Split View: 언어가 생각을 결정하는가 — 사피어-워프 가설의 몰락과 부활

Does Language Determine Thought — The Fall and Rise of the Sapir-Whorf Hypothesis

Introduction — Two Faces of a Fascinating Hypothesis

"The Eskimo language has hundreds of words for snow." "The Hopi language has no time, so the Hopi live time differently from us." You have probably heard some version of these stories. The idea that language is a window onto the world, and that speaking a different language means living in a different world — known in linguistics as the Sapir-Whorf hypothesis — is fascinating enough to have become the premise of the sci-fi film Arrival.

And in the history of scholarship, this hypothesis has ridden a roller coaster. It captivated a generation, crashed into half a century of ridicule, and was revived in the 2000s together with sophisticated experiments. Today we follow that drama, and at the end we get to the question: so what actually changes when you learn a foreign language?

Act One — The Birth of the Hypothesis, and the Exaggeration

The story begins with two figures in early twentieth-century American linguistics: Edward Sapir, a giant of anthropological linguistics, and his student Benjamin Lee Whorf, whose day job was fire insurance inspector. During his insurance investigations, Whorf made an intriguing observation. Because of the phrase "empty drums," people would smoke right next to containers that were in fact full of flammable vapor. A case of a word steering behavior.

Whorf went much further than that. Analyzing the language of the Hopi, a Native American people, he argued that because Hopi lacks a tense system like those of European languages, the Hopi do not conceptualize time as a flowing river. Different language, different cosmology — this is the claim later systematized as linguistic determinism, the strong version of the hypothesis.

The problem was evidence. In 1983, the linguist Ekkehart Malotki, in "Hopi Time," a work of more than 600 pages, showed with massive documentation that Hopi is rich in temporal expressions and tense-like devices. The fate of the Eskimo vocabulary story is even more dramatic. What the anthropologist Boas had mentioned was three or four roots, but as the story was passed along and along, it swelled into dozens, then hundreds. The linguist Geoffrey Pullum summed up the affair in an essay titled "The Great Eskimo Vocabulary Hoax," holding it up as a specimen of academic myths that replicate without verification. For the record, the Inuit languages are agglutinative — a single word can grow as long as a sentence — so the very exercise of counting "words" barely makes sense there.

On top of this, as Chomsky's universal grammar became the mainstream of linguistics, the language faculty common to all humans, rather than the differences between languages, moved to the center of attention. By the late twentieth century, the Sapir-Whorf hypothesis had been pushed into the textbook slot reserved for "refuted hypotheses."

Act Two — The Weak Version Returns, with Experiments

The revival came by shrinking the size of the claim. Not "language determines thought" (a prison of thought), but "language influences the habits of attention and memory" (the defaults of thought). This is the weak version, linguistic relativity, and the experiments of the 2000s took aim at it. Let us look at three flagship cases.

Color — the two blues of Russian. Russian has no single "blue" that bundles light blue (goluboy) and dark blue (siniy). In a 2007 experiment by Jonathan Winawer and colleagues, Russian speakers performed color discrimination tasks that crossed the boundary between the two blues faster than English speakers did. The decisive part is this. When an interference task (rehearsing digits) prevented them from using language, that advantage disappeared. It was not that their eyes saw color differently — linguistic categories were being recruited into the perceptual task in real time.

Space — a language without left. In Guugu Yimithirr, an Australian Aboriginal language, relative directions like left and right are barely used, and all space is expressed in absolute compass directions — north, south, east, west. As in, "your cup is to the north of the table." According to field studies by the team of Stephen Levinson, speakers of this language track compass bearings with astonishing accuracy even in unfamiliar rooms or while on the move. To say what the language demands, you must constantly update an inner compass, and that training reshaped the very profile of their cognitive abilities. The picture is not language locking thought in, but language making you practice a particular kind of thinking every single day.

Grammatical gender — bridges are beautiful and keys are sturdy. In German the bridge (Brücke) is a feminine noun; in Spanish (puente) it is masculine. In a well-known study by the team of Lera Boroditsky, looking at the same photograph of a bridge, German speakers more often picked adjectives like "elegant" and "beautiful," while Spanish speakers more often picked adjectives like "sturdy" and "dangerous." With keys (masculine in German, feminine in Spanish) the pattern flipped. To be fair, though, some experiments in this line have since seen mixed replications, and the size and conditions of the effect are still under debate. The same healthy verification process we saw in the replication crisis post is running here as well.

Beyond these, studies of quantity tasks among speakers of Piraha, an Amazonian language with extremely few number words, and studies of how agent marking in event descriptions affects eyewitness memory point in the same direction. To sum up — language does not change the performance of perception and memory; it changes the wiring and the defaults of attention. The effects are real, measurable, mostly not large, and they grow the more the task recruits language.

Act Three — So What Changes for Us

There are three points where this conclusion meets everyday life.

Learning a foreign language is attention training, not vocabulary memorization. The English article system makes you compute "is this something the listener already knows" in every sentence; the Korean honorific system makes you compute "what is the relationship between the other person and me." A Korean speaker learning Japanese, all of us learning English — we are learning not new words but a new list of information to pay attention to. That is the real reason a foreign language is hard and tiring, and the reason it grows the muscles of thought. It is the same conviction behind this blog supporting parallel reading in Korean, English, and Japanese for every post.

Choosing language is choosing a frame. The lesson of the "empty drums" still holds. Calling people "resources," "manpower," "headcount" versus calling them "people" in a meeting, calling an outage "someone made a mistake" versus "a gap in the system" (blame-free language) — these send the attention of the team to different places. Words do not determine the world, but they do change the defaults of the meeting room.

And one dose of humility. This is the lesson of the Eskimo vocabulary myth. The more fascinating the story, the more it pays to go back to the original source — the habit that the paper series on this blog keeps practicing is needed just as much in popular linguistics.

Closing — Not a Prison but a Home

If the hundred years of the Sapir-Whorf hypothesis had to be compressed into one sentence, it would be this. Language is not the prison of thought but its home. We walk by default along the paths our mother tongue has laid down, but those paths are not the only ones, and for every language we learn, another path opens.

If you are learning a new language, remember that the expression you memorized today is more than a word. It is practice at seeing something different first in the same scene of the world. And the way that practice accumulates is exactly what we already know — spacing it out and retrieving it.