What to read in Japanese

Japanese books and manga, easiest first. A level is counted from the text where the text can be had legitimately, and estimated otherwise; each page says which.

How the numbers are counted

• Words. The text is split into words with Sudachi, and each word is counted in its dictionary form: 待った and 待って both count as 待つ, and older spellings such as 云う count with 言う.
• Common words. "The 5,000 most common words" are ranked by wordfreq's Japanese list (subtitles, Wikipedia, news and the web), folded into dictionary forms the same way.
• Coverage. The share of a text's words that are among the 2,000, 5,000 or 10,000 most common. Particles and auxiliaries count as known; names, numbers and punctuation are left out.
• Level. From the 5,000-word coverage: 1, Beginner, 95% or more; 2, Elementary, 92–95%; 3, Intermediate, 89–92%; 4, Upper intermediate, 86–89%; 5, Advanced, below 86%. The bands are provisional and will be checked as more titles are counted.
• Key words. Words outside the 2,000 most common, most frequent first, checked by hand. Meanings come from JMdict (EDRDG), the dictionary Fusen uses, under CC BY-SA 4.0.
• Estimated levels. Most books still in copyright can't be counted: their text can't be obtained for this site in a way their stores and publishers allow. Their level, marked "(estimated)" and drawn with outlined bars, is estimated with AI from the publisher's free sample and other public information, and their word lists are chosen by hand, without counts.
• Furigana and the other facts at the top of each page are checked against the edition named there.

The text of a book still in copyright is never stored or quoted here.