Mind your manners with the empress
« previous post | next post »
On a Chinese site selling paint, pharmaceutical products, paper industry additives and so forth, we find the colorful Chinglish phrase "fuck the empress." Although the "f" word occurs fairly frequently on this site, the exact citation for the sensational "fuck the empress" is the 14th bulleted item about three quarters of the way down on this page: "Spray to inunction the partition slightly treat to fuck the empress to with the Beat to whet."
The entire website is full of such seemingly whimsical, incomprehensible English (comparable to what results when one turns Google Translate loose on a Tang poem). Since there is a corresponding Chinese original, it is relatively easy to figure out the source of each Chinglishism. "Fuck the empress," however, is so easy and obvious that it is not necessary to consult the Chinese equivalent (though I did just to be sure), since its obvious derivation is as follows:
What is written in Chinese is 干后. In Modern Standard Mandarin (MSM), the first character could be pronounced either gān or gàn. The problem arises because 干 is a simplified character that collapses three traditional characters (干, 乾, and 幹)
In the traditional character system, 干 (MSM gān) is a cyclical or calendrical symbol, 乾 (MSM gān) means "dry," and 幹 (MSM gàn) means "to do," but vulgarly also "to fuck." In the simplified character system, 干 means all of these things, and much else beside. I spelled all of this out in great detail in "The Etiology and Elaboration of a Flagrant Mistranslation" and related posts.
As for where "empress" comes from, it is also due to the collapsing of different characters from the traditional system into a single character in the simplified system. Where formerly we had 后 ("empress") and 後 ("after") — both pronounced hòu, now we have just 后, meaning both "empress" and "after."
If the paint website cited above were written in traditional characters, it would have 乾後, not 干后 as it does now in simplified characters, and there would have been no mistake, not even by a dimwitted software program.
This eyebrow-raising locution can be found on other Chinese sites involving spray cans, paints, and so forth. One of the more spectacular occurrences of "fuck the empress," however, is on a label that explains how to care for a wig made in China.
Quoting just a couple of sections from this unhelpful label:
DRYING: The rap drops the water bead, nature cool fuck.
(VHM: A human being must have intervened to change the first liánggān 涼干 [i.e., 涼乾, "cool dry," "dry in the air," "dry without the application of heat"] to just "Drying," since the identical characers are translated as "cool fuck" later in the same line.)
STYLING: It is cool to fuck the empress, and use the steel needle the comb to comb the original hair style.
(VHM: "It is cool to fuck the empress" comes from liánggān hòu 涼干后 [i.e., 涼乾後 ["after cool drying," i.e., after drying without applying heat].)
Conclusion: if China wishes to avoid such embarrassing mistranslations in the future, it simply needs to spend some of its vast foreign reserves to hire experienced human translators. They wouldn't cost nearly as much, say, as building a rail line from Golmud (in Qinghai) to Lhasa (in Tibet) or from Kashgar (in Xinjiang) to Gwadar (a seaport in Pakistan). Considering the poor quality of available software, plus the nature of the simplified character writing system, the machines are bound to come up with such howlers again and again.
[A tip of the hat to Rory Francisco.]
Declan said,
July 14, 2010 @ 5:42 am
And the first-mentioned 'fuck the empress' has been corrected already :o)
Francisco Pisa said,
July 14, 2010 @ 5:44 am
Shame on all automatic translation software; shame, endless shame!
arthur waldron said,
July 14, 2010 @ 5:52 am
This really is beyond belief! I would never allow a bit of written Chinese to emerge over my name until some real literate educated native speakers had checked it. This also shows a problem with collapsing characters having different meanings into one simplified graph and then relying on the fact that Chinese already speak Chinese to make sense of them. How "fuck" became so common in Chinglish puzzles me a bit. Does it go back to that little dictionary of dirty words that Judy Shapiro produced about twenty years ago?
Scriptor Ignotior said,
July 14, 2010 @ 6:53 am
I have seen fuck turn up for dry in a list of ingredients on a pack of noodles. Yes, it is remarkable how insouciant Chinese translators typically are in these matters, and (in my experience) how blithely they refuse offers to help them avoid such disasters.
A tradition of reposing blind trust in "standard" dictionaries? Whether or not such trust was ever justified in China, it is not now – given the pace of change in the language, and the chaotic state of commercial relations between China and the West. So it seems to me.
John Cowan said,
July 14, 2010 @ 8:14 am
A tradition of not giving a dry fuck what the barbarians think, perhaps.
But whether individual businessmen will pay for expensive human translation has nothing to do with whether the government will pay for expensive railroads.
Mark P said,
July 14, 2010 @ 8:40 am
I don't think they need professional translators to avoid the "fuck" problem. Surely every schoolboy learning English in China has figured that one out.
Ray Girvan said,
July 14, 2010 @ 8:43 am
From the description, "Banana water" appears to be isoamyl acetate.
Ray Girvan said,
July 14, 2010 @ 9:09 am
PS The odd thing is that they could have got a better translation using Google Translate: the output for the same page has peculiarities, but nothing as bizarre as their own translation.
Victor Mair said,
July 14, 2010 @ 9:43 am
Decian, check again.
vanya said,
July 14, 2010 @ 9:57 am
Obviously it's simply time to toss out the ridiculous simplified character system, just another sad legacy of Maoism. Maybe the Mainland can take a lesson from Hong Kong and Taiwan.
stephen said,
July 14, 2010 @ 10:06 am
If China still had empresses, this problem would never have arisen…would it?
That first example, "Spray to inunction the partition…" etc. almost makes a little more sense if you read each word in reverse order.
I wonder about how the characters are presented. On my screen, I see little boxes with things like
5E
72
Is that what everybody else sees on their screens? Where shall I go to learn more about that kind of transcription? Thanks.
Mr. Shiny & New said,
July 14, 2010 @ 10:26 am
@stephen: That sounds like you don't have the Chinese fonts installed. On Windows you can get Chinese fonts if you go into the control panel, regional settings, and install support for Chinese.
stephen said,
July 14, 2010 @ 10:30 am
Thanks.
But I don't read Chinese at all, anyway.
Daniel von Brighoff said,
July 14, 2010 @ 11:19 am
As Ray says, you don't necessarily need human translators, you just need context-sensitive translation software. After all, which meaning of 后 is going to be more common in a corpus of contemporary Chinese? I would imagine "after" by at least an order of magnitude.
(Of course, you do need skilled human translators to build the corpus in the first place. Could there be a GIGO problem of feeding in shoddy English "equivalents" that the software keeps perpetuating?)
Mary Kuhner said,
July 14, 2010 @ 1:07 pm
There's no English context in which "fuck" is not obscene, so simply filtering out that translation from your dictionary would avoid some menu-related as well as paint-related embarrassments. Of course it's not a solution, but surely it would help!
Or–are those Chinese/English dictionaries fulfilling an essential role in translating porn ads? Maybe "fuck" does need to be in there, and "dry fried" would be equally laughable in that context.
Michael W said,
July 14, 2010 @ 3:33 pm
Earlier appearances of 'fuck' on that page are just as interesting:
This part sounds like a warning label for a bizarre sex toy:
Usage hour announcements
(1), the PU is slow to increase to not too many in the water, general for 5% or so, if the exceed deal can result in slow fuck, influence task time.
(2), please adjust according to the season to use the Banana water of our company.
(3), the task empress should cover the screw down, in order to prevent the bucket to vaporize.
Nathan said,
July 14, 2010 @ 7:03 pm
There's little incentive to fix Chinglish; no matter how much we laugh, we keep buying the products.
Scriptor Ignotior said,
July 14, 2010 @ 8:38 pm
There's no English context in which "fuck" is not obscene …
But see windfucker, of course.
Kaykuri said,
July 14, 2010 @ 8:51 pm
[note: please forgive the quick oversimplifications that follow. I speak Korean, not Chinese, and will make the connection below.]
As a former developer of machine translation (MT), my hunch is that what you are seeing here is a segmentation problem (among many others, obviously) in addition to the collapsing of characters. The first trick for parsing languages that do not clearly define the boundaries between words, e.g. with spaces, is to break the text up into its component words for analysis, and this is often done by finding matches in the translation dicitonaries. In the case of liáng/gān/hòu the machine has to decide whether it is looking at "liáng/gān/hòu" (one word), "liáng/gān + hòu", "liáng + gān/hòu", or "liáng + gān + hòu".
The dictionaries can become a problem over time for such languages due to a quantity over quality issue, for two reasons:
1. Because Chinese and related languages are written using characters, which incorporate their own semantic value, individual characters will be listed in dictionaries along with their meanings. But there is a legitimate question as to whether individual characters deserve to be considered "words" in and of themselves. In other words, are they used that way, i.e. singly?
2. There is an incentive to fill the dicitionary with this stuff because MT quality is hard to measure. So how can you prove to your boss or client that you have been productive? Well, you can show that you added 50,000 words to the dictionary in the past month.
The upshot can be a situation whereby one couterintuitively degrades MT quality by adding word fragments to the dictionary, which is the basis for segmentation.
More commonly, at least in Korean, words as they are actually used are comprised of at least 2 characters. Even though each character has semantic value, they are not often used as standalone words. What we have here is a 2-character word liánggān ("to air dry") that has been segmented into its component semantic parts, liáng ("cool") and gān ("dry"). The question then is whether these individual characters are used by themselves when talking about other types of "cool" or methods of "drying" (not to mention "to fuck"). What you usually find is that gān is not used by itself for "to dry" or "to fuck", but rather you find other 2-characters words for various methods of drying or fucking that also include the character gān. If such is the case, then you are better served by removing gān=dry and/or gān=fuck from your MT dictionary, because they will only muck up your segmentation. They may belong in a language dictionary for humans, but not really in a translation dictionary for dumb old machines.
The problem with hòu "after/empress" is related, except that the segmentation is correct, since "after" is a single character preposition. So one has to ask what words are actually used when referring to an empress: hòu by itself or as a component part of another word.
If it isn't used as a standalone word, then take it out of the dictionary. If it is, then you have a legitimate problem and that is one of the many fascinating reasons why MT is really hard.
It is a bit of a mystery why "drying" was translated correctly in the one instance, but I suspect that is because of colon that follows it in that case. Likely the system broke the text at the colon, treating each piece as a separate sentence (this is common). In that case the sentence fragment liánggān is easier to analyze, while other prior segmentation decisions made in the longer string may have contributed to liánggān being split.
Side note: The Korean language is very different from Chinese, but was heavily influenced by it. Korean does not use tones, and although it retains a lot of sino-Korean vocabulary it has (mostly) abandoned Chinese characters in favor of a purely phonetic alphabet. Yet the orthography of Korean preserves the identity of those characters by grouping letters into syllables, thereby also preserving their semantic heritage. The result is sort of a super simplification, so the problem described here as the collapsing of characters is hugely magnified for Korean.
I have seen single Korean syllables/characters listed in MT dictionaries with as many as 14 different meanings. This is disambiguation on a collossal scale, and I had to beg the lexicographers to remove as many of these as possible on the grounds that many are not "words" in the sense of being used on their own (response: "but it's in the dictionary, right here!").
This is one reason that Korean-to-English MT continues to suck worse that Chinese or Japanese-to-English. Chinese has a multitude of characters to clarify things (esp. Traditional Chinese). Japanese utilizes 3 different character sets (and counting?), and the switch from one to the other gives the MT system a clue that something is up. Korean has dealt with this problem somewhat by adopting the use of spaces to a degree, though not as completely as English so segmentation is still required.
Ray Girvan said,
July 14, 2010 @ 11:41 pm
Interesting. A quick skin of references confirms the "windfucker" = kestrel. But I don't get why it's so definitely taken to be that, rather than an example of long s.
deadbeef said,
July 15, 2010 @ 1:23 am
@vanya: "sad legacy of Maoism": I'm sure that's exactly Singapore was going after when they adopted simplified characters.
This is a much longer rebuttal than your short comment deserves, but I think that your view of the history of simplified Chinese characters is seriously simplified and uninformed. Although the standardization of modern simplified Chinese characters is due to the Communists, people had been simplifying their characters long before Mao. Many of the simplified characters come from cursive or grass script forms that have been long in use. It's not a ridiculous idea; many traditional characters are really quite tedious to write in full and even people who were formally educated solely with traditional characters use many simplified forms to save time and effort.
China adopted simplified characters for a noble cause, to increase the literacy rate among the population, by decreasing the difficulty of writing. Whether or not simplified characters played a major or even helpful role, the literacy rate really did improve dramatically in China under Communist rule. If this had happened in an impoverished nation that happened not to go Communist, I bet you wouldn't be damning the idea so zealously.
That said, the merging of characters, which rightfully deserves blame for the dry/fuck issue, can probably be rightfully blamed on the Communists. The dry/fuck merge is arguably bad because they are pronounced differently and neither character is terribly complex in the first place. But you don't often have problems telling the difference between lead as in leader and lead as in lead paint, do you? And the empress/after issue isn't really so bad since they are pronounced the same; you can't see the characters when somebody is talking anyway. The problem is with the computer translation programs. You don't think other written languages need to be "fixed" to avoid bad machine translations, do you?
Despicable as you may find the Chinese Communists, they are not really to blame for the proliferation of the ridiculous translations. Whatever they do won't make you happy anyway; when they try to fix the problem on their own end, somebody laments that the disappearance of charming and humorous signs. The problem is what John Cowan and Nathan alluded to: Chinese businesses use cheap translation software, and it's economical for them.
Nathan Myers said,
July 15, 2010 @ 5:22 am
Given the magnitudes of the populations involved, I suppose we will come to equate "empress" with "after" and "fuck" with "dry" and think nothing of it, and wonder that our grandparents did. After all there are plenty of other alternatives for "fuck", and how useful is that word "empress" any more anyway?
bryan said,
July 15, 2010 @ 10:22 pm
涼干后 [i.e., 涼乾後]
涼 should be 晾.
晾 = to hang on a clothes-line.
涼 = cool, breezy.
涼乾後 should be 晾乾後, meaning "after being hung on a clothes-line", where as in simplified Chinese, it's totally confusing due to meshing together of characters with different meanings with the same characters, which originally had only ONE meaning.
bryan said,
July 15, 2010 @ 10:28 pm
涼乾後 should be 晾乾後, meaning "after being hung on a clothes-line"
I meant "After being hung-dry on a clothes-line" instead.
晾 = to be hung on a clothes-line.
乾 = dry, dries, dried.
晾乾 = to be hung dry on a clothes-line.
後 = after, afterwards
Bob Violence said,
July 16, 2010 @ 2:42 am
晾 is the same in simplified and traditional characters.
vanya said,
July 16, 2010 @ 9:15 am
I think that your view of the history of simplified Chinese characters is seriously simplified and uninformed.
You don't even know what my view is. You admit yourself that the confusion of "dry/fuck" is a legacy of the post 1949 reforms, i.e. "Maoist". I am well aware that people have been simplifying characters for centuries and that some traditional characters are ridiculously hard to write (the "wan" of Taiwan for example). The fact remains that the approach of the CPC towards simplification was driven primarily by ideology and not by logic or ease of use. Your assumption that "China adopted simplified characters for a noble cause, to increase the literacy rate among the population, by decreasing the difficulty of writing" is questionable. Arguably the real motivation was trying to break with the past and create a new Homo Siniticus who would happily follow the Party in all things. Many in the Party would have preferred to throw out characters after 1949 and use pinyin for that very reason. The argument that simplification has improved literacy is also not supported by any empirical evidence. Taiwan and Hong Kong have had no problem achieving literacy levels superior to the Mainland using traditional characters. If you've studied hanzi you know that there's really very little practical difference in the memorization required between simple and traditional – except that simplified characters as promulgated by the CPC are more illogical and easier to confuse.
J.W. Brewer said,
July 16, 2010 @ 3:11 pm
@deadbeef: What's the use of improving literacy in a country w/ no freedom of the press? And, following Vanya and perhaps considering the parallel to Ataturkism, I expect that the Communists might have viewed it as a feature rather than a bug that their newly literate subjects would have considerable difficulty reading texts published before their own rise to power (or texts published in the non-Communist parts of the Sinosphere).
Sampan said,
July 18, 2010 @ 1:18 am
What's the use of improving literacy in a country w/ no freedom of the press? It is to improve one's own life and living situation, and in a mass, being able to influence policies when more are educated.
Just compare China when Nixon visited, and newspapers are pasted on walls for the populace to read, and now, with most major newspapers (even in English) available, not to mention Internet access.
Being communist does not detract China from trying to improve literacy via script change but not language change, and to widen easier learning of the Mandarin dialect via hanyu pinyin for foreigners.
Freedom of the press is an anomaly.
If the press is owned by the government, then it is government-controlled.
If is is own by capitalists, then it is money-controlled.
Someone somewhere will still have to toe the line, whether in obedience to his government, his bosses or the masses.
Otherwise, pornographic photos and stories will be on the front pages of every newspaper in the world because that's what the masses really want, but the masses do not want to be seen to want these stories.
Freedom of the press is about telling stories truthfully, back by facts and empirical evidences.
However, in most societies, freedom of the press equate to publishing what you like, including telling not just lies, but blatant lies, as well as speculative negative stories about political figures they don't like or who are not from their political party affiliations.
And of what value or use is a newspaper when only those animals who are more equal than others can read but not the rest in the farmyard?
Zilch.
John said,
July 18, 2010 @ 9:40 pm
Oh my! Communism? Freedom of the press? Let's not be too silly here. To blame this translation phenomenon on any group or ideology in particular would be like… gee, I can't even think of an example as ridiculous as this.
bryan said,
August 7, 2010 @ 11:02 am
晾 is the same in simplified and traditional characters.
I meant someone actually used the wrong character by replacing 晾 with 涼.
晾 & 涼 ARE NOT the same form no matter what dictionaries you used, Traditional or Simplified Chinese.
晾 = To hang dry or have something like clothes, etc… hung-dried on a clothesline or some other similar means.
涼 = cool; breezy.
Clearly both words have different definitions and could not be interchanged.
涼乾 does not exist as a set phrase in the Chinese language. Only 晾乾 is a set phrase in Chinese.
Sources:
I'm Chinese
http://www.nciku.com/search/zh/detail/%E6%99%BE%E5%B9%B2/24751
bryan said,
September 12, 2010 @ 4:58 pm
I've found a different site, but it has some manner issues with the empress: Over 18, then enter.
July 9th, 2010
Don’t worry about it – it is cool!
Search for the above terms after entering http://adult.engrish.com/