When a friend phones, you know her identity by the sound of her voice. This is the basis for the writer's-voice metaphor, but it’s only a metaphor. The features distinguishing spoken voices refer to the physical dimensions of the sound waves the vocal chords produce, and a pen’s scratch or a keyboard’s click are failed candidates for the voice that’s purported to infuse the scratcher or clicker’s document.
If the spoken voice is the usual metaphor, expression of the writer’s personality is the standard explanation, although it retreats to a murkier metaphor. A real explanation would link the specific characteristics of writing said to constitute voice to specific personality traits. Perhaps someone will someday link writers’ personality traits to expressive style, but before theorists can even speak of a linkage between personality and manner of written expression, they have to know the expressive traits voice comprises. When graphologists, for example, claim styles of handwriting are linked to the writers’ personalities, they have in mind connections like, “If the writer makes her dots above her letter ‘i’ like little circles, she will have histrionic tendencies"; or, “A rising baseline expresses an optimistic outlook.” Writing-voice exponents don’t specify any candidates for the expressive equivalents of circular dots or upward slope, never mind whether they correlate with personality.
Nobody knows how to talk about writers’ voice; yet, some writers manifest a distinct “voice.” Why should being specific about what they manifest be so difficult? My answer is that there’s an obvious solution, but it is, on second thought, obviously wrong—so obviously, that we don’t even consider it; but no other solutions are forthcoming. The obvious solution is that expert writers whose voices are said to differ write sentences distinctive in their length. The rebuttal is that, if voice is worth discussing—if writers can find their authentic voices—then voice can’t be a trait writers adopt as casually as making their sentences longer or shorter. Finally, the mistake the rebuttal commits is ignoring that an expert writer lacks the capacity to change his average sentence length without damaging his expressive capability: optimal average sentence length is immutable.
If you’re like me, when pressed for examples of distinctive voice you think of Hemingway and Faulkner, who are so unlike in the length of their sentences that it overshadows other differences. A second formal difference between them, preference for common versus esoteric words, accommodates different typical sentence lengths: to cohere, long sentences require abstraction. But problematically, average sentence length seems a matter of choice or preference, rather than an inherent personal quality. The idea that finding your voice means achieving stability at your optimal sentence length strikes, at first, as crudely reductionist. Writing teachers often advise students to shorten their sentences, and to the extent this advice helps, it would not seem tantamount to directing students to write in an inauthentic voice. Misleading in this scenario is that we’re talking about students who haven’t "found" their voices—and probably never will. Imagine telling Faulkner to shorten his sentences.
An element of commercial branding probably contributed to polarizing the Hemingway-Faulkner contrast, but I have an example of a professional writer being “told” to shorten his sentences. Science writer Steven Berlin Johnson—on whose casual research the present sentence-length theory of voice is based—found that Malcolm Gladwell’s average sentences were 6.5 words shorter than Johnson’s. His reaction is telling, Johnson declaring, “A 25% drop in sentence length has to alter the reading experience dramatically"; and he joked, “Clearly, the only things separating me from selling ten million copies of my books are those extra 6.5 words per sentence.” While this was overstatement—the writers’ topics no doubt affect their popularity—it probably isn’t entirely false, since a greater number of readers can understand short sentences than can understand long ones. This is why primary-school texts contain very short sentences! Yet, there’s no sign that Johnson—already an accessible writer—tried to make his writing still more accessible by using shorter sentences. Instead, Johnson’s posting focused on each writer's invariant sentence length—evidence that, for the expert writer, optimal sentence length is an immutable trait. For immature writers, the advice to shorten sentences nudges them toward their “authentic voice” or, at least, toward a degree of syntactic complexity they can manage, but it can be taken too far—and often is.
Expert-writers' consistency in their works' syntactic complexity is evidence that mature voice is optimal average sentence length; evidence against this hypothesis is that average sentence length has declined over the years, from 50 words in pre-Elizabethan times, to 29 in Victorian times, to 20 words per sentence, today. (William H. DuBay. (2006) Unlocking Language: The Classic Readability Studies.) If optimal average sentence length is voice, it shouldn't change over generations: if environments change it, then why not training regimens, so that Steven Berlin Johnson could train himself to write shorter sentences—to write more like Malcolm Gladwell?
The objection seems surmountable. In an era when a “good writer” was expected to average more than 30 words per sentence, one who could sustain only 20 would choose a different occupation; today, it can seem the reverse is true. With popular writing style ever increasingly that of marketers, it may seem that those whose genes cause them to write their best using complex syntax will be declared incompetent. But this is unlikely: the unity of language and thought suggests that well-managed syntactic complexity accompanies competent ideational complexity. Before mass advertising arose, the world might have found little use for the master of the simple sentence, but today's complex world still needs complex thinkers.