Translation continues to tax the limits of technology. When Facebook announced back in October that it selected Microsoft Bing as its partner for real-time online translation, expectations were set high. Online social media platforms seem like perfect candidates for translation tools. Professional translators with masters degrees feared the takeover by online translation tools. The initial excitement was short-lived, however, as linguistic enthusiasts revealed that Bing isn’t so good at translating after all.

The community of Artificial Intelligence (AI) researchers and developers have been shooting for flawless machine translation (MT) since the 1950s. Even though MT has come a long way, it is far from reaching fluency. Part of the problem lies in the propensity of MT to provide literal translation instead of language interpretation. One cannot fault computers for this linguistic shortcoming; after all, linguists who have been consulted for the purpose of teaching computers about language haven’t been able to follow Noam Chomsky’s language acquisition theory.

Microsoft’s translation tools, like the disappointing Bing translator on Facebook, use a technology called “Collaborative Translations” centered on MT speed and content awareness. Microsoft admits that error-free, contextually accurate translation is difficult to achieve, so perhaps Bing translator has a lot to learn.

The broken translations that Bing provides on Facebook haven’t stopped it from being integrated in other major websites and services. Trip Advisor, eBay and even Merriam-Webster have chosen Bing as their online translation partner.

Professional translators are naturally bemused at the fact the Bing gets “Lost in Translation”, and it is clear that MT tools probably will not pass the Turing Test anytime soon, but the field of MT is vastly improving, particularly when English is the target language. Bing isn’t likely to give translators a run for their money, but there is one important aspect to consider about the collaborative translation feature and its placement on the world’s most popular online social networking platform. When translations are performed on Facebook, users have the opportunity to correct them and submit them for Bing to learn. These translations could eventually be “Liked” and reviewed by the community, thereby giving Bing even more opportunities to learn, though idioms and colloquialisms will very likely continue to slip through the cracks.

Bing’s currently deficient translation is being generally accepted by the Facebook community, something that isn’t entirely surprising. The human brain can understand even crude translations accurately when they are presented in a social context. Outside of the realm of online social networking, however, MT suffers from low confidence and poor expectations. Professional translators need not worry about Bing taking over their jobs for years to come.

_________________________________________________________________

Emily Matthews is currently applying to masters degree programs across the U.S., and loves to read about new research into health care, gender issues, and literature. She lives and writes in Seattle, Washington.