L5 · Roots & etymology

Statistics over verbal roots and derivational morphology: the MW root inventory, etymology derivation tables, root-oracle agreement, and the Whitney × DCS audit. Part of the statistics census overview (H817 WS1.3).

:::note Trust block. Source: stats_census_register.csv, rows where layer = L5, aggregated from csl-orig v02 and WhitneyRoots. n = . As of 06–12-07-2026. This is the one layer in the register with zero ○ not-started rows — the descriptive base here is fully closed. :::

Headline magnitudes

How to read: log-scale bar of root/etymology counts. Example: the Whitney × DCS audit (935 roots) sits between the 41-pair oracle-agreement matrix and the 2,113-root MW inventory — it audits a large minority of MW's roots, not all of them, and is explicitly capped: unaccented DCS text cannot split verb class I/VI, so corpus root-class verdicts do not appear here as a bare number (see the full table's caveat column-equivalent, the value_display field).

Status breakdown

Full table

Download: stats_census_register.csv (full register, all layers) · Data downloads.

Known ceiling

Per the roadmap's risk register: unaccented DCS cannot split verb class I/VI or IV/passive (accent collapse). Do not over-claim on root-class statistics beyond what the Whitney × DCS audit already flags as capped.

Chart density note: 2 Plot.plot calls (magnitude bar + status bar) — justified per the same heterogeneous-units reasoning as the L1 page.