Key takeaways
- Five tones are distinguishable on isolated syllables, but on an analysis that strips out predictable conditioning there are only two phonemically contrastive tones.
- The light and dark tone categories are conditioned by whether the initial consonant is voiced, so the split is predictable rather than independently learned.
- Checked tones are short and belong to syllables ending in a glottal stop, which again makes them conditioned by syllable structure.
- Word-level sandhi in Shanghainese is left-prominent: the first syllable's tone sets the contour for the whole word and later syllables contribute nothing.
- Because of that, drilling isolated syllables teaches a pronunciation that rarely occurs in connected speech — practise at word and phrase level instead.
Ask how many tones Shanghainese has and you will get five, or two, or eight, depending on who you ask. None of them is wrong. They are answers to three different questions, and once you see which question each is answering, the apparent disagreement disappears — and so does most of the confusion about how to practise.
Five: what you can hear on an isolated syllable
Shanghainese has five phonetically distinguishable tones on single syllables said in isolation.
The five sit inside an older structure. Middle Chinese had four tone categories, each of which split into a dark (陰) and a light (陽) variant, giving eight. Shanghainese has since merged several of those: in the dark category, the rising and departing tones merged into one, leaving three; in the light category, level, rising and departing all merged, leaving two.
Three plus two is five. That is the count you get from a phonetics table, and it is what a careful speaker will produce if you hand them a single character and ask them to say it.
Two: what a learner actually has to contrast
Here is where the number drops, and the reason is worth understanding rather than memorising.
The dark and light categories are not freely chosen. Light tones occur only with voiced initials, and dark tones only with voiceless ones. The conditioning is absolute. If you know how the syllable begins, you already know which category it is in — you are not storing a separate fact.
The checked tones are conditioned too. They are the short ones, and they occur on syllables whose rime ends in a glottal stop. Again: not an independent choice, a consequence of the syllable's shape.
Strip out both kinds of predictable conditioning and what remains is a two-way phonemic contrast — falling against rising — and only in open syllables with voiceless initials.
Which is why many romanisation systems mark only the dark level tone, usually with a diacritic such as an acute or grave accent, and leave everything else unmarked. The unmarked cases are recoverable.
Compared with Cantonese, which maintains six contrasting tones that must be learned and produced independently on every syllable, this is a substantially lighter memory load. Cantonese tones explained shows the other end of the spectrum.
Eight: what the history says
Some sources count the historical Middle Chinese categories rather than the modern contrasts. That gives eight — four tones, each split by the dark/light distinction — and it describes the system Shanghainese inherited rather than the one it uses.
This is where a reader who has consulted three sources and got three numbers gets stuck. All three are describing the same language at different levels: the inherited categories, the surface realisations, the underlying contrasts.
Sandhi: the part that changes how you practise
None of the above is the thing that matters most day to day. This is.
Shanghainese has two layers of tone sandhi: a word layer and a phrasal layer. The word layer is the one to understand first, and it is described as left-prominent.
Left-prominent means the first syllable of the word dominates the tone contour of the whole domain. The underlying tones of every syllable after the first have no effect on the result. The pattern either spreads across the word or shifts along it, depending on which tone the first syllable carries and how many syllables follow.
Sit with what that implies. In Mandarin, a three-syllable word has three tones and you have to produce all three. In Shanghainese, a three-syllable word has one pattern, determined by its first syllable, laid over the whole thing.
This is why Shanghainese is sometimes compared to Japanese pitch accent: the tone behaves like a property of the word rather than of the syllable. It also has a low number of tones relative to other southern Chinese varieties, which strengthens the resemblance.
What this means for practice, and what we changed
The direct consequence is the reason our Shanghainese audio is recorded the way it is.
Lessons in Learn Shanghainese Wu Chinese record whole phrases rather than isolated syllables, and the phrase recording is the primary artefact. Where a single word is presented, it is presented as a unit with its real contour, not as a sequence of syllables that a learner is expected to join up.
The argument is simple and, once you have seen it, hard to argue with: a syllable-level recording of Shanghainese teaches a pronunciation that rarely occurs. The citation tone of a non-initial syllable is overwritten by sandhi in nearly every real utterance that syllable appears in. Drilling it is not merely inefficient, it is drilling something you will then have to stop doing.
The cost of the phrase-first approach is real and worth naming. It is harder to give a learner a clean, isolated model of a single sound, because the sound is embedded in context. It makes the audio more expensive to produce and re-record, since changing one word means re-recording the phrase around it. And learners coming from Mandarin — who are used to syllable-level drilling and find it reassuring — sometimes want the isolated version and do not get it.
We accepted all of that, because the alternative produces confident, careful, syllable-by-syllable Shanghainese that a Shanghai speaker will find distinctly odd.
A practice routine that fits the system
- Learn to hear voicing first. The voiced initial series is the thing Mandarin ears do not have a slot for, and it determines tone category. Get the contrast before you worry about contours.
- Work at two-syllable words, not syllables. Two syllables is the smallest unit where sandhi is visible.
- Copy the whole contour rather than assembling it. Listen, then imitate the shape, the way you would imitate a short melody.
- Watch for the glottal stop at the end of checked syllables. It is short and easy to miss and it changes the tone category.
- Leave the phrasal layer for later. It sits on top of word sandhi, and it is easier to hear once word-level patterns are stable.
If you want the wider orientation first, what a tonal language is covers what tone does across languages, and learning Shanghainese for beginners puts the tone work in the context of a whole course. Other courses in the catalogue are listed under language learning.
Good to know
Frequently asked questions
How many tones does Shanghainese have?
What is tone sandhi?
Why is Shanghainese compared to Japanese pitch accent?
Do I need to memorise a tone for every syllable?
Sources
- Shanghainese (opens in a new tab)Wikipedia — accessed
- Tone sandhi (opens in a new tab)Wikipedia — accessed
About the author
Reign Creative LLC — The independent app studio that builds the apps covered on this site and publishes every article here. On Google Play it publishes as Reign Collective Apps.
We use AI tools to help research, draft and edit our articles. Each article lists the sources its factual claims rely on, and details about our apps come from their live Google Play listings. No article on this site has been reviewed by an independent expert unless the reviewer is named on it.
Spotted an error? Tell us — how corrections are handled is set out in our editorial policy.