When a baby is learning a language some words must be easier to grasp. You show them a banana and say "This is a banana." You show them a train and say "This is a train." But how can we really be sure that the baby grasps the right meaning for a word that does not refer to some very specific object or action, like "coherence"?

The only way we can talk about the meaning of these words is by using other words; but it seems to me that this necessitates the semantics of these words to be "normalized" in the mind of the baby: when you give a definition for an abstract word, it is going to consist of words which the baby has to understand individually before they can understand the definition. But this is an issue because I do not think anyone has proved that we can define all the abstract words in a non-circular fashion starting from the concrete words.

What the baby could observe if they do not understand all the words is the interrelationships between these different words and then it could find concepts which seem to have the same interrelationship.

But then how do we know the baby is going to have the same semantic understanding as us? For instance, if we were to define "lazy" as the opposite of hard-working and the baby knew neither of those words then it might think that "lazy" means shy and "hard-working" means gregarious. This is not the best example because it is very likely that when the baby will have conversations with other people using these words they eventualy will not parse; but my point is there might be a different way to interpret the words such that the conversations will parse to all the parties involved yet they will actually mean different things to all of them.

  • 8
    Ultimately, we don’t. We know that we all refer to (roughly) the same part of the colour spectrum as ‘blue’, but I have no idea if the image your brain creates from your eyes’ input actually matches the image that my brain creates for me. Perhaps the actual cognitive reality that you perceive as blue corresponds to the cognitive reality I perceive as red. No way of knowing, as far as I’m aware. With most words, though, such discrepancies would become clear fairly quickly. Commented Apr 28, 2020 at 0:35
  • @JanusBahsJacquet, the bigger problem is that we can't actually describe for an individual what the immediate input of the retina is to the first brain-like structure (and the retina pre-compiles a fair amount of stuff that is computed by the brain, in the case of sound). If you are vague about what I do, you are even more vague about how we differ in processing inputs.
    – user6726
    Commented Apr 28, 2020 at 1:47
  • @JanusBahsJacquet This is the first time I hear this "colour perception" problem from someone else. The idea occurred to me in my early twenties and I even used 'blue' and 'red' when explaining it to my friend :))
    – tum_
    Commented Apr 28, 2020 at 5:58
  • 1
    Colour is a clear example of when this isn't true because of colour blindness. We do perceive colours differently. But that's a small issue, and if we didn't generally have a share understanding of most concepts language just wouldn't work as well as it does.
    – curiousdannii
    Commented Apr 28, 2020 at 7:35
  • I would recommend 'On Truth and Lies in a Nonmoral sense', an essay by Nietzsche, specially the bit where he goes on about "metaphors".
    – user22430
    Commented Apr 28, 2020 at 12:00

4 Answers 4


We know that words have roughly the same meanings to most members of a language community because we're able to have conversations using them and not become totally confused. You can also flip it around: a language community is a collection of people defined by the fact that they share similar meanings of words.

The correlations aren't necessarily exact, there can be nuances of meaning that some people express that aren't picked up by others. But the meanings are close enough that we still understand each other.

How do we learn these meanings? Much the same way we learn other things about the world. Neural networks (i.e. brains) are remarkable designs that are able to learn patterns of significant complexity, given enough time and examples. Babies start out learning mostly concrete concepts, but as children mature they start to think more abstractly, and notice that these concepts correlate to words that other people use.

Consider a concept like "happy". A baby probably feels this very often, such as when their mother feeds them or hugs them, but they don't have a word for it. But they'll notice that people often use the word in these situations -- "Are you happy now?", "You seem happy". And as they learn to recognize behavior of others, they'll also notice that people say "I'm happy" when they're smiling, and recognize that this is similar to their own behavior. This reinforces the relationships between the word "happy" and all the feelings, sensations, and behaviors that we consider part of this concept.

When you get older, the process becomes easier. You've learned more about the world and what people talk about, so when you hear a new word you usually understand the context, and can infer likely meanings. If you're unsure, you can ask for clarification or consult a dictionary.

There are also feedback effects. When you use a word similarly to other people, your conversation flows smoothly, and you know that you have the proper understanding of the word. But if you have a misunderstanding you'll get confused reactions from other people, and they might even correct you, and you'll correct your understanding. Our social instincts gear us towards figuring out how to get along with others, and speaking the same language is fundamental to this (children on the autism spectrum are slower to learn language because they have trouble recognizing these social cues, or just don't care).

Even your original premise seems somewhat flawed. Parents don't generally tell their children "This is a banana." They just use the word "banana" more often when there are bananas around, and the child picks up on this correlation. However, these associations tend to be more immediate and obvious, so the child picks them up quickly.

  • regarding your first paragraph: is the fact that we don't become confused actually enough to conclude that the words mean the same thing to us? Consider the following example (somewhat overused and not entirely accurate but it sends the message across). An American asks a Brit: "Do you enjoy smoking fags?" The Brit replies: "Very much so." They were not confused yet they might have understood different things.
    – user28572
    Commented Apr 28, 2020 at 16:15
  • A single question and answer is not a conversation. It shouldn't take long to realize you're not talking about the same thing. Contradictions are possible, but they're usually contrived, not natural -- you can find them in comedy routines like "Who's on First". And what I'm really talking about is a lifetime of conversations.
    – Barmar
    Commented Apr 28, 2020 at 16:44
  • I do agree with you on some informal level but what I mean is there is not a rigorous proof either way (that a lifetime of conversations is sufficient to rule out the contradictions). I subjectively believe it is sufficient. You seem to believe it as well.
    – user28572
    Commented Apr 28, 2020 at 18:32
  • I'm not s linguist, just a layman who has read lots of popular science books on the brain, evolution, and language. I don't have rigorous proof of this, but it's what I've learned from these sources.
    – Barmar
    Commented Apr 28, 2020 at 18:36

I should point out that babies don't know the word "coherence" or "epistemology", and children (and adults) often know a phonological form without having mastered the standard dictionary definition. To take some painfully prevalent examples from linguistics, the abstract terms "phoneme" and "markedness" do not mean the same thing to most linguists (and mean virtually nothing to non-linguists).

An approximate meaning is induced from correlating usage facts with occurrence of a word, so that a child can tell from behavior of others that being lazy is not good, and that it has something to do with not doing things that one is expected to do. The lack of correlation between surrounding circumstances and the word "lazy" makes it highly improbable that a child would arrive at the meaning of "shy" for the word "lazy". The word "rude" doesn't clearly correlate with particular perceivable facts, instead it involves inferences about people's feelings about actions – a higher-order abstraction. OTOH they can learn specific instances of "rude" when they get their mouths washed out with soap.

In other words, you may have to look it up in a dictionary to get more than a rough understanding of the semantics of some abstract terms.


We don't always have the same definition of abstract words. This is an outstanding issue in all communication.

Examples abound. Plato's Republic has a discussion of the definition of “justice”. Some thousands of years later it is very difficult to say what “knowledge” is. Mortimer Adler in his book How To Read a Book suggests that the first thing in reading a book carefully is to come to terms with the author, i.e., to understand how s/he is using particular words. C.S. Lewis complained about the meanings of words being watered down in Mere Christianity:

The word gentleman originally meant something recognizable; one who had a coat of arms and some landed property. When you called someone a “gentleman” you were not paying him a compliment, but merely stating a fact. If you said he was not a “gentleman” you were not insulting him, but giving information. There was no contradiction in saying that John was a liar and a gentleman. … But then came people who said-so rightly, charitably, spiritually, sensitively, so anything but usefully-“Ah, but surely the important thing about a gentleman is not the coat of arms and the land, but the behavior? Surely he is the true gentleman who behaves as a gentleman should? … When a word ceases to be a term of description and becomes merely a term of praise, it no longer tells you facts about the object; it only tells you about the speaker’s attitude to that object. … As a result, gentleman is now a useless word.

From a linguistic standpoint, I would suggest that the Lewis quote offers the reason that we're not constantly having discussions about the meanings of terms: in all but the most careful conversations, most people are simply conveying attitudes with words, rather than making substantive statements (synthetic statements). (Contra Lewis, “gentleman” is not a useless word: it conveys a very clear positive evaluation of someone. It's just not precise any longer.)

To extend your examples somewhat: perhaps there is a denotative difference between being “frugal” and being “cheap,” but there is certainly a connotative difference, and that is what people are trying to get at when they use the words.


We know for a fact that words do not mean the same to all of us.

This is true for all words, abstract and concrete.

  • Can you elaborate on the existence of different meanings for words that describe concrete objects? I don't know what meaning "banana" could take, for example, other than the yellow fruit with a peel. Commented Apr 28, 2020 at 14:37
  • How do you know this? What's your proof? The NSM project would counter that some words are universals.
    – curiousdannii
    Commented Apr 28, 2020 at 14:43
  • I didn't downvote you but I have a couple of observations to make: "We know for a fact..." - saying it isn't enough, you should present some evidence to backup your claim. "that words do not mean the same to all of us" - all of them? Or some? Examples? "This is true for all words" - Ah all, then. That's a bold statement. All I need to falsify it is ONE example of a word that means the same to everyone. Like "taxi" or "chocolate". Find me one instance where taxi doesn't mean "a car with a driver who you pay to take you somewhere".
    – Alenanno
    Commented Apr 28, 2020 at 17:28
  • @NuclearWang To play along with your example, a plantain (which is itself describes different things to different people) may or may not be a banana to different people. And I don't think this is a fluke, I really fail to see a single word for which we cannot find such ambiguities. Commented Apr 28, 2020 at 17:55
  • 1
    @SriotchilismO'Zaic Interesting counterpoint, I see that some people might have wider/narrower definitions of certain terms. But for highly technical, well-specified terms, that band of interpretation becomes increasingly narrow, to the point where there's effectively no variation in meaning - I'm thinking of terms like hemoglobin or twenty-seven or sphere. If you refer to a banana as Musa acuminata, there's no interpretation that could include plantains. Commented Apr 28, 2020 at 18:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.