Why the worst idea in linguistics won’t die
The Sapir-Whorf hypothesis is mostly wrong
In 1998, Ted Chiang published “Story of Your Life,” a novella about a linguist named Louise Banks, who learns an alien language with a non-linear structure. In learning it, she begins to experience time the same way the aliens do: past, present, and future collapse into a single act of perception.
Eighteen years later, Denis Villeneuve turned this story into a film called Arrival, and the idea that language can affect how you see reality found its way firmly into the mainstream. Suddenly linguists were being asked at cocktail parties whether the premise of the film was true: could learning a new language really change the way you see the world?
Speaking mainly for myself, I think we linguists were just happy to be noticed, and none of us wanted to give people the blunt truth. So most of us said something like, “It’s complicated.”
But it actually isn’t that complicated. The premise explored by Arrival is that learning a new language can utterly reshape your perception.
It’s a beautiful idea, and it made a great story and a great film. It is also mostly false.
The idea explored in Arrival has a name: the Sapir-Whorf hypothesis, although the name is itself misleading. The linguists Edward Sapir and Benjamin Lee Whorf never co-authored a paper, never jointly proposed a hypothesis, and would probably be surprised to find their names joined to describe an idea that neither of them ever proposed in the form we understand it today. The term was coined in 1954, after the death of both Sapir and Whorf, by another linguist, Harry Hoijer.
What Whorf actually did, working from his day job as a fire insurance inspector in Hartford, Connecticut, was study Hopi and other indigenous languages of the Americas.
From his study, he concluded that the grammars of these languages revealed fundamentally different ways of experiencing reality. The idea caught fire — ironically, given his profession — and has been burning through popular culture ever since.
You’re reading The Dead Language Society, where 40,000+ readers explore the hidden history of the English language. I’m Colin Gorrie: PhD linguist and your guide through 1,500 years of English being weird.
I publish every Wednesday. Paid subscribers get every issue, the full archive, and the content I’m most proud of: practical guides to reading historical texts yourself, honest takes on how language really works, and live book clubs where we read texts like Beowulf and (up next!) Sir Gawain and the Green Knight.
Mostly false. Mostly.
I said that the Sapir-Whorf hypothesis was “mostly false.” The problem is that there are actually two versions of the idea, and it’s easy to conflate them.
The strong version of the Sapir-Whorf hypothesis, linguistic determinism, says that your language constrains what you can think.
It argues that if your language lacks a word or a grammatical structure for a concept, that concept is literally unavailable to you. This is the version behind Arrival, behind Orwell’s Newspeak, and behind every claim that such-and-such language has a word for something we can’t even conceive of.1
It’s also the version of the hypothesis that has been dead in linguistic circles for half a century.
The weak version, linguistic relativity, makes a much more modest claim: that the language you speak can nudge certain cognitive processes, especially under time pressure, and especially in tasks involving memory and categorization.
This version has real evidence behind it. The evidence is interesting but it’s far less dramatic than any science fiction writer would want it to be.
Here’s a quick example of what the weak version actually looks like in practice:
Russian has two colour terms where English has one: siniy for dark blue and goluboy for light blue. Russian speakers don’t consider these versions of a single basic colour, like English speakers consider navy blue and sky blue to be versions of a more basic colour term blue. For a Russian speaker, siniy and goluboy are as distinct as blue and green are to an English speaker.
In 2007, a team of researchers led by Jonathan Winawer tested whether this linguistic difference had any measurable cognitive effect. It did. Russian speakers were faster at discriminating between two shades of blue when those shades fell on opposite sides of the siniy/goluboy boundary than when both shades fell within the same category. English speakers, who lump all of these shades under blue, showed no such advantage.
But, and this is the critical part, when researchers gave participants a simultaneous verbal task, occupying the language centres of the brain, that advantage vanished.
The effect was real, but it was fragile and entirely dependent on active linguistic processing. No one’s perception was permanently altered. No one was trapped in a world created by language.
So the weak version of the Sapir-Whorf hypothesis has real evidence behind it. But how much evidence? And how far does it actually go?
The Russian blue study studied a single perceptual domain. If language genuinely shapes thought, even modestly, at the margins, the effect should show up beyond colour and beyond the confines of the laboratory.
Turns out, it does.
