Inuit Snow Terms: How Many and What Does It Mean?
The claim that Eskimo languages have numerous words for "snow" has often been repeated and has become familiar to the general public in addition to linguists and anthropologists. The point to be made seems to be that "Eskimo" has some indeterminate number of words -- and the numbers given vary in different sources -- for a substance that is described in English and most non-Eskimo languages with a much smaller number of words. The existence of snow terms in Eskimo is most often used to show the relationship between the vocabulary of a language and the physical environment in which that language is used. For some reason it is always the Eskimo example that is brought out to illustrate this situation and not the fact that painters may use a wide array of color terms and carpenters know a lot of words pertaining to nails and other hardware. The Eskimo example has entered the realm of popular mythology, having turned into a scholarly equivalent of the urban legend about the poodle in the microwave: everyone is familiar with the story but the exact details are a little sketchy.
Two publications have refocused attention on the snow example. Linguist Geoffrey Pullum in 1991 published The Great Eskimo Vocabulary Hoax but first, in 1986, Laura Martin, Professor of Anthropology at Cleveland State University, published a report in the American Anthropologist titled "Eskimo Words for Snow: a Case Study in the Genesis and Decay of an Anthropological Example." This interesting article follows the example from the original claim made by Boas in the introduction to the Handbook of American Indian Languages, published in 1911, through its many mutations and transmogrifications as it has been repeated and often amplified in articles and books and all without reference to primary sources of information. The numerous published dictionaries of Eskimo languages were not consulted, and neither were linguists or Inuit. A brief summation of the history of the snow example based on Martin's article follows, showing how the example progressed and took on a life of its own, divorced from any empirical data to support it.
When Boas in 1911 first presented his Eskimo snow terms, it was not in the section of the introduction called "Influence of Environment on Language" as one might suppose but rather in a less enlightening section called "Limitation on the Number of Phonetic Groups Expressing Ideas." The point of the discussion is to show that languages classify things very differently, "that the groups of ideas expressed by specific phonetic groups show very material differences in different languages, and do not conform by any means to the same principles of classification." Boas gives the example of English, where water in various states is denoted by independent, unrelated words, such as lake, river, brook, rain, dew, etc., although another language might conceivably express them by means of derivations from one term. Boas next gives an example comparing snow terms in "Eskimo" to English with its water terms. He gives four "words," aput 'snow on the ground,' qana 'falling snow,' piqsirpoq 'drifting snow,' and qimuqsuq 'snowdrift.' These forms appear to come from a variety of Eastern Canadian Inuktitut, but the source is not given. Boas goes on to say that the same language has a variety of terms for seals.
This passing reference to Eskimo and the fairly modest claim it is intended to support are rarely remembered when the snow example is brought up. It is, however, closely associated with Benjamin Whorf, who in a 1940 article used the example of Eskimo to contrast with English:
We [English speakers] have the same word for falling snow, snow on the ground, snow hard packed like ice, slushy snow, wind-driven snow -- whatever the situation may be. To an Eskimo, this all-inclusive word would be almost unthinkable.... (Whorf 1940; in Carroll 1956, 216)
Whorf does not cite Boas and does not give specific data, but he clearly suggests that Eskimo languages have five or more snow words. The example is taken up again in two textbooks published in the late 1950s, The Silent Language by Edward Hall and Words and Things by Roger Brown, and the example is mishandled again, with no serious attention paid to the linguistic data. Carol Eastman continues the tradition of carelessness in her book Aspects of Language and Culture, basing a discussion of the Sapir-Whorf hypothesis on the snow example, quoting Brown saying that Eskimo has three words for snow and also asserting that "Eskimo languages have many words for snow."
The snow example has also found its way into the press. A number of articles in Time magazine, the New York Times, and elsewhere refer to the quantity of words for snow in Eskimo languages and often pull numbers out of thin air, nine in one case, one hundred in another. As a result of this wide discussion, the snow example is widely known and referred to. Amazingly, it has come to be accepted as a commonplace of linguistics and anthropology. It has achieved this status without the benefit of reference to linguistic facts but based on the assumption that these facts must be found someplace, all despite the existence of published sources of Eskimo lexicon that have been accessible for decades.
Given how widespread the snow example has become, it is surprising that it escaped scrutiny for so long. Not until Pullum published his humorous essay "The Great Eskimo Vocabulary Hoax" in the book of the same name was there anything in print presenting the view of a specialist on Eskimo languages; Pullum reports information from linguist Anthony Woodbury in an appendix. With the exception of Pullum's appendix, Eskimo linguists have ignored this subject, probably seeing it as unworthy of any serious attention. It would be something like an anthropologist writing about whether Eskimos rub noses. But because the example is so widely known and because it has been written about by both linguists and anthropologists, it seems worthwhile to add a perspective informed by Eskimo linguistics.
Martin's article does a fine job of tracking the snow example through its many incarnations, showing how it has been misrepresented and misused. She points out that Eskimo is not a single language, as it seems to be presented by those who have discussed the snow example. She further stresses that the polysynthetic morphology of Eskimo languages renders a discussion of "words" as such almost pointless, since the number of words in these languages is practically infinite due to highly productive patterns of suffixation. It is therefore necessary to establish what would be considered a word in the languages in question, and apparently it must also be decided what is to be considered a "snow term," since Martin and some others see this as part of the issue.
It is correct that Eskimo words are theoretically infinite in number, so that what is phonologically a single word is most often not a lexicalization but a longer combination of elements generated as part of the speech process and not found anywhere in the mental lexicon, much like sentences in more analytic languages like English. These strings of morphemes combined by productive processes can therefore not be considered lexemes; Eskimo dictionaries list word stems that are then subject to inflection or derivation through suffixation. Most noun stems, e.g. aŋun 'man' or aġnaq 'woman,' are also full "words" in the sense that they may stand alone without any affix. Verb stems on the other hand cannot stand alone and require at least an inflection, although it is usually convenient to cite them in stem form. So katak- 'fall' must be inflected to be used in speech, e.g. kataktuq 'it fell' or katakkaa 's/he dropped it.' It can also take adverbial-type suffixes, e.g. katagniaġaa 's/he will drop it.'
Treatment of actual data is ancillary to Martin's primary purpose of demonstrating the careless handling which the snow example has received, and almost as an afterthought, data are given in a footnote which explains what Schultz-Lorentzen's Dictionary of the West Greenlandic Eskimo Language gives in the way of snow terms. She reports that "There seems no reason to posit more than two distinct roots (her italics) that can properly be said to refer to snow itself (and not for example, to drifts, ice, storms, or moisture) in any Eskimo language. In West Greenlandic, these roots are qanik'snow in the air; snowflake' and aput 'snow (on the ground). Other varieties have cognate forms. Thus, Eskimo has about as much differentiation as English does for 'snow' at the monolexemic level: snow and flake."
In a small flaw in her otherwise fine article, Martin does not handle the data particularly well, taking snow terms in West Greenlandic to be representative of those found in all Eskimo languages. The Comparative Eskimo Dictionary lists three reconstructed Proto-Eskimo noun stems which would fit her criterion of referring to "snow itself" and not to other related atmospheric phenomena: *qaniɣ 'falling snow', *aniɣu 'fallen snow', and *apun 'snow on the ground' are the three basic roots found in all Eskimo languages and dialects, except, unfortunately, West Greenlandic, which lacks *aniɣu, whose Inuit reflex is aniu. Another problem with Martin's equation of the two West Greenlandic stems meaning 'snow' with two English words is the inclusion of 'flake' as a basic English snow term. While 'flake' often refers to snow, it is generally used in conjunction with the word 'snow' in the compound 'snowflake' or the phrase 'flake of snow'. It may also be used with a variety of other meanings unrelated to snow, e.g. flake of paint, of dandruff, etc. I find no historical evidence that 'flake' originated as a snow term either, although it seems to have related to ice; it is of Scandinavian origin and may have meant 'disk' or 'floe'. If we disqualify 'flake', that leaves 'snow' as the solitary English snow word (by Martin's criteria), and Eskimo languages with three times that many!
In his essay "The Great Eskimo Vocabulary Hoax," Geoffrey Pullum writes very entertainingly about the snow words example, citing Martin's paper and poking fun at the scholars who have slavishly repeated the claim promulgated by other scholars with no reference to primary data. Pullum is not above overstating the case just a bit: "The truth is that the Eskimos do not have lots of different words for snow, and no one who knows anything about Eskimo has ever said they do." (This second part of his statement is certainly true, since those familiar with the actual data have kept fairly silent.) "Anyone who insists on simply checking their primary sources will find that they are quite unable to document the alleged facts about snow vocabulary (but nobody ever checks, because the truth might not be what the reading public wants to hear.)" A minor quibble with Pullum is that he calls the bungling treatment of the snow example a hoax, even though there was never really any intention to deceive. In an appendix titled "Yes, but how many really?" Pullum is to be applauded for taking the radical step of consulting a bona fide specialist of Eskimo linguistics, Anthony Woodbury. By all accounts Pullum is the first scholar writing on this subject who has thought to consult a specialist, or at least the first one to openly admit doing so. Based largely on Jacobson's Yup'ik Eskimo Dictionary, Woodbury estimates that there are from one to two dozen words (lexemes) for snow, depending on which ones are included.
Looking at data from Inupiaq that I include in an appendix, one can see that it is difficult to decide if there is really a large number of snow terms for a number of reasons, many of which are pointed out for Central Alaskan Yupik by Woodbury. First, many of the terms describe related phenomena such as ice; remember that in counting roots Martin suggested that only those meaning 'snow' be included and not related terms. Next, many of the terms are derived from non-snow terms, e.g. the verb stem natiRvik- 'for snow to drift along the ground' is based on the noun stem natiq meaning 'floor, bottom'. The meaning of the derived stem has shifted somewhat from the stem meaning even though there is still a clear semantic relationship between the two. While it is not a basic unanalyzable stem like qanik, natiRvik- is not a transparent recent derivation either and goes back to a derived Proto-Eskimo stem, *natiquvig-. Puktaaq 'iceberg' is easier to discount, since it is an obvious derivation from the verb stem pukta- 'to float', which itself derives from puge- 'to surface'. Then there are metaphorical usages like mapsa 'snow cornice', which originally means 'spleen', since a snow cornice is meant to remind one of how a spleen overhangs other organs. The snow meaning of mapsa is found only in Alaskan Inupiaq and is clearly secondary. Even if we discount derived words and transparent metaphors, there are still a number of terms like pukak 'granular snow' and reflexes of pirtur 'snowstorm,' which are not obviously related to a more basic root.
Another relevant fact when tallying snow terms involves how current these terms are in the language. Some of them are a part of any Inupiaq speaker's vocabulary, including the three basic terms mentioned earlier, qanik, apun, and aniu. Others are quite specialized, like piqaluyak 'glacial ice from a river', and are likely to be known only by elders and particularly hunters. Many of the terms are not in general use and would not be known to much of the Inupiaq-speaking population. Yet, the lists of terms that exist were compiled within recent decades from people who knew the specialized vocabulary. Then there is the question of how many snow words English has so that a comparison can be made. Sleet, slush, blizzard, and other terms do not include the word snow, just like almost all of the Inupiaq terms that denote some type of snow or ice without including the basic roots that bear those meanings.
Even if we exclude the sorts of terms that some have suggested should not count in our tally of snow terms, it still appears that Inupiaq at least has an extensive vocabulary for snow and ice. It would surely be a surprise if Inuit people did not pay special attention to snow and ice, which are important features of the landscape throughout most of the year. Weather conditions and the state of frozen moisture underfoot are of utmost importance to travelers, hunters, and others, for whom faulty judgment of the terrain can have severe consequences. This particular semantic area demonstrates the detailed knowledge that many Inupiat have about their natural environment, and the example could have easily been something other than snow. An extensive vocabulary exists for both snow and ice, and the claim should make reference to both phenomena. Linguists and others familiar with these languages have always taken it for granted that there is extensive vocabulary for the areas in question.
 In spite of the controversy surrounding the term "Eskimo," I use it because the snow example is almost always attributed to "the Eskimo language," undifferentiated as to which of the six languages and numerous dialects is intended.
 Boas's qana is probably qannik or qanik, piqsirpoq is a verb meaning "there is a snowstorm," and dialect information would be needed to assess the word for 'snowdrift', which appears to be equivalent to qimugyuk in Caribou Eskimo, for example.
 The Comparative Eskimo Dictionary appeared in late 1994, after Martin's earlier 1982 version, but before the 1998 publication of the reader Linguistics at Work by Oaks, in which the most recent version of Martin's article appears. In any case, the stem anigu and its Inuit reflex aniu have been well documented for decades. I use g to represent a velar fricative here and R for a uvular fricative.
 A derived form of the stem anigu appears in Fabricius (1804) as a shaman's word, but this stem is not found in Schultz-Lorentzen's dictionary.
EASTMAN, Carol M. 1975. Aspects of Language and Culture, San Francisco, Chandler.
FORTESCUE, Michael, Steven Jacobson, and Lawrence Kaplan. 1993. Comparative Eskimo Dictionary with Aleut Cognates, Fairbanks, Alaska Native Language Center.
HALPERN, Mark. n.d. The Eskimo Snow Vocabulary Debate: Fallacies and Confusions, unpublished paper.
JACOBSON, Steven A. 1984. Yup'ik Eskimo Dictionary, Fairbanks, Alaska Native Language Center.
MARTIN, Laura. 1986. Eskimo Words for Snow: A case study in the genesis and decay of an anthropological example, American Anthropologist 88, 2(June), 418-423.
MACLEAN, Edna et al. n.d. North Slope Inupiaq Dictionary, unpublished manuscript.
PRUITT, William O., Jr. 1984. Snow and Living Things, in Olson, Rod, et al., Northern Ecology and Resource Management, Edmonton, The University of Alberta Press.
PULLUM, Geoffrey K. 1991. The Great Eskimo Vocabulary Hoax, Chicago, University of Chicago Press, 159-171.
SCHNEIDER, Lucien. 1985. Ulirnaisigutiit: An Inuktitut-English Dictionary of Northern Quebec, Labrador, and Eastern Arctic Dialects, edited by D. Collis, Quebec, Les Presses de l'Université Laval.
SCHULTZ-LORENTZEN, C.W.. 1927. Dictionary of the West Greenlandic Eskimo Language, Meddelelser om Grønland 69, Copenhagen, Reitzels.
SUN, Susan et al. 1979. Kobuk Iñupiaq Junior Dictionary, Anchorage, National Bilingual Materials Development Center.
WEBSTER, Donald and Wilfried Zibell. 1968. Iñupiat Eskimo Dictionary, Fairbanks, Summer Institute of Linguistics.
WHORF, Benjamin Lee. 1940. Science and Linguistics, Cambridge, Technology Review (MIT), 42:229-31, 247-48, reprinted in Language, Thought, and Reality, 1956, John B. Carroll ed., 207-219.
Full citation for the original article:
Kaplan, Lawrence. 2003. Inuit Snow Terms: How Many and What Does It Mean? In: Building Capacity in Arctic Societies: Dynamics and shifting perspectives. Proceedings from the 2nd IPSSAS Seminar. Iqaluit, Nunavut, Canada: May 26-June 6, 2003, ed. by François Trudel. Montreal: CIÉRA -- Faculté des sciences sociales Université Laval.