October 16, 2021

Scientists have created an interpreter that can decode extinct languages

Algorithms are more and more advanced today, and machines and programs created on the basis of them can really surprise people with their capabilities. A great example of this is the latest work of scientists at MIT, a program that can help historians, archaeologists and cultural scientists in their work. The algorithm that drives it can translate words and phrases from languages ​​whose meanings and pronunciation we do not know anymore.

The dying world

Today’s faster and faster changing world has its dark sides. The development of industry destroys nature, the pace of life adversely affects our psyche, and globalization leads to the disappearance of languages ​​with a small range and a small population speaking it. Moreover, throughout history, this process, though much slower, also took place, leaving us with documents that we cannot interpret. And yet in the sources that we are unable to read, there must be a lot of interesting information.

The power of an algorithm

The system was developed under the supervision of Professor Regina Barzilay, an outstanding specialist in algorithms, who recently received an award of 1 million dollars for the creation of tools of this type useful in the diagnosis and treatment of breast cancer. Her team’s algorithm uses key principles of historical linguistics that have been discovered in ‘human’ research into language development.

One of the examples of such rules is the word with the letter “b”, in which as the language develops or the borrowing occurs, it may change into a similar word in the pronunciation of “p”. At the same time, there will be no conversion to a phonetically completely different “k”. On this principle, from hundreds of similar dependencies, an extensive model was created, which as a result not only is able to decode words that no longer appear, but also can find relationships or their lack between the studied languages.

You are not family

As an example of such an action, the result of the comparison of the Iberian language with the Basque language was indicated, in which the system diagnosed that they are not related, which coincides with the recent “classical” studies in this field. Therefore, it will be possible to partially decrypt languages ​​of unknown origin, the algorithm is able to find appropriate dependencies on its own.

At present, the research focused on languages ​​belonging to the Gothic and Iberian groups, with Spanish, Arabic, Hungarian, Turkish, classical Latin and Basque as auxiliary languages. The system is constantly being developed, but it is already outperforming competing solutions.

The fish of Babel

There is no denying that in the field of translating languages, whether living or dead, algorithms are the future. Probably in a few years the vision of Douglas Adams will come true and an electronic version of the Babel fish with “The Hitchhiker’s Guide to the Galaxy” will appear . I am very curious how it will affect our civilization.

It can be assumed that language learning will become largely redundant, which, however, may translate into a smaller, let’s call it, cultural understanding. Learning a foreign language, whether we like it or not, forces us to acquire the ability to think, though rudimentary in terms of its categories, with the translation given by a machine, even the most perfect one, this sphere will cease to exist.

Polish language, difficult language

As it is easy to see, Polish, one of the most difficult languages ​​in the world, is missing among the languages ​​mentioned. However, I hope scientists will take up the gauntlet on this point, because someone finally has to be able to translate it into “digital” language. Who knows, maybe help will come from such an unexpected side as the study of extinct languages ​​and thanks to this we will see  Polish Siri , and the Google Assistant will stop hurting our language so much?

Source: [1] ,  [2]