[derek73]

Closes #326.

What

The suffix's views were two, generation and credentials, and every word that was neither a generation nor a credential fell into credentials, honorifics included: parse("김민준 씨").credentials was '씨'. A form of address after the name now has a view of its own.

	generation	credentials	honorifics
Smith, John, Jr., MD, Esq.	Jr.	MD, Esq.
田中さん, PhD		PhD	さん
王小明先生			先生
雅婷婷婷王二世先生，議員		議員	先生

• The honorific kind (19 words: さん, 様, 씨, 님, 氏, जी, साहब…) and the professional_honorific kind (11: 先生, 教授, 선생님, 社長, 課長, 部長…) compile into a new internal set, HONORIFIC_WORDS. Token.reading reads these words as the new ReadingKind.HONORIFIC.
• A courtesy (Esq.) and a status (ret., ז"ל) stay in credentials, read as post-nominal. CLDR's own examples of its credentials field include Esq..
• Six offices that CLDR's samples file as credentials (議員, 议员, 国会議員, 國會議員, 太平紳士, 太平绅士) are now credential words written as words, as 博士 already was. CLDR files 先生, 教授, さん and 님 as titles, which is the honorifics view's ground.
• ParsedName.honorifics is a view like family_base: a render key ({honorifics}), not a field of as_dict().

Why now

credentials is new in 3.0. Narrowing it after 3.0 ships would change a released view. CLDR's en pattern writes {surname} {generation}, {credentials}, which with the honorific in credentials would render Kim Minjun, ssi once a pack carries patterns.

What moved

• No field of any parse. suffix keeps every word, and as_dict(), replace() and the repr are unchanged.
• Views only. 73 of the snapshot's 6,793 readings record the new view. In 21 of them it sits beside other suffix words (田中さん, PhD).
• Ambiguity.alternative. It names honorific where 2.7.0 and 2.8.0 named post-nominal for six lone Devanagari and Bengali words (जी, साहब…).

Docs

• rules/R8 is renamed for its three views (R8-generation-credentials-and-honorifics-views.md, in both rules/ and decisions/).
• decisions/R8 has a dated 2026-11-03 Should suffix split into generation and credentials? (Smith, Jr., PhD) #326 entry: the options weighed, the CLDR evidence for the split, and the recompute recipes.
• The Open line in decisions/suffix-field-composition is closed.
• R1, the S Background, decisions/vocabulary-model, the RFC, usage.rst, the release log and tools/differential/README.md are updated.
• The snapshot records honorifics where it is not empty (RECORDED_LATER).
• rules_budget.json is re-recorded: R8's Then grew by one clause, which reads the signal it already read, each word's reading kind.

Verification

• Full suite on 3.11 and 3.15, ruff, mypy, build.py --check and the Sphinx doctest build all pass.
• Three independent reviews: code, design docs, and the fix commits.
  • The design-docs review found the CLDR evidence that split professional_honorific.
  • The fix-commit review parsed 90 office-word texts under six configurations (default, FAMILY_FIRST, ZH, JA, strict zh and strict ja) in both trees. Nothing moved except views.

🤖 Generated with Claude Code