Friday, 27 February 2009

Translate between 41 languages with Google Translate

Google Translate recently added Turkish, Thai, Hungarian, Estonian, Albanian, Maltese, and Galician to the mix. The rollout of these seven additional languages marks a new milestone: automatic translations between 41 languages (1,640 language pairs!). This means we can now translate between languages read by 98% of Internet users.

In just a few years, the machine translation group within Google Research has taken its initial research system from two languages to 41 languages and is now handling millions of translation requests a day. For several languages, Google Translate is the first freely available machine translation system for these languages. Of course, there's always room for improvement, and we're working hard to improve translation quality. Our statistical models are built from vast quantities of monolingual and translated texts using automated machine learning techniques.

It's exciting and satisfying to work on a product that can help people access content they may otherwise be unable to understand. We've heard stories of people using Google Translate to help them do business internationally, and we've seen many websites (e.g., New York's Metro Transit Authority) and blogs add the Google Translate My Page Gadget to their pages to make their content more accessible to people from all over the world.

Whenever I personally travel, I do lots of research on the web to figure out what to see and do, and where to stay and eat. With Translate, I'm able to use the cross-language search feature to find and access the latest info (e.g., restaurant recommendations, most recent trains/bus schedules, special events, etc.), which is often only available in the local language.

More importantly, Translate provides people who may not otherwise have a lot of web content available in their own language with access to the wealth of content on the truly worldwide web.