Humor and comedy craft, oriented

A prepared joke has two working parts. The setup gives the listener a context and steers them toward one expected outcome, and the punchline breaks it. The steering happens quietly: the setup makes one reading feel so obvious that the listener settles on it without spotting another. The punch then sends them back over the setup to look for a second reading. One German rating study shows why that reading matters: when an ending clashed with its setup but allowed no second reading, raters found it no funnier than an ordinary ending. Surprise alone is not enough, so give every punch a reading that fits the whole setup.

The piece both readings share is the connector. Often it is a word with two senses, but the double meaning can also live in the grammar of a sentence, in what is implied, or in the wider situation. Aristotle added a test for a word used in a second sense: it works only when that sense fits the facts.

Take "Our puppy is a keen reader. He finishes a book every night, cover included." The setup plants a reader, "finishes" can mean reading to the end or eating, and the closing words tip it into the chewing reading, which fits everything known about puppies. Two habits follow. Cut every word the switch can do without; an early vaudeville manual tells of a monologist who promised five dollars to anybody able to take one word out of his gags without ruining them. And move the trigger to the end, so the surprise comes with the sentence's final words.

The trap is to keep talking once the reveal lands. Adding "which is why the bookshelf now sits on top of the fridge" explains the joke at the moment the laugh should start. Twain made fun of storytellers who say the point twice or tack on an explanation in parentheses, and in one small pilot study of a live set, the comic paused after punchlines and kept quiet until strong laughter started to fade.

A setup that has paid off once can pay again. A tag hangs another punch on the same setup, a topper pushes a step past the line before it, and a callback returns to an earlier line later on, bringing its first laugh back with a new meaning, provided the whole audience saw it the first time. When a joke fails, check the parts in order: did the listener have the fact the point depends on, and can the ending be reread in a way that fits everything before it?

Rule: plant one reading, let a shared element carry a second that fits, put the switch last, and stop.

Misdirection leads a listener to a confident prediction without letting them notice they made one. Listeners fill gaps with whatever a sentence seems to imply, a process called accommodation, and one analysis of jokes describes a teller steering it through word choice and the order in which facts arrive. The same analysis judges that spelling out the inference you want the listener to draw ruins the humor, so hint at the assumption and never state it. Kant held that a jest has to be able to deceive, if only for a moment.

A pattern builds a prediction of its own. Two beats establish it and the third breaks it. Three is a long stage habit rather than a law: an early vaudeville manual calls the third-time payoff a device used for many years, yet its own slapstick example lets the comic fake the blow twice, or even three times, before it lands. Whatever the count, the break goes at the end.

Take "My weekend plan has three steps: clean the garage, sort the closet, then sit back and admire the plan." The chores set the pattern and the last item breaks it. The trap is moving the break. "Admire the plan, clean the garage, sort the closet" uses the same words, and one analysis of jokes judges that a reshuffle of the same material, with nothing new in it, seems to leave a joke less funny and perhaps harder to follow.

Escalation holds one premise fixed while the stakes climb, each step going a little past the last. Exaggeration works best when it enlarges a trait that is really there; Bergson treated it as a means, never the aim. Neighboring moves bend scale, register or literal sense. Understatement describes something huge in small, flat terms, scale mismatch gives a trivial matter grand treatment, register mismatch puts language at the wrong level of formality, and literalization takes a figure of speech at its word. Understatement and scale mismatch both want a straight face, since Twain said the humorous story is told gravely. Even an absurd premise needs rules: once you grant one impossible idea, every step after it should be reasonable.

Escalation has a ceiling. Philosophers since the 18th century have held that once fear, pity or anger takes over, the incongruity stops being funny, and one early manual cautions that if a touching story is turned into a trick, the audience may feel its sympathy was outraged.

Rule: imply the prediction, set the pattern, put the break last, and tie every escalation to something true.

Three old families of theory each explain part of humor. Incongruity theory, which a philosophy encyclopedia calls today's dominant account, says something strikes you as funny when it upsets an expectation or mental pattern. Superiority theory, which led Western ideas of laughter for some two thousand years, says laughter voices a feeling of standing above someone. Hobbes called it sudden glory and gave it two triggers: your own pleasing act, or a flaw you spot in someone else. Relief theory treats laughter as venting built-up nervous energy; almost no scholar today defends it as a general theory, partly because many funny things arouse no feeling to discharge.

Each has a blind spot. A pun need not lower anyone, so superiority struggles with it where incongruity fits. And a broken expectation can bring fear or disgust instead of a laugh. Benign violation theory adds a second condition: something has to seem wrong and, at the same moment, okay, meaning safe or not serious. The theory predicts that a violation too harmless or too severe stops being funny, though it struggles to explain why two listeners split over one joke.

On this theory, psychological distance can tip a violation toward okay: the gap in time or space, how hypothetical it feels, and how socially near the people are. The trap is reading that as more distance, more laughs. Say a coworker's slides froze in front of the team this morning. A joke about it at lunch may feel like a jab while the embarrassment is fresh. The theory's researchers also report a reverse effect: from far enough away, a minor mishap becomes too harmless to register, so the same joke a year later may land less well. One analysis argues that the comfortable range varies from person to person and can split along a power gap, so judge the distance from where the listener stands.

Laughter is a poor stand-in for humor. In observational studies, Robert Provine found that laughter seldom followed jokes and often followed remarks that made no obvious bid to be funny, and that people use it to signal affiliation and agreement. A joke can lose force with repetition: Freud argued that surprise works only once, and that a joke has to be grasped at once, since needing effort to work it out spoils the effect, one account of why an explained joke falls flat.

Rule: use the theories as questions, asking what pattern breaks, who is lowered and whether it feels safe, and never treat a laugh as proof that a joke was made.

Irony runs on a gap between what the words say and what the situation shows. In verbal irony the words state one thing and the speaker means another, frequently the reverse. Context is the key, so a sentence that is plain in one situation can turn ironic in another. Sarcasm points that kind of irony at someone, usually as criticism. Because the meaning lives in the gap, the listener needs a sign that the words are not meant at face value. In speech the sign can be a flat delivery, a slower tempo, a lower pitch or an exaggerated mimicking voice. An 1876 essayist put the condition bluntly: leave a wink, or your remark passes for agreement.

Satire points ridicule at a real target to expose its vice or folly. Dryden contrasted two Roman models. Horace laughs follies to shame with good manners, which Dryden judged the best way to mend manners and the hardest to do well, while Juvenal aims to stir anger. Dryden also wanted each satire to have one chief subject, touching others only in passing, so that its design stays single. Parody aims elsewhere: it copies the recognizable style of one specific writer or work and bends it toward comedy. In a 1994 copyright case, the United States Supreme Court drew the line this way: a parody needs to imitate some original to work, while satire "can stand on its own two feet." Ask whether your joke is really about the source or about the world.

The trap is trusting the reader to catch the wink. On the page, voice and face are gone, which is why writers reach for scare quotes or a winking emoticon, and why a usage guide warned that such crutches talk down to readers. Say you write a mock memo for an office newsletter announcing that every request for a stapler now needs two managers' approval. The closer it sits to real memos, the more clearly it needs a signal. Poe's law holds that without a clear sign of humor, someone will take a parody of an extreme view for the genuine article. The core cue is a clash with the situation, so make the absurd detail unmistakable.

Three smaller devices follow the same logic. Hyperbole is overstatement meant to be recognized as overstatement; once a listener could believe it, it is no longer hyperbole. Litotes affirms something by denying its opposite, and faint praise approves so little that the approval itself becomes a criticism.

Rule: decide what your irony is aimed at, then make the gap between the words and the situation plain enough to see.

A pun is the connector from the first chapter shrunk to one word or one sound. You hear or read the pun word, a second meaning comes to mind silently, and the sentence around it has to support both readings. With a homograph, one spelling carrying two senses, listeners reach the more familiar sense first; one listening study found the second sense caught up about half a second after the word was over. So build the sentence to give the less familiar sense a reason to arrive.

Take "The new tailor is struggling, but he says he will pull through." The tailor puts needle and thread in the listener's mind, so "pull through" can mean both recovering and drawing thread through cloth. The trap is cutting the word that supports the second reading. "My neighbor is struggling, but he says he will pull through" contains the same phrase and no pun, because nothing calls up the sewing sense. A homophone, two spellings that share a sound, carries a different risk. When English puns use an inexact match, studies find the sound shift is usually slight: mostly a consonant changes, while the syllable count and stress stay the same. Stretch the sound too far and the hidden word cannot be recovered. On the page a homophone pun normally prints just one of its spellings, leaving the reader to hear the other.

Puns have long had low standing. Freud observed that they usually sit at the bottom of the rankings of wit, and Hazlitt ranked chance hits on words beneath wit that finds a likeness between things. The remedy is Aristotle's test: the second meaning should fit the facts, not just the sound.

Other devices are built from the same parts. Syllepsis yokes one word to two partners in two senses, as in "He lost his umbrella and, an hour later, the argument"; place last the partner that demands the second sense. A stock phrase predicts its own ending, so its last word is an easy place for a twist, as long as the twisted version still makes sense. Bathos, a drop from the grand to the trivial, can land on a list's last item, and Spencer explained that a petty idea seems pettier when it follows an impressive one. A malapropism is a wrong word used in earnest, in the belief that it means what a similar-sounding word means, so the writer plants it on purpose while the character has to seem unaware.

Rule: give the second meaning a reason to arrive and a fit with the facts, and put the word that delivers it last.

A topic is not a bit. A bit can start from a premise with a stance: something true the audience has seen for itself but never said aloud, plus your view of what is off about it. Aristotle observed that people like hearing a speaker state, as a general truth, what their own experience already told them, and he set the bar for a good saying as true without being commonplace. Build on what the room shares, because a joke resting on private reference can leave outsiders cold.

On a familiar topic, first ideas tend to be the well-worn ones. Research on idea generation finds that later ideas are, on average, somewhat more original, though one study found the effect small and unstable for individuals. So generate a long list before judging any of it; the core rules of brainstorming, quantity first and criticism held back, can carry over to writing alone.

The trap is stopping at the first take. Say your topic is phone alarms. "Hitting snooze is a bad habit" is a topic plus the obvious complaint. Keep going and you might reach a stance: the person who set the alarm last night is treated by the morning you as a stranger whose advice can be ignored nine times. Now there is a premise to break. Name the expectation it sets up, then write the line that overturns it.

Concrete detail carries the claim. Spencer treated it as a settled rule of writing that concrete words strike the mind more vividly than abstract ones, and he credited success to choosing a few particulars that suggest the rest, while a storytelling manual warned that too many bury the point. A comparison must be both unexpected and apt. Aristotle cautioned that an obvious metaphor does nothing, while a far-fetched one is hard to follow, so find the real shared feature before you phrase it. Voicing a character can beat reporting one. Twain thought one old anecdote, told briskly, was barely worth telling, yet performed in character it held a room for ten minutes. If you stretch a true story, stretch it the way the truth already leans, and make the play frame obvious first.

Then let performance shape it. One small study followed two comedians from show to show. Material the newer comic kept seemed to work like a seed, with new lead-ins and follow-ups growing around it, while some new lines turned up once and had vanished by a later show.

Rule: find a stance on something the room already knows, keep generating past your first takes, and grow the bit around what survives in front of people.

An audience hears every line through the person saying it. A comic persona is that steady stage self, and an early vaudeville manual placed this consistency of character just below humor itself in importance: topics can range widely, but each line should leave one clear impression of who is speaking. The practical test is fit. If a joke is strong but would never come out of your persona's mouth, cut it; Aristotle gave speakers similar advice about jests that do and do not suit them. Deadpan is a persona choice too. Studies of spoken humor find that speakers often flag a joke with a smile, raised eyebrows or a nod, and Twain wanted the humorous story delivered with a serious face, which withholds those flags and leaves each listener to catch the joke alone.

Timing is where the silences go and how long they last. Twain called the pause delicate: cut it short and the point goes by unnoticed, stretch it and the audience senses a surprise on the way. A short beat before the punch builds suspense, and the pause after it gives the laugh room. The trap is filling that second silence. Say your first two jokes draw only smiles, so you rush the third and start the fourth on top of it. In one story from a 1914 book of vaudeville anecdotes, a comic talked to a silent hometown crowd for five minutes, stopped for breath, and the house erupted; they had been enjoying him all along and had held back so as not to interrupt. A polite crowd may swallow a laugh you leave no room for. If you prefer to come back in before the room settles, start your next line while the laugh is dying away, and make that line top the one before.

A set is an arrangement, not a list. The opener tells the room who you are, and one early vaudeville manual advised saving the first big laugh until the room is on your side. The closer should be your proven biggest laugh, moved to the end wherever it first sat, and early vaudeville advice held that the right order shows itself only after many tries in front of audiences. Between them, aim for peaks that keep rising and bridges of one short sentence at most. A tight set gets tight by cutting: write more than the slot allows, then drop points to fit.

Rule: settle who is speaking, leave silence on both sides of the punch, and end on the laugh you have already proved.

A joke can be soundly built and aimed at the wrong target. When 30 volunteers rated jokes in one brain-imaging study, the ones belittling a person or group came out both funny and offensive, so a laugh does not settle whether the aim was right. Judge the aim apart from the mechanics, and ask whom the laugh lowers. A philosophy reference work argues that enjoying a joke built on a stereotype does not require believing it, but the joke still keeps the stereotype in circulation. Check whether the joke still works without it.

The room sets the terms. An audience arrives with expectations from the venue and the billing, and in one interview study of workplace leaders, humor that worked often rested on reading the people, the seriousness of the task, and whether the moment was public or private. Teasing stays play only while the person teased can tell it is play, and teasing that leans aggressive can be taken as hostility. Say you pick someone in the front row who did not volunteer and joke about their jacket. The trap is reading the room's laugh as the verdict, when teasing in public can hit harder than the same joke one on one. Watch the person, not just the laugh. At work, a federal guide on harassment names offensive jokes, ridicule and put-downs among conduct that may count as unlawful harassment, when it is linked to a protected trait and is also either a condition of the job or severe or pervasive.

When a joke dies, diagnose before you blame. In one study, 127 students described a time they tried to be funny; 70 percent of those who failed put it down to the other person, while among those who succeeded only 33 percent gave the other person credit. One workplace leader drew a line: if a joke merely falls flat, move on; if someone is visibly offended, apologize at once.

Vaudeville called a joke thief a chooser, a label one early manual ranked worse than liar. Copyright in the United States protects original expression fixed in some tangible form, such as a page or a recording, but not ideas or short phrases. Two comics can arrive at the same joke independently, and copyright practice there treats originality as independent creation: what counts is copying, not resemblance.

Aristotle placed good joking between two failures: the buffoon, who will say anything to get a laugh, and the boor, who never manages a joke.

Rule: judge the target apart from the laugh, read the person as well as the room, and own the miss when a joke hurts.