Chapter 4 — Kanji: How the System Works

Most learners approach kanji as a memorization problem — thousands of arbitrary symbols to be drilled into submission. This framing is wrong, and it leads to years of unnecessary frustration. Kanji are not arbitrary. They are a compositional system built from a limited set of reusable parts, with patterns that predict both meaning and pronunciation. This chapter lays out those patterns so that every kanji you encounter from this point forward arrives as a structure you can analyze, not a shape you must memorize blind.

4.1 What Kanji Are

Kanji are sometimes introduced through their pictographic origins — 山 "looks like" a mountain, 川 "looks like" a river. This is true for a handful of characters, but it is deeply misleading as a general principle. The vast majority of kanji in use today bear no visual resemblance to the thing they represent, and they have not looked pictographic for over two thousand years. If you approach each new kanji expecting to find a picture, you will fail quickly and blame yourself for it.

Here is what kanji actually are: a compositional system. Most kanji are not single indivisible symbols. They are combinations of smaller components — pieces that recur across dozens or hundreds of different characters. The character 語 (language), for instance, contains 言 (speech) on its left side and 吾 on its right, which itself contains 五 (five) and 口 (mouth). Knowing the components lets you distinguish 語 from 話 (conversation), which shares the 言 component but pairs it with 舌 (tongue) rather than 吾.

Three facts about kanji to carry forward:

Most kanji are built from reusable components. There are roughly 200 common components. Once you know them, new kanji become combinations of familiar pieces rather than entirely new shapes.

Kanji carry both meaning and sound. Every kanji has at least one meaning and at least one pronunciation. Most have multiple pronunciations — the Chinese-derived reading (おんよみ) and the native Japanese reading (くんよみ), which are used in different contexts. Section 4.4 covers this in detail.

The 常用漢字 (じょうようかんじ) list contains 2,136 characters. This is the set designated for everyday use in newspapers, official documents, and general publishing. It is a large number, but you will learn them gradually over several years, not all at once. This chapter introduces 25.

4.2 Radicals (部首)

Every kanji is classified under one 部首 (ぶしゅ) — a radical. The radical system is a categorization tool developed for dictionaries: if you want to look up a kanji you have never seen, you identify its radical, count its remaining strokes, and find it in the dictionary under that radical's section. In the age of electronic dictionaries, this lookup function matters less. What still matters enormously is the meaning-hint function.

Radicals frequently indicate the semantic category of a kanji. A kanji with 氵(the water radical) on its left side is very likely related to water or liquid: 海 (sea), 池 (pond), 湖 (lake), 泳 (swim), 洗 (wash). A kanji with 忄 (the heart radical) typically involves emotion or mental state: 情 (emotion), 悲 (grief), 快 (pleasant). This is not a guarantee — exceptions exist — but the correlation is strong enough to be a reliable first guess.

The table below presents 20 of the most common radicals. Learn to recognize them on sight. When you encounter an unfamiliar kanji, identifying its radical is the first analytical step.

The column labeled 変形 (へんけい) shows the compressed form a radical takes when it appears as part of a larger character. 水 becomes 氵 when squeezed into the left side of a kanji. 人 becomes 亻. 心 becomes 忄. These compressed forms are visually distinct from the standalone characters, but once you learn to associate them, recognition becomes automatic.

The column labeled 位置 (いち) indicates where the radical typically appears. "Left side" radicals (called へん) are the most common pattern — the radical sits on the left, and the other component sits on the right.

4.3 Phonetic Components

If radicals hint at meaning, phonetic components hint at sound. This is one of the most useful patterns in the kanji system, and it is routinely undertaught.

Consider the character 青 (blue/green), which has the おんよみ せい. Now look at these characters:

Every one of these characters contains 青 as a component, and every one shares the おんよみ せい. The radical changes — giving each character its distinct meaning — but the phonetic component 青 is preserved, carrying the reading with it.

This pattern holds across hundreds of kanji. The component 工 (こう) appears in 空 (くう, sky), 紅 (こう, crimson), and 功 (こう, merit). The component 各 (かく) appears in 格 (かく, status), 閣 (かく, cabinet), and 客 (きゃく — close, though not identical).

Two important caveats:

The pattern is not perfectly reliable. Some characters containing 青 do not read せい. Phonetic components are a tendency, not a rule. They are useful as a reading strategy — when you encounter an unfamiliar compound, checking for a known phonetic component gives you a reasonable guess — but you must be prepared for exceptions.

At this stage, awareness is sufficient. You do not need to memorize phonetic series. Simply notice when you encounter two kanji that share a component and share a reading. Over time, your brain will accumulate these patterns naturally. Detailed phonetic-component analysis is covered in Stage 2.

4.4 おんよみ and くんよみ

Every kanji entered Japanese with at least one Chinese-derived pronunciation and was then mapped onto at least one native Japanese word. These two layers of reading are called おんよみ (音読み, literally "sound reading") and くんよみ (訓読み, literally "meaning reading"). Understanding when each applies is essential for reading.

おんよみ (音読み)

The おんよみ is a Japanese approximation of the original Chinese pronunciation. It does not sound like modern Chinese — the borrowing happened over a thousand years ago, through Japanese phonology. おんよみ are typically one or two morae long: さん, がく, にち, きん.

おんよみ are used primarily in 漢語 (かんご) — compound words made of two or more kanji. These compounds follow the Chinese pattern of combining meaningful elements into larger words:

• 大 (だい, big) + 学 (がく, study) → 大学 (だいがく, university)
• 日 (にち, sun/day) + 本 (ほん, origin) → 日本 (にほん, Japan)
• 水 (すい, water) + 曜 (よう, weekday) + 日 (び, day) → 水曜日 (すいようび, Wednesday)

When you see two or more kanji pressed together without hiragana between them, the default expectation is おんよみ.

くんよみ (訓読み)

The くんよみ is the native Japanese word that was assigned to the kanji. くんよみ tend to be longer than おんよみ — two, three, or even four morae — and they often have おくりがな (trailing hiragana that marks the inflectional ending):

• 大きい (おおきい) — the kanji 大 carries the root meaning, and きい is the おくりがな
• 小さい (ちいさい) — 小 carries the root, さい is おくりがな
• 休む (やすむ) — 休 carries the root, む is おくりがな

When a kanji stands alone (with or without おくりがな), the default expectation is くんよみ:

• 山 (やま) — standalone, くんよみ
• 川 (かわ) — standalone, くんよみ
• 火 (ひ) — standalone, くんよみ

The General Rule (and Its Exceptions)

In compounds → おんよみ. Standalone → くんよみ.

This rule is correct often enough to be useful and wrong often enough to require vigilance. Major exceptions include:

• 熟字訓 (じゅくじくん): compound words read with くんよみ rather than おんよみ. The word 今日 is read きょう, not こんにち (though こんにち exists in こんにちは). The word 大人 is read おとな, not だいじん. These must be learned individually.
• Names: place names and personal names follow their own reading conventions, which are often unpredictable.
• Mixed compounds: some words use おんよみ for one kanji and くんよみ for another. These are called 重箱読み (じゅうばこよみ, おん + くん) and 湯桶読み (ゆとうよみ, くん + おん).

For now, apply the general rule and learn the exceptions as they come. Context will usually resolve ambiguity — if you know the word, you know the reading.

4.5 First Kanji Set: Numbers 一〜十

The number kanji are among the simplest in the system — many consist of just one to four strokes. They are also among the most frequently encountered. Learn to recognize them instantly.

一　いち / ひと(つ)

部首: 一 (いち, one) — this character is itself a radical.

おんよみ: いち

くんよみ: ひと(つ)

Compounds:

• 一人 (ひとり) — one person ※特別な読み方
• 一月 (いちがつ) — January
• 一つ (ひとつ) — one (thing)
• 一日 (ついたち) — first of the month ※特別な読み方

The simplest kanji: a single horizontal stroke. As a component in other kanji, it often appears as a horizontal line at the top or bottom.

二　に / ふた(つ)

部首: 二 (に)

おんよみ: に

くんよみ: ふた(つ)

Compounds:

• 二人 (ふたり) — two people ※特別な読み方
• 二月 (にがつ) — February
• 二つ (ふたつ) — two (things)
• 二日 (ふつか) — second of the month

Two horizontal strokes. The top stroke is shorter than the bottom.

三　さん / み(つ)

部首: 一

おんよみ: さん

くんよみ: み(つ)

Compounds:

• 三人 (さんにん) — three people
• 三月 (さんがつ) — March
• 三つ (みっつ) — three (things)
• 三日 (みっか) — third of the month

Three horizontal strokes. The middle stroke is shortest, the bottom longest.

四　し / よん / よ(つ)

部首: 囗 (くにがまえ, enclosure)

おんよみ: し

くんよみ: よん、よ(つ)

Compounds:

• 四月 (しがつ) — April
• 四つ (よっつ) — four (things)
• 四人 (よにん) — four people
• 四日 (よっか) — fourth of the month

Note: し is the same sound as 死 (し, death). In many contexts — hotel room numbers, hospital floors, gift-giving — よん is preferred to avoid the association. This avoidance is cultural, not grammatical, but it is pervasive.

五　ご / いつ(つ)

部首: 二

おんよみ: ご

くんよみ: いつ(つ)

Compounds:

• 五月 (ごがつ) — May
• 五つ (いつつ) — five (things)
• 五人 (ごにん) — five people
• 五日 (いつか) — fifth of the month

六　ろく / む(つ)

部首: 八

おんよみ: ろく

くんよみ: む(つ)

Compounds:

• 六月 (ろくがつ) — June
• 六つ (むっつ) — six (things)
• 六人 (ろくにん) — six people
• 六日 (むいか) — sixth of the month

七　しち / なな(つ)

部首: 一

おんよみ: しち

くんよみ: なな(つ)

Compounds:

• 七月 (しちがつ) — July
• 七つ (ななつ) — seven (things)
• 七人 (しちにん / ななにん) — seven people
• 七日 (なのか) — seventh of the month

Note: しち can be confused with いち (one) in rapid speech. For this reason, なな is often preferred in spoken contexts — telephone numbers, ordering, counting aloud. In fixed compounds like 七月, however, しち is standard.

八　はち / や(つ)

部首: 八 (はち)

おんよみ: はち

くんよみ: や(つ)

Compounds:

• 八月 (はちがつ) — August
• 八つ (やっつ) — eight (things)
• 八人 (はちにん) — eight people
• 八日 (ようか) — eighth of the month

The shape is two strokes spreading outward — like a roof or the top of a tent. As a component in other kanji (such as 六, 分, 公), it appears frequently.

九　く・きゅう / ここの(つ)

部首: 乙 (おつ)

おんよみ: く、きゅう

くんよみ: ここの(つ)

Compounds:

• 九月 (くがつ) — September
• 九つ (ここのつ) — nine (things)
• 九人 (くにん / きゅうにん) — nine people
• 九日 (ここのか) — ninth of the month

Note: く is the same sound as 苦 (く, suffering). Like 四/し, this leads to avoidance in certain contexts — きゅう is often substituted. The number 49 (四十九, しじゅうく) is considered particularly unlucky because it can be heard as 死十苦 — though this level of superstition varies by individual.

十　じゅう / とお

部首: 十 (じゅう)

おんよみ: じゅう

くんよみ: とお

Compounds:

• 十月 (じゅうがつ) — October
• 十日 (とおか) — tenth of the month
• 十人 (じゅうにん) — ten people
• 十分 (じゅっぷん / じっぷん) — ten minutes

A cross shape — one horizontal stroke and one vertical. Simple to recognize, and it appears as a component in characters like 早 (early), 古 (old), and 千 (thousand).

4.6 Second Kanji Set: Days and Nature

These nine kanji include the seven days of the week (曜日 names) and two geography characters. You will encounter them constantly.

日　にち・じつ / ひ・か

部首: 日 (にち, sun)

おんよみ: にち、じつ

くんよみ: ひ、か (in dates)

Compounds:

• 日本 (にほん) — Japan
• 日曜日 (にちようび) — Sunday
• 今日 (きょう) — today ※特別な読み方
• 毎日 (まいにち) — every day
• 一日 (ついたち) — first of the month ※特別な読み方

This is one of the highest-frequency kanji in the language. It means "sun" and "day," and it appears in dates, days of the week, and countless compounds. The character is a rectangle with a horizontal line through the middle — originally a depiction of the sun.

As a radical, 日 appears on the left side of characters related to time and light: 明 (bright), 時 (time), 暗 (dark), 暑 (hot weather).

月　げつ・がつ / つき

部首: 月 (つき, moon)

おんよみ: げつ、がつ

くんよみ: つき

Compounds:

• 月曜日 (げつようび) — Monday
• 一月 (いちがつ) — January
• 今月 (こんげつ) — this month
• 月 (つき) — moon

げつ and がつ are both おんよみ, used in different compound contexts: がつ for naming months (一月, 二月...), げつ for other compounds (月曜日, 今月). This distinction is lexical — you learn it through vocabulary, not through a rule.

As a radical, 月 appears in characters related to the body (肉 "meat" and 月 merged historically): 腕 (arm), 胸 (chest), 脳 (brain).

火　か / ひ

部首: 火 (ひ, fire)

おんよみ: か

くんよみ: ひ

Compounds:

• 火曜日 (かようび) — Tuesday
• 花火 (はなび) — fireworks (literally "flower fire")
• 火事 (かじ) — fire (incident), conflagration
• 火山 (かざん) — volcano (literally "fire mountain")

The standalone くんよみ ひ means "fire" or "flame." In compounds, the おんよみ か is standard.

水　すい / みず

部首: 水 (みず, water) — compressed form 氵

おんよみ: すい

くんよみ: みず

Compounds:

• 水曜日 (すいようび) — Wednesday
• 水 (みず) — water
• 水泳 (すいえい) — swimming

As a radical in compressed form (氵), this is one of the most frequently seen components in the entire kanji system. Any time you see three short diagonal strokes on the left side of a character, think "water."

木　もく・ぼく / き・こ

部首: 木 (き, tree)

おんよみ: もく、ぼく

くんよみ: き、こ

Compounds:

• 木曜日 (もくようび) — Thursday
• 木 (き) — tree
• 木村 (きむら) — Kimura (surname, literally "tree village")
• 大木 (たいぼく) — large tree

One 木 is a tree. Two together, 林 (はやし), is a grove. Three together, 森 (もり), is a forest. This is one of the rare cases where the pictographic origin is genuinely helpful — the character looks like a tree with branches and roots.

金　きん・こん / かね

部首: 金 (かね, metal/gold) — compressed form 釒

おんよみ: きん、こん

くんよみ: かね

Compounds:

• 金曜日 (きんようび) — Friday
• お金 (おかね) — money
• 金 (きん) — gold
• 金魚 (きんぎょ) — goldfish

The お in おかね is a polite prefix (お + かね), not part of the kanji reading. You will see this prefix frequently with everyday nouns.

土　ど・と / つち

部首: 土 (つち, earth)

おんよみ: ど、と

くんよみ: つち

Compounds:

• 土曜日 (どようび) — Saturday
• 土地 (とち) — land
• 土 (つち) — earth, soil

A simple character — a cross with an extra horizontal stroke at the bottom. As a radical, it appears in characters related to ground and place: 地 (ground), 場 (place), 坂 (slope).

山　さん / やま

部首: 山 (やま, mountain)

おんよみ: さん

くんよみ: やま

Compounds:

• 山 (やま) — mountain
• 富士山 (ふじさん) — Mt. Fuji
• 火山 (かざん) — volcano
• 山口 (やまぐち) — Yamaguchi (place name/surname)

When appended to a mountain name, the おんよみ さん is standard: 富士山 (ふじさん). When referring to mountains generically, the くんよみ やま is used: あの山 (that mountain).

川　せん / かわ

部首: 川 (かわ, river)

おんよみ: せん

くんよみ: かわ

Compounds:

• 川 (かわ) — river
• 小川 (おがわ) — stream (literally "small river"), also a common surname
• 川口 (かわぐち) — Kawaguchi (place name)

Three vertical strokes of varying height — one of the few kanji where the pictographic origin (flowing water) is still visible.

4.7 Third Kanji Set: People

These six kanji describe people and basic attributes. They appear in elementary vocabulary from the very first lesson of any Japanese course.

人　じん・にん / ひと

部首: 人 (ひと, person) — compressed form 亻

おんよみ: じん、にん

くんよみ: ひと

Compounds:

• 日本人 (にほんじん) — Japanese person
• 一人 (ひとり) — one person ※特別な読み方
• 二人 (ふたり) — two people ※特別な読み方
• 三人 (さんにん) — three people
• 人 (ひと) — person

じん and にん are both おんよみ, used in different compounds. Generally, じん marks nationality or type of person (日本人, アメリカじん, 大人 — though 大人 is read おとな), and にん is used for counting people (三人, 四人). But this is a tendency, not a rule — 一人 and 二人 are irregular readings entirely.

大　だい・たい / おお(きい)

部首: 大 (だい, big)

おんよみ: だい、たい

くんよみ: おお(きい)

Compounds:

• 大きい (おおきい) — big (い-adjective)
• 大学 (だいがく) — university
• 大人 (おとな) — adult ※特別な読み方
• 大切 (たいせつ) — important

The character depicts a person with arms spread wide — literally "big." It is among the most common kanji in the language and appears in hundreds of compounds.

小　しょう / ちい(さい)・こ・お

部首: 小 (しょう, small)

おんよみ: しょう

くんよみ: ちい(さい)、こ、お

Compounds:

• 小さい (ちいさい) — small (い-adjective)
• 小学校 (しょうがっこう) — elementary school
• 小川 (おがわ) — stream
• 小鳥 (ことり) — small bird

The くんよみ varies: ちいさい when used as an adjective, こ as a prefix meaning "small" or "young" (小鳥, 小犬), お in certain words like 小川. These are lexical facts — learn them with the vocabulary.

子　し・す / こ

部首: 子 (こ, child)

おんよみ: し、す

くんよみ: こ

Compounds:

• 子ども (こども) — child, children
• 女子 (じょし) — girl, female
• 男子 (だんし) — boy, male
• 子 (こ) — child

The character is also extremely common in female given names (よしこ, けいこ, etc.), where it carries the reading こ.

女　じょ / おんな

部首: 女 (おんな, woman)

おんよみ: じょ

くんよみ: おんな

Compounds:

• 女の人 (おんなのひと) — woman
• 女子 (じょし) — girl, female
• 女性 (じょせい) — woman (formal)

When used standalone to mean "woman," the reading is おんな. In compounds, the おんよみ じょ is standard.

As a radical, 女 appears in kanji related to relationships and family: 好 (like), 妹 (younger sister), 姉 (older sister), 嫌 (dislike).

男　だん・なん / おとこ

部首: 田 (た, rice field)

おんよみ: だん、なん

くんよみ: おとこ

Compounds:

• 男の人 (おとこのひと) — man
• 男子 (だんし) — boy, male
• 男性 (だんせい) — man (formal)

The character is composed of 田 (rice field) and 力 (power) — the person who works the fields with strength. Note that the radical is 田, not 力, which is a reminder that radical assignment is a classification convention, not always an intuitive semantic match.

4.8 連濁 (れんだく): Sequential Voicing in Compounds

When two words are joined to form a compound, the first consonant of the second word often voices. This is called 連濁 (れんだく). It is one of the most pervasive sound changes in Japanese, and a learner who is unaware of it will systematically mishear compounds and fail to find them in dictionaries.

The pattern is simple to state. A voiceless consonant at the start of the second element becomes its voiced counterpart:

The shift from は to ぱ deserves special note. After a small っ (a geminate), the consonant が or ば is impossible — the voicing target is ぱ instead. This is the same phonotactic rule that makes counter readings shift in Chapter 7 (一本 いっぽん, 三本 さんぼん, 何本 なんぼん).

You have already encountered rendaku without it being named. In this chapter:

• 花火 (はなび) — はな + ひ. The ひ voiced to び.
• 火山 (かざん) — か + やま would be the bare reading, but compounded the おんよみ pair is か + さん, and the さ voiced to ざ. (Sino-Japanese compounds rendaku less, but still do.)
• 小川 (おがわ) — お + かわ. The か voiced to が.
• 金魚 (きんぎょ) — きん + ぎょ rather than きん + きょ. The き voiced to ぎ.

Why This Matters for Comprehension

If you encounter the word はなび in speech and expect 花 (はな) + 火 (ひ) without rendaku, you will not recognize it. You will hear "hanabi" and look for a word starting with hi, find nothing, and assume you do not know it. The same problem affects te-form compounds, place names, and any compound word you encounter for the first time.

The cure is awareness, not memorization. When a compound surprises you with a voiced consonant, recognize that the underlying word has the voiceless form. はなび → はな + ひ. てがみ → て + かみ. たけのこ (bamboo shoot) → たけ + の + こ — and the こ is the same こ as in ki-no-こ (mushroom), where rendaku also operates.

When Rendaku Does Not Apply

Rendaku is a tendency, not a rule, and it has well-known restrictions. You do not need to memorize these now — but recognizing the pattern is useful when you hear a compound and the second element does not voice.

Lyman's Law. If the second element already contains a voiced obstruent (が, ぎ, ぐ, げ, ご, ざ, じ, ず, ぜ, ぞ, だ, ぢ, づ, で, ど, ば, び, ぶ, べ, ぼ), rendaku is blocked. The classic example is 大風 (おおかぜ, gale) — か does not voice to が because かぜ already contains ぜ. Likewise 紙くず (かみくず, paper scrap), not かみぐず, because くず already contains ず. The intuition is that Japanese phonology resists having two voiced obstruents inside one phonological word.

Foreign loanwords (gairaigo). Words borrowed from English, Portuguese, etc., generally resist rendaku. アイス + コーヒー stays アイスコーヒー, never アイスゴーヒー.

Sino-Japanese compounds (kango) rendaku less reliably than native compounds. 国家 (こっか, nation), 一個 (いっこ, one item), 半開 (はんかい, half-open) — kango compounds often preserve the voiceless consonant.

Coordinate compounds (where the two elements are equal partners, "A and B" rather than "A's B") tend to block rendaku: 山川 (やまかわ, mountains and rivers), 草木 (くさき, grass and trees) — though some idiomatic ones do voice: 親子 (おやこ, parent and child).

Recognition Habit

Practical rule: when you encounter a word in writing and the kanji you recognize would normally read with a voiceless consonant at the start of the second element, but the furigana shows a voiced one, that is rendaku. When you hear an unknown word that begins promisingly but ends with an unexpected voiced consonant in the middle, mentally try the unvoiced version of the second element and see if you recognize the morpheme.

Rendaku will become automatic with exposure. The first hundred compound words you learn account for the bulk of the rendaku you will ever need to recognize, because they are also the most frequent words in the language. The goal of this section is to name the phenomenon so you can recognize it as a regularity rather than a string of arbitrary exceptions.

4.9 Recognition Strategies

You now have 25 kanji. Before they become 50, then 100, then 500, establish the analytical habits that will scale.

Strategy 1: Radical → Meaning Category

When you encounter an unfamiliar kanji, identify the radical first. If you see 氵 on the left, expect a water-related meaning. If you see 忄, expect an emotion or mental state. If you see 言/訁, expect something related to speech or language. This will not give you the exact meaning, but it narrows the field dramatically.

Strategy 2: Phonetic Component → Possible Reading

If the right side of a kanji is a component whose おんよみ you know, try applying that reading. If you know 青 is せい and you encounter 清 for the first time, guessing せい is reasonable — and in this case, correct. This strategy fails often enough that you should hold your guess loosely, but it succeeds often enough to be worth trying.

Strategy 3: Context → Disambiguation

The kanji 生 has more than ten readings. In isolation, that sounds impossible to manage. In practice, it is not — because you never encounter 生 in isolation. You encounter it in words: 先生 (せんせい, teacher), 生まれる (うまれる, to be born), 学生 (がくせい, student), 生きる (いきる, to live). You learn the reading as part of the word, not as a property of the character. When you see 先生 in a sentence, you do not stop to wonder which of ten readings applies — you recognize the word.

Strategy 4: Learn Readings Through Vocabulary

Do not sit down with a list of kanji readings and try to memorize them in the abstract. This is inefficient and discouraging. Instead, learn words. Every time you learn a new word, you implicitly learn a reading for each kanji in it. After encountering 日 in 日本, 日曜日, 今日, 毎日, 一日, and 日本語, you will know its readings not because you memorized a list but because you have seen them in context repeatedly. This is how native speakers acquire kanji readings, and it is how you should too.

Vocabulary List

All kanji and compounds introduced in this chapter, with pitch accent notation.

Numbers 一〜十

Days and Nature

People