Microsoft and a left hook to the language barrier

C-3PO was “fluent in over six million forms in communication”. Is Microsoft getting nearer?Image copyright Mark Menzies, reproduced under Creative Commons

Yesterday I read this article in the Telegraph that reported Microsoft’s development of translation software. If you watch the embedded video (also available here), you can see the new software in action. Last month Rick Rashid, Microsoft’s Chief Research Officer, spoke in English and the program computed his voice and presented it as English subtitles on the screen behind him. The subtitles weren’t perfect – some words are wrong and the grammar is shaky – but the meaning is by and large conveyed.

The video saves the best for last. Rashid says a sentence, then the software translates it into Chinese characters and into audible speech in a matter of seconds. I don’t speak Chinese, but judging by the audience’s reaction it was intelligible.

Before I started studying at City University, I was a translator for an online retail company. My main duties were to translate product listings and emails for the customer service team. I brought my dictionaries with me but relied heavily on online translating tools to help.

While I was there I thought that human translators have two main advantages over machines:

  • Humans understand context better than machines. TechCrunch reported in 2009 that the average American computed 100,000 words every day from a variety of different sources. Computers might have more processing power, but humans have years of experience that computers can’t match easily.
  • Humans can provide instant translation. In organisations like the EU, interpreters listen to MEPs and others and provide an oral copy of a speech in real time. Computers didn’t have the ability to handle the intonation, word order and speech patterns of different people quickly enough.

Until now, it seems.

Ordering Words

In German, for example, in many sentences the verb comes last:

Ich habe das Buch gelesen.

This means “I have read the book”. Translated word-by-word, it would read “I have the book read”. With more complicated German sentences, you can have 20 or 30 words before you get to the verb, what actually happened. It’s amazing that a computer can “reorder” a sentence so quickly where necessary, as Rashid explains in the video.

A future for human translators?

So does this mean human translation is on the way out? If Microsoft or its competitors could improve the program’s accuracy, then it’s possible. I doubt translators need to fret just yet though. Assuming Microsoft could perfect this software, it would be a long time before it was affordable on the market.

A better world?

Rashid said: “We hope in a few years, we’ll be able to break down the language barriers between people. Personally I believe this is going to lead to a better world.”

I think he’s right. It would make the world a more understandable place. But I hope, as a language graduate, this doesn’t mean people will stop learning languages.

I think there’s a case for a “universal second language” to be agreed upon and taught in all countries which don’t have it as a mother tongue. You would think English would be the front-runner for that language, although to be absolutely fair you could use a constructed language like Esperanto.

This has the advantage of keeping human conversation natural and fluid, without having to rely on machines. It also means we can preserve the cultural heritage wrapped up in all the world’s languages.


2 thoughts on “Microsoft and a left hook to the language barrier

  1. Very interesting (and well written) little post. As I’m constantly reminded by my Dad, in professional circles there is a big difference between translation and interpretation. I know professional translators already use software to look at the way they’ve translated things in the past/the equivalent of technical terms. At the moment I don’t think computers can replace the speed of human interpreters or the level of equivalence or idiom human translators achieve (but then I’m biased!), but technology keeps getting better!


  2. Thanks Ed, I think you’re right, if computers ever do match idiom and shades of meaning it will take a long time to come in. I didn’t really appreciate the difference between interpreting and translation until I started writing, and even then your comment made that clearer.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s