Structural Register Scatter

H6 asks whether citation style plus grammar marking predicts dictionary family. This page is dictionary-first: it uses generated dictionary structure data only, not corpus frequency, standards exports, or GitHub activity.

Trust Block

• Evidence: data/dictionary-coverage.json joined to data/lexico/microstructure_fingerprint.json.
• Limitations: clusters show structural similarity, not direct descent.
• Validation: generated by npm run build-structural-register; source data by npm run build-coverage and python scripts/lexico/validate_lexico.py.
• Owner repo: csl-atlas.
• Next use: review H6 edge and family-outlier prompts before using the chart as a paper-level lineage argument.

Citation Register x Grammar Marking

The x-axis uses max(taggedCitationPct, inlineItiPct), so indigenous iti evidence is visible instead of being treated as zero <ls> evidence. Tooltips retain both raw components.

Citation Mode By Family

Dictionary Rows

How To Read It

• A cluster is evidence for structural similarity, not direct proof of descent.
• A zero can be a convention gap. It does not automatically mean the dictionary lacks the feature.
• iti and <ls> are different citation registers and should remain visible.
• H4 semantic fields live on /tools/semantic-fields; this chart should not mix topical vocabulary into the H6 structural-register claim.

Generated by npm run build-structural-register. See docs/H6_STRUCTURAL_REGISTER_SCATTER.md. CC-BY-SA-4.0. Companion docs: MICROSTRUCTURE_METHODS.md and MICROSTRUCTURE_FINDINGS.md.