Introduction — The Man Who Overturned Philosophy Twice
There are philosophers who have contradicted themselves. But few have pushed a philosophy to its peak, completed it as a single book, declared with that book that "the problems of philosophy have in essence all been solved," and then dismantled the very heart of that book with their own hands. Ludwig Wittgenstein (1889-1951) was just such a person.
He did philosophy not once but twice. The two periods, commonly divided into the "early Wittgenstein" and the "later Wittgenstein," are so different that they could almost have been written by two different people. The early Wittgenstein saw language and the world as interlocking precisely in a logical picturing relation; the later Wittgenstein saw language not as such a static picture but as something closer to a "game" that people play together as they live.
His life itself reads like a drama. Born into one of the wealthiest families in Vienna, he first studied aeronautical engineering, became captivated by the foundations of mathematics, and finally plunged into the deepest questions of philosophy. He filled the notebooks in his pack while serving in the trenches of the First World War, completed a book, believed he had solved every problem of philosophy, and then left academia to become a village schoolteacher. About a decade later, realizing that nearly everything he had believed was mistaken, he returned to Cambridge.
The question I want to raise in this essay is a simple one. How far can this thing we call "speech," which we use every day, actually reach? Where is the boundary between what can be said and what cannot? And is meaning some jewel lodged somewhere inside a word, or something that lives and breathes only in the way we use words?
Wittgenstein wrestled with this question all his life. And interestingly, today's debate over whether a large language model (hereafter LLM) can be said to "know meaning" overlaps strangely with concepts he refined a century ago. I should state in advance, of course, that this last connection is entirely my own speculation. Wittgenstein himself knew nothing of artificial intelligence.
This essay follows those two revolutions in turn. First we examine the early picture theory and its famous proposition of silence; then we take up the later language games and the theory of use, the private language argument, and family resemblance. Only then will we cautiously ask what that thought suggests for today's artificial intelligence. Let me promise one thing in advance: this essay will not hastily declare either side right. Hastiness was exactly what Wittgenstein himself most guarded against.
The Early Period — the Tractatus and the Picture Theory
A Thin Little Book
Wittgenstein's first major work, the Tractatus Logico-Philosophicus, was first published in German in 1921 (the German title is Logisch-Philosophische Abhandlung), and an English parallel edition appeared the following year, in 1922. The text is astonishingly thin. It is written in a peculiar system, with short propositions numbered 1, 1.1, 1.11, and so on, so that sometimes only a single sentence sits on a page. The form reads almost like a book of mathematical theorems.
The book grew out of thoughts he jotted in a notebook in his pack while serving as an Austrian soldier during the First World War. There is even an anecdote that he refined the manuscript in a prisoner-of-war camp. The book's ambition was almost reckless in scale. By revealing how language pictures the world, he sought to show that the problems of philosophy are in fact pseudo-commotions arising from a misuse of language.
In the years leading to this book, Wittgenstein associated closely with Bertrand Russell at Cambridge. Russell once named this young Austrian the most impressive genius he had ever met. The Tractatus stands on the stream of modern logic running through Russell and Frege, yet it pushes that further into a bold attempt to draw the entire relation of logic, language, and world like a single blueprint. As great as that ambition was, so too was the courage with which he later doubted it.
The World Is the Totality of Facts
The Tractatus opens with a famous first sentence: "The world is everything that is the case." It then immediately drives home the point: "The world is the totality of facts, not of things." Here a key reversal occurs. The world is not a list of things like "stone, tree, person" but a fabric of facts (states of affairs) such as "the stone is next to the tree."
From this view the role of language also becomes clear. The proposition, the basic unit of language, pictures a single fact. Wittgenstein developed this into the idea that would come to be called the picture theory of language.
Why facts first rather than things? Here is one intuition. The word "apple" by itself is neither true nor false, because it asserts nothing. But the sentence "the apple is on the table" can be true or false, because it asserts something about the world. For Wittgenstein, the point where language touches the world is not the isolated word but the proposition, in which words are woven into a single assertion. This dovetails exactly with his seeing the world as a "totality of facts."
A Proposition Is a Picture of Reality
The core analogy of the picture theory runs like this. A proposition depicts reality just as a single picture does. Just as the elements in a picture are arranged in a certain way to show the arrangement of actual objects, so the words (names) in a proposition are arranged in a certain logical structure to show the structure of a state of affairs.
The seed of the idea, it is said, came from a surprisingly trivial episode. Picture a courtroom where toy cars and dolls are laid out on a table to reconstruct a traffic accident. A model car "stands for" a real car, a doll "stands for" a pedestrian. Why can this arrangement depict the scene of the accident? Because model and reality share the same structure. A proposition is the same. A proposition has meaning because it shares the same "logical form" as the reality it sets out to depict. A picture need not resemble its subject down to color or material. It need only share the same logical structure.
[ state of affairs in reality ] [ proposition ]
cat — (is on) — mat "the cat is on the mat"
| | | | | |
object <- shared logical form -> name <- same structure -> name
So a proposition can be true or false. If the picture fits reality, it is true; if it fails to fit, it is false. But either way the proposition has "sense," because the picture is showing something.
Going a little deeper, this idea has one profound appeal. It explains why language can handle "novelty." We can understand a sentence we have never heard before, even at the first moment of hearing it. We grasp at once a sentence like "a purple elephant plays the cello on the library roof." How? Wittgenstein's answer is that we can construct the "logical picture" of the state of affairs the sentence depicts. We have not memorized individual facts; because we know the rules for assembling a picture, we can understand infinitely many new pictures.
In this respect the early Wittgenstein tried to explain the productivity and the systematicity of language at once. The insight that infinitely many propositions can be made from finite names and rules left a lasting echo in many later disciplines that study language.
The Boundary of What Can Be Said
Here a decisive conclusion emerges. For a proposition to have sense, it must share a logical form with reality. Yet that very "logical form" cannot itself be pictured by a proposition. It is like the fact that a picture cannot itself depict, within the picture, the very mode of depicting it uses.
From this Wittgenstein separates what can be said from what can only be shown. Facts within the world can be said. But things like ethics, aesthetics, the meaning of life, God, and logical form itself are not facts that can be pictured by propositions. They cannot be said; they can only emerge or be shown. They are excluded not because they are false. Because they are not the sort of statements that can be true or false at all, they slip through the net of language.
A caution is needed. Wittgenstein was by no means saying that ethics or the meaning of life are "trivial." Quite the opposite. To him, those were precisely the most important things. He was only pointing out that they are not the kind of thing captured by the language of factual statement.
[Structure of the Picture Theory]
proposition (sentence) --- depicts ---> fact (state of affairs)
| |
arrangement of elements <-- same form --> arrangement of objects
Can be said: facts, what is the case in the world
Cannot be said: ethics, aesthetics, meaning of life — only shown
The Meaning of the Silence Proposition
"Whereof One Cannot Speak, Thereof One Must Be Silent"
The Tractatus is built from seven principal propositions, and the last, the seventh, stands alone, the only one in the entire book without subordinate numbers beneath it. That sentence is the famous line that also serves as the title of this essay.
"Whereof one cannot speak, thereof one must be silent."
(The German original is Wovon man nicht sprechen kann, darüber muss man schweigen.)
This sentence is often misread as an expression of mysticism or resignation. But following the logic of the Tractatus, it is not. This silence is not a holding of the tongue born of ignorance but an honest silence that only someone who has clearly grasped the limits of language can reach. It is an exhortation to say clearly what can be said, and not to wring out by force, into words, what cannot be said.
A strange tension hides here. Wittgenstein did not banish the unsayable region from sight; rather, he granted it the highest place. In a letter he compared his task to drawing the limits of ethics "from the outside," not "from within." That is, he did not slot ethics inside the realm of speech; he stepped back to view the whole realm of speech and pointed at what lies beyond it. The silence was also a gesture of reverence toward that beyond.
So this proposition can also be read as a kind of ethical counsel. The more an age overflows with information and assertion, the more it needs the restraint not to reduce, carelessly into words, what words cannot handle. Love, grief, awe, the trembling before death — try to grasp them with a definition and they slip through your fingers. Wittgenstein's silence is closer to a counsel to "respect" such things with silence.
Of course the later Wittgenstein would have re-examined even this proposition of silence. Rather than pushing ethics or aesthetics away at a stroke as "the unsayable," he would have looked at how we actually use the words of that region. That is, the early silence becomes, in the later thought, a "more attentive listening." Yet both attitudes are branches from one root, in that both guard against the easy omnipotence of words. One points with silence beyond the boundary of speech; the other looks more closely within it.
Throw Away the Ladder
The Tractatus contains a famous paradox aimed at itself. The whole book explains the picture theory, yet the picture theory itself is not a fact within the world. So the very sentences of the Tractatus become, by their own standard, meaningless sentences. Near the end of the book, Wittgenstein says that anyone who understands its propositions will come to recognize that they are, in the end, "nonsensical." He urges the reader to climb up on these propositions like a ladder and then, having climbed up, to throw the ladder away.
This is because the Tractatus itself is not a proposition that pictures a fact of the world but an attempt to speak "about" the relation between language and the world. Yet that relation, as said, cannot be spoken and can only be shown. So, strictly speaking, the sentences of the Tractatus are, by their own standard, close to nonsense. They are useful only as a ladder that carries the reader to the right vantage point. This gesture of self-negation is one of the strangest and most honest scenes in the history of philosophy.
How to read this ladder metaphor has long been debated among scholars. One side holds that the Tractatus "indirectly" points at deep truths inexpressible in words. The other, more radically, reads the Tractatus as a therapeutic book that ultimately makes us set aside the very idea of such "unsayable truths." Whichever reading one follows, what is clear is that Wittgenstein presented his most intricate work as something finally "to be surpassed." This structure of self-transcendence may already have foreshadowed, from afar, the turn to the later period.
And He Left Philosophy
Having finished the Tractatus, Wittgenstein believed the fundamental problems of philosophy had been solved. So he left the philosophical world for a time. He worked as an elementary schoolteacher in the Austrian countryside, threw himself into the architectural work of designing a house for his sister, and even worked as a gardener. That a man who could have won fame at Cambridge chose such a path voluntarily shows that his philosophy was not a mere profession but a matter of life.
But the silence did not last forever. From the late 1920s he was drawn back into philosophical conversation, and eventually returned to Cambridge. And on his return, he began to think that his first book harbored a serious mistake.
This return did not happen all at once. Over many years he talked endlessly with colleagues and sharpened his thought against students in lectures. It is no accident that the later work is written as dialogue. His thinking itself grew not from solitary contemplation but from collision with others. In a sense he was already living, in his way of life, the truth he was about to discover — that meaning is rooted in communal practice.
The Turn — Refuting Himself
A Question Cast by a Gesture
There is a famous anecdote. A colleague at Cambridge, an economist, once made an Italian gesture at Wittgenstein, brushing his fingertips outward from under his chin, and asked: "What is the logical form of that?" According to the picture theory, which held that meaning is established only by sharing a structure, every meaning must have a definite logical form. Yet what structure does this gesture picture?
This small challenge, it is said, threw Wittgenstein into deep reflection. Whatever the precise details of the anecdote, it is clear that he gradually turned his back on his own early theory. Does meaning really lie in a picture? Or does it lie in what people do with those words and gestures?
The early picture theory had a hidden premise: that behind every meaningful expression lies a single, strict logical form. But look quietly at everyday language and a suspicion grows that this premise is too narrow. We command, request, joke, and write poems. Might the attempt to cram all these varied acts into the single frame of "a picture of a fact" actually distort reality? The later Wittgenstein pushes precisely this suspicion to the end.
The Turn to the Later Period — Philosophical Investigations
A Second Major Work Published Posthumously
Wittgenstein's second major work, the Philosophical Investigations, was published in 1953, after his death in 1951. That is, this book was not one he published himself while alive but a posthumous work. Its style, too, is the opposite of the Tractatus. Instead of a clean numbering system, it unfolds as a series of dialogical reflections that endlessly question, answer, and doubt themselves. An imaginary interlocutor breaks in to raise objections, and Wittgenstein traces them back in turn.
The later Wittgenstein's first target was none other than himself, that is, the view of language in the Tractatus. He held that we must break free of the simple picture in which meaning arises from pointing to an object.
That does not mean he discarded the early work entirely. To bring his new later perspective into sharper relief, he deliberately stages the early position as an imaginary interlocutor and collides with it. So reading the Investigations is like overhearing a dialogue between the two periods carried on inside one mind.
There is a story behind this book's posthumous publication. Wittgenstein was close to a perfectionist and rarely felt sure that his thought had ripened enough. He revised the manuscript again and again, and in the end never published it in his lifetime. So the Investigations we read is like an unfinished map a thinker refined over a lifetime and left to the world without ever completing it. Perhaps that very incompleteness is the form best suited to a spirit that sought to leave questions rather than answers.
At the opening of the Investigations, Wittgenstein begins by quoting an old philosopher. That passage contains a seemingly natural picture of language acquisition: that a child learns to speak by watching adults point at things and name them. Wittgenstein takes this naive picture as his starting point and step by step reveals how much it leaves out. Pointing and naming is only one piece of language, not the whole. How could we point with a finger at "pain," or "perhaps," or "but"? From this small crack the whole of the later philosophy grows.
The Builder's Language — A Thought Experiment
Early in the Investigations, Wittgenstein proposes that we imagine a very simple language. There is a builder and his assistant. When the builder calls out "Block," the assistant brings a block. At each call of "Pillar," "Slab," "Beam," the assistant brings the corresponding thing. This little language has only these four words.
A question arises here. What is the meaning of the word "Slab" that the builder called out? Is it merely a name tag pointing to some lump of stone lying over there? Wittgenstein's answer is different. The meaning of this word lies in the "work it does" between the two people. It is a command, a signal, a syllable in the flow of working together. If we see a word only as a label stuck onto an object, this living function becomes invisible.
This simple thought experiment takes direct aim at a premise of the early philosophy. The Tractatus quietly started from the picture that every word is a name that "points to" something. But "Slab" on a building site is not a simple name. It is a cry that calls forth an action. Wittgenstein invites us to compare a word to a tool in a toolbox. A hammer, a saw, a ruler, a nail, glue — all are "tools," but each does a different job. So with words. To bind every word to the single function of "name" is as crude and inaccurate as lumping every tool under the single phrase "something that fixes things."
Language Games
Here the famous concept, language games, appears. Wittgenstein stresses that language does not perform only a single essential function, namely "picturing the world." With language we command, ask, promise, thank, curse, pray, joke, and greet. Each of these is a "game" with its own rules and context.
The game analogy matters for two reasons. First, a game has rules, but those rules are kept up by people actually following them together; they are not some eternal form lodged in advance somewhere. Second, a game cannot be separated from the activity of life. Wittgenstein said that "to speak a language is part of an activity, or of a form of life." Language is not a sign system in a vacuum but an act rooted in the way we live.
The game analogy carries one more subtle implication. The rules of a game need not be fully specified before the game begins. We often make up rules as we play, filling ambiguous situations by agreement on the spot. So with language. Meaning does not require that every word's use be fixed exhaustively in advance. A living language always holds a little margin and flexibility, and precisely thanks to that it can adapt to new situations and new uses.
The same sentence carries an entirely different meaning depending on which game it is in. "Fire!" is a warning at a blaze, a command at a shooting range, a line on a theater stage. See a word only as a static dictionary entry and this difference vanishes; see the activity in which the word is set, and its meaning comes alive.
Wittgenstein held that the list of language games can grow without end. He deliberately strings out a long list: giving orders and obeying them, describing an object, constructing something from a report, guessing the course of an event, forming and testing a hypothesis, making up a story, telling and getting a joke, solving an arithmetic problem, translating from one language to another, requesting, thanking, cursing, greeting, praying. The very richness of this list is itself an argument. The attempt to reduce language to a single essence looks narrow again and again before this abundant reality.
The Paradox of Rule-Following
At the heart of the language games lies another thorny question. What exactly is it to "follow a rule"? Wittgenstein gives the famous example of a number series. Suppose we teach someone the rule "add 2" and have them write 0, 2, 4, 6, 8. If after 1000 they jump to 1004, 1008, did they follow the rule "wrongly," or follow "correctly" a different rule from ours?
What this question shows is that the sign of a rule does not by itself fully determine its application. Any finite set of cases is compatible with infinitely many rules. So where does our confidence that we are following the same rule come from? Wittgenstein's answer is that it comes from the practice and training we share, that is, from the form of life we live together. Rule-following is not a solitary calculation in the head but a custom communally upheld. This insight becomes the foundation of the private language argument we examine next.
Meaning Is Use
The phrase that compresses this later thought into a single sentence is precisely "meaning is use." This is a famous gloss on section 43 of the Philosophical Investigations, where Wittgenstein says that in many cases (though not all) the meaning of a word is its use in the language.
The implications of this single step are large. If you want to know the meaning of a word, do not search for some mysterious mental object or Platonic essence the word points to; look at how people actually use the word, in what situations. The meaning of "hello" lies not in some idea in the head but in the custom of exchanging the word when people meet.
But Wittgenstein did not set up this slogan as an absolute formula. He carefully added the qualifier "in many cases." There are cases, as with proper names, where pointing to an object is close to the center of meaning. So "meaning is use" is not another essentialism reducing all meaning to a single frame. Rather, it is closer to a liberating invitation not to see meaning in only one predetermined way. Miss this subtlety and it is easy to mistake the later Wittgenstein for yet another rigid theorist.
The Private Language Argument
The most contested and intricate part of the later philosophy is the private language argument. Wittgenstein asks: is a wholly private language possible, one that only I could know and that is in principle impossible to communicate to others?
Suppose, for instance, that whenever a certain inner sensation arises, I write "S" in a diary to name that sensation. The problem is this. When I next write "S," how can I confirm that I am really pointing to the same sensation? If there is no external standard to compare against, there is no way to tell apart "I used it correctly" from "I feel that it is correct." Wittgenstein compares this to buying several copies of the same newspaper to confirm that the story in it is true. If the ground of verification and the thing verified are the same, the verification is empty. And if there is no standard to separate right from wrong, the very phrase "following a rule" spins idly there.
Wittgenstein's conclusion is this. Meaningful language presupposes rules, and to follow a rule presupposes a practice in which right and wrong can be communally checked. So a wholly private language that is in principle unverifiable does not hold up. This is the argument that supports the heart of the later philosophy: that meaning is rooted not in the private experience inside an individual's head but in public practice. Behind every word we use lies the layered agreement and practice of countless people.
This argument often invites misunderstanding. It will not do to read Wittgenstein as "denying" private sensation or inner experience. He does not deny that we feel pain and feel joy. What he denies is only the idea that we can build meaning by attaching to such inner states a "private label only I can access." That we can put our inner life into words is, paradoxically, because those words gain a place within a public language game. Even when expressing the most intimate things, we lean on the shared asset of a common language.
Family Resemblance — What Do All Games Have in Common?
Finally, let us look at another jewel of the later philosophy, family resemblance. Wittgenstein poses a provocative question. Is there a single essence common to everything we call a "game"?
Picture board games, card games, ball games, the Olympic contests, games of patience played alone, children's games of tag. Some have winning and losing, some do not. Some turn on luck, some on skill. Some are contested by many, some enjoyed alone. Try to find a single feature running through them all, and you fail again and again.
Instead, what we find is a web of overlapping and crisscrossing similarities. It is like turning the pages of a family photo album: there is no single feature common to all, yet the nose resembles the father, the eyes the mother, the gait the grandmother, with resemblances woven this way and that. Wittgenstein called this web of resemblance family resemblance. Many concepts are bound not by the fence of a definition but by such a bundle of resemblances.
This insight is no mere wordplay. It is a fundamental challenge to one impulse Western philosophy has long harbored: the impulse to find, behind every concept, a clear definition of necessary and sufficient conditions. Since Plato, philosophers have asked "what is courage," "what is justice," seeking a single essence common to all cases. Wittgenstein says such an essence is not always found, and that its absence does not make the concept defective.
Our daily life, too, is full of cases of family resemblance. The moment we try to pin down words like "art," "culture," "religion," "love" with a one-sentence definition, we soon hit exceptions. That does not mean we are using these words wrongly. We simply follow the web of resemblance, naturally placing a new case beside the existing ones. Wittgenstein saw this natural ability not as a defect but as the normal working of language.
The concept of family resemblance also acknowledges that the boundaries of language can be blurred. There is no knife-cut boundary between "game" and "not a game." But a blurred boundary does not make the word useless. Wittgenstein asks: "Is a blurred photograph not a picture at all? Is it not often exactly the blurred picture that we need?" The very idea that only a precise definition guarantees meaning was another knot he sought to untie.
Family resemblance among games
chess ── (rules, skill) ── go
| |
(win/lose) (competition)
| |
football ── (ball, many) ── basketball a game of card patience alone
(no winning, yet a game)
A single common essence: none
Instead: a web of overlapping, crisscrossing resemblances
Curing the Diseases of Philosophy
Showing the Fly the Way Out of the Fly-Bottle
The later Wittgenstein redefined the very role of philosophy. Traditional philosophy tried to give correct answers to grand questions: what is mind, what is time, what is meaning. Wittgenstein held that many such questions are not genuine problems but confusions arising from the misuse of language, a kind of mental cramp.
His comparison is striking. The aim of philosophy, he said, is "to show the fly the way out of the fly-bottle." A fly-bottle is a glass jar with a narrow opening; the fly inside keeps bumping against the glass toward the light, unable to get out. The exit is the very opening it came in through. So too with someone caught in philosophical confusion: gripped by a misleading picture, they circle the same spot. The philosopher's job is not to build yet another theory but to show where the wrong turn was taken and so dissolve the confusion.
It is telling that the word Wittgenstein often used here is "picture." For the early Wittgenstein, the picture was the bridge where language touched the world. For the later Wittgenstein, the picture can instead become a trap that holds us captive. "A picture held us captive," he says. When some mistaken analogy or image drives our thinking in one direction without our noticing, we forget how to think outside that picture. Philosophical therapy is the work of raising that captivating picture to the surface of consciousness and freeing us from its grip.
So he said philosophy is "a battle against the bewitchment of our intelligence by means of language." Bringing words back from their metaphysical to their everyday use — that is the cure.
One Case — Where Is Pain
Take one case. Consider the question: where is pain, is it a private inner object? We are tempted to imagine pain as a secret item hidden somewhere in the head. But Wittgenstein invites us to recall how we actually learned the word pain.
A child does not peer inward and name what it finds. When the child falls and cries, adults say "that hurts," and as the behavior, the expression, and the word are woven together, the child learns the use of the word pain. The moment we hunt for a word's meaning in a mysterious inner object, we are trapped in the fly-bottle; the moment we turn to how the word is actually used, we find the way out. This small case connects directly to the private language argument. The word pain has meaning because it is set within a public web of behavior and expression, not because of an actor on a secret stage that only I can peer into.
The Man Called Wittgenstein
Someone Who Lived Philosophy
To understand Wittgenstein, it helps to see his life alongside his thought. He regarded philosophy not as a profession but as a kind of way of life. Though he inherited an enormous fortune, he gave most of it away and chose a frugal life far from luxury. He volunteered for the front in the First World War, and during the Second World War he even worked as an orderly in a hospital.
His writings always carry a certain tension. It is the tension between an almost obsessive craving for clarity and the awareness that what matters most cannot be fully held in words. In one letter he remarked that his book consisted of two parts, and that the more important part was the one he "had not written." It is a characteristic paradox: that the blank surrounded by silence matters more than what is drawn in words.
Why He Sought to Be Right Twice
What is interesting is that Wittgenstein, having refuted the early work in the later, did not dismiss the early work as a "mistaken exercise." He once even conceived of publishing his two phases together. He hoped that the early period's clear drawing of limits and the later period's rich dissolving might stand side by side as lights illuminating each other.
From this attitude we can draw one lesson. To change one's mind is not defeat but becoming more honest. The courage to doubt even the conclusion of which one was once most certain was the very power that made him one of the most influential philosophers of the twentieth century.
On his deathbed he is said to have left the words, "Tell them I've had a wonderful life." For the last words of a man who fought fiercely with himself all his life, stepping toward clarity and honesty, no sentence could fit better. Perhaps what his philosophy hands us is something similar in the end: the posture of pressing on honestly as far as one can say, even if one can never say it all.
Early and Later, at a Glance
Laying out the differences between the two periods in a table makes the breadth of the turn vivid.
| Aspect | Early (Tractatus) | Later (Investigations) |
| --- | --- | --- |
| Core analogy | language as a picture of reality | language as a shared game |
| Where meaning lies | logical correspondence of proposition and fact | the way a word is used, that is, use |
| Unity of language | one essential logical form | many language games, no essence |
| How concepts bind | strict definitions and boundaries | a web of family resemblance |
| Ground of meaning | a picturing relation with the world | communal practice and forms of life |
| Task of philosophy | drawing the limits of language | dissolving confusions bred by misuse |
| Representative proposition | be silent on what cannot be said | meaning is use |
| Private language | not directly treated | argued to be impossible |
One point of balance should be set down, however. Against the common notion that sharply opposes the early and later periods, there is also an interpretation in scholarship that stresses the continuity between them. In that both periods grip the same question, "how does language have meaning," Wittgenstein's thought is at once a break and a single deep, sustained inquiry.
The two periods also agree, unexpectedly, in their attitude toward philosophy. The early Wittgenstein sought to "dissolve" the problems of philosophy by drawing the limits of language; the later, by "untying" the confusions bred by the misuse of language. The method changed, but the conviction stayed constant that the proper task of philosophy does not lie in piling up some grand new metaphysical theory. In this sense one can say Wittgenstein stayed faithful to a single conviction all his life.
A Quick Quiz to Check Yourself
Let us lightly check what we have read. The answers are just below.
Question 1. In what year was the Tractatus first published in German?
Question 2. What is the name for the early Wittgenstein's view of language?
Question 3. In what year was the Philosophical Investigations published, and what
is notable about it?
Question 4. "Meaning is use" is a gloss on which section of which book?
Question 5. The idea that all games share a single common essence — with what
concept did Wittgenstein replace it?
Let us check the answers.
Answer 1. 1921 (in German). The English parallel edition came the next year, 1922.
Answer 2. The picture theory of language.
Answer 3. Published in 1953. The notable point is that it is a posthumous work,
appearing after his death in 1951.
Answer 4. A famous gloss on section 43 of the Philosophical Investigations.
Answer 5. Family resemblance. Instead of a single essence, a web of overlapping
resemblances.
Implications for Today's Artificial Intelligence — From Here on, This Is Speculation
What follows from here is, let me state clearly, not Wittgenstein's claim but my own speculation, an attempt to hold his concepts up against today's problems. Wittgenstein died in 1951, and he had no materials even to imagine today's computers of meaning, let alone large language models. The connections below hold only on the assumption "if we look through his framework," as a thought experiment.
There is a reason for stressing this repeatedly. The temptation to borrow a great philosopher's name to lend authority to a present-day debate is always strong. But that is exactly what Wittgenstein would most have guarded against. He knew better than anyone that even his own concepts grew from a particular context. So please read this chapter not as a place to declare "how Wittgenstein would have seen LLMs," but as a place to test whether, borrowing his concepts, we can ask better questions.
Learning Meaning Through Use
Today's LLMs statistically learn, from vast amounts of text, how words are used together in which contexts. Rather than being injected with a word's dictionary definition, they learn the pattern of the "work that word does" within countless sentences. At this point the later Wittgenstein's slogan, "meaning is use," strangely comes to mind. At least on the surface, both seek meaning not in an abstract definition but in the pattern of actual use.
Let me offer a comparison. Recall how a young child learns its mother tongue. The child does not memorize a grammar book whole. Immersed in a sea of countless utterances, it learns by experience which words call forth which responses in which situations. At least the outward form of this "learning through use" looks faintly like the way an LLM learns. So some feel that Wittgenstein's insight — finding meaning in use rather than definition — unexpectedly echoes, in one form, inside today's machines.
So some regard the LLM as an accidental implementation of a Wittgensteinian view of meaning. The point being that it draws a word's meaning not from an idea in the head but from the web of use.
Interestingly, both camps in this debate enlist Wittgenstein. One side argues, "If meaning is use, then a model that has learned the vast use of human language is, in some sense, participating in a language game." If we need not seek meaning in a mysterious mental picture, then there is no reason to declare that a machine, whose inside we cannot inspect, cannot handle meaning.
But Do Not Get Carried Away Too Quickly
A counterweight must be hung here at once. Wittgenstein's "use" was not a simple pattern of co-occurrence in text. It was a use rooted deeply in forms of life, in the activity of people working, promising, and quarreling together as they live. The reason the builder and his assistant can exchange the word "Slab" is that they are inside the practice of building a wall together.
What an LLM handles is not such embodied practice but something closer to the shadow of text, the product of that practice. And recalling the private language argument raises another question. To "follow" a rule presupposed a practice in which right and wrong are communally checked. Whether the reproduction of a statistical pattern is the same as rule-following in that sense, or something merely resembling it on the surface, is by no means self-evident.
Laying this contrast out in a table makes the outline of the question, rather than a hasty conclusion, come into focus.
| Point to examine | Wittgenstein's "use" | the LLM's "use" |
| --- | --- | --- |
| Where it is rooted | a shared form of life | a statistical pattern of text |
| Checking of rules | within communal practice | within the distribution of training data |
| Body and action | working and promising together | no direct embodiment |
| Point of resemblance | finds meaning in use, not definition | finds meaning in use, not definition |
As the last row of the table shows, there is a clear overlap between the two. But as the three rows above tell us, overlap is not sameness. Not hastily filling in this subtle distance may be the most honest attitude of all.
One thing to add: the very question we pose in this comparison is Wittgensteinian. The question "does the machine understand meaning?" sounds as if understanding meaning were a single something, hidden somewhere, whose presence or absence we could check. But perhaps "understand" is itself a word with a family resemblance. A person understanding a poem, a child understanding a joke, a student understanding a proof — these resemble one another but are not the same. Then when we say a machine understands something, where in that web of resemblance should we place it? Before this question, Wittgenstein would have laid out slowly the various scenes in which the word is used rather than pronounced a verdict.
The Empty Seat of the Form of Life
Let us go a little deeper. For the later Wittgenstein, language always worked against the background of some "form of life." There is his famous remark that if a lion could speak, we would not understand it. The reason is not grammar but that we do not share a form of life with the lion. Stringing out the same words is not the same as playing the same game.
Holding this insight up to the LLM raises a weighty question. Because the model learns from human text, it can mimic human language games on the surface. But the form of life that is the background of those games — hunger and fear, the effort to keep a promise, grief at losing a loved one — the model does not live. Then are the sentences a model produces "participating" in our language game, or precisely reflecting the traces of that game? There is no easy answer to this. And that very difficulty tells us that Wittgenstein's concepts are still a living tool.
Yet room for rebuttal remains here too. Is there a reason to take "form of life" narrowly, as only human biological life? If the very interaction the model continually exchanges with humans constitutes a kind of new practice, might that practice too grow into a form of life? This is an open question Wittgenstein never answered, and that we have not answered either. What matters is that the moment we pose this question, we are already thinking on the road Wittgenstein paved.
Leaving It as an Open Question
In short, Wittgenstein's framework tilts neither readily toward praising the LLM as one that "understands meaning" nor toward dismissing it as "mere mimicry." Rather, his true gift lies not in answers but in sharpening the questions. If meaning is use, where does "use" end and "life" begin? Are a machine arranging words correctly and doing something with those words the same, or different? Before these questions, the honest attitude is careful inquiry rather than hasty verdicts. And this caution may be, perhaps, the most Wittgensteinian posture of all.
When we say "thank you" to a chatbot, what are we doing? When the machine replies "you're welcome," to what language game does that belong? Wittgenstein, rather than answering with an abstract definition, would have urged us to look slowly at the scene.
The Echo Wittgenstein Left
An Influence Across Two Eras
Wittgenstein's thought left its trace on both great streams of twentieth-century philosophy. The early Tractatus deeply influenced the logical positivists known as the Vienna Circle. What is interesting, though, is that even when they adopted the Tractatus as their manifesto, Wittgenstein himself felt they had misunderstood the book's heart, namely the importance of the unsayable.
The later Investigations gave great impetus to the stream called ordinary language philosophy. It is the attitude of resolving the problems of philosophy not by trying to crack them with grand theory but by looking carefully at how we actually use words. In discussions of mind, of meaning, and of rules, his shadow still stretches long today.
His influence did not stay within the fence of philosophy. From anthropology, linguistics, and cognitive science to the very debates around artificial intelligence, insights like "meaning lies in use," "rule-following is communal," and "concepts are bound by family resemblance" have been drawn up again and again. For a man who left only a single formally published book, the echo he left has spread astonishingly far, and long.
The only book he himself published is, in effect, the Tractatus alone. The rest of his vast thought reached us through lecture notes, students' transcriptions, and posthumously edited remains. In a sense, Wittgenstein's philosophy was itself preserved and transmitted through the cooperation of a community. It is a curious coincidence that his thought — which held meaning to be rooted in communal practice — is embodied even in the way his texts reach us.
Not Answers but a Method
Yet the greatest legacy Wittgenstein left may not be any particular doctrine. Rather, it is an "attitude." The attitude of not rushing straight to grand theory before an abstract big question, but first stopping to look at how the words used in that question actually work. The invitation, before piling up theory, to return to the everyday uses we already know.
This attitude is of use beyond philosophy, in our everyday thinking too. When some dispute fails to resolve and circles in place, the real problem is often that both sides are using one key word in different games. Quarrels like "is that really freedom?" or "is that genuine art?" are like this. Before fighting over a definition, Wittgenstein would have asked us to look together at what we actually do with the word.
His way of writing itself resembles this attitude. The later work does not proclaim conclusions. Instead it endlessly poses questions, gives examples, and refutes itself. The reader, rather than being handed a finished doctrine, is invited to think along. To some this looks like a frustrating incompleteness; to others, that is exactly his greatness. For he sought to show us not what to think but how to think.
Conclusion
Wittgenstein's two philosophies climb the same mountain of language by different paths. The early Wittgenstein, at the summit, drew a clear boundary line for what can be said, and counseled honest silence about the territory beyond it. The later Wittgenstein came back down the mountain to look at what people actually do with that language as they live. Meaning is not a jewel lodged in a word but shines only within the activity of people using words together — that was his second realization.
Whether one divides the two periods or joins them, one thing does not change. Wittgenstein makes us stop and look at the thing we have grown too used to see, namely the act of "speaking" itself. We speak every day, yet rarely ask how speech comes to have meaning. His philosophy cracks open precisely that taken-for-grantedness.
As a fish is not conscious of water, we live immersed in language yet rarely see language itself. Wittgenstein is the one who makes us conscious of that water once. Once we see it that way, we realize anew how intricate and mysterious a product of cooperation the plain words we exchange each day really are.
And this realization makes us a little more humble, and at the same time a little more full of wonder. Humble, because we come to admit that a region words cannot reach is always beside us. Full of wonder, because the fact that, with this imperfect tool, we nonetheless reach one another's minds, build a world together, and share love and grief, strikes us anew as astonishing.
And today, in an age when machines have begun to arrange words like humans, that crack asks us once more. What is it to speak, and what is it to know meaning? What Wittgenstein left us is not easy answers but a better way of asking.
If one spirit runs through his life and thought, it is intellectual honesty. He was candid enough to tear down even the conclusion of the book in which his conviction had been deepest. And though he drew the limits of language more sharply than anyone, he never belittled what lies beyond them. On the contrary, he believed the most precious things reside in that silent region.
The humility to speak clearly of what can be said, and to fall silent before what cannot. Only when we set aside the illusion that everything can be captured in words may we finally see the things that truly matter. Perhaps that is the most timely gift a difficult philosopher of a century ago offers our age, which overflows with words.
So next time you fall out with someone over the meaning of a word in conversation, why not pause and recall Wittgenstein? Are we using the same word in the same game, or throwing the same word in different games? That one small pause may, unexpectedly, head off many a fruitless quarrel. That philosophy lives not only in grand theory but can breathe like this in the midst of daily life — that is precisely what the later Wittgenstein wanted to show us.
Food for Thought
1. Is the counsel "be silent on what cannot be said" still valid for us today?
What are the things we ruin by trying to wring them out into words?
2. If you have ever failed in trying to pin a word's meaning down with a
"definition," was that not perhaps a case of family resemblance?
3. If meaning lies in "use," can we really "use" a word for something we have
never directly experienced? What about a machine?
4. If a wholly private language is impossible, how can we convey to one another
the inner feelings that only we ourselves know?
5. If "understand" is itself a word with a family resemblance, where should the
line between human understanding and machine processing be drawn? Is such a
line even necessary?
References
- Stanford Encyclopedia of Philosophy, "Ludwig Wittgenstein": https://plato.stanford.edu/entries/wittgenstein/
- Stanford Encyclopedia of Philosophy, "Private Language": https://plato.stanford.edu/entries/private-language/
- Encyclopaedia Britannica, "Ludwig Wittgenstein": https://www.britannica.com/biography/Ludwig-Wittgenstein
- Ludwig Wittgenstein, "Tractatus Logico-Philosophicus" (1921, first German edition; 1922 English parallel edition)
- Ludwig Wittgenstein, "Philosophical Investigations" (1953, published posthumously)
- Stanford Encyclopedia of Philosophy, "Wittgenstein's Logical Atomism": https://plato.stanford.edu/entries/wittgenstein-atomism/
현재 단락 (1/177)
There are philosophers who have contradicted themselves. But few have pushed a philosophy to its pea...