What a Word Means vs. Why It Means That
An anthropologist walks into a kitchen
Picture this: you land on a small island to study how a community eats. Nothing flashy — just everyday meals. You notice straight away that everyone uses their spoon only for soup, never for the main course. No one puts a fork in their right hand. The rules are as strict as traffic lights, but nobody ever sits you down to explain them.
Now you have two jobs.
First, you could simply write down the rules. “Soup spoon on the outside, fork in the left hand. Meat is cut only after it’s on the plate.” You’d be building a kind of dictionary of table manners — a complete list of which actions count as polite. That’s one investigation.
Second, you could ask a deeper question: Why this set of rules? What about the history, habits, or beliefs of these people makes these rules the ones that govern them? Why not put the fork in the right hand, or eat soup with a regular spoon?
When we talk about language, exactly the same pair of questions shows up — and keeping them apart is one of the most useful moves in philosophy.
What a word means, and what makes it mean that
Take any ordinary word — say “dog.” If I ask, “What does ‘dog’ mean?”, you’d probably answer: it means a furry, four-legged animal that barks. You’ve given me a semantic fact. A semantic theory of a language is exactly that: a big, systematic description of what all its words and sentences mean. Think of it as the world’s most complete dictionary, but one that also explains how the meanings of little pieces — like “bark” and “the” — combine to give whole sentences their meaning.
Now I ask you a scarier question: “In virtue of what facts does the sound ‘dog’ mean that animal?” In other words, what is it about you, your community, and the history of English that makes “dog” carry that particular meaning rather than meaning “cat,” or nothing at all? That’s a metasemantic question. A metasemantic theory doesn’t list meanings; it explains what makes those meanings belong to those words. It’s an investigation into the foundations of meaning — the hidden machinery that makes a string of noises count as language.
Both are real questions. Both are hard. But they are not the same question.
Spoons and sentences: the table-manners analogy
Let’s return to the anthropologist at the dinner table. Her first job — listing which actions are polite and which are rude — is pure semantics for table manners. She’s saying, “In this community, holding a fork in your left hand while cutting belongs to the category correct.” She never asks why that category exists or how it got its power. She just maps the landscape.
Her second job — explaining why these rules govern this group — is the analogue of metasemantics. She might say, “They follow these rules because the elders taught them, and there’s a quiet agreement that guests will be trusted only if they eat like this.” Now she’s not cataloguing the rules; she’s uncovering the social facts that make those rules the real ones.
One anthropologist might be brilliant at the first job and hopeless at the second. Another might switch between them. But nobody would confuse “What are the rules?” with “Why are these the rules?” They are simply different assignments.
Languages work the same way. I can study the semantics of English without ever worrying about what makes “apple” mean 🍎 — I can just record the connection. And I can ask metasemantic questions (“is it because of our mental images? our ancestors’ choices? a chain of baptisms?”) without writing a dictionary.
David Lewis draws a line
The philosopher David Lewis (1941–2001) put the point bluntly: “Only confusion comes of mixing these two topics.” He noticed that people use the phrase “theory of meaning” to talk about both jobs, and that this one label hides a crucial fork in the road.
Lewis said: imagine you want to study a possible language — maybe English, maybe a made‑up code. First, you could treat that language as an abstract system. You’d just write down which symbols go with which worldly things: the symbol “dog” points to dogs, “red” points to red things, “and” points to a way of combining true sentences. That’s a semantic description — it’s like a map of the language, without any mention of flesh‑and‑blood speakers.
Second, you could look at a real human population that uses that system. You’d ask: what psychological or sociological facts make this particular system their language? This second question is metasemantics. It doesn’t change what “dog” means, but it tells you what makes it stick.
If you try to answer the first question by pointing to the second — say, by arguing that “dog” means furry animal because your parents told you so — you’re mixing up the columns. Your parents are part of the story of how the meaning got there, not what the meaning is. The dictionary doesn’t care about your biography.
When the lines get blurry
It’s easy to slip. Suppose two friends argue about the word “justice.” One says, “Justice means fairness.” The other says, “No, justice is what our society has built over centuries of struggle.” The first friend is making a semantic claim — a claim about the meaning of the word. The second is making a metasemantic claim — a claim about the historical facts that gave the word its meaning. Both might be partly right, but they’re not answering the same question, and a shouting match that doesn’t notice the difference goes nowhere.
Even scientists can stumble. Some psychologists study how children learn the meanings of words. That’s a metasemantic project: they want to know what goes on in the brain and the community that makes “mama” mean that person. But if they then say, “A word just is whatever a child associates with it,” they’ve wandered into a semantic theory — and a controversial one at that. Keeping the two projects distinct lets us use insights from each without blurring them into confusion.
So why does this matter to you, right now, reading this on a screen? Because every time you ask, “What does this emoji mean?” you’re doing semantics. And every time you wonder, “How did a smiley face come to mean happiness instead of, say, sarcasm?” you’re doing metasemantics. Both are cool. But knowing which door you’re opening keeps you from walking into walls.
Think about it
- The word “lit” once meant “having light.” Now it means “exciting or excellent.” Is that change a semantic fact, a metasemantic fact, or both? If you and your grandparent disagree about what “lit” means, are you really disagreeing about the same thing?
- Imagine a secret handshake that only your friend group uses. What would a “semantic description” of the handshake look like? And what would a “metasemantic explanation” of why that handshake has its meaning look like?
- If every human vanished tomorrow, would the word “dog” still mean dogs? Why or why not? (Hint: does your answer depend on which of the two questions you’re asking?)
