On Language and Definitions
May 1, 2026 | 3,039 words | 15min read
0. Introduction
Language is a fundamental part of human nature. Some would even say it is what differentiates us from animals. We use it every day: whether through reading, hearing, or speaking. We use it to create beautiful art such as stories and music, but it can also be used in destructive ways, to hurt or to demean.
Yet, even though it is so important, we understand very little about it and often make very basic mistakes when using it. In this article, I want to focus on one of these mistakes, show how it can happen, why it happens, and how we can avoid it in the future.
1. Shared Meaning
But in order to do that, we first need to understand certain things about language. First and foremost: what is the goal of language, and how can we achieve this goal? The goal of language, to put it plainly, is to communicate with each other. For this communication to be possible, it is essential that we share meanings for words. This means that when I speak a certain word, I evoke in you roughly the same images as the ones I have in my head; otherwise, communication becomes impossible.
For example, if I say, “Can you buy an apple for me?” and what I understand by “apple” is a round fruit that is red, slightly sour but also sweet, while the other person understands it as a yellow, sweet, long, and curved fruit (commonly known as a banana), then they will buy that fruit for me. I will be surprised and will not have what I wanted, communication has failed. The shared meaning we have for words is exactly what allows us to avoid this failure.
But with that, the question becomes: how does this shared meaning develop? How can we be sure that it develops properly and that our understanding of words does not diverge?
In practice, this shared meaning is already embedded in society, and as we grow up, it is slowly passed on to us. In this way, we become part of a web of shared meaning. This may still sound abstract, so how does this actually work?
As we grow, from toddlers, to children, to teenagers, to adults, we are constantly exposed to language. Whether through our parents speaking to us as toddlers, reading children’s books when we are young, or later through newspapers and everyday communication, the same process is always happening. Words appear in certain contexts, and initially we have no idea what they mean. But over time, as a word is repeated again and again in similar contexts, for example, the word “apple” appearing whenever a red, round fruit is present, we develop associations with that word.
These associations are roughly similar within a culture because we acquire them from our parents and the media. As we grow up, we participate in this shared web of meaning by communicating and by using words when the same associations arise. So the reason we have shared meaning is that we associate certain things with words, and these associations are, within a society, relatively homogeneous.
However, these associations are not universal; they differ from society to society. Each society has a slightly different understanding of words. This does not mean we cannot communicate across societies, we can use translation to map words from one language to another, but this translation will always be imperfect. Because of differences in history and culture, different societies associate different things with what may seem like the same concept.
To make this more concrete, consider the English word “home” and its German counterpart “Heimat.” Although they appear to have the same meaning, there are important differences. Because Germany was not unified for a long time, poets used the word “Heimat” to advocate for German unification, and this gave the word associations that do not exist in English, where this historical process did not occur.
2. The ‘What is’ Questions and the Problem of Systemization
Let’s switch tracks for a moment. In philosophy, we encounter many questions that are debated extensively, with philosophers taking vastly different, and often mutually exclusive, positions. Just to name a few: nominalism vs. realism, moral emotivism vs. moral realism, Platonic realism vs. idealism, and many more. All of these debates are framed around questions of the form “What is X?”, for example, what are categories? What is morality? What is the thing that holds the world together at its deepest level?
But how is it possible that we arrive at such mutually exclusive positions when we are using common words? Shouldn’t our shared meanings lead us to relatively similar conclusions?
The problem is as follows: in philosophy, when we ask a “what is” question, we often begin with an intuition that pulls us in a certain direction. From there, we try to systematize this intuition by constructing a theory, a set of rules that is internally coherent, i.e., free of contradictions. The problem is that we are trying to systematize something that cannot be fully systematized.
Why can we not develop complete systems for the meanings of words?
There are several reasons. First, humans do not understand words through strict systems or sets of rules. Rather, meaning arises from a loose set of associations. We are therefore trying to systematize something that is, by its nature, not fully systematizable. How do you turn a web of associations into a rigid system?
Second, as discussed earlier, these associations differ from society to society. Even if it were possible to construct such a system, it would have to vary across cultures.
Third, shared meaning is not static. It changes over time. In different historical periods, we have different understandings of words. For example, with the rise of modern medicine, our understanding of what it means to be “dead” has changed drastically. Nowadays, we often say someone is dead when they are brain dead, that is when no brain activity can be detected. In ancient times, people believed someone was dead when their “pneuma,” or breath of life, had left them; later, death was associated with the heart stopping. This is just one example of how meanings evolve.
In fact, within our society, there is a constant process of negotiation over the meanings of words. If I say, “My friend is tall,” and someone else replies, “He is only 1.70 m, that is not tall,” we are negotiating what “tall” means. Both people have different conceptions of the word and try to convince each other that their conception is correct. This process of renegotiation happens continuously, gradually shifting meanings over time.
A more modern example is the word “computer,” which used to refer to any calculating device, including very simple ones. Nowadays, most people think of a desktop or personal computer when they hear the word. This is a natural shift in meaning.
Fourth, and perhaps most importantly, our sets of associations often include contradictory elements. When people try to systematize the meaning of a word, they tend to select one element and ignore the rest. Around this single element, they then build an elaborate theory. But this leads to problems down the line: the theory is challenged, and a new, more complex theory must be constructed to address these issues. This process continues, with theories becoming more intricate and refined, until they begin to feel strange and disconnected from the world they were meant to describe.
3. Examples of Systemization
Let’s ground this a bit more. In particular, I want to focus on the fourth aspect and combine it with an explanation of the previously raised question: how is it possible that we arrive at contradictory philosophical positions when answering “what is” questions? To do this, I will look at two examples: time and morality.
3.1 Time
In his famous paper The Unreality of Time, McTaggart differentiates between two theories of time called the A-series and the B-series. The A-series describes statements like “I am eating an apple,” “In the future I will eat an apple,” and “In the past I ate an apple.” This series is, in other words, categorical: events are classified as present, past, or future. The B-series, by contrast, is relational. It includes statements like “9/11 happened after World War II” or “The Queen dies before the year 2026.” Here, time is described in terms of relations and events are ordered relative to one another using terms like “before” and “after.”
There is one important difference between the two: in the B-series, change is impossible. If one event is before another, it will always be before that event. In the A-series, however, change is possible: an event can be present and later become past.
McTaggart then makes the following argument: for time to exist, change is essential. Only the A-series allows for change, so the A-series is necessary for time. However, the A-series is contradictory. Therefore, time does not exist.
I do not want to elaborate on his argument in full detail here; if you are interested, you can read the original paper or my paper summary of it. But to give an intuition: a time event cannot simultaneously be in the present and in the past. Yet we can say, “I am now eating an apple,” and the next day, “Yesterday I ate an apple.” How do we explain this shift?
One way is to introduce a meta-time, a time about time. We move from “I am now eating an apple” to “Yesterday, I said ‘I am now eating an apple.’” But now change occurs at the level of this meta-time as well, which would require a meta-meta-time, and so on. This leads to an infinite regress. From this, McTaggart concludes that the concept of time is contradictory and therefore unreal.
But the argument itself is not what matters here. What matters is what just happened. Let’s take a step back. We tried to explain time by constructing two rivalry systems, the A-series and the B-series, each driven by a different intuition: that time is categorical and that time is relational. These intuitions pull us in different directions, and from each we build a theory. These theories then conflict or lead to contradictions. Each has its own problems (e.g., no change in the B-series, contradiction in the A-series), and to resolve these problems we must make our theories more complex or introduce additional assumptions.
This is exactly the process described earlier. We start with a set of associations tied to a word, extract particular intuitions, and then build systems around them. But the problem did not arise at the level of argumentation or conclusion, it arose at the very first step: we tried to systematize something that cannot be fully systematized.
3.2 Morality
Let me provide a second example, this time from morality, more precisely meta-ethics, the field that investigates questions like “What is morality?” or “What is its nature?”
There are many theories that attempt to answer this question. I will focus on two: emotivism and realism. Emotivism roughly states that moral judgments are not statements of fact but expressions of the speaker’s feelings. When we say, “You acted wrongly in stealing that money,” we are not stating a fact beyond “You stole that money.” Rather, we are expressing disapproval.
This contrasts sharply with moral realism, which holds that moral statements are statements of fact. According to this view, when I say “You stole that money,” I am not expressing an emotion but referring to a moral fact or rule that has been violated.
This difference has consequences. In emotivism, contradictory moral positions can coexist because people have different feelings. In realism, if two people make contradictory moral claims, at most one can be correct.
Both positions face serious problems. Emotivism encounters the Frege–Geach problem, which pushes it toward abandoning truth-based logic in favor of coherence-based approaches. Realism, on the other hand, struggles to explain how moral facts interact with the world, how we could possibly perceive or access them.
Once again, the question arises: how do we end up with such radically different positions?
The answer mirrors the case of time. What we understand by “morality,” like any word, is a set of associations. Different theorists select different intuitions from this set and build their theories around them, leaving others aside.
For example, someone might focus on the immediate emotional response we feel when seeing a homeless person being beaten, an instinctive sense that something is wrong. From this intuition, emotivism emerges: morality is rooted in emotion. Another person might focus on cases where we are unsure whether something is morally wrong and must deliberate before reaching a conclusion. This slower, reflective process seems more rational than emotional, leading to the idea that morality involves discovering truths, hence realism.
But just as in the case of time, building a system around a single intuition is a mistake. By doing so, we ignore other important aspects of our understanding. The resulting theories are incomplete, and this incompleteness generates problems that require increasingly strange solutions.
It is like trying to assemble a puzzle where the pieces do not quite fit. Instead of reconsidering the overall picture, we start cutting the pieces to force them into place.
4. The Game of Bitter Pill Swallowing and Constitution of Concepts
I like to call this “the game of bitter pill swallowing.” What we are doing, even though we claim to be searching for truth, is constructing worldviews that minimize the problems we personally find most troubling. It becomes a competition to adopt the theory that requires us to swallow the fewest “bitter pills”, whether that means abandoning truth for coherence in emotivism, or positing mysterious faculties to perceive moral facts in realism. To a certain extent, adopting different philosophical positions reveals more about ourselves than about the nature of reality; rather than accepting the position that is closest to the truth, we tend to accept the one that fits us best.
There is another way of understanding what is happening, besides the “set of associations” framework I have used so far. We can instead say that we have certain concepts: like time, morality (or identity), that are not singular, unified entities, but are constituted by many different phenomena.
Take time as an example. When we talk about time, we are actually referring to several distinct phenomena. One of them is change: we observe how things around us transform, leaves turn from green to brown and fall from trees, food spoils, objects decay. We group these observations under the concept of time. But there is also another phenomenon: memory. We can remember things we did in the past, like the apple we ate yesterday or playing in the sand as a child. This too is a phenomenon we associate with time, yet it is clearly distinct from change.
The same applies to morality. It is not a single, unified thing but is composed of multiple phenomena: the emotional response we feel when someone is unjustly harmed, the rational deliberation about whether something is right or wrong, and so on. These are different elements that together constitute what we call “morality.”
What we humans do is observe these different phenomena and group them together under one term. So concepts that appear unified are, in reality, composites of many distinct parts.
One way to avoid the problems with models discussed earlier is to stop focusing on these high-level terms like “time” or “morality” and instead focus on the underlying phenomena themselves, as suggested in the paper The Unimportance of Identity. Instead of speaking about time and building models around it, we speak about change and memory. Instead of speaking about morality, we speak about emotion and deliberation.
This has two advantages. First, it simplifies our models, we have fewer “bitter pills” to swallow. Second, we do not lose anything in doing so. Whenever we would talk about the broader concept, we can instead refer to the specific phenomenon that is relevant in that context. This makes our thinking more precise and less confusing.
5. The Role of Models in Understanding the World
So where does this leave us?
Models and definitions of words are not useless, far from it. They help us understand the world and make sense of it. But they are always only partial perspectives on a concept, never the full picture. They are lenses through which we view reality, imperfect representations that can never fully capture it, partly because our concepts themselves contain tensions and even contradictions that cannot be completely expressed in pure logical systems.
For this reason, we should aim to hold multiple models or perspectives of a concept in mind when investigating it. Each model provides a different viewpoint, and together they can give us a clearer picture. There is a well-known story that illustrates this:
A group of blind men heard that a strange animal, called an elephant, had been brought to the town, but none of them were aware of its shape and form. Out of curiosity, they said: “We must inspect and know it by touch, of which we are capable.” So they sought it out, and when they found it they groped about it. The first person, whose hand landed on the trunk, said, “This being is like a thick snake.” Another, whose hand reached its ear, said it seemed like a fan. As for another, whose hand was upon its leg, said the elephant is like a pillar, like a tree trunk. The blind man who placed his hand upon its side said the elephant “is like a wall.” Another who felt its tail described it as a rope. The last felt its tusk, stating the elephant is something hard, smooth, and like a spear.
Models, then, have their proper use and should not be abandoned. But we should also avoid focusing too narrowly on a single model. A healthy plurality of perspectives gives the best picture of reality. And if we remain aware of the limitations of models, as discussed in this article, they can be extremely valuable tools for understanding the world more clearly.
Source:
reply via email