The BBC reported that the first soldier out of the helicopter full of commandos who attacked and then killed Osama Bin Laden in a pre-dawn raid spoke fluent Pashto. Compare this to the recent job posting by a US-based development contractor searching for a person to serve as strategic communications and outreach advisor in Herat, Afghanistan, whose job is to oversee all capacity-building activities related to improving the skills, competencies and abilities of local officials to deliver effective citizen-oriented communication and outreach strategies. The language requirement is English only.
But how can you develop successful strategies to communicate with a local, uneducated populace if you do not speak their language and know nothing about their culture? You rely on an interpreter, who is probably someone the company recruited from who knows where.
The successful kill operation involved at least one fluent Pashto speaker, while a development company wants to rely on a “communicator” with no local language skills. The whole issue of how the United States handles languages in a war zone dominated by insurgents who are indistinguishable from the local populace and speak in local dialects with bad syntax, poor diction, slang and accents is fascinating to me if for no other reason than that the most money being spent anywhere on earth on languages is in the Middle East by the US military and contractors. According to Wired, the Pentagon spends US$1 billion a year for translation and interpretation services, and much of this language work barely touches what we are doing in the commercial sector.
We are fighting a war of words the United States is having a hard time winning in spite of its technological superiority. I got the distinct impression growing up in a multilingual immigrant household that both the Pentagon, which finances language work, and some of the high-tech contractors carrying out the work referred to languages other than English as an irritant to be dealt with rather than an expression of a people’s culture. No one would, of course, admit to that.
Whatever the case, the military seems to be throwing money at something that may seem more straightforward to the average language-savvy person. “Psst, Military: There’s Already a Universal Translator in the App Store” proclaimed an April 19, 2011, headline in Wired magazine. Wired asks, “On one side of the scale: an app developer that wants to translate Pashto and Dari on your iPhone. On the other: around $50 million in federal money, this year alone, for research into a Cadillac version of the same translation tools. Which sounds like the better deal?”
The Universal Translator SpeechTrans uses Nuance, the world’s largest speech recognition technology. Nuance acquired part of IBM’s speech recognition technology back in 2009, and its current automatic speech recognition algorithms and technology are self-learning and constantly improving. Record a spoken phrase, choose the translation, and SpeechTrans will play it back, displaying both the source and target texts on the screen.
Nuance’s Dragon Medical is already used by the military in the US Army Medical Department, though as an advanced dictation method rather than a translation or interpretation device (Figure 1). As a result of Dragon Medical’s ability to enable direct dictation into electronic health records, the use of voice-recognition software with the military e-health record system is freeing doctors from hours of typing patient notes each week, thus eliminating the need for traditional medical transcription, which saves operating costs in the long run.
However, given the size of the Pentagon budget, there is money to reinvent the wheel. The Pentagon loves to flirt with technology, something I readily saw when I wrote part of my master’s thesis while sitting at a desk given to me by the US army’s civic action command.
What the military has wanted all along is a silver bullet —that magic piece of technology allowing instantaneous translation of any language into any language without any glitches. They want a Star Wars protocol droid — a C-3PO fluent in over six million forms of communication.
This is not new, for this has been going on since 1954 when IBM crunched a bit of Russian text into its English equivalent. A Georgetown University professor who worked on the project predicted the computerized translation of entire books in “five, perhaps three years hence.” So, here we are 57 years later, and no one has mastered literary machine translation (MT), let alone instantaneous speech-to-speech translation complete with all its slang, broken sentences and spoken poetry.
No one has mastered that yet, but this has not stopped the Pentagon’s far-out-there research branch DARPA to start BOLT (Boundless Operation Language Translation), which is slated to be sophisticated enough that it can understand foreign slang. By the way, for those of you considering sending in a proposal, the first step to getting money from the Pentagon is to have a workable, alert-sounding acronym; of course, you need to be registered and have a retired military guy as your representative as well.
The research and development company Raytheon BBN Technologies has formed a team to meet DARPA’s GALE (Global Autonomous Language Exploitation) program whose goal is “to develop and apply computer software technologies to absorb, analyze, and interpret huge volumes of speech and text in multiple languages, eliminating the need for linguists and analysts and automatically providing relevant, distilled actionable information to military commands and personnel in a timely fashion. Automatic processing ‘engines’ will convert and distill the data, delivering pertinent, consolidated information in easy-to-understand forms to military personnel and monolingual English-speaking analysts in response to direct or implicit requests.” Excuse me while I catch my breath. To this end, Raytheon BBN, which, by the way, is a member of TAUS, has assembled a team of companies, including Language Weaver and MIT, to pursue this goal by the rather traditional means of speech-to-text technology followed by MT and so on.
But given that the military needs something now, why not officially do what some soldiers are doing already — buy the SpeechTrans app for US$19.95 in the universal version that allows translation in multiple languages? However, critics say this only works when you are in places where there are transmission towers, not the hostile deserts where US troops are fighting. Wired adds that “the Army really, really wants to equip its soldiers with smartphones. It’s just also having a hard time saying goodbye to the expensive high-tech development projects that smartphones appear to make irrelevant.”
This has all come a long way from where I began with portable translation devices, namely the Seiko handheld translators. I remember being confused as to how to get out of a Tokyo park, which was totally surrounded by water. The only way out was a bridge that I couldn’t find on the way back. So, I typed in bridge in English, and the corresponding term appeared in katakana. I showed it to the nearest human, shrugged my shoulders, and he pointed to the bridge off in the distance, and I followed his sign language and eventually got out. That was high-tech portable language translation ten years ago, but hey, it solved my problem. Whether this current batch of expensive research and development will do any better remains to be seen.