The five Cantonese phrases that do 80% of the work
First time? Do this
- Audio is built in. Tap any phrase to hear it in a native voice — pre-recorded and cached, so it just works offline, no setup. (The voice bar can still pick a phone voice as a backup.)
- Phrases — drilled by situation: the cha chaan teng, dai pai dongs, directions, the ferry. Hit play, then say it back out loud.
- Sounds — tones and characters, tap to hear how each one sounds. Ten minutes here pays off the most.
- Customs — the unspoken rules. This is where tourists actually give themselves away.
- Practice — hear / read / recall / speak. Use it on the plane.
Right thing, right moment.
Tap a chip to jump to a scene — or just search. Common holds the go-anywhere basics (hello, thanks, where's the toilet); a few phrases like drinks live under more than one scene.
You can type in either language — both sides are searched.
Common — useful everywhere · greetings · basics · the toilet
Greetings & manners · the social handshake
At the restaurant · ordering & dining
Bars & nights out · your scene
Wine & spirits · tasting & buying
Finding your way · asking directions
Taxis & getting around · rides & fares
On the water · ferries & boats
Shops & getting by · everyday
Sea & scenery · quick vocab
Numbers · prices & counting
- Tens and ones glue into a single compact word. 21 is simply 二十一 (yih-sahp-yāt, "two-ten-one") — no linking word needed, unlike English "twenty-ONE".
- 萬 (ten-thousand), not "thousand", is the pivot. Chinese counts in blocks of four digits: after 千 (yāt-chīn, 1,000) the next new word is 萬 (maahn) for 10,000, not a bigger "thousand". 120,000 is thought of as "12 maahn" (十二萬), not "120 thousand" — reading a Chinese price in English needs this regrouping.
- There's no native word for "million" — it's built from 萬. 1,000,000 is 一百萬 (yāt-baak-maahn, literally "one hundred ten-thousand") rather than an atomic word of its own — always work it out as 100 × 10,000.
- 100 and 1,000 both need a leading 一 (one). Cantonese always says the leading "one": 一百 (yāt-baak, 100) and 一千 (yāt-chīn, 1,000), never a bare 百 or 千 alone in speech.
- 二 and 兩 are both "two" — but not interchangeable. 二 (yih) is used in numerals and counting sequences like 二十 (20), while 兩 (léuhng) is used with a measure word for quantities — 兩個人 ("two people"), 兩蚊 ("two dollars"). Picking the wrong one is one of the first mistakes learners make.
Same characters as Mandarin, a totally different soundtrack.
Cantonese keeps six tones and word endings that Mandarin dropped centuries ago, so a phrase that looks the same on a menu can sound nothing alike. Read the CAPS in this kit as pitch, not punch.
Knowing what a letter sounds like isn't the same as being able to make it. These four are where English mouths actually go wrong: what to do, what to stop doing, and a native recording you can slow down and copy.
What your mouth doesEvery syllable carries a pitch shape — flat-high, rising, mid-flat, low-falling, low-rising, low-flat — and changing the pitch changes the word entirely, not just the emphasis. Keep your throat relaxed and let the pitch glide smoothly.
The mistakeEnglish speakers try to show meaning by getting louder or leaning on a syllable harder — Cantonese tone is about pitch shape, not volume, so a loud flat tone still sounds like the wrong word.
What your mouth doesPush a light puff of air out for the "p" sound, and keep zero breath for the "b" sound — both are technically voiceless in Cantonese, so breath is the only cue that tells them apart.
The mistakeEnglish speakers rely on vibrating vocal cords to tell b from p — that cue doesn't exist here, so a breathy "g" where none belongs can land as a different syllable entirely.
What your mouth doesWhen a syllable ends in p, t, or k, your mouth reaches the closing position — lips together for p, tongue on the ridge for t, back of tongue up for k — and simply stops there. No puff of air escapes at the end.
The mistakeEnglish speakers instinctively release the final consonant with a little burst of air; in Cantonese that release turns a clipped, closed syllable into something that sounds like an extra vowel got added.
What your mouth does唔 (m̀h, "not") is just a nasal hum — press your lips together and hum through your nose, similar to an "mm-hm", held level with no vowel sound at all.
The mistakeEnglish speakers want to add a vowel (like "muh") in front of it, but that changes the word — practise saying just the hum by itself first.
- Tone changes the word, not the mood. Cantonese runs six tones across every syllable — said with a rising pitch, gáu is "nine" (九); said high and flat, gāu is a different word altogether (交, "to hand over"). Pitch is grammar here, not emotion.
- No stress accent like English. Cantonese doesn't lean on one syllable in a sentence the way English does — every syllable keeps roughly equal weight, so the CAPS in this kit mark pitch prominence, not loudness.
- Breath, not voice, splits consonant pairs. b/p, d/t and g/k are told apart by a puff of air, not by vibrating your vocal cords the way English b/p does.
- Endings stop dead — no release. Syllables ending in -p -t -k close and hold; nothing bursts out the end the way English "stop" does.
- One character, one syllable, one meaning. Every Chinese character is exactly one spoken syllable and carries its own meaning — words are built by stringing two or more characters together, not by adding endings or prefixes.
The rules nobody prints on the menu.
Your Cantonese rarely outs you. These do. Read once, and you'll move differently.
Learn the tea-cup finger tap
When someone refills your tea, tap two fingers lightly on the table — 叩指禮 (kau-JI-lai, "finger-knock etiquette") — instead of saying thank you every single time. The story traces to an emperor travelling in disguise; it's still everywhere today.
Don't stand your chopsticks up in rice
Sticking chopsticks upright in a bowl of rice echoes the incense sticks burned at a funeral. Rest them across the bowl, or on a chopstick rest, between bites.
Tipping isn't expected
Restaurants, taxis and hotels in mainland China don't expect a tip, and offering one can just confuse the person you hand it to. Good service is thanked with words, not cash.
Match the toast, or say so upfront
At a meal with baijiu, someone will raise a glass with 飲勝 (YUM-sing, "cheers") — literally "drink to victory", and it means drain the glass. If you'd rather sip, say 隨意 (CHEUI-yi) before the first toast, not after.
Skip clocks and the number four as gifts
A clock as a gift — 送鐘 (SUNG-zung) — is an exact homophone for "seeing someone off at their deathbed", and the number 四 (sei) sounds just like "death". Avoid both as gifts or quantities.
Cash is nearly obsolete
Almost everyone pays with a phone, scanning a QR code through WeChat Pay or Alipay for anything from a taxi to a single bottle of water. Carry a little cash as backup — small vendors don't always want to break a big note.
Offer and receive with both hands
Business cards, gifts, and money are handed over and accepted with both hands, or at least a supported right hand — a small gesture that reads as respectful rather than formal.
Meals are shared, not ordered solo
Dishes land in the centre of the table for everyone's chopsticks, rice is often more of a background filler than the main event, and whoever is hosting steers the best pieces toward their guests.
Don't clear your plate completely
Finishing every last bite as a guest can read as a signal the portions were too small. Leaving a little behind is the polite way to say you're full and happy.
"Face" shapes how requests land
Public criticism, a flat refusal, or making someone look foolish in front of others costs 面子 (MIN-ji, "face") — for you and for them. A soft "maybe" or a redirected question is often a polite no.
The plates and glasses worth going out of your way for — each with the name as you'd say it out loud.
Test the four channels.
Understanding Cantonese by ear, reading, recall, and your own mouth. Quick rounds — do them twice.