This question has puzzled me since I have no training in ancient Chinese and ancient phonetics. What is puzzling me are the following facts:

  1. The tones in different regions are quite different. If they are from Ancient Chinese, why are they not more consistent?

  2. When Chinese characters were borrowed by Japanese and Korean around 1000 years ago, the tones are not picked up by their languages. Did Chinese actually have fixed tones at that time, or not?

  3. Ancient Chinese had codas (Cantonese codas like [p], [t], [k]) but modern Chinese doesn't. Is it possible that the information in the codas is transformed to tones? Otherwise, a lot of information would have been lost.

Any ideas?

  • 2
    Modern Chinese has codas, thank you very much, in its Cantonese, Hakka and Hokkien versions.
    – dda
    May 30, 2014 at 14:05
  • @dda Those may be considered dialects of Chinese (especially in Chinese) but they aren't "modern Chinese". Even if Wikipedia prefers to distinguish Mandarin as "Standard" Chinese and uses "Chinese language" to talk about the entire family, the rest of the English speaking world doesn't, as you well know. Wikipedia itself redirects "modern Chinese" to the Mandarin article.
    – lly
    Jul 7, 2021 at 5:02

3 Answers 3


Linguists divide pre-modern Chinese broadly into two periods: Old Chinese and Middle Chinese. I wanted to preface my answer by noting that Bernhard Karlgren used the term "Ancient Chinese" to refer specifically to Middle Chinese, and it appears that your questions seem to be referring to Middle Chinese as well, though I will be making a note about Old Chinese in my answer.

Middle Chinese is documented as having tones (in particular by the Qieyun in 601) while Old Chinese has been postulated as being atonal. I will go into more detail as I address your questions in order:

  1. The tone contours of the various modern Chinese varieties can appear to be superficially quite different, but the tones themselves are actually quite consistent in the way they evolved from the tones of Middle Chinese. Words that belong to particular tone categories tend to remain in the same tone category across the various modern Chinese varieties, even though the ways the tone itself is pronounced has diverged among differing dialects.

    Take for example, the word 平, which generally always has the 陽平 tone across all of the modern Chinese varieties; however, the 陽平 tone itself has evolved into having, for instance, the contour 35 in Standard Mandarin but contour 21 in modern Cantonese.

    Various tone categories can end up getting split or merged though during the evolution of any particular Chinese variety due to various phonological environments, so for instance, Middle Chinese 陰去 and 陽去 tones both ended up merging into one 去 tone in Mandarin (with contour 51), while they remained separate in Cantonese (with contours 33 and 22, respectively).

    One area where tone mergers were not very consistent is the loss of the checked tones (入) in Mandarin. Words with this tone ended up being redistributed to the other 4 modern Mandarin tones seemingly randomly in Beijing Mandarin (and thus also in Standard Mandarin which is based on Beijing Mandarin); however the redistribution in other dialects of Mandarin appear to more consistent in this regard (refer to the table in this article). One explanation for this inconsistency is that checked tone words in Beijing Mandarin ended up being pronounced in a variety of different tones once the distinguishing feature of checked tone words (namely the stop consonant codas) were lost and the confluence of various dialects in the capital reshaped the tones with which these words were pronounced in different ways.

    Refer to the table in this article for how the various modern Chinese varieties map the original tones of Middle Chinese into their modern tones and tone contours.

  2. Japanese and Korean belong to separate language families that are not related to Chinese even though they imported many Chinese words. The time period during which most Chinese words were imported fall squarely during the Middle Chinese period, which was documented as already featuring tones. Japanese and Korean though, are not tonal languages, and because they don't distinguish lexical tone, it's not unexpected that they would also not distinguish tones in the words they imported.

    For a parallel example, English is an atonal language that has also imported a few words of Chinese origin, such as bok choy from Cantonese and feng shui from Mandarin, and the tones with which these words are originally pronounced are similarly ignored in English.

  3. As I mentioned above, the words with stop codas were classified under the checked tone in Middle Chinese, and when the stop codas were lost in Mandarin, their tone contours ended up being redistributed to the other tones. Strictly speaking, yes, this did end up resulting in a loss of information leading to a possible increase in ambiguity; however, this is not an uncommon occurrence in the evolution of any language. Distinguishing features have been lost in the history languages across the world, but their speakers generally make up for it in other ways. Mandarin, in particular, lost many of the distinguishing features that were present in Middle Chinese, leading to a reduced syllable inventory, but it has compensated by increasing the number of disyllabic words in its vocabulary.

    It is interesting to note, though, that loss of certain codas in the transition from Old Chinese to Middle Chinese did indeed get transformed to tones. It is postulated that Old Chinese was actually not a tonal language, and words ending in a glottal stop () in Old Chinese ended up cheshirizing to the 上 tone and words ending in -s to the 去 tone. See this article for more details.

    Similarly, a later tonal distinction caused by voiced vs. unvoiced initial consonants in Middle Chinese lead to the split of the four 平上去入 tones into two registers, 陰 and 陽, leading to 8 tones (this marked the transition from Early Middle Chinese to Late Middle Chinese). The voiced vs. unvoiced distinction has disappeared in most modern varieties of Chinese (a notable exception being the Wu dialects), but the tonal distinction still remains in many of them.

  • 1
    Just a note that a lot of people consider Japanese dialects to make use of pitch accent (rather than syllable weight), so usually two "tones" in the dialect in the sense that two words with the same phonology outside of pitch can then be distinguished by pitch alone. It is conceivable that Middle Chinese's tones had some influence on Japanese, whether limited to loanwords or not, and that this influence has become masked over time. Unfortunately I do not know much more than that. But it is probably good to anticipate people who say "but Japanese has tones too!"
    – njahnke
    May 28, 2014 at 14:20

As an addendum, two brief comments regarding how tones are reflected in the languages that borrowed a lot of vocabulary from Chinese:

  1. According to the Wikipedia page on ‘Sino-Xenic pronunciations’, “[m]ost Middle Chinese tones were preserved in the tones of Middle Korean, but these have since been lost in all but a few dialects.” The source cited seems reputable. It seems Korean initially did retain the tones!

  2. There is another language which borrowed large amounts of vocabulary in the period of Middle Chinese: Vietnamese. Since this language has 6 tones as well as checked syllables, it was able to preserve the tone categories quite faithfully. This means that the tone classes of Sino-Vietnamese vocabulary items and Cantonese words line up at least 80% of the time, at least that was what I found in an informal comparison of about 200 characters that I did recently.

For example, words of the 陰 上 class correspond to the SV ngã tone:

晚 MC mjonX = Cant maáhn = SV vãn

雅 MC ngaeX = Cant ngáh = SV nhã

But 陽 上 words correspond to SV hỏi tone:

早 MC tsawX = Cant jóu = SV tảo

普 MC phuX = Cant póu = SV phổ

(MC pronunciations taken from William Baxter's list.)


Wikipedia is your friend:

Tones See also: Four tones The Qieyun classified characters in four parts according to their tone: even tone (píngshēng 平聲), rising tone (shǎngshēng 上聲), departing tone (qùshēng 去聲), and entering tone (rùshēng 入聲). The "entering tone", also known as a "checked tone", actually refers to syllables characterized by a final stop consonant (/p/, /t/, or /k/) rather than a distinct pitch.[48]

It is difficult to determine the exact contours of the other tones. Karlgren interpreted the names literally as level, rising and falling pitches, respectively.[48] The oldest known description of the tones is found in a Song dynasty quotation from the early 9th century Yuanhe Yunpu 《元和韻譜》 (no longer extant): "Level tone is sad and stable. Rising tone is strident and rising. Departing tone is clear and distant. Entering tone is straight and abrupt."[m]

Every topolect has their own ancient Chinese as well. For example 四川话 has 蜀语 as its "Classical Chinese" - so your assumption that ancient Chinese is root of all modern Chinese would be wrong. Chinese has a lot of variants - that does not mean that they stem from each other or that they are strictly speaking even dialects.

入声 was originally considered a tone - it is your so-called "coda" as you can see from the quote from Wikipedia above. This can still be found in a few Sinitic topolects today, which you obviously already know.

  • Thanks for your comments. Actually, the problem is way more complex than Wikipedia can answer.
    – Danke Xie
    May 28, 2014 at 4:40
  • From Qieyun, it looks like around 600BC, there was 3 tones and a category of words with stopping consonant (p,t,k). The last one was not a tone at that time, but might have turned into a fourth tone later. Anyway from the information, I can see that around 600BC, there was tones and different stopping consonants. Before that, not much was known. It is left for imagination whether tones were used first, or stopping consonants.
    – Danke Xie
    May 28, 2014 at 4:56
  • The Qieyun is circa 600 AD, not BC. A big difference.
    – neubau
    May 29, 2014 at 1:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.