• Australian Army soldier Akran Saleh (right) acts as an Arabic translator between Major Timothy Fricker (left) and an Iraqi Army officer at the Junior Leaders Course graduation ceremony at Taji Military Complex, Iraq. Credit: Defence
    Australian Army soldier Akran Saleh (right) acts as an Arabic translator between Major Timothy Fricker (left) and an Iraqi Army officer at the Junior Leaders Course graduation ceremony at Taji Military Complex, Iraq. Credit: Defence
Close×

Patrick Durrant | Sydney

Raytheon subsidiary BBN Technologies has been awarded US$4 million to provide automatic speech recognition, machine translation, text-to-speech, and optical character recognition software licenses for one year to the US Army's Machine Foreign Language Translation System (MFLTS) Program Office.

This is the first large scale fielding of the system, that enables personnel to converse with Iraqi Arabic and Pashto speakers and understand foreign language documents and digital media on three platforms:

  • an Android handheld,
  • a Windows laptop, and server-based systems,
  • a major intelligence system used at the battalion level.

 

The potential for faster interaction and avoiding misunderstandings should be a priority at a time when this is routinely flagged as conducive to future war

 


“Our military needs to converse with foreign language speakers and understand the situation around them," Raytheon BBN Technologies MFLTS program manager Martha Lillie said. "Giving them the tools to converse fluently, exchange information and understand printed material helps them accomplish their missions.”

The US Army began the MFLTS program in 2011 to provide language translation capabilities at the point of need across all Army echelons in all environments.

As Directorate of Future Land Warfare researcher Dr Allison Sonneveld wrote in a Land Power Forum piece entitled Finding a place for machine translation: Speech-to-speech technologies and the implications for the future land force, it’s worth considering the implications of developing a speech-to-speech machine translation (S2SMT) capability in the ADF.

“Since it takes time to learn a language, a portable S2SMT capability would save resources. At the Defence Force School of Languages (DFSL), long courses that teach basic to intermediate proficiencies on social, military and operational subjects range from 16 to 46 weeks.”

Dr Sonneveld points to the Defence White Paper having acknowledged the need to “expand cultural and language capabilities to increase … effectiveness in operating in the region and collaborating with international partners”.

“Given the long lead time that language training can entail, S2SMT is appealing for its ability to address a potential language capability gap,” she said.

Other factors that S2SMT would mitigate are the danger and exhaustion faced by the scarce number of locally engaged linguists. In Afghanistan, interpreters risk torture and death for working with coalition partners and the high volume of information requiring translation on operations is very demanding.

Of course there is a downside to S2SMT, according to Dr Sonneveld.

“S2SMT cannot fully replace the knowledge that locally engaged staff provide, and this can be central to Army’s situational understanding. It’s no replacement for interpreting, but the potential for faster interaction and avoiding misunderstandings should be a priority at a time when this is routinely flagged as conducive to future war.”

As noted by Dr Sonneveld, Army, in conjunction with the DST Group, had planned to conduct trials late last year with a commercial off-the-shelf device, the Voxtec SQ.410. ADM is seeking an update on the outcome of the field trials which we will publish in due course.

comments powered by Disqus