Interestingly, one of the first things I noticed when I poked around looking at writing advice for dialog was that practically everyone spends a lot of time focusing on the part of the scene that’s not dialog – that is, on the speech tags and stage business. Those things are really important parts of the way you present your dialog, but you can do them to perfection and still have a tin ear for the speaking part.

Dialog is imitation speech. That means that above all else, it has to sound like something a person might actually say, which is why the one piece of advice you see over and over is this: if you’re having trouble with your dialog, read it out loud. It’s good advice, but if you have a truly tin ear, it may not be the place to start.

The other really common piece of advice is to listen to real people talk. Eavesdrop on the bus, in restaurants, at the mall, even take notes if you can get away with it. This sounds like reasonable advice, and it does seem to work for some people; the trouble is that dialog is an imitation of the way people speak, not a transcription of it. It’s a slightly-idealized, simplified model, not word-for-word and um-for-um dictation.

Listening to real people talking can help one get a notion of the differences in syntax and vocabulary and rhythm that make up the elusive thing called “the character’s voice.” I’ve never found eavesdropping to be terribly useful as a way into the writing part, though, because real conversations were never much help when it came to extracting that idealized, simplified model that I needed for my stories. Not even if I went through and cut the ums and ers and digressions and cleaned up half or more of the sentence fragments.

What did and does work, for me, was studying plays, screenplays, and movies. This ought to be obvious, but it wasn’t for me and it doesn’t seem to be for many, many other folks. Shakespeare is particularly useful, because you have not only the scripts, but also multiple films of many of them, which means you can study the dialog on the page and several different ways that different actors delivered the lines. Listening to radio plays, or movies where the picture is turned off so that all you get is the dialog, is also really informative.

Mostly, though, I read plays. All kinds of plays by all sorts of playwrights. I read them out loud with the play-reading group, out loud in my office, silently in my living room.

The thing about plays and screenplays is that the scenes are nearly all dialog. They’re also in a format that means the non-dialog parts – the speech tags and stage business – drop out of the way. A scriptwriter doesn’t have to worry about whether to use adjectives or where to put the speech tags, because the format is going to be “MARIA (angrily): That’s my hat.” So the writer isn’t as likely to be distracted by things that aren’t dialog.

The other thing about plays and movies is that they are already written in that idealized, somewhat simplified imitation-of-real-speech that you want for dialog in a story. This means that you don’t have to sort out which bits of vocabulary and syntax and so on are things that need to be cut (because they make things run on too long) and which bits are part of the specific character’s voice and therefore absolutely necessary.

Once you have a feel for the rhythm and syntax of speech in plays, then you start reading your own dialog out loud. If that doesn’t seem quite enough, try recording yourself saying it and then listen to it, or have someone else read it out loud while you listen. Start with just the dialog – no speech tags or stage business – because that’s the part that has to sound like speech. The tags and stage business and description that goes around the dialog is presentation, and it’s not the part people are talking about when they say someone has a tin ear for dialog. (It is the part they mean when they say someone has a tin ear for syntax or rhythm or narrative or just in general, but that’s not what I’m talking about here.)

The other thing that leads a lot of folks to say “he/she has a tin ear for dialog” is actually not so much about the dialog itself as it is about the characters who speak it.  I have quite a bit to say about it, so I’ll leave that part for next post.

The first common mistake, especially for science fiction writers, is to go to one extreme or the other – either the writer unthinkingly uses nothing but current real-life idioms and catchphrases, like “once in a blue moon,” or the writer uses nothing but their own made-up idioms.

If you look at common, ordinary American speech in 2012, you will find common phrases and idioms that arose long ago, like “keep it under tight rein” or “give it free rein”…but you will also find ones of much more recent vintage, like “I don’t have the bandwidth to do that today.” A future society that uses only those idioms and catchphrases that are currently in use in English implies that nobody has invented a catchy new turn of phrase in the intervening time, that none of the catchphrases or idioms in use in other cultures will migrate into the English language, and that none of the current idioms will shift in meaning. This is unlikely, to say the least.

The second problem is that while people occasionally use clichés, idioms, and other such turns of phrase in their conversations, most of us don’t use them constantly. And a novel gives the impression of things happening very quickly. It is very likely that in real life someone would say “time flies” to one person and then, a couple of days later, “add his two cents” to another conversation. In a novel, those two conversations, days apart, can quite easily take place on consecutive pages. This gives the illusion that the characters are talking in clichés all the time, even when they aren’t. And a conversation like the following would be ridiculous: “Harry! Haven’t seen you in a blue moon.” “It has been a coon’s age, hasn’t it? How’s your aunt?” “Fit as a fiddle. Her son is always in trouble though, and keeps leaving her holding the bag.” “And him born with a silver spoon in his mouth! What’s the world coming to?” “You hadn’t heard? I thought it was the talk of the town.”

The third problem is that realistic dialog in a novel is not a transcript of the way people actually talk. If you were to tape-record a conversation on a bus, or over dinner, and then transcribe it exactly, you would have a lot of boring, unrealistic-reading stuff like “Well, you know, it was, um, that other one – yeah, the pink, and the thing, er, went…I dunno, it, um, went. So I said, like, um, wow, I mean really. But, um, then it stopped.”

Dialog in novels is a model of the way people speak in real life. It leaves out all the “ums” and “ers” and “likes” and “I means” and the rest of the verbal static that people use in real-life conversations. Dialog also tends to be a lot more coherent – you don’t see as many sentence fragments or dangling bits as you do in real-life conversations. And you don’t see as many idioms.

There are several ways of getting around these problems. One author had a modern-day character moving around in time; the people he talked to expressed amazement at his cleverness and wit every time he used a phrase like “time flies when you’re having fun,” because they’d never heard it before. Several books have postulated a future in which different social groups were based on different past eras, and tried to use the language, dress, and social customs of the time period they were re-creating (rather like the Society for Creative Anachronism, or some of the Civil War re-creation groups).

A more common solution is for the writer to invent his/her own idioms, ones that would arise out of the kind of society (different from ours) that the characters live in. In our society, “too many cooks spoil the broth;” in a seafaring or spacefaring culture it might be “you can’t have two captains on a ship;” in a culture of traveling merchants, it could be “the more people involved in bargaining, the worse the deal.” An excellent idea-generator for this sort of thing is IDIOM’S DELIGHT, by Suzanne Brock, which lists a whole bunch of Spanish, French, Italian, and Latin phrases that are just as commonplace and “clichéd” in those languages as ours are in English. (My favorite is Italian – instead of “I see you once in a blue moon,” they say “I see you once every death of a pope.”)

…“and what is the use of a book,” thought Alice, “without pictures or conversations?”

-Lewis Carroll, Alice’s Adventures in Wonderland

Dialog occupies an odd place on the list of fundamental fiction-writing skills. It’s a component of nearly all fiction, but it’s not absolutely necessary (Hatchet and My Side of the Mountain, for instance, both have only one character present for most of the book; there is thus almost no dialog). Many people, like Alice, prefer dialog-heavy stories; others warn sternly about “talking heads” or sneer at dialog-heavy books as being “too easy” or a cop-out of some kind.

Yet dialog is one of the more flexible – and therefore complex – aspects of writing. Dialog can be used to describe people, places, and things, to convey the speaker’s personality and background, to advance or explain the plot, to provide background and backstory – anything that narrative does, in fact. The writer has to pay attention to all the normal concerns, like pacing, in addition to some things like keeping it clear who the speaker is, which is only a problem with dialog. And there are some things, like voice, that become more important in dialog because the writer is dealing with, potentially, as many different voices as there are characters.

In fact, the way characters talk inevitably tells the reader a lot about them. Consider the following untagged talking heads:

“I ain’t doin’ it, and that’s flat.”

“Have I requested any such thing of you?”

“Um, well, I don’t like saying this, but it certainly sounded to me as if you did. Ask, I mean. Though of course I may be mistaken; still, somebody will have to water the roses while you’re away – it’s such a lovely garden, it would be a shame to let it go – and I do think – “

“I ain’t watering no damned flowers. Sissy job.”

“No, no, there are ever so many men who are florists. It’s just like farming, really. Sort of. Isn’t it? Don’t you think?”

“Your defense of my position leaves a great deal to be desired. In the first place, garden maintenance has very little to do with being a florist, and in the second, I have still not requested that he perform any.”

“Good. You got some sense, anyways.”

Now consider what you know or can guess about these people just from their dialog…starting with how many of them are present in the conversation. If I did my job right, it should be pretty clear that there are three people talking, and it should also be clear whether A, B, or C is saying each line. The reader can, I think, make at least tentative assumptions about the relative social class and education level of each speaker, the fact that they aren’t complete strangers, how well they get along, and the general personality type of each. Also, one ends up with a pretty good idea of their various opinions of gardening.

That’s quite a bit to get out of seven dialog exchanges, and there may be some other things in there as well that I’m not noticing because I put it in unintentionally. Theoretically, one could tell many stories using only dialog (and I don’t mean just plays). Normally, though, untagged dialog is a technique that’s used only briefly, for reasons of variation or emphasis or pacing.

The point I wanted to make here, though, is twofold: first, that the most effective dialog is frequently the sort that could work without any speech tags at all (whether or not it has any) because each character has a unique voice that is obviously or subtly different from that of every other character in the book; and second, that the most effective speech tags, description, and stage business are the sort that add something more to that already-effective dialog.

If I’m over-simplifying, I’d say that there are two kinds of things that fall into that second category: stuff that’s already there in the dialog, and stuff that isn’t and can’t be in the dialog. In other words, you can take the personality or emotions or whatever that’s already implied by the way the dialog is phrased, and emphasize it with stage business or a speech tag: “I don’t – I can’t – oh, dear, oh, dear.” Her fingers twisted and untwisted the curtain cord in time with her stammering. This can be perilously easy to overdo, though, and overdoing it often weakens the impression the author wants to make, rather than strengthening it.

Stuff that can’t be in the dialog is, in part, fairly obvious. It’s possible to have one character describing a room or a landscape in detail in his/her dialog, but it’s difficult to justify doing more than once (and if one does do it more than once, it starts looking obvious and overdone pretty quickly). The same goes for commenting on another character’s actions.

Less obvious are things like contrasting tone of voice. “You are a rotten, scheming bastard!” seldom needs a speech tag of “he shouted,” but if the speech tag is “she said admiringly” it is absolutely required.

When I finally did start to think about the way characters talked, I was at first bewildered by some of the advice I was getting. “Characters will choose different words depending on their personalities, cultural background, age, class, education and training, and so on,” I was advised. “Two characters should never say the same thing in the same way.” Then I’d look at a simple statement like “That’s a mistake” or “The house is on fire!” and wonder how else to put it. “That’s wrong” didn’t seem different enough to carry all that freight, and I couldn’t see any of my characters choosing words like “The domicile is ablaze!” (though someone who did might be interesting to write about).

What I didn’t realize for a long time is that I had the emphasis wrong. I thought it was “Two characters should never say the same thing in the SAME WAY,” when I should have been looking at it as “Two different characters will never say the SAME THING in the same way.”

Speech patterns are as much about WHAT is said as they are about the WAY it is said. “Madam, will you do me the honor of granting me your hand in marriage?” and “Hey, baby, why don’t we get hitched?” are both proposals of marriage, but that’s not all they are. There’s a lot more information in each of those sentences than just “Will you marry me?”…and it’s different information, depending on who the speaker is and what they think is important in addition to the basic question they’re asking.

A lot of that additional information has to do with the speaker him/herself. You can tell quite a lot about the two people who are proposing in the paragraph above – the first one uses formal, traditional language and is perhaps a little stuffy, while the second is slangy and informal. One can easily picture the first in a tuxedo, on his knees with a diamond ring in a box, while the second seems more likely to be sporting an untrimmed beard and a tie-dye T-shirt. One can, of course, set up circumstances in-story in which it would be the hippy in the tie-dye shirt using the formal language and the stuffy gent in the tux who’s being slangy, but if all you have is the dialog, that isn’t what first springs to mind.

In any exchange of dialog, each of the characters has a lot more going on than just the basic information they’re supposedly telling the other person. They have personal agendas; they have emotional reactions that they may not be able to – or want to – hide; they have ingrained ideas about the proper way to behave and speak (both in a grammatical sense and in terms of good manners). All of these things will affect what they say and how they say it.

What and how are often a lot harder to distinguish than first appears. When I made my first deliberate foray into giving characters different speech patterns (in The Seven Towers) I thought I was concentrating on how they spoke: Amberglas in a rambling, roundabout fashion; Vandaris using colorful swears, Ranlyn in a slightly archaic formal style, etc. But in order to ramble or swear or be archaic, I had to add things to whatever the basic underlying dialog was. And what got added depended on the character.

Amberglas  couldn’t just say “Don’t move; you’re injured.” If I wanted her to ramble, I had to add some things for her to ramble about. So what could have been a short, simple, straightforward line of dialog became “You really shouldn’t do that, especially if you’re not feeling well, which I can see you aren’t, what with that hole in your side and so on.  I assume you realize that, though one can never tell.  People can be so very odd.  There was a man I used to know, who always wore his boots on the wrong feet for one day out of every month.  So I thought I’d mention it, in case you didn’t.”

What people say isn’t just syntax and word choice (though those are an important piece of how they say things). It’s about what’s important to them – manners, image, people and things they’re worried about or afraid of, attitude toward the listener, and a host of other things. The more urgent the situation, the more of this stuff gets stripped out of the dialog, but one can’t write an entire book with people saying nothing but “Help!” or “Fire!” or “Duck!”

If you haven’t ever thought about this stuff before, syntax and word choice are a good place to start – things like having one type of character use shorter, less complex sentences and words of fewer syllables often work well.  Look at the way Shakespeare did it:  you wouldn’t mistake any of the rude mechanicals’ speeches for those of the nobility. Or you can try doing what I did – picking a cast of characters several of whom have exaggerated or extreme speech patterns that are very different from each other. Nobody else talks like Amberglas, so it was really easy to tell if I’d gotten her dialog wrong or if her style was creeping into someone else’s dialog inadvertently.

Nowadays, I do this mostly by instinct, and on a much less obvious (I hope) level. But that’s where I started.

The terminology has bled over from acting and visual media into prose writing, but it means the same thing. The difference comes in how a writer indicates the pause. An actor hesitates; a writer has to actually say “he hesitated.” A director has the camera cut away from the fight for just a second to show the horrified look on a bystander’s face; the exact same interruption for a writer runs into all sorts of viewpoint and pacing considerations (Would the first-person narrator actually notice a bystander’s reaction when he’s dodging punches? Would it distract him if he did? Is it too flat and generic to say “The bystander looked on in horror”? Is it going to be too much of an interruption to give a couple of sentences or paragraphs of description of the bystander?)

On the other hand, writers have a couple of useful tools that actors don’t. Punctuation, for instance. Standard punctuation is meant to indicate differences in tone and timing; there’s a reason that the period is also called a “full stop.” Commas are shorter pauses – just enough for a breath – while semi-colons and colons indicate longer breaks, dashes more of an interruption, and ellipses a hesitation or fading out.

Punctuation gets even more useful for indicating beats when writers use it in non-standard ways. This has to be done with a light hand, or it looks as if the writer is simply ignorant of standard punctuation rather than doing it on purpose. Still, the ability to write “‘Put. It. Down.’ He scowled – she lifted it higher – a flurry of motion; a crash; a fading cry…then silence, and curtains blowing through the broken fourth-story window.” makes it all but impossible for a fiction writer to stick strictly to correctly punctuated sentences. It’s hard to pull off effectively, though, if one doesn’t know the standard rules and usages to begin with. For those who are doubtful, or who want an engaging refresher course, I recommend Karen Elizabeth Gordon’s The New Well-Tempered Sentence.

Sentence fragments and short paragraphs can also provide beats, especially when they are a) not overused and b) in sharp contrast to whatever is around them. That is, a sentence fragment in the middle of an action paragraph composed of relatively short sentences will provide a less strong beat than one that occurs in the middle of a long description. Compare:

He dodged left. The bear dodged right. He ran for the tree. The branch was just out of reach. He jumped. Missed. And the bear was on him.

A row of ornate picture frames lined the back of the mantelpiece. Most of the pictures were of a single person in dark, old-fashioned clothes, but two were of couples, and one showed a family grouping of three adults and five children. A white candle-stub stood in front of each picture, trailing cold wax and bits of blackened wick across the gray stone. Except for one. At the far end, half-hidden behind the portrait of a stern-faced matron in black, stood the picture of a ten-year-old boy in a baseball uniform, glaring at the unseen photographer…and at the empty space in front of the picture where his candle should have been.

In dialog, the speech tag can act as a beat, especially if it is longer than “he said” and/or comes at the beginning or in the middle of a line. “You insist that I say it? All right, then,” he said doesn’t have a beat in it (this is what people really mean when they claim that “said” is invisible as a speech tag). But “You insist that I say it?” he said. “All right. then.” has a very short beat in the middle, because the dialog is interrupted just a little, even if it’s only by “he said.” And “You insist that I say it?” He looked down. “All right, then.” has a longer beat, because the reader has to switch from dialog to the character’s actions and back.

Beats in dialog can come at the beginning or the end of a line, too, to indicate the pacing and the rhythm of the conversation. There are a couple of things to watch out for here: one is to not get carried away and put a beat somewhere in each and every line. Another is to vary your placement. Even if it’s the sort of conversation where there’s a dramatic pause between several lines in a row, you can make it look varied by putting the first beat at the end of the first line and the second beat at the beginning of the third line: “I’m coming with you,” she said firmly, and waited.//”You’re not going to give up on this, are you?”//She smiled. “Do I ever?” reads more smoothly, in most cases, than “I’m coming with you,” she said firmly.//”You’re not giving up on this, are you?” He sighed and shook his head.//”Do I ever?” she said with a smile.

The two problems that caught my attention were said-bookisms and character tags (aka epithets). Said-bookisms are all those substitutes for “said” that some writers use obsessively in their speech tags – he shouted, she demanded, he stated, she whispered, etc. Character tags are those little descriptive phrases that replace the character’s name when the writer needs to identify a character by more than a pronoun – “the brown-eyed man,” “the younger woman,” “the scarred man,” “the second undergraduate.” They can also be little bits of stage business, habits that the character has that supposedly distinguish him/her from other characters – popping his gum, fiddling with her cigarette, flipping his baseball hat, twirling her knitting needles.

Usually, these two techniques are treated separately in how-to-write advice, with said-bookisms being dealt with under dialog and character tags under either characters or narrative/description. But they’re both techniques for doing the same thing – labeling something (a line of dialog or a character) so that whatever is going on is clear – they’re both extremely useful when used correctly, and they’re both overused or misused all too often.

Part of the reason for the mis- and over-use, I think, is that it doesn’t occur to a lot of people that it is just as possible to overuse a technique as it is to overuse a specific word. It usually takes a little longer for the reader to notice and get irritated by an overused technique, but the aggravation factor tends to escalate rapidly after that. And once a reader is sensitized, she/he is likely to find even appropriate, innocuous dialog and character tags annoying, possibly to the point of giving up on the story.

Fixing the problem begins, as always, with diagnosis. If you can’t see it, you can’t fix it. And since these two problems are related, whenever you catch yourself doing one, it’s probably worth checking to see if you’re doing the other.

Once you realize what you’re doing, the first thing to do is consider why you did it. If the only reason for choosing a different word is that you are avoiding “said” or “George” or “she,” then delete the substitute and put back whatever you were avoiding, and see how it reads. If it still makes you twitchy, find an alternative technique – rephrasing for clarity, using stage business instead of a dialog tag, or whatever.

Sometimes, though, it’s important that a character mumbles or shouts or whispers or whatever. If it is, leave the tag and get as much variation as you can by changing where it goes in the line – there’s a very slight difference in how a given line will read, depending on whether “he mumbled” is at the beginning, in the middle, or at the end of the dialog.

And consider the paragraph that reads: “The captain stared out the window at his new spaceship, waiting for someone to speak. The tall, black-haired man shifted uncomfortably as the silence stretched. Finally, John said, “Looks better than I expected for a refurbished tanker.” The ex-navy man nodded once as if to emphasize the words.”

If that author was simply trying to avoid “he” or “John,” then this would be a lot clearer if all those character tags except the first were replaced with “he” (unless there really are four different people standing around and not just tall, black-haired, ex-navy Captain John). But if the author was trying to work in a bunch of description and background information without stopping to info-dump…well, it didn’t work very well here, but taking it all out is clearly not going to do the job.

So we’re back to rephrasing. “Tall, black-haired, ex-navy Captain John stared out the window…” would, I think, be somewhat better (and certainly less confusing) than the original paragraph, but it does rather overload the opening with adjectives. I’d prefer to start with the name – “Captain John stared out the window…” – and then fill in physical description either in one blunt, straightforward sentence – “He was a tall man with black hair and gray eyes, whose rigidly erect carriage proclaimed him ex-navy” – or by working it into stage business or internal dialog (“He ran his fingers nervously through his black hair as the silence stretched, wondering if they’d given him this piece of junk because he was ex-navy.”)

The thing about said-bookisms and character tags is that they’re relatively easy to do. It took me a lot less time to write the first example paragraph than it did to do the “fixed” versions of the sentences. This means that even experienced writers are likely to find things like this creeping into their first drafts, especially when they’re writing fast to get something down before it evaporates, and because they are actual techniques (and not just mistakes), they can’t just be automatically taken out in the rewrite. One has to consider them carefully and decide. It can be a right nuisance, but it’s worth the effort.

Slowly but surely, I got better at making my characters’ speech sound more natural. At first, they all still sounded pretty much the same, but after a while the voices started to be more individual. The process was both gradual and exaggerated – early on, I had one or two characters per book who had strong, unmistakable voices (nobody would confuse Telemain or Amberglas with any other character in Talking to Dragons or The Seven Towers, respectively), but everyone else still used the same speech patterns.

As I worked on it, I got better at making more subtle distinctions between my characters’ speech patterns. A lot of it was instinct – as I got more sensitive to distinctions in speech, a particular line would “feel wrong” for a particular character until I rephrased it. And then I hit dialect.

Dialect, according to the dictionary definition, is a variation on standard speech that has its own grammar, vocabulary, or pronunciation. It’s the pronunciation part that drives writers (and sometimes readers) to distraction.

Pronunciation is an integral part of speech, and it’s especially important for dialect. Yet non-standard pronunciation is really difficult to render on the page (unless you use the International Phonetic Alphabet, which few readers are familiar with). Oh, there are a few things that work pretty well – a character who drops the final “g” or initial “h” on words like “writin’” or “‘ospital” isn’t hard to show. But it quickly gets murky after that. Phrases like “Whatcha doing?” and “kinda hard” work on the page, but they get very old very quickly. If you use them with too heavy a hand, they can really turn off a lot of readers, even if they are the only non-standard speech your characters use. And when you get to full-blown phonetic respellings like “I wad be laith to rin an’ chase thee/ Wi’ murd’ring pattle!” (from Robert Burns’ poem), they can be practically unintelligible.

The thing people forget is that, like everything else in writing, dialect is mostly illusion. It’s important that it be convincing, not that it be an accurate reproduction in every aspect and at all times.

Phonetic dialect has actually got two strikes against it:  the difficulty readers have in reading it (see Burns, above) and the fact that different readers will “decode” the phonetic dialect in different ways, no matter how hard you try to make it clear. People speak with various regional accents, and any respelling is going to be filtered through those accents. For someone who speaks with a Southern accent, “lakh” is not a phonetic respelling of “like;” “like” is how you spell that word that Northerners pronounce “lyke.” At best, a phonetic rendition of a Southern accent is not going to work for them; at worst, they’ll find it actively insulting. And it is generally a very bad idea to insult a sizeable chunk of one’s potential readership.

I did a bunch of experimenting and came to the conclusion that by and large, dialect works best on the page when I use non-standard syntax and sentence structure, rather than trying to respell it.  Nobody had any doubt that my character Renee D’Auber was a Frenchwoman with a noticeable accent, yet she does not speak one word of French or have one bit of phonetically respelled dialog anywhere in either of the two books she appears in.  Huckleberry Finn speaks with a pronounced dialect, but only about one word per page is respelled (and since the book is first-person, everything in it counts as dialog for these purposes).  Manny in THE MOON IS A HARSH MISTRESS likewise speaks in an odd syntax, but not in a respelled one.  Keith Laumer was a master at this technique – every one of the many alien races in the Retief stories uses a different scrambled syntax. They end up all being clearly and obviously aliens speaking an alien language, yet their dialog is seldom hard for the reader to understand.

There are, of course, exceptions. One of the more obviously useful ones is if you have a minor character whose accent is so thick that neither the viewpoint character nor the reader is supposed to understand what he’s saying without paying careful attention. Some writers even play with this if the minor character starts recurring regularly; they’ll lighten up on the respelling as the viewpoint character gets more used to interpreting the accent, but they keep the syntax scrambled as a reminder that the character is speaking with an accent.

The first most important thing to remember about dialog is that it is a model of speech, not a transcription. I’ve talked about that before on this blog, so I won’t repeat myself in detail, but I think it’s worth at least mentioning here.

The second most important thing to remember about dialog is that it is communication between two or more characters. This means that it is almost always made up of short exchanges, back and forth. Unless one of your characters is giving a lecture, like the detective in a classic murder mystery doing his summing-up, you should expect a page of dialog to have paragraphs that are mainly one to three lines long. There’s usually lots of white space as a result; in fact, one of the classic tests for whether the characters’ speeches are running on too long is to print out a page and tape it to the wall, then walk across the room so that you can see the pattern of the paragraphs and how much white space there is on the page. These days, you can get the same effect by reducing the font size:

Description vs. dialog

Above is an example. On the left is a page of descriptive paragraphs; on the right, a page of dialog. Shrinking the font makes it instantly obvious which is which – and you can see immediately if your dialog is bogging down in long speeches, and take steps to break it up.

The second classic trick for checking your dialog is to read it out loud. This lets you know whether it sounds right in general; it also is an easy way to identify tongue-twister phrases that no one would actually ever say.

If you’re having trouble figuring out how to do dialog generally, try reading some plays or screenplays. Out loud, so you are seeing and hearing the words at the same time, and can get a feel for how the words-on-the-page work when spoken aloud and vice versa. If you really want a workout, get hold of the screenplay for any movie that has lots of dialog, read it aloud, and then watch the movie while following along with the script. Even if you’re not having trouble, paying a little extra attention to passages of dialog in your favorite movies and novels will very likely give you some useful ideas.

The next thing to think about is the difference in the speech patterns of your various characters – the way each particular person phrases things, depending on their individual personalities and backgrounds. You can do this either by consciously coming up with speech tics (like having a character who never uses contractions, or who always ends their sentences with “yeah?”), which can be effective in small doses but which gets really annoying to read when every character in a story has one, or you can come up with broader ways of distinguishing your characters’ voices (Shakespeare had all his noblemen speak in unrhymed iambic pentameter, and their servants and more ordinary people just any-which-way. The lyricist for Man of La Mancha gave Don Quixote complex sentences and syntax ["I am I, Don Quixote, the Lord of La Mancha; destroyer of evil am I!"] and his servant short, simple sentences and no words of more than two syllables  ["I'm Sancho! Yes, I'm Sancho! I follow my master to the end!"])

Or, you can just look at different speech patterns in real life.  Take the same sentence of dialog/information, and rephrase it in as many different ways as you can:

“I think you’re making a mistake.”

“That’s wrong, dumbo.”

“I believe, sir, that you are in error in this instance.”

“Um, do you think…I mean, is that really the way you want to do that? Because it doesn’t look quite right to me.”

“That ain’t no way to do that there thing.”

“Kiddo, you got that upside down and backward.”

“I’m afraid that’s not going to work.”

“A guy could have some problems, doing things that way.”

“You’re screwing up again! Honestly, can’t you do anything right?”

…and on, and on.

I was on my third book before I started trying to do this consciously, and my first few efforts were exaggerated (Telemain in Talking to Dragons, Amberglas in The Seven Towers) because it was the only way I could be sure I was keeping them consistent. More subtle variations took me longer to get the hang of. Most of the time now, speech patterns and character voices are automatic  for me – I know when I’ve used a word or a turn of phrase that a particular character just wouldn’t say, that’s all, so I fix it immediately. But at the beginning, it required a lot more conscious attention. So don’t worry if it takes a while.

But dialog in fiction isn’t a transcription of actual speech. It’s not even an imitation of real speech. It’s a model of the way real people talk to each other, a model that’s both simplified and exaggerated so that it can be clear and read easily while still sounding “real.” Real conversation is full of ums and ers and other placeholders. It digresses and repeats as the speaker tries to order his thoughts and find the right way to phrase them. We’re willing to put up with that in person because we have no choice…and besides, most of us do it ourselves. In fiction, though, most of us won’t put up with the rambling way we talk in real life.

On the other hand, dialog is still speech, not narrative. It is therefore a lot less formal. This means the writer has a lot more leeway to use things like sentence fragments, improper syntax, unusual punctuation, and phonetic spelling. Sentences tend to be shorter and less complex. Contractions are welcome. The rhythm of the sentences is much more important; it’s a large part of what makes dialog feel like real speech. When I was writing “Roses by Moonlight,” I used the same Shakespearian trick I’d first learned writing Snow White and Rose Red, and marked all the rhythms of the Faerie Queen’s dialog to make sure she spoke in iambic pentameter. Even though, for that short story, I kept her word choices modern, the Shakespearian rhythm gave her just enough of a different feel from the other characters (I thought) to make her stand a little apart the way I wanted.

Dialog isn’t just what the characters say; it’s how they say it. A lot of the difficulty people have in writing dialog comes in trying to get the delivery across. The first, most obvious, and usually easiest way is to just tell the reader how the line is being said by adding an adverb to the speech tag:  “Nice job,” he said smoothly. Unfortunately, this method is so obvious and so easy that it has been overused to the point where one of the common “rules” for writing has become “Never use adverbs in speech tags.”

I, of course, wouldn’t say “never.” Adverbs have their place in speech tags, and one of them is to indicate when the character’s tone of voice and/or demeanor is at odds with what he/she is saying: “That’s the ugliest thing I’ve ever seen,” she said admiringly. What you really don’t want are adverbs that don’t actually add any new information: “You’re out!” he shouted loudly is redundant; “You’re out!” he shouted hoarsely isn’t.

Sentence fragments, punctuation, syntax, etc. are all tools the writer can use to convey the way a line of dialog is delivered. So is the placement of the speech tag and/or stage business. Consider the following:

“What have you done? I don’t believe you. You’ve ruined everything!” he said, taking the report and staring down at it.

He took the report and stared down at it. “What have you done, I don’t believe you, I don’t believe you, I don’t – you’ve ruined everything!”

“What have you done?” he said, taking the report. “I-I don’t believe you.” He stared down at the pages. “You’ve…you’ve ruined everything!”

“What have you done?” He took the report and stared down at it. “I don’t believe you. You’ve ruined everything!” he said with a grin.

The first line groups all the dialog together, with speech tag and stage business at the end. It implies some simultaneity of action and dialog, because “he” is presumably taking the report at the same time as he’s speaking, but that’s about all. In the second line, with the stage business first as a separate sentence, it’s clear that first he takes the report and looks at it, then he panics and speaks. The panic is conveyed by stringing all the dialog sentences together with commas as a run-on sentence and repeating the middle one; it gives the feeling that the speaker isn’t pausing for breath during his denials.

The third line builds slowly, and the speaker comes off as stunned, rather than panicked.  First the speaker is worried (indicated by the emphasis on “done”). Then he takes the report and is in denial (that stammering – obviously, this is bad news). Then he stares down at the pages and it slowly sinks in (the elipses indicate more of a pause between the repeated words, compared to the n-dash in “I-I don’t believe you” earlier). The last line manages to completely reverse the implication of the first three (that this is someone getting bad news) by emphasizing “believe” and adding that grin at the end – obviously, this is someone who agrees that everything should be ruined. (And in context, in a story where we already know who is talking to whom and what is being ruined and why, it might not even need the grin to be clear.)

In all three of the last lines, the pacing of the dialog is conveyed by the way in which it is punctuated and broken up. Emphasizing different words by italicizing them, repeating words, deliberately using run-on sentences, breaking up the dialog with speech tags or stage business in different places – each one changes the implied delivery.

First-person viewpoint is the “I” viewpoint: “I hate pickled beets. I’ve always hated them. But Ma thinks they’re good for what ails you, so whenever I’m sick, I get pickled beets.”

A lot of people jump straight to first-person when they start writing, because it looks easy. For quite a while, first-person was so over-used by beginning writers that it got a really bad reputation as something only an amateur would try. There are still traces of that around, some places.

But first-person isn’t as easy as it looks, and there are a lot of possible varieties. “Plain” first person is the most common – something written as if the reader is riding along in the narrator’s head. There’s the subtly different form in which the narrator is writing everything down immediately after the fact (or years later). Then there’s the as-told-to, where the first-person narrator is telling the story to someone (possibly the reader; possibly another character) and the reader is listening in. Diaries, letters, memoir, stream-of-consciousness - all different formats requiring slightly different approaches, but all first-person.

The thing that’s most difficult for a lot of writers to grasp about first person is that they are not the putative narrator. When I say “I did this or that” in normal everyday life, I mean me, the person currently sitting here typing. But when my first-person narrator says “I did that,” the “I” doesn’t mean me-who-is-typing. “I” means the character.

This is so obvious that to most folks it goes without saying. But if one doesn’t say it or think it or pay attention to it, one is likely to find that habit takes over. All my life, “I” has meant me-who-is-typing, and that’s a lot of habit to overcome. It’s no wonder that a lot of first-person narrators sound (and think and act) a lot like their authors. (It is also no wonder that a lot of readers leap to the conclusion that anything written in first person is autobiographical, or at least reflects the writer’s opinions and errors of knowledge, rather than the character’s – but that’s a rant for another time.)

It can help to pick a first-person narrator who has a strong voice of their own – one that is unlike the author’s natural voice. It can also help to pick a character who is significantly different from the author in some way – age, sex, ethnicity, ability/disability, etc. But these things only help if the author thinks about them and the ways they’ll affect the character’s voice and opinions and attitudes; when the author doesn’t think, you get the young black woman protagonist who sounds oddly like a middle-aged white author (and who more than half the readers don’t even realize is black until nearly the end of the book. If then.)

A strong voice helps because first-person is written in the voice of the character – in a lot of the varieties, the narrative is supposed to sound like dialog, like the viewpoint character telling you the story. A question always comes up when the viewpoint character has an accent or uses dialect or pidgin as their normal speech pattern, because it is a writing truism that too heavy a hand with dialect or phonetic respelling can make something almost unreadable (the poetry of Robert Burns, anyone?).

But dialog in any book isn’t an accurate transcript of the way people really talk. Dialog leaves out the ums and ers and most of the sentences that trail off into nowhere and a lot of the digressions and speech tics that happen in real life conversations. It’s a model of the way people talk…and first-person narrative is even more so.

When you’re writing first-person, you are inside that character’s head (or nearly) all the time. People don’t sound to themselves as if they have an accent. Inside their heads, that’s just how everyone talks. I’m not the one with the accent; it’s my friends from the deep South, from New England, from Scotland who have regional accents. The French I speak in my head sounds just fine to me; it’s everyone else who knows instantly from my accent that I’m a native English speaker. So a writer can skip most of the phonetic respelling aspects of doing accent in narrative, which instantly makes everything a lot more readable, and stick with word order and idiosyncratic word choice to convey the narrator’s speech patterns.  (This also works well in many cases for the dialog of characters other than the narrator who have accents.)