Here’s an interesting article from Newsweek. It talks about new translation machines that apparently seek to make instant translations of any language available without the need for human intervention. I have some fairly strong views on why such a machine is by definition very limited in what it could do, but for now I’ll reserve those comments.
Alex Waibel doesn’t understand Chinese, but he can read street signs when in Beijing. A team of engineers led by Waibel at Germany’s Karlsruhe University has developed a heldheld device called the Sign Translator. It uses an integrated camera and software that recognizes, and translates into English, about 3,000 Chinese characters.
What do you think about this? I’d like to hear your opinions. Do you think such a device could work? What are the limitatioins of such a machine?
Ken Carroll 凯恩
Device long over due, with all the previous hype about AI, and slowly developing Voice translation devices nice to see some serious progress in area the world most definitely needs. Perhaps it will help knock down some of the barriers we have with other language outside of our own.
I’d like a program like that on my phone, for the tricky characters I can’t find in my dictionary.
Yeah, it’s a totally cool idea, but there’s so much stuff in Chinese that totally defies machine translation. The example that I’ve had stuck in my head lately is the band name “重塑雕像的权利.” Without context, there’s nothing to tell you whether it means “the right to rebuild statues” or “rebuilding the rights of statues.” (The band chose the English name “Re-establishing the Rights of Statues;” the native speakers I’ve run the phrase past have all interpreted it as “the right to rebuild statues.”) This kind of stuff is even more of a problem than machine segmentation of text.
Also, I remember once feeding Adso the first couple of poems in the 诗经 and watching it segfault.
I often use electronic translators to get “a feel” for something sent to me in a foreign language. This usually tells me just enough to know whether it is worth the money to hire a real translator to translate. I do not see how a devise like this could ever go beyond the abilities of electronic translators, which work less well for Chinese than for languages closer to English. Even asssuming a devise like this could get up to 80% accuracy, is it good enough to be heading in the wrong way 20% of the time?
interesting stuff
looks like writing will be figured out first, before speech - i’m glad i never wasted any time on characters
unfortunately it seems technology is perpetuating characters instead of helping them go extinct
謝謝你的幫忙!
Wo du shu zhong wen yi nian ban, dan shi, wo xue hen duo zhong wen!
I have been searching and searching for adequate teaching material, both in stores and on the web. Wode baba gao su wo… “wo tingshuo podcasts hen bang o! ni yin gai ‘check it out’…” hehe, ta bu hui shuo zhong wen, dan shi, wo te bie xi huan lian xi
you shi hou… wo kan zhong wen dian ying, ne ge bang mang. Xie xie nide bang mang, wo kan ding wo hui shuo zhong wen ‘fluently, soon’.
I am new to this, so I don’t know where to post my suggestions. I think I’ll just post it here.
I would like you to do a lesson/dialogue on school subjects…
ex: a school day… two friends dicuss their subjects and their day and if they like their positions.
Thanks!
This discussion’s prob dead by now, but I’d say that somthing like that can be great as a phrase book type thing, and its a step in an interestion direction. On the other hand, linguistics has some way to go with understanding how the human brain processes language before natural language processing will work really effectively. There is also some problem in the understanding of speech without training a program to understand your own speech. A computer can’t cope with the little differences between people’s speech. It’s the same as the computer can’t easily know what letters are written below, but a human can see it easily.