As readers of this blog will know from prior posts (e.g. here and here), I've been very interested for a long time in the powerful role of language and culture in shaping media behavior in Korea. Preference for the Korean language is one of the big reasons that Naver continues to control such a large share of search, despite its limited scope when compared with Google. Today's alerts brought links to two very interesting sources.
The first is an article in The Independent by David Bellos entitled "How Google Translate Works." Rather than trying to develop an algorithm to discover the meaning (syntax and vocabulary) of a particular passage, Google uses a statistical approach that leverages its vast collection of written language and translations of the same writings into many languages. As Bellos notes,"It uses vast computing power to scour the internet in the blink of an eye, looking for the expression in some text that exists alongside its paired translation." The entire article is well worth reading.
Another source is the Localization Industry Daily, published by Cloudwords. That publication provides a number of interesting avenues through which to explore the burgeoning localization industry. More on this topic in future posts.
It seems obvious that Google is leading the effort to develop machine translation, with some gratifying results. However, it is equally apparent that the industry has a long way to go and that its most difficult challenges lie here in Asia, given the difficulty of translating Asian languages to and from English and other Latin-based languages.