The claim
A24, with context C = (u, s, a, k, g) (TERM-031): translation changes the words while trying to keep the speaker’s relation to the audience, (u, s, a, k, g) → (u′, s, a, k′, g′); reattribution keeps the words and changes the footing, (u, s, a, k, g) → (u, s′, a, k, g) (CASE-018). “A translation may achieve pragmatic equivalence by compensating for changes in the social coordinates of an utterance through changes in its linguistic coordinates. In other words, lexical infidelity can be required for pragmatic fidelity.”
A18 had the observation without the mechanism: “fidelity is sometimes achieved through lexical deviation, not despite it.”
Where it went
The manuscript states the consequence (Abstract: an expression “can change almost every word while preserving the interpersonal relation enacted by the utterance”) and its Case III manipulates social alignment with objects held fixed, but the hypothesis that wording changes because the social coordinates changed, and in proportion to them, is not stated. Nothing in owner turns U1–U13 argued against it.
What it does not say
That any lexical change is licensed. The claim is that some are required, and which ones is set by the change of context.
Note of 2026-10-10: a documented case
CASE-066, found by a search the owner asked for, is a documented instance of the mechanism. Dolly’s one sentence of ты amid вы in Anna Karenina I.4 is kept and glossed by Dole (1886, “the familiar tui (thou)”), and carried by added words by Garnett (1901, “Stiva”) and the Maudes (1918, “our children”). The change is to the code’s address system, k and g, not to the speaker. Tolstoy himself says вы and its dictionary equivalent vous differ in coldness (III.14). So the claim no longer rests for its mechanism on the fictional CASE-018 alone, the point of CLAIM-229.
What it does not supply is the defeat condition’s data. Every judgement in the case about what the renderings convey is a translator’s or the author’s, not a reader’s, and no reader or reception study of T/V compensation into English was found. It also says nothing about proportion, and “our children” may carry a different relation, not the same one. The case’s last section sketches a reader test from its four versions, usable as items in CASE-055.
Note of 2026-10-29: what holds it back
No reader evidence that a compensating rendering keeps stance and force better than one that keeps the wording. CASE-018 is invented (ARG-013, Rejected), and CASE-066’s judgements are the translators’ and Tolstoy’s. The reader test CASE-066 sketches, Russian readers as the reference against four English versions, would meet or defeat the claim.