Conventional dialogue in fiction follows a formula: “I am saying something,” said Character X, “and nobody can stop me.” That part in the middle, said Character X, is called dialogue attribution. The phrase attributes the dialogue in quotation marks to Character X, using the verb said. Writers write attribution all the time, but why? Why would anyone ever write dialogue attribution?
Many of the articles about dialogue attribution, or even just about dialogue, focus on the quality of the attribution rather than questioning the existence of attribution itself. The bugbear here, for many good reasons, is the adverb. Writers caution other writers against sentences like this:
“Maybe,” she said vehemently.
The problem with adverbs in dialogue attribution is the same as with adverbs anywhere: they are either redundant or, more dangerously, they mask the presence of a weak verb. So, instead of “said vehemently,” it would seem, I should use a more precise verb, liked “shouted.” But this is no better:
“Maybe,” she shouted.
Why isn’t this better? Well, dialogue is an odd case where the adverb’s appearance masks not the use of a weak or imprecise verb, but an error of another order. “Said” is weak and imprecise, but it is acceptable in dialogue attribution — in fact, much of the conventional wisdom here is that you should only use said, which is fine as far as it goes.
The problem, of course, it that it doesn’t go far enough. But before I get to that, let’s identify the real problem here, which the adverb tries to mask: “Maybe” is not something you say vehemently, not something you shout. (Normally. It doesn't do all the work itself, when maybe it should.)
The dialogue itself is weak and imprecise, and the adverb is attempting, after the fact, to mask the problem, by telling the reader how they should have read the dialogue, since it was too poorly written to have been read properly in the first place.
So what’s the lesson? Restrict dialogue attribution to “he said” or “she said” and so on? It almost sounds good — except that the very presence of dialogue attribution weakens dialogue and muddies character action.
Moreover, dialogue attribution is always unnecessary if dialogue is well-written.
What Dialogue Attribution Does
This may seem obvious, even tautological: dialogue attribution attributes dialogue. Let’s interrogate this further. The function of “Bob said” is to tell us that Bob said something. Why would a writer ever need to do this?
The answer, of course, is if Bob is saying something that another character could have said. There are two basic situations where this might occur. The first is the most common:
Action in the Scene is Too Minimal
In other words: the context of character actions, the geography of the scene, or other factors that boil down to what we shall call action remains underdeveloped. For our purposes, we’ll call action the prose that surrounds the dialogue and describes what characters are doing and where they are located relative to one another, and so on.
When there is enough character action, dialogue attribution becomes unnecessary, because readers always assume that the character taking action in the text closest to the dialogue is the speaker of the dialogue, except in rare cases. For example:
“You’re dead meat.” Bob ran his tongue along the knife’s blade. Sarah winced.
How do we know that Bob said “You’re dead meat”? We don’t. We assume it. Why don’t we assume that Sarah threatened Bob, then Bob defiantly ran his tongue along a knife’s blade, and big-talking Sarah winced in response? Because Bob’s name is closer to the dialogue.
It is really that simple, in most cases. Notice that we don’t need “Bob said.” In fact, let’s rewrite to add it.
“You’re dead meat,” Bob said. He ran his tongue along the knife’s blade. Sarah winced.
Not only does “Bob said” add nothing, it detracts. I have to put a comma after “meat” instead of a period, so it is less declarative. The reader does a split-second replacement of “He” with “Bob” in the second sentence, so that sentence’s tone shifts down to seem ever-so-slightly less threatening.
One of the great “rules” of writing is that every word you add pulls power away from every other word. “Fewer words, better words” is the editor’s credo. (For conventional writing; with experimental work, you may have other editing goals, as we shall see below.) Cutting the dialogue attribution entirely is good editing practice in this sense.
It also helps you avoid a problem in first drafts, which is that if I was used to writing things like “Bob said,” I would probably write this:
“You’re dead meat,” Bob said. Sarah winced.
I don’t actually need the line “Bob ran his tongue along the knife’s blade” — the coolest part of the section. I don’t need to describe the character’s action, because I wrote “Bob said.” So, maybe I won’t. If I write dialogue attribution, I can get away with crap like this. But I cannot get away with this:
“You’re dead meat.” Sarah winced.
Sarah didn’t say it — Bob did — so I don’t want the reader getting confused. So I need boring old dialogue attribution, or I need character action. We’ve already seen which works better. Or I could do this:
“You’re dead meat.”
Here, at least, the blandness of the scene (without Bob’s tongue on the knife’s blade) is more obvious, and something I would catch in revisions, probably. But if “Bob said” accomplishes precisely as much as empty white space, then it is easy to see how worthless “Bob said” truly was.
Now what’s the second problem?
Characters Say Things Other Characters Could Say
If written well, the way a character speaks and the things a character says should be so specific to them that dialogue attribution is unnecessary. We know Bob is speaking because he says things like “The world has a slowness to it and though we rush we cannot overtake this,” while Sarah says stuff like “You sound like how Cormac McCarthy might sound if I bashed in his skull.”
Where’d you get that pistol? she called.
At the gettin place.
Did you buy that thing?
No. I found it.
She sat up on the sofa. Llewelyn?
He came back in. What? he said. Quit hollerin.
What did you give for that thing?
You dont need to know everthing.
I told you. I found it.
No you never done no such a thing.
He sat on the sofa and put his legs up on the coffeetable and sipped the beer. It dont belong to me, he said. I didnt buy no pistol.
You better not of.
Here you can see that McCarthy is only using dialogue attribution to kill time where a pause should go (it is more or less the equivalent of writing “he paused”). Look at all the character information that McCarthy packs into this exchange, subtextually, without stating any of it.
Because of how they talk, we now know (or, rather, we can assume) that they are married, that she’s younger than him, that they don’t have a lot of money, that they live in the American South, that she trusts him but not entirely, that he won’t lie to her even though he doesn’t feel the need to tell her the whole truth.
When Characters Don’t Talk to Each Other
Not only does jettisoning dialogue attribution force you into a position where you need to better describe character action and write better dialogue that is more specific to the characters, it also operates like a poetic constraint that can help you develop experimental dialogue forms.
One of my favourite tricks is threading strings of dialogue together where characters are talking to themselves but appearing to have a conversation on the surface. Here’s an excerpt from a story called “Costa Rican Green”:
“It is good to see you. It’s been so long.”
“It’s good to see you too.”
“The wind is picking up.”
“Did you sleep well?”
“The wind is salty, so strange.”
“Here, let me show you some pictures.”
“I want to stop and write out some postcards.”
“Maybe I can send that letter.”
“This is my dog — but you’ve seen my dog in pictures already.”
“It would be nice to swim tomorrow, if it’s warmer.”
“Have you ever gone snorkelling?”
“I wonder how the water tastes, if it tastes like the wind.”
The next line of dialogue in this scene is quite long. What I’ve done here also is cut all action around the dialogue so that it appears with no context. The result is an ambiguity about who is speaking (although it is more clear if you have read the rest of the story).
In this moment, I am starving the reader of action and scene description because I want to pack it into the chunk of dialogue that appears next. Here are the next lines in the story — note how the action of the characters is being suggested by the dialogue and the scene is described here as well.
“Look at that. Those houses. They’ve fallen into the sea. All that stone. Just broken off like that, one wall standing, the rest below. Come here, look. Do you see? I wonder how it happened. Solid stone, must be cement. See down there, in the water, parts of the floor and the roof, sticking out. There must be five houses. Maybe there were more before. I wonder when it happened. Look out those windows. All those grey walls, then the road and the trees beyond it. Or if you look the other way, into the house, there’s no glass in the window, and nothing inside, just the sky and the sea where there should be a wall. No doors. All the doors must have faced the sea. Think of all those doors now, buried in the sea bed. I wonder if they are opened or closed. Do you think that anybody was hurt? Where did they go, after? I don’t see any other houses around here. We’re too far from the town, just the beaches and the cemetery and these houses. Put that camera down. I wonder about those doors.”
In this moment, the line “Put that camera down” clarifies who has been speaking but not until the end. Of course, the recording character (with the camera) is also the narrator (recording now through text) so I am playing a bit of a metafictive game here. The ambiguity about who has been speaking gets clarified at the end because then it draws attention to the silence of the other character (he picked up the camera and stopped talking) and the dynamic between the two.
You can play around with the form of the dialogue and experiment with its construction more fully when you have more room for ambiguity. Dialogue attribution, by its very nature, robs the scene of ambiguity.
Here is the most normal case of dialogue over-attribution:
“Where did that pistol come from?” Her voice trembled with curiosity as she yelled to him in the other room.
How does a voice “tremble with curiosity”? Never mind the awkwardness of all of this. Notice how McCarthy did it better.
So far, I’ve only discussed how worthless dialogue attribution is, and how it is worse than worthless in that it actively encourages bad writing. But what if I wanted to break my own rule? Here’s how I would do it: by over-attributing, by going to an extreme with dialogue attribution.
Why might I do this? Well, there could be any number of reasons. Here’s an example of how you might over-attribute dialogue for a particular effect:
“No,” she said as she plunged the knife. She said it again. She said it again and then said it again. She said it again and again, and again and again, and she said it again and again and again and each time she plunged the knife.
Here the dialogue attribution starts to operate like a species of onomatopoeia. Sometimes, when you do something, it is best to overdo it, if you do it at all.
Of course, you should never underestimate the power of not doing it at all.