Should a strict choice still read PhD, Dr and Prof? (parser_for(locales.PT, strict=True).parse("Ana Souza, PhD") → given 'PhD') #1005

What happens

A strict choice reads only the chosen languages' files, plus native-script vocabulary. mul.toml, the file for words shared across languages, is treated like any other language's file (vocabulary-model-rfc.md, "File-strict"), so a strict choice drops it unless it is chosen too. en.toml holds most of the credentials and salutations that are written everywhere. So under strict pt or es:

Dr. with its period stays a title by its shape (rules/H2). Since #1004, the spaced Ph. D. follows phd, so both spellings leave together.

How often. Of ClinicalTrials.gov's 100,908 officials writing PhD or Ph.D., the suffix holds it in 100,560 under Parser() and in 45,469 under strict pt. Measured 2026-11-02 on master b2b75948, in the #1000 comment. The RFC's strict-losses table already lists what each strict choice drops on the local corpora (vocabulary-model-rfc.md, under "Strict moves little"): nine of ten rows lose a bare dr, three prof, and several phd and md. It concludes that "a strict choice made for real data is in practice strict of the language and en, or needs those words in its own file". The RFC leaves this open: "entries may later move between en.toml and mul.toml, or into the files of the languages that write them, so that strict en can shed problem words; which words those are is not yet known."

Two questions

1. Which words are every language's? For example, the credentials written after names in every country (phd, md, mba, msc?) and the abbreviated dr and prof already in mul.toml. Decide by corpus frequency per country: for each en.toml credential and salutation, its incidence in the fielded local corpora by country (ORCID, PSC, PatentsView), and how much of each strict choice's loss it accounts for.
2. Should a strict choice read mul.toml? Two shapes:
   • (a) Strict always reads mul.toml, as it reads native-script vocabulary, and the words in question 1 move there. Strict en then cannot shed them.
   • (b) mul.toml stays one more language's file, and a caller who wants the shared words chooses mul alongside their language (strict=True with pt and mul). This needs a spelling for choosing mul, and documentation.

Before deciding, measure the strict-losses table again with each option simulated, and count how many of each option's gains are words that are also names somewhere (a moved word keeps its name readings' gates only where those sit in a file the choice reads, per the RFC's "The derived markers under strict").