Taiwan Mandarin
⌘ K
Display

Saved on this device.

Pronunciation
how readings are shown
Changes apply instantly.

infrastructure for learners of Taiwan's traditional Chinese

Accurate, structured data on Taiwan's 4,974 traditional Chinese characters, and an MCP server for your LLM tutor.

Taiwan Mandarin is a well-sourced reference you can browse — and a typed JSON API and MCP server you can hand to Claude, Cursor, or any LLM client, so your AI tutor answers from real data instead of guessing.

for people · the website01

Browse the dictionary

Look up any of 4,974 characters — by character, pinyin, or English meaning.

⌘ K
Or jump straight to today's entry →
for your AI tutor · MCP + API02

Plug it into your LLM

One line of config and Claude, Cursor, or any MCP-aware client answers from this dictionary instead of guessing.

{
  "mcpServers": {
    "taiwan-mandarin": {
      "url": "https://mcp.taiwanmandarin.com"
    }
  }
}
works with:Claude DesktopCursorClineGooseany MCP client
MCP server · for your AI tutor

Hand the dictionary to your LLM.

One line of config and your model sees the same data you do — real frequency ranks, readings and TOCFL levels, looked up rather than recalled. Works with Claude Desktop, Cursor, Cline, Goose, or anything that speaks MCP.

① setup

Add to your MCP config

In claude_desktop_config.json, Cursor's settings, or your client's MCP panel.

{
  "mcpServers": {
    "taiwan-mandarin": {
      "url": "https://mcp.taiwanmandarin.com"
    }
  }
}
Restart your client and it will discover the Taiwan Mandarin tools automatically. No keys, no auth — the server is read-only and free for personal use.
② in use
excerpt · Claude, Taiwan Mandarin connected

Your AI tutor, with real data

You

How do I tell apart from ? They feel the same to me.

AI · with Taiwan Mandarin
↳ called character('你'), character('您')
Both are second-person pronouns, but is the polite form — used with elders, customers, or anyone you want to show respect to. is the everyday version.
· everyday "you" · TOCFL Novice 1
nín· polite "you" · TOCFL Novice 1
By frequency

The ten most common characters

Frequency-ranked from Taiwan's COCT corpus. Learn the first hundred and you're past roughly 40% of everyday written Chinese.

Taiwan · COCT
See all ranked ›
  1. #1
    open ›
    de·˙ㄉㄜ
    Grammatical particle indicating possession or modifying a noun; used to form nominal expressions; can also indicate affirmation or emphasis at the end of a sentence.
  2. #2
    open ›
    ·
    The number one; single; a; as soon as; entire; whole; all; a "one" radical in Chinese characters; also used for clarity when spelling out numbers.
  3. #3
    open ›
    shì·ㄕˋ
    To be; yes; correct; right; true; to affirm; to acknowledge; a surname; a thing; indeed.
  4. #4
    open ›
    ·ㄅㄨˋ
    Function word; no; not; un-; negative prefix; used to form questions or emphasize a statement; sometimes used for rhythm.
  5. #5
    open ›
    yǒu·ㄧㄡˇ
    To have; to exist; to possess; to own; abundant; plentiful; old; intentional; a bound form indicating possession or accompaniment.
  6. #6
    open ›
    zài·ㄗㄞˋ
    To exist; to be located at or in a place; to be in the process of doing something; to depend on; to observe.
  7. #7
    open ›
    rén·ㄖㄣˊ
    person; people; mankind; human being; someone else; other people; each person; a certain type or identity of person; character or disposition; surname.
  8. #8
    open ›
    ·ㄨㄛˇ
    Pronoun; first-person singular: I, me, my; also we, us, our in compounds and formal usage.
  9. #9
    open ›
    ·ㄊㄚ
    Function word; third-person singular pronoun: he, him, his; also means other, another; separate; different.
  10. #10
    open ›
    le·˙ㄌㄜ
    Particle indicating completed action or change of state; to finish; to understand; completely; to become clear.
Common radicals

Twelve radicals to start with

These twelve appear in a large fraction of all characters. Spotting them is the fastest way to make a character feel familiar.

About the data

Sourced & free.

Taiwan Mandarin pulls together established Chinese-language datasets — frequency counts from Taiwan's COCT corpus, character data from Unihan, and readings from the Taiwan Ministry of Education's dictionary via Moedict. The English definitions are written by a language model, grounded in those same sources and documented here. Everything is free to use through the JSON API and the MCP server.

Characters
4,974
Radicals
214
Proficiency
TOCFL
Data sources
5 datasets