The most common mistake writers make with dialogue is trying to make it sound real. This seems reasonable. Characters are people. People talk. Surely the goal is to capture how people actually speak: the hesitations, the repetitions, the half-finished sentences, the mundane exchanges that fill ordinary life. Record a conversation between two real people and transcribe it faithfully, the thinking goes, and you will have authentic dialogue.
Transcribe a real conversation and what you will actually have is something close to unreadable. Real speech is circuitous, redundant and largely without purpose. It fills time as much as it communicates. It contains vast stretches of phatic communion: the 'how are yous' and 'not bads' and 'anyway, I should let you go' that oil the wheels of social interaction but carry no dramatic weight whatsoever. Real speech, on the page, is boring.
Screenplay dialogue is a performance of naturalism, not naturalism itself. It creates the impression of how people speak while doing things that real speech almost never does: advancing the story, revealing character, establishing conflict, carrying subtext, and doing all of this with an economy that keeps the script moving. The best screen dialogue feels real and effortless precisely because the craft behind it is invisible. The moment the craft shows, the dialogue stops working.
David Mamet, one of the most distinctive dialogue writers in American cinema and theatre, puts it plainly: the purpose of dialogue is not to communicate information. It is to make the audience wonder what is going to happen next. Everything else, the character revelation, the conflict, the texture of a particular world and a particular voice, is in service of that forward momentum. Dialogue that stops the story in order to be clever, or literary, or impressive, is dialogue that has forgotten what it is for.
If the purpose of dialogue is not simply to communicate, the obvious question is: what is it doing instead? The answer is several things at once, and understanding each of them separately helps you diagnose what is wrong when a scene is not working.
First, dialogue reveals character. Not through direct statement, the character who announces their own qualities is always unconvincing, but through the specific words a person chooses, the things they avoid saying, the way they respond under pressure, the gap between what they claim and what they do. Every character should have a distinct voice: a particular rhythm, a characteristic vocabulary, a habitual way of approaching conversation that reflects who they are and where they come from. When you can read a line of dialogue and know which character said it without a name attached, the voices are working.
Second, dialogue creates and sustains conflict. Conversation in drama is almost never simply an exchange of information between people who are cooperating fully and honestly. It is a negotiation, a performance, a battleground. Characters want things from each other. They withhold, deflect, manipulate, test, challenge and occasionally reveal. Even in scenes that appear to be calm and friendly on the surface, the most interesting dialogue has an undercurrent of tension: something that is not being said, something that is at stake, something that could shift at any moment.
Third, and perhaps most importantly for the page, dialogue carries subtext. What characters say and what they mean are rarely identical. The gap between the two is where the real drama lives, and where the audience does the most active and pleasurable work of interpretation. A scene in which everything is said directly, in which characters express exactly what they feel and want with complete transparency, is a scene without subtext, and without subtext it is almost always a flat scene, no matter how well the individual lines are written.
What characters say and what they mean are rarely identical. The gap between the two is where the real drama lives, and where the audience does the most active and pleasurable work of interpretation.
Subtext is one of those concepts that writers understand in the abstract but often struggle to produce in practice. The principle is simple enough: characters talk around what they really mean, and the audience reads between the lines. The execution is harder, because it requires you to hold two conversations in your head simultaneously: the one on the surface and the one underneath.
The classic example is the scene in which two characters discuss something apparently trivial while the real subject of the conversation is something neither of them can bring themselves to name directly. A couple arguing about where to go for dinner when the real argument is about whether the relationship has a future. Two colleagues discussing a project when what is actually being negotiated is who has power over whom. The surface conversation provides cover. The subtext provides meaning.
Harold Pinter built an entire body of work on this principle. His characters talk in circles, repeat themselves, discuss nothing of apparent importance, and yet the scenes are dense with threat, desire and unspoken history. The silences in a Pinter play are as loaded as the words. What is not said has as much weight as what is said, sometimes more. For screen writers, Pinter is a masterclass in what dialogue can carry when it trusts the audience to do the interpretive work.
Producing subtext in your own work requires you to ask, for every scene, what each character actually wants and what they are actually willing to say. Those two things are rarely the same. The gap between them is the subtext. Write the surface conversation, then check whether the gap is present and whether it is doing genuine dramatic work. If the characters are saying exactly what they mean, ask why. Sometimes directness is the right choice, particularly in moments of crisis or revelation. But it should be a choice, not a default.
One of the clearest signs of a developing writer is dialogue in which all the characters sound like the same person, usually the writer themselves. The vocabulary is consistent across characters regardless of background, education or personality. The rhythm of speech is uniform. The wit, if there is wit, is distributed evenly. Reading the script, you could swap character names at random and lose very little.
Distinctive voice is not simply a matter of adding verbal tics: having one character say 'y'know' and another say 'absolutely' and a third speak in complete sentences. Those are surface differences. Genuine voice goes deeper, into the way a person constructs thought in language, the things they reach for as metaphor, the lengths to which they will go to avoid certain words or topics, the speed at which they process and respond. Voice is character expressed through the specific texture of speech.
A useful exercise is to take a scene you have written and remove all the character names. If you cannot tell who is speaking from the dialogue alone, the voices are not yet distinct enough. Another approach: write a short monologue for each of your main characters in response to the same question, something neutral and open-ended. The differences in how they answer, in what they choose to say and what they omit, will reveal where the voices are and are not individuated.
It is also worth remembering that voice includes silence. Some characters talk a great deal. Some say very little. Some answer questions with questions. Some go long distances around a subject before arriving at it, or never arrive at all. The quantity and shape of a character's speech is as much a part of their voice as the specific words they choose. The character who speaks rarely and precisely will have a very different presence in a scene from the one who fills every silence with words.
Distinctive voice is not simply a matter of adding verbal tics. Genuine voice goes deeper, into the way a person constructs thought in language, the things they reach for as metaphor, the lengths to which they will go to avoid certain words or topics.
On-the-nose dialogue is the term for lines in which characters say exactly what they mean, feel or want, with no gap between surface and subtext. It is the most common dialogue problem in early drafts, and the hardest to see in your own work, because on-the-nose dialogue often feels clear, purposeful and emotionally direct. The problem is that clarity purchased at the cost of subtext is almost always the wrong trade.
'I am angry with you because you lied to me and I feel betrayed' is on the nose. So is 'I have always been afraid of intimacy because of what happened with my father.' These lines explain the character's psychology rather than revealing it through behaviour. They remove the interpretive work the audience would otherwise do, and with it, much of the engagement and pleasure the scene might have generated.
The fix is usually not to cut the content but to find an oblique route to it. What would this character actually say in this moment, given who they are and what they are willing to admit? They might say something apparently unrelated that carries the weight of what they cannot express directly. They might deflect with humour, or attack rather than defend, or go quiet at the precise moment the other character needs them to speak. The emotional content remains. The directness goes.
There are moments where directness is earned and necessary. The scene where a character finally says the thing they have been unable to say for the entire film can be one of the most powerful moments in a story, precisely because the audience has been waiting for it. But that power depends entirely on the indirection that preceded it. The direct statement lands hardest after a long conversation in which nothing was said directly at all.
Screen dialogue tends to work best when it is short. Not uniformly short, not without rhythm or variation, but tending toward brevity in a way that prose fiction rarely requires. There are several reasons for this.
The first is pace. A script that is heavy with long speeches will read slowly and almost certainly play slowly on screen. The camera, the actor's face, the music, the edit: all of these carry meaning in film. Dialogue has to share the load rather than carry it alone. A line that takes thirty seconds to deliver is a line that is asking a great deal of the actor and the audience, and it had better be worth the ask.
The second is that long speeches often indicate a writer explaining rather than dramatising. When a character delivers a lengthy exposition of their backstory, their feelings, or their view of the world, the writer is usually doing work that the story should be doing through action and situation. The impulse to explain is understandable, it feels like clarity, but it almost always produces exactly the problem it is trying to solve: the audience disengages, precisely because they are being told what to think rather than given something to feel.
The short line also has a particular power in moments of confrontation or revelation. When two characters have been circling something for a scene, the moment one of them finally names it can carry enormous weight if it arrives in a few words. The longer the buildup, the shorter the landing tends to need to be. The screenplay is full of lines of five words or fewer that are among the most remembered in cinema. That is not a coincidence.
There is no substitute for reading your dialogue aloud. This sounds obvious, and yet many writers do not do it, or do not do it consistently enough to hear what the page cannot show them.
Reading aloud reveals several things that silent reading conceals. It exposes lines that are grammatically correct but unspeakable: sentences that look fine on the page but cannot be said naturally, that trip the tongue, that require a breath in the wrong place. It reveals where the rhythm is wrong: where the speech patterns are too uniform, where a conversation has no variation in pace or energy, where a monologue has no internal arc. It shows you where the subtext is absent, because lines that carry no subtext tend to feel flat when spoken, no matter how purposeful they appeared in writing.
Better still, find someone else to read it with you. Even a rough table read with a friend or fellow writer will reveal things about the dialogue that you simply cannot hear when you are reading it alone. You know what you intended every line to mean. Another reader does not, and what they do with the lines in performance will quickly show you where the intention and the execution have diverged.
Some writers cast their table reads as carefully as they can, finding people whose rhythms and sensibilities match the characters, not to get a performance but to hear the material in voices that are not their own. Others record themselves and listen back. The method matters less than the discipline. Dialogue that has never been spoken aloud is dialogue that has only been half-tested.
Every script has information it needs to convey to the audience: backstory, world-building, the rules of a particular situation, the history between characters. The challenge is doing this without grinding the story to a halt while the characters explain things to each other that they would, in reality, already know.
The old term for clumsy exposition is 'As you know, Bob': the scene in which one character tells another something both of them clearly already know, purely so that the audience can hear it. 'As you know, Bob, our company was founded in 1987 by your father, who left the business to you on his death.' No one says that. It is exposition wearing a costume.
The most effective expository dialogue arrives in conflict. When two characters disagree about something, the argument naturally surfaces the information the audience needs. When a character is under pressure, explaining themselves or defending their position, the backstory emerges as part of the drama rather than as an interruption to it. The information is still delivered, but it is delivered in motion, as part of something that is already dramatically alive.
Billy Wilder's oft-quoted principle was that exposition should be delivered to someone who does not want to hear it. The resistance of the listener creates the conflict that makes the scene work. It is a small but potent piece of craft wisdom, and it applies far beyond pure exposition: the best conversations in screenplays are almost always ones in which at least one participant would rather not be having them.
If you are writing a first draft, let the dialogue come without too much interference. Get the scene down, get the characters talking, and worry about craft in the revision. The first draft is for finding out what the scene is. The rewrite is for making the dialogue do its job properly.
In revision, read every scene of dialogue with three questions in mind. Does each character sound distinctly themselves? Is there something being said beneath the surface of what is being said? And does every line move the scene forward, either by advancing the conflict, deepening the character, or tightening the pressure? Any line that fails all three tests is a candidate for cutting or replacement.
Dialogue is the most immediately pleasurable element of a script for many writers. It is the place where a particular kind of wit and music can live, where the specificity of a world and its people becomes most vividly present. But it is also the easiest place to hide: to fill pages with lines that sound good but do nothing, that perform depth without having any. The discipline is in holding both things at once: the pleasure of the voice and the rigour of the purpose. When you get that balance right, the dialogue stops being something the audience notices and becomes something they simply believe.