Last updated: November 3, 2026

Ludwig Wittgenstein (1889 to 1951) changed the philosophy of language twice. In his early Tractatus Logico-Philosophicus (1921) he argued that meaningful sentences are logical pictures of possible facts, so that whatever cannot be pictured, ethics, religion, the meaning of life, cannot be said at all. In his later Philosophical Investigations (1953) he abandoned the picture theory and argued that the meaning of a word is, in most cases, its use in the language: words are tools in many different “language games”, woven into human activities and forms of life. Both books share one conviction. Most philosophical problems arise from misunderstandings of how our language works.

That is a different claim from the one usually attached to his name online, that language determines how we think. The idea that each language shapes the thought and perception of its speakers is linguistic relativity, better known as the Sapir-Whorf hypothesis, and it has its own history, its own evidence and its own problems. This post untangles the two, because the difference matters, and because the honest answer to the question in the meme is more interesting than a slogan.

Short Answer: Yes, With an If

The meme borrows a scene from The Simpsons. A worried man grabs another by the shoulders: “Mr. Philosopher, I need to know. Does language shape the way we think?” In the second panel, the man being asked now has Wittgenstein’s face pasted on, and he answers with professional calm: “Short answer: ‘yes’ with an ‘if’. Long answer: ‘no’ with a ‘but’.”

It is a good joke and also, unusually, a good summary of the research. Yes, language influences thought, if you mean that the words and grammar we use make some distinctions easier, faster or more habitual. No, language does not imprison thought, but it shapes what we notice and how we argue. And the person on the right of the meme would probably have objected to the question itself, which is the most Wittgensteinian part of the joke. Asking whether “language” shapes “thought” assumes that these are two separate things, one acting on the other. Wittgenstein spent his later career doubting exactly that kind of picture.

Two Wittgensteins

Wittgenstein was born in Vienna into one of the richest families in Europe, studied engineering, and arrived in Cambridge in 1911 to work with Bertrand Russell on the foundations of logic. He wrote most of the Tractatus as a soldier in the Austrian army during the First World War, believed it had solved the problems of philosophy, and left the subject to become a village schoolteacher in rural Austria. When he returned to Cambridge in 1929, it was because he had come to think the book was wrong.

The result is that “Wittgenstein’s philosophy of language” refers to two very different views, usually called the early and the later Wittgenstein. Students sometimes encounter quiz questions asking which philosopher expanded linguistic philosophy, and Wittgenstein is the usual answer, but the honest version is that he founded one of its major traditions and then spent twenty years dismantling it. Both phases shaped the century. The Tractatus inspired the logical positivists of the Vienna Circle; the Investigations fed the ordinary language philosophy of Oxford and much of what philosophers still do when they ask what a word means.

The Tractatus: Language as a Picture of the World

The Tractatus is a short book of numbered propositions that begins: “The world is everything that is the case.” The world consists of facts, facts consist of arrangements of simple objects, and a meaningful proposition is a picture of a possible arrangement. Just as a model of a car crash in a courtroom represents the accident because its toy cars are arranged like the real ones, a sentence represents a state of affairs because its elements are arranged like the elements of reality. If the arrangement exists, the sentence is true; if not, false.

The analogy is not ours. According to his wartime notebooks, Wittgenstein got the idea from a report of a Paris court case in which a traffic accident was reconstructed with miniature cars and dolls. A model can represent a situation because there is a correspondence between its parts and the parts of what it depicts. Language, he concluded, works the same way, only with the logical form hidden under the surface of everyday grammar. That last point mattered to him: ordinary language disguises thought, “just as the outward form of the clothes” hides the form of the body beneath them, and philosophy’s task is a critique of language that brings the hidden logical form to light.

The theory has a dramatic consequence. Only sentences that picture possible facts make sense. The propositions of logic are empty tautologies, and the sentences of ethics, aesthetics and religion picture nothing at all. They try to say what can only be shown. Wittgenstein did not think these things unimportant. On the contrary, he wrote to a publisher that the book’s point was ethical and that the important part was the one he had not written. The famous last line draws the boundary: “Whereof one cannot speak, thereof one must be silent.”

This is the context of the most quoted sentence in the book: “The limits of my language mean the limits of my world.” It is not a claim that English speakers and Russian speakers live in different worlds. It belongs to a dense passage about logic and solipsism, and the “language” in question is not English or German but the one logical language that any meaningful language shares. We have a whole post on what Wittgenstein actually meant by that line.

Meaning Is Use: The Language Games

The Philosophical Investigations opens by attacking a view of language that Wittgenstein found in Augustine and, implicitly, in his own earlier book: the idea that words are names and sentences combinations of names. To show how limited that picture is, he invents a tiny language. A builder calls out “block”, “pillar”, “slab”, “beam”, and his assistant brings the stone he has learned to bring at that call. Is “slab” here a name, an order, a request? It is whatever the activity makes it. The meaning lies in the use.

From there Wittgenstein introduces the term “language game”, meant to bring into prominence that speaking a language is part of an activity, or of a form of life. His own examples are a deliberately motley list: giving orders and obeying them, describing an object, reporting an event, speculating, making up a story, play-acting, guessing riddles, making a joke, translating, asking, thanking, cursing, greeting, praying. There is no single essence that all these share, any more than all games share one feature. Some games have winners, some do not; some involve skill, some luck. Instead there is a network of overlapping similarities, which he calls family resemblances.

The slogan that came out of this is “meaning is use”, and it needs one qualification Wittgenstein himself made: for a large class of cases, though not all, the meaning of a word is its use in the language. To understand the word “pain”, “game” or “justice”, you do not look for an object it names or a definition that captures every case. You look at how it is used, by whom, in which activities, with which consequences. That is why legal language, medical language and the language of a football commentary can share words and still mean different things by them.

The Beetle in the Box and Philosophy as Therapy, in Brief

The later Wittgenstein drew two famous conclusions from the idea that meaning is use. The first is the beetle in the box (Philosophical Investigations §293). If everyone had a box with something called a beetle inside, and nobody could ever look into anyone else’s, the thing in the box would drop out of the language game altogether. Words for sensations such as pain, Wittgenstein concluded, cannot get their meaning from a private object that only the speaker can inspect.

The second is his view of philosophy itself. Many philosophical problems, he argued, arise when language goes on holiday (§38), when words are lifted out of the practices that give them their sense, and the philosopher’s task is to show the fly the way out of the fly-bottle (§309). We treat both ideas in depth, with everyday examples, in our post on Wittgenstein’s language games and linguistic confusion. Here they matter for one reason: they explain why Wittgenstein would be suspicious of the claim that language imprisons thought, the question the rest of this article turns to.

Did Wittgenstein Think Language Shapes Thought?

Not in the sense of the popular question. Wittgenstein never argued that speakers of different languages perceive colours differently or that grammar locks thought into channels. He was not doing empirical linguistics at all. What he did argue is more radical in one way and more modest in another. Concepts, he thought, are bound up with practices: “to imagine a language means to imagine a form of life.” Our agreement in using words rests on agreement in ways of living, in what we find natural, funny, obvious or absurd. If a lion could talk, he remarked, we could not understand him, not because lion grammar is strange but because a lion’s form of life is too different from ours.

He was also fascinated by what he called seeing an aspect: the moment when a drawing that was a duck suddenly becomes a rabbit, although nothing on the page has changed. The duck-rabbit, which he borrowed from the psychologist Joseph Jastrow, shows that seeing can involve interpretation, and that our concepts can change what we notice in the same object. That is a real link between language and perception, but it is about concepts and attention, not about English versus Hopi. If you want the duck-rabbit on your chest, that is what the designs below are for.

Sapir, Whorf and Linguistic Relativity

The claim people usually mean when they ask whether language shapes thought comes from two American linguists. Edward Sapir argued in 1929 that the “real world” is to a large extent unconsciously built up on the language habits of a group. His student Benjamin Lee Whorf went further, writing in 1940 that we dissect nature along lines laid down by our native languages, and claiming that the Hopi language, which he said lacked words and grammar for time as a flowing quantity, gave its speakers a different metaphysics of time. The label “Sapir-Whorf hypothesis” was attached later by others, and comes in two strengths. Linguistic determinism, the strong version, holds that language fixes the limits of what we can think. Linguistic relativity, the weak version, holds that language influences habitual thought and perception.

The strong version has fared badly. Whorf’s claims about Hopi were challenged in detail by the linguist Ekkehart Malotki, whose book Hopi Time (1983) documented extensive Hopi vocabulary for time. People can think about distinctions their language lacks, children think before they speak, and Steven Pinker’s The Language Instinct (1994) made the case that thought runs on something deeper than any particular language. But the weak version has made a real comeback in experimental psychology, and the evidence is worth looking at case by case.

The picture that emerges is roughly what the meme says. Language does not build the walls of thought, but it lays down well-worn paths. What your language forces you to mark, such as tense, gender, direction or evidence, you get very good at noticing. What it leaves unmarked, you can still think about, you just do it less automatically. The linguist Guy Deutscher put the point neatly in Through the Language Glass (2010), borrowing an idea from Roman Jakobson: languages differ less in what they allow you to say than in what they oblige you to say.

Newspeak, Arrival and a Language Without Numbers

Fiction has always loved the strong version of the hypothesis. In the appendix to Nineteen Eighty-Four (1949), George Orwell explains the purpose of Newspeak, the official language of his totalitarian state: by shrinking the vocabulary and stripping words of unwanted meanings, the regime hopes to make heretical thought literally impossible, because there will be no words in which to think it. Ted Chiang’s story “Story of Your Life” (1998), filmed as Arrival (2016), runs the experiment in the other direction. A linguist learns the written language of alien visitors, whose sentences are composed all at once rather than word by word, and gradually begins to experience time the way they do, with past and future present together.

Both stories dramatise exactly the claim that most researchers today reject: that language sets hard limits on what can be thought. The real evidence points to something subtler. A widely discussed case is the Pirahã, a small Amazonian people studied by the linguist Daniel Everett, whose language appears to lack words for exact numbers. Experiments by Peter Gordon (2004) and later by Michael Frank and colleagues (2008) found that Pirahã speakers could match quantities exactly when the objects were in front of them, but struggled to remember exact large quantities over time. Language did not prevent them from perceiving how many things there were. It seems instead to work as a cognitive technology, a tool that makes certain kinds of thinking, such as holding an exact number in memory, much easier.

That is a more Wittgensteinian picture than Orwell’s. Words are not a cage around thought, but tools in a practice, and a community that lacks a tool will find some tasks hard that others find trivial. The danger Orwell saw was real, but a Wittgensteinian would doubt that Newspeak could ever be completed, because people keep inventing new uses for old words, and that inventiveness is precisely what language games are.

Weak or Strong? What the Experiments on Language and Thought Show

Researchers usually distinguish two versions of the Sapir-Whorf hypothesis. The strong version, linguistic determinism, says that language fixes the limits of what its speakers can think. The weak version, linguistic relativity, says only that the habits of a language nudge attention, memory and categorisation in certain directions. Almost nobody defends the strong version today. The weak version has real support in some areas and has failed to replicate in others, which is exactly what makes it interesting.

Some of the most famous findings have not held up well. In 2001 Lera Boroditsky reported that Mandarin speakers, whose language often describes time vertically, thought about time differently from English speakers; later attempts to repeat the effect, including studies by David January and Edward Kako (2007) and by Jenn-Yeu Chen (2007), did not find it. A widely cited study on grammatical gender suggested that German speakers, for whom the word for bridge is feminine, describe bridges as elegant and slender, while Spanish speakers, for whom it is masculine, describe them as strong; a preregistered replication by Anne Mickan, Maren Schiefke and Anatol Stefanowitsch (2014) found no such effect.

Other effects seem more robust, especially where a language forces its speakers to keep track of something all the time, such as the cardinal directions in Guugu Yimithirr or the two basic blues in Russian discussed above. The linguist Dan Slobin proposed a useful way to put it: languages shape thinking for speaking. Because you must mention certain distinctions to say anything at all, you get practised at noticing them. That is a far cry from Orwell’s cage, and much closer to Wittgenstein’s picture of language as a set of tools that make some jobs easy and others awkward.

What Wittgenstein Adds to the Debate

So where does that leave the philosopher in the meme? Wittgenstein would have found the Russian blues experiments interesting and slightly beside the point. His question was not how much one language tilts perception compared to another, but what it is for any word to mean anything at all, and why we keep getting tangled in our own concepts. In that sense he goes deeper than Whorf. He suggests that the line between language and thought is not a wall with influence passing through it, but a set of shared practices in which both are embedded from the start.

That also explains why the meme’s answer is so Wittgensteinian. “Yes with an if, no with a but” refuses the yes-or-no question and points you back to the uses of the words in it: which language, which kind of thought, measured how? Wittgenstein would have recognised the move. As he put it in the preface to the Investigations, he did not want his writing to spare other people the trouble of thinking. For more on the way language games can be weaponised in modern media, read The Gettier Problem and the Filter Bubble, and for where his work sits in the split between the two great traditions of modern philosophy, see Continental vs Analytic Philosophy.