Chinese Tones Explained: How to Actually Learn Them

A Mandarin tone is a pitch contour over the duration of a syllable — the shape of the pitch, not its absolute height. This matters: a child, a bass and a soprano all produce the same four tones, scaled to their own ranges. What you are learning is relative movement.

The standard notation uses a five-point scale, where 5 is the top of a speaker's comfortable range and 1 the bottom. First tone is 55: high and flat. Second is 35: starting mid and rising. Third is 214: starting low, dipping lower, rising at the end. Fourth is 51: falling sharply from high to low.

The neutral tone has no contour of its own. It is short, light, and takes its pitch from the syllable before it — which is why it is easier to think of it as an absence of tone than as a fifth one.

Mandarin has around 400 syllables without tone and around 1,300 with. Tones are therefore carrying roughly two-thirds of the language's capacity to distinguish syllables. Dropping them does not degrade your Mandarin slightly — it removes most of the phonological system.

The classic examples are real. 买 mǎi is buy and 卖 mài is sell, which is an unfortunate pair to confuse in a shop. 问 wèn is ask and 吻 wěn is kiss. 睡觉 shuìjiào is to sleep and 水饺 shuǐjiǎo is dumplings.

Learners often notice that they are understood anyway and conclude tones are optional. Context does rescue you a great deal of the time. But the listener is doing the work of reconstructing what you meant, and a conversation where one party is continually disambiguating is tiring — which is why speakers with poor tones find that natives switch to English, or to simpler topics, without ever saying why.

Tone sandhi is where most learners come unstuck, because the tone you memorised is often not the tone you say.

Third tone before third tone: the first becomes a second tone. 你好 is written nǐ hǎo, said ní hǎo. 很好 is hén hǎo. 我想买 involves three third tones and gets grouped by phrase structure rather than mechanically.

Third tone before any other tone: it is reduced to just the low portion, with no rise. This 'half third tone' is the most common realisation of third tone in running speech, and it is why learners who drill the full dip-and-rise sound laboured. 好吃 hǎochī, 老师 lǎoshī, 请问 qǐngwèn all use the half third.

不 bù becomes bú before a fourth tone: 不是 bú shì, 不去 bú qù, 不对 bú duì. Elsewhere it stays fourth.

一 yī is first tone alone and in ordinals, yí before a fourth tone (一个 yí gè, 一样 yíyàng), and yì before first, second or third tones (一天 yì tiān, 一年 yì nián, 一起 yìqǐ).

None of these changes are written, except that dictionaries often show the sandhi form for 不 and 一 because it is regular and predictable.

The standard first-week drill has you produce four tones on one syllable in a fixed ascending order. It teaches you to produce a memorised sequence. It does not teach you to identify a tone you hear in isolation, or to produce the right tone on an arbitrary word in a sentence — which are the two things you actually need.

The tell is familiar: a learner who can recite the four tones flawlessly, and who cannot say whether a word they just heard was second or third.

The second failure is that tones are drilled in isolation and used in combination. Mandarin words are mostly two syllables, giving twenty tone-pair combinations, and each pair has its own rhythm. A learner who has only practised single syllables has to assemble each pair from scratch every time, which is too slow for connected speech.

Train identification first. Have someone or something produce a syllable and say which tone it was. This is harder than production and it is the bottleneck — until you can hear tones reliably, you cannot self-correct, and every hour of production practice reinforces whatever you are currently doing.

Drill in pairs, not singles. Work through the twenty tone-pair combinations with real words: 中国 (1-2), 老师 (3-1), 学生 (2-1), 汉语 (4-3), 谢谢 (4-neutral). Each pair has a characteristic rhythm, and internalising them as units is what makes connected speech possible.

Always learn the tone with the word. Storing 学校 as 'xuexiao, school' and adding tones later means learning it twice, and the toneless version will keep resurfacing. Store xuéxiào from the start.

Use synchronised audio and text. Listening to a passage while watching which word is being spoken gives you tone patterns attached to specific words in real time, at natural speed. This is far closer to how tones are actually used than any isolated drill.

Shadow whole sentences. Speaking along with a recording rather than after it forces your pitch to track a natural contour continuously, which is the only way to get sentence-level prosody rather than a sequence of correctly-toned syllables.

Record yourself and compare against the model. Tone errors are almost invisible from the inside — you hear what you intended. The recording does not.