Last updated December 1, 2017 at 2:09 pm
Every time you speak, you are either inventing new words, in which case you’re likely to be my three-year-old daughter, or you’re more likely copying something you’ve heard someone else say.
As in any case where there is copying, there’s the chance for changes to occur over time. This change might be random, or there might be some sort of bias that leads speakers to favour one form of a word over another. But how do you tell the difference between random change or selection?
Researchers from the University of Pennsylvania drew their inspiration from another field of study which regularly deals with this problem, population genetics. Geneticists know that gene frequencies within a population change even without natural selection, a process called genetic drift. The researchers in this study used the tools of population genetics to test whether the change in frequency of word forms over time, for example the evolution that’s seen for the verb form waked to woke, was due to random drift or evolution by selection.
They started out by analysing more than 100,000 texts containing more than 400 million words spanning a timeline from 1810 to 2009, looking for changes to verbs from their regular form (for example lighted) to their irregular form (lit), or vice versa (wove -> weaved). The prevailing theory is that language evolution selects for regularization of verbs, because they’re more economical and take less brain power to work out- sticking an ‘ed’ on the end of a verb is easier than having to think about changing the whole internal structure of the word. This is seen in changes like the observed change of smelt -> smelled as the past tense of smell over the last 200 years.
The change to irregular form of the verb, like sneaked -> snuck, is more mysterious. The authors suggest that it could be because they have a rhyming pattern with other present/past verbs. For example, the change from dived to dove occurred around the same time as the invention of cars in the 20th century, and the rise of the use of drive/drove.
Of all the past-tense verbs analysed in this study, most were found to change purely due to random drift rather than selection, and rare words had more drift than common words. What is actually going on at the individual level, and the cognitive and social factors that lead to selection acting rather than chance is now an open area for future research. This approach will allow linguists to delve further into the forces that shape our language.