Very interesting. A computer can no doubt look for patterns more quickly than we, and perhaps more thoroughly. Maybe more importantly, it's not limited to the languages it's expert in, as humans are.
Also of interest they mention Basque. It's already declared as a language isolate- not related to *any* other living language. Perhaps the AI will be able to find something it's connected to.
I was reading yesterday about an interesting hypothesis that "Iberia" (as in the penninsula) and Caucasian Iberia (around where Georgia and Armenia are today) have some ancient relationship. It is postulated that the Basque people are from the Caucasus, but I don't know how much evidence there is to back that.
Caucasian languages have been proposed as relatives of Basque, but most linguists find the theory unconvincing. More expansively, it is seen a member of the Dene Caucasian Family, which include a number of language families throughout Asia and into the New World. https://en.wikipedia.org/wiki/Den%C3%A9%E2%80%93Caucasian_languages
It will be interesting what AI comes up with, because people have been trying to find similarities to Basque "by hand" for a long time without convincing success.
Maybe, but maybe not. If the assumption holds that there are only a certain number of sound shifts that are probable, and we know that Linear B is Greek, then Linear A should be a language that sounds something like Greek letters spoken in a different order. (E.g., P/B shifts might happen, but probably not P/K shifts.) Then you can come up with a set of probabilistic models for what Linear A would have sounded like.
Then you can try to see if there are any known languages that sound like that. That might at least give you a linguistic family, and maybe from there you can start looking for roots of words that have set meanings across the family. (E.g., 'right' in English has a root that is very similar to concepts of right, justice, law, and order in all the Germanic languages.)
You might at least be able to begin to make sense of the broad content of Linear A writings.
5 comments:
Very interesting. A computer can no doubt look for patterns more quickly than we, and perhaps more thoroughly. Maybe more importantly, it's not limited to the languages it's expert in, as humans are.
Also of interest they mention Basque. It's already declared as a language isolate- not related to *any* other living language. Perhaps the AI will be able to find something it's connected to.
I was reading yesterday about an interesting hypothesis that "Iberia" (as in the penninsula) and Caucasian Iberia (around where Georgia and Armenia are today) have some ancient relationship. It is postulated that the Basque people are from the Caucasus, but I don't know how much evidence there is to back that.
As an aside to that, I came across a thread on twitter of various Georgian and Caucasian folk dances that was really charming.
Caucasian languages have been proposed as relatives of Basque, but most linguists find the theory unconvincing. More expansively, it is seen a member of the Dene Caucasian Family, which include a number of language families throughout Asia and into the New World. https://en.wikipedia.org/wiki/Den%C3%A9%E2%80%93Caucasian_languages
It will be interesting what AI comes up with, because people have been trying to find similarities to Basque "by hand" for a long time without convincing success.
And we've been trying to crack Linear A for a long time, too.
I don't know that the computer can work on Linear A. To compare, it needs to know what is being said to compare the structures, phrasing and so forth.
Linear A may forever be a mystery.
Maybe, but maybe not. If the assumption holds that there are only a certain number of sound shifts that are probable, and we know that Linear B is Greek, then Linear A should be a language that sounds something like Greek letters spoken in a different order. (E.g., P/B shifts might happen, but probably not P/K shifts.) Then you can come up with a set of probabilistic models for what Linear A would have sounded like.
Then you can try to see if there are any known languages that sound like that. That might at least give you a linguistic family, and maybe from there you can start looking for roots of words that have set meanings across the family. (E.g., 'right' in English has a root that is very similar to concepts of right, justice, law, and order in all the Germanic languages.)
You might at least be able to begin to make sense of the broad content of Linear A writings.
Post a Comment