Saturday, July 21, 2007

Brad Pitt, Bill Murray and statistical machine translation

What do Brad Pitt, Bill Murray and statistical machine translation have in common, you may ask?

You are about to find out after this little detour. Do you read and write chinese? I don't. However I can pretend I do, and that's thanks to the recent english to chinese online translation service provided by Google. A relative from Taiwan asked us to explain the two following english sentences.

1:Grapes of Wrath by John Steinbeck is a classic American novel.
2:Greek island-hopping makes for a very varied vacation.

You can see the transation in chinese in my reply to that email. That's a straight copy and paste from Google Translate. So what is the big deal? Well, I was asked sometimes later by another relative of mine - and native speaker - if it was true that I had simply used a translation made by a machine. Why? Because the translation was surprisingly good.

Current statistical machine translation programs have made amazing progress. Google trains its machines with billions of documents from the United Nations that were translated in many languages by real human translators. Yes this takes our tiny Java program discussed in Activity 4 - Translator to a completely different level.

Can online translation accessed through an iPhone make travel in foreign countries easier? If you answer yes, you may want to watch again the movies Babel figuring Brad Pitt, and Lost in translation with Bill Murray. Cultural shock goes beyond translation issues... Brad Pitt and Bill Murray may still not fit in. They would be able to carry a conversation with a local for a medical emergency in Morroco and visit Tokyo via its many subway lines without actually understanding Kanji characters.

No comments: