Pronunciation guide
Chinese (Mandarin) pronunciation guide
Mandarin has a small, strict syllable inventory and four tones, which means pronunciation is a closed system you can learn exhaustively — roughly 400 syllables times four tones — but every one of those tones carries meaning, so there is nowhere to hide.
How the Chinese (Mandarin) sound system works
Mandarin syllables have a rigid structure: an optional initial consonant, an optional glide, a vowel, and a very limited set of finals (only n, ng and r can close a syllable). There are around 400 distinct syllables ignoring tone, and about 1,300 with tone — a genuinely finite set. The consonant system contrasts unaspirated and aspirated stops rather than voiced and voiceless, which is why *b* and *p* in pinyin are both voiceless to an English ear.
The 5 sounds English speakers get wrong
These are ordered by how much damage they do. The first two or three are worth more of your time than everything else on this page combined, because they are the ones that change meaning rather than merely signalling an accent.
the x/j/q series
xie, jia, qi
Why it goes wrong. Alveolo-palatals. English speakers substitute "sh", "j" and "ch", merging them with the zh/ch/sh series.
How to fix it. Tongue blade flat and high against the hard palate, tip behind the lower teeth. Contrast *xi* against *shi* directly.
the zh/ch/sh/r series
zhong, chi, shi
Why it goes wrong. Retroflex. English "sh" sits between these and the x/j/q series, so it sounds wrong for both.
How to fix it. Curl the tongue tip up and back. These are the sounds English speakers most often flatten.
the ü vowel
nü, lü, jué
Why it goes wrong. Merged with /u/. Written u after j, q, x, y, which hides it.
How to fix it. /i/ with rounded lips. Note *ju* is really *jü*, and *nu* and *nü* are different words.
the syllabic finals of zhi, chi, shi, ri, zi, ci, si
shi, zi, ri
Why it goes wrong. Learners read the i as /i/. It is not — it is a syllabic continuation of the consonant.
How to fix it. Say the consonant and simply voice it; do not add a vowel. *Shi* has no "ee" in it.
tone 3 and tone sandhi
nǐ hǎo, hěn hǎo
Why it goes wrong. Learners produce a full dipping tone in isolation and never learn that two third tones in a row change the first to second tone.
How to fix it. In connected speech tone 3 is usually just a low tone, not a full dip. And *nǐ hǎo* is pronounced *ní hǎo*.
Stress, rhythm and tone
Four lexical tones plus a neutral tone, and tone is not decoration — *mā*, *má*, *mǎ*, *mà* are four unrelated words. Tone sandhi rules change tones in context, most importantly the third-tone rule and the special behaviour of 一 and 不. Mandarin is syllable-timed with roughly even syllable duration, and the neutral tone syllables are shortened and reduced.
What the writing system does and does not tell you
Characters carry meaning, not pronunciation — roughly 80% are phono-semantic compounds giving a hint of sound, but the hint is unreliable after centuries of sound change. Pinyin is the standard romanisation and is fully systematic, but it uses letters in ways that mislead English readers: q is /tɕʰ/, x is /ɕ/, c is /tsʰ/, and the vowel written *e* varies by context. Learning to read pinyin correctly is a distinct skill from learning to read the Latin alphabet.
Recurring mistakes
- Merging the retroflex and alveolo-palatal series.
- Reading pinyin with English letter values.
- Producing tones correctly in isolation but flattening them in sentences.
- Ignoring tone sandhi.
- Adding a vowel to the syllabic finals.
A four-week practice plan
- Week 1 — Tones in isolation, then in two-syllable combinations. All sixteen tone pairs, systematically.
- Week 2 — The two sibilant series, contrasted: *xi/shi*, *ji/zhi*, *qi/chi*.
- Week 3 — Pinyin traps: ü after j/q/x/y, the syllabic finals, and the several values of e.
- Week 4 — Sandhi in running speech, especially third-tone sequences and 一/不.
Teaching or learning Chinese (Mandarin) with Teachee
Teachee supports Chinese (Mandarin) as a lesson language
(zh-CN, zh-TW). During a lesson you
record a sentence, and it comes back transcribed, translated and — where
the phonetic data exists — with a transcription your student can read.
- Translation
- Text-to-speech voices
- IPA word dictionary (not currently available for Chinese (Mandarin))
Where automatic phonetics are not available, the recording still is — and for a language like Chinese (Mandarin) a teacher’s own recording of a word, attached to that word, is worth more than any generated transcription.
Frequently asked questions
Is Chinese (Mandarin) hard to pronounce for English speakers?
Among the hardest. The specific difficulties are the x/j/q series, the zh/ch/sh/r series, the ü vowel — not the language as a whole. Mandarin has a small, strict syllable inventory and four tones, which means pronunciation is a closed system you can learn exhaustively — roughly 400 syllables times four tones — but every one of those tones carries meaning, so there is nowhere to hide.
Does Chinese (Mandarin) spelling tell you how to pronounce a word?
Characters carry meaning, not pronunciation — roughly 80% are phono-semantic compounds giving a hint of sound, but the hint is unreliable after centuries of sound change. Pinyin is the standard romanisation and is fully systematic, but it uses letters in ways that mislead English readers: q is /tɕʰ/, x is /ɕ/, c is /tsʰ/, and the vowel written *e* varies by context. Learning to read pinyin correctly is a distinct skill from learning to read the Latin alphabet.
Where does the stress go in Chinese (Mandarin)?
Four lexical tones plus a neutral tone, and tone is not decoration — *mā*, *má*, *mǎ*, *mà* are four unrelated words. Tone sandhi rules change tones in context, most importantly the third-tone rule and the special behaviour of 一 and 不. Mandarin is syllable-timed with roughly even syllable duration, and the neutral tone syllables are shortened and reduced.
What is the fastest way to fix a Chinese (Mandarin) accent?
Work on the specific contrasts that carry meaning rather than on the accent in general. Week 1 — Tones in isolation, then in two-syllable combinations. All sixteen tone pairs, systematically.