Tag: prose

Posted on in Craft

Dialogue Mechanics: Punctuation and Attribution

Aside from actually figuring out the content of your character’s dialogue, you also need to know how to say who said what—dialogue attribution—and how to punctuate it. This post is all about these technical issues. It isn’t about how to write the content of dialogue, just how to express who is saying it.

Dialogue Punctuation

The current convention is to use double-quotes around spoken dialogue, so this post will be primarily about how punctuation should work within this convention. It’s worth noting that there are other options. For example, you can ditch the double-quotes, and instead use an initial em-dash to indicate speech:

—You’re not going anywhere, she said.

The em-dash for dialogue might raise eyebrows, but it is an option. Andre Alexis, for example, has used this punctuation for dialogue in his work. Another option is to ditch punctuation entirely:

You’re not going anywhere, she said.

If you ditch punctuation, you need to be extra careful with your writing to make sure it is obvious who is speaking. Cormac McCarthy and Margaret Atwood are two authors who have gone in this direction.

We’re going to go with the standard convention of double-quotes:

“You’re not going anywhere,” she said.

This is the standard: The quoted speech ends with a comma inside the double-quotes; the attribution is not capitalized unless it is a proper name; each new speaker sets off a new paragraph.

Those are the basics, so now we can look at different implementations, special cases, and how things might go wrong.

Dialogue attribution mid-sentence:

“And if frogs had wings,” she said, “they wouldn’t bump their ass when they hopped.”

The attribution comes in the middle of a quoted sentence, so we don’t need to capitalize the first word in the second piece of quoted dialogue, which is set off by a comma after the attribution.

This technique has the added effect of creating a subtle/implied pause in the speaker’s speech; the pause isn’t stated, but the reader feels it.

Dialogue attribution between spoken sentences:

“I don’t think that’s a good idea,” Gregor said. “Let’s go to the park instead.”

In this case, we end the attribution with the period. The second quoted dialogue starts with a capital, since it is a new sentence.

Dialogue attribution before spoken sentences:

Hagrid said, “Not in the fire-pit, you dolt!”

The quoted dialogue is a new sentence, so we get a capitalized first letter. The quoted dialogue is set off by a comma after the attribution.

Dialogue attribution with exclamation or question mark:

“Not in the fire-pit, you dolt!” she said.

For exclamation marks and question marks, just pretend that they end with a comma instead.

Implied attribution via action:

Sometimes, instead of explicitly attributing dialogue to a character, we have some action or event described in the same paragraph, and by that means imply who is speaking:

“That’s what I call a brew.” Hagrid stirred the cauldron. “Now where’s my eye of newt?”

For this implied attribution, you have to close the preceding dialogue. It is a mistake to leave the comma, as in the following erroneous construction:

✗WRONG✗ “Now that’s what I call a potion,” Hagrid stirred the cauldron. “Now where’s my eye of newt?” ✗WRONG✗

Dialogue interrupted by narrator:

“When I said I was hoping for a warm welcome” —passing laser-beams singed Darva’s helmet— “this isn’t what I had in mind!”

The interruption belongs to the narrator, and so the em-dashes are placed outside of the quotes. We can read this as a continuous spoken line, without an interruption, and the narrator only functions to add detail. You can use this same construction when the dialogue is actually interrupted, however…

Dialogue interrupted by action:

“Now this—” Darva fired her blaster and ducked behind the barricade “—is what I call a firefight!”

The break in the dialogue can optionally be represented by placing the em-dash within the quotes. This is not a firm rule. I have seen it both ways. It is acceptable to punctuate a break in dialogue using em-dashes outside the quotes, even if it is an action that causes a pause in the spoken dialogue (as in the previous example). The reader can tell from context whether the speech was interrupted.

Dialogue interrupted by another speaker:

“But mom, I just thought—”

“I don’t care what you thought!”

The interrupted dialogue is cut off with an em-dash.

Dialogue interrupting the narration:

Around the campfire, some of the warriors traded bravado— “I once killed two orcs with one swing!” —and others chewed their mutton.

You set off the interrupted narration the same way you would use em-dashes for an ordinary interjection, except that you contain the whole quoted dialogue within.

Dialogue that trails off:

“I just thought that…”

Ellipses indicate a speaker that trails off.

“speaker/attribution” versus “attribution/speaker”:

Should we go with:

“Sure thing,” Aspen said.

Or:

“Sure thing,” said Aspen.

Technically, both are correct. However, unless you have a reason for doing otherwise, you should probably go with the first formulation. The second can sound slightly archaic, which is easier to hear if we replace the named entity with a pronoun:

“Sure thing,” said she.

It’s grammatical, but it sounds archaic. One exception is if we are using a long description in place of a name. It can be awkward to wait for the end of a long description before reaching the attribution:

“It’s just not my day,” the tall man with the overcoat and the handlebar mustache said.

In cases like these, it would be better to put the attribution first, followed by the long description.

Alternative Dialogue Attribution:

The standard dialogue attribution verb is “said”.  Some people like to spruce up their dialogue by using alternatives like “continued”, “replied”, “stated”, “joked”, “answered”, and so on, or by adding adverbs, as in “said tersely”, or “said angrily”. As a matter of subjective taste, I would caution against such alternative dialogue attributions. They have their place, of course, but they are easy to overdo, and easy to do badly. If you want to give an overworked submissions editor a quick reason to put your story in the reject pile, excess or needless alternative dialogue attribution is a good way to do it. There are a few reasons for this.

For the most part, “said” is invisible to the reader, functioning more-or-less like punctuation. The reader passes over it quickly, and it doesn’t get in the way of reading. It keeps the pace quick. By contrast, synonyms like “stated” or “explained” or “answered” or “replied” add syllables and slow pacing without offering anything in return. This category of alternatives should be ruthlessly cut in edits. When you deviate from “said”, you should have a good reason for doing it, because it is always a trade-off with pacing.

Other alternatives attempt to add extra color. Words like “joked” or “pleaded” offer additional shades of meaning. In many cases, these should also be avoided. They are often redundant, since it should be obvious from the surrounding context and the content of the dialogue whether something is a joke or a plea, for example, so you aren’t getting anything by using these terms. They are also “telling” instead of showing—don’t tell us a character joked or pleaded; show us that it is a joke or a plea.

You also see alternatives that specify the manner in which something is said, like “shouted” or “whined” or “wheezed” or “screeched”. In many cases, these should be avoided. If you can’t tell that something was shouted, for example, that might be a problem with how the dialogue or the surrounding passage is written; write the scene and the dialogue so the dialogue sounds like shouting. As for wheezing and screeching, these sort of things can be useful for characterizing a manner of speech, but they need to be used in moderation. If your established baseline is “said”, and suddenly a character “screeches”, it will feel more screechy. Conversely, if you constantly use alternatives, the reader will begin to gloss over them, and they will have less effect. Your ability to use alternatives for effect depends on you using them sparingly and judiciously.

All of this applies as well to adverbial modifiers on “said”. You could write “said tersely”, or you could just write terse dialogue—the terseness should be in the dialogue, so explicitly indicating that it is terse is redundant, and it is also “telling” instead of showing. You could say “said angrily” or “said wearily”, or use any of a variety of emotion-laden adverbs on “said”, but in all cases this will constitute “telling” instead of showing; a better strategy is to write the scene in such a way that the emotion is shown instead of told. If the reader can’t tell that someone is angry or sad or happy without being explicitly told, this might indicate a problem with how the scene is written.

Of course, there are exceptions to all of this. The most important thing is to be controlled and judicious about your use of language. Developing craft is not about mindlessly following rules; it is about understanding the underlying rationales for the “rules” so that you can use whichever techniques are most effective for your story.

Final Words

Thanks for reading, and I hope you found this article on dialogue punctuation and attribution useful!

Posted on in Craft, Uncategorized

“Show, Don’t Tell” explained

This is one of the most misunderstood rules in all of prose fiction.

It’s important to understand that every single sentence you write is simultaneously telling one thing (the explicit/referential meaning of the sentence) and showing others (that which is implied or inferred by the sentence).

So, the famous example from Chekhov of “showing” by way of “light glinting in broken glass” is indeed “showing” the reader the light of the moon, because that is not made explicit but rather inferred by the reader. However, the sentence is “telling” us that there is broken glass and that light is glinting in it.

The general recipe for “show, don’t tell” is this: think about whatever it is you want the reader to feel or imagine or understand, and then don’t write that thing. Instead, write around it, so that the reader can feel what you are getting at by way of implication or inference.

Showing is always more evocative than telling, since the reader supplies the information from their own imagination, which is necessarily more vivid than reading it on the page. Showing uses more of the reader’s brain, and places the target image in their imagination, not just within their linguistic processing.

“Show, don’t tell” applies to all aspects of writing, including theme, tone, character, emotion, plot, setting, description, etc. You can show any of these things, or you can tell them.

For example, to “tell” emotion is to say, “John was angry”. To “show” the same emotion is to have John punch someone, or tell them to fuck off. Likewise we can identify “telling” for character, as in “John was the sort of guy to get angry easily”, for setting, as in “they were standing in a grocery store”, for description, as in “it was really dark”, and so on, all of which can be translated into “showing” statements by picking out the right sort of evocative details—details which imply or allow the reader to infer that which you intended to convey.

Many people wrongly equate “show, don’t tell” with a distinction between “scene” and “summary”. But this is only “show, don’t tell” as it concerns plot. It’s true, to “show” the plot is to write a scene, and to “tell” the plot is to write a summary. But this is only the the principle as it applies to the dimension of plot. “Show, don’t tell” is much broader than that. Similarly, many people wrongly interpret “show, don’t tell” as a caution against info-dumps or “expository lumps”. But exposition is only “show, don’t tell” as it applies to backstory, setting, or other information. Summary “tells” the plot, but it can simultaneously show other things; likewise, info-dumps or exposition, while it “tells” backstory or other information, can show other things. (For example, an info-dump about a historical artifact will “show” something about the narrator, or in the cases of dialogue, about the speaking character.) In both cases, an author might choose to use these techniques to “tell” something while effectively/artfully “showing” something else.

The question isn’t whether to show, it’s first what to show, and then how to show it. If you are writing fiction, there are many things you will want to show, including character, theme, emotion, plot progression, world-building (if you’re writing in the speculative mode), backstory, etc, etc. Ideally, you want to show multiple things with each sentence. A well-crafted story reveals multiple layers through the surface level of the text; a single line of description can simultaneously develop character, theme, and plot, without saying any of them directly. It shows multiple layers of meaning.

Many people wrongly believe that “showing” takes more words than “telling”. This is completely wrong. Effective showing is often shorter than telling. In fact, Hemingway (who along with Chekhov is one of the two most important authors for the “show, don’t tell” principle) talked about the “iceberg” theory of literature, where most of the story is hidden below the surface level of the words. Hemingway uses lots of simple sentences, but buries the emotion and character and theme below the surface, showing the story by hiding it “between the lines”, so to speak.

I think there are probably two reasons why people think “showing” is shorter than “telling”: the first, because they wrongly believe that “show, don’t tell” just means to use scene rather than summary, and of course scene is always longer than summary; the second, because the wealth of examples of “show, don’t tell” that you can find online are people taking simple descriptive statements, like “Sally was friendly”, and then replacing them with additional detail. These examples are indeed a form of “showing”, but it’s a limited aspect of showing that is probably better talked about in terms of “general/vague/abstract” language versus “specific” language, or more generally, the concept of the ladder of abstraction. Showing should in general be shorter than telling, because it is about cutting things out and letting the reader intuit or infer the meaning. Often, during editing, this can literally mean just deleting entire sentence rather than rewriting them. First drafts will often contain pairs of sentences, where the first tells and the second shows, something like: “Frank was distracted and he burnt the toast. He smelled burning toast and rushed to hit the button.” In this example, the first sentence can be deleted entirely, because it is shown by the second. If you go through and delete the first of these kind of paired sentences, your manuscript will be made shorter by following the principle of “show, don’t tell”. In other cases, showing is shorter because fewer words, rightly used, can convey multiple layers of meaning. This is probably easiest to see in the case of punchy dialogue, which can simultaneously convey characters, relationships, and plot.

There are many literary techniques for “showing”.  For example, the techniques of metaphor, irony, understatement, ambiguity, and unreliable narration all depend essentially on what is not stated by the writer. Each of them, in their own way, refrains from telling directly, and instead shows just enough for the reader to comprehend the meaning on their own—these techniques all rely on the reader to do some work to find the underlying meaning, where it’s hidden between the lines. They are all examples of showing, rather than telling.

In literature, the written text includes narrative gaps that are filled in by the reader. According to some critics, in particular reader-response critic Stanley Fish, this is the distinctive feature of literature: plain-language is referential and expository, whereas literary language reveals additional meaning through intentional interpretive gaps. According to this view, “show, don’t tell” is not just advice on good writing—it is the essence of literature. Plain language puts the meaning in the surface level of the words (it “tells”) and literary language puts the meaning in interpretive gaps filled by the reader (it “shows”).

Before writing anything, you should figure out what it is you are hoping to express. You can think of what you want to express as the “target”—it could be a character trait, an emotion, a theme etc. The goal of “show, don’t tell” is to write in such a way that you express the “target” without saying it explicitly.