Metaphor
A Socratic walk-through of metaphor — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why do we explain almost every abstract idea in physical terms?
Notice what just happened in that heading. A question is something we started with, as though it were a place. We will go over the argument, arrive at a conclusion, and hope nobody shoots it down. None of that is poetry; it is the plainest available English. So here is the oddity: when we talk about things with no shape, no location and no temperature — ideas, time, importance, disagreement — we reach almost helplessly for words about shape, location and temperature. Is that a habit of style, or is something structural going on?
Reasoning it through
REASONING #Test whether it is decoration by trying to remove it. Say what you mean by "a deep argument" without the depth. "Substantial"? That is weight. "Weighty"? Worse. "Profound"? Latin for deep. The physical vocabulary is not sitting on top of a literal version we could fall back on. There is often no literal version.
Next, check whether the physical words come in random assortment or in families. Take how we speak about disagreement: your position is indefensible, you attack a weak point, I shoot down your claim, I defend mine, you gain ground, someone wins. Every one of those comes from armed conflict, and they interlock — if arguments have positions and attacks, they can also have retreats, and you will understand "she withdrew that claim before I could get to it" instantly, though you may never have heard it. That is the crucial observation. It is not a list of frozen idioms; it is a mapping that generates new expressions and lets you infer with them. This is what Lakoff and Johnson called conceptual metaphor: not a figure of speech, but a systematic correspondence between one domain and another, of which the individual phrases are the visible residue.
Now ask which way the correspondences run. We describe time in the vocabulary of space — a long meeting, the week ahead, moving a deadline back, looking forward to Friday. Do we describe space in the vocabulary of time? Barely, and where we do it is by measuring travel rather than by importing time's structure. The same asymmetry holds elsewhere: quantity is described as height (prices rise, output is down) far more readily than height is described as quantity; affection is described as warmth more readily than warmth as affection. The traffic runs consistently from the concrete and directly experienced toward the abstract.
Why would that be? Consider what an abstract domain lacks. It has no edges, no parts you can point at, no directions. There is nothing to reason with. A concrete domain has all of that, plus a lifetime of practice reasoning about it. Borrowing its structure gives you inferences for free: if progress is motion along a path, then obstacles, detours, distance covered, and being stuck all become available, and each one means something about the abstract thing. The metaphor is not describing the abstract domain so much as lending it a skeleton.
That is also why a metaphor is never innocent. A skeleton fits some things and not others. If argument is war, disagreement has sides and a winner, and a conversation in which both people ended up understanding more is hard to score. Nothing about disagreement requires that reading; the mapping supplied it.
The analogy
THE ANALOGY #Imagine navigating unfamiliar country with a map drawn of somewhere else that happens to share its coastline. It is enormously better than no map: it gives you a grid, distances, names for places you had no names for, and it lets you reason about ground you have never walked. It also quietly hands you rivers and roads that are not there, and you will not notice which is which until you fall into one.
A traveller can always look up from the map and compare it with the terrain, whereas for a genuinely abstract domain there is no unmediated look available — the metaphor is not a rough substitute for direct access, it is much of the access we have.
Clarifying the model
THE MODEL #Two clarifications keep this from being overclaimed. First, "systematic" does not mean total. Mappings are partial by nature: time borrows space's ordering and extent but not its two-way traffic, which is why the future can be ahead of us without our being able to walk to it. Any mapping highlights part of the target and hides the rest, and different languages and communities partition it differently — Mandarin speakers use a vertical axis for time alongside the horizontal one, and some languages place the past in front, as the known, rather than behind.
Second, the strongest form of the theory is not settled. It is one thing to say the language is systematically structured this way — that is straightforwardly observable in the data. It is a stronger claim that speakers actively run the mapping in real time when they use a conventional phrase, and stronger still that abstract thought is grounded in bodily simulation. That embodiment literature has had mixed experimental support: several of the neat demonstrations, particularly the priming studies linking physical warmth or cleanliness to social and moral judgement, have replicated poorly. So hold the linguistic finding firmly and the cognitive mechanism more loosely. The systematicity of the mappings is not in doubt; how much of it is live in the head at the moment of speaking is.
A picture of it
THE PICTURE #How to readThe two boxes on the left are concrete domains we know from direct experience; the two on the right are abstract ones we have no direct experience of. Read each arrow as the borrowing, and then read across the member lists, line by line: traveller becomes person, path becomes career, destination becomes goal, obstacles become setbacks. The point of the picture is that the correspondence is element-by-element rather than a single loose resemblance — and that the arrows only ever point rightwards, which is the asymmetry the whole account rests on.
What became clearer
WHAT CLEARED #We describe abstractions physically not because we are being colourful but because abstractions come without structure, and a concrete domain lends them one — systematically, element by element, so that the inferences come along with the words. That borrowing is productive, which is how you understand a phrase you have never heard, and it is selective, which is how a metaphor can smuggle in a whole way of scoring a situation before anyone argues for it. What remains genuinely open is how much of the mapping is running in the mind at the moment of use, as against how much is simply frozen into the language.
Where to go next
ONWARD #- How a shift of metaphor changes what counts as a solution — crime as a disease against crime as a beast.
- Which conceptual mappings recur across unrelated languages, and what the exceptions suggest about their source.
Key terms
TERMS #| Term | What it means |
|---|---|
| Conceptual metaphor | a systematic correspondence between a concrete source domain and an abstract target domain, of which everyday expressions are the surface trace. |
| Source and target domain | the domain lent (usually concrete and directly experienced) and the domain borrowing it (usually abstract). |
| Mapping | the element-by-element correspondence itself, which carries inferences and not just vocabulary. |
| Embodiment | the stronger claim that such mappings are grounded in bodily and sensorimotor experience and remain active during comprehension. |
Every term the collection defines is gathered in the glossary.