What Can We Really Do With Words?

You’re at the dinner table. Your younger brother reaches for the last piece of pizza, and you say, “I call it.” Did you do something by saying that? Sure, you made sounds. But did you actually claim the pizza? Something weird is going on here. The words “I call it” aren’t describing anything—they’re not telling us about the world the way “The pizza is pepperoni” does. They’re changing the situation. After you say them, your brother knows he can’t take that slice without starting a fight.

Now think about this: what if all our sentences work a bit like that? What if saying something is never just describing the world, but always also doing something? That’s the kind of question that fascinated a British philosopher named J.L. Austin, who died in 1960 but left behind ideas that still bother and delight philosophers today.

The Problem with Pretending Words Are Just Labels

Here’s a strange thing philosophers noticed: we tend to imagine that language works like a labeling machine. We have words, we have things in the world, and when the words match the things, the sentence is true. When they don’t, it’s false. Simple, right?

Austin thought this was deeply wrong—not just a little wrong, but the kind of wrong that sends philosophers down dead ends for decades.

The problem is that sentences don’t carry their truth or falsity around with them like a nametag. Take the sentence “France is hexagonal.” Is that true or false? Well, if you’re a general planning an invasion and you just need a rough shape, it’s good enough. But if you’re a cartographer drawing a precise map? It’s not just false—it’s the wrong kind of thing entirely. It’s a rough description, and saying it’s “true” or “false” misses the point.

Austin pointed out that we judge statements in all sorts of ways that don’t reduce to true/false. A description can be exaggerated, vague, misleading, too concise, roughly right, or fair enough for the purpose. When someone says “Belfast is north of London,” is that true? Well, it depends on how precise you need to be. For some conversations, it’s fine. For others, you’d want to say “well, technically it’s northwest, and the exact relationship depends on which parts of each city you’re measuring from.”

This isn’t nitpicking. Austin thought that philosophers, in their hurry to answer big questions, often ignore these nuances. They grab onto a simple model of how language works and then build elaborate theories on top of it—theories that collapse when you notice that real people don’t use words that way.

How to Do Things With Words

Austin’s most famous discovery came when he noticed a whole category of sentences that don’t even try to be true or false. He called these performative utterances—sentences that are actions.

Consider these:

“I name this ship the Queen Elizabeth” (said while smashing a bottle against the hull)
“I bet you sixpence it will rain tomorrow”
“I take this person to be my lawfully wedded spouse” (said during a wedding ceremony)

When you say these things in the right circumstances, you’re not describing something you’re doing. You’re doing it. The words themselves are the action. Nobody who says “I name this ship” is reporting on a naming that’s happening somewhere else. They’re performing the naming right there.

Austin noticed something else about these sentences: they can go wrong in ways that have nothing to do with being false. A performative can be unhappy or infelicitous. If you try to marry someone when you’re already married, the ceremony might be a misfire—the act doesn’t actually happen. If you say “I promise to pay you back” but you have no intention of doing so, that’s an abuse—the act happens, but it’s hollow or insincere.

So right away, Austin had two different ways sentences can fail: they can be false (like saying the cat is on the mat when it isn’t), or they can be unhappy (like trying to name a ship when you’re not authorized to do so).

But here’s where it gets interesting. Austin started trying to separate sentences into two neat piles: constatives (which are true or false) and performatives (which are happy or unhappy). And he failed. Every attempt to draw the line kept breaking down.

Why? Because the same sentence can be used to do different things on different occasions. And, more surprisingly, even apparently pure statements of fact can be assessed for happiness or unhappiness. If you say “The cat is on the mat” when you don’t actually believe it, that seems like a kind of abuse—an insincere statement. And performatives can be assessed for correctness relative to the facts: if you warn someone that the bull is about to charge, but the bull isn’t actually about to charge, your warning is mistaken, not just unhappy.

The distinction between describing and doing, Austin concluded, isn’t built into sentences themselves. It depends on how we use them, in what circumstances, for what purposes.

Three Kinds of Acts

From the wreckage of his failed distinction, Austin built something more powerful: a way of analyzing all the different things we do when we speak.

When you say something, you’re performing at least three different kinds of acts at once:

First, there’s the locutionary act: simply producing a meaningful sentence. If you say “I’ll be home for dinner,” you’ve produced words with a certain meaning. That’s the locutionary act.

Second, there’s the illocutionary act: what you’re doing by saying those words with a certain force. You might be promising to be home, or stating that you’ll be home, or warning that you’ll be home. Same words, different acts. The illocutionary act is where the real action is—it’s the point where language and action meet.

Third, there’s the perlocutionary act: the effects your words have on your audience. You might persuade them to wait dinner, or alarm them, or reassure them. These effects happen because of what you said, but they’re not the same as what you said.

This might sound like academic hair-splitting, but it matters. Philosophers had tended to think that the most important thing about language was whether sentences were true or false. Austin was saying: no, the most important thing is what people do with sentences. Truth and falsity are just one way of evaluating what we do—and not always the most interesting way.

What Does “I Know” Actually Do?

Let’s apply this way of thinking to a concept that’s crucial in philosophy: knowledge.

Normally, philosophers treat “I know that such-and-such” as a description of a mental state. You have a belief, it’s true, and you have good reasons for it—that’s what knowledge is. When you say “I know,” you’re reporting that you have this special kind of justified true belief.

Austin thought this was backwards. He noticed that saying “I know” behaves a lot like saying “I promise.” When you say “I promise to pay you back,” you’re not describing a promise that exists somewhere else. You’re making the promise. The words are the act.

Similarly, Austin suggested, when you say “I know the train arrives at 3pm,” you’re not describing some inner state. You’re giving your word that the train arrives at 3pm. You’re offering a guarantee. You’re making yourself responsible for that information.

This explains some strange features of how we use “know.” If someone says “I know it’s raining, but I might be wrong,” that sounds contradictory. Why? Because if you’re giving your word, you can’t simultaneously say “but my word might be false.” And if someone claims to know something, you can ask “How do you know?”—you’re asking for their credentials, their basis for giving this guarantee. You don’t ask “How do you believe?” You ask “Why do you believe?” The difference matters.

This doesn’t mean Austin thought knowledge wasn’t real—he wasn’t saying knowledge is just a social performance. He was saying that our talk about knowledge is a kind of action, and that philosophers had been misled by assuming it was just description.

How Knowing Actually Works

Austin also had interesting things to say about what knowledge is, beyond how we talk about it. He thought knowing was a basic way of grasping how things are—not a combination of belief plus extra conditions.

Here’s a key point: Austin thought our ability to know things depends on two things working together: acumen (our judgmental skill) and opportunity (the circumstances that give us access to the facts). Both are needed, and both are imperfect.

Our judgmental capacities are inherently fallible. We can always make mistakes. But—and this is crucial—that doesn’t mean we can never know anything. A good machine is still liable to break down sometimes, but good machines don’t break down often. Similarly, our minds and senses are fallible, but they work well enough in the right circumstances to give us genuine knowledge.

This puts Austin against a long philosophical tradition that says: if you can possibly be wrong, you don’t really know. The tradition says we should only count something as knowledge if it’s absolutely certain—if it’s impossible to be mistaken. Austin thought this was a recipe for concluding that we know nothing at all, which is obviously false.

If you watch a pig in good light, at close range, for several minutes, and you say “That’s a pig,” you know it’s a pig. Could you be wrong? Maybe if it’s a really convincing robot pig. But that’s not how we actually talk about knowledge. “Enough is enough,” Austin wrote. “It doesn’t mean everything.” You don’t have to rule out every bizarre possibility to know something. You just have to have enough evidence for the purposes at hand.

The Problem of Freedom

One more area where Austin’s approach bears fruit: the question of whether we act freely.

Philosophers have argued for centuries about whether human actions are free or determined—whether we genuinely could have done otherwise, or whether everything we do is forced by prior causes. Austin’s response was characteristically sideways: instead of tackling the big abstract question directly, he looked at how we actually talk about actions and excuses.

He noticed something odd. We have a rich vocabulary for describing ways actions can go wrong or be less than fully free: accidentally, unintentionally, inadvertently, involuntarily, by mistake, under duress, without realizing, while sleepwalking, and so on. These are excuses—they’re things we say to take partial responsibility off ourselves.

But here’s the thing: we only need excuses when something is out of the ordinary. If you sit down in a chair in the normal way, nobody asks whether you sat voluntarily or involuntarily. That question only arises when something is weird—if you were hypnotized, or forced, or didn’t realize the chair was there.

Austin’s slogan was: No modification without aberration. You don’t need to add “voluntarily” to normal actions. The normal case already counts as free and responsible. The job of words like “voluntarily” and “involuntarily” is to mark departures from the normal case.

This matters because many philosophers had assumed that every action raises the question “Was it done freely or not?” and that we need a general theory to answer that question for all actions. Austin thought this was like assuming that every statement raises the question “Is it true or false?”—it ignores all the nuance and context that actually matter.

What It All Adds Up To

Austin wasn’t trying to build a grand philosophical system. He was trying to get philosophers to pay attention to the fine details of how we actually use language, rather than imposing simple theories that ignore those details.

His big insight was this: language is a form of action, not just a way of describing the world. When we speak, we’re doing things—making promises, giving warnings, claiming knowledge, offering excuses. And to understand what we’re doing, we need to look at the specific circumstances, the specific words, the specific intents and purposes of the people involved.

This doesn’t make philosophy easier. In some ways, it makes it harder, because you can’t just apply a simple formula to decide whether a statement is true or an action is free. But Austin thought that was the whole point: the world is richer than our theories, and our job is to do justice to that richness, not to ignore it.

Appendix: Key Terms

Term	What it does in this debate
Performative utterance	A sentence that doesn’t describe something but does something—like promising, betting, or naming
Constative utterance	A sentence that does aim to describe something and can be true or false
Felicity / Happiness	The way performatives are judged—whether they succeed or fail, not whether they’re true or false
Illocutionary act	What you’re doing by saying something (promising, warning, stating) with a particular force
Locutionary act	Simply producing a meaningful sentence
Perlocutionary act	The effects your words have on listeners (persuading, alarming, reassuring)
Excuse	A way of saying an action wasn’t fully free or responsible—accidentally, unintentionally, by mistake
No modification without aberration	The idea that we only need special labels (“voluntarily,” “intentionally”) when something is unusual; normal actions don’t need them

Appendix: Key People

J.L. Austin (1911–1960): A British philosopher who served in military intelligence during World War II and later became a professor at Oxford. He argued that philosophers should pay careful attention to ordinary language rather than building abstract theories that ignore how people actually talk.
P.F. Strawson (1919–2006): A philosopher who debated with Austin about truth. Strawson thought “is true” was just a way of agreeing with a statement, not describing it as having a special property.

Appendix: Things to Think About

Austin said that “I know” works like “I promise”—it’s giving your word, not describing a mental state. But if someone says “I know” and they’re wrong, what’s the difference between being wrong about a fact and breaking a promise? Are they equally blameworthy?
If truth depends on the purpose of the conversation, as Austin suggests, then can the same sentence be true in one conversation and false in another? What would that mean for things like scientific facts, which we usually think are true regardless of who’s talking?
Austin says normal actions don’t need to be called “voluntary”—that label is only for unusual cases. But if everything is determined by prior causes, is any action “normal”? Or does determinism make every action potentially weird?
Think about a time you said “I’m sorry.” Were you describing a feeling you had, or were you doing something—making an apology? Can you apologize without meaning it? Does that make the apology “unhappy” in Austin’s sense?

Appendix: Where This Shows Up

Everyday arguments: When someone says “That’s not what I said” or “You’re taking that out of context,” they’re using Austin’s insight that meaning depends on circumstances and purposes, not just words.
Legal language: Courts spend enormous energy figuring out what a law or contract “does” with its words—whether it’s a promise, a threat, a warning, or a command.
Artificial intelligence: When we try to build AI that understands language, one of the hardest problems is getting computers to grasp that sentences perform actions, not just convey information. An AI that can pass a true/false test might still have no idea what you’re doing when you speak.
Social media: The phrase “I’m not saying X, I’m just saying Y” is a classic attempt to control what illocutionary act you’re performing—to insist you’re just stating, not accusing or judging. Austin would say that what you’re doing depends on context, not just your words.