Others liked the book, too—and Pohl was pressured for a sequel. But the second book, Beyond the Blue Event Horizon, fell flat on its face. Why? Because Pohl had to shoehorn the character he’d built for a very specific job into a different story. Robinette, absolutely perfect for Gateway, was a fish out of water in the follow-up story about the discovery of a human child on an ancient alien space station.
Clearly, your character must fit your premise—but it’s also important that you not make the fit too comfortable.
Everybody knows Steve Austin, the fictional test pilot who lost an arm and both legs in an aircraft crash and was rebuilt with super parts so that he could undertake secret missions. Austin first appeared in Cyborg, a mediocre novel by Martin Caidin, and was played by Lee Majors in the wonderful, Hugo-nominated movie The Six Million Dollar Man.
Why was the novel just so-so but the movie glorious? Simple. In the novel, Steve Austin was a colonel in the United States Air Force. When he was asked to undertake his first mission as the bionic man, he told his new secret-agent bosses, “You have a job to do. It’s serious, in many ways it’s dirty, in some ways it stinks, but having worn the blue suit [an Air Force uniform] for a long time, I understand and even appreciate what you do. You will receive my absolute cooperation.”
Ho hum. Screenwriter Henri Simoun saw that Caidin had missed the essential conflict. For the movie version, he changed Colonel Austin to Mister Austin, one of six civilians in the U.S. astronaut program. Simoun’s Austin fights those who are trying to make him an obedient little robot every step of the way—making for much better drama.
(When The Six Million Dollar Man became a TV series, the producers went back to Austin being an Air Force officer, and the show degenerated into mindless adventure.)
I almost made the same mistake Caidin did in my novel The Terminal Experiment, which is about the discovery of scientific evidence for the existence of the soul. My first thought had been to have a protagonist who had undergone a metaphysical bright-light-and-tunnel near-death experience. But that would have been absolutely the wrong choice. A person with that background would be predisposed to believe in the existence of the soul, accepting any proof too readily. No, what was called for was a skeptic—someone who had stumbled on the existence of the soul while looking for something else, and who would be bothered by the discovery. The lesson is simple: your main character should illuminate the fundamental conflict suggested by your premise.
And, of course, that means that you shouldn’t start with a character and then go looking about for a story; it’s a lot easier to do it the other way around. First, come up with your premise (for instance, “I want to write about a telepathic alien who can read subconscious instead of conscious thoughts”). Then you ask yourself who could most clearly dramatize the issues arising from that premise (“There’s this guy, see, who’s been suppressing terrible memories of the suicide of his wife”).
After that, head for your keyboard and build the character to your specifications, for that one specific job. (In this case, the story has already been done brilliantly; it’s Solaris by Stanislaw Lem.) Of course, you have to add subtleties and quirks to give your character depth, but if you do it right, only you will ever know that underneath the real-looking skin, your hero is actually a made-to-measure robot…
Point of View
New writers are often baffled when trying to choose a point of view for their stories and novels. But, actually, the choice is easy. Over ninety percent of all modern speculative fiction is written using the same POV: limited third person.
“Third person” (“she did this; he did that”) means the story is not told in first person (“I did this”), or the always-irritating second person (“you did this”). That’s easy enough. But what does “limited” mean?
It means that although the narration refers to all the characters by third-person pronouns (he, she, it), each self-contained scene follows the viewpoint of one specific character. Consider this example, which is not limited but rather is omniscient third person, in which the unseen narrator knows what all the characters are thinking:
“Hello, Mrs. Spade. I’m Pierre Tardivel.” He was conscious of how out-of-place his Quebecois accent must have sounded here—another reminder that he was intruding. For a moment, Mrs. Spade thought she recognized Pierre.
In the opening of the paragraph, we are inside Pierre’s head: “He was conscious of how out-of-place…” But by the end of the paragraph, we’ve left Pierre’s head and are now inside another character’s: “Mrs. Spade thought she recognized Pierre.”
Here’s the same paragraph rewritten as limited third person, solely from Pierre’s point of view.
“Hello, Mrs. Spade. I’m Pierre Tardivel.” He was conscious of how out-of-place his Quebecois accent must have sounded here—another reminder that he was intruding. There was a moment while Mrs. Spade looked Pierre up and down during which Pierre thought he saw a flicker of recognition on her face.
See the difference? We stay firmly rooted inside Pierre’s head. Pierre is only aware of what Mrs. Spade is thinking because she gives an outward sign (“a flicker of recognition on her face”) that he can interpret.
Think of your story’s reader as a little person who rides inside the head of one of your characters. When inside a given head, the reader can see, hear, touch, smell, and taste everything that particular character is experiencing, and he or she can also read the thoughts of that one character. But it takes effort for the little person to move out of one head and into another. Not only that—it’s disorienting. Consider this:
Keith smiled at Lianne. She was a gorgeous woman, with a wonderfully curvy figure.
All right: we’re settling in for an encounter with a woman from a man’s point of view. But if the next paragraph says:
Lianne smiled at Keith. He was a handsome man, with a body-builder’s physique.
Hey, wait a minute! Suddenly we’ve jumped into another head, and immersed ourselves in a whole ’nuther set of emotions and feelings. Not only have we lost track of where we are, we’ve lost track of who we are—of which character we’re supposed to identify with. Although at first glance, omniscient narration might seem an ideal way to involve the reader in every aspect of the story, it actually ends up making the reader feel unconnected to all the characters. The rule is simple: pick one character, and follow the entire scene through his or her eyes only.
Of course, we usually want some idea of what the other characters in the scene are thinking or feeling. That can be accomplished with effective description. To convey puzzlement on the part of someone other than your viewpoint character, write “he scratched his chin” or “she raised an eyebrow” (or, if you really want to hit the reader over the head with it, “she raised an eyebrow quizzically”—“quizzically” being the viewpoint character’s interpretation of the action). To convey anger, write “he balled his hands into fists,” or “his cheeks grew flushed,” or “he raised his voice.” There are very few emotions that aren’t betrayed by outward signs.
Still, in real life, there are times when you can’t tell what someone else is thinking—usually because that person is making a deliberate effort to keep a poker face. If you’ve adopted the omniscient point of view, instead of a limited one, you can’t portray such things effectively. Here’s a limited point of view:
Carlos looked at Wendy, unsure whether he should go on. Her face was a stony mask. “I’m sorry,” he said again. “So very sorry.”
That’s much more intriguing than the omniscient version:
Carlos looked at Wendy, unsure whether he should go on. Wendy thought Carlos had suffered enough and was going to forgive him, but for the moment she didn’t say anything. “I’m sorry,” he said again. “So very sorry.”
In the former, we feel Carlos’s insecurity, and we have some suspense about how things are going to turn out. In the latter, there is no suspense. (And, of course, omniscient narration is death—if you’ll pardon the e
xpression—in mystery fiction: the reader must be kept ignorant of what the various suspects are thinking, or else it will be obvious which one is guilty.)
Note that I’ve suggested keeping in one character’s head for each individual scene. However, you can freely switch viewpoint characters when you change scenes (either at the end of a chapter, or with a blank line within a chapter). Many novels have separate plotlines intertwined, with each of them having its own viewpoint character. But what happens when individuals who have been viewpoint characters in disparate plotlines come together in the same scene? Whose POV do you choose then?
In most cases, it’ll be whichever one is at the heart of the action of that particular scene. But there are exceptions. One big one is when someone who has been a point-of-view character is about to die. See, the central conceit of modern fiction is that it’s actually a form of journalism: the tale you are reading is an account of something that really happened, and the author’s job has simply been to interview one witness per scene to the events being described. Well, if your main character dies in a scene, how did he or she subsequently relate his or her feelings to the journalist-author? Even if the dying character has been your viewpoint character throughout most of the story, it’s best to be inside another person’s head as you watch him or her expire.
(One of the great violations of the journalistic-storytelling model comes from the movie Citizen Kane, which, ironically, is a film about journalism: the whole movie revolves around trying to discover the meaning of Charles Foster Kane’s dying word, “Rosebud.” But the film clearly shows Kane dying alone, with no one witnessing him saying it. Unless you’re a genius comparable to Orson Welles, don’t try to get away with this in your own fiction.)
There are other times when you’ll want to choose someone besides your protagonist as the POV character for a scene or two. No person really knows how he or she is perceived; you may find it illuminating to do an occasional scene from a secondary character’s point of view, so that the reader can see your hero as others do. Philip K. Dick did this brilliantly in The Man in the High Castle. One of the novel’s main characters, Ed McCarthy, is trying to interest a merchant, Robert Childan, in buying some jewelry he and his partner have designed. Ed seems clever and in control in the scenes leading up to the sales pitch to the merchant—but when it comes time for the actual pitch, Dick plants us firmly inside the merchant’s head, and we see Ed McCarthy in a new light:
[McCarthy] wore a slightly-less-than fashionable suit. His voice had a strangled quality. He’ll lay everything out, Childan knew. Watching me out of the corner of his eye every second. To see if I’m taking any interest. Any at all.
For each scene, choose your point-of-view character with care. Stick with that one person throughout the scene—and you’ll find that readers are sticking with your story all the way until the end.
Dialogue
Writing convincing dialogue is one of the hardest things for new writers to master. In fact, it’s so rarely done well in any form of fiction that when it is done right, people rally around it. The movie Pulp Fiction, Terry McMillan’s novel Waiting to Exhale, and the TV series My So-Called Life were all remarkable in large part because of how believably the characters spoke.
Here’s the kind of dialog you read in many beginners’ stories:
“What happened to you, Joe?”
“Well, Mike, I was walking down the street, and a man came up to me. I said to him, ‘What seems to be the difficulty?’ He replied, ‘You owe me a hundred dollars.’ But I said I didn’t. And then he hit me.”
Here’s how real people talk:
“Christ, man, what happened?”
“Well, umm, I was goin’ down the street, y’know, and this guy comes up to me, and I’m like, hey, man, what’s up? And he says to me, he says, ‘You owe me a hundred bucks,’ and I’m like no way, man. In your dreams. Then—pow! I’m on the sidewalk.”
See the differences? Most people’s real dialog tends to contain occasional profanity (“Christ”), to be very informal (“guy” instead of “man,” “bucks” instead of “dollars”), and to have lots of contractions and dropped letters (“goin’,” “y’know”). Note, too, that when relaying an event that happened in the past, most people recount it in the present tense (“he says to me,” rather than “he replied”).
Also note that in the first example, the speakers refer to each other by name. In reality, we almost never say the name of the person we’re talking to: you know who you’re addressing, and that person knows he or she is being addressed.
A few other features of real human speech demonstrated in the second example above: when relaying to a third party a conversation we had with somebody else, we usually only directly quote what the other person said; our own side of the conversation is typically relayed with considerable bravado, and the listener understands that what’s really being presented is what we wish we’d had the guts to say, not what we actually said. We also tend to act out events, rather than describe them (“Then—pow! I’m on the sidewalk”). Indeed, without the acting out, the words often don’t convey the intended meaning. The speaker was probably standing on the sidewalk throughout the altercation, of course; what he meant by “on the sidewalk” was that he was knocked down.
Now, which of the above examples is better? Well, the second is clearly more colorful, and more entertaining to read. But it’s also more work to read. A little verisimilitude goes a long way. Dropped final letters are rarely shown in fictional dialog (they’re usually only employed to indicate an uneducated speaker, although in reality almost everyone talks that way), and vagueness about verbs (“I’m like” instead of “I said”), verbalized pauses (“umm”), and content-less repetitions (the second part of “He says to me, he says”) are usually left out. In a short story, I might perhaps use dialog like the second example above; in a novel, where the reader has to sit through hundreds of pages, I might be inclined toward some sort of middle ground:
“Christ, man, what happened?”
“I was going down the street, and this guy comes up to me, and I’m like, hey, man, what’s up? And he says to me, ‘You owe me a hundred bucks,’ and I say ‘in your dreams.’ Then—pow!—he knocks me on my ass.”
Of course, not all your characters should talk the same way. I read one story recently in which there were dozens of lines of dialog like this:
“Interchangeable?” he said. “What do you mean the characters are interchangeable?”
We have the attribution tag between an initial word and a sentence that repeats that same word. This is clearly being used to denote confusion—and works fine once or twice, but grates if the same dialog device is employed more than that in a given story—especially by multiple speakers. Assign distinctive speaking patterns to single characters.
One trick is to come up with a word or two that one character—and only that character—will use a lot (in my The Terminal Experiment, the character Sarkar loves the word “crisp,” using it to mean anything from well-defined to delicate to appealing to complex); you might also come up with some words your character will never use (in Starplex, I have a character who hates acronyms, and therefore avoids referring to the ship’s computer as PHANTOM).
Profanity is also important. Terence M. Green’s rule: you can’t worry about what your mother will think of your fiction. But, again, not all characters swear the same way, and some may not swear at all (in The Terminal Experiment, I have a Muslim character who never swears, although the rest of his speech is quite colloquial).
It’s tricky handling characters who are not native English speakers. No matter what language they’re speaking, people tend also to be thinking in that language. It’s common to write a French character saying things like, “There are beaucoup reasons why someone might do that.” But at the time the person is speaking, his brain is thinking in English; it’s as unlikely for him to slip into French for a word as it is for a computer running a program in FORTRAN to suddenly swi
tch over to BASIC for a single instruction. Instead, if you want to remind the reader of the character’s native tongue, have the character occasionally mutter or think to himself or herself in that language.
The best way to learn how real people talk is to tape record some actual human conversation, and then transcribe it word for word (if you can’t find a group of people who will let you do this, then tape a talk show off TV, and transcribe that). You’ll be amazed: transcripts of human speech, devoid of body language and inflection, read mostly like gibberish.
To learn how to condense and clean up dialog, edit your transcript. For your first few attempts, try to edit by only removing words, not by changing any of them—you’ll quickly see that most real speech can be condensed by half without deleting any of the meaning.
Finally, test your fictional dialog by reading it out loud. If it doesn’t sound natural, it probably isn’t. Keep revising until it comes trippingly off your tongue (yes, that’s a cliche—but remember, although you want to avoid cliches in your narrative, people use them all the time in speech).
A couple of matters of form that seem to elude most beginners: when writing dialog for a single speaker that runs to multiple paragraphs, put an open-quotation mark at the beginning of each paragraph, but no close-quotation mark until the end of the final paragraph. And in North America, terminal punctuation (periods, exclamation marks, and question marks) go inside the final close-quotation mark: “This is punctuated correctly.”
Get your speech-attribution tags in as early as possible. There’s nothing more frustrating than not knowing whose dialog you’re reading. Slip the tag in after the first completed clause in the sentence: “You know,” said Juan, “when the sky is that shade of blue it reminds me of my childhood back in Mexico.” And when alternating lines of dialog, make sure you identify speakers at least every five or six exchanges; it’s very easy for the reader to get lost otherwise.