Brian Conley founded Alive in Baghdad back in 2005, using Iraqis to create weekly video packages reporting the news from the ground. Since then, his group, Small World News, has rolled out projects in Mexico, Gaza, Tehran and other locations in the Middle East and Latin America.
His latest projects are Alive in Egypt, a project to crowdsource tweets phoned in via Twitter and Google’s speak2tweet tool during the #jan25 uprising, and its successor, Alive in Libya. Now, however he’s working on something on a larger scale: machine-generated subtitling for aggregated video coming out of conflict areas. But the stumbling blocks for automation in the Arabic world are formidable.
Given the flood of information coming out of even a communications-compromised country like Libya, the need to massively but reliably translate is obvious. Given how much of the information is video – a lot of the reporting done around the hashtag uprisings has relied on participant-generated video – the ability to translate into subtitles is increasingly urgent.
Obstacles to Automation
Small World News has partnered with Universal Subtitles to subtitle translations onto video coming out of Libya, Bahrain, Algeria and elsewhere. Ideally, a volunteer would transcribe the audio from a video and machine translation could be employed to do fast translations and positioning.
There are two reasons, endemic to the Middle East, that make that impossible as things currently stand.
As Conley said in a recent conversation:
“A lot of our translators are second-generation Arabic speakers. They’re fluent but they can’t read or write Arabic. I can, but I’m not fluent.”
Even those who can read and write fluently are hamstrung by the keyboard issue. Conley said he himself has three different virtual keyboards, that map his QWERTY keyboard to Arabic script. Each of them has different positioning. There currently exists no standard for QWERTY-to-Arabic keyboard mapping.
Today, the United Nations issued a press release detailing the Secretary General’s conversation with Qaddafi. Although today is February 21, the press release was dated February 12. Whoever translated it did not reverse the dating convention (right-to-left in Arabic, left-to-right in English).
As Conley put it:
“There’s a lack of interoperability within that space that creates a technical barrier to communications.”
These specific barriers means that Arabic-speakers cannot automate much of what they do online. That in turn means a developer or company will have to expend more work-hours, and therefore spend more, to do the same project that would be cheaper for a developer using the Latin alphabet.
The Prestige
Counterbalancing these structural obstacles is the increased prestige that online communications technologies are seeing, due to their use in the uprisings, across the whole of Arabic societies. (To the point that one Egyptian has named his new-born daughter, “Facebook.”) Yamli, a kind of Arabic Google, which is run by one of Small World News’ volunteers, has seen its traffic skyrocket in recent weeks.
Should the uprisings that are shaking the countries of the Middle East wind up stabilizing them, there will exist a huge opportunity for technology entrepreneurs – individual developers, local companies small and large and international operations.
Job one may be standardizing virtual keyboard layouts and automating the transition between Latin and Arabic documents. Whoever does this may give birth to an Arabic technical renaissance.
Photo by telecentre.org