I came across this interesting article in the Washington Post describing how Google Translate and Yahoo! BabelFish have improved their machine translations by using statistical instead of linguistic models. That is, with the massive abilities of computers today, better translations can be obtained by having them statistically analyze passages in two languages and generate translations on that basis rather than trying to linguistically analyze words and phrases. It does make sense given the idiomatic nature of so much of speech, but they are wise to go back now and try to provide some linguistic enhancement to their translations.
Give it a try using Hebrew or Greek (but remember that it is modern and not ancient).
Google Translate
Give it a try using Hebrew or Greek (but remember that it is modern and not ancient).
Google Translate
Actually, there are more recent linguistic models (e.g. Construction Grammar) which have basically rejected rule based linguistic frameworks as accurate models for how language. Construction Grammar takes what you term "the idiomatic nature of so much speech" as a basic starting point for explaining how language works on a larger scale. And they've done so quite successfully. Much of what's been done in a frameworks like Construction Grammar has far more in common with the statistical/usage-based approaches used by machine translation.
ReplyDeleteFundamentally, the problem in early machine translation research was the inadequacy of the existing linguistic models.
Deus Ex Machina: Texts translated by machines should be read only by machines (in the bathroom on the pot).
ReplyDeleteYou said right about the using statistical instead of linguistic models,and both of these sites are doing a great sort of job and they are facilitating the people a lot,and also the thing is that the translation technology for Babel Fish is provided by SYSTRAN, whose technology also powers a number of other sites and portals,great event this is.
ReplyDeleteNice! Interesting detail that I have been looking for on the Internet and that I found in your post.
ReplyDeleteEnglish to Greek translation