Chinese Tone Tutorial
The Universal Tone System
By James Campbell

Introduction
Universal Tone System
Order of Tones
Tone Contours
Tone Chart Reference Numbers
Tone Categories Have More Than a Two-Way Division
Vietnamese Tones
Tones in Other Dialects
Endnotes

The order or counting of tones in each language varies. One must understand that no matter how you count them, they are all a subset of one and the same system. Mandarin 3rd tone is not going to equal the 3rd tone in Cantonese, Taiwanese, or Shanghaiese. You have to use a universal system in order to compare not just between languages but between dialects as well. That universal is the original Chinese tonal system from which all modern dialects evolved.

This system consists of 4 tone categories. They are called Píng 平, Shǎng 上, Qù 去, and Rù 入. In the charts below I list them in order as 1, 2, 3, and 4 across the top of each chart. Eventually, each tone split into a higher and a lower one (Yīn 陰 and Yáng 陽 respectively). These are called Yīn Píng, Yáng Píng, Yīn Shǎng, and so on. Normally words starting with voiced consonants all dropped to the low tone. Translated names you may come across may be level (píng), ascending (shǎng), departing (qù), and entering (rù). Some writers refer to positions in the tone table with the translation such as upper departing or lower departing, etc, instead of using the Chinese terms yīn qù or yáng qù, and so on.²

Here is a poem about the Ancient Chinese tones as found in Kāngxī Zìdiǎn (康熙字典):

平聲平道莫低昂，
上聲高呼猛烈強，
去聲分明哀遠道，
入聲短促急收藏。

In Ancient Chinese, the Píng tone was probably a mid-flat tone as found in present-day Changsha Xiang or Guilin Mandarin. The Shǎng tone is described as a high flat tone as found in Jinan Mandarin. The Qù tone is described as a long and low tone as found in present-day Chengdu Mandarin. The Rù tone is described as having a short, clipped sound, about half the length of a Píng tone, as found in Yangzhou Mandarin and Nanchang Gan. The fact that some modern dialects have the same tones as the ancient ones is completely by coincidence, as these tones have probably changed many times before they reached their present forms.

Mandarin doesn't have any voiced onset consonants left except for m, n, l, and r. Mandarin only has one tonal split (in Píng--the 1st and 2nd tones), so anybody can do the following check: open your dictionary (or follow the links here) and find how many words starting with m, n, l, and r are in the Mandarin First Tone (Yīn Píng) and how many are in the Second Tone (Yáng Píng). Except for those very few exceptions (like mā-ma) that have occurred, most words follow the rule--they've dropped to the Second Tone.

There are also many other words beginning with consonants that only occur in 1st, 3rd, and 4th tones in Mandarin because these words evolved from ancient unvoiced consonants. Examples of this are those words that begin with pinyin ‘g’. A few exceptions are characters originally having shifted from Rù tone (such as 國 guó).

Most of the modern Chinese languages function in this way. And if you can understand this system, it makes it a lot easier mapping out your tones when switching between languages.

The tonal contours themselves are actually irrelevant to this system. Of course, the contours are the most important for speaking, but with contours, there’s no way we can match up the tone categories. Tonal contours change over time easily, and are vastly different from place to place. So the category that tones belong to has no relation with a tone’s contour. It has more to do with the phonology of the syllable, such as the features of onset consonants, middle vowels, and rhymes, which determine the tonal category a character belongs to.

The way the tones are counted and listed in each of the modern languages is also completely irrelevant to this universal system I just described.

Cantonese normally counts the tones in the following order:

Mandarin:

Minnan (Southern Min):

Wu (Shanghai):

Hakka:

All the characters in the language are categorized into one of the 8 tones in the system. Although it is apparent that not all eight tones exist anymore in the languages, we have to take into account the merges that have occurred. No matter what language or merged tone the character 八 is in, we should always consider this an Yīn Rù character, even though in some languages like Mandarin, it is now pronounced in first tone (Yīn Píng).

In order to read the numbers in the charts, you have to know how they represent tone contours. We use a system developed by Chao Yuanren 趙元任. This system is based on 5 pitches from low (1) to high (5). Most tones are written with two numbers: the starting pitch and the ending pitch. Those with only one number are shorter, clipped tones. Sometimes these short tones have a change in pitch, so they are underlined to indicate their shortness and to differentiate them from other long tones.

5---5 high
4---4
3---3 mid
2---2
1---1 low

The charts in the following section have the tone contours written according to this system. Mandarin’s first tone is 55, second tone 35, third tone 214, and fourth tone 51. 55 means it is high and flat. 35 means it starts mid (3) and rises (to 5). 214 means it dips low (21) then rises to mid-high (4). 51 means it starts high (5) and drops all the way low (1).

Since in some cases tones have merged or they are equal to the contours of another tone, instead of leaving the space blank in the chart, I could write in the tone contour that it equals. In most cases, such as the Mandarin and the S. Min, the blank Yáng tones have merged with their Yīn counterparts, so it's not necessary to write them in. I leave them blank. Of course there are still characters that belong to those Yáng tones, but they are just read the same as those they've merged with.

But in the case of Shanghaiese, it's not that simple. Notice that there are no tones in the 2 category (Shǎng). So, we know that Chinese has Shǎng tones, but what happened to them, where did they all go? So here it is necessary for me to write in where they've gone. If I write the following:

Wu (Shanghai):

It is unclear to me which tones have merged. In fact it looks like a whole bunch of tones are exactly the same. So which ones have merged? The best way is to reference the merged tones to the one(s) they have merged to. In order to this, we just use the coordinates in the chart. Those coordinates are:

For Shanghai Wu, 2A = 3A, and 1B = 2B = 3B. In 2A's place we write 3A, and in 1B and 2B's place we write 3B:

Wu (Shanghai):

So Mandarin could be written in this way, but it looks a little cluttered because of all the merges that have occurred:

We have already seen that each tone category can be divided into Yīn and Yáng divisions. And we have already seen that these divisions are based on the feature of the onset consonant: i.e. whether or not it is voiced. However, each tone category can be divided more than four ways each. Let's look at the most important divisions that you'll come across in the tone charts:

First let's assign the Chinese names to the Yīn and Yáng divisions by their phonological names. Unvoiced (Yīn) consonants are called qīng (清), and voiced (Yáng) consonants are called zhuó (濁). Notice that 清 in Ancient Chinese starts with an unvoiced consonant, and 濁 starts with a voiced consonant. We can further divide these two by aspiration: quán (全) and cì (次). Notice that 全 is unaspirated and 次 is aspirated, also by their own pronunciation in Ancient Chinese. For voiced consonants, 全 and 次 differentiate obstruents and sonorants rather than aspiration. So by combining these four terms, we get the following:

全清 unvoiced and unaspirated (-v, -asp)
次清 unvoiced and aspirated (-v, +asp)
全濁 voiced and obstruent (+v, +obs)
次濁 voiced and sonorant (+v, +son)

The Yīn Rù tone shifts in Mandarin need much finer distinction than the four mentioned above. However, the Yáng Rù are a little clearer. The 次濁 (sonorants) have merged with 3A, for example the characters 目 and 日; and the 全濁 (obstruents) merge with 1B, such as 白 and 達. However, explanation of these Rù tone merges have not been included in the tone charts.

Vietnamese has an interesting tone merge that one should pay close attention to. This does not appear to have occurred in any Chinese dialect. First let's look at the names of Vietnamese tones and how they line up with the Chinese system:

bằng (không dấu) = Yīn Píng
sắc = Yīn Qù, Yīn Rù
huyền = Yáng Píng
hỏi = Yīn Shǎng
ngã = Yáng Shǎng
nặng = Yáng Qù, Yáng Rù

One will find much fewer Yáng Píng characters in the huyền tone in Vietnamese than one finds in Chinese Yáng Píng. The reason for this is the merge of 次濁 with 清 (Yīn) characters, leaving only 全濁 to be read in huyền tone. This would account for mostly words that start with nasals, and a few other consonants.

Vietnamese tones are produced together with glottal stricture. In this case, an IPA symbol "ʔ", the glottal stop, has been introduced into the tone charts as in the following chart for Việt Hà Nội. Note that the Rù tone category has merged with the Qù category:³

So here are the tone charts for several dialects. In the master tone charts on this site, the charts appear in linear format in order to more easily compare data between dialects.

Guangzhou Yue:

The tone split in the 4th category can be explained by the presence of an /a/ or /ɑ/ in Ancient Chinese syllables. Such syllables have fallen into their own tone category, the lower Yīn Rù, a phenomenon that belongs only to Yue dialects.

So how many tones does Guangzhou Yue (Cantonese) have? By looking at this chart, we can see there are more than one way to count them. First we can count the categories. There are four traditional categories. Each category has two divisions traditionally, except Yue, which has an extra in the fourth category, yielding 9 divisions, or tones. But if we count all the contours, there are 10, since there are two ways to read the first tone. However, if we compare the different contours, we will notice that all three divisions in the fourth category match one of the other contours, so 10 - 3 = 7 different tone contours. Many textbooks will not actually consider the two 1st tone readings, since it has disappeared in some areas, therefore it is common to see citation of 9 tones, and 6 individual contours. This is a confusing matter, but it must be kept in mind that any number (6, 7, 9, 10) is correct, it just depends what you are counting.

Taipei Minnan:

Again here, the counting of tones may be confusing. Taiwanese Minnan cites tones 1-8, but the 6th tone (2B) is not used. In addition to just the tones listed here are tone sandhi, which are quite abundant in Min, Wu, and Jin dialects. In the case of Min, most sandhi appear to match other base tones that exist, but in the case of many Wu and Jin dialects, these sandhi do not match any other base tone, therefore creating the existence of more tones that aren't generally counted in these charts. In some cases there may be twice as many tone sandhi as there are tones. Each of the Minnan tones undergoes sandhi to another tone (7 sandhi), but the 4th and 8th tones (4A and 4B) have two sandhi possibilities adding another two, or 9 cases of sandhi in Minnan. However, three sets of 2 sandhi match (4A & 3A → '53'; 4B & 3B → '21'; 1A & 1B → '33'), meaning 6 are reduced to 3 separate sandhi, while there are 3 others, for a total of 6 different tone sandhi contours. Each of these 6 tones match one of the base tones ('53' = 2A, etc.), eliminating the chance of any extra tones occurring. Therefore, including all sandhi, the Taipei Minnan dialect presented here has a total of 7 tones, no matter how you count it.

Hailu Kejia

	1	2	3	4
Yīn (A)	42	13	21	55
Yáng (B)	44	A1	22	32

2B has merged with A1 meaning all characters with Ancient Voiced Onset consonants in the second tone category are now pronounced with a 42 tone contour in this dialect.

This is the end of the tutorial. With this information you should be able to better understand the charts presented on this site. If you have further questions, please write to James

¹ The information about the tone system presented here has long appeared in the literature on Chinese historical phonology. Most publications mention it in one form or another, and I have just combined much of the information into one source here. As reference, the first time the tone system appeared in the literature was in the History of the Southern Dynasty written by Yue Shen (441-513 AD), Yao Xie, and other scholars and poets. At present, I do not have a copy of this work.

² Píng, Shǎng, and Qù are collectively known as Shu-sheng (舒聲). Shǎng, Qù, and Rù are colletively known as Ze-sheng (仄聲).

³ Vietnamese tone contours on this site are based on my own findings and are not cited in the source material. An important issue has come to my attention: In the dictionary, 越汉辞典 Từ Điển Việt Hán, tonal contours are provided for the standard (northern) dialect, which I copy here: (55 / 11 / 214 / 135 / 35 / 11). Dylan Sung has posted an updated page for Vietnamese tones here which has several links to references online. Upon closer look, one finds that three of these tones are exactly the same as in Mandarin, however when speaking Vietnamese and Mandarin, it will be obvious to the speaker and listener that the tones in these two languages are completely different. Not only are these three tones cited incorrectly, but the rest of the data is also inaccurate if an analysis is done on Vietnamese tone contours. Vietnamese is not a Sinitic language like the other dialects listed on this site, and I take into consideration tonogenesis features in Vietnamese that differ such as creaky voice, clarity, and glottal stricture, which is pronounced as part of the tone contours and so should be written in. This site intends to provide authoritative information and even though tone contours conflict in some places, if it is proven that one is incorrect, I will remove that information from the site, and I do welcome readers and reviewers to submit such information. I am without doubt in this case that my Vietnamese tone contours are correct.