The great linguist Chao Yuen-Ren once wrote an essay in Chinese using only words which (in Mandarin) would be transliterated as shih (using Wade-Giles; shi in pinyin). You can see the text in characters and two transliterations, read the translation (“A poet by the name of Shih Shih living in a stone den was fond of lions…”), and hear both Mandarin and Cantonese readings here (in Cantonese, of course, the words do not all have the same consonants and vowels). It’s really quite a tour de force. (Thanks for the link go to P. Kerim Friedman, who got it from Muninn.)

Addendum. See also Suzanne McCarthy’s discussion of Chao’s thoughts on reading Chinese, with copious quotes.


  1. That’ real fun, you could extend it by making a “Southern Mandarin” version [i. e. where ‘ten’ (shi) and ‘four’ (si) are homophones].
    Incidentally, is anyone still using Gwoyeu Romatzyh in the US (a good example, by the way, for your discussion on digraphs vs. untypable diacritics)?

  2. i. e. where ‘ten’ (shi) and ‘four’ (si) are homophones
    (Less the tone, of course; so make it “quasi-homophones”.)

  3. haha,real fun!

  4. In Taiwan, people buying ten of something in a noisy market would cross their fingers and make the “ten” sign just so they didn’t get four.
    There’s an incredible wealth of wordgames, puns, visual puns, etc. in Chinese, some of shich are woven very deeply into the culture as taboos or as elements of wedding ceremonies and funerals.
    At this point I always recommend Arthur Smith’s Proverbs and Common Sayings of the Chinese (1886, frequent reprints). I also have a book called “qi shi” (Strange Poems) which consists entirely of word-game poems.

  5. Lin Yutang’s Chinese-English Dictionary used Gwoyeu Romatzyh together with a four-corners index, making the book almost entirely unusable for 95%+ of the people who would ever want to use it.

  6. Right, while I know Lin’s dictionary by fame, I never used it, but not for that reason. The four-corners 四角 method invented by Wang Yunwu (the erudite director of Taiwan shangwu yinshuguan/Commercial Press, who was an autodidact himself) is still pretty popular, maybe more for indices and catalogues than dictionaries. Sometimes, there is an index by stroke number for those who don’t master the method (i.e. didn’t learn the little poem): that’s the case in the Zhonghua shuju index nominum for the 24 dynastic histories 廿四史. On the contrary, even modern reprints of Wang zhongmin’s catalogue of shanben 善本 have only a four-corners index.
    Now, the Ricci Institute’s Dictionnaire français de la langue chinoise uses both Gwoyeu Romatzyh and four-corners, but, fortunately, along with other more current transcription and character-checking methods.

  7. Initially, I just wanted to present this “activist”, but well-done GR site I just found (I shoulda check if LH mentioned it before). Among many exciting features, it reproduces Chao Yuen-Ren’s partial translation of Alice in Wonderland.

  8. Chao’s phonemicist slogan “pronounce same things the same, and different things different”, is one I’d love to enforce on reconstructors of archaic Chinese. Chao believed that “correct pronunciation” was not terribly important, as long as you mispronounced the same phoneme the same way every time.
    When reading Archaic Chinese, the most important things you need to know are which sounds are the same and which different. You really don’t need to know whether the p was voiced or not.

  9. Crazy!!!! It’s like La Disparition on crack!!!! 🙂

  10. The page states “With his essay, the eminent linguist tried to spoof the idea of converting Chinese character text into a phonetic (e.g. Latin alphabetic) system.” This is misleading. Y.R. Chao was making a point about classical Chinese (or “Literary Sinitic”), not about any alleged inadequacy of romanization to represent Mandarin or any of the other Chinese languages.
    He was also being characteristically playful.
    For an excellent discussion of the stone lion piece — along with a translation into the modern vernacular that is completely comprehensible in pinyin — see The Historical Evolution of Chinese Languages and Scripts, by Zhou Youguang. I hope to have this particular excerpt up on my Web site within a month or two.

  11. Mark, that is some great site you have. I see you live in Banqiao (officially Banciao), that makes me miss Taiwan.
    Of course, you’re right, Chao’s tour de force can be seen as an argument in the case for baihua (vs. wenyan).
    I have got Zhou Youguang 周有光’s Shijie wenzi fazhan shi 世界文字发展史 (A History of the Evolution of World Scriptures, Shanghai jiaoyu cbs, 1997), but I wasn’t aware of the one you mention (I’d kept in mind Zhou was a member of the pinyin commission, though).

  12. Specifically, Chao was spoofing the idea of using alphabetic script to write down the contemporary pronunciations of Classical Chinese texts. An alphabetic transcription of the archaic pronunciations of these text could be intelligible, though there are difficulties to the degree that some authors were pretty disengaged from the spoken language, and many of them thought in terms of already-obsolete pronunciations.

  13. Tangent: I’ve just received permission to put up about half of Y.R. Chao’s delightful translation of the Humpty Dumpty chapter from Through the Looking Glass. I plan on presenting it in various en face versions: Gwoyeu Romatzyh and Hanyu Pinyin transcription, G.R. and characters, G.R. and English, etc. Expect this in about a month.

  14. I attended a Chinese program hosted by Princeton University and discovered that beginning students of Chinese at Princeton are still taught Gwoyeu Romatzyh (in addition to pinyin). Professor Perry Link (The Tiananmen Papers) is fond of the system and thinks that it helps students’ pronunciation and, as far as I remember, he teaches the first-year class in alternating years.

  15. It’s very clever, but I’m dubious about its use as ‘alphabets won’t work for Chinese’ evidence – constructing artificial examples to prove your (any) point isn’t that difficult.
    I only mention this as it’s not difficult to find the above poem cited around the web, with no mention that it was written purely for the purpose of ridiculing romanisation. If someones wants to ridicule romanisation, fine, but do it with real world materials.

  16. Actually, it’s not “written purely for the purpose of ridiculing romanisation” — see Mark’s first comment above.

  17. Songlian, thank you for that information. I wonder if there is any evidence that GR helps students more than pinyin (I would tend to believe that tones graphically represented by diacritics rather than mute letters are easier to learn, but I never met anybody using GR, so I don’t have any arrested opinion on this).
    Do they teach zhuyin fuhao as well? After my short stay in Taiwan, I see that as more necessary than multiplying the romanizations (everyone I met with at least an elementary education mastered the zhuyin kana, but was unable to use properly one single existing romanization system; in the papers or personal messages, final exclamative particles are written directly in zhuyin fuhao, and the fact that many young people now exchange emails entirely written with that system prompted a new debate on alphabetization and the disappearance of hanzi).

  18. Roddy,
    As Mark said, a baihua translation of the text in pinyin would be “completely comprehensible“, and therefore not as funny. It is the proper characteristics of wenyan (monosyllabism, concision, elisions fading the context that would help to distinguish homophone characters) that make any romanization unsuitable to it.

  19. Jimmy:
    A few years ago someone did a study that compared tonal spelling and tone marks in teaching Mandarin pronunciation. It concluded, “The results clearly indicated that GR did not lead to significantly greater accuracy [than Hanyu Pinyin] in tonal production. Indeed, the use of GR reflected slightly lower rates of tonal production accuracy for native speakers of both American English and Japanese.”
    Here’s the article: “Tonal Spelling versus Diacritics for Teaching Pronunciation of Mandarin Chinese” (requires JSTOR access).

  20. “see Mark’s first comment”
    “As Mark said”
    Whoops . . .

  21. Thanks, Mark, that would confirm my feeling toward GR. Unfortunately, I cannot register to JSTOR to read the article.

  22. In fact, there is at least one case of an official and still current use of GR in the People’s Republic of China: in the absence of both Hanyu pinyin diacritics and characters, the province of Shānxī 山西 (1st tone on the first syllable) can not be distinguished from the province of Shănxī 陝西 (3rd tone). Hence the spelling Shaanxi for the latter, where the transcription of the first character actually respects the rules of GR (the double vowel indicates the third tone), something that, believe it or not, never occurred to me until today (while readwriting in my unconnected shūfáng 書房, I mean, shufarng). The complete GR transcription would be Shaanshi, which is somehow confusing for someone used to pinyin.
    As for zhuyin fuhao ㄓㄨㄧㄣㄈㄩㄏㄠ, another argument in their favor is the convenient way they are used in editions of classical texts annotated for students, aka duben 讀本 (the most prestigious collection of this kind is the « turquoise-cover » one published by Sanmin shuju), just like Japanese furigana, i.e. with the transcription of each character on its left side (the text being in columns). Pretty tiring when you don’t need it, but quite precious when you are stopped by a rare character or when a particular character is to be read differently than usual. I have never been taught the system and would never use it myself to transcribe Mandarin, but the passive knowledge I acquired by using these editions (which were strictly verboten by some of our teachers) has been useful in more than one occasion (note that, unlike my example above, tones are noted in zhuyin fuhao with diacritics similar to the pinyin ones, except for the first tone, which is not noted at all).
    The Guoyu ribao 國語日報 (National Language Daily), a daily paper for children (the great Hu Shi calligraphed the title), uses systematically zhuyin fuhao in parallel with the characters. And the youth happen to really use it. However, I’m not sure I would encourage teaching it to students who don’t already master a syllabary script.

  23. The Guoyu ribao 國語日報 (the official English title, which I ignored, is Mandarin Daily News). Oddly enough, most features don’t use zhuyin fuhao (I suspect it was too difficult to maintain the parallelism).
    Still, you can admire Hu Shi’s calligraphy on the upper left corner.

  24. Erratum:
    “with the transcription of each character on its right (not left) side.”
    Sorry (jamais deux sans trois ?).

  25. (One more correction: my fu should read ㄈ, not *ㄈㄩ, which is absurd since ㄩ is the -ü. I told you it’s a passive knowledge.)

  26. I really love the San Min Duben series. I must have six or eight of them. If you don’t learn Chinese by rote as a kid there are lots of rare characters whose pronunciations you never learn. So every time you see them, you look them up. But not often enough that you actually learn them (if you’re me).
    Some characters in proper names are so rare that apparently nobody knows how they are pronounces. A great XX-c scholar was named Liang Shu-ming or Sou-ming, and last I heard it was undecided. The first emperor of 3-kingdoms Wei was named Cao Bi or Cao Pi.
    In part this may have been an i=effeiciency move, to avoid the neuisance involved in tabooing the Emperor’s personal name throughout the kingdom.

  27. A rough version of the excerpt I mentioned earlier is up now: Humpty Dumpty. The Chinese characters and the Hanyu Pinyin are going to have to wait because I’m not in the mood (and don’t have the time) to type in 15 pages of hanzi anytime soon.

  28. Zizka,
    Alternative reading is a common phenomenon.
    In fact, that’s where some of the most obvious differences between Mainland putonghua and Taiwan guoyu lie. The first time I heard lèsè instead of lājī 垃圾 (garbage, trash), I thought it was some dialectal form, then I heard it repeatedly on « national » radio (usually when talking about lese hanjian 垃圾函件, aka « spam »), looked it up in local dictionaries, and learned that both readings were equally acceptable and « official » in so-called Mandarin. Same thing for bāokuò (Mainland) vs. bāoguā (Taiwan), « to include ».
    Some of those differences, I guess, may be due to regional variants, some others to the existence of a « literate » or « old » reading (noted as jiudu 舊讀, nothing to do with the reconstructed phonology of ancient Chinese). That is the case of the « cart » as a chess piece (ju and not che 車) and of, among many other possible examples, the character 單, normally pronounced dan1, but which should be read shan4 as a surname.
    The shu in Liang Shuming 梁漱溟 can indeed be read sou (that would be a less « ordinary » reading), though I never personally witnessed that. As for the pi in Cao Pi 曹丕, I was not aware of the alternate reading you mention (nor can I find it in my dictionaries, but that’s no proof, since this is they are notoriously lacking in this regard). Let me add to your examples the name of historian Chen Yinque 陳寅恪 whose name’s last character is usually read ke (but in Guangzhou, where he taught and where I spent some time, every scholar knows it’s que).
    The fun with reading variants stops when they are used as a power-tool by the elite, something I did witness more than once. Otherwise, I’d say they are irrelevant as long as intelligibility is preserved (I’m all for historical phonology, but that’s certainly not the most fiable material to base it upon).
    The way I see it, « tabooed characters » (I and other sinologists prefer to say « forbidden/prohibited »), are a slightly different concern. What you write is as important as the way you read it (which increases the risk of execution for scribes who are not careful enough about homophony, homography, and literary connotations ; one emperor’s tendency to paranoïa critique is often deadly). For instance, the Song prohibition of xuan 玄 lead to the adoption of yuan 元, a different character with a different reading, but close enough to guess easily the original one.

  29. The fun with reading variants stops when they are used as a power-tool by the elite, something I did witness more than once.
    OK, you’re going to have to expand on that. My curiosity is definitely piqued.

  30. Mei wenti, LH. I have to go now, but I’ll be back later with a few examples.

  31. In my English, “later” obviously means “in two months”.
    Actually, I did start writing a comment with examples from my personal experience, but it got too lengthy (and maybe a bit too personal). I put it aside, and because of “offline” events, never got to finish it. Unfortunatelly, I have now several deadlines that force me to renounce to it.
    I apologise sincerely to anyone who may have been looking forward to it (I emailed Language Hat about it).
    Talk about “high expectances” (yes, now I know it’s “expectations”)!

  32. I’m afraid I’m running behind schedule as well. But I finally have a beta version of the “shi shi shi” English translation and Pinyin transcription I promised a couple of months ago. (Jimmy, e-mail me if you see this. I have something you might be interested in.)

  33. Mark,
    The Laozi knows no schedule or deadline: Daqi wancheng 大器晚成!
    It’s good to get to see Zhou Youguang’s chapter, this is some great job you did. I just sent you an email to the address indicated on your “Contact” page. Sorry in advance for any silliness that might occur in it, but it is now 02:20 AM here, and thesis writing has eaten my day.

  34. Jonathan Weed says

    If anyone finds this on google and wants to know:
    I’m a first year Chinese student at Princeton and can confirm that we use GR in the first year classes. I’m not sure whether it’s due to Lin Laoshy’s (Perry Link’s) exacting standards or the GR, but all of my Chinese friends comment especially on how precise my tones are. Anecdotal evidence, to be sure, but it seems to work for the Princeton EAS department.

Speak Your Mind