Few novelists today would want a character to say, “It’s a good, much better rest than I’ve ever known.” This is not only because few modern figures contemplate death by guillotine, but also because writing styles have changed drastically since Charles Dickens wrote A tale of two cities in 1859. So, how does the literary style evolve? Surprisingly, the clues are found in seemingly insignificant words, such as “to” and “that”.
By analyzing how writers use such “contentless” words, mathematician Daniel Rockmore and his colleagues at Dartmouth College in Hanover, New Hampshire, were able to conduct the first large-scale “stylometric” analysis of literature.
Words without content are indicative of the writing style, Rockmore says. While two authors can use the same words to describe a similar event, they will use “syntax glue” without content to link their words in a different way.
Using Project Gutenberg’s digital library, Rockmore’s team analyzed 7,733 English-language books written since 1550, tracking how often and in what context words without content appeared. As you might expect, they discovered that writers were heavily influenced by their predecessors.
They also found that as the canon of literature grew, the scope of older works diminished. The authors of the early periods wrote very similarly to each other, the researchers found, likely because they were all reading the same small body of literature. But as the modern era approached, when more people wrote and more works were available from many eras and styles, the styles of authors were still very similar to those of their immediate contemporaries. “It’s as if they find dialects in time,” says Alex Bentley of the University of Bristol, UK, who was not involved in the study. “Content is what sets us apart, but words without content put us in different groups. “
That writers are more influenced by their contemporaries than by the great works of the past is interesting, says Rockmore, because it calls into question the significance of “classical” literature. Style-wise at least, we might not be so heavily influenced by the classics after all.
Journal reference: Proceedings of the National Academy of Sciences, DOI: 10.1073 / pnas.1115407109