Two men can trade forty clever lines and leave the room exactly as they entered it. One true sentence, said at the wrong moment, can end a friendship or start a life.
Most manuscripts I read don’t have a dialogue problem in the usual sense. The lines are often good. The men spar easily, the rhythm snaps, the jokes land, and every exchange ends on a button a television writer would envy. Then the scene ends, and nothing between the men has moved. They know what they knew before. They want what they wanted before. Neither has risked anything, and the reader turns the page with the pleasant, hollow feeling of having watched two talented people perform.
Polished banter is seductive because it looks like chemistry and reads fast. It also costs the characters nothing, and dialogue that costs nothing can’t change anything. Conversation becomes action when somebody uses words to try to change another person—to get something, hide something, protect something, or end something—and the attempt carries a price whether it succeeds or fails. Characters talk too much when their talk has stopped being an attempt and become a display.
This essay is about four tools that put the cost back, plus a few habits that take it away. Evasion makes the reader watch a man work to avoid an answer. Interruption and timing turn the rhythm of an exchange into evidence about power and fear. Unequal knowledge lets one man speak while the reader hears what he’s hiding. And the badly timed honest sentence breaks a scene open because it arrives when nobody can afford it.
The research behind these tools comes mostly from linguistics, conversation analysis, and social psychology. Those fields study real talk, and fictional dialogue isn’t transcription. Real conversation is full of repetition, false starts, and dead air no novel could survive. What the research offers a fiction writer is a map of the forces real speakers navigate without noticing. Readers navigate the same forces every day, so they recognize them on the page even when they couldn’t name them. The craft conclusions are mine, and I’ll mark where the evidence stops.
Nobody says what he means, and readers know it
The philosopher H. P. Grice proposed that conversation runs on a cooperative assumption. Listeners expect speakers to say enough but not too much, to say what they believe true, to stay relevant, and to be clear. When a speaker visibly breaks one of those expectations, the listener doesn’t conclude that the speaker is broken. The listener infers a meaning beyond the words. Grice called that inferred meaning implicature.[1] If a man asks whether his partner liked the dinner and hears “The wine was excellent,” he learns more about the food than any direct review could tell him.
Implicature is the native language of romance. A great deal of what two people feel for each other travels in what they decline to say, what they say instead, and what they say at a length or in a place that doesn’t fit the stated subject. The reader decodes all of it automatically, because the reader has been decoding it since childhood.
Real speakers also mislead each other routinely. In diary studies by Bella DePaulo and her colleagues, college students reported lying in about one of every three social interactions and community members in about one of every five, most often to protect feelings, smooth a moment, or present themselves well.[2] The figures come from self-report and reflect two U.S. samples, so they shouldn’t be treated as a universal rate. They still suggest how unusual perfect candor is. A scene in which two men say exactly what they mean, at the moment they mean it, in complete sentences, describes a kind of conversation people rarely have.
Polished banter fails the same test from the other direction. It isn’t candid either, but it misleads nobody, because both men understand that the wit is a performance and neither tries to get anything through it. The surface sparkles and nothing moves underneath. Hemingway’s image for prose built on omission was the iceberg, whose dignity of movement comes from the seven-eighths that stay underwater.[3] Banter that’s all surface has nothing below the waterline. The reader senses that emptiness before naming it.
The useful question for any exchange is what each man is trying to do with his words. Asking for help while appearing not to ask is an action. Testing whether the other man remembers is an action. So is warning him off, inviting him in, buying time, changing the subject before it reaches the wound, or making him say something first. When every line in a scene serves one of those purposes, the dialogue moves even if every line is quiet. When the lines serve only the author’s wish to sound clever, the dialogue stalls even if every line is brilliant.
Banter is a truce, and truces end
Penelope Brown and Stephen Levinson built an influential theory of politeness around the idea of face—the public self-image every speaker wants to protect. Requests, refusals, criticism, and confessions all threaten face, either the speaker’s or the listener’s, and much of ordinary language is machinery for softening those threats or avoiding them.[4] The theory has been debated and refined across cultures for decades, and nobody should treat it as a law of human speech. It does give writers a useful lens on what banter accomplishes.
Banter lets two people stay close while protecting both faces. Teasing can carry affection without the risk of declaring it. A shared joke can acknowledge attraction while preserving deniability. A barbed exchange can register a grievance without making anyone apologize. In early chapters, banter often does real work, because the men need a way to approach each other that neither has to defend.
The trouble starts when the truce becomes permanent. If every charged moment dissolves into wit, the reader learns that the men will always choose the safe register, and the dialogue stops generating suspense. The reader no longer wonders what either man will say, because the answer is always something clever.
Meet Dace, who runs the repair shop at a struggling marina, and Corwin, the lawyer who has come to manage its sale. Here’s their third conversation, written as banter.
“You’re in my light,” Dace said.
“You don’t have a light. You have a bulb and a grudge.”
“The grudge works fine.”
“So I’ve noticed.” Corwin leaned on the bench. “Does it come with a warranty?”
“Lifetime.”
It moves, it sounds like chemistry, and it accomplishes nothing. Now let the truce break.
“You’re in my light,” Dace said.
“You don’t have a light. You have a bulb and a grudge.”
“The grudge works fine.” Dace set down the wrench. “Are you closing the shop?”
Corwin had a joke ready. He let it go. “I don’t know yet.”
The second version keeps the banter long enough to show the habit, then lets one man step outside it. Dace’s question threatens both faces at once. It exposes his fear and forces Corwin either to lie, to evade, or to say something true. Corwin’s abandoned joke is the most important line in the exchange, and it’s a line nobody speaks. The reader sees him decide that this moment is too serious for the register they’ve been using, which tells the reader the relationship has just changed.
Banter earns its place when it sets up the moment one man refuses to continue it. The writer should know, for every witty exchange, what truth the wit is holding off, and should plan the scene where somebody finally lets that truth through.
Evasion is an answer
Janet Beavin Bavelas and her colleagues studied equivocation—messages that are deliberately unclear about content, sender, receiver, or context. Their central finding was situational. People equivocate most when every direct answer available to them would be harmful, such as when a friend asks for an opinion of a gift the speaker dislikes. In those avoidance-avoidance conflicts, equivocation isn’t a personality flaw. It’s a reasonable response to a situation with no good clear answer.[5]
That finding gives a writer a precise tool. An evasive line becomes dramatic when the reader can see both of the bad answers the character is trying to avoid. Evasion without visible alternatives reads as coyness, or worse, as an author withholding information. Evasion between two visible bad answers reads as a man in a trap.
Return to Corwin. The buyer has told him privately that the repair shop will be demolished within the year. If Corwin tells Dace, he breaches a confidence and may lose the sale his firm needs. If he tells Dace the shop is safe, he lies to a man he’s begun to care about. When Dace asks whether he’s closing the shop, Corwin’s “I don’t know yet” is an equivocation, and the reader can see what it costs because the reader knows both answers he’s refusing to give.
Evasion also tells the listener something, which keeps the exchange from becoming one-sided. Dace may not know why Corwin dodged, but he heard the dodge. A good evasion scene tracks both men. The evader works to give away as few facts as possible, and the listener works to read what the effort means. If Dace answers “That’s a lawyer’s answer,” he hasn’t learned the truth, but he’s learned that there is a truth, and the relationship now contains a secret both men know exists.
Evasion comes in several shapes, and varying them keeps a long book from repeating itself. A man can answer a different question from the one asked, or answer with a question of his own. A technically accurate reply that leaves out what matters is another option. So is changing the subject so abruptly that the change becomes the answer. Or he can retreat into the role that protects him—the lawyer, the sergeant, the host—and speak only in that register. Each shape reveals something different about how this man protects himself.
The limit on evasion is reader patience. A character who equivocates in every scene about the same secret starts to look less trapped than lazy, and the book starts to look like it’s stalling. Each evasion should cost more than the last, either because the stakes have risen or because the listener has grown less willing to accept it. Eventually one of them has to break the pattern.
Timing is meaning
Conversation analysts have spent half a century showing how precisely ordinary talk is timed. In a foundational 1974 paper, Harvey Sacks, Emanuel Schegloff, and Gail Jefferson described turn-taking as a system that tends to minimize both gaps between speakers and overlaps in which two people talk at once.[6] Because the system leans toward smooth transitions, departures from it carry meaning.
Delay is the clearest case. Kobin Kendrick and Francisco Torreira analyzed responses in recorded telephone calls and found that the timing of the most common responses didn’t differ much by type, but that once a gap reached roughly 700 milliseconds, dispreferred responses such as rejections became clearly more common, and longer delays tracked with qualified, hedged answers.[7] Sara Bögels, Kendrick, and Stephen Levinson found in a brain-recording study that listeners’ responses to a “no” differed depending on whether it came right away or after a pause, suggesting that people anticipate trouble from timing alone.[8] Anita Pomerantz’s earlier analysis of agreement and disagreement described the same pattern qualitatively. Disagreements tend to arrive late, softened, and prefaced, while agreements arrive fast and plain.[9]
All of this research measures fractions of a second in recorded speech, and a novel can’t reproduce milliseconds. What fiction can reproduce is the reader’s sense that an answer came too slowly, too fast, or not at all. A beat of description between question and answer creates delay. An answer that starts with “Well” or “I mean” signals the hedge the research associates with trouble. An answer that arrives before the question finishes signals something else entirely.
“Do you want me to stay tonight?”
Dace wiped his hands on the rag, then wiped them again. “It’s late to drive.”
The rag does the work of the pause. Dace’s answer is technically a yes, but it arrives late, it arrives through logistics, and it declines to say want. The reader hears all of that because readers hear delay on the page exactly as they hear it across a table.
Interruption runs the other direction. A man who cuts someone off has taken the turn, and taking the turn is an exercise of power. Who interrupts whom, and who allows it, says a great deal about a relationship. A senior officer who finishes a junior soldier’s sentences is showing rank. A lover who interrupts the other man’s apology to stop him from groveling may be showing mercy, or may be refusing to hear something he isn’t ready for. The interrupted line is often more revealing than the interrupting one, because the reader gets to imagine how it would have ended.
“Corwin, if the buyer—”
“Don’t ask me about the buyer.”
“I wasn’t going to ask you. I was going to tell you I don’t care what—”
“Dace.”
Corwin interrupts twice. The first interruption protects his secret. The second protects Dace from making a promise Corwin knows he can’t let him keep. The reader may not know yet what Dace was about to say, but the reader knows it mattered enough for Corwin to stop it, and that knowledge carries the scene.
Silence belongs here too. The unanswered question can be the loudest line on a page, provided the reader knows what answer is being withheld and what it costs the silent man to withhold it. A silence the reader can’t interpret is just white space. A silence the reader can interpret becomes a refusal, a confession, or a plea, depending on what came before it.
Unequal knowledge turns talk into suspense
Alfred Hitchcock explained suspense to François Truffaut with a bomb under a table. If a bomb suddenly explodes during a conversation, the audience gets a few seconds of surprise. If the audience knows the bomb is there and the characters keep chatting, the most ordinary conversation becomes unbearable for fifteen minutes.[10] Dialogue works the same way. A scene in which one man knows something the other doesn’t, and the reader knows what he knows, turns every innocent line into a near miss.
Unequal knowledge is the fastest way to give quiet dialogue stakes. Once the reader knows that Corwin has seen the demolition plans, Dace’s offhand remark about expanding the shop next spring lands hard. Corwin’s answer, whatever it is, becomes a choice between lying, deflecting, and confessing, and the reader watches him make it in real time.
“Next spring I’m knocking out that wall,” Dace said. “Double bay. Two boats at once.”
Corwin looked at the wall for a long time. “It’s a good wall.”
The line is nearly meaningless to Dace. It’s nearly unbearable to the reader. That gap between what the listener hears and what the reader hears is the engine, and it runs without a single dramatic word.
Unequal knowledge also explains how lovers talk in public. Herbert Clark and Edward Schaefer showed in experiments that speakers can design utterances so that an intended addressee understands them while an overhearer doesn’t, by drawing on knowledge the two of them share and the overhearer lacks.[11] Couples do this constantly, and secret couples do it with their safety at stake. A reference to a place, a borrowed phrase from an earlier scene, or a joke only one other person could decode lets two men say something intimate in a crowded room.
For M/M romance, this is far more than a stylistic trick. Men who can’t be openly together in a given setting—a unit, a family dinner, a workplace, a hometown—often speak in exactly this double register, and the reader’s ability to decode it puts the reader inside the couple. I’ve written about how rank shapes what men can risk being seen to mean in Writing the Lives Between Duty and Desire. Coded speech is one of the ways that cost appears line by line. The danger, of course, is that the code fails. An overhearer who decodes the private meaning is one of the most efficient crises a writer can build.
Unequal knowledge between the reader and both men works too. Dual point of view can let the reader know what each man wants while neither knows what the other wants, so that every polite exchange becomes two people missing each other at close range. The risk is the one every writer of mutual pining knows. If the reader can see both hearts and the obstacle is weak, the reader starts wondering why the men don’t simply talk. Unequal knowledge creates suspense only while the reason for silence holds.
The badly timed honest sentence
Every scene so far has been about men avoiding the truth. The most powerful dialogue tool is the moment one of them stops avoiding it at the worst possible time.
In Brown and Levinson’s terms, a blunt truth is a face threat delivered without the usual softening. Most of the time, people soften by instinct. When a character stops softening, the reader knows that something inside him has given way, and the timing of that collapse decides its cost. The same sentence spoken in private, at dawn, after a long conversation, is a confession. Spoken in front of the buyer, the family, the unit, or the ex, it’s an explosion.
The buyer was still talking about drainage when Dace said, “He knows you’re tearing the shop down. He’s known for a week.”
Dace has learned the truth somewhere offstage, and he chooses to say it here, where it will cost Corwin the most. The sentence is honest and it’s an act of aggression. It also exposes Dace, because the reason he’s angry enough to say it in public is that Corwin’s silence hurt him more than a business decision should hurt a stranger. In one line he wounds Corwin, damages the sale, and reveals how much he cares.
A badly timed honest sentence doesn’t have to be cruel. It can be a declaration of love blurted in the middle of a fight about something else. It can be the admission “I was scared” during an argument both men expected to be about logistics. It can be a man saying “I’m gay” in a room where he’d planned never to say it. What makes it work is the gap between the sentence and the setting. The reader feels the room change.
Three conditions make the badly timed sentence land. The reader has to have watched the truth being held back, so the release has weight. The setting has to raise the cost of saying it, so the timing matters. And the scene has to stay in the room long enough afterward to show the consequence. A sentence that blows up a dinner and then cuts to the next morning has thrown away the most valuable minutes it created. Stay for the silence, the first response, and the moment someone tries to recover the old register and can’t.
I’ve argued in Rebuilding After the Break that an apology works when it names the harm accurately instead of performing remorse. The same principle applies to every honest sentence. Precision is what makes it expensive. “I’m sorry about everything” costs nothing. “I knew about the shop and I let you plan the wall anyway” costs a great deal, and the reader believes it for that reason.
Two guarded men in one conversation
M/M romance often puts two men in a room who were both raised to say as few vulnerable words as possible. That’s a craft opportunity and a craft trap. The opportunity is obvious. When neither man will name what he feels, every gesture toward naming it becomes an event. The trap is that two silent men can produce a book in which nobody says anything, and the reader is left assembling an entire relationship from glances.
What follows is my own observation from writing these men rather than anything the research above establishes. Guarded men often talk most freely when their hands are busy and their eyes are elsewhere. Two men fixing an engine, driving at night, cleaning weapons, or splitting wood can say things neither could say face to face, because the task gives them somewhere to look and an excuse to stop talking at any moment. The shared work lowers the cost of honesty without removing it. A writer who needs a guarded man to say something true can often get there by giving him a job to do while he says it.
Dace handed him the impeller without looking up. “My dad sold this place once. Bought it back two years later. Never said why.”
Corwin fitted the part. “Did you ask?”
“Every day for two years.”
Neither man looks at the other, and neither stops working. Dace has told Corwin what the shop means to him, and Corwin has heard it, and the engine gives them both a place to put their eyes while it lands. The exchange would be unbearable across a table. Over an engine it’s merely hard.
Guarded men also need contrast to stay readable. If both speak in the same clipped register, the reader can’t tell their silences apart. Give each man his own way of not talking. Dace deflects with facts and stays put. Corwin deflects with jokes, and when the jokes fail, he leaves the room. When their evasions differ, the reader can track who is retreating and how far, and the moment one of them abandons his usual method becomes visible.
Finally, guardedness has a history, and the dialogue can carry it. A man who learned in the barracks, the locker room, or his father’s house that certain words were dangerous will avoid those words with a precision the reader can learn to see. When he finally uses one, the reader who has watched him avoid it for two hundred pages feels the full weight of the choice.
When the talk should stop
Characters often talk too much because the writer doesn’t trust the reader. The writer fears the meaning won’t land, so the characters explain it to each other. They articulate their feelings in paragraphs, summarize what just happened, and name the subtext the scene already delivered. Each explanation costs the reader a discovery.
A useful test is to delete the line in which a character explains his own feeling and see whether the scene still makes the feeling clear. Often it does, because the evasion, the delay, the interruption, or the action already carried it. When the deleted line was doing real work, it usually needs to move later, after the reader has felt the feeling, so that the explanation arrives as a cost instead of a caption.
The opposite failure is dialogue that talks to the reader instead of the other man. Two men who share a history don’t remind each other of it in complete sentences. “As you know, we haven’t spoken since your brother’s wedding” tells the reader something by making the characters behave strangely. If the reader needs the fact, let it surface through friction. One man gets the date wrong and the other corrects him too fast, or one refers to the wedding by a private name that tells the reader it went badly. Information delivered through conflict costs the characters something. Information delivered through recital costs only the reader’s patience.
Action can replace a line outright. A man who can’t say he’s staying can unpack his bag. A man who can’t say he’s afraid can check the lock twice. A man who can’t apologize can do the chore the other man always does. These substitutions aren’t evasions of dialogue. They’re dialogue in another channel, and they often hit harder because the character didn’t choose words he could take back.
When both men are speaking in a charged scene, the mechanics matter too. Two male speakers can blur, and a reader who has to stop and work out who said the line has left the scene. I’ve covered that line-level problem in He, Him, and the Other Man. Clean attribution is part of what lets subtext work, because the reader can only decode what a man means once the reader knows which man said it.
Revising dialogue that costs nothing
These questions help locate where a conversation has stopped working as action.
- What is each man trying to get, hide, protect, or end in this exchange?
- What truth is the banter holding off, and where does somebody finally let it through?
- When a man evades, can the reader see both of the bad answers he’s avoiding?
- Where does an answer come too late, too fast, or not at all, and does the page show it?
- Who interrupts whom, and what does the interrupted line reveal?
- What does one man know that the other doesn’t, and does the reader know it too?
- Which honest sentence is being held back, and where would saying it cost the most?
- Which lines explain a feeling the scene has already shown?
- If you cut every clever line, what’s left, and is it enough?
That last question stings, but it’s the most useful one on the list. Clever lines are fine. They’re a pleasure to write and to read. The problem only arrives when cleverness is all the scene has. Strip it out, and if the men are still trying to do something to each other, the scene was alive and the wit was decoration. If nothing remains, the wit was the scene, and the scene needs a reason to exist.
Make them pay for every word
Good dialogue in a romance isn’t measured by how good the men sound. It’s measured by what talking costs them. The evasion that leaves a secret visible, the pause that turns a yes into a confession, the interruption that saves or silences, the knowledge one man carries while the other plans a future that can’t happen, and the true sentence spoken in the worst possible room all share one quality. Somebody had to pay for them.
Let the men be sharp and funny when they’re protecting themselves. Then put them somewhere wit can’t protect them, and make one of them say what he’s spent the book not saying. Readers will forgive a great deal of quiet for one line that costs a man something he can’t get back.
Sources
These works inform my craft argument; none is offered as proof of how a fictional scene must unfold. The conversation research measures recorded real talk, which fiction compresses and shapes.
- H. P. Grice, “Logic and Conversation,” in Syntax and Semantics, vol. 3, Speech Acts, ed. Peter Cole and Jerry L. Morgan (Academic Press, 1975), 41–58. ↩
- Bella M. DePaulo, Deborah A. Kashy, Susan E. Kirkendol, Melissa M. Wyer, and Jennifer A. Epstein, “Lying in Everyday Life,” Journal of Personality and Social Psychology 70, no. 5 (1996): 979–95, https://doi.org/10.1037/0022-3514.70.5.979. ↩
- Ernest Hemingway, Death in the Afternoon (Charles Scribner’s Sons, 1932). ↩
- Penelope Brown and Stephen C. Levinson, Politeness: Some Universals in Language Usage (Cambridge University Press, 1987), https://doi.org/10.1017/CBO9780511813085. ↩
- Janet Beavin Bavelas, Alex Black, Nicole Chovil, and Jennifer Mullett, Equivocal Communication (Sage, 1990). ↩
- Harvey Sacks, Emanuel A. Schegloff, and Gail Jefferson, “A Simplest Systematics for the Organization of Turn-Taking for Conversation,” Language 50, no. 4 (1974): 696–735, https://doi.org/10.2307/412243. ↩
- Kobin H. Kendrick and Francisco Torreira, “The Timing and Construction of Preference: A Quantitative Study,” Discourse Processes 52, no. 4 (2015): 255–89, https://doi.org/10.1080/0163853X.2014.955997. ↩
- Sara Bögels, Kobin H. Kendrick, and Stephen C. Levinson, “Never Say No … How the Brain Interprets the Pregnant Pause in Conversation,” PLOS ONE 10, no. 12 (2015): e0145474, https://doi.org/10.1371/journal.pone.0145474. ↩
- Anita Pomerantz, “Agreeing and Disagreeing with Assessments: Some Features of Preferred/Dispreferred Turn Shapes,” in Structures of Social Action: Studies in Conversation Analysis, ed. J. Maxwell Atkinson and John Heritage (Cambridge University Press, 1984), 57–101. ↩
- François Truffaut, with Helen G. Scott, Hitchcock (Simon and Schuster, 1967). ↩
- Herbert H. Clark and Edward F. Schaefer, “Concealing One’s Meaning from Overhearers,” Journal of Memory and Language 26, no. 2 (1987): 209–25, https://doi.org/10.1016/0749-596X(87)90124-0. ↩