Pinyin: Tones, Pronunciation and Spelling Rules

Pinyin is a spelling system for the sounds of Mandarin. Each syllable of spoken Chinese gets one written form, and that form carries the tone as a diacritic. It was adopted in mainland China in 1958 and became the international standard for romanizing Chinese in 1982, which is why airports, dictionaries, phone keyboards and passport names all over the world use it.

What pinyin is not: it is not a replacement for Chinese characters, and it is not a phonetic alphabet you can read with English letter values. 汉字 (Hànzì) are the writing system; pinyin is the pronunciation annotation on top of it. A Chinese newspaper is printed in characters, not in pinyin. Pinyin appears underneath, beside, or above the characters when someone needs to know how they sound.

This distinction matters for how you learn. You can read a page of pinyin out loud without knowing what a single word means, and you can understand a sentence of characters without being able to pronounce it. The two skills are separate, and pinyin only trains the second one in one direction: characters to sound.

A Mandarin syllable is built in three layers, and pinyin spelling encodes all three.

The initial (声母, shēngmǔ) is the opening consonant. Mandarin has 21 of them, and a syllable may also start with no consonant at all — 爱 ài, 安 ān, 二 èr. In pinyin spelling a syllable with no initial simply starts with a vowel letter, and that is your signal that there is nothing to pronounce before the vowel.

The final (韵母, yùnmǔ) is the vowel part, and it can be a single vowel (a, o, e, i, u, ü), a vowel plus a nasal ending (an, ang, en, eng, in, ing, ong), or a compound (ai, ei, ao, ou, ia, iao, uai, and so on). Finals are the part learners under-rehearse. An initial said wrong is usually still recognizable; a final said wrong changes which word you said.

The tone (声调, shēngdiào) is the pitch contour carried across the whole syllable. It is not a decoration and it is not optional. In Mandarin, tone is part of the word in the same way that a consonant is: māi is not a word, mǎi means to buy, mài means to sell, and no listener will confuse the two.

• Initial: the opening consonant, or nothing.
• Final: the vowel part, including any nasal ending.
• Tone: the pitch shape, written as a mark over one vowel.

The standard description is four tones plus a neutral tone. The four are written with marks; the neutral tone is written with nothing.

First tone is high and level. Think of holding a single musical note without letting it drift. It is the tone English speakers find easiest to produce and easiest to forget to sustain — the pitch tends to sag halfway through.

Second tone rises. It starts in the middle of your range and climbs. The closest English analogue is the intonation of a surprised "what?" — but shorter, and without the vowel change.

Third tone is the one that gets caricatured. It drops to the bottom of your range, stays low, and then comes back up slightly. In connected speech it is often just the low part with no rise at all, which is why learners who have practiced the full dip-and-rise in isolation sound wrong in sentences.

Fourth tone falls sharply from high to low. It is short and decisive. The English analogue is a curt command syllable, but the fall needs to be steeper and faster than English intonation normally goes.

The neutral tone is short, flat and unstressed. It takes its pitch from the syllable before it rather than having one of its own. It appears constantly: the second syllable of 谢谢 xièxie, the second of 朋友 péngyou, the particle 吗 ma, the 子 in 名字 míngzi.

• First tone: 妈 mā, mother. High, flat, sustained.
• Second tone: 麻 má, hemp. Rising.
• Third tone: 马 mǎ, horse. Low, then a slight rise.
• Fourth tone: 骂 mà, to scold. Sharp fall.
• Neutral tone: 吗 ma, question particle. Short, no mark.

Only one vowel in a syllable carries the mark, and the choice is not arbitrary. There is a fixed priority order, and once you know it you never have to guess again.

The order is: a first, then o, then e. If the final contains an a, the mark goes on the a — 好 hǎo, 快 kuài, 小 xiǎo. If there is no a but there is an o, the mark goes on the o — 我 wǒ, 多 duō. If there is no a and no o but there is an e, the mark goes on the e — 写 xiě, 谢 xiè.

If the final contains none of a, o or e, the mark goes on the last vowel of the pair iu or ui — 六 liù, 会 huì, 对 duì. When i and u appear together without a third vowel, this rule handles both cases.

Two smaller rules finish the job. A single i keeps its dot when a tone mark is added, so it is written 你 nǐ rather than replacing the dot, and a lone ü keeps its umlaut unless the initial is j, q, x or y — after those four, ü is written as plain u. That is why 女 nǚ keeps the umlaut but 去 qù does not, even though both are the same vowel.

The practical consequence: if you are ever unsure where the accent belongs, scan the syllable for a, then o, then e, then take the last vowel. That covers every syllable in the language.

This is where most beginners lose a month. Pinyin uses the Latin alphabet, but it assigns several letters values that English does not use, and reading them with English instincts produces words that no Chinese speaker recognizes.

c is "ts", never "k" and never "s". 菜 cài is "tsai", not "kai". The letter is always aspirated: a puff of air follows it.

q is not "kw". It sits close to English "ch", but with the tongue flat and forward behind the lower teeth rather than curled back. 七 qī is close to "chee" said with a tight, forward tongue.

x is not "ks". It sits close to English "sh", again with the forward tongue position. 西 xī is close to "she" with lips spread.

zh, ch, sh are the curled-back versions of j, ch, sh. The tongue tip lifts toward the roof of the mouth. 知 zhī, 吃 chī, 是 shì — all three involve a tongue position English does not use for these sounds.

z, c, s are the flat, forward versions, made with the tongue behind the upper front teeth. 字 zì, 次 cì, 四 sì.

r is the hardest single letter. It is not the English r. It is a soft, voiced, buzzing sound closer to the s in "measure" or the ge in "rouge", with the tongue curled back.

ü is a vowel English does not have: say "ee" and then round your lips without moving your tongue. 女 nǚ, 绿 lǜ, 去 qù.

• c → ts (菜 cài)
• q → ch, tongue forward (七 qī)
• x → sh, tongue forward (西 xī)
• zh ch sh → curled-back j ch sh (知 吃 是)
• z c s → flat ts-like series (字 次 四)
• r → soft buzzing, not English r (日 rì)
• ü → ee with rounded lips (女 nǚ)

Two tone changes are obligatory in normal speech, and both are invisible in the spelling. The written tone stays as it is; the spoken tone shifts.

The first is third-tone sandhi. When two third tones occur in sequence, the first becomes a second tone. 你好 is written nǐ hǎo and spoken ní hǎo. 很好 is written hěn hǎo and spoken hén hǎo. With three third tones in a row the grouping depends on the words, not on a simple left-to-right rule, which is why 我也很好 takes practice rather than arithmetic.

The second is the behaviour of 不 bù and 一 yī. 不 is fourth tone in isolation, but before another fourth tone it becomes second: 不是 is spoken bú shì, 不要 is spoken bú yào, 不错 is spoken bú cuò. Before first, second or third tone it stays fourth: 不吃 bù chī, 不行 bù xíng, 不好 bù hǎo.

一 yī behaves similarly but less simply. Before a fourth tone it becomes second: 一个 yí ge, 一样 yí yàng. Before first, second or third tone it becomes fourth: 一天 yì tiān, 一年 yì nián, 一起 yì qǐ. When it means the number one itself — as in a count, or at the end of a number — it stays first tone: 第一 dì yī.

The point for a reader is not to memorize exceptions but to stop trusting the written tone as a description of the sound. Written tone is the dictionary form. Spoken tone is what the language actually does, and it follows rules.

Pinyin has three everyday jobs, and knowing which one you are looking at tells you how much precision to demand.

The first is input. Chinese phone and computer keyboards are pinyin keyboards: you type the letters, the software offers characters, you pick one. This is why pinyin has to be unambiguous enough to be typed — and why typing fluently is a different skill from pronouncing well. Plenty of people type fast with mediocre tones.

The second is annotation. Children learn to read with pinyin printed above the characters. Dictionaries list the pinyin beside every entry. Language-learning subtitles place pinyin above or below the Chinese line so a learner can follow the sound while seeing the characters.

The third is naming. Street signs, place names, brand names and personal names in passports are romanized with pinyin. This is the one context where the tone marks are usually dropped, because tone marks are awkward to print on signage and in databases. The result is a pinyin string that has lost its most important information, which is exactly why a sign saying BEIJING does not tell you how to say it.

When you meet pinyin with no tone marks at all, treat it as a spelling hint rather than a pronunciation guide. It tells you which letters, not which pitch.

Mainland China uses pinyin. Taiwan uses zhuyin fuhao (注音符号), also called bopomofo, a set of symbols derived from ancient character components rather than from Latin letters. The two systems describe the same sound system and the same tones; they spell it differently.

Zhuyin writes each initial and final as its own symbol, and it places the tone mark to the right of the syllable rather than above a vowel. Because there are no Latin letters involved, zhuyin does not inherit the English letter-value problem: a learner of zhuyin never has to unlearn reading c as "k".

In practice, if you are learning Mandarin and your materials use pinyin, stay in pinyin. Mixing the two systems early is more confusing than useful, because you end up with two spellings for every word and no automatic mapping between them. If you later need zhuyin — for Taiwanese materials, or for typing on a Taiwanese keyboard — it is a small additional step once the sounds themselves are solid.

Pinyin is a small system. The total inventory is 21 initials, roughly three dozen finals, and five tones. The reason people spend months on it is not the size of the inventory but the absence of feedback: you can produce a wrong tone indefinitely without anyone correcting you, because a wrong tone usually still gets a polite nod.

Start with tones in isolation, on one syllable, until you can hit all four on demand without thinking about it. Use a syllable whose meaning you do not care about, so you are listening to pitch rather than to vocabulary.

Then do minimal pairs. mā má mǎ mà is the standard drill, but it is more useful to drill pairs that actually collide in meaning: 买 mǎi to buy against 卖 mài to sell; 问 wèn to ask against 闻 wén to smell; 四 sì four against 死 sǐ to die. When the two words are ones you would actually say, the tone stops being an exercise.

Then read pinyin aloud from a text you have not seen, out loud, before you know the meaning. That is the skill the annotation is for: converting letters to sound without a dictionary in the loop.

Finally, check yourself against real speech rather than against your own memory. Listen to a Chinese sentence you can also read in characters, and match your reading to what the speaker does with pitch. The gap between your reading and their speech is the exact list of things to fix.

• Drill the four tones on one syllable until they are automatic.
• Drill meaning-bearing minimal pairs, not just mā má mǎ mà.
• Read unseen pinyin aloud without looking up meaning.
• Compare your reading against real speech, not against memory.