File last modified:

More Than You Want To Know About

Chinese Tone

Chinese makes use of syllable pitch patterns to distinguish between words. Two syllables which are identical in their vowels and consonants but vary in tone are just as different as if they varied in consonants or vowels. Some "tonal languages" consider only the pitch of a syllable relative to that of other syllables, but Chinese makes use of both relative pitch and pitch change across the syllable. Mandarin, for example, has a some syllables with a high even pitch, and others with a low pitch that drops yet lower and then rises slightly. Still others drop from high to low or rise from low to high.

The Eight Tone Categories

Different Chinese dialects have different numbers of tones. But a useful traditional analysis of Chinese tone across all dialects and at all periods assumes a system of eight named tone categories. Four are referred to as yīn 阴 and four yáng 阳.

Within each of these divisions we find tones named "even" (píng 平), "rising" (shǎng 上), "going" (qù 去) and "entering" (rù 入). The entering category includes all syllables ending in a stop consonant (P, T, K, and in some dialects glottal stop). The actual pitch pattern of an entering tone is often identical with one of the non-entering tones. (Usage note: The character 上 is read shǎng, not shàng, when it is the name of a tone.)

Chinese dictionaries (or other works) sometimes place a small half-circle beside a character to show that it is to be read in a yīn tone and a half-circle with a line under it to show it is to be read in a yáng tone. Some characters can be read in more than one tone with a difference in meaning (e.g., ꜀中 zhōng = "middle" as against 中꜄ zhòng = "hit a target"; ꜀操 cāo = "grasp, operate, or speak" as against 操꜄ cào = a vile obscenity that you should blush from reading). Therefore some works indicate the less common one with the appropriate half circle. (James Legge did this in his classic bilingual edition of the Confucian classics.)

Most traditional expositions lay this out as a nested squares, with the yīn tones in the inner square and the yáng ones in the outer square. (The Chinese characters here are merely to show how the diacritics are positioned. The traditional numbering in Mandarin, Hokkien, and Cantonese is provided for those who have studied one of these languages.)

So if there are eight tones, how can different dialects have other numbers? The answer is that the eight tones just mentioned are categories, not sounds. Eight is the abstracted, cross-dialectal analysis. It is convenient to think of the various dialects as having "merged" some of the categories, which is why some tone numbers (such as H2 or M3 and M4) occur in both a yīn and a yáng box. We will get to this in a moment.

The Actual Sounds

The actual sounds of tones in these various categories may vary considerably from one dialect to another (and often have nothing to do with the names), but normally a syllable that is in a particular category in one dialect is in the same category in all dialects. For example, in Běijīng 北京 Mandarin a "yīn even" (阴平) tone is high and unchanging in pitch, and that is the official standard. But in the Mandarin of nearby Tiānjīn 天津, it may fall and rise in some contexts. It is still "yīn even" and retains its contrasts with other Mandarin tones. Having a firm grasp of the categories and knowing that the categories, rather than the actual sounds, constitute the shared system can be very helpful to the foreign user of Chinese facing dialectical variation.

Oblique tones. In classical poetry, the even (píng 平) tones, both yīn and yáng, stand together in contrast to all the others taken together, which are collectively called "oblique" (zè 仄). Poetic lines may then exhibit defined patterns of alternation between even and oblique tones.

Students of modern Chinese usually refer to the tones of any particular macro-dialect (such as Mandarin or Cantonese) by macro-dialect-specific numbers, although this tends to obscure the cognate relationship across dialects. For example the "yáng even" tone is numbered two in Mandarin but four in Cantonese and five in Hokkien. (There is a table of correspondences at the bottom of this page.)

Tone Sandhi

Sandhi (pronounced: SUN-dee) refers to modifications in a speech sound caused by its adjacency to particular other speech sounds. (For example, in American English the S in "this year" tends to be spoken as SH because of the sound that follows it.)

In most dialects of Chinese, the actual pronounced pitch pattern of a syllable may also vary by context through a process of tone sandhi. For example, when Mandarin has two third tones occurring in sequence without a syntactic break between them, the first is mutated to sound like a second tone. In Hokkien, students learn that every tone has two pronunciations, the "citation form" and the "combining form." The citation form is used when the syllable stands alone or is at the end of a syntactric structure, while the combining form occurs before any other syllable when there is no syntactic break to prevent it.

Naturally the details can become quite complicated. It is conventional when transcribing most dialects alphabetically to write the unmodified tone and leave it to the reader to apply appropriate pronunciation rules.

As a broad generalization, Cantonese is the only kind of Chinese without significant tone sandhi. I have the impression that tone sandhi is most complex in the "Foochow" (Fúzhōu 福州) dialects of northern Fújiàn 福建.

Reduced Tone Inventories

In many (most?) dialects, fewer than eight tones are distinguished. For example, in Hokkien, there is no difference between yīn rising and yáng rising tones, and both are numbered 3 (or called simply rising 上, although the actual sound is usually a sharp fall). In Mandarin, there is also no difference between the yīn rising and the yáng rising tones (both numbered 3) or between the yīn going 去 and yáng going tones (both numbered 4). And most dialects of Mandarin lost entering (rù) tones centuries ago, "reassigning" those syllables to other tone categories and picking up some other distinctions instead. (The entering tones, recall, are the ones on syllables that end in P, T, K, or glottal stop. Mandarin has lost such endings.)

In some dialects of Cantonese there is a third kind of entering 入 tone, and the entering tones are traditionally numbered 7,8, and 9, thus exceeding the limits of the eight-tone scheme. However because the entering tones are identical in sound to some non-entering tones in modern standard Cantonese, they are not given separate numerical designations in the new Jyutping system used in Hong Kong, but share tone numbers with identically sounded non-entering tones. (I.e., tones 7, 8, and 9 are written as 1,3, and 6 respectively.)

In Cantonese this simplification is possible because most Cantonese dialects, including Hong Kong, do not exhibit tone sandhi. In many dialects of Hokkien, in contrast, entering tones also sound identical to some non-entering ones, but it is NOT possible to number them identically, because different sandhi rules apply.

Tone Negation

Finally, in some situations tone on selected unstressed syllables may be negated, a phenomenon known as the qīng 轻 or "light" tone. In English such syllables may sometimes be called "toneless." Obviously they have pitch, but they are normally brief and unstressed, and the pitch does not vary from the beginning to the end of the syllable. The pitch itself is a function of the tones which precede and/or follow the toneless one.

For example, the "toneless" de in the Mandarin phrase tā de jiā (他的家, "his family") is low in pitch because it occurs between two yīn even tones (pronounced high and even), while the de in shéi de gǒu (谁的狗, "whose dog" is high because it occurs between a yáng even tone (rising in pitch) and a rising tone (falling in pitch and then rising again).

I have the impression that tone negation is commoner in Mandarin than in other macro-dialects, that it is greater in Běijīng than elsewhere, and that it is more frequent in lower-class than in educated usage. However it occurs to some extent in all Chinese.

A special kind of tone negation is said to occur in the lower Yangtse river basin (including Shànghǎi) in which the tone of the first syllable of a syntactic structure is extended across the whole of the structure, masking the underlying tones of the other syllables.

Here is a table showing the tone correspondences among modern Mandarin, Cantonese, and Hokkien as they are usually numbered. The letter A is used to hold the usual diacritic. For Hokkien AH shows an entering tone.

The diacritics have been standardized for a century in Mandarin and longer yet in Hokkien. Jyutping uses numbers, not diacritics. In Cantonese both Yale and Meyer Wemp systems remain in use, and both Hong Kong and Guangzhou pronunciations have been taken as standard. (Yale-Guangzhou is shown here.) Note that the H shown for Cantonese indicates the lower tone register, not, as in Hokkien, the entering tone status.)

Return to top.