|
|||
06-05-2010, 03:28 PM
Quote:
The trouble with thinking you understand the gist of the translated text is that if you don't know the original language, and if you don't have some other information to serve as a sanity check, you can't know whether the gist of the text has been preserved, corrupted, or completely garbled. For example, given the source here, it seems likely that the original sentence discussed some relationship between biology and natural history, but that's all you can conclude without further work. Not only is the meaning obscured, but the English generated is just plain ungrammatical. It would be prudent to assume that going the other way would likely produce equally ungrammatical Japanese. Personally, not knowing Japanese myself, I'm a big fan of machine translation aids. They've helped me to understand a lot of Japanese text that would be otherwise unreadable. And in the process of trying to squeeze answers out of these tools, I've learned some features of the language. But the problem is far from solved. Google Translate in particular needs vast amounts of work. I find the situation rather frustrating, since it seems clear to me that Google's machine translation system in particular seems to be dropping a lot of useful information on the floor which in principle it should be able to use to refine the result. It also doesn't tell you whether there is any ambiguity in the particular translation generated or what alternate translations for words or phrases might be available, so you could select something more appropriate for the context. |
Thread Tools | |
|
|