Google Planning Statistical Machine Translation for Additional Languages?
Google is about to launch statistical machine translation for languages other than Arabic, according to Garrett Rogers at ZDNet. Would you want to stake your reputation on translations that are often more confusing than edifying?
There's a school of thought that says any translation is a start, but statisitcal machine translation still can't get cultural references without human intervention, and leads to strange and overly dramatic translations. They build up their own dictionaries of words and phrases by comparing many pairs of documents that translate each other, known as "parallel texts."
Even the wrong translation of a positive post could have unforeseen consequences on your firm's reputation, says Sloan Friedman, president of SRF Global Translation, which specializes in mindful, nuanced, professional multilanguage translations. Statistical machine translation is a good tool for professional translators, who can add nuance and localize the language, but it's got a long way to go until it's adequate for business.
Email this to a friend