The dictation fingerprint

Status: Already answered

Status is derived only from the shepherd-authored triage/prediction data above -- community submissions and claims are a separate overlay and can never change it (see the participation panel below).

This is a proposed connection between two domains, generated by a language model. It is not an article and not evidence: it sits below the evidence/publication boundary. A quantitative prediction and a named kill-dataset are attached (when registered) so the claim stays falsifiable rather than merely evocative.

Claim (verbatim)

The dictation fingerprint. A scribe copying from an exemplar in front of him makes errors of the eye — confusing letters that look alike — whereas a scribe taking down a text read aloud, as in dictation or the pecia system of mass university production, makes errors of the ear, confusing words that sound alike. Because these two production modes stamp copying errors with different signatures, the mode itself should be recoverable from error typology alone, without external documentation of how a manuscript was made. The prediction is a bimodal distribution: the fraction of phonetically-confusable versus visually-confusable errors across collated witnesses should split into two clusters, with the high-phonetic mode concentrating at known dictation and pecia mass-production centres and periods — 13th–14th-century university towns — and the low-phonetic mode in monastic visual-copying contexts.

Prediction clause (verbatim)

The fraction of phonetically-confusable (vs visually-confusable) copying errors across collated witnesses is bimodal, with the high-phonetic mode clustering at known dictation/pecia mass-production centers and periods (13th–14th-century university towns) and the low mode in monastic visual-copying contexts.

Kill-dataset (verbatim)

Kill: a unimodal error-typology distribution, or no association between the high-phonetic mode and known production contexts.

In the atlas

This conjecture is bridged, as an L1 lead, onto these Inferpedia subject pages.

Provenance

Run: Fresh agent generation · model: claude-fable-5

Authored by the shepherd session (Claude Fable 5) as the recorded instrument, drafted 2026-07-04 in the session scratchpad (fresh_conjectures_draft_20260704.md) after Phase A was launched and BEFORE any triage literature search for this pilot; imported immediately after Phase A deployment and before the B2 triage pass began, so the fresh-lane ModelRun timestamp precedes all triage ModelRuns. Novelty unverified: the author cannot rule out prior formulations in the literature; these enter the same triage lane as the imported harvest.

Novelty / leakage triage

already answered in the literature

The shepherd's own fresh item, judged leaked by its own triage: inferring production mode from phonetic-versus-visual error typology IS the method of the classical dictation debate — Skeat's 'The Use of Dictation in Ancient Book-Production' (1957) and the Codex Sinaiticus dictation controversy turn on exactly this discrimination. Worse for the conjecture, Junack's critique (ancient scribes self-vocalized while copying, producing phonetic errors without a lector) undermines the clean bimodality premise. The pecia-clustering systematization was not located, but the core join is published. Retained as an instrument-calibration point: the triage catches its own author.

Skeat, 'The Use of Dictation in Ancient Book-Production', Proceedings of the British Academy 42 (1957) — The core method
'An Unseen Paleographical Problem with Milne and Skeat's Dictation Theory of Codex Sinaiticus' — The debate incl. Junack's self-vocalization critique

Its literature citations feed the frontier as source leads (2 leads below the evidence/publication boundary, not yet reviewed).

Predictions

Open registered 2026-07-04 calibration prediction (parent triage: leaked/adjacent)

Calibration-tier OPEN prediction (triage: LEAKED — this is substantially Skeat's dictation-debate method; registered only as a calibration of whether the old method survives a modern corpus-scale test): the phonetic-vs-visual error-type fraction across collated witnesses is bimodal, with the high-phonetic mode clustering at known pecia/university production contexts.

Resolution criteria: Resolvable against large collated corpora with error typology (e.g., annotated Greek/Latin collations). SUPPORTED if a two-component mixture fits the phonetic-fraction distribution better than one component (likelihood ratio p < 0.01) AND high-phonetic membership predicts documented pecia/university origin better than chance. KILLED if the distribution is unimodal. MANDATORY confound control from the triage (Junack): self-vocalized copying produces phonetic errors without dictation — the analysis must include an error class that discriminates lector-mishearing from self-vocalization, else INCONCLUSIVE.

Known priors disclosure: The registrant knew of the Skeat dictation debate before generating this item (confirmed leaked at triage — the instrument caught its own author). No corpus-scale bimodality test has been seen.

Weigh in

No community feedback yet.

Add your take

Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and never changes this conjecture's triage label or status above.