Reading extinct languages through cryptographic analysis & Google

Ideas, thoughts, and resources from the permanently curious.

Living

World

Politics

Reading extinct languages through cryptographic analysis & Google (6/1/05)

Recently Google unveiled it's most recent machine translation efforts. Their feeding Rosetta stone-like texts into a computer and having the computer learn to read and translate on its own, rather than trying to teach it. This approach is apparently a closer approximation of how we learn languages and so the results are already far beyond previous methods.

Very cool, no doubt. Using this method Google should be able to put together a nice list of most common rules and ways of doing stuff that the majority of languages use. Having done that, Google could then apply those rules to a language for which no Rosetta stone exists; one we don't understand. The implications of the computer figuring out just a little bit of that unknown language would be huge, a fact demonstrated by the explosion of our knowledge of Ancient Egypt shortly after their language was deciphered.

Wikipedia has a good list of undeciphered languages, should anyone at Google be reading this. In particular, figuring out Old European Script would seem to be a boon to anyone trying to figure out the Proto-Indo-Europeans.