What you will get from this article
- What the claim that language determines thought (linguistic determinism) says, and what has been taken as evidence
- How Pinker checks Whorf’s famous examples (“empty” drums, Hopi time, Eskimo snow words)
- What remains after the checking, and where caveats apply
The answer first. Pinker calls the idea that thought is the same thing as language a “conventional absurdity”: something that goes against common sense yet is believed by everyone. The first half of chapter 3 tests the evidence traditionally cited, namely Whorf’s argument, one item at a time. Most of it turns out to be weak anecdote or error. Pinker nonetheless grants that language can have weak effects on thought.
Starting point: Orwell’s Newspeak (pp. 55–56)
Chapter 3 opens with the appendix to Orwell’s Nineteen Eighty-Four. Newspeak was meant not only to express the Party’s worldview but to make all other modes of thought impossible. The word free could only be used in sentences like “This dog is free from lice,” and its political sense would vanish as a concept, becoming nameless.
Pinker notices Orwell’s own caution: “at least so far as thought is dependent on words.” Is thought dependent on words? That question is the theme of the whole chapter (p. 56).
“A conventional absurdity” (pp. 57–58)
The idea that thought is the same thing as language is an example of what can be called a conventional absurdity
Pinker’s reply rests on ordinary experience.
- We sometimes say or write a sentence and realize “that’s not what I meant.” Then what we meant must exist apart from what we said.
- When we hear or read, we remember the gist, not the exact words.
- If thoughts depended on words, how could a new word ever be coined? How would a child learn a first word? How would translation be possible (p. 58)?
Sapir and Whorf (pp. 59–60)
The hypothesis is tied to Sapir and Whorf. Sapir, a student of Boas, belonged to a school that showed non-industrial peoples have complex and valid worldviews. He observed that in Native American languages speakers are made to attend to different aspects of reality. English makes you attend to tense via -ed; Wintu makes you mark, on the verb, how you know (seen versus heard) (p. 59).
Whorf, an inspector for the Hartford Fire Insurance Company and an amateur scholar, took this much further. He wrote that “we dissect nature along lines laid down by our native languages” (pp. 59–60).
He said the idea came from his fire-prevention work.
- A worker tossed a cigarette into an “empty” gasoline drum full of vapor, and it exploded.
- Another lit a blowtorch near a “pool of water” that was in fact decomposing tannery waste giving off flammable gas.
Whorf claimed the word empty means both “without contents” and “null and void, inert,” so workers did not notice the danger (p. 60).
Pinker’s check
Pinker rereads the example (p. 60): the worker was fooled by his eyes, not by English. Gasoline vapor is invisible; a drum full of vapor looks just like a drum with nothing in it.
Next, Apache. Whorf said the Apache It is a dripping spring would be, in English, “As water, or springs, whiteness moves downward,” adding “How utterly unlike our way of thinking!” (p. 60). The psycholinguists Lenneberg and Brown made two points (pp. 60–61).
- Whorf did not actually study any Apaches; it is not clear he ever met one. His claims about Apache psychology rest entirely on Apache grammar, hence circular. Apaches speak differently, so they think differently; how do we know they think differently? Just listen to how they speak!
- Whorf made deliberately clumsy word-for-word translations. With equal grammatical justification the first sentence could be rendered “Clear stuff—water—is falling,” and the English “He walks” could be rendered “As solitary masculinity, leggedness proceeds” (p. 61).
Pinker adds Mark Twain’s mock word-for-word English of a German speech (verbs piled at the end, etc.) to show that the same trick would make the German mind look strange (p. 61).
Colour words (pp. 61–63)
Of Whorf’s “kaleidoscopic flux of impressions,” colour is the most eye-catching. Wavelength is continuous; physics has no breaks at red, yellow, green. Languages differ in colour inventory (Latin lacks generic “gray” and “brown,” Navajo merges blue and green, Russian has distinct words for dark blue and sky blue). “Language puts the frets in the spectrum” (p. 62).
But Pinker says physiologists find a basis for colour boundaries.
- The eye has three kinds of cones, and neurons are wired to respond best to contrasts of red against green, blue against yellow, black against white.
- It would be preposterous to think language could reach down and rewire the retina and ganglion cells (p. 62).
Indeed humans the world over (and babies and monkeys) colour their perceptual worlds using the same palette, which constrains vocabularies. Languages disagree on the wrappers in the sixty-four-crayon box, such as burnt umber and turquoise, but agree much more on the eight-crayon box (fire-engine red, grass green, lemon yellow).
- A language with only two colour words: black and white (dark and light).
- Three: black, white, red.
- Four: black, white, red, plus yellow or green.
- Five: both yellow and green. Six: blue. Seven: brown. More than seven: purple, pink, orange, gray.
So languages add colour words in a fixed order (p. 62). The clinching experiment was with the Dani of the New Guinea highlands, whose language has just black and white. The psychologist Rosch found that the Dani learned a new colour category based on fire-engine red faster than one based on an off-red. The way we see colours determines how we learn words for them, not vice versa (pp. 62–63).
Hopi time (p. 63)
Whorf wrote that Hopi contains no words, grammatical forms, or expressions referring to “time,” to past or future, and that the Hopi had no notion of time as a smoothly flowing continuum (p. 63).
But the anthropologist Ekkehart Malotki studied Hopi at length and showed that Hopi speech does contain tense, metaphors for time, units of time (days, days of the week, parts of the day, yesterday and tomorrow, months, lunar phases, years), ways to quantify time, and words like “ancient,” “quick,” “long time,” “finished.” The culture had a horizon-based sun calendar, exact ceremonial day sequences, and sundial-based timekeeping (p. 63). Pinker blames Whorf’s limited, badly analyzed sample and his leaning toward mysticism.
The Eskimo snow myth (pp. 64–65)
“Eskimos have hundreds of words for snow” is well known. Pinker calls it “the Great Eskimo Vocabulary Hoax.”
Contrary to popular belief, the Eskimos do not have more words for snow than do speakers of English.
- In print the number has been four hundred, two hundred, one hundred, forty-eight. One dictionary gives two. Counting generously, a dozen or so; English is not far behind with snow, sleet, slush, blizzard, avalanche, hail, hardpack, powder, flurry, dusting (p. 64).
- The story did not come from anyone who studied the Yupik and Inuit-Inupiaq languages. According to the anthropologist Laura Martin’s account, Boas casually mentioned in 1911 that Eskimos use seven unrelated roots for snow, Whorf embellished it, and textbooks and popular books inflated the count with each retelling (p. 64).
- The linguist Pullum speculates the story got out of control because it fits so well with many “exotic” facets of Eskimo culture.
- He also notes that even if there were many snow roots, it would not be intellectually interesting. Horsebreeders have names for breeds, sizes and ages; printers have many names for fonts. No one writes that “in the culture of printers fonts are of great enough importance to split the conceptual sphere” (pp. 64–65).
Pinker notes the irony: linguistic relativity came out of the Boas school as part of a campaign to show that non-industrial cultures were as complex and sophisticated as European ones. Yet such anecdotes appeal through a patronizing willingness to treat other cultures’ psychologies as weird and exotic (p. 64).
What the controlled experiments showed (pp. 65–67)
If the anecdotes are hollow, what about controlled studies? Pinker says that thirty-five years of research have shown very little (p. 65).
- Most experiments test banal versions of the “weak” Whorfian hypothesis, namely that words can have some effect on memory or categorization. Some worked, but that is hardly surprising.
- In a typical colour-chip memory test, colours with names are remembered better, but unnamed colours are remembered fairly well too. Subjects remembered the chips in two forms, a visual image and a verbal label, and two kinds of memory are better than one (p. 65).
- When subjects had to say which two of three chips go together, they often paired chips with the same name. But they may just have thought, “I’d probably call those two ‘green’ and that one ‘blue,’ and that seems as good a reason as any” (pp. 65–66).
The only really dramatic finding came from the linguist Alfred Bloom. English has a subjunctive for counterfactuals (If John were to go to the hospital, he would meet Mary); Chinese lacks one and must put it circuitously. Bloom wrote stories containing sequences of implications from counterfactual premises and gave them to Chinese and American students. 98% of Americans correctly answered that the events had not occurred; Chinese students gave the correct answer only 7% of the time. Bloom concluded that Chinese makes it very hard to entertain hypothetical false worlds (p. 66).
But the cognitive psychologists Au, Takano, and Liu were not convinced (p. 67).
- Bloom’s stories were written in stilted Chinese.
- Some of the science stories were genuinely ambiguous.
- Chinese college students tend to have more science training than American students and so were better at detecting ambiguities that Bloom himself missed.
When these flaws were fixed, the differences vanished.
Summary
- Linguistic determinism is a “conventional absurdity”: thought is not the same thing as language.
- Whorf’s examples did not survive checking: the “empty” drum (an optical trap), Apache (circularity and deliberately awkward glosses), Hopi time (refuted by Malotki), Eskimo snow (a hoax).
- For colour, physiology drives words, not language perception.
- The controlled experiments showed only weak effects. Even the dramatic report (Bloom) disappeared once the flaws were fixed.
Cautions
- The book is from 1994. Later research has continued to show limited effects of language on perception and cognition, so a simple “completely refuted” is no longer the majority understanding. This is a summary of Pinker’s argument, not a statement of the current state of research. (ToL’s note; check specialist literature.)
- I have not checked the originals of Whorf and Bloom, or the Hopi and colour studies.