Dave Coplin, chief envisioning officer at Microsoft UK (what, were you really expecting an article about Steve Ballmer?), recently wrote an Financial Times article questioning the value of foreign language education (registration required). His article also provides a strong signal about what his company intends to do with its Skype acquisition.
"Should we bother to teach children a foreign language, when big data-powered technology could soon do [real-time speech translation] better and much more efficiently?," Coplin asks.
By that logic, I suppose that we can end that time waster called physical education because of Segways. Take the kids out of karate class because of bombs. Give up dentistry because of blenders and straws. Skip history, government, philosophy, and every other conceivable pursuit of the liberal arts because of Wikipedia.
Looking beyond the Microsoft executive's (mildly successful) attempt to garner some attention with a bold attack on the all-powerful foreign language education lobby -- or what's left of it after the sequester -- there are some interesting technology implications behind his incredible claim. We should soon expect a mashup between Skype and Douglas Adams' Babel fish that may forever transform customer service and enterprise workforce interactions.
The confidence that Coplin shows in the technology gives me the impression that speech-to-speech translation already works wonders in certain scenarios. We can imagine the most obvious customers: the governments and spy agencies quietly training their attention on phone calls placed by persons of interest. After all, a court-approved wiretap is much more effective when you don't have to wait for a translator. The commercial and civilian sectors would be next in line. Some situations we might anticipate (or which may already exist in embryonic form):
- You speak on the phone to a customer service representative (CSR), who gets to see a transcript of what you're saying as you're saying it. The CSR doesn't miss anything and asks for confirmations when necessary. The company keeps a text-searchable record of the call.
- You speak on the phone to a CSR for whom English is a second language. The real-time transcript guides the CSR to the appropriate range of canned answers. The company lowers its costs by having access to a larger labor pool.
- You speak on the phone to a computer, with any nonstandard queries handled behind the scenes through Amazon Mechanical Turk or an equivalent tool, reducing the hands-on labor component to a minimum.
- You speak through Skype to a friend, colleague, or business partner, using real-time subtitles or dubbing to overcome language barriers.
- You can attend a conference presented in a second language with more assurance and comfort.
- You can speak to someone working in retail customer service anywhere in the world.
- Throughout your travels, whatever you hear gets automatically translated into English and whispered into your Bluetooth-enabled earphone.
There are limits. I don't expect machine translation to be particularly helpful at making you seem witty and charming when you're having drinks with the team during your next overseas posting. (See: CIO as Chief Immigration Officer.) Nor would I expect it to help you ace that big job interview. The accuracy rate is a major constraint on how the technology will be used. Even 99 percent isn't good enough if there are high stakes associated with what you're hearing.
Furthermore, machine translation skips the nonverbal aspects of communication. If I'm in Japan, how will translation software infer what level of politeness I should use in any given situation? If I'm in Italy, will the software tell me what to do with my hands? What about the Gallic shrug?
Finally, as a dues-paying member of the Modern Language Association, I must register my strong disapproval of Coplin's vision of the end of language education. An illustrative quote from a 2007 MLA report describes a spectrum of views of language:
At one end, language is considered to be principally instrumental, a skill to use for communicating thought and information. At the opposite end, language is understood as an essential element of a human being's thought processes, perceptions, and self-expressions; and as such it is considered to be at the core of translingual and transcultural competence.
Guess where I land on this spectrum.
What's your take on the future of language-translation technology in the enterprise? Also, do you think Microsoft is meshuga for opposing foreign language education? Espero leer sus pensamientos en los comentarios.