How many Japanese words (and kanji) do you need?

Why is it hard to count Japanese words?

Ordinary Japanese is written without spaces, so where one word ends is partly a decision. Is 持っています (is holding) one word or two? Dictionaries, textbooks and word-counting software answer differently, and so do the series in the Contos library, whose spacing isn't yet uniform: some attach particles to the word before them, most don't.

Verbs multiply the problem. In the free A1 and A2 chapters, 言う (to say) appears 112 times, as 言います, 言った, 言いました, 言いません and more. Counted by spelling it is several words; counted by dictionary form it is one. The frequency lists page explains these units for any language.

Kanji are easier to count: a character is a character. So this page leans on kanji counts from the library, and treats word counts as approximate.

How many words and kanji does each level take?

Neither of the usual scales gives a number. The CEFR describes levels by what you can do, not by vocabulary size. The JLPT used to publish word and kanji lists, but since its 2010 revision it says a list of vocabulary, kanji and grammar items "was not necessarily appropriate", because it tests the ability to communicate, not memorised items. The lists you find online for N5 to N1 are unofficial estimates.

What the JLPT does describe is the reading: N5 expects sentences in hiragana, katakana and basic kanji; N4 expects passages on familiar daily topics in basic vocabulary and kanji. Contos sets its own thresholds in known words: A2 at 500, B1 at 1,500, B2 at 4,000. For kanji, here is what the free chapters at each level actually contain.

Two things stand out. Each level adds only about 200 to 250 new kanji, and the kanji you met lower down already cover the large majority of what is on the page. Across all four levels the free chapters use 1,068 different kanji, about half the 2,136 on the jōyō list. The jōyō list, set by cabinet notice in November 2010, is the government's guide to kanji use for writing modern Japanese in everyday life; it replaced a 1981 list of 1,945, adding 196 characters and removing five.

Which kanji and words should you learn first?

The most frequent ones, and frequency is steep. In the free A1 chapters, the 50 most common kanji account for 54% of all kanji on the page, the top 100 for 72% and the top 200 for 89%. At the other end, 128 of the 368 A1 kanji appear only once. Learn the common ones deliberately and let the rest come and go until they repeat.

Learn kanji through words. Two common kanji often make a third word you already need:

All three are in our list of your first 100 Japanese words: 手紙 appears in 12 of the 48 free A1 and A2 chapters. The pattern also shows the catch. Knowing a kanji's meaning helps you guess a compound; its sound in the compound still has to be learned.

How do you get to 500, 1,500 and 4,000 words?

1. 0 to 500: kana and the core. Read A1 stories in the polite form. Save the particles, the verbs ある, いる, する and 言う, and the nouns that keep appearing.
2. 500 to 1,500: plain forms and everyday kanji. At A2 the plain form takes over the narration, and you will meet about 200 kanji you didn't see at A1. Save whole words, not single characters.
3. 1,500 to 4,000: the long tail. B1 and B2 chapters are longer and narrated in the plain past. Finishing a series gives you repeats; spaced repetition catches what reading misses.
4. Count what you recognise, not what you can write. Reading needs you to know a word when you see it. Writing kanji by hand is a separate skill.

For the hours each stage tends to take, see how long it takes to learn Japanese.