Testimony Without a Global Translator: Co-Reference, Faithful Transmission, and Shared Evidence Between Embedded Observers

This is the fourteenth paper of a series that redescribes physical reality as a causally consistent history of information updates, and the middle panel of its third triptych: reference, testimony, and objectivity. The thirteenth paper fixed, inside a history with no global interpreter, what a single token is of, what it says, under which scheme, and whether it misrepresents, by separating a standing representation architecture from target-relative provenance and by separating apparent content, referent, and full referential content. It also previewed, but did not analyse, an inter-observer translation whose provenance, reliability, and non-uniqueness were left for the present paper. The twelfth paper analysed when a single embedded observer may trust its own records, defining trust as layered warrant: objective process adequacy, a reflective adequacy credence, and a reflective entitlement conferred by a warrant bridge. This paper takes up the deferred inter-observer translation and lifts that three-layer warrant from one observer to a community. It asks under what conditions embedded observers with distinct standing representation architectures refer to the same target, transmit content faithfully, take one another's reports up as evidence, and count agreeing reports as genuinely independent corroboration. It argues that neither shared sign-forms, successful communication, nor informal community convergence secures formal co-reference, faithful transmission, warrant, or corroboration. Its contribution is threefold. First, it develops a typed architecture of testimony. It lifts the standing representation architecture to an observer specification, separates referent alignment into a task-relative comparison frame from content translation between content spaces, and separates the standing translation architecture from its individual translated tokens. Two lineages are kept apart throughout: target provenance, from a world target to a source record, and testimonial source provenance, from a source record through a report to a recipient inscription. Testimony is defined not as a kind of entity but as a cross-observer relation requiring a report role, a source lineage, and preserved source attribution at a declared grain, with uptake as the further step that turns received testimony into evidence. Second, it develops a typed status structure connected only by explicitly declared directional entailments. Agreement entails co-reference, faithful transmission entails testimonial source provenance, anchor adequacy entails target provenance of the designated evidential anchor, and objective corroboration entails concordance and co-reference among the relevant reports. No converse holds in general. Genuine disagreement, compatible difference, talking past one another, reference failure, cross-frame incomparability, reference indeterminacy, translation failure, faithful disagreement, and misleading agreement are therefore distinguished rather than treated as one undifferentiated form of disagreement. Third, it lifts the twelfth paper's three-layer warrant to a community: objective testimonial chain adequacy, a recipient-side adequacy credence, and testimonial warrant conferred by a recipient bridge. Each is relative to the claim for which the testimony is used, so adequate evidence that A asserted p is separated from adequate evidence that p. A parallel three-layer treatment of corroboration separates objective corroborative structure, a recipient's corroborative credence, and bridge-conferred corroborative warrant. There is no global translator supplying a unique mapping from outside the history. Translation is a history-grounded process whose architecture has a formation history, reliability, and failure modes, while its individual outputs bear token-level provenance and correctness statuses. Comparison verdicts are relative to declared frames and translation paths. Faithful transmission certifies neither target provenance nor the correctness of the source. The account of corroboration is limited to concordant testimonial evidence rather than a general theory of evidential convergence. Inter-observer invariance and objectivity are deferred to the third panel.