In search of lost languages
Is computer technology a new Rosetta Stone? The work of researchers at the Massachusetts Institute of Technology (MIT) seems to be proving that it is.
Translation techniques are closely related to evolution in linguistics and a deeper understanding of the languages currently used in the world. It is impossible to continue moving forward in this field without knowing more about the so-called “dead” languages that came before our current verbal and written communication systems.
It was with this in mind that, in 2021, Regina Barzilay and Jiaming Luo, researchers at the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), embarked on a translation software project to decipher ancient languages that today’s linguists have so far been unable to figure out. The team hopes to create a tool that can decipher these lost languages using only the few available segments of text and some contextual and historical clues.
The Phonetic Prior system makes it possible to unravel these language structures on the basis of scientific rules of linguistics that detect patterns in their evolution over the centuries. For instance, the symbols used in related languages tend to have a similar distribution and regularity; related words tend to have a similar character order; and so on.
This knowledge, using machine learning analysis, comparison and data processing, makes it possible to decipher a previously unknown language.
One of the researchers’ test cases was Linear B a form of writing used by the Minoan civilisation on the Greek island of Crete between the 14th and 12th centuries BC. A few centuries later, Ancient Greek became the most widely spoken and written language in the region.
Using these contextual clues and available data on the language of the ancient Greeks, the programme compared characters and possible phonemes with the later language and was able to decipher much of the meaning of the text by correlating linguistic patterns.
At the moment, this is the limitation of this technology, as it is necessary to have some information about the language that came after the one being analysed. Be that as it may, this is an important development in the field of linguistics that could lead to the emergence of new tools for use in professional translation.
This research could lead to further advances in machine translation technologies used daily by the general public (with search engine tools) and translators working with CAT (computer assisted translation) apps.