Cantonese Characters

Cantonese has no alphabet of its own. It is written in Chinese characters, and in Hong Kong and Macao those are Traditional characters. There are really two ways to write it. Newspapers, schoolbooks, laws and most signs use standard written Chinese (書面語), the same written language as Mandarin, which Hong Kong people simply read aloud in Cantonese: 他是我的朋友 is said taa1 si6 ngo5 dik1 pang4 jau5. In chat, advertising, comics, gossip columns, menus and street talk, people write written Cantonese, the words as they are actually spoken: 佢係我嘅朋友 keoi5 hai6 ngo5 ge3 pang4 jau5, "he is my friend".

Written Cantonese uses mostly the same characters as Mandarin, read in Cantonese: 粵語 jyut6 jyu5 is the language's formal name, 廣東話 gwong2 dung1 waa2, Guangdong speech, what most speakers call it. But where Cantonese grammar and everyday words differ, it needs characters of its own, and a large share of them carry the mouth radical 口 on the left, a sign that the character stands for a spoken sound: 嘅 咗 啲 嗰 喺 嘢 嚟 嘥. Those are the characters people mean when they search for Cantonese characters that don't exist in Mandarin, and they are the heart of this page.

In Guangzhou and the rest of Guangdong, Cantonese is written in Simplified characters, so 話 becomes 话 and 廣東 becomes 广东; that version has its own collection, Cantonese (Simplified). This page follows Hong Kong usage.

Cantonese characters vs Mandarin: 佢 嘅 咗 唔 冇 喺

The small words carry most of the difference. 佢 keoi5 is he or she (Mandarin 他); plurals add 哋 dei6: 我哋 we, 你哋 you all, 佢哋 they. 係 hai6 is "is" (Mandarin 是), 唔 m4 not (不), 冇 mou5 not have (沒有), written as 有 (have) with its two inner strokes taken out. The possessive is 嘅 ge3 (的), this and that are 呢 ni1 and 嗰 go2 (這, 那), at is 喺 hai2 (在), and 咗 zo2 marks a finished action (了): 食咗飯未呀？ "have you eaten yet?"

Everyday verbs follow: 睇 tai2 look, watch; 嚟 lei4 come; 攞 lo2 take; 搵 wan2 look for (also typed 揾); 畀 bei2 give (also 俾); 瞓 fan3 sleep; 諗 nam2 think. Then the words that make Cantonese sound like itself: 嘢 je5 thing, 靚 leng3 pretty, 叻 lek1 clever, 攰 gui6 tired, 嬲 nau1 angry, 啱 ngaam1 right, correct, 掂 dim6 sorted, as in 搞掂 (done, handled), and 乜 mat1 or 咩 me1 for "what".

At the end of a sentence Cantonese adds a particle that sets its tone, and every one has its own character. 啦 laa1 urges (快啲啦! hurry up!), 喇 laa3 says something has changed, 喎 wo3 reports or corrects (佢話唔嚟喎, he says he isn't coming), 㗎 gaa3 explains or affirms, 啩 gwaa3 means "I suppose", 咋 zaa3 "only", and 咩 turns a statement into a doubtful question (唔係咩？ isn't it?). 㗎 sits in CJK Extension A, so a few older fonts show a box; 架 is the stand-in some people type.

Cantonese alphabet? Jyutping, Yale and the tone symbols

Cantonese is written in Latin letters only in romanization, and the system you will meet most is Jyutping (粵拼), published by the Linguistic Society of Hong Kong in 1993. It uses plain letters and writes the tone as a digit after each syllable: 香港 hoeng1 gong2, 你好 nei5 hou2, 唔該 m4 goi1. Cantonese has six tones, numbered 1 to 6 in Jyutping; the traditional count is nine, because syllables ending in -p, -t or -k, such as 十 sap6, were once counted as tones of their own. Textbooks often print the numbers raised, hoeng¹ gong², and the superscript digits ¹ ² ³ ⁴ ⁵ ⁶ are here to copy.

The older Yale romanization, still used in many English-language textbooks and dictionaries, marks the tone with accents instead: a macron for the high level tone (ā), an acute for the rising tones (á), a grave for the falling tones (à), and an h after the vowel for the low tones. So 香港 is Hēunggóng, 粵語 Yuhtyúh and 你好 néih hóu. Cantonese also has syllables made of a bare m or ng, and Yale marks those too: 唔 m̀h, 吳 ǹgh, 五 ńgh. Unicode has ready-made ń and ǹ, but no m with a grave, so m̀ is m followed by the combining grave accent U+0300.

The spellings on Hong Kong street signs, Tsim Sha Tsui, Mong Kok, Sheung Wan, are neither of these: they come from the government's own older romanization of place and personal names, which has no tone marks at all.

How to type Cantonese characters

Most Cantonese characters are in the ordinary Traditional Chinese character set and come up in any Hong Kong input method. On Windows 10 and 11, go to Settings › Time & language › Language & region and add Chinese (Traditional, Hong Kong SAR), which adds Microsoft Quick by default; add Microsoft Changjie from Language options. On a Mac, add a Cantonese or Traditional Chinese input source under System Settings › Keyboard › Input Sources. Many people in Hong Kong type with Cangjie (倉頡), which builds a character from its shapes, or with handwriting input, drawing the character with a finger on a phone or a pad. Learners who type by sound use Jyutping keyboards, available for iPhone, Android and computers, including the free Rime engine.

The harder ones are the characters that Hong Kong needed beyond the old Big5 standard. The government's Hong Kong Supplementary Character Set (HKSCS) collected them, and they were later encoded in Unicode, many far out in CJK Extension A and B: 㗎 and 㩒 in Extension A, 𨋢 (lift), 𠝹 (to cut) and 𡃁 (youngster) in Extension B. Input methods and fonts may not have them, so copy them from this page; in HTML write 𨋢 for 𨋢 or 佢 for 佢.

For Yale romanization, the accented vowels are on every Latin keyboard. On Windows with US-International, type ' then the vowel for á and then the vowel for à; on a Mac, Option+e or Option+ then the vowel, and Option+a then the vowel for the macron ā on the ABC Extended layout. On a phone, hold the vowel and slide to the accent. The combining grave for m̀h is U+0300: in Word type 0300 right after the m and press Alt+X, or copy it here.

Where Cantonese is spoken

CLDR estimates about 82 million Cantonese speakers. Most, around 74 million, live in China 🇨🇳, in Guangdong and parts of Guangxi, where Guangzhou, the old Canton that gave the language its English name, is the centre. Hong Kong 🇭🇰, with about 6.6 million speakers, and Macao 🇲🇴, with about 550,000, are the places where Cantonese is the everyday language of government, television, film and Cantopop, and where it is written in Traditional characters. Abroad, CLDR counts nearly 780,000 speakers in Canada 🇨🇦, in Vancouver, Richmond and Toronto above all, and there are old Cantonese communities in Malaysia, Singapore, the United States, Britain and Australia; the first Chinatowns of North America spoke Taishanese, a related Yue language from the same part of Guangdong.

Cantonese is one of the Chinese languages: a Mandarin speaker cannot follow spoken Cantonese without learning it, though both write in the same characters. Its neighbours have their own pages: Chinese and Traditional Chinese for Mandarin, Hakka Chinese, whose speakers share 佢 and 係, Min Nan Chinese (Hokkien) and Wu Chinese. For the emoji and symbols of the places themselves, see the Hong Kong SAR China, Macao SAR China, China and Canada collections.