Japanese Pitch Accent Patterns for Everyday Words

The four patterns and what the particle does

Standard Tokyo pitch accent works in morae, the short beats that Japanese words are built from, and each mora is either high or low. Every word has at most one drop from high to low, and the four pattern names simply describe where that drop sits. If you want the concept explained from scratch, the beginners' pitch accent guide on this blog covers that; here the focus is on the patterns themselves and the everyday words that share them.

In the table, capital letters mark high morae and lower-case letters mark low ones, and the same convention is used throughout this post. The particle column is the important one. Two of the four patterns sound identical until a particle follows, so the particle is where you prove which pattern a word belongs to.

Heiban and odaka are the pair to watch. Sakura and hashi meaning bridge both start low and rise, and on their own they sound the same kind of word. Add ga and they separate: sakura ga stays up, hashi ga falls on the ga. Learn every noun with a particle attached and this problem never gets a chance to grow.

Hearing the single drop

Learners often try to hear every rise and fall, which is exhausting and unnecessary. Tokyo Japanese only ever rises in one place, between the first and second mora, and it only ever drops once. So there is exactly one question to ask about a new word: does the pitch fall, and if so, after which mora? Everything else about the contour follows automatically from that answer.

The easiest way to isolate the drop is to hum. Say the word with a particle, then hum the same thing with your mouth closed. Without consonants and vowels to distract you, the melody is all that is left, and the step down is obvious. Try it with the pairs below, which contrast a heiban noun with an accented one in the same frame.

The minimal pairs everyone learns

A few pairs of words are spelled identically in kana and separated by pitch alone, and they are the fastest way to convince your ear that the drop carries meaning. The patterns below are the standard Tokyo ones you will find in accent dictionaries. Say each row with its particle, because for hashi the bridge and chopsticks difference only shows fully once ga or o follows.

Kami meaning hair takes the same odaka pattern as kami meaning paper in Tokyo speech, so that pair is one that pitch does not separate and context has to. That is a useful reminder that pitch accent removes some ambiguity, not all of it, and that Japanese speakers rely on the sentence around a word as much as on its melody.

Two-mora nouns: the largest set

A two-mora noun can only do three things: stay flat, drop after the first mora, or drop after the second, which you hear on the particle. Heiban is roughly the biggest group and atamadaka the next, with odaka the smallest but still full of everyday words. Because there are so few options, it pays to learn these short nouns in named sets rather than one at a time.

• Heiban, high through the particle: 水 mizu, 風 kaze, 口 kuchi, 飴 ame (candy), 端 hashi (edge), 柿 kaki (persimmon).
• Atamadaka, high then low: 雨 ame (rain), 箸 hashi (chopsticks), 猫 neko, 春 haru, 秋 aki, 海 umi, 本 hon, 今 ima.
• Odaka, high until the particle drops: 橋 hashi (bridge), 花 hana, 山 yama, 川 kawa, 犬 inu, 夏 natsu, 冬 fuyu, 紙 kami (paper), 石 ishi, 町 machi.

Verbs in -masu: one rule for every verb

Dictionary forms are unpredictable. Iku is heiban, nomu drops after the first mora, and taberu drops after be, so every new verb needs checking. The polite forms are the opposite: they follow one rule regardless of the verb. In the -masu form the pitch is high up to and including ma, then drops on su. Every verb, godan or ichidan, regular or irregular, does exactly this, so tabeMAsu, ikiMAsu, and benkyō shiMAsu all share the same ending melody.

The other polite endings are just as regular. The past -mashita drops after the same ma, so it sounds like tabeMAshita. The negative -masen drops after se, so it is tabemaSEn, and the -mashō suggestion drops after sho, giving tabemaSHOo. Since beginners spend most of their speaking time in these forms, mastering four endings gives your verbs correct pitch almost everywhere.

i-adjectives: the drop before the final i

Most i-adjectives are accented, and the drop sits on the mora just before the final i. Takai is taKAi, samui is saMUi, atsui is aTSUi, and longer adjectives simply move the drop along: ōkii is oOKIi and omoshiroi is omoshiROi. In front of desu the adjective keeps its own drop, so takai desu still falls after ka and desu stays low.

A smaller group is heiban, with no drop at all: akai, amai, and tsumetai are common examples, and they stay high straight through. Because the accented group is larger, treat the drop before i as your default guess for a new adjective and note the flat ones as you meet them.

Loanwords, clock times, and minutes

Katakana loanwords lean heavily towards a drop early in the word. Short ones are very often atamadaka: TErebi, KAmera, BAnana, BAsu, PAn, HOteru, TAkushii, and REsutoran all fall after the first mora. A well-worn group has settled into heiban instead, including paSOKON, koNBINI, and aMERIKA, so treat atamadaka as the guess and heiban as the exception to note.

Clock times are one of the tidiest sets in the language. With -ji the drop lands immediately before ji, so ichiji is iCHIji, rokuji is roKUji, and nanji is NAnji. When the mora before the counter is a special one, such as the n in sanji or the second half of the long vowel in jūji, the drop moves one mora earlier, giving SAnji and JUuji. Minutes with -fun and -pun behave the same way, which is why so many of them come out atamadaka: GOfun, SAnpun, JUppun.

Why the drop matters more than the height

Listeners are not measuring how high your voice goes. They are tracking where it falls, because that location is what distinguishes one word from another. A low-voiced speaker and a high-voiced speaker produce completely different frequencies for the same word, and both are perfectly natural, so absolute height carries no information at all. The size of the fall varies too: a calm speaker may drop only a little, while an excited one drops a lot.

This is good news for nervous learners, who often do the opposite of what helps. They push the high morae up dramatically and then place the fall on the wrong mora, producing a word that is both foreign and hard to place. A small, relaxed fall in the right spot sounds native; a theatrical one in the wrong spot does not. Aim for location, keep the range modest, and let the melody be small.

One more thing to expect: after a drop inside a phrase, the words that follow tend to sit a little lower than they would on their own, so a sentence does not keep returning to the same ceiling. Do not fight this. Match the model speaker's overall shape and concentrate on placing each drop where the word wants it.

How to mark accent in your own notes

Marking pitch only works if you use one system everywhere. The simplest is the number that accent dictionaries use: the number is the mora after which the pitch drops, and zero means no drop. Under that system ame meaning rain is 1, hashi meaning bridge is 2, kokoro is 2, tabemasu is 3, and sakura is 0. The number tells you everything, because the rise is always between the first and second mora and the rest follows.

• Write the number in square brackets after the word, such as 心 kokoro [2], and keep the same brackets on flashcards so the pattern is reviewed with the meaning.
• If you prefer a visual mark, put a small downward arrow or a backslash after the mora that drops: ko-ko↘ro, ha-shi↘ ga.
• Always write nouns with a particle when they are odaka or heiban, since that is the only place the two patterns differ: 花 hana [2] ga, 桜 sakura [0] ga.
• Record verbs in the -masu form and note only the dictionary-form pattern separately, because the polite forms follow the rule above and do not need marking.
• For a new i-adjective, write the drop before i as the default and only mark the flat ones explicitly, such as 赤い akai [0].
• Keep a short list of the loanwords that broke the atamadaka guess, and review it once a week.

Check yourself: record, compare, and get coached

Your ear is the last thing to catch your own mistakes, so build a routine that does not depend on it. Record ten nouns with ga attached and ten sentences in the -masu form, then play the recording back while humming along. Wherever your hum and the recording disagree about the step down, mark the word and re-record it against a native model, listening for location rather than height.

After that, take the same lines into a live conversation with your AI Sensei in Unihongo's immersive 3D classroom, where the pronunciation coaching flags the words whose drop landed early or late, and the session report lists the corrections so you can add them to your marked notes. Short daily rounds of record, compare, and correct move patterns from something you know into something you say.

Read each check line once as a hum, once slowly with the drops exaggerated for placement, and once at normal speed with a modest range. If the third version keeps every drop where the second one put it, the pattern has moved into your speech.