TED “Open Translation Project” is not so open

by Suw on May 16, 2009

I prick my ears up every time I hear words like “open” and “translation” and “project” used together in a sentence in case it’s something that might be interesting to the Welsh language community. So when I saw that TED, the insanely expensive Technology, Entertainment, Design conference, was providing English language transcripts of its often amazing talks, I thought that was a great opportunity for the Welsh community to create the Welsh language subtitles.

It might seem a bit stupid for people who basically all speak English to translate an English transcription into Welsh so that other people who basically all speak English could listen to it in English and watch it in Welsh. In fact, such a thing would be a great boon to learners as it would allow us to hear in English but read in Welsh and start to join up the dots inbetween. (The next step on from that would be to have more material in Welsh with English subtitles.) Personally, I would like to see much more bilingual stuff online because it helps learners develop not just their vocab but also their sense of grammar, mutation and idiom.

Many of the TED lectures are both short – about 20 minutes – and fascinating. Great material for translation because they’re interesting in their own right and just the sort of thing that people love to watch. I believe you learn much more when you’re engaged, so such a resource would be very valuable.

But the Open Translation Project turns out not to be very open at all. Once you’ve registered, you have to request a transcript to translate, which will supposedly be sent to you within 1 – 2 business days. You can’t just download one that you fancy and get on with it, it has to be sent to you by someone from TED. And you have just 30 days to complete it – what happens if you don’t is not specified. I have so far seen no sign or mention of any online translation or collaborative working tools, so it looks like if you want to work with others you have to figure that out yourself. (I haven’t been accepted into the programme yet, so maybe that’s just stuff I am unable to see.)

Beyond that, as a new member of the translation project, I just received this email:

Dear Suw,

Thanks so much for registering to be a TED translator and requesting your first talk. We’re eager to get you started! But we have a few questions for you first. As you know, TED doesn’t require translators to have formal language or translation skills. We do however, ask that you be fluently bilingual. It’s so important that your language skills gives you the ability to faithfully translate the words of speakers, capturing not only the vocabulary, but also the tone, style and personality. TED speakers are at the edge of their fields, and therefore the edge of language. Being current, as well as fluent, is key.

So the following questions are for you, as much as us. They provide us a way to gauge your experience, knowledge and fluidity with both English and the language into which you’re translating. We require these answers for translators in languages that are new to TED, and for which we have neither in-house knowledge nor a stable of volunteer translators.

1) What language do you want to translate into?

2) Is this your native language? If not, how and where did you learn it?

4) How often do you speak this language? Do you use it professionally, personally, or both?

5) How often do you read in this language? Do you read news? Novels? Personal correspondence?

6) How often do you write in this language?

7) If English is not your first language, how and where did you learn it? How often do you speak, write and/or read English?

8.) Do you have other colleagues, family or friends who can assist you on the translation of tricky or culturally-specific words and phrases?

9) What is your profession?

10) Why do you want to translate for TED?

Thank you for taking the time to answer these questions, and for volunteering for the TED Open Translation Project!

The TED Team

That doesn’t seem very open to me. That seems very much like I’m going to be judged on my ability to speak and write Welsh, and that my acceptance into the project is going to depend on my capabilities rather than, say, my ability to gather a kick-ass team of Welsh speakers to collaboratively – and openly – translate. Furthermore, what does my profession have to do with my ability to translate? What does my motivation matter? And what happens if the TED Team don’t like my answers? Do I get summarily booted out?

I feel rather insulted by these questions, not just because they are intrusive, but because they see the translation process through an outdated and judgemental lense. As a learner of 10 years, I’m not too bad at Welsh, although I write it better than I speak it. I probably could not create a perfect translation of any but the simplest texts. But what I can do is create a flawed translation that others, whether more experienced learners or native speakers, can then polish up. This idea that making a start so that others can help finish up is a well established way of working collaboratively, and it can produce great results. It’s what Wikipedia relies upon, it’s what Pledgebank encourages. By showing the community that I have committed to an action, I’m more likely to find people willing to help me finish it.

TED’s approach to translation has been disappointing to say the least. They have used the word ‘open’ as a buzzword, a way to put a gloss on what is an old-school project that is snobbish, closed, and controlling. I know that someone is bound to leave a comment saying “but that is the only way to get high quality translations”, but that’s just not true. Communities of passionate people are capable of great things, and there are many passionate Welsh speakers online who could come together to do flawless work.

Question is, will TED let us?

Tim Morley May 16, 2009 at 12:28 pm

I can understand where TED are coming from with this. I got the same email from them yesterday, and I’ve duly flashed my credentials and sent it back for approval.

I agree with you that it is possible to produce high quality work collaboratively, be it translation or something else, but you’d have to admit that it’s also quite possible — and more frequently the case — that the result is incomplete, badly done, half-arsed and unreadable; and this, presumably, is not the image that TED is hoping to project to the non-English speaking world.

If they don’t have an in-house team to do QA on the finished work — or to distinguish a work in progress from something worth publishing — then I think it’s quite reasonble to ask for some credentials.

Suw May 16, 2009 at 12:56 pm

I’m afraid I disagree about credentials ensuring anything. A native speaker could still produce a rubbish translation, and credentials do nothing to prevent half-arsed, unreadable work. There’s a saying in programming that with enough eyes, all bugs are shallow. Collaborative translation is not always easy, but it does provide an opportunity to get enough eyes on the project that mistakes are more likely to be spotted.

All asking for credentials does is weed out drive-by participants who are unlikely to really do much, but as they are self-weeding-out by their nature – the fact that they’d be unlikely to finish and submit a translation – I’m not sure that TED are achieving anything. Other than feel that they are retaining control, perhaps.

I’d rather see a genuinely open translation project that used the power of collaborative technology to welcome all comers and help ensure that the results are high quality. It might take a bit more effort, and be a little bit harder, but it would actually be a valuable project for them to engage in. Particularly if it was open source. The, maybe, they could use the words “open translation project” and actually mean it.

Tim Morley May 16, 2009 at 7:21 pm

You’re right that the credentials requested don’t actually prove much; you can speak a language quite competently without having any clue about translation. The fact that there are a couple of (not really very arduous) hoops to jump through will, as you say, weed out drive-by participants before they have chance to do any damage.

You say that collaborative translation is a “bit harder”, but I’d say it’s an order of magnitude harder to do properly. See, for example, the trainwreck that is the Rosetta project at Launchpad.net; at least, it was two or three years ago, the last time that I wasted dozens of hours of my time using it. Maybe it has improved radically since, but all it achieved for our team was to pollute our well researched, corrected and proof-read work with hundreds of “corrections”, incorrect additions, and very little of any actual substance; and then it removed all the comments from our translation files where we’d carefully documented our research, and also left no indication of who was responsible for which edits. Thanks guys, three months wasted.

Now, if Canonical can get it so badly wrong — when so many things about Ubuntu seem to be spectacularly right — I can’t blame anyone else for doing it traditionally.

It’s an interesting problem, and one that I think can be solved, but I’ve yet to see a good implementation of a solution. Have you? I’d love to give it a once-over if you have.

Charbax May 16, 2009 at 8:51 pm

What’s holding Google up for providing automatic subtitle generation for all Youtube videos and also automatically translate them all?

We don’t care if speech recognition has mistakes, or if automatic translation has mistakes! Youtube has hundreds of millions of users, let them all participate in correcting the speech recognition and translation mistakes in some type of Wiki sort of way! And let the translators even be able to earn money as people can pay to request faster subtitle translation correction to be posted by an international group of open translators!

When Google activates automatic subtitle generation and translation, Youtube is going to be 10x huger even!

Tim Morley May 17, 2009 at 12:17 pm

Turns out, actually, that once you’ve jumped through TED’s hoops, you’re given a URL to translate one of their videos via an unmodified DotSub interface. And DotSub is a collaborative translation platform, so you’re free to share the URL with other people whom you’d like to help you, or, I suppose, publish it on your blog to solicit help from the world and his dog.

Suw May 17, 2009 at 1:17 pm

Tim, I think the key to collaborative translation is to have people in the team that you trust, rather than letting it be a total free-for-all. Lord knows I’ve come across learners who think they know better than native speakers, and native speakers whose grip of grammar is way below that of learners. It’s easy for well-meaning people to screw up a translation, and it’s easy for stubborn people not to listen to others who do know better.

So the key is building trust within a team, creating strong bonds between co-translators, and facilitating positive interactions. That means the interface is important as it has to allow for good team management and not just present the text for translation. Given the right tools, I see no reason why it should be “an order of magnitude” harder, but to be honest, I don’t know if the right tools exist – I haven’t seen them, but then, I haven’t looked that hard.

I found out this morning that I had managed to jump through TED’s hoops and am now in the DotSub interface, which I can’t say I like very much. It’s only vaguely collaborative, and is missing many of the social features I would expect to see in a truly collaborative platform. Colour me deeply unimpressed.

If anyone is interested, I’ve started translating here:


Charbax, I think you’re overestimating what current technology can achieve.

Carl Morris May 20, 2009 at 12:43 am

These people clearly do not realise that Welsh learners could be among the most potent forces in the known Universe. Not to minimise the role of a kick-ass team or of experts, but I have noticed a particular zeal among Welsh learners for starting online projects.

(I wonder if that’s mirrored in other languages? Maybe it’s a minority language thing?)

At the beginning of these projects it’s just about having a lone zealot to smear a blank page into life. When enthusiasm is more important than knowledge!

The fine tuning can wait. I think it was Eric Raymond who originally said “given enough eyeballs, all bugs are shallow” – love it. If anyone hasn’t read The Cathedral and the Bazaar essay, then do so.

I liked the idea of dotSUB when I first saw it and watched a few videos. But when I visited your translate link today and tried to enter some text it didn’t even save at all on my browser (Firefox). It’s got some errant JavaScript refreshing every couple of seconds.

Therefore I can’t work with dotSUB, sorry, at least for the moment. I’ll be back when it works, even if it’s tad clunky.

Flawed translation is my middle name.

Cathy Ma July 8, 2009 at 9:29 pm

It seems that TED is falling back into the challenges where Wikipedia’s predecessor Nupedia faced – created a high bar of entry to ensure quality of the user created content is up to certain standard.

By all means that is all with good intention, however if their objective is to get high quality content fast, it’s much better to follow the low bar of entry system but introduce a few extra layers of volunteer reviews before the content goes live.

‘With enough eyeballs, all bugs are shallow’.

In that case, ‘With enough eyeballs, all typos are rid-dable’.

Comments on this entry are closed.

Previous post:

Next post: