The Four Pitch Patterns | Japanese Grammar

Once you accept that Japanese accent is melody, not loudness — the idea developed in Pitch Accent: What It Is — the whole system turns out to be remarkably tidy. Every word in standard Tokyo Japanese belongs to exactly one of four accent patterns, defined entirely by whether there is a downstep and where it falls. Learn these four shapes and you have a mental filing cabinet for every word's pitch, plus a safe default for the ones you have not looked up yet.

The four patterns have traditional names built from 高 ("high") and the position of the peak: 平板 heiban, 頭高 atamadaka, 中高 nakadaka, and 尾高 odaka. This page walks through each one with the particle が attached — because, as you will see, が is the only place some of these patterns can even be told apart.

The two rules that generate all four

Everything below follows from the two Tokyo rules introduced in the overview:

The first two morae differ in pitch — a word opens either low→high or high→low.
There is at most one downstep (ꜜ), and after it the pitch stays low.

Given a word of n morae, the downstep can sit after mora 1, after mora 2, … after mora n, or nowhere at all. That is exactly n + 1 possibilities — and they collapse into the four named families below. We mark the downstep with ꜜ and describe the melody as a string of H(igh) and L(ow), one letter per mora, including the particle.

1. 平板 (heiban) — accentless, stays high

Heiban means "flat." There is no downstep at all. The word starts low on mora 1, rises to high on mora 2, and simply stays high — including onto any particle that follows. This is the most common pattern in the language and the safe default when you do not know a word's accent.

Word	Reading	Word alone	With が
桜 (cherry blossom)	sakura	sa(L) ku(H) ra(H)	sakura ga → L H H H

桜がきれいに咲いたね。

sakura ga kirei ni saita ne

The cherry blossoms have bloomed beautifully, haven't they? (桜 heiban: rises and stays high, が high.)

犬が嬉しそうにしっぽを振っている。

inu ga ureshisō ni shippo o futte iru

The dog is wagging its tail happily. (犬 inu is heiban: i-L nu-H, が high.)

💡

When you meet a new everyday noun and cannot check its accent, guess heiban — low first mora, then high all the way through and onto the particle. It is the statistically likeliest pattern and the least conspicuous if you are wrong. Just remember the first mora is low; do not start high.

2. 頭高 (atamadaka) — high head, immediate drop

Atamadaka means "head-high": the first mora is high, and the pitch drops immediately after it. The downstep is after mora 1, so everything from mora 2 onward — including the particle — is low.

Word	Reading	Word alone	With が
猫 (cat)	neꜜko	ne(H) ko(L)	neꜜko ga → H L L

猫が窓の外をじっと見ている。

neꜜko ga mado no soto o jitto mite iru

The cat is staring intently out the window. (猫 atamadaka: high, then everything drops.)

Atamadaka is the pattern English speakers over-apply, because a high-then-falling shape resembles English initial stress. It is a real Japanese pattern — but only for the words that actually have it (猫, 雨 aꜜme "rain," 恋 koꜜi "love"), not for everything.

3. 中高 (nakadaka) — a peak in the middle

Nakadaka means "middle-high": the word starts low, rises, and then drops somewhere in the middle — after some interior mora, but before the last one. It needs at least three morae to exist (you need a mora before the peak and a mora after the drop).

Word	Reading	Word alone	With が
お菓子 (sweets)	okaꜜshi	o(L) ka(H) shi(L)	okaꜜshi ga → L H L L
卵 (egg)	tamaꜜgo	ta(L) ma(H) go(L)	tamaꜜgo ga → L H L L

お菓子がまだたくさん残ってるよ。

okaꜜshi ga mada takusan nokotteru yo

There are still lots of sweets left. (お菓子 nakadaka: rises on か, then drops.)

卵が一つ割れちゃった。

tamaꜜgo ga hitotsu warechatta

One of the eggs broke. (卵 nakadaka: peak on ま, drop after.)

お母さんはもう出かけたよ。

okāsan wa mō dekaketa yo

Mom already went out. (お母さん okāsan is nakadaka: peak on か, then low.)

4. 尾高 (odaka) — high to the very end, drop on the particle

Odaka means "tail-high": the word rises and stays high all the way to its last mora, and the drop happens on the following particle. Said in isolation, an odaka word looks just like a heiban word — low, then high to the end. The difference is invisible until a particle appears.

Word	Reading	Word alone	With が
男 (man)	otokoꜜ	o(L) to(H) ko(H)	otokoꜜ ga → L H H L

さっきの男がもう一度来たよ。

sakki no otokoꜜ ga mō ichido kita yo

That man from before came again. (男 odaka: high to the end, then が drops.)

The key insight: odaka vs heiban lives on the particle

Here is the fact that makes particles the diagnostic for the whole system. Odaka and heiban words are identical on the word itself — both are low-then-high, high to the end. They diverge only on the following particle: heiban keeps it high, odaka drops it low.

The cleanest proof is a real minimal pair — 花 and 鼻, both hana, both L-H alone:

Word	Type	Word alone	With が	Meaning
花	odaka	ha(L) na(H)	hana ga → L H L	flower
鼻	heiban	ha(L) na(H)	hana ga → L H H	nose

花がとてもきれいに咲いている。

hana ga totemo kirei ni saite iru

The flowers are blooming beautifully. (花 odaka → が drops low.)

花粉症で鼻がむずむずする。

kafunshō de hana ga muzumuzu suru

My hay fever is making my nose itch. (鼻 heiban → が stays high.)

💡

Because 花 and 鼻 sound the same alone, the only audible difference is what happens to が. This is exactly why you should learn every noun with が attached — it is the one context that pins down whether a word is odaka or heiban, a distinction the word can otherwise hide completely. The full set of these pairs is on Pitch Accent Minimal Pairs.

Summary table

Pattern	Downstep	Example (＋が)	Melody with が	Particle が
平板 heiban	none	桜が sakura ga	L H H H	high
頭高 atamadaka	after mora 1	猫が neꜜko ga	H L L	low
中高 nakadaka	mid-word	お菓子が okaꜜshi ga	L H L L	low
尾高 odaka	after last mora	男が otokoꜜ ga	L H H L	low

Notice the pattern in the last column: only heiban keeps が high. Every accented pattern (atamadaka, nakadaka, odaka) drops the pitch by the time が arrives — they just drop at different points. This page covers nouns; verbs and adjectives conjugate and follow their own accent rules, which are treated in Pitch Accent in Verbs and Adjectives.

Common mistakes

Mistake 1: Applying atamadaka to everything. English initial stress makes learners drop the pitch after the first mora on every word, turning heiban 桜 into an atamadaka "SÁ-ku-ra."

❌ さꜜくらがさいた（atamadaka: SA-drop）

saꜜkura ga saita

Incorrect — 桜 forced into a high-first, dropping shape.

✅ 桜がやっと咲いた。

sakura ga yatto saita

The cherry blossoms have finally bloomed. (heiban: low first, then high through が.)

Mistake 2: Not dropping が on odaka words. Treating 男 or 花 like heiban and keeping が high, which merges them with 鼻-type words.

❌ はなが（odaka word, but が kept high）

hana ga (が high)

Incorrect — 花 'flower' pronounced like 鼻 'nose'; the が should drop.

✅ この花がいちばん好き。

kono hana ga ichiban suki

I like this flower best. (花 odaka: が drops.)

Mistake 3: Starting heiban words high. Forgetting that even the accentless pattern opens low on mora 1. Saying "HH-H" from the first beat sounds unnatural; it must be L-then-H.

Mistake 4: Using loudness or length instead of pitch to mark the peak. Punching お菓子 with a loud, long か instead of a higher か. Keep every mora equal in force and length; move only the melody.

Mistake 5: Assuming a word's citation pitch never changes. The drop you memorize is stated relative to a following particle. In longer phrases and compounds, accents shift and merge — the four patterns describe the word in isolation with one particle, which is the right starting point, not the whole story.

Key takeaways

Four patterns: heiban (no drop, が high), atamadaka (drop after mora 1), nakadaka (drop mid-word), odaka (drop on the particle).
An n-mora word has n + 1 possible accents; they sort into these four families.
Heiban is the most common pattern and the safe default — but it still starts low.
Odaka vs heiban is audible only on the particle (花 vs 鼻) — so learn nouns with が.

Now practice Japanese

Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.

Start learning Japanese→