Novels with a certain structure are more likely to be classics

Why do some novels captivate generation after generation, while others hit bestseller lists upon publication, then fade away? A study suggests the answer may partly lie in the structure of their words and sentences.

Past research has explored the content that makes some books become bestsellers, with an affectionate male character often being a winning factor, but little is known about what contributes to a novel still being popular a century later.

To learn more, Leyao Wang at York University in Toronto, Canada, and her colleagues analysed 300 English-language novels published between 1909 and 1923. Half of these made the top 10 bestseller list of Publishers Weekly, a trade magazine that started in 1872, within a year of publication, such as The Fortunate Youth by William John Locke and Black Oxen by Gertrude Atherton.

But these novels don’t feature in the top 10 of users’ past, current or future read lists on the site Goodreads, which provides data on this according to a novel’s year of publication. The other half, such as The Mysterious Affair at Styles by Agatha Christie and Anne of Avonlea by Lucy Maud Montgomery, remain popular on Goodreads today, regardless of whether they were bestsellers in their day or not.

The team trained an AI model to differentiate between these two categories, based on 70 per cent of all the selected novels. When the model scanned their full texts, looking for any patterns in word usage, it found that those that were bestsellers initially tended to be longer overall, but also used more conversational words, such as “yeah”, “oh” and “OK”. These books were also heavier on punctuation.

These features may make the novels relatively easy to read quickly, boosting their immediate appeal, according to the researchers, even if their content isn’t strong enough to still excite readers today.

The books that are still popular with modern readers were shorter, but contained longer sentences and more complex words. The researchers wondered if this means they demand more of our attention, which may make them more memorable, perhaps prompting people to re-read or recommend them.

When the AI model was tested on the remaining 30 per cent of the books, it separated them into the two categories with around 70 per cent accuracy.

“Although many other factors also likely contribute to timelessness – such as themes, marketing choices, author reputation – we demonstrate that timelessness can be predicted without taking into account these more obvious non-textual influences,” according to the researchers. Understanding the different factors that give a novel lasting appeal “would be immensely powerful for publishers and authors alike”.

But Dorothy Hale at the University of California, Berkeley, says that the impact of a novel needs to be studied over a period longer than 100 years before it can really be judged as timeless, and that even the popularity of long-standing classics changes over time. “Many might consider Shakespeare the ultimate timeless English author, but the current trend in US colleges and universities is to eliminate the Shakespeare requirement from the English major,” she says.

Creating science fiction: an immersive writing course, England

Explore the world of science fiction and learn how to craft your own captivating sci-fi tales on this immersive weekend break.