Japanese verb conjugation looks enormous and is actually small. There are three verb groups, and one of them contains exactly two verbs. Once you can sort a verb into its group and produce its stem, every form you will ever need — polite, past, negative, te-form, potential, passive, causative — comes off the same handful of rules. Here is the whole system in one place.
The only hard part
Telling Group 1 from Group 2. Everything downstream is mechanical. Learners who struggle with conjugation almost always have a group-identification problem, not a conjugation problem — so that is where to spend your effort.
The three groups
Group 1 — 五段 / u-verbs (most verbs)
Dictionary form ends in any う-row sound: う つ る く ぐ ぬ ぶ む す.
買う · 待つ · 取る · 書く · 泳ぐ · 死ぬ · 遊ぶ · 読む · 話す
Group 2 — 一段 / ru-verbs (the easy ones)
Ends in る, and the syllable before る is an i- or e-sound. Drop る, add everything.
食べる · 見る · 起きる · 教える · 寝る
Group 3 — irregular (two verbs, total)
する (to do) · 来る (to come). Memorise them and you are done forever.
The る-verb trap
A verb ending in る is not automatically Group 2. Check the sound before る:
- 食べる — べ is an e-sound → Group 2
- 取る — と is an o-sound → Group 1
- 帰る — え-sound but Group 1 anyway. A genuine exception, and 入る, 走る, 知る, 切る and 要る do the same with an い-sound. The common ones are a short list: learn them as a set.
The masu-stem: your universal starting point
Almost every polite and compound form builds from the ます-stem, so this is the single most valuable transformation to make automatic.
- Group 1: change the final う-sound to its い-sound. 読む → 読み, 書く → 書き, 話す → 話し.
- Group 2: drop る. 食べる → 食べ, 見る → 見.
- Group 3: する → し, 来る → 来.
From that stem: add ます (polite), ました (polite past), ません (polite negative), たい (want to), ながら (while), やすい / にくい (easy / hard to), and more.
LenGaki has 445 grammar points across N5 to N2, each with its structure, worked examples and a breakdown of every particle in the sentence.
Browse the grammar libraryThe te-form, and why it is the gateway
The て-form is the most-used conjugation in Japanese. It joins clauses, forms the progressive, makes requests, and grants permission. Group 2 and 3 are trivial; Group 1 requires the sound changes below, and they are worth drilling until they are reflex.
| Ending | Becomes | Example | た-form |
|---|---|---|---|
| う · つ · る | って | 買う → 買って | 買った |
| む · ぶ · ぬ | んで | 読む → 読んで | 読んだ |
| く | いて | 書く → 書いて | 書いた |
| ぐ | いで | 泳ぐ → 泳いで | 泳いだ |
| す | して | 話す → 話して | 話した |
| 行く (irregular) | 行って | 行く → 行って | 行った |
The free lesson
The た-form follows exactly the same rules as the て-form. Swap て→た and で→だ. So learning the te-form gives you the plain past tense at no additional cost — and the plain past is what N4’s 〜たら, 〜たことがある and 〜たほうがいい all attach to.
The full form map
Using 読む (Group 1), 食べる (Group 2) and する (Group 3):
| Form | 読む | 食べる | する | Level |
|---|---|---|---|---|
| Polite ます | 読みます | 食べます | します | N5 |
| Polite past | 読みました | 食べました | しました | N5 |
| Polite negative | 読みません | 食べません | しません | N5 |
| te-form | 読んで | 食べて | して | N5 |
| Plain past た | 読んだ | 食べた | した | N5 |
| Plain negative ない | 読まない | 食べない | しない | N5 |
| Want to たい | 読みたい | 食べたい | したい | N5 |
| Potential | 読める | 食べられる | できる | N4 |
| Volitional | 読もう | 食べよう | しよう | N4 |
| Passive | 読まれる | 食べられる | される | N4 |
| Causative | 読ませる | 食べさせる | させる | N4 |
| Conditional ば | 読めば | 食べれば | すれば | N4 |
| Imperative | 読め | 食べろ | しろ | N4 |
The potential/passive collision
Notice that Group 2 potential and passive are identical: 食べられる means both “can eat” and “is eaten.” This is not a mistake in your textbook. Japanese resolves it by context and particles.
- Potential: 私はさしみが食べられる。— the object usually takes が.
- Passive: さしみは猫に食べられた。— an agent marked with に is present.
In casual speech, Japanese speakers increasingly shorten the potential to 食べれる to avoid the ambiguity. It is widespread, natural, and still marked wrong on the JLPT — so recognise it, do not write it.
How to practise so it becomes automatic
Conjugation is a motor skill, not knowledge. You are not trying to be able to work it out; you are trying to not have to.
- Group identification drills first. Take 50 verbs and only sort them into groups. No conjugating. Speed matters more than care.
- One transformation at a time, across many verbs. Twenty verbs to て-form beats one verb through twenty forms — you are training the rule, not the word.
- Say them aloud. The sound changes (んで, いて, って) live in your mouth. Silent reading trains recognition and leaves production untouched.
- Then use them in sentences. One real sentence per form. This is what moves it from drill into language.
LenGaki tags every verb with its part of speech, so you can filter the vocabulary list down to verbs only and drill group identification against real N5–N1 words.
Get the verbsLenGakiで学ぶ:Japanese grammar
この内容を、すべて学べる
- JLPTの単語7,788語(N5〜N1)、読みと音声付き(N5は無料)
- 音読み・訓読み・書き順の練習付きの漢字2,057字
- 例文と間違えやすい表現の比較付きの文法445項目
- 本番形式のN5・N4模試。時間制限付きで、聴解は実際の音声で出題(N5の第1回・第2回は無料)
The summary
Three groups, one of which is just two verbs. One stem that generates most polite forms. One set of sound changes that generates both て-form and past tense. Everything at N4 and above hangs off plain form, which you get free from the て-form rules. It is a small system that rewards drilling far more than studying — get the group identification instant, and the rest follows.




