Cantonese Converter (Basic)

BASIC CANTONESE CHINESE CONVERTER TEXT

About this tool

Convert between Mandarin and written Cantonese (嘅、咗、喺、佢) — basic dictionary version.

Basic Mandarin ↔ Cantonese converter using a curated dictionary. 普通话粤语转换。Runs entirely in your browser.

RT-TXT-043 · Text Tools · Reviewed May 2026

Direction · 方向

Input · 输入

Output · 输出

0 / 5,000 characters · 0 words matched in dictionary

All conversion runs in your browser. No text is sent to our servers. 全部转换在浏览器本地完成。

Accuracy note · 精度说明: Basic dictionary-based conversion — approximately 75% accuracy on short, common phrases. Complex grammar, idioms, or long-form content may convert incorrectly. For professional or long-form translation, the AI-powered Cantonese translator (Premium, coming soon) will be substantially more accurate.

Need higher accuracy? · 需要更高精度？

Our AI-powered Cantonese translator (coming to Premium) handles long-form text, complex grammar, and idiomatic expressions with much higher accuracy.

How to Use the Cantonese Converter

Pick a direction

Mandarin → Cantonese is the default. Use the dropdown to flip to Cantonese → Mandarin if you need the reverse. The mode determines which side of the dictionary applies.

Type or paste your text

Up to 5,000 characters. Live conversion as you type, debounced at 200 ms. The output appears in the right pane. The stats line shows how many words got matched in the dictionary — higher = more confident conversion.

Check what was matched

If the "words matched" count is low for a long input, much of your text passed through without conversion — typically because the dictionary doesn\'t cover the vocabulary. For short, common phrases, expect ~75% of words to match.

Copy, download, or upgrade

Use Copy or Download for the result. For long or complex text where this basic version isn\'t accurate enough, click the Notify me button to be alerted when the AI-based Premium translator launches.

Why Cantonese Conversion Is Harder Than Simplified ⇄ Traditional

A common assumption: "converting to Cantonese" just means converting Simplified Chinese to Traditional. That misses the point. Written Cantonese (書面粵語) isn\'t the same language as Mandarin written in Traditional script. It has its own vocabulary, its own grammatical particles, sometimes different word order, and a set of characters that don\'t exist in standard Mandarin writing at all. A converter that only swaps Simplified for Traditional produces something Cantonese readers can decipher but that doesn\'t sound right — a tell-tale "translated by a Mandarin speaker" register.

The clearest evidence is the five Cantonese-only characters that mark grammar: 嘅 (possessive — replaces Mandarin 的), 咗 (perfective aspect — replaces 了), 喺 (locative "at" — replaces 在), 佢 (third-person pronoun — replaces 他/她/它), and 啲 (plural / quantifier marker — replaces 些/点). These don\'t appear in Mandarin writing. They were codified for written use largely through the Hong Kong Supplementary Character Set in the 1990s, formalising what Cantonese speakers had been writing informally for decades.

"Cantonese is spoken natively by approximately 86 million people worldwide, with major communities in Guangdong, Hong Kong, Macau, Malaysia, Singapore, and diaspora populations across North America and Europe." — Ethnologue 2024, Cantonese language profile

What this basic tool can and cannot do

This is a Phase 1, dictionary-based converter. It carries roughly 195 Mandarin↔Cantonese word/phrase mappings — the highest-frequency substitutions that cover daily conversation. It applies them longest-match-first (so 不知道 → 唔知 wins over a character-by-character pass), then runs OpenCC to normalise the character set to Hong Kong Traditional. On short, common phrases (greetings, simple questions, food orders, directions) it hits about 75% accuracy. On long-form text, idiomatic expressions, song lyrics, or anywhere word order genuinely changes between Mandarin and Cantonese (e.g. Mandarin 正在吃 → Cantonese 食緊), it falls short. The honest framing: this is a helpful assistant for casual writing, not a translator for publishing.

Best uses: chat messages to Cantonese-speaking relatives, short comments on Cantonese-language social media, learning the most common substitutions as you go. Not suitable: literary or professional translation, song lyrics, complex sentences, anything where word order or idiom matters. For those cases, the AI-based Cantonese translator coming to our Premium tier (sign up for notifications above) will use modern LLM-based translation that handles word order, idioms, and context-aware vocabulary.

The ASEAN Cantonese community

Singapore\'s 76% ethnic Chinese population is overwhelmingly Hokkien and Teochew in heritage, but a sizeable Cantonese-speaking minority — especially among older generations — still consumes Hong Kong TVB drama, Cantopop, and Cantonese-language films daily. In Malaysia, Cantonese is the dominant Chinese dialect in Kuala Lumpur, Ipoh, and parts of Penang — ~22% of Malaysians are ethnic Chinese, and the KL Cantonese-speaking population is one of the largest outside Greater China. For these communities, a tool that lets a younger Mandarin-educated speaker write recognisable Cantonese to a grandparent or to an HK contact has practical daily value, even at 75% accuracy.

The Phase 2 LLM-based translator will support proper word-order changes, idiomatic expressions, and register adjustment (formal Cantonese newspaper register vs casual Cantonese chat). When it launches, opting in via the button above gets you the first email — no spam, no marketing, one launch notification and an unsubscribe link.

10 Things to Know About Cantonese

86 million Cantonese speakers. Globally — including diaspora in Malaysia, Singapore, North America (per Ethnologue 2024). Major communities in Guangdong, Hong Kong, Macau.

Written Cantonese is its own language. Not just Mandarin in Traditional script — written Cantonese (書面粵語) has unique grammar, particles, and vocabulary.

5 unique Cantonese characters. 嘅、咗、喺、佢、啲 — none of these exist in standard Mandarin writing. They mark possessive, aspect, location, pronoun, and plural.

Penang Hokkien isn't Cantonese. Penang's Chinese community speaks Hokkien (Min) — often mistakenly grouped with Cantonese (Yue). Different language branches entirely.

Hong Kong since 1842. Cantonese became dominant in HK after British colonisation; Mandarin only mandatory in schools post-1997 handover.

Malaysian Cantonese capital. Kuala Lumpur historically had the largest Cantonese-speaking population in ASEAN. Ipoh and Penang follow.

Tonal complexity. Cantonese has 6–9 tones (depending on analysis); Mandarin has 4 + 1 neutral. Tonal distinctions carry semantic weight.

Hong Kong Supplementary Character Set. HKSCS adds ~5,000 Cantonese-specific characters to Unicode — codifying written Cantonese for digital use.

Cantopop's golden age. 1980s Hong Kong Cantopop is credited with mainstreaming written Cantonese in song lyrics — Sam Hui, Anita Mui, Leslie Cheung.

Why this tool is "basic". Word-order changes (正在吃 → 食緊) require grammatical parsing, not just word substitution. Phase 2 (AI) will handle this.

Frequently Asked Questions

This Phase 1 tool uses a curated dictionary of ~195 Mandarin↔Cantonese word/phrase mappings, applied longest-match-first before OpenCC normalises the character set. It hits roughly 75% accuracy on short, common phrases (greetings, simple sentences, daily speech). For long-form text, idiomatic expressions, or anything where word order changes between Mandarin and Cantonese (e.g. 正在吃 → 食緊), the AI-based Cantonese translator coming to Premium will be substantially more accurate.
The dictionary covers the most common ~195 words and phrases that genuinely differ between Mandarin and written Cantonese. Many Chinese characters are identical in both (人, 中, 文, 大, 小 etc.) and need no conversion. The "words matched in dictionary" stat tells you how many substitutions happened — if it's low for long text, the output is mostly pass-through and not really Cantonese yet.
Not well. Lyrics use compressed syntax, idioms, and tonal/rhyming choices that depend on context the dictionary doesn't capture. The output will be readable but won't feel native. Premium AI translation will handle lyrics significantly better — sign up for notifications when it launches.
The Chinese Converter handles script + regional vocabulary (Simplified ↔ Traditional, Taiwan vs Hong Kong word choice). It assumes both sides are Mandarin and just changes characters/vocab. This Cantonese Converter additionally substitutes Mandarin words for their Cantonese-language equivalents — different vocabulary (the verb "to give" becomes 俾 instead of 给), unique characters (嘅、咗、喺), and pronouns (佢 instead of 他). Use the Chinese Converter for SG → TW/HK publishing; use this one when you want the result to actually read as Cantonese.
Use the 👎 Not helpful button in the sidebar, or email the team via the Contact page. Each missed mapping is a candidate for v1.1 dictionary expansion. The dictionary is maintained by RECATOOLS as a public-facing artefact — community-suggested entries are reviewed and added in periodic updates.
No. The dictionary (a small JSON file) and the OpenCC bundle (~1.1 MB shared with the Chinese Converter) load once in your browser, and all conversion runs locally. Your input text, output text, and the conversion process never touch our servers. This matters for personal messages, legal drafts, or anything covered by data-residency requirements.
No firm date yet — we're building the LLM router and quality-evaluation pipeline first. Click the "Notify me" button on this page to be added to the launch list. We'll email you when it's ready. No spam, no marketing — one launch email and you can unsubscribe at any point.
In the reverse direction (Cantonese → Mandarin), characters not in the reverse map pass through unchanged. Then OpenCC tw→cn converts them to Simplified Chinese form. So a rare Cantonese-only character will either stay as-is (no Mandarin equivalent) or get its character form changed to Simplified. The "words matched" stat shows what was understood vs passed through.
Written Cantonese is overwhelmingly written in Traditional script — Hong Kong and Macau use Traditional, and Cantonese vocabulary characters (嘅、咗、喺、佢、啲) are themselves Traditional forms. Even Mainland Guangdong sources typically use Traditional for Cantonese content. Converting Cantonese to Simplified would be unusual and potentially confusing.
Limited. The dictionary covers formal-to-conversational written Cantonese but doesn't include heavy internet slang (e.g. 「我都係」、「你冇野呀？」 specific to forum/social-media usage). For these, the Phase 2 AI translator will handle context-dependent slang better. Suggestions for slang entries are welcome via the feedback flow — we add high-frequency terms in dictionary updates.

Cantonese Converter (Basic)