Ever since I read about a computer model developed by researchers at Stony Brook University to predict the success of novels based solely on writing style, I’ve been pondering why this study upsets me so. I have an analytical mind. I believe problems, situations, and tasks can all be broken down into component parts that make them easier to solve, understand, complete. But to assert that a piece of art can be so disassembled goes against everything I believe about art. Good art is more than the sum of the parts.
According to the study, “Success with Style: Using Writing Style to Predict the Success of Novels,” statistical analysis of writing style can distinguish successful literature from unsuccessful literature with 84% accuracy. Therefore, the researchers say, distinct linguistic patterns shared among successful literature, at least within the same genre, make it possible to predict which works will be successful. Wow. Wouldn’t that be an editor’s or writer’s dream? Don’t worry about whether a book grabs you in the gut or not. Just run it through a statistical model and you’ll know whether it’s any good.
Well, that’s not exactly what the study says. The researchers admit that they’re not trying to separate success based on quality from success based on popularity. To define success, they relied mainly on the download counts available at Project Gutenberg, which houses more than 40,000 books available for free download in electronic format and provides a catalog containing brief descriptions of these books. The researchers explain that determining whether the high download counts are due to one reason or another is not practically possible. For a small number of novels, they considered awards and Amazon’s sales records to define success.
But, whether the book is any good or not, they can predict whether it will sell or at least be added to readers’ libraries.
For comparison purposes, the researchers divided novels into genres. As examples of word choices they uncovered, they say less successful adventure novels use verbs that describe actions or emotions (“want,” “promise,” “cry,” “shout,” “jump,” “glare”). More successful adventure novels use thinking verbs, including “recognized” and “remembered.”
In the parts-of-speech category, the study shows that prepositions, nouns, pronouns, determiners, and adjectives are predictive of highly successful books, whereas less successful books have more verbs, adverbs, and foreign words. Also, more successful books tend to have more complex sentence structure, and less successful books rely more on simple sentence construction.
Some of these findings are not only counterintuitive, they run contrary to lessons taught in writing classes. But, if this is what the researchers found, this is what they found. I won’t argue with that.
As a reader, I think what disturbs me about this study is that I feel manipulated. Could I love a book written by a computer if it had the right word choices, the right parts of speech, the right sentence constructions? Could I love another person if the person had the right appearance, the right intelligence, the right talents? What is it about the book or the person that gives the entire being its spark?
The Stony Brook researchers say early in their report that the success of a book may depend upon many factors, including interestingness, novelty, style of writing, engaging storyline, social context, and luck. Their goal was to establish a purely quantitative method for measuring the connection between stylistic elements and literary success. And they’ve achieved that goal.
But I have to ask, isn’t style one of the elements that make an author’s writing unique? What if every author ran his or her novel through this program and made adjustments in the writing to meet the success criteria? Would all the books I read sound alike?
I think that to fully understand the significance of the study and to accept it, you have to emphasize its definition of a successful book. A successful book is not necessarily a great book or even a good book; it’s a book that people will add to their collection and maybe read. I think the study says more about our common baseline for digesting literature than it does about our ability to embrace an acquired taste for something different.
Vanilla is the most popular flavor of ice cream, so vanilla ice cream sells the most. If you choose to make vanilla ice cream, it will probably be successful. Thank goodness ice cream makers choose to offer more than vanilla.