You, too, have an optimal sentence length

Plain-writing proponents advise you to check your documents’ average sentence lengths to guarantee against excess: Bryan Garner recommends an average of 20 words per sentence, and some plain writers recommend 15. Since every writer has an optimal average sentence length, the better advice is to use your own optimum rather than an arbitrary standard. I find, in fact, that when the average sentence length departs from my average, the document needs more work. Nobody has previously explained why writers consistently prefer a certain average sentence length, but inasmuch as “writer’s voice” is mostly sentence length, an explanation could help writers find their “true voice.”

I assume excellent writers prefer their strengths to their weaknesses, and I hypothesize that optimal sentence length is a trade-off between two abilities integral to writing: abstraction and sequencing. Typically, we construct sentences by abstraction and paragraphs by sequencing. Constructing a coherent sentence requires abstracting a suitably deep idea, but linking sentences to form cohesive paragraphs requires attending to their sequential relations. Long sentences capitalize on the writer’s ability to entertain a complex abstraction to be stated in words. Short sentences capitalize on the writer’s ability to link ideas in successive sentences. To make the most of their ability to entertain complex abstractions, writers strong on abstraction compared to sequencing will write long sentences, and to make the most of their ability to sequence thought, writers strong on sequencing compared to abstraction will write short sentences.

The distinction between abstraction and sequencing sounds somewhat like right and left hemisphere, but it isn’t. Here, we’re not talking about whether the internal processing is simultaneous or serial but whether the output is a unified abstraction or a sequence. The dimension of relative strength in abstraction compared to sequencing most resembles construal level: abstraction being far (resulting from abstract construal) and sequencing near (resulting from concrete construal). Personal consistencies in tendency to think far or near are shown, as in the finding that people who wake up late and prefer to work at night (“night owls”) tend to think far.

One educator’s questionnaire estimates your position on what amounts to far versus near thinking, conceived as Global-versus-Sequential learning style. (Hat tip: Words, Ideas, and Things.) I’d be interested in anyone’s results measuring their sentence lengths and testing their Global-Sequential position. My average sentence length is 25 and Global-Sequential learning-style score is 7 (moderately high Global). 

If you apply this test, bear in mind these caveats: 

1. Short documents will diverge from your average due to random statistical fluctuation.

2. Some documents should diverge from your optimum when the need to write in a particular voice outweighs achieving your highest literary quality.

3. Any document will not only be more interesting but also clearer if you vary the sentences' length.