summary of the essay, in the premier issue of the periodical.

------------------ (1991), The Great Eskimo Vocabulary Hoax . U. of Chicago Press. Contains the fullest and most authoritative version of the essay.

Sapir-Whorf Hypothesis : go to CCMS - Communication studies, cultural studies, media studies infobase by Mick Underwood

Woodbury, Anthony C. (1984), “Eskimo and Aleut Languages,” in David Damas (ed.), Handbook of North American Indians V: Arctic . Washington, D.C., Smithsonian Institution.

Yule, Henry & A. C. Burnell (1886), Hobson-Jobson: The Anglo-Indian Dictionary . London, 1886; reprinted Wordsworth Editions Ltd, 1996.

Article Footnotes

1I owe the term to D. W. Brogan, the best foreign observer of America since Tocqueville. I had thought that it was already in use in the form Métis, meaning one of mixed Eskimo-European descent, and so stated in earlier versions of this paper, until I was disabused of this false notion by Mr P. D. Petersen, who kindly explained that metic is from the Greek metaoikos, and Métis from Latin misticius, an altogether different word.