Nihongomachi

What’s on the site

The idea

I'm studying for the N5 myself, aiming for a 2027 sitting, and I wanted the resource I couldn't find. There's no official word list since 2010, so every site draws its own line around N5, and knowing a word isn't the same as passing the exam. I built a site that names its sources, follows the real exam structure and is free to study.

How it’s built

Vocabulary and kanji follow the Tanos N5 lists, and grammar covers the 84 points on JLPT Sensei's list plus five more, grouped by what they do. AI drafts the first pass of the data, then every kanji reading is checked against KANJIDIC and every item page carries a "Report it" link. Each word, kanji and grammar point has its own page, about 1,000 in total, linked into clusters and built to be found in search.

Stack Next.js · structured study data · spaced repetition · KANJIDIC · IndexNow

Skills it shows

• Technical SEO: site architecture, internal linking, metadata and indexing for about 1,000 pages, with one focused page per question a learner might search.
• AI workflow: AI drafts the data at speed, and a trusted reference decides what is safe to publish.
• Content architecture: organising 814 words, 103 kanji and 89 grammar points into topics, groups and one learning route.
• Product ownership: setting the goal, the audience and what not to build, including retiring XP and streaks to stay focused on the exam.

What I learned

• A fast first version needs a slow second pass. Research into the exam is what made the site useful.
• AI drafts need a source of truth. Checking against a trusted reference makes the output safe to publish.
• When no official list exists, naming the lists you follow builds more trust than claiming to be definitive.
• Scope is a feature. Doing N5 properly before touching N4 keeps quality high and the work finishable.

Behind the scenes

Why I built it

The goal is simple: pass the JLPT N5 myself, and build the resource I wanted while studying for it. I'm aiming for a 2027 sitting, so every page has to work for me as a learner first.

The first problem is the syllabus. There has been no official JLPT word list since 2010, so each study site draws its own line around N5. A beginner can't easily tell what's actually in scope.

The second problem is the gap between knowing and passing. Knowing a word is not the same as answering an exam question about it. The exam also sets section minimums, so a strong vocabulary score can't rescue a missed listening minimum.

My primary audience is English speaking beginners preparing for the N5. My secondary audience is anyone who lands from a search needing one word, kanji or grammar point explained clearly.

What it deliberately doesn't do:

• It doesn't charge. Everything on the site is free to study.
• It isn't official. It's independent and not affiliated with the Japan Foundation or JEES, who run the JLPT.
• It doesn't teach kanji handwriting. At N5 you need to read kanji, not write them, so that's where the effort goes.
• It doesn't chase streaks or points. I built XP and ranks early on, then retired them to keep the focus on the exam.
• It doesn't go past N5 yet. One level done properly beats five done thinly.

The methodology

The site runs on a short set of rules. The core of them is on the About page.

1. Follow the lists learners already use. With no official list since 2010, vocabulary and kanji follow the Tanos N5 lists, the ones most study sites use.
2. Use a checklist, then go further. The grammar section covers all 84 points on JLPT Sensei's N5 list, for 89 points in total.
3. Check readings against a dictionary. Every kanji reading is checked against KANJIDIC, an open kanji dictionary.
4. Write everything ourselves. All explanations, example sentences and practice questions are original.
5. Build around the real exam. Section timings, pass marks and minimums come from the official JLPT site, and the study route mirrors the exam's sections.
6. Make corrections easy. Every word, kanji and grammar page carries a "Report it" link, and the contact page is always one click away.
7. Meet beginners where they are. Every word shows kana, romaji and English, with audio, so learners who can't read kanji yet can still follow.
8. Stay free and independent. Free to study, no affiliation with the exam bodies, and every source credited under its open licence.

Where the content comes from

Nihongomachi leans on four public references for scope and accuracy. Everything built on top of them, from explanations to practice questions, is written from scratch.

Not covered yet: sources for reading passages, listening audio and mock test content, which arrive with those features.

The hard problems

A free study site looks simple from the outside. These were the parts that weren't.

Choosing a syllabus when none exists. With no official list since 2010, I adopted the Tanos lists for words and kanji. For grammar I used JLPT Sensei's 84 points as a checklist, but grouped my 89 by what they do rather than alphabetically.

Trusting AI drafts. I used AI to draft the first pass of word, kanji and grammar data so I could move quickly. Fast isn't the same as right, so every kanji reading is checked against KANJIDIC and every item page has a "Report it" link.

Rebuilding a version one that came together too fast. The first version was built quickly, without deep research into the exam. I restructured it around the exam itself: three sections, real timings and pass marks, and one route from kana to mock test.

Same sound, different word. Many N5 words share a reading, like kiru (to wear) and kiru (to cut), or atsui for hot weather and hot to the touch. Each needs its own readable page address, so I separated them by meaning. An edge case or two still carries a number, and that's on my list.

Writing for English speakers from zero. Some of the clearest N5 explanations I found were written for Chinese speaking learners, who already know most kanji. I used them for research only, then wrote original English guides for true beginners, with regional details stripped out.

Letting go of gamification. I built XP, streaks and ranks early on, and they were fun to make. I retired them for now so the study loop stays focused on passing the exam.

Technical SEO and organic traffic

Search is how most learners will find Nihongomachi, so I treat it as part of the product, not a finishing touch.

Match pages to real searches. Each list page targets a phrase learners actually type, like "JLPT N5 vocabulary list" or "JLPT N5 kanji list". Every word, kanji and grammar point also has its own page, about 1,000 in total. A search for one item lands on an answer, not a scroll through a list.

Build clusters, then link them. The N5 hub links to vocabulary, kanji and grammar, and each list links to its items. Word pages link to their kanji, grammar pages link to the points they're often confused with, and breadcrumbs lead back up.

Get the foundations right. Every page has a canonical address, a clean readable URL, and a title and description written from its own content. Pages load fast and read well on a phone. My own audit flagged gaps, such as structured data and a sitemap, and I'm working through them.

Get indexed fast. I verify the site in Google Search Console and Bing Webmaster Tools, and use IndexNow so new and changed pages are picked up quickly. Clear, well structured answers also give AI search tools something accurate to cite.

Earn trust in public. The About page explains where content comes from, every source is credited under its licence, and pass marks cite the official JLPT site. A clear "not affiliated" note and a report link on every item help too. Named authorship is the next signal to add.

Make things worth linking to. Complete free lists, the exam on one page and grammar lessons with practice questions are the kind of reference a teacher or study forum can link to. That's a slower path than chasing links, but it compounds.

What's next

• Practice in the real exam formats. Questions for each section, so learners rehearse the exact question types they'll face.
• Reading and listening. The two stops on the study route still marked "coming next", including listening audio.
• Timed mock tests. Full length tests scored against the real pass marks and section minimums.
• Beyond N5. More guides first, then the same treatment for N4 and the higher levels.