You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A verification audit of PR #676 paired all 1,054 units against pre-merge main and checked each change class mechanically. The audit confirmed the sweep's core invariants and found three defects, all in the fifteen units the sweep edited by hand rather than through its checked scripts. This PR repairs them.
What the audit confirmed
Every citation line is byte-identical to pre-merge main, and all six source reconstructions remain exact. The 175 changed units divide into 63 frame strips, 37 onym substitutions, and 75 English-only reorderings, with nothing outside those classes. The English-order commit touched no Phi or gloss line. Every automated frame strip preserved its inner speech exactly and stands where Morris neither reopens a quotation nor re-attributes. Every onym substitution sits on a citation carrying a personal name rather than an epithet. A pronoun-fidelity scan across all changed units cleared everything except the two defects below.
The three repairs
Chapter 5, the friend-Bob line. Morris writes "and some, as you heard about friend Bob yonder". The sweep's onym rewrite produced Phi saying mia ... hea, I heard, under derived English saying "you have heard": unfaithful to Morris in the Phi and internally inconsistent between the layers. Dick is speaking, reminding Guest of what Guest heard, so the hearer is thia. The unit now reads whu thia roa to hea ra melu ne selomi, about whom you heard, with ra carrying Morris's "yonder" and the English matching. The inherited pre-sweep unit had the same wrong hearer, so this error survived one fidelity sweep and was then rebuilt rather than caught; only the layer mismatch made it visible.
Chapter 6, the patted horse. Morris re-attributes mid-speech: "even you like to work, old fellow--sometimes," said he, affectionately patting the old horse. The sweep's own rule keeps a frame wherever Morris re-attributes, but this unit was one of the two hand strips, and the hand applied the strip without the citation check the script enforced. The frame is restored, with the derived English in Morris's quote-first order.
Chapter 3, the broke-in line. The sweep's quote-order pass converted attribution-first English by matching a fixed verb list, and "broke into the talk" contains none of those verbs, so the line kept a colon-announced quote. It now uses the same shape as the sweep's other compound attributions: narrative sentence, then the quote with trailing attribution.
Scanner false positives, verified and left alone
Six audit flags were confirmed false: two case-sensitivity misses in the scanner itself, one plural possessive, "I suppose" correctly rendering the evidential ke, Morris's "It is said that" being hearsay inside the speech and correctly carried by ti, and a single Morris attribution whose citation text falls in a continuation unit while the kept frame in the previous unit already carries it.
Checks
validate_examples.py 0 errors, 0 warnings. validate_sentences.py parses the three touched chapters with 0 errors, and both repair sentences were parser-tested standalone before entering the file. All six source reconstructions remain exact. The phonetic-neighbour baseline moves by exactly the attestations the repairs predict, one mia fewer and one shia more, and is committed. The site builds.
Both "broke in" units in chapter 3 now render conversational entry as entry rather than as an English break calque, by the maintainer's ruling.
The inherited muo haolu to pukate imported the European rupture idiom whole (interrupt is Latin break-between), and imported its syntax with it: per the pukate entry's own usage note, the subject is the thing that breaks, so the sentences claimed that Dick himself broke. Both units now read muo haolu to reshi koema, came swiftly into the talk, using the entry verb chapter 1 already applies to the house door and the corpus's established fast-manner shape. The unit where Morris writes "Here Dick broke in:" keeps his attribution-first English; his exact wording carries the idiom in the adjacent line either way.
A family sweep prompted by the same question found the other break and sharp figures in the six chapters licensed by their entries: keno pukate breach of custom, pukate haoni breaking voice, and the tiso uses for wits, towers, and the wasp all sit inside their stated ranges. One borderline case remains unruled and unfixed: chapter 6's to pa tiso haolu for Morris's "said rather sharply", where the entry licenses edges and keen perception but not manner of speech.
Checks: validators 0/0, parser clean on the chapter, all six reconstructions exact, baseline unchanged, site builds.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A verification audit of PR #676 paired all 1,054 units against pre-merge main and checked each change class mechanically. The audit confirmed the sweep's core invariants and found three defects, all in the fifteen units the sweep edited by hand rather than through its checked scripts. This PR repairs them.
What the audit confirmed
Every citation line is byte-identical to pre-merge main, and all six source reconstructions remain exact. The 175 changed units divide into 63 frame strips, 37 onym substitutions, and 75 English-only reorderings, with nothing outside those classes. The English-order commit touched no Phi or gloss line. Every automated frame strip preserved its inner speech exactly and stands where Morris neither reopens a quotation nor re-attributes. Every onym substitution sits on a citation carrying a personal name rather than an epithet. A pronoun-fidelity scan across all changed units cleared everything except the two defects below.
The three repairs
Chapter 5, the friend-Bob line. Morris writes "and some, as you heard about friend Bob yonder". The sweep's onym rewrite produced Phi saying
mia ... hea, I heard, under derived English saying "you have heard": unfaithful to Morris in the Phi and internally inconsistent between the layers. Dick is speaking, reminding Guest of what Guest heard, so the hearer isthia. The unit now readswhu thia roa to hea ra melu ne selomi, about whom you heard, withracarrying Morris's "yonder" and the English matching. The inherited pre-sweep unit had the same wrong hearer, so this error survived one fidelity sweep and was then rebuilt rather than caught; only the layer mismatch made it visible.Chapter 6, the patted horse. Morris re-attributes mid-speech: "even you like to work, old fellow--sometimes," said he, affectionately patting the old horse. The sweep's own rule keeps a frame wherever Morris re-attributes, but this unit was one of the two hand strips, and the hand applied the strip without the citation check the script enforced. The frame is restored, with the derived English in Morris's quote-first order.
Chapter 3, the broke-in line. The sweep's quote-order pass converted attribution-first English by matching a fixed verb list, and "broke into the talk" contains none of those verbs, so the line kept a colon-announced quote. It now uses the same shape as the sweep's other compound attributions: narrative sentence, then the quote with trailing attribution.
Scanner false positives, verified and left alone
Six audit flags were confirmed false: two case-sensitivity misses in the scanner itself, one plural possessive, "I suppose" correctly rendering the evidential
ke, Morris's "It is said that" being hearsay inside the speech and correctly carried byti, and a single Morris attribution whose citation text falls in a continuation unit while the kept frame in the previous unit already carries it.Checks
validate_examples.py0 errors, 0 warnings.validate_sentences.pyparses the three touched chapters with 0 errors, and both repair sentences were parser-tested standalone before entering the file. All six source reconstructions remain exact. The phonetic-neighbour baseline moves by exactly the attestations the repairs predict, onemiafewer and oneshiamore, and is committed. The site builds.