What the Hell?


Update: I’m so amazed by this wiz-bang page translator I added the buttons at the top of the second column. Those who have a better working knowledge of those languages are invited to comment on whether the machine translations are too craptacular to be functional.


  1. My wife says it’s actually not very good, but better than Babelfish or Altavista. It gives you the basic idea, but none of the subtlety. I would say that for Japanese, Chinese, and Koreans who speak no English and want to know what we’re thinking and writing about over here, it lowers the language barrier to a degree.

    I’m sure that better linguists than me will ridicule it, but I’ll keep it as long as my visitors’ log shows me that people are using it (and it does). I actually discovered this through my visitors’ log, when I noticed that people in Japan were using it to read one of my posts on the Guano Rocks.


  2. I used to try to translate Korean newspaper articles … I eventually gave up.

    It was a combination of things … intricacies of the Korean language and the distressing paucity of dictionaries that are written to an American audience. (The ‘Essence’ and ‘Prime’ series of dictionaries are written for Korean readers.)

    The conclusion that I came to was that if one didn’t already know what the article was about, it became an almost impossible (for me, anyway) task to translate … at least anything beyond elementary school-level materials.

    Machine translators have a lot going for them, but they are faced with the same problem, i.e., which of the, say 5 or 6, different definitions given for a particular word, does the translator select for any particular instance.

    I’ve found a useful ‘acid test’ is to take articles at the Dong-A Ilbo website and run the ‘Korean Text’ page through the translator and then compare the resulting output product with the translation provided by the Dong-A Ilbo on its ‘English Text’ page.

    After trying this on several Dong-A Ilbo articles, the impression that I have is that the machine generates the same type of ‘garbage’ that I would … but a lot faster … and I mean a LOT faster.