New cunning linguist computer has got ancient tongues licked

Boffins build system to reconstruct man's protolanguages

Boffins have put together a new computer system that attempts to translate protolanguages, the ancient "parent" tongues from which modern languages evolved.

The sophisticated Rosetta Stone-like system can quickly reconstruct the languages of yore from today's vocabularies with 85 per cent accuracy, we're told. The system's designers reckon it can outpace human linguists who painstakingly reconstruct protolanguages from the words we all know and use today.

With the exception of Latin - the parent of the Romance language family* - and a few others, written records of protolanguages tend to be rather rare, forcing experts to analyse modern speech to derive the parent languages. Specifically, linguists group together words with common meanings and study changes in pronunciation, among other techniques.

“We’re hopeful our tool will revolutionise historical linguistics much the same way that statistical analysis and computer power revolutionised the study of evolutionary biology,” said Alexandre Bouchard-Côté, statistics professor at the University of British Columbia and lead author of the study.

“And while our system won’t replace the nuanced work of skilled linguists, it could prove valuable by enabling them to increase the number of modern languages they use as the basis for their reconstructions.”

The new system, designed with help from colleagues at Berkeley, analyses sound-changes at basic phonetic unit level so it can operate at a much greater scale than previous computer tools.

The researchers reconstructed a set of protolanguages from a database of more than 142,000 word forms from 637 Austronesian languages for a study that will be published next week in the Proceedings of the National Academy of Sciences. ®

* The group includes Spanish, Italian, French, Romanian, Neapolitan, Ladino and many more.

Sponsored: 5 critical considerations for enterprise cloud backup