Skip to content

Sound Roots: The Phonetic Component Inside the Kanji

The other half of the root system. Compounds combine whole kanji; most kanji are themselves built from a meaning part and a reusable sound part.

Part of the Roots of Japanese project. See also: Catalog | The Data | Verb Roots

The root inside the character

The rest of this project treats whole kanji as roots that combine into compound words. But there is a second, smaller root system inside each character.

Most kanji are phono-semantic compounds (形声文字, keisei moji): they pair two parts.

  • A meaning part (the 意符, usually the radical) tells you roughly what the kanji is about.
  • A sound part (the 声符, the phonetic component) tells you its on-reading.

Take 清 (clear):

Part Role Contributes
氵 (water) meaning "to do with water / liquid"
青 (SEI) sound the reading セイ

So 清 reads セイ and means "clear (as water)". The sound part 青 is doing the same job as a Latin root: once you know it carries セイ, it hands you the reading of every kanji it appears in.

How common is this?

Very. It is the dominant way kanji are built, though the exact share depends on how you classify the edge cases:

  • By Shirakawa's classification of the 2,136 Jōyō kanji, about 61% are phono-semantic, rising to 72% across the more advanced second half of the list.
  • Counting Chinese characters more broadly, the usual figure quoted is 80 to 90%.

Either way, the phonetic component is the single highest-leverage pattern for reading kanji you have never seen, in the same way productive kanji are the highest-leverage pattern for vocabulary.

Sound series: learn one, unlock many

A sound series is a family of kanji that share one phonetic component and therefore share an on-reading. A few clean ones:

Sound part Reading Kanji that share it
青 セイ 清 (clear), 晴 (fine weather), 精 (refined), 請 (request)
包 ホウ 抱 (embrace), 泡 (bubble), 砲 (cannon), 胞 (cell), 飽 (be full)
司 シ 詞 (word), 飼 (raise animals), 伺 (inquire), 嗣 (heir)
工 コウ 江 (large river), 攻 (attack), 功 (achievement), 項 (clause)
白 ハク 拍 (beat), 泊 (lodge), 迫 (urge), 伯 (count/uncle)
半 ハン 伴 (companion), 判 (judge), 畔 (shore)

Learn that 青 carries セイ and you can read 清, 晴, 精, and 請 on sight. This is the within-character mirror of the compound roots: the same "learn the part, generate the whole" logic, one level down.

The catch: a hint, not a rule

Sound parts are reliable enough to be worth learning, but they are hints, not guarantees. Studies put their predictive accuracy somewhere in the 40 to 70% range, for a few honest reasons:

  • Inconsistent series. The same sound part can give different readings. 各 yields カク in 格 (status) and 閣 (cabinet), but キャク in 客 (guest), ラク in 落 (fall), and ロ in 路 (road).
  • Voicing (連濁 / rendaku). A reading can pick up a voiced consonant in compounds, so セイ can surface as ゼイ, ハク as バク.
  • Layered borrowing. Kanji entered Japan from different regions and eras, so a phonetic that was consistent in Chinese can map onto more than one Japanese on-reading.

The practical rule: when you meet an unfamiliar kanji, find its sound part, guess the on-reading, then confirm. You will be right far more often than chance, and the misses teach you the exceptions.

Where this fits

This project now maps roots at two scales:

  • Inside the character (this page): the phonetic component that drives the reading.
  • Across characters (the Catalog and The Data): productive kanji that drive compound vocabulary, plus the Verb Roots that drive compound verbs.

Same idea, three scales: find the reusable part, and a whole family comes with it.

Sources