The 4 tones of Chinese (plus the neutral tone): the guide to finally mastering them

In English, voice pitch expresses emotion or intent: we rise at the end of a question and fall to make a statement. But the word itself doesn't change: "cat?" and "cat!" are the same animal.

In Chinese it's different: pitch is part of the word. The same syllable ma can mean "mum" (mā), "hemp" (má), "horse" (mǎ) or "to scold" (mà). Changing the tone changes the word, just as changing a vowel does in English ("bit" / "bat").

You'll often hear: "Chinese people will understand from context." That's true for one mistake, but a sentence where every tone is wrong becomes very hard to follow, even for a patient native speaker. Chinese has only about 400 syllables: without tones, homophones become unmanageable.

The good news: there are only four tones (plus a neutral tone), and they form a very regular system. With good training at the start, they become second nature.

Linguists describe tones on a scale from 1 (lowest) to 5 (highest) of your voice. The first digit shows where the tone starts, the last where it ends.

• Tone 1, 55: high and level. Start at the top of your range and stay there, like a sustained note.

• Tone 2, 35: rising. Start mid-range and rise to the top, like a surprised "Huh?" or "What?".

• Tone 3, 214: dipping. Start fairly low, drop right to the bottom of your voice, then rise slightly.

• Tone 4, 51: falling. Start high and drop sharply, like a curt command: "Stop!", "No!".

• Neutral tone: short, light, with no contour of its own. Its pitch depends on the preceding tone.

Important: these pitches are relative to your own voice. A deep male voice and a child's high voice each have their own "5" and "1". What matters is the movement and the range.

Textbooks draw tone 3 as a big "V" (214). In reality, in normal speech, the full tone 3 is only pronounced when it stands alone or at the end of a sentence.

Everywhere else it becomes a half-third tone: a low, flat tone (21 or 211), with no final rise. For example, in 很忙 or 我是, the third-tone syllable just stays low. And before another third tone, it turns into a second tone (see the tone sandhi guide).

So remember: tone 3 is first and foremost a LOW tone. The rise is secondary. Many learners exaggerate the "V" in every word, which makes their Chinese sound choppy and artificial. Think instead "low, squashed, slightly creaky voice".

This is the tone that takes longest to master. If you can produce a truly low third tone, you are halfway there.

The neutral tone (轻声 qīngshēng, "light sound") is not really a fifth tone: it is a short, weak syllable that has lost its original tone. It has no mark.

Its pitch depends on the preceding tone:

• after tone 1 → fairly low (妈妈 māma);

• after tone 2 → mid (朋友 péngyou);

• after tone 3 → rather high (你们 nǐmen, 我的 wǒ de);

• after tone 4 → very low (谢谢 xièxie, 爸爸 bàba).

It mainly appears:

• in particles: 的, 了, 吗, 呢, 吧, 着, 过;

• in suffixes: 们, 子 (桌子 zhuōzi), 头 (木头 mùtou);

• in the second syllable of reduplicated words: 妈妈, 哥哥, 看看;

• in many common words: 东西 dōngxi (thing), 衣服 yīfu, 知道 zhīdao, 喜欢 xǐhuan.

Watch out for pairs: 东西 dōngxī = "east and west"; 东西 dōngxi = "thing, stuff". The neutral tone can change the meaning!

Tone errors aren't random: they come from the habits of your native language.

Mistake 1: sentence intonation overrides tones. English rises at the end of questions and falls at the end of statements. Apply that to Chinese and a 4th tone at the end of a question turns into a rising tone, and everything collapses. In Chinese, every syllable keeps its own tone, even in a question (the question is marked by 吗 or word order instead).

Mistake 2: stress instead of tone. English speakers stress one syllable per word and reduce the others ("PHO-to-graph"). In Chinese, every syllable (except neutral ones) carries roughly equal weight. A stressed syllable often comes out as a 4th tone by accident, and unstressed ones lose their tone.

Mistake 3: tone 1 too low. Many learners pitch tone 1 in the middle of their voice, where it blurs with the start of tone 2. Tone 1 should be clearly high, almost sung.

Mistake 4: a weak tone 4. Tone 4 must fall from the top to the bottom, fast. Starting it mid-range makes it sound like a tone 3.

Mistake 5: mixing up tones 2 and 3. Both involve a rise. The difference: tone 2 rises straight away; tone 3 first dips very low. When in doubt, exaggerate the dip in tone 3.

Mistake 6: learning words without their tones. This is the number-one cause. If you memorise "mai" instead of "mǎi" (to buy) or "mài" (to sell), you will have to relearn everything later.

Isolated tones are easy. The real challenge is sequences: most Chinese words have two syllables. Teachers recommend drilling the 20 combinations (4 tones × 4 tones + 4 combinations with a neutral tone) using real words.

Take the table below and read each row aloud, slowly, then at normal speed. Record yourself, compare with Hanlu's audio, repeat. The hardest combinations are usually 2+3, 3+2, 4+1 and 1+4, because they require big pitch jumps.

Note the 3+3 combination: it is pronounced 2+3 (你好 = "ní hǎo"). This is an automatic rule explained in the tone sandhi guide.

Here is a simple, effective routine.

1. "Ma-ma-ma-ma" (1 min). Say mā má mǎ mà, then shuffled: mà mǎ mā má. Exaggerate: tone 1 very high, tone 3 very low, tone 4 falling from the top.

2. Hand gestures (2 min). Draw each tone with your hand: flat for tone 1, upward for tone 2, a dip for tone 3, a downward chop for tone 4. Gestures genuinely help your brain link melody to word.

3. Minimal pairs (3 min). Practise words that differ only by tone: 买 mǎi / 卖 mài, 汤 tāng / 糖 táng, 问 wèn / 吻 wěn, 睡觉 shuìjiào / 水饺 shuǐjiǎo.

4. Combinations (3 min). One row of the 20-combination table per day.

5. Shadowing (1 min). Play a short sentence from a Hanlu dialogue and repeat it at the same time as the audio, like an echo. This is the best way to pick up natural melody.

Above all: record yourself. You can't hear yourself properly while speaking. Comparing with native audio is often an eye-opener.