We think of archeology as dusty, khaki-on-khaki work. If we want to learn something about our ancestors, we need to dig up their cultural artifacts and piece together, not just some broken piece of pottery, but how that piece of pottery mattered to a particular group of people in its historical context.
But could big data simplify some of this manual labor? Apparently so. Because UBC researcher Alexandre Bouchard-Côté has developed an algorithm that’s capable of sifting through several regional sister languages and deducing what their mother tongue used to sound like.
The study analyzed 637 Austronesian languages (a family of languages that includes Hawaiian and Fijan) and was then able to suggest both how earlier protolanguages might have sounded, and which phonemes were most likely to change. When compared to the humans who reasoned the same information manually? The algorithm got things right about 85% of the time–or, at least what we can assume is right.
Interestingly enough, this sort of big data analysis allows us to look further into our past. Analyzing the divergence of the word “wheel” across European regions can actually track settlement rates. I can’t help but wonder if archeology is going to get a whole lot more interesting.
[Hat tip: The Dish]