The four tones, explained (and why yours are probably wrong)

By Carson 木林森 · August 2026

妈, 麻, 马, 骂. Same letters if you strip the marks off, and four completely different words: mom, hemp, horse, to scold. The only thing that changed is a rise-or-fall in pitch across the syllable — the tone. Most beginner pages introduce the four tone marks and move on; the two mistakes that actually stall people are learning the marks as static accents instead of ear-trained pitch shapes, and never meeting the sandhi rules that quietly change how tones sound once real words run together.

A tone is a pitch shape, not an accent mark

Write the tone as a small mark over the vowel and the mark's shape pictures the pitch: ā is high and flat (1st), á rises (2nd), ǎ dips low then rises (3rd), à falls sharply (4th), and an unmarked a is neutral — short and light, with no shape of its own, as in the second syllable of 妈妈 māma. The marks are a memory aid for something you have to train by ear first: 妈 mā "mom" and 骂 mà "to scold" are the same syllable with opposite pitch, and mixing them up isn't a foreign accent, it's a different word.

A second four-way family makes the same point: 汤 tāng (soup), 糖 táng (sugar), 躺 tǎng (to lie down), 烫 tàng (scalding hot). Ordering soup and saying "scalding" instead is a tone mistake, not a pronunciation quirk — which is exactly why tone has to be trained as sound before it's memorized as a mark.

The rule almost every beginner page skips: third-tone sandhi

Real speech strings syllables together, and a few combinations shape-shift when they touch. The most common one, by far: when a 3rd tone is immediately followed by another 3rd tone, the first one turns into a rising 2nd tone in speech. 你好 is written nǐ hǎo but sounds like ní hǎo — the most-said phrase in the language, and most learners never get told why the pinyin lies. Dictionaries and textbooks keep the underlying 3rd-tone mark; only the sound changes.

不 and 一 change tone depending on what follows

Two of the most frequent words in Mandarin shape-shift on their own schedule:

不 bù is 4th tone (falling) on its own, but rises before another 4th tone: 不是 sounds like bú shì. Before anything else it stays falling: 不好 bù hǎo.

一 yī is 1st tone (flat) alone. Before a 4th tone it rises: 一个 yí ge. Before tones 1, 2 or 3 it falls: 一天 yì tiān.

Both are sound changes only — the character on the page never moves, which is exactly why a page of tone marks alone can't teach them. You have to hear the pair, not read it.

Tone pairs: the actual rhythm of Chinese

Most Chinese words are two syllables, so real listening and speaking runs on tone pairs — the melody of syllable 1 plus syllable 2 — not on isolated single tones. Native speech is a stream of these two-note tunes. Nine common words, and the pair each one drills:

One more detail most learners never hear named: a 3rd tone before tones 1, 2 or 4 only does its low dip, with no rise at the end — the "half-third." Listen for it in 老师 lǎoshī and 米饭 mǐfàn. Before another 3rd tone, it's the sandhi rule above instead: the first one rises, as in 你好.

From GraphChinese's data: tones aren't a warm-up paragraph before "real" lessons start — they're three explicit topics inside the 19-topic HSK 0 foundation layer at the base of a 829-topic graph: the four tones by ear, the sandhi rules above with real audio, and a tone-pair drill. Every exercise plays a recording first and asks you to name or type what you actually heard, because tone is trained by ear, not memorized as a diacritic.

Why this is worth getting right before you have vocabulary

Toneless pronunciation habits fossilize with repetition exactly the way correct ones do, and unlearning a wrong habit costs more than building the right one from the start would have. Training tones while you still have almost no vocabulary to be wrong about means every word you learn afterward arrives with its tone already attached, instead of needing to be retrofitted with one months or years later. See the full beginner roadmap for why tones are rung one, and the HSK levels guide for what "HSK 0" covers before HSK 1 even starts.

Frequently asked questions

What are the four tones in Chinese?

A pitch shape attached to each syllable, not decoration on top of it: 1st is high and flat (mā), 2nd rises (má), 3rd dips then rises (mǎ), 4th falls sharply (mà), plus a short, toneless neutral tone. Same letters, different pitch, different word — get it wrong and you've said a different word, not mispronounced the right one.

What is third-tone sandhi (why does 你好 sound like ní hǎo)?

When a 3rd tone is immediately followed by another 3rd tone, the first turns into a rising 2nd tone in actual speech, though it's still written with its dictionary 3rd-tone mark. 你好 is written nǐ hǎo but sounds like ní hǎo — the most common sandhi rule in the language.

Do 不 (bù) and 一 (yī) change tone?

Yes. 不 is 4th tone alone but rises before another 4th tone (不是 → bú shì), staying falling elsewhere (不好 bù hǎo). 一 is 1st tone alone but rises before a 4th tone (一个 → yí ge) and falls before tones 1, 2 or 3 (一天 → yì tiān). Both are sound changes only; the character never changes.

Can I skip tones now and fix them later?

Not cheaply — toneless habits fossilize the same way correct ones do, and unlearning takes longer than learning right the first time. Training your ear on tones before you have much vocabulary means every new word arrives with its tone already attached.

Related guides: How to learn Chinese from absolute zero · The HSK levels, explained · How long it actually takes · The comment + question method