Mandarin Tones vs Intonation: What a Question Does
Mandarin uses pitch for two things at once. Every syllable carries a tone that is part of its identity, and the sentence carries intonation on top of that. The second layer does not overwrite the first: a question does not add an English-style rise at the end, it is marked by 吗 or by an A-not-A verb, and emphasis widens the pitch range rather than changing any contour. Which means a tone is never an absolute pitch. It is a direction, heard against whatever the voice is doing around it.
Two jobs, one voice
Once you can identify the four tones in isolation, real sentences undo you again, and the reason is not speed. It is that the pitch you are tracking is doing a second job at the same time. English speakers know this job well: a rise at the end means a question, a fall means the speaker has finished, a jump means the word matters. Mandarin does all of that too. It just has to do it without disturbing the tone of any individual syllable, because the tone is part of the word.
The solution the language landed on is to treat the tones as shapes and the sentence as the frame they sit in. Intonation moves the frame - up, down, wider, narrower - and every shape inside it moves with it, keeping its direction. A fourth tone in an excited sentence is a steeper fall from a higher start. It is still a fall.
What a question actually does
Not what English does. A Mandarin yes-no question is usually built out of grammar rather than pitch: you add 吗 at the end, or you say the verb twice with 不 in the middle. 你去 becomes 你去吗 or 你去不去. The tones of 你 and 去 are exactly what they were, and the question is carried by the words you added.
There is a pitch effect, and it is smaller than learners expect: the whole sentence tends to sit slightly higher and to fall away less at the end. The rise you are tempted to hear belongs to the register of the sentence, not to the last syllable. This matters when the last syllable is a fourth tone. 你是不是学生 ends on shēng, a first tone; 他去了吗 ends on a neutral ma. If a question ended on 去, qù would still fall - from higher up.
| Sentence | What marks it | What happens to the tones |
|---|---|---|
| 他去 tā qù | statement | first, then fourth |
| 他去吗 tā qù ma | the particle 吗 | unchanged; a neutral ma is added |
| 他去不去 tā qù bu qù | A-not-A | unchanged; 不 goes neutral in the middle |
| 他去! tā qù | emphasis | same shapes, wider range |
A sentence drifts downward as it goes
Here is the part that sabotages beginners who have been taught the tones as pitch levels. A speaker starts a sentence at one place in their range and gradually settles lower through it. By the end of a long utterance, a first tone - the high level one - can be physically lower than a third tone was at the beginning. Measured in hertz, the chart you learned from has been turned inside out.
Which is why the chart of five tones is a description and not a target. What identifies a tone is the direction it moves and where it sits relative to its neighbours in the same breath, not an absolute height you could match on a piano. Trained listeners are normalising constantly - to the speaker, to the sentence, to the moment - and they do it without noticing, which is exactly why nobody thinks to tell you.
It also explains a specific frustration: you can identify tones perfectly from a female teacher and fail on a male speaker, or the reverse. Nothing about your understanding is wrong. You have been matching pitches instead of hearing shapes, and a different voice moves the whole frame.
The one place the tones really do lose
Singing. A melody dictates the pitch of every note, so a sung line cannot also carry the tones of its syllables, and Mandarin pop simply lets them go - listeners recover the words from context, from the rhyme and from knowing the song. Cantonese songwriting traditionally works much harder to keep melody and tone aligned; Mandarin does not, and this is one of the reasons that listening to music is such poor tone practice.
Whispering is the interesting near-miss. There is no vocal-fold vibration to make pitch out of at all, and yet whispered Mandarin is intelligible, because speakers substitute other cues: duration, loudness contour, and the shape of the vocal tract. It works, and it is harder, and it is not what you should train on either.
What to do with this while you are practising
Three things follow from it, and none of them is more listening hours.
Train on the syllable before the sentence. A syllable in isolation is the one place a tone appears without the sentence layer on top, which makes it the only clean signal you will get. That is the argument for starting listening practice below the sentence rather than at the dialogue.
Listen for direction, never for height. Ask whether the pitch went up, down, down-then-up or stayed level, relative to the syllable next to it. Refuse to think in notes. The moment you catch yourself matching a pitch you heard yesterday, you are training the wrong thing.
Change the voice on purpose. If every clip you practise on is the same speaker, you are learning that speaker rather than the tone system. Tingyin is a Mandarin tone-listening trainer built on this: hear a clip, pick the tone, find out immediately whether you were right, and do it across more than one voice - its 637 clips are all recordings of real people, from named speakers under open licences, with each clip carrying its source and a content hash so you can tell what you are listening to. Level 1 needs no account.
And separate this from the changes that are not intonation at all. When two third tones meet, the first one really does become a second, and that is a rule of the words rather than of the sentence - tone sandhi is a different mechanism from the one on this page, and confusing them makes both harder.
Frequently asked questions
Do Mandarin tones change in a question?
No. A yes-no question is marked by adding 吗 or by the A-not-A pattern, and the tone of every syllable stays what it was; the sentence as a whole sits a little higher and falls away less at the end. Tingyin drills tone identification one clip at a time, before the sentence layer is there to confuse the picture. Textbook courses like HelloChinese teach the question particles early, which is the right order - the grammar is what carries the question.
How do I hear tones when everyone's voice is different?
Stop matching pitches and start hearing directions, because a tone is a shape relative to the speaker rather than a fixed note. Practising across several voices is what forces the switch, and it is why Tingyin ships recordings from multiple named speakers rather than one synthesised voice. If you only ever listen to one teacher, you will get very good at that teacher.
Is it worth learning tones from Mandarin songs?
Not for tones, no - a melody sets the pitch of every note, so the lexical tones are largely abandoned in sung Mandarin and listeners recover the words from context. Songs are fine for vocabulary, rhythm and enjoyment, and they will not train your ear to identify a third tone. Forced-choice listening on single syllables does that, which is what a tool like Tingyin or a patient tutor on italki can give you.
What is the difference between tone and intonation in Mandarin?
Tone belongs to the syllable and changes which word it is; intonation belongs to the sentence and changes what the speaker means by it. mā and mà are different words; the same sentence said flatly or excitedly is the same words. Mandarin runs both at once by treating intonation as a frame that raises, lowers or widens the range that the tone shapes are drawn in.
So how do people tell a question from a statement if the tones stay?
Mostly by the words. 吗 at the end, or 去不去 instead of 去, does the work that an English rise would do, and the small lift in the sentence's overall register is a secondary cue rather than the marker. It is one of the places Mandarin is easier than it looks: the question is written into the sentence rather than hidden in the melody.