Machine translation for the military

The World

by Alex Gallafent

The US military has been producing specialist interpreters for decades – people who spend years learning Arabic, Chinese and other challenging languages.

They’re the kind of people Rye Barcott depended on. He’s a former Marine captain who served in Iraq. For Barcott, there’s no substitute for having a flesh-and-blood interpreter by your side when you’re gathering intelligence, specifically, a trusted linguist who can translate all aspects of a conversation – the verbal and the non-verbal.

But as Barcott points out, few units, especially in the infantry, have access to human interpreters. Machine translation could help, with basic information at least.

“When a Marine or Army unit’s out patrolling and they need to ask a person for permission to enter their property for example, it’s far better to have a machine than to not have anything,” Barcott said.

In recent years, the US military has adopted a policy of trying to give everyone who deploys some knowledge, something to hang their hat on.
Apps

“Hello. And thank you for your interest in learning more about V Communicator Mobile – the first deployed and fielded mobile military application.”

This is one of a growing range of products out there designed to bridge the military language gap. Ernie Bright is the product manager for V Communicator Mobile, which runs on a modified iPod Touch. Bright said the focus was on assisting soldiers, not distracting them.

“We knew that they were needing a more mobile easy-to-use product that would allow them to learn key tactical phrases, and also be able to just pick it up and use it,” Bright said.

It’s impressive, quickly translating a phrase such as ‘do you have any weapons?’ into the equivalent language in, say, Pashto.

But it’s not the only translation device out there vying for a potentially lucrative military contract. Another mobile app is called SpeechTrans.

According to the company’s co-founder, Yan Auerbach, “It allows you to speak into the device in your native language; it will translate and speak aloud the translation instantly.”
Dialects

It’s equally impressive. SpeechTrans draws on an archive of voice samples, but it doesn’t yet cover every language or – crucially with languages such as Arabic – every dialect.

Yan Auerbach tells me his company is now in the running for a Pentagon project called BOLT (Boundless Operational Language Translation). The goal for BOLT is lofty: technology that would recognize not just dialects but localized slang too.

That still leaves a lot of missing information: gestures and body language. It’s something Ernie Bright and the V Communicator Mobile team have been thinking about.

On their device, attention is paid to non-verbal communication too. For every piece of language you get an animated video, demonstrating the appropriate gesture. It’s meant to help soldiers learn the language too.

“They can remember that this phrase coincides with this gesture or this body language,” said Bright.
Interpreting gestures

That’s fine for stock phrases at checkpoints and the like. But soldiers have to interpret gestures as much as they have to reproduce them. And they have to do it in the blink of an eye.

Could a machine ever do that work too?

When it comes to translation technology, no-one yet knows where the lines will be drawn. All languages – all communications – are built on ever-shifting sands of meaning. That meaning can’t ever be fully, totally captured by technology.

But as translation tech accelerates it might fool us into believing otherwise. And so for the military the dream of a universal translating machine will be ever-more tantalizing.

The trick will be knowing when to use the gadgets. And when to put them aside.

Less than .05% of listeners will donate. Can we count on you?

Our coverage reaches millions each week, but only a small fraction of listeners contribute to sustain our program. We still need 224 more people to donate $100 or $10/monthly to unlock our $67,000 match. Will you help us get there today?