Picture by Bartek Ambrozik
In real life, it is usually easy to tell who is speaking. We see their lips move. We hear their voice. We are surrounded by signals that help us concentrate on the substances of what is being said.
In good fiction, we want the reader to feel immersed in the story. We want them to feel as if they are actually hearing the words being spoken. But, of course, our reader is not actually there, and does not have any of the usual signals.
In order to get over this we make use of dialogue attribution. In its simplest form, this simply means writing “X said” after the dialogue: “This is dialogue attribution,” X said. The two other most common means are indicating who was spoken using “beats”, short descriptions of thought or action, and structuring the dialogue so that the reader infers who has spoken from the text or the order of speech.
I’ve distilled a list of guidelines that I use now when thinking about dialogue attribution, after lots of useful conversations on Linkedin*. It is not the sexiest subject in the world. But learning about it has improved my writing.
1. Normally you should use “X said.”
Not “X answered” or “X replied”, and certainly not “X shivered”, “X barked” or any other combination X and a verb that doesn’t actually conjugate from “said” or, worse yet, can’t be physically done by the speaker.
It is tempting to substitute colourful descriptive words for “said.” Line after line of X said/Y said looks boring. A couple of the first stories I wrote, when I started this experiment a few months back, were guilty of using descriptive verbs to avoid repetitive dialogue attribution. It feels exciting and dynamic when you are writing: “I’m talking!” He exploded.
But this is really sloppy writing**. It drags the reader’s attention away from the dialogue itself to your excitable descriptive word, and worse yet to a part of the text that is telling the writer how the character feels rather than showing them. The reader should as much as possible feel like they are hearing the conversation themselves, not reading you telling them about a conversation.
An exception is where you need a verb to make the dialogue clearer. X whispered has a place here, as can, as it happens, X replied. But be careful. Be sure you really need it, and you are not just beating your reader over the head with the obvious stick.
Meanwhile, there are better ways to deal with the problem of excessive “he saids” (see beats and inference below). Unless you are absolutely sure, stick with “said.”
2. X said is usually better than said X
I resisted this one at first, because “said X” seemed more natural to me than “X said.” And if I was comfortable with it, why would it matter which way round I put them? I asked Linked in*. This sparked quite a debate. At least one person told me that I should write X said because English is a subject object language, whatever that means. Another explained that the preference for X said comes from journalism, because it makes it easier for editors to move your sentences about.*** Some people told me to replace “said” with exciting descriptive verbs (see above for why I disagree). Finding myself both more informed and more confused, I performed an experiment and went and looked at the last three books I’d read.
I wasn’t surprised to find that Marquez in A Hundred Years of Solitude religiously used “X said”; this was, after all, a man who systematically cut all “-ly” adverbs from his writing. Bank’s Hydrogen Sonata had some interesting bits where he wrote the dialogue between his giant intelligent starships as if it was a play, which was a neat trick to deal with the many hard-to-remember names, but elsewhere he pretty much uniformly used “X said.” Finally, I looked back at Susan Coopers The Dark is Rising, which I thought being an older text would be more in line with the more literary style I imagined I had been cultivating with my use of “X said.” When I counted them up, though, I found that although she used more “said X”’s than Banks or Marquez , X said was her go to attribution.
So my assumption wasn’t looking very sound, and while my investigation was hardly scientific it did get me thinking. It all relates back to the subject/object point. We are used to seeing words arranged in this order, and we are so used to seeing dialogue tagged with “X said” that it fades from our vision. We don’t notice it. We simply notice who is speaking. The reason I had thought “said X” was more common was not because it was more common, but because I noticed it more often.
But being unobtrusive is a good quality in dialogue attribution. We want the reader concentrating on what is being said. We want them to hear the voices. Every time we play with the standard format, we draw the reader’s attention away from the story (even if it is only a little bit). This is why “X said” is generally superior.
Of course, it is a guideline, rather than a rule. “Said X” might be a good way to link into a beat or another action: “’Could you tell it was me speaking?’” Said X as he climbed the staircase.” There are lots of clever things you can do with this more passive phrasing to exploit the generated shift in the reader’s attention, or to alter the pace. But unless you are deliberately doing something clever, better stick with X said.
3. Beats: “Who said that?” X had emerged from the back room.
The term beat has other meanings in the context of writing, but usually refers to a small unit of action, thought or feeling that contributes to the story.
In the context of dialogue attribution, beats can serve a double purpose of both progressing the story and allowing the reader to determine by inference who is speaking. It can also be much more pleasing to read, because most readers like to figure stuff out for themselves. It makes them feel engaged in the story. There’s a whole bundle of cognitive poetics research I should write about at some point to back up that argument, but for the time being I shall simply say “eat that, haters of Steven Moffat’s Doctor Who.”
The danger of beats is that it is not certain who has spoken. If X is walking in on a conversation between Y and Z, it is quite plausible that the Y is the one talking when X comes into the room. It just isn’t as clear as “X said.” Asking the reader to figure it out can snap their engagement, especially if there isn’t actually an answer in the text (we really can’t know if X is or isn’t the one speaking when he emerges from the back room). Moffat haters, I take it back.***
Another issue with beats is that they change the pace of the dialogue; they invite a breather, they break up the beat (hence, I suspect, the name). This makes them an excellent way of dealing with the boredom of endless “X said, Y said”s. Mixing up standard attribution with beats allows you to control the pace of the dialogue and thus the tension of the scene.
4. Inference: “Was that you, X?” “Yes, it was me!”
The big sister of beats. The context of your story and any preceding dialogue gives clues to the reader. If the reader has a clear mental picture of what is going on they may not need a dialogue tag to know who is speaking.
The most obvious example of this is when two people are speaking to each other. After the initial “X said/Y said” the reader can safely follow who is speaking because the speech will alternate. This can get a little dry, however, so you want to think about using beats to break up the rhythm.
Dialogue might also contain dialect, verbal tics, or content that identifies the speaker: when he says “Yes, my master”, you can be sure it is Darth Vader speaking to the Emperor and not the other way around.
Like with beats, relying on context has the danger of causing confusion for the reader. But it can also be smooth, stylish, and obey the writerly virtue of saying no more than needs to be said.*****
5. Use the same names in the same scene
I wrote a story a couple of months ago I which one of the characters was Petty Officer Kendall. After introducing him with his full title, I switched between calling him Kendall, PO, or “the Petty Officer.” Switching between his titles seemed like a good way to avoid boring repetition.
When I gave it to a reader, the feedback that came back was “who is the PO? Is it the same person as Kendall?”
You can change the name you call a character when attributing him dialogue between scenes, perhaps to illustrate a changing point of view or a development in the character. But when the reader is concentrating on the dialogue you should do everything you can to make the attribution clear. That means using the same name every time you attribute dialogue to them in a scene.
I went back through the story and changed all the tags to “Kendall said.” The story was better for it.
Does any of this really matter?
Yes it does. Good fiction defaults to immersion. It succeeds by making us feel like we are there; like we are the ones hiding behind the door as Catherine declares that she has never loved us, or that we are the ones who still dream of Manderley. If you want to distance the reader it should be done deliberately and artfully: confusing them about who is speaking or telling them X is shouting angrily rather than showing it in the words he uses are really crappy ways of doing breaking their immersion. By which I mean it will work, their immersion will be broken, but they won’t like it. The reader may not realise that poor dialogue attribution was the final straw that caused them to put the book down and do something different. Switching to “X said” will not win you any prizes. But, like in basketball, you can’t win if you don’t score the easy points.
* This article would not have been possible without the help of the Linked-in group “Authors, Writers, Publishers, Editors, & Writing Professionals (no religious/ political discussion)”. I doubt I have done justice to their many useful contributions, but I have learned a great deal writing about them.
**Plus, what I have literally said with ““I’m talking,” he exploded” is that directly after speaking the man blew up. Yuck. The reader has to figure out that this is not what you mean, and even after they have they have this silly image in their head. Probably not the best thing.
*** The second answer (handy for editors) is, of course, linked to the first (subject/object language), which is why the journalist and the grammatician****** were both right and I was wrong. It took me a little while to understand that out though.
**** I don’t really. Steven Moffat is a brilliant writer (I wonder how many impressionable young boys had a crush on Julia Sawalha because of Press Gang and Mr Moffat’s writing?) There’s a whole other conversation to be had about how different types of reader want differing levels and even types of complexity, and demanding that your reader draw the greater part of the meaning from inference may put some off. That’s ok though. I like Umberto Eco and struggle with Dostoyevsky, my wife liked Crime and Punishment but found the Name of the Rose pretentious. Both of us like JK Rowling. I can’t get through five pages of a Clive Cusslar novel but he’s one of the biggest selling and most popular authors in the world. Some of us like figuring things out, some of us need things to make sense, and some don’t care either way: your writing style needs to adjust to meet your reader (or, your writing style will attract different readers). This is part of what makes stories good. But don’t diss the mighty Moff, because Doctor Who has never been better.
***** An economical virtue that I completely ignore in this blog. Oh well.
****** Grammatician is excellent word for a person who studies grammar. It was not in my spellchecker, and I had to check it on the internet. There I also discovered “Grammarian”, which is also a very good word, and possibly more appropriate in the sentence, but I was too scared to use it. Grammarian. Brrrrr.
