Obama had narrowly won a second term, stimulating a sustained new wave of people power worldwide, helped by a rage-against-the-machine trend following a series of massive computer virus attacks.
The inexorable rise of the Chinese economy over the previous decade led to a large volume of translation work being taken over by Asian language service providers (LSPs), riding on the back of low wages and widespread but low-quality machine translation (MT) technology.
In an effort to claw back their business, other providers began to criticize translation quality and propose as an alternative very efficient, real-time, high-quality, human networks of translation built around soft technologies.
On the R&D front, insight into the translation process in the human brain using advanced fMRI scans had led to the development of powerful programs that rapidly train people to input spoken translation to the network in near real time across hundreds of language pairs.
The SubberTime Crowd. Dolsie Cabrero and Sami Trabulsi are taking a smoothie during a break in discussions at the Sub Dub hack event being held this year in South Africa. The Crowd are road mapping their technology development for the next two years, with the idea of offering an open source platform for huge numbers of people to tag subtitle chunks for easy search and recovery by translators. Dolsie is an active member of the EdVid network dedicated to multilingualizing premium lectures of educational content in near real time. Sami oversees the technical standards group that are forging a new version of the LNetwork standard, which enables ad hoc language tool developer networks to interoperate with each other in real-time to speed up testing and debugging.
Dolsie: Hey Sami, good to see you in person after so much virtualizing. What d’ya think of South Africa for this year’s venue?
Sami: Terrific for me. I can keep in touch with my little boy back home in Tripoli in the same time zone for once. A bit cold at night, though.
Dolsie: Right. Manila’s still 6 hours ahead for me. By the way, I liked that demo of the keyboard adapter for the West African Languages community. Much better to have a real piece of hardware than the so-called virtual keyboards that no one much liked.
Sami: Absolutely. It’s been like that with the Arabic keyboard for a long time as well: a mashup of Latin alphabet and Arabic using the same keys for different input directions. And kids were never encouraged to learn to touch type in Arabic due to the price of the equipment. Now that young Arabs are getting together as a community to debug the keyboards and come up with better solutions, they’ll transform the way language work gets done in their countries. Once you get good cheap tools built with user input, you can unblock the interface logjam and get people writing and especially translating more easily.
Dolsie: It’s funny, isn’t it, looking back at the old days when everyone thought machine translation would rule the roost and put translators out of work. Basically intellectual technologies of the sort we are using and developing are much better at helping people, than taking their place. Once whoever it was managed to persuade telecoms companies to make always-on-everywhere services available for a fraction of the cost they were ten years ago, we were all able to work together to handle some of these really hard problems. Like translating online meetings in real time using touch-typists and text to speech output that really undercut the old interpreter price model.
Sami: Did you get involved in that too?
Dolsie: Well yes, that’s how I got into educational subbing/dubbing using the same ‘type to talk’ model we built in the network. We’ve been trying out some fantastic new advances from Piotr and Angelika demoed here in Cape Town, that allow you to correlate the real time translation of speech with a search engine that finds any document or even video references the speaker was talking about for later checking. Very handy in multilingual legal or pharma contexts, apparently. You just help people do their jobs better.
Sami: if I look back, what I’ve been most impressed with is how standards can be established in such a short time, using these new standards building platforms that apply machine learning to do the boring bits and human/machine testing to do the sensitive, hard bits. In the old days they used to sit around in committees for years to agree on a standard. Nowadays you can cut out a lot of the waiting time by letting the infrastructure learn and model what standards you need and then circulate it on the social networks to get rapid buy-in and ownership.
These days the Chinese are moving into West Africa and setting up educational institutions, so we’re trying to populate the cloud with what we call ‘new-language TMs’. The idea is for people with mobile phones in West Africa to generate parallel speech corpora in their spare time between French/English and Yoruba, Igbo, Wolof and Peul using a friend as a consecutive translator, chunk by chunk. We can give points to the crowd to help transcribe and mark up the corpora of chunks, using the standards we’ve developed for sharing and pipelining texts. These will then be used to build new translation resources really quickly to support Chinese and local language communications.
Dolcie: Doesn’t MT work at all for this? I thought they’d broken the “sparse resource” barrier for some of these MT systems at least.
Sami: Sure it’s possible, but there’s been that big worry about the quality of international communication after the translated spam fiasco, and the large international agencies and others have been talking about “cognitive pollution of the environment”, so there’s less interest in pushing MT on its own. And earning points and easy micropayments have made it really worth while getting people with a gift for language work to join the community at a very low entry cost. Especially in these multimedia gigs you’re involved in, I bet.
Dolcie: Absolutely. Trouble is it’s gradually going to wipe out business for most of the LSPs. They’re trying hard to promote their global management skills to keep out the rising crowd, but translation management is just becoming a press-button function in a smart internet. Local speed translators can quickly create a flash crowd to handle jobs for clients much faster and cheaper than LSPs. Anyone can share anything and in any case content is changing so rapidly the big boys just can’t keep up! Everyone just expects a translation – they don’t want to know how it got there!
Sami: Well, you can’t stop people power. I love it.
BACKGROUND TO THIS ARTICLE
This ‘day-in-the life’ story describes a possible future scenario in the translation industry given certain conditions. Download the complimentary TAUS Vision Report found below to see how a group of the leading global buyers came up with this and other possible future scenarios for the translation industry. If you missed the first scenario we published a few days ago you can find it here.
DOWNLOAD THE ACCOMPANYING REPORT
Scenario Based Planning: Planning for an Uncertain Future
A TAUS VISION REPORT
Author: Rahzeb Choudhury, TAUS, and Greg Oxton, The Consortium for Service Innovation
Download this complimentary TAUS Vision Report to see how a group of the leading global buyers came up with this and other possible future scenarios for the translation industry.
