The Word Is Not the Thing
Published:
A word can refer to a mountain.
The word is not the mountain.
A map can represent a city.
The map is not the city.
A photograph can show a person.
The photograph is not the person.
This distinction seems obvious. Yet many conceptual errors begin when representation and represented object are allowed to blur.
Korzybski’s Warning
Alfred Korzybski popularized the phrase:
the map is not the territory.
The same principle applies to language. A word is a representational tool. It selects, compresses, and organizes features of what it refers to. It never becomes identical with the thing.
Names
Consider the name:
“Venus.”
The word consists of letters. The planet consists of rock, atmosphere, and physical processes. The name and the planet have completely different properties.
Confusing them is a category mistake.
Properties Do Not Transfer
The word “elephant” is eight letters long.
An elephant is not eight letters long.
An elephant may weigh thousands of kilograms.
The word “elephant” does not.
Properties of symbols do not automatically transfer to referents.
Representation Is Selective
A representation never includes everything.
A subway map preserves station order and line connections. It may distort geographical distance, direction, and scale.
A good representation omits. Selection is part of usefulness.
Abstraction
Words usually represent categories rather than complete objects.
The word “tree” ignores enormous differences among individual trees.
Language compresses diversity into reusable concepts.
This is abstraction.
Categories Shape Perception
Once we learn a category, we begin noticing the world through it.
A botanist distinguishes plant types that a novice groups together. A musician hears harmonic distinctions another listener misses.
Representations guide attention. They do not simply mirror reality.
Labels Can Reify
Naming a pattern can make it feel like a separate thing.
We invent a label, then begin talking as if the label refers to a sharply bounded entity.
This is reification.
Concepts are useful. Their boundaries may still be conventional or fuzzy.
“Intelligence”
Take the word “intelligence.”
It seems to name one thing. But the term may bundle memory, reasoning, learning, planning, and social skill.
A noun can hide multidimensional structure.
Language can make continua look like objects.
“Species”
Biology provides another example.
“Species” is indispensable. But different species concepts emphasize reproductive isolation, ancestry, or morphology.
The category is scientifically useful without always having one perfectly sharp natural boundary.
Words Can Outlive Their Models
Scientific vocabulary can persist while underlying theories change.
“Atom” originally meant indivisible.
Modern atoms have internal structure.
The word survived. Its theoretical meaning changed.
Language is historically layered.
Same Thing, Different Words
A single object can be represented in many ways.
Morning Star.
Evening Star.
Venus.
Historically, people did not always know these referred to the same celestial body.
Different descriptions can converge on one referent.
Frege: Sense and Reference
Gottlob Frege distinguished roughly between:
- reference: what an expression refers to,
- sense: the mode in which it is presented.
“Morning Star” and “Evening Star” can have the same reference but different senses.
This explains why identity statements can be informative.
“Venus Is Venus”
This statement is trivial.
“Morning Star is the Evening Star” was once an astronomical discovery.
If meaning were only reference, the two statements would seem equivalent.
Sense matters.
Empty Reference
Some meaningful expressions may lack real-world referents.
“Unicorn.”
“Sherlock Holmes.”
“The present king of France.”
Language can remain meaningful even when reference is absent or complicated.
Meaning is not simply pointing.
Fictional Objects
Fiction creates stable representational structures.
We can truly say:
“Sherlock Holmes is a detective”
within the fictional framework.
Representation supports discourse about nonactual entities.
The relation between language and reality is flexible.
Models Are Not Reality
Scientific models are representations too.
A model may use point masses, frictionless planes, or ideal gases.
These objects need not exist literally.
Their value lies in controlled approximation.
The model is not the target system.
Equations
An equation can describe a physical regularity.
The symbols on paper are not the process itself.
The equation F = ma is not a force.
It represents a relationship among quantities under a theoretical framework.
Data Is Not the Phenomenon
A temperature reading is not temperature itself.
A brain scan is not thought itself.
A survey response is not belief itself.
Measurement produces representations.
Scientific reasoning must keep this distinction visible.
Digital Twins
Modern engineering creates detailed digital models of factories, aircraft, and cities.
A digital twin can predict behavior.
Still, the simulation is not the physical system.
Representation may become extremely rich without becoming identity.
Language Creates Distance
Representation lets us think about things that are absent, dangerous, impossible, or abstract.
This distance is one of language’s greatest powers.
A symbol lets us manipulate a model instead of the object itself.
But Distance Creates Risk
Because symbols are easier to manipulate than reality, we can mistake verbal coherence for truth.
A theory can sound elegant.
A label can sound explanatory.
Words can produce an illusion of understanding.
Naming is not explaining.
The Word “Cause”
Saying “X caused Y” does not by itself reveal mechanism.
The sentence may hide correlation, mediation, or confounding.
Language can compress causal claims faster than evidence supports them.
The Word “Natural”
Calling something “natural” can smuggle evaluation into description.
Natural may mean common, nonartificial, evolved, or healthy.
The label can obscure distinctions.
Conceptual hygiene requires asking what the word actually means in context.
Linguistic Categories and Reality
Do categories reflect structures in the world?
Often, yes.
But not perfectly.
Some categories track robust natural regularities. Others are strongly shaped by human purpose.
Representation is constrained by reality and constructed through use.
Representation Without Identity
The core principle is:
a representation can be accurate, useful, and causally powerful without being identical to what it represents.
This matters across language, science, computation, and cognition.
Why This Matters for Self-Reference
Once symbols themselves become objects of discourse, the distinction becomes even more important.
We need to know whether a sentence is talking about snow, talking about the word “snow,” or talking about its own truth.
Each shift changes logical level.
The Philosophical Lesson
Words do not merely mirror the world.
They carve, group, emphasize, and relate.
Their power comes from standing apart from what they represent.
Without that distance, representation would be impossible.
The Next Question
How do we keep track of these levels formally?
We need one language to talk about objects, and sometimes another level of language to talk about the first language.
That leads to:
object language and metalanguage.