|
Abstract:
Cognates are words in different languages that are similar in form and meaning, which usually means that they have some common origin. An example of a cognate pair is English "father" and German "Vater".
In computational linguistics, cognates are used for bitext alignment and inducing translation lexicons. In historical linguistics, there are important for reconstructing proto-languages.
My research is focused on automatic identification and alignment of cognates from wordlists representing related languages. In order to achieve that goal, I have found myself increasingly compelled to use probabilistic and machine-learning methods.
This talk will not be about new machine-learning techniques, but rather about the application of the existing techniques to a problem in linguistics. The presentation should not be longer than 30 minutes, with the remaining 20 minutes devoted to discussion.
Links: