Suffix-acronym collisions: rai and cha out, ba marked, #454 by design

[derek73]

Two issues, one question asked of different words: when a post-nominal acronym is also somebody's name, does the entry get marked, removed, or left alone? No parser code changes — two existing forks carry it, and both were confirmed on the pre-bundle tree with a throwaway Lexicon override before any wordlist was edited.

The criterion. decisions.md#vocabulary-collisions C-i decides ship-or-mark. This adds the third answer and says when to reach for it: the four entries that carried the ambiguous marking before this change (ma, do, ed, jd) are all TWO letters, which is not a coincidence — two letters is too few for an acronym to be anything but common, so both readings are live and a flag is honest output. A longer credential that is tenuous or specialized and collides with a common name has no second reading worth carrying, and is removed.

rai and cha leave SUFFIX_ACRONYMS (#342). Both arrived in the 2019 bulk Wikipedia post-nominal import and neither was reviewed against a surname. Aishwarya Rai reads family Rai — 1.4.0's own answer — where 2.0.0 through 2.2.0 read suffix Rai with no family name at all; Lala Lajpat Rai reads middle Lajpat, family Rai. The cost is stated and pinned: John Smith RAI reads family RAI, and Ahmad Jayadi, CHA reads given CHA, family Ahmad Jayadi, the comma being an ordinary family comma once nothing after it is credential vocabulary. Lexicon.default().add(suffix_acronyms={"cha"}) puts either back.

ba joins the ambiguous subset. BA is a common credential and Ba a real surname (Vietnamese; Senegalese Fula) — the ma/Ma shape exactly. Anna Ba reads family Ba flagged; John Smith BA keeps suffix BA and gains the flag; John Smith B.A. stays an unflagged suffix through the period gate; Smith, BA reads given BA, which is the Smith, Ed cost S2 already accepted.

#454 closes by design. mc has no vowel and is not a borne name, and the Scottish prefix never detaches from the name it belongs to, so Donald Mc is not a name anyone writes. Mc Donald → family Mc Donald and John van Mc → family van Mc hold on the never-given particle membership (#360), which is the one that actually bears on the example. What #454 got right is fixed instead: rules.md#P6's Accepted clause promised that a comma-less name keeps its positional reading, with a de-shaped example that could never falsify it. Measured, it fails on exactly the two words in both the particle vocabulary and the UNAMBIGUOUS suffix vocabulary — mc and vd — and the clause now says so and names them. No example line moves, so corpus_rules.jsonl regenerates byte-identical.

Measurement. Five corpus names move, every one from the radar-tier corpus_issues.jsonl, in three diff shapes with one cause: Aishwarya Rai {family, suffix}; Lala Lajpat Rai and John Smith RAI {middle, family, suffix}; John Smith, RAI and Ahmad Jayadi, CHA {given, family, suffix}. The ba marking moves none. Aishwarya Rai is the one name whose diff DISAPPEARS at 1.4.0 — that release read family Rai too — so the fix(#342) NOT WANTED rule was deleted as its own comment instructed and the 1.4.0 gate lists four names under the replacement where the three 2.x ledgers list five. Gate: 367 / 263 / 175 / 37 intentional at 1.4.0 / 2.0.0 / 2.1.0 / 2.2.0, 0 unexplained, 0 radar unclassified, corpus sizes unchanged.

Also carried: Derek's 2026-08-23 amendment from #342, verbatim, on the comma-suffix-arc Declined entry; the Excluded (MAIDEN_MARKERS) z domu bullet rewritten as LIFTED (the phrase shipped 2026-08-26 via #434, and the bullet had still called it pending); the contested-membership Open: block emptied and saying so; and docs/customize.rst brought in line with C-i — it told callers to leave an uncertain word out, which is the inverse of "under uncertainty, default to AMBIGUOUS".

Closes #342
Closes #454

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.66%. Comparing base (266154d) to head (75cd103).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #515   +/-   ##
=======================================
  Coverage   98.66%   98.66%           
=======================================
  Files          45       45           
  Lines        3232     3232           
=======================================
  Hits         3189     3189           
  Misses         43       43           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.