Chinese pronunciation for English speakers: the sounds that trip you up

Most Mandarin sounds are already in your mouth

Mandarin builds every word from a short list of syllables. Our pinyin chart has 403 of them, and each starts with at most one consonant and ends in a vowel, n or ng.

English is far looser. It lets up to four consonants end a syllable, as in strengths. Mandarin allows just n or ng, plus an r in some Beijing words.

Most of the pieces will feel familiar. Here's how each pinyin consonant and vowel compares with English:

Eight consonants work just as they do in English, and seven more need only a small change, like an extra puff of air. The real work sits in two rows of the chart, plus three vowels.

The new sounds are in almost every sentence

You can't dodge these sounds by learning around them. We turned 33.5 million words of Chinese film and TV subtitles into pinyin and counted every syllable.

Nearly two in five contained at least one sound English doesn't have. The single word 是(shì), to be, makes up 2% of all the syllables people say on screen.

Here's how often each tricky sound turns up:

Source: our count over SUBTLEX-CH, with pinyin by word. Unstressed syllables, like the possessive de, don't count for e, because their weak "uh" exists in English. How we counted.

So start with the tongue sounds. Together, zh, ch, sh, r, j, q and x begin almost three syllables in ten, while erhua barely registers.

Eight sounds cause most of the trouble

Each tricky sound has an English sound that feels close enough. That near-miss is what your mouth reaches for, and it's what an accent is made of.

Here are the eight, the English sound they get mistaken for, and the fix:

Our summary of the sound values in Lee and Zee's IPA description of Standard Chinese, checked against English. The fixes are standard teaching cues.

Notice that two rows share one English near-miss. English j, ch and sh stand in for two different sets of Mandarin sounds, and that's where most of the confusion starts.

Hear the difference in pairs

Train your ear before your mouth. In a classic study, Japanese adults trained only to hear English r and l also said them better afterwards.

A review of 32 such studies found that this kind of training works and that the gains last. It uses pairs of words that differ in just one sound, like few, 少(shǎo), and small, 小(xiǎo).

We built a sound lab from the recordings in our pinyin chart. Here's how to use it on one pair:

• Listen. Play both words a few times, and notice what changes.
• Test your ear. Take the ear test, where the lab plays one word and you pick which you heard.
• Say it. Record yourself saying one word of the pair.
• Compare. Play the model and your recording back to back, and adjust.

I'd start with whichever pair makes you hesitate, then work through the rest. One caveat: the lab uses one computer voice, while the studies used many real ones. Once a pair feels easy here, listen for it in songs, shows and conversations, where the sounds come from many mouths.

Pinyin spelling misleads English eyes

Pinyin was adopted in 1958 to write Mandarin, not to help English readers. The official scheme borrowed Latin letters and gave several of them sounds English spells another way.

That's why 七(qī), seven, looks like "kee" to English eyes. Here are the spellings that trip readers up most:

Source: the Scheme for the Chinese Phonetic Alphabet, 1958, with sound values from Lee and Zee, 2003.

Read pinyin as a code for Mandarin sounds, not as English spelling. Once you know these ten traps, most of the rest says what it looks like.

Curl your tongue for zh, ch, sh and r

The sounds zh, ch, sh and r share one tongue position. Lift the tip of your tongue to just behind the bony ridge above your top teeth.

These sounds are often called "retroflex", which means curled back. But a classic study of Beijing speakers found only a slight curl: the tongue tip rises behind the gum ridge, and the underside stays out of it.

Here's where your tongue goes for each group, with the words to try:

Mandarin r uses the same position as sh. Hold sh, switch on your voice, and you get a buzz close to the s of measure: that's the r of 人(rén), person.

If your zh, ch and sh still sound English, check your lips. English sh usually pushes them forward; Mandarin sh doesn't need to. The good news is that adults can learn the difference: in one study, American learners of Mandarin made their Mandarin sh measurably different from their English one.

Flatten it for j, q and x

The sounds j, q and x have no English twin. Rest your tongue tip behind your bottom teeth, then push the middle of your tongue up toward the roof of your mouth, and smile.

If that feels abstract, build x from a sound you already make:

• Start from yes. Say the y of yes and hold it.
• Whisper it. Push air through that position instead of voice: that's close to x.
• Add a stop. Start x with a light t and you get j; release it with a strong puff and you get q.

You can hear the difference as the pitch of the hiss. We measured it in our pinyin chart's recordings of s, x and sh. Here's where each one sits:

Source: our measurements of SayMei's pinyin chart recordings, one computer voice. How we measured.

Every x we measured sat above the typical sh and below every s. So if your x sounds like English sh, brighten the hiss: tip down, middle up, lips spread.

Hearing it takes practice. In a cross-language study, English-speaking adults told q from x less reliably than Mandarin speakers did.

Round your lips for ü

English ears tend to hear ü as oo. In a 2023 study, English-speaking listeners filed Mandarin ü and u under the same English vowel, the oo of food.

To make it, say ee and freeze your tongue. Then round your lips as if for oo. What comes out is ü, as in green, 绿(lǜ).

Pinyin hides it, too. After j, q, x and y, the two dots are dropped, so go, 去(qù), and rain, 雨(yǔ), both have ü.

Practise the u and ü pair in the sound lab until road, 路(lù), and green, 绿(lǜ), sound like two different words.

English has no twin for e or the buzzing i

Mandarin e is ü's mirror image: a rounded vowel with the rounding taken away. Say oh, keep your tongue where it is, and spread your lips into a flat line. That's the e in to drink, 喝(hē), and hungry, 饿(è).

The second vowel hides behind the letter i. After z, c, s, zh, ch, sh and r, the i isn't ee: it's the consonant's own tongue position, held and voiced. So to be, 是(shì), isn't the English word "she".

These two are hard to hear, not just to say. In a 2026 study, English-speaking listeners heard Mandarin e as a vowel with no English match, and the authors called for targeted training on the buzzing i.

Use the e and u pair in the lab for the first, and the s and sh pair for the second: four, 四(sì), against to be, 是(shì).

The endings n and ng change the word

Apart from the r of erhua, only two consonants can end a Mandarin syllable, n and ng, and swapping them changes the word. Meal, 饭(fàn), isn't to put, 放(fàng).

You can already say both endings, as in sin and sing. The trap is the vowel in front of them. Before ng, the a moves to the back of the mouth, so to help, 帮(bāng), sounds darker than class, 班(bān).

These endings carry a lot of weight: in our subtitle count, a quarter of all syllables end in n or ng. You'll also hear variation. Many speakers in Taiwan blur the endings in and ing, so new, 新(xīn), and star, 星(xīng), can sound alike.

Finish each syllable on purpose: the tongue tip up for n, the back of the tongue up for ng.

Erhua, the Beijing r, can wait

Erhua adds an r sound to the end of a syllable, as in here, 这儿(zhèr), and where, 哪儿(nǎr). Beijing speech is full of it.

You'll hear less of it than you might expect. In our subtitle count, erhua turned up in about one syllable in 600.

It's also regional. Erhua is very rare in Taiwan, and many southern speakers avoid it when they speak Standard Chinese. They say 这里(zhèlǐ) for here instead.

Learn to recognise it, and use it in the few words where you hear it. It can wait until the other sounds feel solid.

What to do next

• Train your ear first. Run the sound lab's ear test on one pair a day until you rarely miss.
• Start with the tongue sounds. The sounds zh, ch, sh, r, j, q and x come up most, so fix them before anything else.
• Read pinyin as a code. Keep the spelling table above handy for your first few weeks.
• Record and compare. Say each word after the model, then play both back to back.
• Put the sounds in sentences. A sound you can say alone can still slip in fast speech, so practise it in real phrases.

Prefer paper? The free Mandarin Pronunciation Guide PDF covers the same sounds with drills and an answer key.