Our methodology

Most JLPT sites hand you a flat list and a wall of practice questions. Yatta is built differently: every grammar point, word, and kanji is a node in a connected knowledge graph, and we use that structure to do the one thing flashcards can't — tell you exactly what to fix and in what order. Here's how it works, how the content is made, and what is genuinely our own.

The core: ontology-based diagnosis

The heart of Yatta is a hand-built ontology — a map of how Japanese grammar actually connects. For each pattern we record three kinds of relationship:

• Prerequisites — what you need to understand first. You can't really grasp the passive without the godan ending shifts; you can't use 〜たら well without the plain past.
• Confusable pairs — patterns learners mix up (は vs が, 〜ば vs 〜たら vs 〜なら, 〜ために vs 〜ように), so a mistake on one can point you to the other.
• Extends / advanced form — how a pattern grows into harder ones across levels, so progress follows the grammar's own logic instead of an arbitrary list.

Take a free mock test and we run your answers through that graph. Instead of just a score, you get a one-time weakness map: which themes are shaky, which specific points to review, what to build first (prerequisites), and which look-alikes you keep confusing — each one linked straight to its guide and drill.

How it all fits together

That same connected database of Japanese — grammar, vocabulary, kanji, and example sentences — powers everything else on the site. From that base we do three things:

1. Organize — we connect grammar and vocabulary into the network of relationships above and group them by theme and level.
2. Explain — we write friendly, accurate guides on top of that structure: meaning, formation, nuance, common mistakes, natural examples, and hand-written insights that go past the textbook.
3. Practise — we generate the question bank from the same data, and even derive the wrong answer choices from our confusion relationships, so a careless mistake is a meaningful one you can learn from.

Our own work (and copyright)

The structure and the writing are ours. Specifically, the following are created and organized by Yatta and are our original work, © Yatta:

• The ontology — the relationships between grammar points and words: prerequisites, confusable/similar pairs, “extends/advanced form of”, and thematic and semantic groupings. We built these connections ourselves; they are not taken from any third party.
• The diagnosis logic that turns your test answers into a personalized weakness map.
• The selection, ordering, and curation of example sentences for each point.
• All explanations, comparisons, study guides, editorial insights, and other written text.
• Every question in our question bank, including how wrong-answer choices are derived from our own confusion relationships.

Please don't republish or resell this work without permission.

Where the underlying data comes from

The raw linguistic data we build on comes from open and public-domain resources. We are grateful to these projects:

• JMdict / EDICT — Japanese–English dictionary data (vocabulary, meanings), from the Electronic Dictionary Research and Development Group, used under licence.
• KANJIDIC2 and open kanji datasets — kanji readings, meanings, stroke counts, and JLPT-level mappings.
• The publicly documented JLPT content scope (the long-standing community and former official level lists) — used only to decide which items belong to which level.

We do not reproduce copyrighted textbooks, and we do not use or copy past JLPT exam papers — those are copyrighted, so every practice question here is generated from our own data instead. (JLPT® is a registered trademark of the Japan Foundation and JEES; Yatta is independent and not affiliated with them.)

How content is created and checked

Guides are drafted from the structured data and then edited for clarity, accuracy, and natural English before publishing. Every guide carries a named author and a visible “last updated” date, we add hand-written insights to flagship topics, review samples across each category, and revise pages as we spot improvements or receive corrections.

Found a mistake?

We genuinely appreciate corrections — accuracy is the whole point. Please reach us via the contact page and we'll fix it.