I am wondering about tones. Specifically, wondering if there are cases where a tone shifts from one vowel to the next, perhaps in some language like Mandarin Chinese or Vietnamese, if not some African language. I don't know much about anything outside of Thai and Mandarin Chinese tones at this point.

But say you have a word like /ʃiaŋ/. In Mandarin Chinese, there are several tone variants of that word, like xiǎng in Pinyin, which is roughly from my knowledge just /ʃia˨ŋ/, where tone 3 in Pinyin is roughly a low tone. But is there anything like this?


Basically, where it changes on a per-vowel basis within a single syllable? If not, how does the tone behave in various languages, or specifically just Mandarin Chinese?

I would like to know what happens when you mix tones with long vowel sequences, like hoia or hoauai or something complicated. Does anything like this exist?

  • 1
    Xiang is /ɕaŋ/, not /ʃiaŋ/. In isolation, the third tone will be /ɕaŋ˨˩˦/, with the tone going from mid-low to low to mid-high over the duration of the syllable – this is known as a contour tone. The low quality (more precisely mid-low to low) appears in context due to tone sandhi. Jul 5, 2021 at 17:47
  • 1
    Note that in Mandarin Chinese just one vowel per one syllable is allowed, and what's written in Pinyin as xiǎng is pronounced in Beijing as /ɕjɑŋ²¹⁴/, /ɕjɑŋ˨˩˦/, just one vowel /ɑ/, there's no other vowel to shift the tone to.
    – Yellow Sky
    Jul 5, 2021 at 17:55
  • Just depends on whether you think a contour tone is really a sequence of tones / tonal targets, and if so whether you think they attach to the vowels. For Thai, Moren and Zsiga have argued for a version of this view (they say the targets attach to morae) but FWIW I'm entirely unconvinced by their analysis. You're still limited to two in a row because you can't have more than two morae in a syllable and you can't have more than one tone (counting contour tones) per syllable.
    – rchivers
    Jul 5, 2021 at 18:15
  • @rchivers - Japanese allows three morae per syllable: “entered” 入った (はいった) haitta has 2 syllables hait-ta with the first syllable having 3 morae: ha.i.t-ta — 4 morae all in all.
    – Yellow Sky
    Jul 5, 2021 at 18:33
  • 2
    @LancePollard You’re conflating letters and sounds. Letters are irrelevant, they’re only used when we need to communicate language through a soundless medium. Vowels and consonants are not clearly separable from a phonetic point of view – glides can have properties of both, for example – but in most languages, they behave differently systematically. In Mandarin, a syllable can have only one vocal peak (= vowel), otherwise it’s two syllables; compare 癌 ái [ai̯˧˥] ‘cancer’ (one vowel, one syllable) with 阿姨 āyí [a˥.i˧˥] ‘aunt’ (two vowels, two syllables). Also, z is a vowel [z̩] in bzzz. Jul 6, 2021 at 8:59

1 Answer 1


There is such a thing as "tone shift", which is especially widespread in Bantu languages (see Kisseberth & Odden (2003) in The Bantu languages). Shifting can be by a vowel (mora), a syllable, or multiple syllables. Shifting by a vowel vs. a syllable are the same thing, except when you are dealing with long vowels / diphthongs, and syllable-internal shifts from one mora to the next are not common, but they exist. However, as K&O point out, "shift" can arguably always be reduced to assimilation (spread) plus disassociation, and there are numerous cases where spreading is more general and disassociation is more restricted, giving a mixed case where you have both spread and shift. Please note that a "shift" in tone is a movement from one underlying position to a different surface position.

Your question / example seems to be more about contouring patterns within a syllable. There are two sub-parts to the question. The first is, how many tonal targets can there be in a syllable? That is kind of a highly language-specific question, but there are languages known to have three tones on a single short vowel. There has been a bit of distraction in the literature over the difference between length and duration, because it takes longer to produce a HLH contour on a vowel that to produce a simple H, so you have to apply phonological criteria to decide if a vowel is long, versus has "greater duration" (than...). The absolute max in the attested contours as far as I know is a four-tone complex in claimed to be Lomongo, but I won't vouch for that, but three-element contours are common enough.

The second part is, to what extend can the movement-points in a multi-toned syllable be contrastive. As far as we know, there is only one kind of short rise or short fall, assuming a language with just H and L tones (with 3 levels, you can have many more kinds of rise and fall). You also do not find multiple types of rise and fall on long syllables, even though [v́v̂] vs [v́v̀] would be perfectly sensible from a theoretical perspective. There may be functional and historical reasons why such a contrast doesn't exist. (There is a controversial claim about Shilluk, which would have a separate paper to review). One formal explanation for the syllable-internal restriction is that tones are properties of syllables, not individual vowels.

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.