What you will be able to do
- Hear and produce the aspiration contrast — b/p, d/t, g/k, z/c, zh/ch, j/q — without relying on voicing.
- Place the tongue for all three sibilant rows: dental z c s, retroflex zh ch sh r, palatal j q x.
- Read the empty rime correctly in zi ci si and zhi chi shi ri.
- Produce r as a motionless voiced sh, with no rounding, no tap and no uvula.
Chinese words in this text can be activated to show a dictionary gloss.
Three rows, one system
Aspiration is the axis, not voicing
English and French separate b from p by whether the vocal cords are working. Mandarin does not. Both b and p are voiceless; the difference is the puff of air.
- Unaspirated: b, d, g, z, zh, j — the sound is released cleanly, with no breath after it.
- Aspirated: p, t, k, c, ch, q — a strong puff follows the release.
bà and pà differ only in that puff. Hold a strip of paper an inch in front of your lips: it must jump on pà and stay still on bà. A French speaker whose p carries no puff will be heard as saying ; an English speaker who voices the b of at the start of a word will usually be understood, but the p at the top of the pair is where the real information sits.
Three places for one kind of noise
Where English has a single s and a single sh, Mandarin has three rows, each built the same way: unaspirated affricate, aspirated affricate, fricative.
Row 1 — dental: z c s. The tip of the tongue is at the back of the upper teeth. z is ts with no puff, c is ts with a hard puff, s is a plain hiss. zì, cì, sì.
Row 2 — retroflex: zh ch sh r. The tip of the tongue curls up and back toward the front of the hard palate, leaving a narrow channel. zhī, chī, shì. This row has a fourth member: r is sh with the vocal cords switched on and nothing else changed. rén, rì, rè.
Row 3 — alveolo-palatal: j q x. The exact opposite gesture. The tip of the tongue rests down, against the back of the lower teeth, while the blade rises to the hard palate. jī, qī, xī.
Say the ladder down each column and the system becomes audible: zī zhī jī — cī chī qī — sī shī xī. Nothing changes in the ladder except where the tongue is.
The spelling cannot mislead you
This is the fact that saves you, and few textbooks state it plainly. j, q and x occur only before i and ü. z c s and zh ch sh r never occur before either. The two groups are in complementary distribution, so the letter that follows always tells you which row you are in.
- ji, jia, jie, jiu, jian, jiang, ju, jue, juan, jun — always row 3
- zhi, zha, zhe, zhu, zhuan, zhong — always row 2
- After j, q, x and y, the letter u is always ü: jū is jü, qù is qü, xū is xü, yú is yü. The two dots are dropped because nothing could be confused; after n and l they are kept because nǚ and nú are different words.
The empty rime
After z, c, s, zh, ch, sh and r, the letter i is not the vowel of see. It is the consonant itself held on, with the voice added: zi, ci, si, zhi, chi, shi, ri. Say a long s, then switch the voice on without moving anything — that is sì. A learner who reads shì as she has produced a different syllable of Mandarin, not an accent.
r, specifically
Mandarin r is the one initial that three different first languages get wrong in three different directions.
- English speakers round the lips and use almost no friction, which makes it too soft. Keep the lips flat.
- French speakers go to the uvula. The tongue must be at the front of the palate, not at the back of the throat.
- Arabic speakers tap or trill. Mandarin r does not move at all: the tongue sets in the sh position and holds.
Test it directly: say a long sh, then hum through it without moving the tongue. That is r. rènshi contains both members of the pair, one after the other.
Four tests you can run alone
- Paper test — aspiration. Paper jumps on p, t, k, c, ch, q; stays still on b, d, g, z, zh, j.
- Mirror test — say zhī then jī. The jaw and lips must not move; only the tongue travels from back to front. If your lips round on zhī, an English or French r has crept in.
- Finger test — for j, q, x, the tip of your tongue should be reachable behind the lower teeth. If it is up, you are producing row 2.
- Hum test — for r, a long sh with the voice added, and nothing else changed.
A three-minute daily protocol
Ladder first: zī zhī jī, cī chī qī, sī shī xī, three times. Then six minimal pairs with a coin toss to decide which you say, recorded and played back: zhī/jī, chī/qī, shī/xī, zì/zhì, cì/chì, sì/shì. Then six words at speaking speed: zhīdào, chīfàn, shíjiān, rènshi, qǐlái, xuéxí. Three minutes a day for two weeks will do more than an hour once.
How to work through it
Read the ladder, not the listread
Read the three rows as a grid rather than as fourteen separate letters. Then say the ladder down each column three times: zī zhī jī, cī chī qī, sī shī xī. Only the tongue position changes between the columns.
Six pairs, twice eachaudio
Listen to zhī/jī, chī/qī, shī/xī, zì/zhì, cì/chì, sì/shì and commit to an answer before the second playing. The pair you keep failing names the row your first language does not have.
The paper and the mirrorspeaking
Run test one on 爸 bà against 怕 pà, 大 dà against 他 tā, 该 gāi against 开 kāi. Then run test two on zhī and jī. Both tests are visual, which is why they work without a teacher in the room.
Words chosen for their consonantsvocabulary
Eight words picked for what they make your tongue do. Say each five times before you think about its meaning; the meaning is already in your notes, the gesture is not.
Three minutes, every dayspeaking
Set a timer and run the protocol: ladder, pairs, words. Record the last thirty seconds and listen once. Do not repeat the recording until tomorrow — the value is in the daily return, not in the length of the session.