NihongoLevel is a Japanese content difficulty database. We analyze anime, manga, and videos to determine their JLPT difficulty level, so Japanese learners can find content that matches their current ability.
We process Japanese subtitles through natural language processing (SudachiPy) and compare the vocabulary against 46,000+ JLPT-classified words. Each title gets a difficulty score based on the 90% coverage threshold — the JLPT level at which a learner would understand 90% of the words, enough to follow the story with occasional lookups.
Research by Laufer & Ravenhorst-Kalovski (2010) shows that 90% vocabulary coverage represents "adequate comprehension with support" — exactly the N+1 sweet spot where learning happens. You understand enough to follow along, but encounter enough new words to grow.
NihongoLevel is built by Lionel Deveaux, a developer and Japanese learner based in France. The project grew out of tools built for personal study and is part of a suite including YomiCheck (text difficulty analyzer) and MiruCheck (video difficulty scorer).
Questions, suggestions, or data corrections? Reach out via GitHub.