How does a metaphor become a dead metaphor?

When ghosting entered romantic English around 2010, it did the work of a live metaphor. Readers saw a vanishing presence, a haunted absence, a person reduced to cold air. By 2026 most uses read flat: the verb means to disappear from someone's life without explanation, and the ghost-image is gone. That is a metaphor's death, compressed to about fifteen years. Most take centuries. A few take a decade. The variables that control the speed are predictable enough that a writer can usually tell, on a single sentence, whether they are reaching for a live figure, a dying one, or a phrase already past saving.

What Actually Causes the Death: The Cognitive Mechanism

A live metaphor works by activating a source domain (a sea, a journey, a body part) and mapping its qualities onto a target (an emotion, a situation, an object). Every time a listener hears the figure, they perform a small inference: they take the source, find the relevant features, and apply them to the target. That inference costs something. With repetition, the listener stops paying it. The target meaning gets stored directly as a sense of the word, and the source-domain activation drops out of the process entirely. By the end the expression is processed as a polyseme, one of the word's literal meanings, not as a figure.

Linguists call this lexicalization (sometimes conventionalization): the figure settles into the dictionary as an ordinary sense of the word.

George Lakoff (The Death of Dead Metaphor, 1986) argued that what dies isn't the metaphor itself but the mapping between domains. A figure like pedigree qualifies as fully dead in his framework because the mapping is gone. No competent speaker connects pedigree to its source: literally "foot of a crane," from pied de grue, after the bird-track marks medieval clerks drew on lineage charts to indicate descent. The word has been cut loose from its source domain entirely.

Max Black (Models and Metaphors, 1962) took the harder line: a fully dead metaphor isn't a metaphor at all. It is an ordinary lexical item that happens to have a metaphorical history, in the same way muscle (from Latin musculus, "little mouse," for the way muscles ripple under the skin) is no longer a metaphor but just a word. Black's position is that calling such words metaphors confuses etymology with semantics.

This is a live disagreement, not a settled question. Lakoff's mapping-fade account treats death as a continuum; Black's literalization account treats it as a categorical exit from the class of metaphors. The next sections work in Lakoff's terms because the gradient is more useful for a writer trying to diagnose a sentence, but the closing section returns to the dispute.

The Gradient: Live, Dying, and Dead

A figure passes through three stages along the way to lexicalization. The names vary by handbook (some sources use moribund or sleeping for the middle stage), but the gradient itself is stable.

A live metaphor is one in which the source domain is still doing work in the listener's head. All the world's a stage (Shakespeare, As You Like It, 1599) works because the reader sees a stage: boards, players, an audience, an exit. The mapping is active, unprompted, and the figure carries the inference.

A dying metaphor sits in the middle. The figure is recoverable if pointed at, but it doesn't fire automatically. Time is running out, grasp an idea, the bottom of the hour, a stitch in time: the source is there, but a competent reader processing the sentence at speed will not stop to see liquid leaving a vessel, a hand closing on an object, a pail with a bottom, a needle. Reminded of the source, they recognize it. Left alone, they read past it. This is the band Orwell complained about (Politics and the English Language, 1946) when he wrote that dying metaphors "have lost all evocative power and are merely used because they save people the trouble of inventing phrases for themselves."

A dead metaphor has lost its source entirely. Pedigree, muscle, daisy ("day's eye," after the flower's habit of opening at dawn and closing at dusk): no native speaker hears the original image. The mapping is unrecoverable without external knowledge.

The test runs in one line: if a competent reader has to be told the etymology to recognize the figure, it's dead. If reminding them of the source lifts the image, it's dying. If the image is already there unprompted, it's live.

Stage	Source-domain activation	Example	Test
Live	Active and visible	All the world's a stage (Shakespeare, As You Like It, 1599)	The image is in the reader's head unprompted
Dying	Recoverable on prompt	Time is running out; grasp an idea	A brief reminder lifts the figure
Dead	Gone	Pedigree; muscle; daisy	Etymology has to be supplied for the image to register

Why Some Metaphors Die Faster Than Others

Four variables govern how quickly a metaphor moves down the gradient.

Frequency of use. The most obvious factor. High-traffic phrases lexicalize within a generation. Ghosting, in the romantic sense, was minted around 2010 and is already moving from live to dying in 2026 prose: a careful reader still feels the spectral source on slow attention, but most uses now read as a flat verb meaning "to disappear from someone's life without explanation." A few decades of dating-column traffic was enough.

Source-domain accessibility. Figures whose source remains part of everyday life stay live longer. A stage is still a vivid image because we still go to plays. A journey still works because people still travel. Figures whose source has dropped out of common life die fast once the source disappears: put through the wringer (the wringer of a washing machine, which most readers under fifty have never operated), bite the bullet (battlefield surgery before anesthesia), a stitch in time (mending clothes by hand). Once the world the source pointed to is gone, the mapping has nothing to map from.

How recognizable the original word is. When the source word is still intact and obvious, the figure can stay dying for centuries. Backbone of the company has been in use since the 17th century and still recovers its image on a moment's attention because backbone still means backbone. When the word fuses or shortens, the figure dies quietly. Goodbye is a contraction of God be with ye (Elizabethan), and no English speaker hears the original blessing in the modern word. The compression closed the door on the source.

Genre traffic. Where the figure circulates matters. Figures that live mostly in journalism or political prose die faster, because those genres prize fluency and recycle stock phrases at high speed. Figures preserved in literary or liturgical use stay alive longer because their reading contexts reward attention. The King James Version (1611) has kept whole categories of biblical metaphor (the apple of his eye, a thorn in the flesh, a wolf in sheep's clothing) closer to dying than dead in English-language prose, even as those same figures faded out of secular speech, because religious repetition is slow, attentive repetition rather than the fluent kind.

How Do You Tell If You've Written a Dead Metaphor?

Three tests, in order of usefulness.

The image test. Read the sentence aloud and ask whether the image actually appears in your head. You wrote the company is bleeding cash. Do you see blood? If yes, the figure is live in this context, and it may be worth keeping. If no, the figure is dead, and you have a choice: either replace it with a phrasing you control, or commit to the image and extend it (the company is bleeding cash, and no one in finance has bothered to find the wound). The test fails the same way for the writer as it does for the reader, which is what makes it useful.

The etymology test. Can you state the source domain unprompted? If you wrote broadcast and you can't say where the word came from (sowing seed by hand, casting it broadly across a field), that figure is dead in your usage. Same for deadline (the line drawn around a Civil War prison camp; prisoners crossing it were shot), pedigree, muscle. Dead etymology is fine in itself. The point of the test is just to confirm the figure isn't doing figurative work for you.

The replacement test. Swap the figure for its literal equivalent. If the sentence loses no force, the figure was dead and you weren't getting any work out of it. The deadline is tomorrow and the cutoff is tomorrow mean the same thing with the same weight, which tells you deadline is an ordinary noun for you in this sentence. If the swap drains the sentence, the figure was alive and pulling its weight.

There is one asymmetry worth keeping straight. Dead metaphors are not automatically bad writing. A truly dead metaphor like deadline or broadcast is just a word; it does its job, you move on, no one is the worse for it. Orwell's complaint was specifically with dying metaphors: the band where the figure is faded enough that you don't notice it but recognizable enough that you reach for it as a ready-made phrase instead of inventing your own. Grasp an idea, the iron fist of, toe the line, play into the hands of: these are the ones Orwell wanted writers to notice. Most popular writing advice flattens this and tells writers to avoid dead metaphors generally. That advice misses the actual line. The thing worth catching is the dying figure you reached for because it was nearby, not the dead one that has settled into being a regular word.

Are Dead Metaphors Even Metaphors Anymore?

This is the taxonomic question, and it has not been settled. Black's Models and Metaphors (1962) takes the cleanest line: a fully dead metaphor isn't a metaphor at all but an independent lexical item. Muscle on this account is no more metaphorical than tree. The metaphorical history is just history, the way the sinister in modern English carries no live association with the Latin for "left-hand-side" even though the word came from there.

Lakoff (1986) pushes back. Even in cases where no individual phrase feels figurative, the conceptual mapping often persists below conscious awareness and shapes how speakers reason about the target domain. We grasp arguments, lose trains of thought, build cases, attack positions, defend claims. No single use of these verbs feels metaphorical, but the cluster reveals an underlying mapping (ARGUMENT IS A PHYSICAL CONTEST, IDEAS ARE OBJECTS) that organizes the entire vocabulary. On Lakoff's view, the death of a particular phrase doesn't entail the death of the conceptual structure behind it. The mapping is still there, just no longer visible at the surface of any one expression.

A third angle from the philosophy-of-language side: it depends on what level you're asking about. The speaker's experience reading the sentence, the dictionary's record of the word, and the history of where the word came from can each give a different answer to "is this still a metaphor." The question isn't broken; it just needs to specify which level it's being asked at.

For analysis, the distinction matters and the disagreement is real. Different theoretical commitments about what metaphor is generate different answers about what dies and what survives. For a writer in front of a draft, none of it matters. The practical question is not is this metaphor dead as a category claim. It is is the image alive in this sentence, for this reader, right now. That question has the merit of being answerable on a single reading, and it survives every taxonomic dispute about whether dead metaphors are still metaphors at all. The same phrase can be alive on one page and dead on the next. Diagnosis happens locally, sentence by sentence, which is the same standard that decides whether a working metaphor is doing real work anywhere else in a draft. The category is interesting; the sentence is the work.