Evidence · explorer

The coded corpus

Interrogate the data yourself. Scope honesty first: this explorer operates over the coded specimen dataset, not full-text search of the multi-million-word corpus, and it never displays a comparative claim it hasn’t earned.

Current state of the dataset: 22 specimens are coded across the catalog (eight closure devices, four force multipliers) — the launch pool of ≥150 coded utterances is under construction, and every specimen currently carries a verification-pending flag. Density figures below are [descriptive — F2 baseline comparison pending]: they describe this corpus and compare it to nothing, because the comparison test (F2) has not been run.

Specimens per device

CodeDeviceFamilyCoded specimens
D1 Persecution transfer Identity fusion 2
D2 Referee removal Referee destruction 2
D3 Evidence laundering Ownership evasion 2
D4 Savior monopoly Stakes inflation 2
D5 Nominal verdict Referee destruction 1
D6 Heads-I-win priors Referee destruction 2
D7 Retroactive frame control Ownership evasion 2
D8 Apocalyptic license Stakes inflation 1
D9 Moral recategorization Stakes inflation 2
D10 Zone flooding Stakes inflation 2
D11 Superlative saturation Ownership evasion 2
D12 Voice merger Identity fusion 2

query: all specimens in devices.json, grouped by device · [descriptive — F2 pending]

Planned facets

At v1.0 this page gains: timeline (specimens per quarter), co-occurrence heatmap (devices within the same speech), venue-type facets, and CSV download of any slice — all precomputed at build, filtered client-side, no server. The full dataset ships on the data page.