Language Log

« previous post | next post »

Paul Bickart has called my attention to the latest issue (June 23, 2010) of New Scientist which has, in its "Feedback" column, a frustrated discussion of what seems to be a strange translation from Chinese:

==========

THE oddest things can send Feedback on quests of epic work-avoidance. Take Julian Parmegiani's attempt to buy a professional video light on eBay, which led him to the LED 5001 at bit.ly/daLLlW and the seller's "How to use" instructions: "Install the battery into the battery jar, when heard a weak 'pyridaben carbazole' sound the installation is completed." Julian would like to know what a weak pyridaben carbazole sound is.

Surely, we thought naively, in this age of instant access to all the world's true facts (and even more of the other kind) it would take only a moment to find the answer. Or any answer. All, arguably, part of Feedback's job.

The search results we found to "pyridaben carbazole sound" directed us, with one exception, to the same instructions Julian found. But we did discover that carbazole is also known as dibenzopyrrole and diphenylenimine and several other things. Carbazoles in general can be synthesised (though this is but a WikiFact) by reacting naphthols and aryl hydrazines using sodium bisulfite. Does this reaction make a sound? Is it squelchy? Crackly?

We found plenty of other references to "the carbazole sound". Examples include: the instructions for what appears to be a clandestine listening device from China; a description of an ailing knee-joint on a Chinese medicine website; and – perhaps importantly, for here's the exception mentioned above – a strange, oddly translated blog which contains this sentence: "He heard the guard outside of a person who asked, and then latch issued pyridaben carbazole sound, the door opened."

So we have a sort of triangulation on what it might sound like (if we use our imaginations). But are we, as the anti-intellectual but neurologically correct phrase goes, "over-thinking it"? Do the characters that say "carbazole" in a Chinese language also stand for something like "graunch" or "bzzzt"? Could it merely be a typo in Chinese?

We even resorted to leaving the computer and asking, face-to-face, a video and lighting technician. She had never heard of pyridaben or carbazole. Surely someone can put us out of our misery? Perhaps, given the internet's failure to provide a satisfactory answer, a handwritten postcard would be the way to do so.

==========

How to solve this misery inducing conundrum?  I began by looking up the Chinese for pyridaben, which is written 噠蟎靈  in characters, and carbazole, which is written 咔唑.  Pronounced in Modern Standard Mandarin (MSM), these would respectively be dāmǎnlíng and kǎzuò.  I actually suspect that these chemical compound terms would sound better in Cantonese than in MSM, as do so many Western terms that have been transcribed into Chinese, such as McDonalds, taxi, park, t-shirt, and so on and so forth.  This is particularly the case, as here, when we encounter a high percentage of characters with mouth radicals, of which Cantonese is inordinately fond.  However, for the convenience of those who know some Mandarin but little or no Cantonese, I will keep dāmǎnlíng and kǎzuò in MSM, though I shall return to the matter of Cantonese briefly again below.

Next, I had a hunch that 噠 was meant to indicate some such onomatopoeic expression as "snap, crackle" and 咔唑 was another onomatopoeic expression for "click," rather than carbazole (the chemical compound name which it generally transcribes).  I made the further assumption that the translator (most likely a machine acting without much human supervision) dragged along the 蟎靈 of 噠蟎靈, when all that was wanted in the original Chinese draft was the 噠 sound.  Incidentally, since Chinese onomatopoeia has a predilection for repeated or multiple syllables, I suspected that 噠 would more naturally and more frequently occur as 噠噠 in actual use.  No problem with 咔唑 because it is already a bisyllable.

Now, in sleuthing the mysteries of Chinglish, the next step is to find verifiable, unmistakable instances of the suspected misuse or error.  That is to say, we want to locate a Chinese text whose English translation contains the particular Chinglish usage we are seeking to explicate.  Here I enlisted the aid of Jonathan Smith, who is a master of Chinese Web searches.  In the following paragraphs, I shall summarize Jonathan's findings.

On this site, called Biz News, there is a Chinese news story (itself bizarre and error-filled) followed by what appears to be largely a machine translation into English.  The story contains these sentences (note there are problems with the Chinese to begin with) and translation:

====

Jìnrù gōngsī, zài zhǐdǎo xīnwén gōngzuò rényuán gēnjù yǐ chíxù nèi, nán shǒuxiān fāxiàn le yīgè xiázhǎi de guòdào, yīgè xiǎo yuànzi liǎng liàng huòchē “chē”, yuànnèi chuánbō dādā kǎzuò de jīqì shēng. Láodòngzhě de huòwù zhuāng xiāng shǔ yī dìbǎn shàng, yìn shàng kuāng de “yánsè Xī'ān qí gōng mào yǒuxiàn gōngsī” xiāoxi.

进入公司，在指导新闻工作人员根据已持续内，南首先发现了一个狭窄的过道，一个小院子两辆货车“车”，院内传播哒哒咔唑的机器声。劳动者的货物装箱数一地板上，印上框的“颜色西安奇工贸有限公司”消息。

Into the company, the press in accordance with guidance staff has been going inside, go south first and found a narrow aisle, a little yard two parked van “truck”, nosocomial spread pyridaben carbazole carbazole pyridaben the machine noise.

====

In other words, the onomatopoetic combination of the Chinese dādā kǎzuò 哒哒咔唑 has turned into the chemical names here.  We have a clear case of Chinese onomatopoeia shifting to chemical compound in English.

Thus, we have already achieved the main goal of our Chinglish analysis, which is to prove our hypothesis about how the mistranslation arose.  The next step is more constabulary and penal, in that we wish to identify the particular translation program that was responsible for the error (e.g., substituting a chemical compound name for an onomatopoeic expression), as we did with the notorious case of GAN1/4 several years ago.  See also Ben Zimmer's post here.

However, it is not always possible to nab the culprit who is responsible for originating a specific Chinglishism, since there are so many translation software programs in circulation, and not all of them are freely available online.  Furthermore, often it is a human being who launches a specific translation error, which is then picked up by others who know little English, but are desperately aiming to please their boss by wrenching the Chinese text they've been given by him / her into what they hope resembles that foreign tongue.  Once their Chinglish translation is posted on the Internet, it is picked up by other clueless translators, and soon — like a virus — "pyridaben carbazole" becomes standard for what was intended to be "crackle, click," etc.

Nonetheless, let us proceed to examine the ways various online engines treat this particular line from the bizarre Biz News article:

Yuànnèi chuánbō dādā kǎzuò de jīqì shēng <院内传播哒哒咔唑的机器声> ("from within the yard could be heard the tapping and clacking of machinery" [giving the correct translation so readers will know what it should be]).

Google: "Hospital-borne da da sound machine carbazole"

Yahoo: "In the courtyard disseminates clip clop clip clop the carbazole machine sound"

Sogou:  (same as Google)

Iciba: "Close to the spread of 机器声 唑 哒哒 there"

Netat: "Yard inner propagation Da Da carbazole machine sound"

Dict.cn: "Nosocomial transmission of JiQiSheng DaDa clicks."

We encounter lots of "carbazole" from 咔唑 in these machine translations.  This is understandable, since 咔唑 is well established as the transcription of the chemical compound carbazole, though I have found ten occurrences of kǎzuò shēng 咔唑聲 ("a kǎzuò [i.e., clicking] sound"; this is otherwise almost invariably machine rendered as "carbazole sound").  The last example even provides "nosocomial" for yuànnèi 院内 ("inside the yard" [should not be translated as "inside the hospital"]) as above.  Still, none of the online translation programs we have examined contains all of the errors present in our colossal specimen:  Yuànnèi chuánbō dādā kǎzuò de jīqì shēng <院内传播哒哒咔唑的机器声> ("nosocomial spread pyridaben carbazole carbazole pyridaben the machine noise").

This is better–a company with both English and Chinese versions of "installation instructions" (for its car lights, etc.) offers the following pair:

7. Ānzhuāng liánjiē qì shí yǒu kǎdā shēng, qǐng jiǎnchá liánjiē qì huò jiēxiàn duān jiēchù shìfǒu liánghǎo. 安装连接器时有咔哒声，请检查连接器或接线端接触是否良好。

7. Installed connector pyridaben carbazole voice from time to time, please check the connector or terminal is a good contact.

(Eng: http://www.huashenhid.com/en/newsdetail.aspx?news_id=60;
Chn: http://www.huashenhid.com/newsdetail.aspx?news_id=60 )

Here we have what was probably the ur-form from which dāmǎnlíng 噠蟎靈 and kǎzuò 咔唑 ultimately derive, namely kǎdā 咔哒 ("crackle," etc.).  This disyllable occurs tens of thousands of times on the Web, virtually always indicating a sound.  Sometimes it is reduplicated for emphasis:  kǎdā kǎdā 咔哒咔哒.  At other times it is is followed by the word shēng 聲 ("sound") so that there is no mistaking what is meant:  kǎdā shēng 咔哒聲 ("a crackling sound").   A few programs (Google 咔哒 = "a click") can do this disyllable as a sound; others can't (Yahoo 咔哒 –> "carbazole clip clop," which does have a nice ring to it!).

We haven't found an online engine that produces "pyridaben carbazole" directly from kǎdā 咔哒; perhaps the source for that is one of the popular pieces of translation software.  However, the gross evolution of the mistranslation under investigation must be 咔哒 –> "pyridaben carbazole."  But it is puzzling that kǎdā 咔哒 should go to "pyridaben carbazole" and not to "carbazole pyridaben."  In the transmission from kǎdā 咔哒 to "pyridaben carbazole," an inversion to dākǎ 哒咔 may have occurred or, more likely, the inversion occurred after the amplification of kǎdā 咔哒 to kǎzuò 咔唑 and dāmǎnlíng 噠蟎靈.  In any event, as in the Biz News instance, pyridaben and carbazole are themselves liable to inversion in the same sentence.

Finally, it seems that dā 哒 would make a more evocative sound-symbolic syllable with its final -t as historically and in Cantonese, though it is managing to hang around in Mandarin.  The rat-tat-tat of machine-gun noise in Cantonese would be daat6-daat6-daat6 哒哒哒, whereas in Mandarin it is dā-dā-dā (rat-tat-tat Mandarinized would be ra-ta-ta, with all final stops elided).  Similarly, compare Cantonese kaat1/kaa1-zo6 咔唑 ("crackle") as sound-symbolism with its Mandarin equivalent kǎzuò.  (N.B.:  I have consistently transcribed 咔 as kǎ, since that is how it is listed in many dictionaries and how it is given by the online references I have consulted, but I think that it would usually be pronounced as kā in actual speech when used for sound symbolism.  I started out transcribing 咔 as kā, but then changed all the kā to kǎ, or at least I tried to do so; forgive me if I may have missed a few.)

Somebody please put the folks over at New Scientist out of their misery by letting them know that, when it comes to Chinglish, there's always an explanation, and it usually has to do with one or more of the following:  a. poor translation software, b. semantic interference (translating the surface signification of the characters when they are meant merely to render sounds, c. sheer incompetence.  In this case, it was probably a combination of all three.

[Thanks to Bob Bauer for the Cantonese pronunciations.]

June 30, 2010 @ 8:03 pm · Filed by Victor Mair under Lost in translation

Permalink