Sentences have been getting shorter for two hundred years, and dialogue is why
Somewhere between Chaucer and Chaucer's modern reader, the English sentence lost most of its length, and one plausible reason for that has less to do with attention spans than with how much of a novel is spoken out loud by its characters.
Start with the single clearest case: Agatha Christie. Writing in 2018 on the blog Language Log, the linguist Mark Liberman measured the proportion of Christie's prose that falls between quotation marks, and tracked it across her career. In 1920, about 49.2 percent of her text was dialogue. By 1972, that figure had climbed to 79.4 percent. For comparison, Virginia Woolf's To the Lighthouse sits at just 3.3 percent dialogue, while a G.K. Chesterton text in the same rough comparison runs to 60.1 percent. Liberman is careful to call this a rough measure on a small sample, not a controlled study, and the numbers should be read that way: suggestive, not definitive.
A separate, larger study lines up with the suggestion anyway. .txtLab at McGill, in a 2018 study applying the methods of the Stanford Literary Lab to roughly 5,000 English-language novels, found that readability, measured by the Flesch score, has risen steadily across two centuries of English fiction. That rise correlates with the growth of dialogue at 86 percent. Books have gotten easier to read, in other words, largely because more of what's on the page is people talking, and talk reads faster than narration.
Sentence length itself tells a similar story from another angle. A widely circulated 2023 compilation by the writer Arjun Panickssery lines up some vivid examples: Chaucer and Spenser averaging sentences of roughly 49 to 50 words, against J.K. Rowling and D.H. Lawrence down at 12 to 14. Even Warren Buffett's own shareholder letters shortened, from 17.4 words per sentence in 1974 to 13.4 by 2013. These are striking numbers, and worth a caveat: they're drawn from a handful of hand-picked authors, not a controlled sample, so the exact slope of the decline shouldn't be taken as precise.
For that, there's Rudnicka, whose peer-reviewed work (first published in 2018, with an arXiv version posted in 2025) measured sentence length across English text more broadly and confirmed the same long-run decline, while also finding real differences by genre. The direction Panickssery's colorful examples point to holds up under more careful counting. The exact numbers, understandably, don't match exactly, because a curated set of famous names and a full statistical corpus are never going to agree down to the decimal.
Put the pieces together and a reasonably coherent picture emerges, even if no single study proves the whole chain: as dialogue's share of the novel grew, average sentence length fell, and readability rose, three measurements that all move in the same direction across the same two centuries, from three independent research traditions that weren't necessarily talking to each other.
Rudnicka's finding about genre differences is worth returning to here, because it complicates the tidy version of this story. If sentence length falls by different amounts, and at different speeds, depending on whether a text is a novel, a newspaper, or a personal letter, then "English prose got shorter" isn't really one trend. It's several related but distinct trends, moving at different rates, that happen to point the same general direction when averaged together. Whether dialogue caused the shortening, or both are effects of some other shift in taste that also happened to make characters talk more, none of these sources claims to settle. What they agree on, from a nineteenth-century blockbuster mystery writer's fifty-year career to a twentieth-century investor's annual letters, is the direction of travel. Sentences got shorter. People, on the page at least, started doing more of the talking.