The only place the full dip-and-rise really shows up: low, then a rise at the end. This is the version textbooks draw, and it is the version you will use least.
Chinese tones
Four words. One syllable — ma — in all four. Mother, hemp, horse, to scold. The consonant and the vowel are identical; the pitch is the entire difference.
That is the thing English speakers most want to be optional about Mandarin, and it is the thing that is least optional. A tone is not an accent laid over a word. It is part of the word, the way the e in “bet” is part of “bet”.
The good news is that the system is small and completely regular: four tones, one neutral, three sandhi rules, twenty pair patterns, and you have the whole of it. Everything below can be played. Reading about pitch is not how anyone learns pitch.
The four tones, plotted and playable
Each line below is a real pitch contour, drawn from the Chao tone-letter numbers underneath it — 5 is the top of your comfortable range, 1 the bottom. Pick a tone to hear it, and switch the syllable to hear the same four shapes on a different word. That swap is the point: the contour is the tone, not the word.
Hear the tones on
And the fifth: the neutral tone
It has no contour of its own. It is short, quiet, and its pitch is inherited from the syllable in front of it — low after a first tone, higher after a fourth. Do not give it a pitch on purpose.
Read the notation and the picture together and the whole system falls out in about a minute. First tone, 55: flat along the top, held. It should feel unnaturally sustained, almost sung — that is the mistake to make, because learners reliably let it sag.
Second tone, 35: starts in the middle and is still climbing when the syllable ends. If your rise arrives early and levels off, a listener hears something between a second and a first. Fourth tone, 51: top to bottom, fast, and short — it is the briefest of the four, and a polite, gentle fall is the single most common way to under-produce it.
Third tone, 21(4) is the one everybody is taught wrong, and it has its own section below. The bracket in the notation is doing real work: the final rise is optional and usually absent.
| Tone | Mark | No. | Chao | Shape |
|---|---|---|---|---|
| First tone | ā | 1 | 55 | high and flat |
| Second tone | á | 2 | 35 | rising |
| Third tone | ǎ | 3 | 21(4) | low |
| Fourth tone | à | 4 | 51 | sharp fall |
| Neutral tone | a | 5 | — | short and weightless |
Tone marks, tone numbers, and how to type them
Two notations, one language. Marks (mā) are what dictionaries, textbooks and subtitles use. Numbers (ma1) are what keyboards and search boxes accept. Learn to read the marks; use the numbers to type.
Type the tone as a digit after each syllable. 5 (or no digit) is the neutral tone. v and u: both stand in for ü, the way a pinyin keyboard expects.
nǐ hǎo ma
- No tones
- ni hao ma
- Zhuyin
- ㄋㄧˇ ㄏㄠˇ ˙ㄇㄚ
Where the mark goes
hao3hǎo
an `a` is present, so it takes the mark
xie4xiè
no `a`, so the `e` takes it
gou3gǒu
`ou` is the exception — the `o` takes it
jiu3jiǔ
no a/e/ou, so the last vowel takes it
lv4lǜ
`v` is how you type ü on a pinyin keyboard
nv3nǚ
same, and ü keeps its dots under the mark
The mark sits on a vowel, so you read the tone at the same instant you read the syllable — that is the whole advantage over numbers, and it is a large one. A tone number sits after the syllable, which means your eye finishes reading ma before it learns what pitch to put on it. That half-beat is enough to keep tone as an afterthought, and tone as an afterthought is exactly the habit that fossilises.
Numbers earn their place in three situations: typing on a keyboard with no way to enter ā, searching (most dictionaries accept hao3), and writing tones down quickly by hand. Some people also find them easier for a first pass at a new word. None of that is a reason to read in numbers.
The one real trap is the neutral tone. It is written with no mark at all, which is visually identical to “nobody bothered to mark the tone”. 妈妈 māma and 妈 mā are not the same vowel treatment; the second ma is short, quiet and pitched by what came before it. When you see an unmarked syllable inside a word, that is information, not an omission.
| Plain | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| a | ā | á | ǎ | à |
| o | ō | ó | ǒ | ò |
| e | ē | é | ě | è |
| i | ī | í | ǐ | ì |
| u | ū | ú | ǔ | ù |
| ü | ǖ | ǘ | ǚ | ǜ |
The third tone is not a dip
Textbooks draw it as a valley — falling, then rising — and beginners dutifully perform the whole rollercoaster on every syllable. In connected speech the third tone is almost always just low. Hear the three cases side by side.
The half-third, and by far the most common. It is simply low — a flat, slightly creaky rumble at the bottom of your range, with no rise at all. Roughly two thirds of the third tones you say will be this.
It rises, becoming indistinguishable from a second tone. Native speakers do this without noticing and the written pinyin is never changed to match.
Count the environments and the arithmetic is stark. The full dip needs the syllable to be stressed and final and not followed by another third tone. Everywhere else you get the half-third — a flat, slightly creaky rumble at the bottom of your range with no rise at all — or the sandhi rise. The version you were drilled on is the rarest of the three.
The practical fix is a relabelling, not an exercise. Call it “the low tone” in your head. Learners who make that one change stop producing the bouncing, sing-song third tone that marks a beginner from across a room, and they stop stretching syllables to fit a contour that was never going to be there.
There is a diagnostic worth running on yourself: say 我很好 at conversational speed. If it takes you noticeably longer than “I’m fine” does in English, you are performing three dips where a native speaker produces one rise and two low syllables.
Tone sandhi: three rules that cover almost everything
Tones shift when they collide, in ways that are entirely predictable and that the written pinyin never records. Toggle between what is written and what is spoken, and play each one to check.
3 + 3 → 2 + 3
Two third tones in a row are physically awkward — you cannot sit at the bottom of your range twice without the word collapsing. The first one rises instead, becoming indistinguishable from a second tone.
The pinyin is never rewritten — dictionaries, textbooks and this page all still print nǐ hǎo. The change is a fact about speech, not about spelling, and native speakers are not aware they are doing it. With three in a row the grouping decides: 我很好 goes wǒ hén hǎo, because 很好 is the unit.
Sandhi looks like an extra layer of rules to memorise and is really the opposite: it is your mouth taking the path of least resistance, and it happens to you whether or not you know the rule. Two third tones back to back are genuinely awkward to produce. Two fourth tones back to back are tiring. The language has smoothed them out, and if you say the words fast enough you will smooth them out too.
So the rules are worth knowing for listening more than for speaking. You will hear ní hǎo and need to recognise the word written nǐ hǎo. You will hear bú shì and need to find 不是 in your memory. Without the rules, a chunk of what you have carefully memorised simply does not match what reaches your ears.
Drill the 20 tone pairs, not the 4 tones
Mandarin words are overwhelmingly two syllables, so the melody your mouth produces is a pair — and there are only twenty combinations. Play a single cell, or play a whole row or column to hear one tone held constant while the other moves. Every word here is HSK 1–3 vocabulary.
| 1st ↓ / 2nd → | |||||
|---|---|---|---|---|---|
This is the single highest-leverage change most learners can make to how they practise. Drilling mā má mǎ mà on a loop rehearses a unit that hardly occurs: isolated single syllables are rare in real Mandarin. The pair is what your mouth will actually have to execute, and a pair is not the sum of its parts — 2-1 (明天 míngtiān) and 1-2 (中国 Zhōngguó) feel like completely different shapes to produce.
Anchor each cell to a word you already like. Once 中国 is your 1-2 anchor, every new 1-2 word inherits its melody — “ah, it’s a 中国-shaped word” — and you have stopped computing tones one syllable at a time. Four or five solid anchors will carry most of your vocabulary.
Then reverse the direction. When a flashcard shows 希望, say it aloud with tones before you reveal the pinyin. Tone knowledge that only works when you can see the answer is not pronunciation; it is trivia about pronunciation.
- Weeks 1–2 · listen before you speak. Work the ear trainer below daily. Perception has to lead production — you cannot reliably say a contrast you cannot yet hear, and trying is how a wrong tone gets grooved in.
- Weeks 2–3 · shadow, exaggerated. Play a pair, repeat it immediately, overshooting the contour: higher highs, lower lows, sharper falls than feel polite. Recorded learners undershoot almost without exception, so exaggeration lands about right.
- Week 4 · take it to sentences. Everyone’s tones are fine on single words and dissolve at sentence speed. Shadow whole sentences slowly, keeping the contours intact. Speed is the last thing to add, not the first.
- Forever · never learn a word without its tone. This is the habit that decides everything. It costs seconds per word now and saves relearning thousands of words later.
Minimal pairs: can you actually hear it?
Two words, identical consonants and vowels, different tone. There is nothing to fall back on — you either heard the pitch or you guessed. Ten pairs, and the ones you miss are your daily drill list.
Ear trainer · pair 1 of 10
0 / 0 correct
One of the two below. Listen, then pick.
The full list, to play at your own pace
The classic, and a genuinely costly one to get wrong in a shop. Third versus fourth: low and level against a hard drop.
请问 (qǐngwèn, “excuse me”) is the single most useful phrase you will say to a stranger. Mis-tone it and you have said something else entirely.
First versus second — flat against rising. On a menu, the difference between a starter and a dessert.
A two-syllable pair: 3-3 against 4-4. Ordering “I want to sleep” at a dumpling stall is the standard learner story for a reason.
The same two tones in the other order — 2-1 against 1-2. Proof that the pair, not the individual tone, is what carries the word.
First against fourth. The widest contrast of the four and the first one most learners get reliably.
你姓什么 (“what’s your surname?”) is asked of you constantly. Worth having solid.
3-3 against 3-4, and note that 老板 is really spoken láobǎn — sandhi and a minimal pair at the same time.
3-neutral against 4-4. Both syllables move, which is why this one stays hard long after the single tones are easy.
Only the second syllable moves — neutral against fourth — and the characters differ by one radical. This is the level of contrast you are training for.
The four mistakes that fossilise
Every one of these is reversible early and expensive later, because a wrong tone that has been repeated a thousand times has to be unlearned before it can be relearned.
Learning tones as trivia
Storing “xīwàng, tones 1-4” as a fact rather than a sound. It passes flashcard review and fails the moment you open your mouth. If you cannot sing the word, you do not have it — say every new word aloud before you rate the card.
Dropping tones at sentence speed
Tones survive the single word and dissolve in the sentence, because attention goes to grammar and vocabulary and there is none left over for pitch. The fix is slow shadowing of full sentences with the contours kept intact; speed added afterwards keeps them.
Putting English intonation on top
Raising your pitch at the end of a question will silently overwrite the final tone of the sentence. Mandarin marks questions with words — 吗, 呢, or a verb-not-verb pattern — not with melody. Let the particle do the work and leave the pitch alone.
Waiting until later to “add tones”
The most expensive of the four, and the most common plan. There is no later: a word learned without its tone is a word learned wrong, and every review reinforces it. Learning the tone with the word costs a couple of seconds each time.
Questions people ask
- How many tones does Mandarin Chinese have?
- Four tones plus a neutral tone. The first is high and flat (mā), the second rises (má), the third is low (mǎ), and the fourth falls sharply (mà). The neutral tone (ma) is short and unstressed and takes its pitch from the syllable before it. Some counts say five and some say four — they are describing the same system, differing only on whether the neutral tone counts as a tone of its own.
- What are the four tones of Mandarin with examples?
- Using the syllable ma: 妈 mā (first, high and flat) means mother; 麻 má (second, rising) means hemp; 马 mǎ (third, low) means horse; 骂 mà (fourth, falling) means to scold. The segments are identical in all four — the pitch is the entire difference between the words.
- Is the third tone really a dip?
- Rarely. Textbooks draw the third tone as a fall-then-rise, but in connected speech it is almost always just low: a flat, slightly creaky rumble at the bottom of your range. The full dip-and-rise appears mainly when the syllable is stressed and final, or said in isolation. Thinking of it as “the low tone” fixes more third-tone problems than drilling the textbook contour does.
- What is tone sandhi in Chinese?
- Tone sandhi is a tone changing because of the tone next to it. Three rules cover everyday Mandarin: two third tones in a row become second-plus-third (nǐ hǎo is spoken ní hǎo); 不 bù rises to bú before a fourth tone (不是 is bú shì); and 一 yī becomes yí before a fourth tone and yì before the others (一个 yí ge, 一天 yì tiān). The written pinyin is not changed to match.
- Should I learn tone marks or tone numbers?
- Learn to read tone marks — they are what dictionaries, textbooks and subtitles use, and the mark sits on the vowel so you read the tone at the same moment you read the syllable. Tone numbers (ma1, ma2) are a typing convenience, useful for entering pinyin on a keyboard and for search. Knowing both takes an afternoon; only the marks need to be automatic.
- Do Chinese people actually notice if I get tones wrong?
- Yes, in the same way an English speaker notices if you swap a vowel. Context often rescues a single wrong tone in a sentence, which is why beginners get away with more than they expect, but tones are not an accent laid over the word — they are part of the word. Wrong tones cost the listener effort, and enough of them make you genuinely hard to follow.
- How long does it take to learn Chinese tones?
- Hearing the four tones reliably takes days to a few weeks of focused listening. Producing them correctly on single words takes a few weeks more. Keeping them intact in fluent sentences takes months and keeps improving for years. The uneven timeline is normal — the failure mode is not slowness, it is learning words without their tones and having to relearn them later.
- What is the best way to practise Chinese tones?
- Drill tone pairs rather than single tones. Mandarin words are mostly two syllables, so the melody your mouth actually produces is one of twenty pair patterns, and practising a lone mā never rehearses that. Pair the drilling with minimal pairs (mǎi vs mài) for the ear, and shadow real audio out loud, exaggerating the contours — recorded learners consistently undershoot.
Hear every word before you memorise it
A tone you only ever read is a number you have memorised. Every word and example sentence in MandarinAI comes with audio, and the review asks you to say it before it shows you the pinyin — so tones stay attached to sounds from the first repetition rather than being bolted on later. The free plan keeps your deck forever.
Keep going
- The full pinyin chart — every syllable Mandarin actually uses, with the tones applied. Tones are half of pronunciation; this is the other half.
- Bopomofo (zhuyin) — Taiwan’s phonetic script, which writes the same four tones with different marks and puts the neutral tone first.
- The beginner’s roadmap — with tones settled, the rest of the pronunciation system is small and the road ahead is mostly vocabulary.
- Graded readers — where tones stop being an exercise and start being how you hear a sentence.