Should Baron, Bishop, Judge and the other title words borne as names read as names outside a title phrase? (Baron Davis → title 'Baron', no given name) #913

Baron Davis reads title 'Baron', family 'Davis', with no given name and no ambiguity reported. The same happens with every word below: when it opens a name, it is taken as the title and the given name is lost (Deacon Jones, Judge Smith, Lady Gaga, Marquis Johnson).

#894 built the mechanism that fixes this, rules/H9 (the title phrase). A title word with a given-name reading is a title only inside a title phrase, where a style or modifier before it, a title that attaches to it, or of after it keeps the title. Everywhere else it is a name word, reported as title-or-name where no comma marks the given name. Phase 2j then moved seventeen words through it, one batch at a time: wing, mahdi, bard, buddha, siddha, gyani, king, sultan, sultana, shaikh, prince, princess, and five spellings of sheikh (#896 through #912). Each batch gave the word a dated given reading, measured every local corpus, and recorded its cost in decisions/H9. #348 asked the design question and is closed; this issue holds the words that are left.

The words still read as titles

The table lists title words of three or more ASCII letters that, opening a name, are read as the title with no given name left (<Word> Johnson). Each has at least 1,000 US given-name births. Counts are from tools/name_usage.py --no-wikidata on 2026-10-27; SSA is a lower bound, since years under 5 births are not published.

• royal, navy, rock and knowledge have the counts too, but already read as given names when they open a name, so they are not listed.
• queen and sir address by the given name, so the word behind them becomes the given name and the family is lost instead (Queen Latifah → title 'Queen', given 'Latifah').
• Should a word that is both a title and a given name report an ambiguity? (Baron Davis → title 'Baron', no given name) #348's own examples have small counts: count has 5 births, chief 348, and hon none, though hon has 2,424 surname bearers. Count Basie and Chief Keef are stage names, which birth records do not count.

The surname half, under family-first order

Under FAMILY_FIRST, the word opening a name is where the surname stands, so a title borne as a surname loses it: Graf Steffi reads title 'Graf', family 'Steffi', with no report. These titles have the most US surname bearers:

decisions/salutation-titles kept graf, greve and knight as titles on purpose, reasoning that the title claim acts only in front, where these surnames are rarely written. Under a family-first order, the front is exactly where they are written.

Questions

1. Which words move, and in what order? A word moves by gaining a given reading, as in phase 2j, and the counts above are where a batch starts. Some may be birth-record curiosities rather than names in use, to be checked by reading the bearers: author, lawyer, majesty and sailor.
2. Does a word with low birth counts but well-known bearers under stage names (count for Count Basie, chief for Chief Keef) move?
3. Should the surname half get a mechanism of its own, such as a family reading that makes a title phrase-only under a family-first order? Or should it wait for the packs (Vocabulary model and language packs (tracking) #747), where the reading order is declared?
4. duke is not title vocabulary, so Duke of Wellington gives given 'Duke of Wellington'. Should duke, and the other excluded peerage words earl and don, come in as phrase-only titles, as dean did in Read a title that is also a given name as a title only inside a title phrase (#752, phase 2i) #894?

To recompute the first table:

uv run python -c "from nameparser import Lexicon; print('\n'.join(sorted(w for w in Lexicon.default().titles if w.isascii() and w.isalpha() and len(w) > 2)))" | xargs uv run python tools/name_usage.py --no-wikidata --json