Chinese Tone Classifications


Chinese is a tonal language, and like other tonal languages, for a given sound, the variance in the pitch creates a different representation of meaning to the listener. Since the tones are often used to distinguish otherwise homophonous words, it need some kind of representation when the written in a romanised form. Using characters, there is no tone distinctions on the written page, and so depends on the erudition of the readers themselves.

In English, the use of tones is rather limited. They can be used to indicate surprise, perplexity, or a number of other emotions, but they are rarely indicated by any sort of special marking. The same word 'Now' and 'Now?' indicates two possibilities where the addition of a question mark alters the reader's intonation of the word when spoken aloud. For Chinese dialects, tones are used for every single syllable. They have neutral readings, in which the word stands alone and read, or they may change tone (tonal sandhi) when they come together to for words of more than one syllable.

It is, therefore, necessary to represent them in some way, so as to distinguish unambiguously to the reader of a body of romanised text, the tone of the words used, so that the meanings are clearer.

A short history of the recording of Chinese tones

The introduction of Buddhism by Indian monks into China in the the early part of the first millenium meant that the sounds of Sanskrit had to be transcribed into Chinese somehow. The monks devised a method whereby Sanskrit or Pali syllables can be split up into two parts, the initial and its ending. The initial is the first consonant; the ending is the remaning part of the syllable. It is akin to using an alphabet in that it orders all words into some set and defined order. Using this method, an unknown word can be represented using two known sounds, one for the inital and one for the ending of a given syllable and was known as FanQie ( ).

When it was later applied to Chinese sounds, Shen Yue ( ) (441-513 AD Scholar of the Kingdom of Liang , alias XiuWen ) invented a system that could distinguish the single syllable words of Chinese by their tones, in naming four categories of sounds called the Si Sheng ( ). Thus it gave a finer resolution for Hanzi categorisation in his work Si Sheng Pu . It is the these tones which we shall present here.

Traditional tone categories from the fanqie rhyme tables

The method of FanQie meant that the rhymes could be sorted out into the Si Sheng tones. We give their names in the the three Chinese dialects of Hakka, Cantonese and Mandarin.

Four Tones

Why are these the names that were used?

The names themselves were representative of the pitch of the sounds they intended to cover at the time of their coinage. So, the words Ping, Shang and Qu tones had tones I, II, and III repectively.

Tone I was one in which the pitch of the sound remained level, i.e., at a constant pitch.
Tone II was one in which the pitch of the sound rose from a lower pitch to a higher one.
Tone III meant the pitch of the sound fell from a higher pitch to a lower one.
Tone IV was ended in one of the three types of stops, -b (-p), -d (-t) or -g (-k).

Today standardised Mandarin Chinese no longer has the IVth tone. It has lost these three endings. However, Hakka and Cantonese still retain the endings and so have Ru type tones.

There is a tale that mentions the King of the Kingdom of Liang asking what the work of the Si Sheng Pu meant after Shen Yue's presentation. A compatriot and colleague of Shen Yue, called Zhou She ( ), replied to the King, , which is "Tian Zi Sheng Zhe" or "The son of heaven is divine and wise" and proceeded to explain using these four characters the meaning of the tones. It turns out that the four characters themselves are representative of the class of tones in which they once belonged to at the time of the Six Dynasties, of which Liang was one. (See end of page for acknowledgements.)

Yuen Ren Chao's Pitch Representation

The representation of tonal pitch contours as numbers is attributed to the late Yuen Ren Chao. He devised this to cover the tonal aspects of the Chinese language and other tonal languages. It consists of five arbitary levels, rather like the use of staves in music score. They were labelled from the bottom upwards, 1 through to 5 inclusive. As with the music score, the lowest line represents the lowest pitch, whilst the highest line, the highest pitch. The variance of the pitch could be captured using the reference pitch numbers by observing the starting, middle and end-points of the tone. The numbers were then enclosed in two forward slash marks. For example, /55/ would be a high level tone, whilst /11/ is a low level one. /53/ is a high falling tone, /35/ is a mid rising tone, whilst /31/ is a mid falling tone. /214/ is a tone which starts low, falls and then rises again. Short tones can also be represented as a single number for instance a short mid level is /3/. There are tone discriptions which are more complex for instance, /2342/.

Example of the pitch contour /42/ in stave form.

Modern Dialects and

Let us analyse the tones of this 'phrase of praise', which our three modern dialects gives us in relation to the four categories above.

wordMC Tonetone typesoundtone typepitch contoursoundtone typepitch contoursoundtone typepitch contour
Itden 1/33/tin /55/tian 1/55/
IItsu 3/32/tsi /35/zi 3/214/
IIIsen 453sing /33/sheng 4/51/
IVtset 5/3/tsit /3/zhe 2/35/

We see from the table above that the characters used are still in their original tone classifications in Hakka and Cantonese, and to an extent in Mandarin. Prof. Majorie Chan pointed out this relation with respect to Cantonese in a message to the discussion group in January 1988, in which I have extended the idea to cover Hakka also.

The Hakka Dialect used here is the same as that within this site, ie, Satdiugok/Shataukok/Sathewkok. It has the peculiarity of having a falling tone contour for the rising or Shang tone. The Cantonese dialect's tone contours are reproduced from two sources, Dr. C. F. Lau, and Professor Marjorie Chan. The Hakka tone contours came from Hashimoto's The Hakka Dialect, and Henry Henne's Sathekok Phonology.

As we can see, the in Hakka, Cantonese and Mandarin, there is agreement in the tones types used. In today's Hakka and Cantonese dialects, there are more than the four tones. Hakka in general has six tones. Cantonese has nine. We shall discuss them in the following section.

Due to the loss of endings in Mandarin, we notice that the last word in the phrase is not of the Ru type, but of a second type of the Ping type. The original Ping tone split into a high pitch register and a low pitch register in all three dialects. The characters which had Ru tones in Mandarin in the past merged into the classifications of Ping, Shang and Qu which is what we see today. The Ru tones in Hakka split into two pitch registers, high and low. In Cantonese it split into three, high, middle and low. Cantonese had futher splitting of the Shang and Qu tones into high (Yin ) and low (Yang ) registers, hence today we see the nine tones.

Hakka and Cantonese Tone Contours.

Hakka Tones
of Tone
Tone Contours
SaTdiuGok MoiYen
1yim pin sangl
mid level/33//33/
2yong pin sanglow level /11//11/
3song sangmid falling /32//31/
4hi sanghigh level /53//55/
5yim ngip sangs
mid falling /3//31/
6yong ngip sanghigh rising /5//44/

Cantonese Tones
of Tone
Tone Contours
Hong KongGuangZhou
1yam ping singl
high level/55//53/
2yam seung singmid rising /35//35/
3yam hoey singmid level/33//44/
4yeung ping singlow level/11//21/
5yeung seung singlow rising/13//24/
6yeung hui singlow level/22//33/
7yam yap sings
high level/55//55/
8jung yap singmid level/33//44/
9yeung yap singlow level/22//33/

Here I have supplied the 'standard' dialect tones of MeiXian/MoiYen Hakka and GuangZhou/GwongJou Cantonese for comparison of the tone contours. As you can see they are slightly different from its other version. All the data for the Mandarin tones are found in the first table above, but let us compare the three dialects together under one table, according to the four tone categories. The GuangZhou tone contours come from Jerry Normans's "Chinese" (1988) p. 218, table 9.6. The Hong Kong tones are supplied by Lau Chun Fat in a message to the Hakka Forum, entitled "Cantonese Tones".

Hakka, Cantonese and Mandarin Tone Contours

Four Tone
SaTdiuGok HakkaHong Kong CantonesePuTongHua Mandarin
Tone TypeChinese
1/33/ 1/55/ 1/55/
2/11/ 4/11/ 2/35/
3/32/ 2/35/ 3/214/
III4/53/ 3/33/ 4/51/
5/3/ 7/55/NO
Ru tones in
Standard Mandarin
6/5/ 9/22/

Please note that I am comparing SaTdiuGok/Shataukok/Shatoujiao Hakka, with Hong Kong Cantonese, and textbook Pinyin. They are the three dialects of which I am familiar with.

For Type I Ping tones, we see there is a consistant high and low level tone distinction in all three dialects.

Type II Shang Tones are peculiar in Hakka, as we often see a falling tone contour as in the case above. In Cantonese and Mandarin, these tones are rising, which is consistant with the name description. Note that in both the high and low distinctions in Cantonese, they are indeed rising from a lower level to a higher one.

Tone Type III, the Qu tones, are as one would expect, falling, though the higher of the two pitch contours for Cantonese is strangely level.

As we have mentioned, the Standard Mandarin dialect has no Type IV or Ru tones. In Hakka , we see that the high and low pitch Ru tones are reversed in comparison to the the Cantonese tones. Cantonese has three distinct levels, high, low and a mid level short or staccato tone. If we group the two higher pitch tones in Cantonese to the low pitch Hakka, and low pitch Cantonese with the high pitch Hakka tones, then we see a flip flop relation, which is borne out quite well in speech. When a Hakka says a word which has a low Ru type tone, the Cantonese is often in the higher pitch, and this can only be taken as a rule of thumb, since it is not 100% accurate.

I have included a mention of Chinese poetical prosody categories, Ping and Ze - see next section below.

Rhyme table in the Four tone classes, Chinese Prosody

The rhyme words below are in a defined order. They are grouped into rhymes across and they belong to the same tone class vertically, in the Ping, Shang and Qu classes. On the other hand, the Ru class, which has its own group of endings. The romanisation used here is Wade Giles, and is from Weiger's "Chinese Characters", and the higher and lower Ping distinctions came from the Japanese Kanji Dictionary, , from Bennesse.

Ending RhymesEnding Rhymes


ung u ü, iü, etc
4ihih, ei, i é, etc
7ue, o, ai
8ia, ia
9ai, uai, yehe, ieh, üeh
10ei, ueiiao, ieh, o, uo, ao
11en, in, ün, un ai, ei, ê, i, ieh, ih, o, uo, ü
13an, uan, yen
14i, ih
15a, ia, o, uo


ieh, yeh
17ao, iao
18There are 106
characters here.
In the

dictionary, there
are no romanised
forms. When a Kanji
character is given,
the rhyme and tone is
represented using
one of these
20e, o
21a, ai, ya, ua
22ang, iang, uang
23eng, ing, iung
26iu, ou
27en, in, un
28an, ien

The structure of a Chinese poem is such that the two categories, Ping and Ze come into play. For four line stanza, each line having seven characters, the last character of the first, second and fourth line must belong to the same rhyme category. That is, if the last character of the first line ended in a Ping rhyme category, then the second and last line must also end in a Ping rhyme. If the last character of the first line ended in a Ze class rhyme, then second and last must also end in a Ze class rhyme. Notice that the Ping category is by itself, but split into a higher group, and a lower group. (I am not sure about the significance of this, but I would hazard a guess that this showed the high and low register tone register, as we see the splitting of the Ping tone also into high (yin) and low (yang) ping tones. In Hakka, the tone contours are somewhat different to those of pre modern times, so a direct comparison on the pitch gives inconclusive, if not misleading, results.) The Ze category includes the other tone classes, Shang, Qu and Ru, where Ze means "unlevel".


I would like to thank Dr. C.F. Lau and the contributors to the participents in this, who have replied to my request for the origins of the tone category names, especially (alphabetically) Daniel Bryant MENG@UVVM.UVic.CA , Marjorie Chan , Richard S. Cook , Olli Salmi , Shiangtai Tuan . Any mistakes here are totally my own.

Hakka Links : Go To Dylan's homepage
This page was created on Wednesday 28th January 1998.
It was recently updated on Thursday 12th February 1998

© Dylan W.H.S. 1996-1998
You can e-mail me here on anything you like!