[derek73]

The third title batch of phase 2j (#752): buddha, siddha and gyani (#869), the first halves of Nepali compound given names, gain declared given readings, so each is a title only inside a title phrase. guru stays a title, for a ne pack in phase 3 (decisions/vocabulary-model, 2026-10-24).

What changes

input	1.4.0 – 2.6.0	now
Buddha Bahadur Tamang	title Buddha, given Bahadur	given Buddha, middle Bahadur
Gyani Maya Gurung	title Gyani, given Maya	given Gyani, middle Maya
Joshi, Siddha Raj	title Siddha, given Raj	given Siddha, middle Raj
Mr Buddha Tamang	title Mr Buddha	title Mr, given Buddha
His Holiness Buddha Ram	title His Holiness Buddha	unchanged (a style is a partner)
Giani Zail Singh / Guru Prasad …	title	unchanged

No listed phrase holds any of the three, so nothing is reported. The other sites read as for mahdi (#898):

• ending the name;
• alone after a full name and a comma;
• past a second comma;
• before a family comma, where H6 no longer peels the word (Buddha Tamang, Ram → family Buddha Tamang, 1.4.0's reading).

Costs (rules/H9, Accepted)

None is written by any name in the eight local corpora or the committed ones:

• The Sikh title spelled Gyani before a name reads as a name word (Gyani Zail Singh → given Gyani). Its usual spelling Giani has no given reading and keeps the title.
• A title before the word that is no style, modifier or attaching title is no partner (Lord Buddha Tamang → given Buddha; family Buddha under a family-first order).
• The epithet with its article reads as the Mahdi does: Ram Tamang, the Buddha → given the Buddha, and The Buddha Ram → given The Buddha.

Measured

Against master 021028b2 (after #898), py3.11, every local corpus with --base; counts and token shapes only.

• nepal-candidates:
  • 121 romanized names moved: 117 the word read out of the title alone, and 4 gaining or losing a report.
  • 2 lose H4's title-or-name on a name that was all vocabulary (Gyani Lama).
  • 2 gain the suffix-or-name any name with a trailing credential word and words to spare gets (Buddha Lama ma reads as Ram Lama ma).
  • Romanized/Devanagari pairs: 128 now alike, 0 apart.
• ctgov-officials: 4 moved, all fixes. Two of them are one official written twice, whose credential master read as the given name.
• No moves: PSC (1,039,174 scored), fec-individuals, tse-donors, jrc-names, the committed corpora, CLDR/GND agreement.

Docs

• rules/H9:
  • One example added.
  • A new Accepted bullet for a religious title that opens given names in another tradition.
  • The connective Precedence bullet now covers a period-marked word before and: Dean. and Mary Smith reads given Dean. and Mary, as 1.4.0 through 2.6.0 did. The H2 Precedence bullet names that exception and now covers the whole leading run (Mr Buddha. Tamang → title Mr Buddha., as master read it).
  • The rule budget is re-recorded (H9 626 → 660 words; H2 gains a Precedence bullet).
• rules/H2: a P3 Precedence bullet for the same older gap (Xyz. and Mary Smith → given Xyz. and Mary), which H2 never stated.
• decisions/H9: a dated entry with the counts, costs and measurement.
• decisions/vocabulary-model: a pointer entry.
• Release log: a new bullet.
• Readings on the words:
  • buddha: SSA 6, Wikidata 4.
  • siddha, gyani: no count; each note cites nepal-candidates (12 and 38 pairs; Buddha 83).
• Case rows: 10.

Checks

• uv run pytest and --python 3.15: 26,625 passed each.
• Also: -n0 -m serial, mypy, ruff, sphinx-build -b doctest.
• A design-docs review in two rounds over a 2,145-parse grid. Its findings (the partner clause, family-first fields, unrecorded H6/epithet/and sites, an example that did not reproduce, then H2's and H9's claim that the period shape always wins, and the Accepted bullet's period spelling) are fixed here.

guru is deferred to a Nepali language pack and tracked on #753, the phase 3 pack issue, whose ne-pack comment covers guru and श्री.

Closes #869. Refs #752, #853

🤖 Generated with Claude Code