One thing which I've been wondering about, since I don't know anything about how such submissions are used (whether they're reviewed by human advisors, fed into some program, or more or less ignored), is what sorts of submissions we, the customers who utilize Google Translate's Web interface, can make when we're prompted to "Contribute a better translation?"
What is most helpful to the developers working on Google translate? What do you developers WANT to see happen? For example, I just opened the Web interface, and submitted: "What sort of feedback most helps the developers at Google translate?" Eenie, meenie, mynie, moe: okay, translate that into German: "Welche Art von Feedback meisten hilft übersetzen die Entwickler bei Google?" which is pretty good, I think, although I think "hilft am meisten" might be a bit better--but not being a native speaker of German, I wouldn't submit any suggestion at all. Or, for another example, I looked up "Shostakovich," spelt in Russian, on google.ru, picked out a page from a Russian Web site, and copied out a substantial-looking sentence for submission to Google translate: [quote] Композитор вновь впал в немилость в 1948: его музыка была названа формалистической и декадентской. Как примеры «типичного формализма» (т.е. искусства, непонятного народу и чуждого ему) упоминались Восьмая и Девятая симфонии, написанные в период 1943–1945. [unquote] I asked for an English translation, and got this one, also pretty good: [quote] The composer once again fell into disfavor in 1948: his music has been called the formalistic and decadent. As examples of "typical formalism" (ie art, incomprehensible to the people and alien to him) mentioned Eighth and Ninth symphonies, written between 1943-1945. [unquote] (By the way, I notice a bit more ease in copying this--so I think the developers have already modified some of the javascript controls to make it a bit less difficult to copy and paste results, something I commented upon in another thread...I am not sure, because I'm using an earlier version of Opera, at the moment, than I normally use--it may just be that the javascript interpreter in that earlier version is NOT interpreting a buggy script and forgiving the same errors as the later one is!) To return to our muttons, as the French say, this English translation is quite good, and I understand the general point of the sentence perfectly, but of course, "ie" is not written correctly: it's the abbreviation for "id est" and needs periods: "i.e."--not to mention the fact that it's not an English translation, but a Latin one--the English for "т.е." would be "that is." More to the point, I am GUESSING that the Russian sentence is supposed to be translated into the passive voice: [hypothetical output] As examples of "typical formalism" (that is, art, incomprehensible to the people and alien to them) were mentioned the Eighth and the Ninth Symphonies, written between 1943-1945. [end hypothetical output] I am fairly certain that none of these "improvements" would be of any help to the developers at Google Translate. As a much more difficult example, let me take two languages which are impossibly remote from one another, and extremely difficult to inter- translate. A Japanese "mangaka" or Comic-Book artist/author maintains a "blog" on which he describes some of his projects. In a recent post, he spoke about a translated edition of one of his "picture stories," which was published in France, and is describing how a particular part of his comic book ran afoul of France's obscenity laws and had to be altered for publication in France, even though nothing in Japan's obscenity laws prevented it from being published in his own country. Here's a passage from his Japanese blog: [quote] どこかというと、クレスケンスが幼少時の思い出話をするシーン。ここの、父親に犯され云々のくだりが、ちょいとあちらの出版コードに抵触してしまうとの ことで、台詞やコマ割りはそのままに、絵のみ４コマ変更しました。海外向けに絵をいじったのは、これが初めて。 御参考までに、その修正箇所があるペー ジを下に載せますので、日本語版をお持ちの方は、宜しかったらお見比べあれ。 [unquote] I submitted it to Google translate, and, in spite of the horrible remoteness of the two languages, and the colloquial, chatty tone of the author--and even the use of a proper name of one of the characters in the comic-book about which he is talking, a Roman gladiator named "Crescens!"--Google translate came back with, if not a terribly articulate bit of English, at least something that gives SOME idea of what he was talking about: [quote] Somewhere and say, a scene in which the childhood reminiscences Kuresukensu. Here, the passage of being raped by her father yada, and that publication would violate the code and there's little, dialogue and panel layout is intact, only four changed their picture frames. The Overseas tweaked the picture, the first time. As a reference control, so that the page is below the corrections, you have the Japanese version better, you can compare it 宜Shikattara. [unquote] Now, the funny thing is, you can grab the phrase translated as "Somewhere and say" and search GOOGLE using the transliteration of it, "doko ka to iu to" and you will find an extremely orderly explanation in Google Books of how the Japanese use this phrase. After looking that up, I see that what the Japanese author would be saying in proper English would be: [hypothetical] If you would like to know where [in the book this happens], . . . [end hypothetical] and the first phrase highlighted by the "Contribute" pop-up, "どこかというどこかというと、クレスケンスが幼少時の思い出話をするシーン。" would be better rendered, "Specifically, the scene in which Crescens speaks of his memories of his childhood." or "To be specific, the scene in which Crescens talks about his memories of childhood." So there are a number of things one wants to "feed back" into Google's self-improvement program, and I'm not quite sure HOW someone like me can best use this mechanism to be most effective and helpful in improving the difficult problem of machine translation. I've described this at some length, because I don't believe there has been much detailed discussion of it in this group, and I'd like to hear what Google engineers have to say in response. Perhaps they can use the examples I've given here to elaborate. I think most customers are very grateful for this service, which is badly, badly needed, and WANT to do as much as they can to help contribute to better translation--but, like me, they don't know how. The huge success of "EDICTJ" is testimony to how much can be done when people help and contribute corrections and suggestions--tell us how WE can best help YOU when we "contribute better translations. -- You received this message because you are subscribed to the Google Groups "General" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-translate-general?hl=en.
