Archaeologists have translated almost every language humans have ever used. Yet eight or so elusive languages remain a mystery. Until now, researchers have not successfully used a computer to assist with deciphering languages. It has been left, largely, to scholarship and often takes years of human logic and intuition to reveal the mysteries of ancient scripts.
But now researchers from MIT and USC have designed a computer system that does successfully model the logic and intuition of a human to decipher a language.
The method starts by making some assumptions about the language and the word. In this article by Larry Hardesty of MIT, he lays out what those assumptions are:
- the language being deciphered is related to a known language
- that there is a systematic way to map the alphabet of one language onto the other
- the mystery language has root words that are similar to a known language
- the mystery language may have similar word parts, such as prefixies and suffixes
Languages share common ancestors, so even extinct languages have similarities to ones we still use today. If you speak more than one Romance language, you know what I’m talking about. Romance languages like French and Spanish both came from Latin, so if you speak one Romance language, chances are you'll have an easier time learning a second one.
Computers don’t normally make this type of assumption, because they are designed to deal strictly with facts. Humans, however, are excellent interpreters and can quickly put language in context.
But this new language deciphering method (Download PDF) translated the ancient Semitic language Ugaritic to Hebrew in just a matter of hours. Ugaritic is extinct, but has already been translated into modern Hebrew, so researchers were able to gauge how accurate their computer system was.
For starters, the Ugaritic alphabet has 30 letters, and the system correctly connected 29 of them to their Hebrew counterparts. And about one-third of Ugaraitic words have Hebrew root words, and the system correctly identified 60 percent of them.
Such a computer program could help archeologists finally read the text of ancient languages that to this day remain a mystery, and it could work for automated translations systems, such as Google Translate, which so far only works for 57 languages. It's only a matter of time until every language can be translated. Even Klingon.
A replica of the Rosetta Stone. Photo by Juan Naharro Gimenez/Getty Images
Tags: Archaeology, Computer Software, Computers, History
comments ( )