Onyomi vs Kunyomi: Why Kanji Have Multiple Readings

The kanji 山 ("mountain") can be read as yama or as san. On its own, it's 山 (やま). In a word like 富士山, it's ふじさん. Same character, two different pronunciations — and this isn't an exception. Nearly every kanji has multiple readings, and knowing which one to use is one of the biggest early hurdles in Japanese.

This article explains why that happened, what the two kinds of readings actually mean, and how to pick the right one when you meet a word for the first time.

Where the Two Readings Came From

Before Japan imported Chinese characters, Japanese already had its own spoken words. When kanji were borrowed from China (starting around the 5th–8th centuries CE), the Japanese didn't throw out their native vocabulary — they kept both.

• Kunyomi (訓読み) — "meaning reading." The native Japanese word that was already in use for that concept, attached to the imported kanji. The kunyomi of 山 is yama, which is simply the pre-existing Japanese word for mountain.
• Onyomi (音読み) — "sound reading." An approximation of the original Chinese pronunciation at the time of borrowing. The onyomi of 山 is san, derived from the Middle Chinese reading shān.

Because Chinese characters were borrowed in several waves — and the Chinese pronunciation itself kept changing — many kanji ended up with multiple onyomi: an older one, a newer one, and sometimes a specialized one used in Buddhist terminology. On top of that, a single kanji might represent multiple native Japanese words, giving multiple kunyomi too. The result is messy, but the system behind it is logical.

The Rule That Covers Most Cases

Despite the complexity, one pattern explains the majority of kanji readings you'll encounter:

In other words: when a kanji appears alone or with hiragana attached (called okurigana — the hiragana tail that follows the kanji), you usually use the Japanese native word. When two or more kanji sit together as a compound (called a jukugo — a word made by sticking two or more kanji together, 熟語), you usually use the Chinese-derived reading. This pattern alone will steer you correctly for a large share of everyday vocabulary.

Side-by-Side Examples

Notice the common pattern. Standalone 山 is yama, but inside 火山 it's -zan (onyomi, slightly voiced). Standalone 食べる uses okurigana and kunyomi; the compound 食事 uses onyomi.

Why Onyomi Readings Sound So Short

Onyomi readings are usually one or two beats (linguists call these "morae" — roughly the length of one kana character): san, sui, jin, shoku. That's because they're approximating a single Chinese syllable. In Japanese they often take forms like these:

• Ending in -n: 山 さん, 本 ほん, 人 じん
• Ending in a long vowel (-ou, -uu, -ei, -ai): 校 こう, 空 くう, 生 せい, 大 たい
• Ending in -ki, -ku, -chi, -tsu: 食 しょく, 日 にち, 月 げつ

Kunyomi, by contrast, can be any length — because they're native Japanese words. 湖 みずうみ (lake, 4 beats) is one kanji with a long kunyomi.

Common Exceptions to the Rule

Even strong rules in Japanese have exceptions. Being aware of them is part of becoming comfortable with readings.

Kun-kun compounds

Some kanji compounds use both kanji in their kunyomi form, especially for native concepts.

• 手紙 (てがみ) — letter (hand + paper, kun + kun)
• 雨空 (あまぞら) — rainy sky
• 昼寝 (ひるね) — nap

On-kun or kun-on compounds (traditionally called jūbakoyomi / yutōyomi)

These are mixed-reading compounds — one kanji takes its onyomi, the other its kunyomi. The traditional names literally mean "lunchbox reading" and "hot-water reading," and are just old labels for this pattern. You don't need to memorize the names — just know these mixed readings exist.

• 台所 (だいどころ) — kitchen (on + kun)
• 朝晩 (あさばん) — morning and evening (kun + on)

These exist, but they're a minority. The default assumption for compound = onyomi is still your best starting bet.

Ateji (当て字) — kanji chosen for sound only

Some words use kanji purely for their sound, ignoring the usual meaning rules. 寿司 (すし) (sushi) is a classic example — the kanji were picked because they sounded like "sushi," not because of what they mean. These words are learned as fixed vocabulary.

Rendaku — voicing changes in compounds

When a kanji attaches to another in a compound, its first consonant sometimes voices: k→g, s→z, t→d, h→b. This happens to both onyomi and kunyomi.

• 時 alone: とき (time). In 時々: ときどき — toki becomes -doki.
• 人 onyomi jin, but in 三人: さんにん — the second kanji's onyomi slot takes nin.

Rendaku isn't fully predictable, but it's regular enough that you'll start to hear when it should and shouldn't happen.

How to Actually Study Readings

Learn readings through words, not through kanji in isolation

A beginner-level kanji card that lists "生: せい・しょう・いきる・なま・うまれる・はえる・き" will feel impossible. You don't need to memorize those readings as a flat list. Learn 学生 がくせい (student), 先生 せんせい (teacher), 生まれる うまれる (to be born), and 生ビール なまビール (draft beer) — each as a word. The readings arrive attached to meanings, which is how your brain actually keeps them.

Trust the rule of thumb for new vocabulary

When you encounter a kanji compound you've never seen, your first guess should be: use the onyomi for each kanji. You'll be right most of the time. If a word has a kanji followed by hiragana, assume kunyomi.

Prioritize the frequent reading, not the rare one

Many kanji technically have 5+ readings, but 1–2 account for the overwhelming majority of real usage. Learning the top reading first means you can read most texts you encounter; the rare readings can wait.

Expect to confirm, not to guess perfectly

Even advanced learners sometimes misread an unfamiliar name or technical term. The goal isn't flawless prediction — it's a strong default guess and a habit of checking when uncertain. A dictionary lookup or hover translator is a normal part of reading Japanese, not a learning failure.