Should parser_for(locales.VI).parse("Giáo sư Nguyễn Văn An") read title Giáo sư, the spelled-out Vietnamese titles? #1113

locales.VI (#1108) reads the abbreviated academic titles written before a Vietnamese name: TS, ThS, GS, PGS and the stacked PGS.TS.. The spelled-out forms read as names:

Each title is two syllables (giáo sư, tiến sĩ, thạc sĩ, phó giáo sư, bác sĩ, kỹ sư, dược sĩ, luật sư), and a title reading is one word. Several syllables are common given or middle names (Tiến, Sĩ, Bác), so a plain title reading would take them.

Proposal, in rule shape

• When: a name under locales.VI opens with a spelled-out Vietnamese title.
• Then: the two syllables read as one title.
• Because: the title is a phrase of two words, and a syllable that is also a name is a title only inside one (rules/H9, the title phrase).
• How: each syllable gets an academic reading written default = false in vi.toml, beside a given reading that keeps it a name outside the phrase, and attaches links each first syllable to its second. Prototyped through a caller's lexicon on Should parser_for(locales.VI) read Nguyen Van Anh as family Nguyen, middle Van, given Anh, and Le Van Anh as family Le? #1108's branch (2026-11-06):

Open

• A comma between stacked titles. CLDR 48's vi file writes Giáo sư, Tiến sĩ Nguyễn Văn An, and the comma splits the phrase: title Giáo Tiến sĩ, family sư. Reading it needs the comma rules, not vocabulary alone.
• The unaccented spellings (Giao su, Tien si, Thac si) need their own readings. si is en.toml's credential and the others need bearer counts, per nameparser/vocab/AGENTS.md.
• The three-syllable phó giáo sư needs a phrase of three words, which attaches doesn't make today.
• The ranks (đại tá, thiếu tướng) have the same shape. The deputies' data writes them in the position, never the name.

Measure: no local corpus writes a spelled-out title inside a name. UIT-ViNames and vn-assembly's official names carry none, and the deputies' titles are composed abbreviated. A corpus of names as Vietnamese text writes them (signature blocks, press bylines) is the evidence this needs before vocabulary lands.

Related: #1108, #1112.