Machine translation has been in the news lately, so I thought it might be interesting to conduct an experiment. I've chosen four different Chinese texts (excerpts from a novel, a film and two newspaper interviews), translated them into English with Google Translate, and added my English translations (three of which have appeared on Paper Republic in the past year). I'm sure most of the translators in our forum have their own machine-translation stories...hope you'll share. That's not to say that machine translation is pointless: ten years from now, we will be taking this a lot more seriously. But in the meantime, we might as well have our bit of fun.
(1) a novel - Yan Lianke, Dream of Ding Village (original translation stub on my translator bio)
Chinese: 六是有了这病必得死，是人世上的新绝症，花多少钱你都治不愈。七是这病其实也才刚开始，大爆发要到明年、后年才来到。那时候，死个人就像死只麻雀样、飞蛾样、蚂蚁样。现在死个人像是死条狗。狗在世上比飞蛾、麻雀贵重得多。
Human translation: The sixth thing was that if you got it, you died. AIDS was a new and incurable disease, and no amount of money could save you. But the sickness had only just begun. That was the seventh thing. The real explosion wouldn't come until next year, or the year after next. That's when people would start dying like sparrows, or moths, or ants. Right now they were dying like dogs, and everyone knows that in this world, people care a lot more about dogs than they do about sparrows, moths or ants.
Google translation: With this disease Betrader 6 is dead, is a human world, a new fatal disease, how much money you have unhealed governance. 7 is that the disease is actually just beginning, the outbreak until next year, the year after it arrived. At that time, only the dead person as dead sparrow-like, moth-like, ant-like. Like a dead person is now dead dogs. Dog in the world than the moths, the game is much more valuable.
(2) film dialogue - Robin Weng (Weng Shouming), Fujian Blue
Chinese: 去出国挺危险的，我前几年去台湾，从平潭东澳过去，一般平潭仔去台湾都是从东澳偷渡过去，不少人没上岸就被抓起来，在“大陆人民宜兰处理中心” 关个一年半载再遣返。那里的管教员最坏,日本的要好一点……捷克的最人性化，每星期六还放三级片给我们看。
Human translation: Going abroad is risky. A few years ago I snuck into Taiwan, went across the straits at Pingtan. That's how the locals stow away. A lot get caught before they reach shore. I spent a year and a half in lockup, and then I got deported. Taiwanese guards are the worst. The Japanese are better, but the Czechs are the most humane. Every Saturday, they let us watch porn.
Google translation: Quite dangerous to go abroad, I go to Taiwan a few years ago, from Pingtan eastern Australia in the past, the general Pingtan Aberdeen to go to Taiwan are illegal immigrants from the East Australia in the past, many people were arrested and did not come ashore in the "dealing with people on the mainland Ilan center "off a year or so and then return. Corrections where the worst of Japan a little better ... ... the Czech Republic the most humane, also released three films every Saturday to show us.
(3) news interview - Han Han on the role of the public intellectual (see PR archive)
公共知识分子和公共厕所有时候是一样的，供人临时发泄，泄完了还不打扫，并且必须是免费的，你要是收五毛钱草纸费就还得踹你两脚墙。一个城市如果没有公共厕所，那么很多人只能在街上拉屎了。所以这个角色有时候其实很可悲的。但是，如果整个城市的人，哪怕家里有卫生间的人都来这个公共厕所大便，那么这个社会也许是乐观的。
Human translation: "Sometimes being a public intellectual is a lot like being a public toilet. Anyone can stop by and take a piss for free, and they don't have to clean up afterward. If you try to charge them 50 cents for toilet paper, they'll bitch about it and start kicking at your walls. But a city's got to have public toilets, otherwise people just crap in the streets. It's a pretty pathetic role sometimes, but if everyone in the city, even those who have their own bathrooms at home, comes to take a dump in the public toilet, well then... maybe there's some hope for this society yet."
Google translation: "Public intellectuals and public toilets are sometimes the same, for people to temporary vent, vent cleaning is not finished, and must be free, if you received fifty cents toilet paper costs have to kick your feet on the wall. If there are no public toilets of a city, then a lot of people can only shit in the streets of the. Therefore, this role is very sad sometimes. However, if the entire city of people, even a toilet at home who came to this public toilet stool, then the society may be optimistic."
(4) news interview - Wu Wei, deputy director of the State Council Information Office/head of China Books International (see PR archive)
Chinese: 贾平凹作品里有很多陕西方言，我们说普通话的人都不懂，外国人就更加不懂了。再比如阎连科的《受活》，2004年版权就卖出去了，到现在也没有翻译出来。为什么啊，人家不会翻，不懂你的方言。
Human translation: "Jia Pingwa's books contain a lot of Shaanxi dialect that we Mandarin-speakers don't understand, dialect that foreigners are even less likely to understand. Another example is Yan Lianke's Shouhuo [The Living/The Joy of Living]. The translation rights were sold in 2004, but the book has yet to appear in translation. The reason is that they can't translate it - they just don't understand the dialect."
Google translation: Jia Pingwa works, there are a lot of Shaanxi dialect, we say that people who do not speak Mandarin, foreigners even more comprehension. Yan Lianke another example of "subject to live" in 2004, the copyright had sold out, and it has not translated. Why ah, other people will not turn, do not understand your dialect.
Comments
To be fair, Google has gotten noticeably better for Chinese over the past few years. The big problems seem to be that their parallel text corpus is just nowhere near the size of corpi for European languages, and that since it's statistically based rather than rules-based, the system doesn't actually "know" anything about Chinese grammar or sentence-parsing. A fair amount of the corpus seems to come from UN documents and news headlines, which is why the system will generally render any phonetic character that happens to be the start of a country name (e.g. 巴) as the name of that country. This happened in one of my favorite examples of this, which I wrote about a couple of years ago:
今天晚上在回家的路上，看到一位遛狗的老太婆边抖着缠在她腿上的狗链，边对身旁的小京巴儿骂骂咧咧：“你他妈好好儿走啊！你他妈不好好儿走我再也不带你出去！”
Me: On my way home this evening, I saw an old lady out walking her dog, which was winding its leash around her legs, and yelling at the little Peke at the same time: “Dammit, walk right! If you don’t walk right, I’m not going to take you out anymore!”
Google: This evening on his way home, walk the dog to see a side Douzhao the old lady in her legs wrapped around the dog chain, Pin right side of the small Palestinian children Mamaleilei Beijing : “you damn thoroughly go ah you damn not properly take I do not take you away!”
Brendan, March 12, 2010, 4:06a.m.
Just so I don't forget this anecdote: On the ten-hour train ride from Wuhan back to Beijing two days ago, I passed a gas station with the following sign:
加油站 onlinetranslation 加油站 onlinetranslation
Canaan Morse, March 12, 2010, 5:08a.m.
Several things about Google Translate.
I didn't know about until 6 months ago. For someone who often depends on translation for a living, that is pathetic.
Now that I do, I must say that like those examples cited above, Google Translate can turn out some real "classics." They often read suspiciously like the faintly poetic copy one sees on those neat Japanese tea-shirts.
Experts can argue about this for eons, but I see translation as largely breaking down into a few principal types: Commercial, technical and creative writing (advertising, literature, etc.)
For commercial and technical texts, Google Translate is very useful. Not to use it would be stupid. I now run all my commercial texts through Google, and it has literally reduced my translation time by 50 percent. I just translate on my own with the occasional glance at the Google version, and almost every paragraph provides 1-2 interesting options or just downright more accurate translations that I am then free to incorporate in my "version."
I suspect that Google Translate is one of the tools out there that is pushing translation fees ever flatter, because referring to it -- not using it verbatim! -- allows non-native speakers of the target language, in particular, to "generate" markedly better copy.
Let's face it: there's a fair amount of pretty mundane wordage even in the literature we translate. For those phrases and sentences, Google Translate offers a lightning fast reference which can be somewhat useful.
The key concept is: as a "reference," not a ready-made translation. Judicious judgment IS required.
I suspect many of us literary translator types find machine translation hopeless because we focus on the screwed up sentence structure and hilarious misinterpretations of slang.
But in my opinion, it is not "hopeless," and it will be useful -- even for translating some types of creative writing between Chinese and English -- in well under a decade.
One of the reasons I say this is because of the radical transformation of contemporary Chinese. In terms of new vocabulary, idioms and even sentence structure, 21st century Chinese has become much, much more like European languages than I would had thought possible in my life. I've translated a dozen or so modern Chinese authors (short pieces and one full-length novel) since the early 2000s, and I find their language, style and even worldview increasingly urban, modern and sometimes downright "Westernized." (con't below)
Bruce, March 12, 2010, 10:08a.m.
...That being the case, I would expect that something like Google Translate, which seems rather Euro-centric and handles translation between European languages fairly well, will soon be much better equipped to handle modern Chinese than it is right now. I predict that -- like it or not -- this will allow more at least partly bilingual people (both native Chinese speakers and native English speakers) to turn out their own computer-assisted translations of creative writing.
Then the emphasis will be on their ability to edit, repackage and spice up the writing, rather than doing the grunt work entailed in "just" translating it.
Chinese Books, English Reviews
Bruce, March 12, 2010, 10:08a.m.
Google translate will soon spell the end of translation memory such as Trados, which is great because it's a horrible, cumbersome piece of software that is artificially made difficult so that translators can claim Trados expertise.
I don't think literary translators will ever have to worry about loosing their jobs, but then again, who really goes into literary translation for the money?
ff, March 12, 2010, 10:20a.m.
@ff
I don't think literary translators will ever have to worry about loosing [sic] their jobs
Actually, English-to-Chinese literary translators are more vulnerable than Chinese-to-English.
Why do I say so? Two reasons:
The marketplace demand for all kinds of English-to-Chinese translation is massive, on the increase and largely unmet, creating a need that can be satisfied to some extent with machine-translated copy that is then "edited";
Chinese readers of many types of writing, particularly news but magazine-length essays too, are quite used to very "mechanical" translations of text into Chinese. The quality of translated copy in Taiwan is much higher than in the mainland, for instance. Due to years of military occupation by the Japanese (1895-1945) and later "cultural" occupation by the Americans (some would say!), knowledge of foreign languages and cultures is much greater and translators and readers simply would not stand for much of the rubbish that appears daily in mainland web sites and international digests parading as "translated copy."
As a result of reading so much poorly translated Chinese text, and also because young people in China are fascinated with English and more than willing to use "trendy" Chinese that sounds like it was invented by a speaker of Chinglish, there is now a huge amount of translationese appearing in all sorts of Chinese texts.
All of these factors make "mechanical" English-to-Chinese translation more or less acceptable in the PRC.
Chinese Books, English Reviews
Bruce , March 13, 2010, 12a.m.
@ff Actually, I don't think CAT software will be threatened by Google translate, because they serve different purposes. CAT software is used for building up translation memory for specific clients so that proprietary technical terms can be consistently used, whereas Google translate is basically a gigantic TM available to the public. I think companies will still prefer translation agencies managing their TMs privately rather than turning them over to Google.
@Bruce That is an interesting technique, I will have to try it. I think the key to making it work is that you translated first on your own, then referred to the machine translation to double check. I have been doing this when I find translations of Chinese laws and regulations online that are clearly done by non-native speakers.
In my experience, skilled non-native speakers can do a very good job with a good TM on technical and legal documents, and I am willing to bet that most agencies experienced at this have much better TMs than Google. These types of documents just need a native English speaker to go in and edit.
The situation is hopeless, though, for commercial (cultural) and literary translation, as these require a native English speaker starting from scratch.
But I am wondering if we look at the lower end of the spectrum, such as the more trashy novels that come out in China, like the emperor and concubine romances, if a good TM + Chinese translator + English editor might not be able to churn out a large volume of work in not much time.
Jeff, March 13, 2010, 2:12a.m.
At That Time, Only the Dead Person as Dead
by Google Translate
Sparrow-like, moth-like, ant-like.
Like a dead person is,
now dead dogs.
Dog in the world than the moths,
the game is much more valuable.
wayne, March 13, 2010, 11:41a.m.
Damn: paper-republic.org doesn't really like line breaks, does it?
wayne, March 13, 2010, 11:42a.m.
I never know how to format line breaks on Paper Republic, either (which is one reason I haven't posted more poetry - it always comes out wonky.) I'll ask Eric about adding info on how to format line breaks on posts and comments.
Cindy Carter, March 14, 2010, 4:42a.m.
Friends don't let friends make single line breaks.
But if friend did let friends make single line breaks, they would tell them do it by leaving two blank spaces at the end the line.
I have taken the liberty of fixin' the poem above, and will add this to the comment instructions…
Eric Abrahamsen, March 14, 2010, 3:25p.m.
Like a dead dog is,
my comments now more valuable.
wayne, March 15, 2010, 9:31a.m.