On'yomi vs. Kun'yomi: Rules and Patterns to Predict Kanji Readings on the JLPT

Japanese kanji characters possess two distinct types of readings: Chinese-derived on'yomi (音読み) and native Japanese kun'yomi (訓読み). For multiple-choice JLPT exams from N5 through N1, determining which reading applies to an unfamiliar word often separates passing scores from failing ones. Fortunately, reading choices are governed by reliable phonetic and structural patterns: multi-kanji compound words usually take on'yomi, trailing hiragana indicates kun'yomi, and shared phonetic radicals allow you to accurately predict pronunciations. This guide breaks down the core rules and exceptions.

On'yomi and Kun'yomi at a glance

When the Japanese adopted Chinese writing characters between the 5th and 8th centuries, ancient Japan had a spoken language (Yamato kotoba) but no indigenous written script. Consequently, every imported character inherited two sound systems:

1. On'yomi (音読み — "Sound reading"): Approximations of how the character sounded in classical Chinese at the time of borrowing.
2. Kun'yomi (訓読み — "Meaning reading"): Native Japanese words attached to the Chinese character to represent its underlying meaning.

On the JLPT, the first question set of the Language Knowledge section tests whether you can select the exact correct reading for an underlined word in a sentence.

Rule 1: The Compound Word Rule (Jukugo = On'yomi)

When two or more kanji sit together side by side without any intervening hiragana, the word is a compound (熟語, jukugo). More than 85% of compounds use on'yomi readings for all characters.

• 学生 (gakusei — student): 学 (GAKU) + 生 (SEI)
• 電話 (denwa — telephone): 電 (DEN) + 話 (WA)
• 病院 (byōin — hospital): 病 (BYŌ) + 院 (IN)
• 案内 (annai — guidance): 案 (AN) + 内 (NAI)
• 経済 (keizai — economics): 経 (KEI) + 済 (ZAI)

JLPT Test Tip: If you see an unfamiliar two-kanji compound in an exam passage, eliminate multiple-choice options that use long native Japanese words. The answer is almost always a crisp on'yomi sound.

Rule 2: The Okurigana Rule (Trailing Kana = Kun'yomi)

When a kanji is followed by trailing hiragana (送り仮名, okurigana), it represents an inflecting Japanese verb or adjective. These words almost always use the native kun'yomi reading.

• 食べる (ta-beru — to eat): 食 uses the kun'yomi ta
• 飲む (no-mu — to drink): 飲 uses the kun'yomi no
• 新しい (atara-shii — new): 新 uses the kun'yomi atara
• 静かな (shizu-kana — quiet): 静 uses the kun'yomi shizu
• 忙しい (isoga-shii — busy): 忙 uses the kun'yomi isoga

The stem kanji carries the core meaning, while the trailing hiragana conjugates for tense, politeness, and conditionality (〜た, 〜ない, 〜ば).

Rule 3: The Standalone Noun Rule (Single Character = Kun'yomi)

When a single kanji stands entirely by itself as a noun with no trailing okurigana and no adjacent kanji, it generally uses its native Japanese kun'yomi:

• 山 (yama — mountain) vs. 富士山 (fuji-san, Mount Fuji, on'yomi)
• 川 (kawa — river) vs. 河川 (kasen, rivers/waterways, on'yomi)
• 空 (sora — sky) vs. 空気 (kūki, atmosphere/air, on'yomi)
• 雨 (ame — rain) vs. 降雨 (kōu, rainfall, on'yomi)
• 魚 (sakana — fish) vs. 水族館 (suizokukan, aquarium, on'yomi)

There are a few exceptions for formal concepts borrowed directly from Chinese (such as 本 hon, book; 肉 niku, meat; 門 mon, gate), but the overwhelming majority of single-character words in daily speech use kun'yomi.

Rule 4: The Phonetic Radical Hack

This is the most powerful shortcut for intermediate and advanced JLPT candidates (N3 to N1). In Chinese character formation, roughly 80% of kanji are phono-semantic compounds (形声文字, keiseimoji):

• One part (the radical) provides the general meaning.
• The remaining part provides the on'yomi pronunciation.

Once you recognize the phonetic component, you can predict the on'yomi of dozens of kanji you have never studied:

Example: The 寺 (JI) family

The character 寺 (ji — temple) acts as a phonetic anchor:

• 持 (扌 hand + 寺): On'yomi is JI → 持参 (jisan, bringing)
• 時 (日 sun + 寺): On'yomi is JI → 時間 (jikan, time)
• 待 (彳 step + 寺): On'yomi is TAI / JI → 期待 (kitai, expectation)
• 詩 (言 words + 寺): On'yomi is SHI / JI → 詩人 (shijin, poet)

Example: The 青 (SEI) family

The character 青 (sei — blue/green):

• 清 (氵 water + 青): On'yomi is SEI → 清潔 (seiketsu, clean)
• 晴 (日 sun + 青): On'yomi is SEI → 晴天 (seiten, fair weather)
• 静 (青 + 争 struggle): On'yomi is SEI → 静止 (seishi, standstill)
• 精 (米 rice + 青): On'yomi is SEI → 精神 (seishin, spirit/mind)

When sitting an upper-level reading section, recognizing phonetic components prevents panic when facing unfamiliar N2 or N1 characters.

Sound modifications on the JLPT

Even when words follow the rules, sound shifts happen at character boundaries. Test creators intentionally test these phonetic changes:

1. Sequential Voicing (Rendaku — 連濁)

When two native words form a compound, the initial consonant of the second element often becomes voiced (濁音):

• k → g (木 ki + 陰 kage = 木陰 kokage, tree shade)
• t → d (手 te + 紙 kami = 手紙 tegami, letter)
• h/f → b (花 hana + 火 hi = 花火 hanabi, fireworks)

2. Gemination (Sokuonbin — 促音便)

When an on'yomi ending in つ (tsu) or ち (chi) meets a character beginning with k, s, t, or h, the final syllable contracts into a small っ (tsu):

• 学 (GAKU) + 校 (KŌ) = 学校 (gakkō, not gakukō)
• 出 (SHUTSU) + 発 (HATSU) = 出発 (shuppatsu, not shutsuhatsu)
• 日 (NICHI) + 記 (KI) = 日記 (nikki, not nichiki)

The Exceptions: Jukujikun and Mixed Readings

The JLPT frequently includes "trap" words that violate the standard compound rule:

1. Jukujikun (熟字訓): Words where the kanji compound as an entire unit is assigned a native Japanese word, regardless of the individual character readings:
   • 今日 (kyō — today, not kon-nichi)
   • 明日 (ashita — tomorrow, not mei-nichi)
   • 昨日 (kinō — yesterday, not saku-jitsu)
   • 部屋 (heya — room, not bu-oku)
   • 大人 (otona — adult, not dai-jin)
2. Yutō-yomi (湯桶読み — On + Kun): Compounds where the first character uses on'yomi and the second uses kun'yomi:
   • 場所 (ba-sho — place: ba is kun'yomi, sho is on'yomi)
   • 夕刊 (yū-kan — evening edition: yū is kun'yomi, kan is on'yomi)

How to study readings effectively

1. Never study isolated reading lists: Memorizing that 生 has twenty readings without knowing which words use them is wasted effort.
2. Learn one compound per reading: Anchor the on'yomi to a core noun (先生 for 生) and the kun'yomi to the primary verb (生きる).
3. Verify with timed mocks: Test your ability to identify correct readings under pressure by practicing with a timed mock exam.

FAQ

Why do some kanji have multiple on'yomi? Because Japan borrowed words from China across different historical dynasties and regions:

• Go-on (呉音): Borrowed during the 5th–6th centuries (common in Buddhist terms, e.g., 明 = myō).
• Kan-on (漢音): Borrowed during the Tang Dynasty in the 7th–9th centuries (the standard academic reading, e.g., 明 = mei).
• Tō-on (唐音): Borrowed during the Song/Ming dynasties (e.g., 明 = min in 明朝 minchō).

How can I tell if a reading is on'yomi or kun'yomi in a dictionary? Japanese dictionaries conventionally write on'yomi in Katakana (セイ, ショウ) and kun'yomi in Hiragana (い.きる, う.まれる). The dot in kun'yomi entries indicates where the okurigana begins.

Do names follow standard on'yomi/kun'yomi rules? Japanese family and given names (nanori, 名乗り) frequently use special archaic readings that do not follow standard compound rules (e.g., 東 = Azuma, 一 = Hajime). However, the JLPT does not test obscure personal name readings.

What is the easiest way to recognize on'yomi syllables? On'yomi syllables almost never have three or more mora. Furthermore, any reading containing a contracted glide (kya, kyo, sha, shu, cho, ryo) or ending in -n, -tsu, -ku, -ki, or -chi is almost certainly an on'yomi reading.

Keep studying

• How many kanji for each JLPT level
• JLPT vocabulary spaced-repetition plan
• The JLPT N5 kanji list
• JLPT N4 vs N3: what changes between levels
• How the JLPT is scored: scaled scoring explained

Official sources checked

• JLPT test item categories and linguistic competence — official item specifications for Language Knowledge (Kanji Reading and Orthography).
• JLPT sample questions for Language Knowledge — official question patterns for Kanji reading multiple-choice items.

Vocabulary and kanji totals are study estimates, not official JLPT syllabus counts.