The English Language Literally Evolved at Random

When language evolves, most might assume it evolves for a reason. But new research upends that idea—a lot of the time, language evolves completely at random.

Linguists pooled their expertise with evolutionary biologists and tracked changes to individual words as if they were genes. Using databases of thousands of English-language texts that generations of their predecessors had spent decades parsing, annotating, and digitizing, the teams from the University of Pennsylvania pored over archives dating back to the 12th century. They focused on patterns like consistency among conjugated verbs—tracing the word "dove" as its meaning evolved to include the past-tense of "dive" and not just the bird.

By contextualizing and tracking each development ("dived" to "dove"; "quitted" to "quit") they were able to show for the first time that not all changes are due to selective pressures; often they just happen. Nature, in other words, is at play—not nurture.

Selective pressures can refer to either cognitive forces (some words, such as any that are commonly used or that rhyme, are just easier to remember than others) or social ones (we tend to use words that are reflective of our social class). The overwhelming majority of verbs that the team analyzed—especially rare verbs—showed zero trace of selective pressures. A study detailing the findings was published today in the journal Nature.

"So there's a range of forces. This study is showing that even just randomness can cause languages to change, sometimes a lot of change," said senior author and UPenn biology professor Joshua Plotkin. "Just talking about verbs—the rarer the verb, the stronger the amount of randomness, so that tells us there are rare words now where an alternative form is going to win out, and maybe we can make some predictions."

Celebrated American linguist Leonard Bloomfield once said that you can never see the changes in a language as they're happening. Plotkin's colleagues point to their collaborative method of analyzing the historical databases as the way we make that statement obsolete. But when we turn from history to the present day, it kind of already feels like language is evolving at accelerated enough speeds to be registered by the naked eye.

Consider the absurdly short half-life of slang on social media, or the speed with which definitions of words like "because" and "literally" were amended to reflect new usages. We really can see language evolving, in real time—or at least, we can hear it.

"This is based on written texts," Plotkin said. "But a lot of interesting change is based on just how we speak, right?"

Historical English-language texts, no matter how detailed, will still be written from a fairly homogenous perspective that doesn't account for things like cultural dialects or African American Vernacular English, meaning they lack the informality and diversity that make a language like English so rich in the first place. UPenn has databases containing decades of recorded speech from Philadelphia and other American cities. Plotkin and the other biologists and linguists plan to join forces again to apply the same methods they used on the text files to the audio ones.

"In Philadelphia, people drink woo-der," Plotkin said. "I'm from Boston, and we have wah-ta. But 40 years ago we had woo-der. That's exactly the ripe area of study. So that's the fun thing to do next, learning about phonetic evolution."