Provenance. §1 = reviewed unit (candidate gold, Watson-grounded; inter-annotator agreement is an LLM-panel proxy for human IAA). §§2–3 = candidate (proxy IAA, not human-calibrated). §§4–73 = machine-assisted, unreviewed heuristic exploration. Interpretive layers are valid-but-heuristic rule-sets — coverage ≠ accuracy, not a digital edition. Editorial statement · Error budget
73 sections 6903 tokens 14 comparable HTML builds syntax + semantics + pragmatics

01. Layered Span Lab

Formal tracks align speech position, clause contour, phrase role, construction, and morphology.

02. Word Microscope

A close-reading interface for one word's grammatical addresses and partners.

03. Discontinuous Arcs

Selected-word arcs expose separated Latin partners and nested membership.

04. Contour Ribbons

Clause bands turn the Ciceronian period into visible terrain.

05. Metro Graph

Roles become routes; the selected word becomes an interchange station.

06. Wave Model View

The sine/cosine positional-encoding analogy as computational model view.

07. Semantic Roles

Words become participants in event frames: agent, patient, goal, anchor, circumstance.

08. Referent Chains

Cicero, judges, Sicilians, Verres, and role nouns form visible mention chains.

09. Pragmatic Map

Speech acts, audience targets, discourse moves, and information status become lanes.

10. Rhetorical Heatmap

Ethos, pathos, logos, attack, defense, and institutional pressure become measurable heat.

11. Multidimensional Word Lab

The integrated microscope: syntax, semantics, pragmatics, rhetoric, morphology.

12. Watson Philologist Expert

Machine readings and Watson-reviewed corrections are compared token by token, with grammar hooks and source basis visible.

13. Philology Review Queue

Corpus-wide adversarial review candidates are filtered by priority, trigger, status, and Watson basis.

14. Watson Full Fix Dashboard

The Watson specialist pass applies deterministic corpus-wide fixes and shows unresolved philological work.