the manual

Kanji Build user guide

version 1 · updated 26 February 2026

Kanji Build is not a flashcard app. It is a relationship map for the ~2,800 characters used in modern written Japanese. Every screen answers one of three questions:

• What is this kanji made of? — visual and semantic decomposition.
• Where else does this piece appear? — the productive radical & phonetic families.
• What sounds and words fall out of that structure? — on'yomi phonetic series and the compound vocabulary that uses them.

The guide below runs through every feature, links straight into the app so you can try each one, then finishes with three worked learning journeys that chain the features together. The historical background — oracle-bone script through the modern Jōyō list — sits inside the expandable panels so you can dive in when you want to and skip past them when you don't.

1. Play the 90-second guided tour — it drives the real app and shows the three lenses in motion.
2. Open the Home page and pick a browse mode: JLPT, Strokes, Radicals, or Phonetic.
3. Tap any kanji to open its focus card. Read its etymology, then follow either the “built from” row (its components) or the “used in” row (larger kanji that contain it).
4. Mark what you know with the New / Learning / Known / Hide buttons — this feeds the “ready to learn” highlight and the compound previews everywhere else.
5. When you spot a family that all shares one reading, jump into the Phonetic Atlas to see the rest of that sound family.

Every feature in Kanji Build resolves to one of three ways of looking at a character. Knowing which lens you're using at any moment is the single most useful thing to internalise.

The Home page is the main index. A single browse by toggle at the top switches which lens you're navigating through:

• JLPT — N5 → N1. The path is chaptered into stages that reveal each level's new kanji only after its prerequisites are marked known. This is the closest thing to a traditional textbook progression.
• Strokes — buckets of 1 → 14, then 15+. Useful for the novice tracing period and for hunting a character whose reading you've forgotten. Secondary sort freq vs alpha reshuffles inside each bucket.
• Radicals — 382 productive semantic components, sorted by frequency or by the number of Jōyō kanji they generate. Tap any radical to see every kanji it appears inside.
• Phonetic — jumps to the Phonetic Atlas (see below) for browsing by shared on'yomi driver.

Below each mode selector sits a progress ribbon: how many characters in the current view are marked known, learning, or still new. Kanji tiles are ringed in the same colours so the state is obvious at a glance.

Every tile leads here. The card is a single scroll and layers the three lenses in order:

Header

Giant serif glyph, meaning gloss, and its JLPT badge. Below the header sits the animated stroke sequence — tap to replay. The strokes are drawn in the correct pedagogical order (top-down, left-right, enclose-then-fill) so you can trace along.

On'yomi & kun'yomi

Two rows of reading chips — on'yomi (Sino-Japanese, katakana) and kun'yomi (native Japanese, hiragana). Each on'yomi chip is a link that opens the Phonetic Atlas pre-filtered to that reading, so you can immediately see who else shares this sound.

Katakana origin chip (if applicable)

A small chip appears when the character is the historical source of a katakana glyph (e.g. 加 → カ). See the kana etymology section below for the full explanation.

Built from · Used in

Two horizontal rows of component chips:

• Built from — every KanjiVG element inside this glyph, tappable. The interactive tree beneath renders the same decomposition graphically.
• Used in — every larger kanji that contains this one as a component. This is the “learn 1, unlock N” row. A subtle ready to learn flag lights up when all of that kanji's components are already marked known — a strong signal that the compound will be trivial to add next.

Phonetic driver & siblings

If this glyph belongs to a keisei-moji (phonetic-semantic compound) family, the driver character appears here with its siblings. Reading Chinese wasn't random — around three quarters of Jōyō kanji were formed this way, and the driver almost always survives as the on'yomi. See the historical section for the mechanics.

Compounds

A live list of real words that use this kanji, colour-coded by the status of their constituent kanji: green if you know all pieces, amber if some, grey if you're missing components. This is the same “ready to learn” heuristic applied to vocabulary — click any word to open its vocab focus card.

Progress & mnemonic

A shared control at the bottom lets you mark the kanji as New, Learning, Known, or Hide. Above it, a plain-text editor stores your mnemonic — a private note that shows up wherever the kanji appears in the app. Notes sync to your account when you sign in.

The same card shape as the kanji focus, but scoped to a multi-character compound. It splits the word into its constituent kanji chips, shows the reading and translation, exposes the same progress controls, and lets you jump into any component character with one tap. Reached by tapping any word in a compound list or via the search bar.

The Phonetic Atlas is a single-page index of every on'yomi in the dataset, grouped by gojūon initial (ア, カ, サ, タ, ナ, ハ, マ, ヤ, ラ, ワ). Under each initial:

• Every reading has a driver — the seed character whose shape the whole family inherits. Drivers are drawn in shu red.
• Under each driver, the family members that share the reading are listed. A filter above the atlas lets you show only readings with 3+ members, or only readings whose driver is itself absent from the Jōyō list (“ghost drivers”).
• The reverse lookup panel accepts a katakana glyph and returns every phonetic driver that produces that sound — useful for the “I heard shou but which kanji?” case.

The atlas is the fastest way to understand why beginners legitimately confuse 待, 持, 時 and 詩 — they all inherit タイ / ジ / シ from the same driver 寺.

The Study page is a spaced-list of the characters you have currently marked as learning, ordered so that the ones whose components you already know surface first. Tap any tile to jump into its focus card, then use the progress buttons to promote it to Known or send it back to New.

The Journal is a chronological feed of every mnemonic you've written and every character you've promoted to known. Think of it as a lab notebook — a record of your thinking, not a review deck. Each entry links back to its focus card.

Every katakana glyph was originally a shorthand corner of a specific kanji. Kanji Build surfaces this in two places:

• On kanji cards — a chip on the source glyph says “this is where the kana X came from”. Example: 加 → カ. If the kanji still reads with that sound today (a “retained sound”), the chip flags it and offers to jump you into the phonetic family.
• Reverse lookup in the Atlas — type a kana into the Phonetic Atlas reverse panel and see every kanji whose historical simplification produced that shape. Roughly half of the modern katakana set has a productive kanji source, so this is a real navigation tool, not a curiosity.

Four states, applied per character and per compound. They drive the “ready to learn” highlight, the compound colour-coding, and the visible progress ribbons on Home.

Progress lives locally by default and syncs to your account once you've signed in — see the account section.

The search bar in the header accepts three inputs: a single kanji, a hiragana / katakana / rōmaji reading, or a whole compound word. Results are sectioned by kind (single characters first, then multi-character words) and each result opens its focus card. There is no fuzzy matching by design — searches are deterministic so “I meant that one” never happens.

The 90-sec demo button in the top header (and the “take the tour” buttons on the Welcome page) drives the real app through a scripted seven-beat tour. It is the fastest way to see all three lenses in motion. The demo can be resumed from any beat by starting it again — it fast-forwards past anything you've already completed.

• Sign in — via Google. We ask only for your email, name and avatar (see the privacy policy). Signing in enables cross-device progress sync.
• Lifetime unlock — a single £20 payment via Stripe removes the demo gate. The refund policy is set out in the terms.
• Install to home screen — the Install app button on the Welcome page turns Kanji Build into an offline Progressive Web App on Android, iOS, Chrome and Edge.

Kanji Build's features only make sense against the actual history of the script. These panels are optional — expand whichever ones catch your eye.

Three worked walk-throughs that chain the features into a pedagogical arc. Each takes ten to twenty minutes; each teaches you a family of characters, not just individual ones.

Bushu (部首)

Radical. A recurring sub-component used as a dictionary index key.

Jōyō / Jinmeiyō

The official Japanese lists of “daily use” kanji (2,136) and “permitted for personal names” kanji (863). Kanji Build carries both, for 2,828 total.

Keisei-moji (形声文字)

A phonetic-semantic compound. One component gives the meaning, another gives the sound. Around three quarters of Jōyō kanji are formed this way.

On'yomi (音読み)

The Sino-Japanese “sound reading” imported from Chinese. Traditionally written in katakana.

Kun'yomi (訓読み)

The native Japanese “explanatory reading” grafted onto the imported glyph. Traditionally written in hiragana.

Man'yōgana (万葉仮名)

The 8th-century convention of using kanji purely for their sound — the direct ancestor of both hiragana and katakana.

Phonetic driver

The seed character whose sound is inherited by an entire keisei-moji family. Rendered in shu red in the Phonetic Atlas.

Ready to learn

A visual flag applied to a kanji whose components are all already marked known — a signal that the character is one small step away.