The linguist’s Drosophila: Experiments in language change

  • Gareth Roberts
  • Linguistics Vanguard, August 2017, De Gruyter
  • DOI: 10.1515/lingvan-2016-0086

Why conducting experiments to investigate language change is not as hard as you might think

What is it about?

Language changes all the time. However, it changes relatively slowly, and over much longer time periods than a laboratory experiment typically lasts. It has therefore long been assumed in the fields of historical linguistics and sociolinguistics (which are devoted to investigating language change) that we can't really use laboratory experiments to investigate processes of language change (though we might be able to conduct experiments on related questions). Here I argue that this assumption is obvious, but ill-founded. In fact, in a parallel field of study devoted to the cultural evolution of language, whose adherents are often interested in kinds of language change that take a particularly long time, a number of experimental approaches have been developed. These could profitably be applied to the study of language change more generally. I lay out how these methods work, suggest how they could be applied to change in modern language, and argue why they should be more broadly applied, and why more traditionally trained sociolinguists and historical linguists should be involved in them.

Why is it important?

Those historical linguists who have doubted that experiments could be used to investigate processes of language change seem at the same time to wish that they could be so used. This paper should thus be cheering to them. And experimental methods are important – they are arguably the bedrock of science, where we can test our theories in a replicable way. Introducing them into a field where they have previously not been used therefore opens up extremely important new vistas for that field. At the same time, it is important to be aware that the researchers already conducting such experiments are often not trained as historical linguists or sociolinguists and are often not aware of a host of important things that those linguists know about. The primary goal of this paper is to encourage these two groups of researchers to interact much more closely. If this happens, the consequences for the scientific study of language are huge.


Gareth Roberts
University of Pennsylvania

In many ways my career so far has been about taking methods from language evolution and applying them to questions that sociolinguists and historical linguists care about. This has led to several enjoyable and fruitful interactions and collaborations. I am already not alone, but I very much hope that this paper will encourage more such interactions to happen.

Read Publication

The following have contributed to this page: Gareth Roberts