sethg Wrote:I think he might have culled them from the Tanaka Corpus, which is notoriously bad.From what I understand, most of the problems with the current Tanaka Corpus are bad translations into English or incorrect kanji usage. Aside from that, most (not all) of the Japanese text should be fine. Jim Breen himself has said so:
Jim Breen Wrote:Apart from typos and 変換ミス the Japanese is usually OK, but the English is sometimes either a horrible literal translation, or completely wrong.The corpus is also several years old, and he said that it's considerably better than it was before. For example, I use an old version of the Tanaka Corpus in WaKan, and when I see a mistake in it, I usually go to WWWJDIC to correct it... only to find that it's been corrected already.
) You'll note that he says 'often copied out of textbooks' -- I do wonder whether in some of those cases the Japanese text was overly influenced by the English it was a translation of. (after all it's trying to convey the sense of the English original to a native Japanese speaker, not vice versa).
