One vocabulary curriculum for a whole school
In my first semester at Dongsung Elementary I built a master word list meant to last: five grades, four proficiency levels per grade, 20 lists in all, about 800 words total. Second semester and every term after it reuse the list instead of starting over.
This page covers how the tests are sequenced and generated. For how the 800 words themselves get chosen — the six weighted gates and the hard rules on top of them — see My Creations.
Who did what. I designed the curriculum, the difficulty progression, and the test sequencing. The code is Claude's: I directed it to build the Python and lxml pipeline that produces print-ready tests and answer keys from a single source, and to track down a recurring formatting defect in the test template.
What I designed
- A difficulty progression aligned to CEFR levels, across the four levels.
- Test sequencing built on retrieval practice and spaced repetition. Word order is re-randomized on every test, and each test deliberately overlaps with the ones before it so words return after a gap.
- Uniqueness constraints across the proficiency groups, so lists do not quietly repeat each other.
- An audit of the full set for duplicates across grades and for translation ambiguity, the kind of word whose translation can mislead a student.
Why one source matters
Every test and answer key is generated from the same master list. Fix a word once and every printed document changes with it. The list was built to survive past the first semester, and reuse was a design requirement from day one.
How the scores get used
I level students with TOEIC Bridge scores. Vocabulary test results then regroup them into three differentiation tiers, which I call mild, medium, and spicy, inside cooperative groups of four based on Kagan structures.
For the classroom side of this, see the team points system.