Lost in Machine Translation

While I’ve been traveling over the last week or so, loads of people sent me a link to this wonderful image of a sign in China reading “Translate Server Error” which has been written up all over the place. Thanks everyone!

Billboard saying

It’s pretty easy to imagine the chain of events to led to this revealing error. The sign is describing a restaurant (the Chinese text, 餐厅, means “dining hall”). In the process of making the sign, the producers tried to translate Chinese text into English with a machine translation system. The translation software did not work and produced the error message, “Translation Server Error.” Unfortunately, because the software’s user didn’t know English, they thought that the error message was the translation and the error text went onto the sign.

This class of error is extremely widespread. When users employ machine translations systems, it’s because they want to communicate to people with whom they do not have a language in common. What that means is that the users of these systems are often in no position to understand the output (or input, depending on which way the translation is going) of such systems and have to trust the translation technology and its designers to get things right.

Here’s another one of my favorite examples that shows a Chinese menu selling stir-fried Wikipedia.

Billboard saying

It’s not entirely clear how this error came about but it seems likely that someone did a search for the Chinese word for a type of edible fungus and its translation into English. The most relevant and accurate page very well might have been an article on the fungus on Wikipedia. Unfamiliar with Wikipedia, the user then confused the name of the article with the name of the website. There have been several distinct citings of “wikipedia” on Chinese menus.

There are a few errors revealed in these examples. Of course, there are errors in the use of language and the broken translation server itself. Machine translations tools are powerful intermediaries that determine (often with very little accountability) the content of one’s messages. The authors of the translation software might design their tool to avoid certain terminology and word choices over others or to silently censor certain messages. When the software is generating reasonable sounding translations, the authors and readers of machine translated texts are usually unaware of the ways in which messages are being changed. By revealing the presence of a translation system or process, this power is hinted at.

Of course, one might be able to recognize a machine translation system simply by the roughness and nature of a translation. In this particular case, the server itself came explicitly into view; it was mentioned by name! In that sense, the most serious failure was not that the translation server worked or that Wikipedia was used incorrectly, but rather that each system failed to communicate the basic fact that there was an error in the first place.

23 Replies to “Lost in Machine Translation”

  1. This is just a special case of the all too widespread error made by computer programmers of providing meaningless error messages. The average user doesn’t speak SQL so there’s no point in displaying the raw database error message to him. And if you’re providing a program to translate Chinese into English (presumably because the user doesn’t speak English), which language do you think you ought to display the error message in?

  2. Dan – how does that not ever occur to anyone? Displaying the error message in the original languge shouldn’t be that hard to do and it would help so much.

  3. Does it not occur to anyone to ask a native English speaker to proofread your copy before printing?  How hard is it to scan it to someone who speaks English?

    1. Jack hi,
      As a fairly technical, well read, cultured Englishman with a very good general knowledge and a wide life experience I have tried to get proof reading work with the Chinese company’s I come in contact with, and there are many .
      Its an impossible task, they believe that the translation is ” Good Enough ” and nothing will persuade them other wise.
      Even when they lose a sale due to the documentation quality.

      1. Stu – Not to nit-pick, but I count at least 8 grammatical errors in your comment. Perhaps you just weren’t cut out for the proofreading job.

  4. A local Chinese restaurant where I live has a humorous typographical and spellcheck error on its menu resulting in a dish called “beef in empirical sauce”.  What did it taste like you might ask?  Well, you’ll just have to try it and find out.

    1. This may have a reasonable explanation: the chef does not master the recipe at all, and the outcome is different every time s/he tries to prepare the sauce.

  5. The answer is not to omit the SQL or whatever tech-speak is relevant. It’s valuable information the developers can use to help support the user who runs into problems, even if all the user can do is copy and paste it into email sent to the developers. What’s important is providing reasonable natural language errors IN ADDITION to the tech-speak.

    The problem with providing error messages in the “original” language is the volume of error messages which must be translated. A boggling variety of failures can occur while any non-trivial computer program is running. Often, there are so many possibilities that errors aren’t reported well internally within the program, much less to any human, much less to any human other than the programmer, much less in seventeen languages. Oftentimes, nobody even knows all the errors which might occur, or if it’s possible to know what they all are, it’s impractical to create circumstances under which they all occur so that messages for them can be designed well in one natural language.

    Software is harder than you think and easy to complain about.

    One thing people could do better easily is make sure error messages are clearly designated in the “original” language. Then at least the user would know THAT the operation failed even if the nature of the failure is incomprehensible.

  6. I am from China, I am so sorry to see that.

    Why did they use a online dictionary? and they even didn’t know they’d lost their internet connection? Stupid guys.

  7. I’m afraid “Chinese sucks” ignorance of Evolution and Natural Selection appears to match their ignorance of Chinese science and technological innovations over the millenia. :-(

    You really should go and read Joseph Needhams “Science and Civilisation in China” and learn how the Chinese invented many things before the West had considered them possible.

    The Wikipedia page on the history of science and technology in China gives a taste of this with “matches, dry docks, the double-action piston pump, cast iron, the iron plough, the horse collar, the multi-tube seed drill, the wheelbarrow, the suspension bridge, the parachute, natural gas as fuel, the raised-relief map, the propeller, the sluice gate, and the pound lock”.

  8. Only a thousand years? Why thank you for your flattery, but it was more like ten, a hundred thousand. Western languages, on the other hand, took the whole thousand years and is a mishmash of about fifty languages; at least Chinese was standardized in the Qin dynasty.

  9. why are you proud of something you don’t even contribute to? So your granpa was smart. Yay. Are YOU smart? Didn’t think so..

  10. “And if you’re providing a program to translate Chinese into English (presumably because the user doesn’t speak English), which language do you think you ought to display the error message in?”

    Using a machine to translate from your native language to one you don’t understand is a bad idea anyway – at least without checking it. Using it to translate into your native language, on the other hand, generally works well. So it’s not unreasonable for the translation software to display errors in the destination language.

  11. The problem here is that the xlation software probably has a “source language” and “target language”.  It also has a translation backend server to which it sends the text.  If it can’t get at the translation server, the only thing it can do is output a standard error message.  I guess it would be best if the client errors were themselves localized – so that “translate server error” was printed in the “source language” instead of English, but that’s pretty expensive and a bit too much to expect for what should be a last-chance error handler for a rare situation.

  12. He who builds translation systems, writes error messages for the intended users. Internet gives access to anyone with access… and no one except the world speaks all languages.

  13. this is hilarious. in belize i saw a few mistranslated signs. (to english)which is interesting, english is the official language there and quite widespread

  14. I agree I’ve seen this translation dotted around a fair bit, maybe there’s a very inaccurate auto translation tool that’s really popular in China. Either way, I think these companies should try using <a rel=”follow” href=”http://www.rosettatranslation.com”>translation agencies</a>. Real language professionals wouldn’t make such careless mistakes and they can apply the human factor that machines can’t. I feel for the restaurants that now have to re-print all their menus because the translations are completely useless but it goes to show that language accuracy shouldn’t be underestimated really.

  15. Two points:

    – The simplest check that I can think of for translation software is to have it translate the results back into your native language. You might or might not pick up on the empirical sauce, but you’d probably notice ‘Translation Server Error’, at least.

    – Websites that display SQL or database-related error messages are begging to be hacked. They’re holding up a sign that says “Anything you enter here will be processed by our database without being sanitised by the application first! Have fun!”

Leave a Reply

Your email address will not be published. Required fields are marked *