Phonetic Components in Kanji (音符): The Hidden Reading Hint in 75% of Kanji
Phonetic components in kanji are the sound-bearing pieces inside a 形声 (keisei) phono-semantic compound. One piece signals the meaning category. The other, the 音符 (onpu), signals the on'yomi.123 Learning to spot the onpu turns a wall of unfamiliar characters into a network of reading families. One known kanji can often predict the on'yomi of half a dozen others.4
The same word 音符 (おんぷ) more commonly means "musical note" (quarter note, eighth note). In kanji-formation linguistics, it names the sound-mark component of a 形声 character. Both senses are attested under the same headword in Wiktionary.1
Overview
What a phonetic component is, in one sentence
In a 形声 kanji, one component (the 意符, ifu, "meaning mark") signals a broad meaning category. Another component (the 音符, onpu, "sound mark") signals the on'yomi. The 音符 is what this article calls the phonetic component.123
The Japanese Wikipedia 形声 article describes the pairing as combining "a symbol representing the category or type of thing (意符) with a symbol representing pronunciation (音符) to form a new character."2 The morphology of both words is transparent: 意 ("meaning") + 符 ("mark") gives "meaning mark," and 音 ("sound") + 符 gives "sound mark."1
How this article is scoped
This article covers the sound side of a 形声 kanji. Separate articles in this pillar cover the meaning-side mechanics, the strict 部首 definition, the radical-vs-component distinction, the positional taxonomy of へん and つくり, the on'yomi historical strata, the on+on / kun+kun / jūbako / yutō compound-reading decision flow, and the working top-N onpu inventory.
The transferable skill is reading the on'yomi from a kanji's right side before the dictionary tells you what it is. The worked examples use three series: 青, 反, and 工. The reliability discussion comes at the end so the limits of the technique are explicit.
How a phonetic component works
The 意符 and 音符 split inside a 形声 kanji
A 形声 kanji is built from two named components: an 意符 that signals a broad meaning category, and an 音符 that signals the on'yomi.23 Both are linguistic terms with transparent morphology, and both appear together in Japanese descriptions of how 形声 characters work.12
The pairing is reciprocal. The meaning-side and sound-side treatments are two views of the same composition. This article zooms in on the 音符 view, and the meaning-side companion article handles the 意符 view in detail.
Why the phonetic usually sits on the right
In a left-right 形声 kanji, the usual structure is 意符 on the left and 音符 on the right.45 The Kanji Code formulates the convention as a workflow rule: in left-right compositions, the phonetic component tells you the on'yomi and sits on the right side.4
The convention is a tendency, not a law. Phonetics in ashi (bottom) and kanmuri (top) positions are well-attested but rarer. A handful of kanji put the phonetic on the left and the semantic on the right; the worked series below includes 功 as a clean example of that flip.6
A worked decomposition: 清 = 氵 + 青
清 ("clear, pure, clean") is a 形声 kanji. Its 意符 is 氵 (the water radical, a hen-position variant of 水), and its 音符 is 青.8 The on'yomi of 青 alone is セイ (kan-on) and ショウ (go-on), and the on'yomi of 清 is the same セイ and ショウ.98
The prediction "the right side is 青, so this kanji reads セイ" gives you 清's on'yomi before you meet 清 in a dictionary. That is the entire reading-prediction workflow, compressed into one kanji.
清流で泳ぎたいです。10
"I want to swim in a clear stream."
How big a share of kanji this covers
The share of kanji that are 形声 depends on how strictly "phonetic" is defined. The literature reports a range rather than a single number.23411 Roughly 75–80% of jōyō kanji are phono-semantic, depending on which source you count. Conservative counts land near 67%, mid-range learner sources settle on 80%, and the broadest historical counts (including kanji whose phonetic component is now opaque) exceed 90%.23411
| Source band | Figure | Counting basis |
|---|---|---|
| Conservative learner sources | about 67% | strict modern 形声 inventory11 |
| Mid-range learner sources | at least 80% | jōyō 形声 inventory4 |
| Historical-inclusive counts | over 90% | counts kanji whose phonetic is now opaque23 |
The 75% in this article's title is a widely-cited learner-facing headline that sits between the conservative and mid-range bands. The body keeps the spread visible so the reader is not misled by any single number.
Worked phonetic series
The 青 series: セイ (and ショウ)
青 itself reads セイ (kan-on) and ショウ (go-on). It has kun'yomi あお and あおい, and senses "blue, blue-green; green (of grass and plants); young, immature."9 As an 音符, 青 anchors a productive series across 清, 晴, 精, 請, and 静.9
| Kanji | 意符 | On'yomi | Sense |
|---|---|---|---|
| 清 | 氵 (water) | セイ, ショウ | "clear, pure"8 |
| 晴 | 日 (sun) | セイ, ショウ, ジョウ | "to clear up, fine weather"12 |
| 精 | 米 (rice) | セイ, ショウ | "refined, essence, spirit"13 |
| 請 | 言 (speech) | セイ, ショウ, シン (tō-on) | "to ask, to request"14 |
| 静 | 青 + 争 | セイ, ショウ, ジョウ | "quiet, still"15 |
The セイ / ショウ doublet across the series is a go-on / kan-on stratum split. ショウ is the go-on layer, セイ is the kan-on layer, and both survive in different compounds.981213 請 carries a third reading シン from the tō-on stratum, visible in 申請 (しんせい, "application").14
精神を集中する。18
"Concentrate the mind."
申請を出しました。19
"I submitted an application."
静 is the most marginal member of the series. Wiktionary lists its composition as ⿰青争 without explicitly naming 青 as the phonetic. Traditional 漢和辞典 do treat 青 as the onpu on the strength of the shared セイ reading.15 The safe formulation is "静 contains 青 and reads セイ" rather than "静 is a 青-phonetic kanji."
Toyoda, Firdaus, and Kano measured the conditional reliability of 青 as an onpu across 200 jōyō tokens. セイ accounts for 55%, ショウ for 38.5%, and three other readings for the remaining 6.5%.20 The combined "セイ or ショウ" coverage is over 93%. So the prediction "the right side is 青, read it with a sibilant plus ei or ō vowel family" is essentially reliable. The precise tone-layer choice is not.
The 反 series: ハン
反 itself reads ハン, ホン, and タン. It has kun'yomi そる and かえす, and senses "anti, oppose, reverse, return."21 As an 音符, 反 anchors a clean five-member jōyō series whose 意符 rotates through five different Kangxi heads.21
| Kanji | 意符 | On'yomi | Sense |
|---|---|---|---|
| 坂 | 土 (earth) | ハン, バン (also ベン, ホン) | "slope, hill"22 |
| 板 | 木 (wood) | ハン, バン, ヘン | "board, plank"23 |
| 版 | 片 (slice) | ハン, バン, ヘン | "edition, printing block"24 |
| 飯 | 食 (food) | ハン | "cooked rice, meal"25 |
| 販 | 貝 (cowry, money) | ハン | "to sell, trade"26 |
All five members carry ハン as the primary kan-on reading. In every case, 反 sits in tsukuri (right) position.2223242526 The series is one of the cleanest right-side defaults in jōyō kanji.
大阪に行きました。27
"I went to Osaka."
黒板に書いてください。28
"Please write on the blackboard."
朝ご飯を食べます。29
"I eat breakfast."
その本は出版されました。30
"That book was published."
販売が始まりました。31
"Sales have begun."
板 takes a voiced バン variant in 黒板 (こくばん, "blackboard").2328 The base reading is still ハン. The compound environment selects the voiced variant under the same voicing principle covered in Rendaku in Kanji Compounds: Why 紙 Becomes -gami. The kun'yomi さか (坂), いた (板), and めし (飯) are native-Japanese readings layered onto the kanji long after the on'yomi was inherited. The onpu has no purchase on them.222325
The 工 series: コウ
工 itself reads コウ (kan-on) and ク (go-on). It has kun'yomi たくみ, and senses "work, labour, worker, craft, skill."32 As an 音符, 工 anchors a five-member jōyō series. Four members share コウ as the primary kan-on reading, and the fifth supplies it.32
| Kanji | 意符 | On'yomi | Sense |
|---|---|---|---|
| 江 | 氵 (water) | コウ | "river, Yangtze; inlet"33 |
| 紅 | 糸 (silk, by extension dye) | コウ, ク | "crimson"34 |
| 虹 | 虫 (insect, by extension small creature) | コウ | "rainbow" (also kun'yomi にじ)35 |
| 攻 | 攵 (striking) | コウ | "to attack"36 |
| 功 | 力 (strength) | コウ, ク | "achievement, merit"6 |
紅 and 功 also carry the ク go-on variant, mirroring the same split visible in 工 itself.32346 The コウ / ク alternation is another go-on / kan-on result of the on'yomi historical strata.
工場で働いています。37
"I work at a factory."
紅茶を飲みます。38
"I drink black tea."
大学での専攻は何ですか。39
"What is your major at the university?"
功 is the series' structural exception. In 功, the phonetic 工 sits on the left and the semantic 力 sits on the right.6 This is the cleanest demonstration in the article that the right-side default for the onpu is a tendency, not a rule. 虹 is the cleanest demonstration that the phonetic principle scales to less common kanji. にじ is the form a learner more often meets in spoken Japanese, but 工 still does its job on the on'yomi side.35
What to look for when you spot a new series
You can apply this three-step workflow to any unfamiliar kanji that looks like a candidate 形声 character.4740
- Decompose visually. In a left-right kanji, the right-side piece is the candidate onpu. In a top-bottom kanji, the bottom piece is the candidate in some series. Treat the principle as "the non-radical, non-meaning piece" rather than a strict position rule.45
- Match against your inventory. Check whether the candidate is itself a standalone kanji you already know the on'yomi for. If yes, predict that the unfamiliar kanji shares that on'yomi.47
- Test against the dictionary. If the prediction matches, you have read the on'yomi from the onpu before meeting the kanji. If it does not, you have caught one of the failure modes covered in the next H2.411120
This procedure is the article's transferable skill. The three series above are the practice ground.
The reliability of phonetic prediction
The working accuracy: about 60%
Around 60% of jōyō kanji contain a component that meaningfully hints at the on'yomi to a modern reader.42 Component-level reliability is lower than headline coverage. Toyoda, Firdaus, and Kano measured the 青 onpu across 200 jōyō tokens at 55% for セイ, 38.5% for ショウ, and 6.5% for other readings. The aggregated per-onpu average across the inventory lands near 40%.20
The two figures are not in conflict. The 60% counts kanji that have a meaningful onpu at all. The 40% measures how often a randomly chosen onpu's most common reading is the one the whole kanji takes.4220
The onpu is a hint that narrows the search space, not a guarantee. Mori and Nagy's 1999 study found that learners' phonetic-component strategy is independent from context-based inference. The combined strategy consistently outperforms either one alone.43
Perfect series, on-echo series, and noisy series
The reliability landscape sorts into three tiers, each with a worked example in jōyō.
Perfect series. Every member shares both the onpu shape and the on'yomi. No member has an extra on'yomi attached. The 包 series (包, 抱, 泡, 砲, 胞, 飽) and the 司 series (司, 伺, 詞, 飼, 嗣) both behave this way: every member reads ホウ or シ respectively as the primary on'yomi.54445 In morg.systems' formulation, "if you know the on'yomi of one of those kanji, you will know with 100% accuracy the on'yomi of all of those kanji in the series."5
On-echo series. Two or more jōyō kanji share both an onpu shape and an on'yomi, but the series is not as clean as the perfect tier. Some members carry extra on'yomi, and some non-jōyō members may diverge.41 Joy o' Kanji coined the term "on-echo" for this tier and reports that about 20% of jōyō kanji belong to a series with on-echoes. The canonical example is 安, 案, 按, 鮟 sharing the on'yomi アン around the 安 onpu.41
Noisy series. The onpu shape recurs, but the on'yomi varies. The prediction "this kanji reads like its onpu" fails. Most series sit in this tier, and the canonical broken-series example is handled in the next H3.
Why the prediction breaks: sound change, simplification, and folk etymology
Four failure modes account for most broken predictions.
(1) The onpu's own on'yomi shifted between go-on, kan-on, and tō-on layers. The 青 series picks up doublets like セイ / ショウ in 清, 晴, and 精. 請 adds a third tō-on reading シン.98121314 The series is not broken in any deep sense, since every member is still "in the 青 family." But no single reading covers every member.
(2) The kanji was simplified or reshaped, and the modern shape no longer transparently contains the original onpu. Some kanji underwent post-Tang reshaping in Chinese and further simplification in the 1949 Japanese 当用漢字字体表. Modern shinjitai sometimes hides the onpu that the older glyph carried transparently.2
(3) The kanji is not actually 形声. It may be 会意 (ideogrammic compound), 象形 (pictograph), or 指事 (simple ideograph), in which case the right-side piece was never a phonetic.2
(4) The kanji is a Japan-coined kokuji whose author re-used an onpu but assigned a fresh reading, or whose onpu sits in an unexpected slot. The 働 and 腺 cases in the next H3 show the opposite pattern, but this failure case still exists in the broader kokuji inventory.
Modern coinages keep the rule visible
The phonetic-component principle is productive, not archaeological. Newly coined Sino-Japanese kanji explicitly inherit their on'yomi from a phonetic component already in the inventory.48
働 ("to work") is a kokuji coined in Japan. It has a phono-semantic structure of semantic 亻 (person) + phonetic 動.49 The on'yomi ドウ is taken from the kan'yō-on of 動. Wikipedia's "On'yomi" article phrases the inheritance plainly: "the character 働 ('to work'), which has the kun'yomi hatara(ku) and the on'yomi dō … in both cases, these come from the on'yomi of the phonetic component, respectively dō (動)."484950
腺 ("gland") is a kokuji coined by Udagawa Genshin in the early 19th century as a medical translation term.4851 Its on'yomi セン is taken from the kan-on of 泉, the standalone kanji for "spring, fountain."485152
Both cases are deliberate Japanese-side applications of the onpu principle. The inventor of the new kanji chose a known phonetic component to give the new kanji a predictable Sino-Japanese reading. This is the strongest evidence in the article that the 音符 logic is a working system, not a historical curiosity.484951
Good to know
The phonetic only signals the on'yomi, never the kun'yomi
A kanji's kun'yomi is the native Japanese word later attached to the meaning; the onpu has no purchase on it. The kun'yomi さか (坂), いた (板), めし (飯), and にじ (虹) are independent native-Japanese readings. They were layered onto each kanji long after the on'yomi was inherited.212223253235
Reading every right-side piece as a phonetic
明 is a left-right kanji whose right-side piece is 月. But predicting that 明 reads ゲツ is wrong. 明 is 会意, an ideogrammic compound of 日 "sun" plus 月 "moon." It is not 形声, and neither piece is the phonetic.2 明 reads メイ and ミョウ from an independent etymology. The "right side is the onpu" rule applies to 形声 kanji, not to every left-right kanji.
Surname-and-given-name framing for onpu and ifu
The phonetic component is the kanji's surname: it tells you the reading family the character belongs to. The semantic component is the given name: it tells you which family member this kanji is. The Kanji Code uses this framing in plain English to teach the onpu / ifu pair.47
音符 is dictionary metavocabulary, not conversational vocabulary
Native speakers outside a teaching context are unlikely to use 音符 to refer to the phonetic component of a kanji. The everyday reading of 音符 is "musical note."1 In a kanji-formation context, a Japanese teacher or 漢和辞典 foreword is more likely to reach for the kanji-category name 形声 than to isolate the 音符 by name.2
High-leverage onpu are themselves common kanji
青, 反, 工, 寺, 包, 司, 方, and 白 all appear as standalone jōyō kanji and as productive onpu across multiple derived kanji.921324445 Learning these "high-leverage" kanji first is the main productivity argument for an onpu-first study order.
Prevalence is not reliability
A common claim like "80% of kanji are 形声, so 80% of kanji let you predict the on'yomi" mixes up two different numbers. Roughly 80% of jōyō kanji originated as 形声 compounds;24 of those, only about 60% carry an onpu that meaningfully hints at the on'yomi to a modern reader;42 and inside that 60%, component-level reliability averages around 40% per onpu.4220 Prevalence and reliability sit on two different axes.
See also
- Reading 生: The Kanji With Over 150 Attested Readings
- Top 50 Kanji Radicals by Frequency: The 70% Coverage List for Jōyō Kanji
- Jukujikun (熟字訓): When a Compound's Reading Is Assigned to the Whole Word, Not the Kanji
- Ateji (当て字): Kanji Chosen for Sound, Not Meaning
- Rendaku: When K Becomes G in Compound Words