This page is a discussion of Chinese hanyu pinyin 汉语, the Roman alphabet system for spelling the sounds of Chinese words, which is used as an auxiliary script for Chinese, alongside the familiar Chinese characters.

There is no doubt about the usefulness of pinyin, and in fact it has made possible the integration of Chinese characters into the modern digital world.

• Text Entry

  Pinyin is by far the most used entry method for typing Chinese. There are alternatives: the 五笔 wubi system, the 倉頡 cangjie system, and the 四角號碼檢字法 four-corner system are all based on the characters (not their sounds), and they're all still in use, although uncommon. In Taiwan, zhuyin fuhao, which is also phonetic, is often used. But pinyin is the most used in the PRC, and the easiest to learn.

• Collation

  Pinyin is used to look words up or to put them in alphabetical order. Again, there are alternatives in traditional Chinese literacy, for example the 康熙 kangxi dictionary scheme with its 214 radicals - but they are difficult to use and to learn.

• Pronunciation

  Pinyin is used to spell the pronunciation of words. As with the cases above, there is a traditional method called fanqie 反切, based on specifying other words with the same pronunciation, or combining words with the same elements of pronunciation, as if we were to say that "cat" is pronounced with the first sound of "cow" and the final of "hat". But it's much easier to use a phonetic alphabet.

• Pedagogy

  Pinyin is used to teach Chinese, both to natives and to foreigners. Characters can be taught to people who already know the meaning and pronunciation of the words, but it is difficult to learn all three - meaning, pronunciation, and idiosyncratic written form - at the same time.

But all these advantages derive from the adoption of a sound-based spelling, not from the specific system. Even the Wade-Giles system of 1892 offered these benefits, and worked well for most of a century. But Wade-Giles was always intended as a system for specialists, and as China awoke from its Great Slumber, reformers became more ambitious: they wanted a true phonetic spelling system that would replace characters for the general population.

This dream was shared by many of the luminaries of this epoch - for example Sun Yatsen, Mao Zedong, Qu Qiubai, and Lu Xun - and it was made a priority of both the Republic of China and the People's Republic of China. Two original romanizations were developed in this period: Gwoyeu Romatzyh in China and Latinxua Sinwenz in the USSR, both intended to replace characters. A non-Roman phonetic alphabet, Zhuyin Fuhao 注音符號, was also invented in this epoch (see the page on Bopomofo). But in the end, the nationalist agenda promulgated by Chiang Kaishek won out, and the reform was limited to simplifying characters and using pinyin only in an auxiliary role. It's ironic that the Chinese Communist Party, Chiang's enemies, are so loyal to his agenda!

During the Second World War, American pilots were flying missions over China against the Japanese, and they sometimes found themselves in the middle of nowhere after being shot down, bailing out, or crash-landing. In order to enable them to talk to their Chinese allies on the ground, the US Army began to teach pilots some Chinese, using a romanization developed at Yale University: the Yale Romanization. This was the first ever practical romanization of Chinese, and the first one to use diacritics for tones. But it was designed for Americans, and needed to be adapted for Chinese use.

Thus was born Hanyu Pinyin 汉语拼音, fathered by a non-linguist banker, Zhou Youguang, who had lived in New York and seen the Yale romanization. He and his team did a good job ... but, in my opinion, they made some mistakes. On this page, I will discuss the shortcomings of pinyin itself, as an alphabet. In several aspects, the Yale romanization was better. The ROC government in Taiwan also recognized the problems with Hanyu Pinyin, and tried to replace it twice, with Mandarin Phonetic Symbols II in 1984, and with Tongyong Pinyin 通用拼音 in 2002. But in the end, they were forced to accept Hanyu Pinyin, despite its flaws, because of its widespread use.

The main problems with all these romanizations is simple: the Roman Alphabet doesn't have all the letters needed to write Chinese. The missing letters would represent the sounds of the retroflex and palatal affricates and sibilants, and the "apical" vowels. There are some other sounds in Standard Chinese that are pronounced slightly differently from common European use, but that's not a problem: many Roman letters, like j c z r, are pronounced differently in different European languages, too. There were also some bad design choices made for pinyin.

Here is my list of the flaws of Hanyu Pinyin:

1. The three syllables -iu -ui -un should be spelled -iou -uei -uen, as they are pronounced. The abbreviated versions are not ambiguous - there no other syllables spelled -iu -ui -un - but there is no need to save a letter every once in a while, especially when that letter is the one that should bear the tone.

2. The letter ü is not in the English alphabet, and it's not on most keyboards. Why not use the unused letter v to represent this vowel (as many people already do)? That also makes it easier to write the tone mark above it.

3. The letters j q x represent palatal sounds, while the digraphs zh ch sh represent the matching retroflex sounds. But they never occur in the same context! We could use the single letters to spell the retroflex sounds, too. Yes, they sound different, but so do many other pinyin letters, like i a e. The letters j q x should always be followed by a y: jy qy xy that replaces the i of pinyin and turns the ordinary u into a ü.
   Interesting side note: As recently as 1892, when the Wade-Giles romanization was developed, the palatal sounds j q x were pronounced like the velar sounds g k h in Beijing dialect, which is why they romanized the name of the city as Peking - that's how it was pronounced then! But in Chinese, those velar sounds have become palatal, as has happened in many other languages. For example, in Latin civitas, the c was pronounced as a k. Over time, it became a ky before i and e, then a ch, and finally, we pronounce the word city with an s! The ci in words like suspicion is pronounced as a sh - that's the same phenomenon. So we can predict, with some confidence, that in the not-too-distant future, the Chinese j q x will be pronounced just like the zh ch sh are now pronounced, and all that will be left is the odd spelling: pinyin users will have to remember to spell zh ch sh as j q x before i and ü as a spelling rule.

4. In pinyin, the medials y and w are spelled i and u. But those are vowel letters, and the medials are not vowels, as we can see when there's no initial: they're spelled with semivowels as in yan and wan, for example. Using y and w not only unites wei with dui - they rhyme! - but it also helps identify which letter is the vowel: xi'an might be confused with xian, but it wouldn't be confused with xyan.

5. The apostrophe in Xi'an is not a punctuation symbol that we write to separate the a from the i; it's a letter! It spells the sound written in IPA as ɰ, the semivowel glide that corresponds to the vowel ɯ, or it may be pronounced as a glottal stop ʔ. It can be omitted after a space or an initial consonant - so, most of the time. But we need it when there's no initial and it's not initial.

6. Most of the rhotacized finals that result from erhua can simply append an -r to the normal final. However, pinyin doesn't write the changes this triggers, and it should. Final i and n are lost before -r, and final ng becomes a nasalization sign, written ∼. The result is closer to the pronunciation, but some finals become ambiguous (as they are in speech).

7. The letter o should be reserved for mid vowels: the finals -ao -iao should be spelled -au -iau. The finals -ong and -iong and their naked versions weng and yong can stay as they are, since they follow the pronunciation.

8. Why leave out the w in bwo pwo mwo fwo? To save a letter? If we always spell it -wo when it's pronounced as a rounded o, then we can respell the -e final as -o when it's pronounced as a back ɤ, as it often is. To spell the interjection that pinyin spells ê, just use e - it rhymes with the -ie finals.

   There are a few other cases that we would like to fix, if we could add a few more letters to the alphabet:

9. The is in zi ci si and the is in zhi chi shi ri don't sound at all like the normal i in yi bi pi mi di ti ni li j qi xi, and they shouldn't be spelled the same way. (They don't sound like each other, either, but they're close.) There's a good letter for that sound: the undotted ı of Turkish. So we'd spell them zı cı sı and zhı chı shı rı.

10. For the ng final, there is a lovely letter: the eng ŋ. When this ng final is made diminutive with the addition of r, we can spell the resulting nasalized vowel as -∼r it definitely doesn't sound like it's spelled in pinyin: -ngr.

11. The initial r is not an r at all: it's the voiced equivalent of sh, written zh in some transcription schemes. The IPA has a nice letter for it: ʒ.

12. The medial y in yu yue yuan yun is not really a y (and the u's are really ü's, too): it is the semivowel written ɥ in IPA, the semivowel equivalent of ü.

As you can see on the Chinese page, most of these problems don't come up in Musa, simply because Musa has all the letters we need. In Musa, we have letters for both retroflex zh ch sh r and palatal j q x. We have different letters for the apical vowels after retroflexes and the palatals, and different letters for rounded and unrounded o. And we have vowel letters for the sounds of the finals, not just their phonemes.

In addition, Musa writes Chinese in Fangzi gait, so that syllables hang together as a unit. The elements within a character are arranged differently for different tones, so different words are easier to recognize. But the written form still indicates the pronunciation, not the meaning!

Musa for Chinese >