Hanja–Hangul Converter

Reading Hanja in Korean: one sound per character, and the rules that move it

Every Chinese character used in Korean has a Sino-Korean reading of exactly one syllable: 韓 is 한, 國 is 국, 大 is 대. That is what makes Hanja comparatively easy to read aloud — and what makes a character-by-character lookup almost right. The places where it goes wrong are the ones this page is built around.

The initial sound law

Standard Korean spelling does not let a word begin with ㄹ, or with ㄴ before the vowels ㅣ, ㅕ, ㅛ and ㅠ. 한글 맞춤법 articles 10 to 12 spell those readings differently at the start of a word: 女子 is 여자 although 女 is 녀, 歷史 is 역사 although 歷 is 력, 老人 is 노인 although 老 is 로. Inside a word the character keeps its own reading, so the same 女 is 녀 in 男女, 남녀. Two further rules follow from the same articles: 렬 and 률 after a vowel or ㄴ are written 열 and 율 (比率 비율, but 確率 확률), and a compound or a word with a prefix-like character starts a new word in the middle (新女性 is 신여성, not 신녀성). The last one cannot be seen from the characters, so the page offers it as a choice rather than guessing.

Customary readings and numbers

Some words are written with a sound the character does not normally take. Article 52 of the spelling rules lists them in pairs: 承諾 is 승낙 but 許諾 is 허락; 萬難 is 만난 but 困難 is 곤란; 十日 is 십일 but 十月 is 시월; 五六十 is 오륙십 but 六月 is 유월; 木材 is 목재 but 木瓜 is 모과 (한글 맞춤법 제52항). The page applies exactly the words that article and the National Institute of Korean Language's commentary on it print, and marks each one; it does not extend the list by guesswork. Numbers follow a rule of their own: in a number written out in Hanja, each digit after 十, 百 or 千 is spelled as if it began a word, so 六千六百六十六 is 육천육백육십육 — the article's own example (제11항 붙임 5).

Characters with more than one reading

Only 176 of the 8,525 characters here have more than one reading once the start-of-word forms are set aside — but they include some of the commonest, and there the reading belongs to the word. 金 is 금 (gold) in 金額 and 김 as a family name; 北 is 북 (north) and 배 (to turn away, as in 敗北, 패배); 不 is 불 but 부 before ㄷ and ㅈ (不動産, 부동산) — a rule the National Institute of Korean Language calls regular, with 不實 부실 as the one exception before ㅅ. The page applies the rules it can state, starts every other choice from the reading on the Ministry of Education's list of 1,800 basic hanja, and marks the character so the choice is visible.

A child's name that uses a character outside the permitted range is recorded in the family register in Hangul only. — translated from 가족관계의 등록 등에 관한 규칙, Article 37(3)

Hanja for names

Article 37 of the family-registration rules limits the Hanja a Korean name may use: the 1,800 education hanja, the Supreme Court's own table of additional characters (Table 1, each tied to a stated reading), and the variant forms listed in Table 2 (Article 37 ① and ②). The permitted count has grown in steps, each recorded in the rule's amendment history: 5,761 from July 2013, 8,142 from January 2015, 8,279 from December 2018, 8,319 from February 2022 and 9,389 from 11 June 2024 (제정·개정이유, law.go.kr; Supreme Court announcement, 23 May 2024). The list this page marks comes from Unihan, and Unicode says its copy predates the last two steps: the 40 characters added in 2022 and the 1,070 added in 2024 are not in it (UAX #38, kKoreanName) — none of the three characters the court's 2024 announcement named (㖀, 疋 read 아, 䬈) carries the mark. So a character marked as an education hanja is settled; one marked from the name list was on it; and an unmarked character may well be allowed. For each syllable the page shows how many characters the current table lists beside how many Unihan marks, so the gap is visible rather than hidden.

The limit itself still stands. On 29 April 2026 the Constitutional Court dismissed a parent's challenge to it, brought after a daughter's name written with 婡 was registered in Hangul only; four justices dissented (2023헌마282).

Why the same character can be two code points

South Korea's KS X 1001 character set, which fixed the Hanja most Korean software could type for decades, arranged its 4,888 characters by reading and encoded a character once for each reading it listed. Unicode kept those 268 duplicates as compatibility ideographs, so U+F914 is 樂 as encoded under 낙. Normalising text silently turns it back into the ordinary 樂 and loses the reading, so the page reads such a character as typed.

The commoner problem runs the other way. A character typed on a Hong Kong, Taipei or Tokyo keyboard is often a different code point from the Korean one — 研 rather than 硏, 為 rather than 爲, 黄 rather than 黃, 青 rather than 靑 — and Unicode's Korean readings sit on the Korean form only. Measured against Unihan, 556 of the 5,402 characters in Big5's frequent level and 239 of Japan's 2,136 常用漢字 had no Korean reading of their own. The page now reads such a character through the Korean form that Unicode's variant data, or the old form printed beside it in Japan's 常用漢字表, names, and it says which form it used; that leaves 466 and 9 without a reading. It never treats the typed character as being on a Korean list: 青 is labelled as the school hanja 靑 written at another code point.

Korean readers in Southeast Asia

Korean names and company names reach readers in Singapore and Malaysia in three scripts at once — Hanja on a name card, Hangul in a news report, romanized on a passport. Reading the Hanja in Korean and romanizing the result, as this page does, is how those three are matched up.