Pitch Accent in Verbs and Adjectives

Nouns carry a fixed accent you memorize once. Verbs and adjectives are different: they conjugate, and their accent moves as they do. The good news — and it is very good news — is that the movement is almost entirely rule-governed. Every Japanese verb belongs to one of exactly two accent classes, and so does every i-adjective. Learn what each class does across the conjugation table, and you can predict the pitch of any form of any verb from a single fact: is this word accented, or is it heiban?

That is a genuine shortcut, and most textbooks never mention it. They print 食べる with a pitch mark and leave you to guess what happens in 食べます, 食べない, 食べた. This page gives you the two patterns instead of the endless list. If you have not met the four accent shapes yet, start with Pitch Accent Patterns.

We mark morae H/L and put ꜜ where the pitch drops.

Verbs: two classes

The dictionary form tells you the class

• Unaccented (平板 heiban) verbs have no downstep in the plain form — the pitch rises and stays up.
• Accented (起伏 kifuku) verbs have a downstep on the second-to-last mora (the penultimate) of the plain form.

In 食べる and 泳ぐ the fall lands on the mora right before the last one (べ, よ) — that "penultimate" placement is the accented class's signature. 遊ぶ simply never falls.

What conjugation does to each class

Here is the whole story in one table. Follow the two verbs across the forms.

Two patterns fall out of this:

The heiban verb stays flat everywhere — except in -ます. 遊ばない and 遊んだ have no downstep at all. But 遊びます suddenly drops on the ま. That is not an exception to memorize; it is a property of the ending itself.

The polite -ます ending is always accented on the ま, for every verb in the language. Look at both columns: あそびまꜜす and たべまꜜす both fall after ま. This is why the polite forms of all verbs share the same closing rhythm — 行きまꜜす, 飲みまꜜす, わかりまꜜす, all identical at the tail. Once you hear it, you hear it everywhere.

The accented verb keeps its downstep anchored to the stem. In 食べる the fall sits on べ, and it stays on べ through the negative (たべꜜない) and the past (たべꜜた) and the te-form (たべꜜて). Only -ます pulls it away to ま. So for an accented verb you really only track one thing: where the stem's downstep lives, plus the universal -ます rule.

Adjectives: the same two-class logic

I-adjectives split into the same two classes. They are distinct in the plain form and — contrary to what many books claim — they stay distinct through conjugation too; what changes is where an accented adjective's downstep sits.

The plain forms behave like the verbs: 赤い is heiban (flat, stays up), 高い is accented on the penult か. But the -く row does not neutralize the two classes — this is the point almost every book gets wrong. An accented adjective's peak leaps to the first mora: 高い (drop after か) becomes 頭高 in 高く — たꜜかく, H·L·L. A heiban adjective, by contrast, stays flat: 赤い stays flat in 赤く — あかく, L·H·H, with no fall at all. So in the -く form the two classes remain audibly distinct: front-drop (accented) versus flat (heiban). The real learner-catch here is that leap — an accented adjective's downstep jumping forward onto the first mora — not a flat word magically acquiring a fall.

It is tempting to reach for the verbs' -ます as an analogy here, but the analogy runs backwards. The polite -ます genuinely neutralizes verbs: it drops on ま for every verb, so heiban and accented verbs converge on one shared shape. The adjective -く does the opposite — it preserves the class split, because it only moves the accent of words that already have one. A heiban adjective has no accent for -く to move, so 赤く stays flat (あかく, L·H·H). Don't import the -ます intuition; -く is not a neutralizer.

The past -かった keeps the two classes distinct as well, but here the picture is genuinely murkier and deserves far less of your attention. As a broad tendency an accented 高かった drops earlier (たかꜜかった) than a heiban 赤かった (あかかꜜった) — yet the exact placement, and even whether a heiban adjective takes a clear drop at all, wobbles from speaker to speaker. This is the corner of the system where you will hear the most variation, so it is not worth stressing over. The robust, high-value point is the one from the -く form: -く does not neutralize the classes — an accented adjective jumps to 頭高 (高く → たꜜかく, H·L·L), while a heiban adjective stays flat (赤く → あかく, L·H·H).

How this differs from English

English stress is fixed to the word and rides along through inflection: PHÓtograph keeps its stress idea even as it shifts in photÓgraphy, but within a single word like eat → eats → eating → ate, the stressed syllable never moves — it is always the one lexical stress. Learners import that instinct and give a verb one "stressed" mora that they then hold constant across every conjugation. Japanese does the opposite: the downstep genuinely relocates depending on the ending. 食べる drops on べ, but 食べます drops on ま — the peak physically moved. Accepting that the accent is a property of the whole conjugated word, not a fixed syllable you carry around, is the mental shift this page is really about.

Common Mistakes

Key Takeaways

• Every verb and every i-adjective is either accented or unaccented (heiban) — just two classes.
• Accented verbs drop on the penultimate mora of the plain form (食べꜜる → べ); heiban verbs never drop in the plain form.
• The polite -ます always drops on ま, for every verb — the universal rule that gives all polite forms the same tail.
• Heiban verbs stay flat through -ない and -た; accented verbs keep the downstep on the stem there.
• Adjective -く does not neutralize the classes: an accented adjective jumps forward to 頭高 (高く → たꜜかく, H·L·L), while a heiban adjective stays flat (赤く → あかく, L·H·H).
• Don't drag one fixed "stress" across the conjugation the way English does — the downstep really moves.