Subscribe to the Open the Future feed
Translating Opacity
Andrew Revkin asked what I thought about his arguments for greater development and use of automated language translation technologies. In his piece "The No(w)osphere," Revkin writes:
As the human population heads toward nine billion and simultaneously becomes ever more interlaced via mobility, commerce and communication links, the potential to shape the human journey — for better or worse — through the sharing of ideas and experiences has never been greater. [...].But language remains a barrier to having a truly global conversation...]
Automated translation remains clumsy, at best, these days. (One perfect illustration is the website "Translation Party," which translates an English phrase into Japanese, then translates it back to English, then back to Japanese, until it reaches "equilibrium" -- a point where the English and the Japanese auto-translate back and forth precisely.) Linguistic accuracy is a much harder problem than technology pundits of a few decades ago had expected. Nonetheless, as Revkin points out, there are a number of projects out there that suggest that a future of relatively useful automated translation is probably fairly near.
Here's the twist: I suspect that a less-than-perfect system would be better than an idealized perfect translation. Why? Because an imperfect system would require us to speak more simply and in a more straightforward fashion, with fewer culture-specific idioms and convoluted sentences, as we do today with our current tools. Working with people for whom English is not their primary language, I know that I need to speak and write in a way that doesn't lend itself to unintended ambiguity or confusion. If I knew that an automated system could be tripped up by overly-complex language, I'd be as careful and precise as possible.
But in everyday conversation, we don't tend to speak carefully and precisely. Correspondingly, an effectively perfect system would let us slip into the kinds of discussion and writing patterns that we use with other native speakers. I suspect that, counter-intuitively, this would lead to more confusion and friction, as meaning is culturally-rooted. A perfect translation of the denotation of a word or phrase may not carry the correct connotation; moreover, the translated word or phrase may have a very different connotation in a different culture.
In other words, translation technology that offers results that make sense linguistically, and carry the proper surface meaning of the words and phrases used, could well be close at hand. But translation technology that offers results that have the same meaning in both languages, especially with complex or idiomatic phrasing, probably awaits the arrival of relatively strong machine intelligence. Simply put, it would require software that understood what you meant, not just what you said.
We should be careful not to get these two outcomes confused. The more that we expect our translation tools to convert meaning, not just phrasing, the more likely we are to be unhappy with the results.
Comments
Having fed the translation party with some phrases of your post, I found that single words suffice to fail reaching a fixed point and that oscillations of different periods are easily obtained, e.g. with
The latter also leads to an admittance of guilt in the first backtranslation, stating
I was counterintuitive, to further increase the confusion and conflict that has taken hold of it will lead to cultural meaning.
Then I tried a shorter sentence with simple structure:
I really enjoyed the obsession with repetition of numbers in the translations which converged toward a phrase fit for Monty Python
We have two, one two three one result, the two must be careful not to confuse the two.
Bottom line? Three cheers for the linguistic king of the ring
Posted by: manoser | February 10, 2010 4:03 AM
I see Translation Party as illustrating the fun of getting to know each other through an implicitly funky translation widget on, say, Facebook. This all struck me while I was in Istanbul reporting a story on the impending earthquake there. I met a bunch of kids (video: http://j.mp/IstanKids) who were greeting me by asking, "Facebook?" I thought of them playing Farmville, or a muliplayer game, with my son back home. Simple vocab. Playfully wrong translations. Still beats Esperanto.. : )
Posted by: Andrew Revkin | February 10, 2010 7:47 PM
@Andrew: Automated translation remains clumsy, at best, these days. (One perfect illustration is the website "Translation Party," which translates an English phrase into Japanese, then translates it back to English, then back to Japanese, until it reaches "equilibrium"…
Translation Party does not show that machine translation is clumsy. Translation Party shows that if a phrase is repeatedly translated between 2 languages, frequently, the equilibrium is nonsense. Possibly, the first translation is excellent.
@Andrew: If I knew that an automated system could be tripped up by overly-complex language, I'd be as careful and precise as possible.
@Andrew: In other words, translation technology that offers results that make sense linguistically, and carry the proper surface meaning of the words and phrases used, could well be close at hand.
Many companies successfully use machine translation to translate technical documents.
If text is optimised for machine translation, machine translation usually gives satisfactory translations (http://www.international-english.co.uk/mt-evaluation.html). The quality of the translations is dependent partly on the languages. The results for English to Welsh are not good. (The results are not published. The results will be available in approximately 1 week.)
Posted by: Mike Unwalla, TechScribe
|
February 11, 2010 1:34 AM
Google's "Babel Fish" translator will in never solve the language problem. Not only does it discriminate against anyone who cannot afford a mobile phone, but against minority language groups as well.
There are 6,800 languages worldwide, not fifty-two !
Moreover, if I met a native in Borneo, and he said to me in Hakka "I've lost my mobile phone" how would I understand him :) And how many starving Africans can afford a mobile phone !
As English loses its economic power, the answer is not for us to move to Mandarin Chinese, but to Esperanto which puts all speakers on an equal footing.
Have a look at http://www.lernu.net or http://www.esperanto.net
Posted by: www.facebook.com/profile.php?id=1004522862
|
February 11, 2010 5:43 AM
I think the 'answer' to this is the rise of an international trade pidgen, like the Lingua Franca of the Mediterranean, or East Africa's Swahili. An obvious candidate is some form of reduced grammar English, like Globish, which is actually taught to workers who do not share common languages by many multinationals, like oil companies.
Rather than seeking perfect translation from language A to language B, members of those two languages would converse and condust business in Globish.
Posted by: Stowe Boyd
|
February 16, 2010 1:58 PM
What's more, it's expected that the world will hit five billion mobile phone subscriptions this year. Some of those will certainly be multiple accounts for individuals, but within the next five to ten years (barring calamities), it's safe to expect that almost everyone who wants a mobile will have one. Babel for All, or something.
Posted by: Chris L | February 19, 2010 1:09 PM