The family unit
The smallest teaching unit is a character family, not a single character. Every design decision must trace back to at least one of the seven axioms below. The upshot: a standard module needs no new infrastructure — only the existing research norms (database schema, the six-section template, review classes) turned into a teaching order by these axioms. Method
1. Common reading first
2. Design reading
3. Ablation
4. Transfer
Seven design axioms
- Family unit: the smallest module is a character family with at least one minimal pair, not a single character.
- Two-track order: every unit runs common reading → design reading → ablation → transfer.
- Pairs: new characters come in pairs or groups; a lone character needs a reason.
- Layer order: form → meaning → rationale, matching the database's three layers and the six-section template exactly.
- Room for insight: each unit has a task before the reveal, and no explanation may come before it.
- Transfer exit: each unit ends by predicting unseen characters, counted in the main measure.
- Visible review: every design reading shows its review status; candidates stay out of exercises and tests.
The six-section template
Research entries and teaching units share one structure: headword → form layer → common reading → design reading → family → review status. Submitted explications follow it too (see the submission guide).
First prototype: 田 / 由 / 甲 / 申 / 神 Planned
Step one, the study of the underlying thinking, is done: five primitives and seven axioms. Step two is the prototype module, starting with the family 田, 由, 甲, 申, 神. 田 and 由 are the ablation minimal pair: the same parts (囗 + 十), differing only in which sense of 十 the structure selects. A confirmed correction: the reading “reaching upward” belongs to 由, not 甲; 由 → 申 → 神 is a derivation chain attested in the manuscript. The unit's design readings follow the book and stay out of exercises and tests until the author has reviewed them.
Trial plan: 10–30 real learners, assessed on acquisition, retention, transfer and insight, with an explicit go/no-go gate; course production scales up only after the trial passes. The rule is smallest closed loop first — one complete unit working before anything is rolled out.
