Should Armitage JO read family 'Armitage'? A surname followed by initials, with no comma, reads the initials as the family name #1137

The gap

Citation styles (Vancouver/PubMed, and Russian and Eastern European bibliographies) write a byline as the surname, then the initials, with no comma. nameparser reads by position, so the surname becomes the given name and the initials become the family. Measured on master 13e7314:

Armitage JO          → given 'Armitage', family 'JO'
Armitage J.O.        → given 'Armitage', family 'J.O.'   (suffix-or-name reported)
Armitage J. O.       → given 'Armitage', middle 'J.', family 'O.'
Gomes A.R.S.         → given 'Gomes', family 'A.R.S.'     (suffix-or-name reported)
SAFIN R.R.           → given 'SAFIN', family 'R.R.'
McWilliams John P.   → given 'McWilliams', middle 'John', family 'P.'
Armitage, J.O.       → family 'Armitage', given 'J.O.'   (right: the comma settles it)

This is the Latin-script twin of #995 (Иванов И.И. under locales.RU). #995 is scoped to the RU pack and leaves the default alone. This issue asks about the default parse, where the shape turns up regardless of language.

Measured: it goes both ways

ORCID credit names (orcid-researchers, 356,956 entries) of the form one word + initials, checked against the holder's own given and family boxes. "Surname first" means the word is in the family box and the first initial matches the given box:

The "unclear" rows are mostly surname first when read by eye (Wagner F, BISWAS P. K., Ben-Salem S.), but the boxes don't confirm it. Most of the given-name-first rows are South Indian or Bangladeshi names, where an initial stands for the father's name or the family name (Gomathi KG, Kavitha S, FATIMA M.J.). That is the same convention #864 kept for Nepali paired initials (Ram Bahadur B.K. → family 'B.K.'). So surname first is the majority, roughly 2:1, but the minority is real and concentrated in particular regions.

The OpenAlex sample in ourresearch/human-name-parser-gold-standard has 37 rows where its labels read this shape surname first.

It collides with #659

#659 asks whether an all-caps word in a mixed-case name marks the family (DUPONT Jean). Here the all-caps word is initials, not a surname, and the two have the same shape:

Zhi LI      → family 'LI'     (right: LI is the surname, as #659 would have it)
Hongwei LIU → family 'LIU'    (right)
Islam MS    → suffix 'MS'     (wrong: MS are the initials, read as the credential)
Armitage JO → family 'JO'     (wrong)

Two or three capital letters can't be told apart by shape: LI and MS are written the same way. A #659 rule would reinforce the current misreading of Armitage JO, and a rule for this issue would reverse Zhi LI. The two need to be decided together, or each needs to exclude the other's case explicitly. The dotted form (Armitage J.O.) has no #659 collision, but it does collide with #864.

Options

1. Leave the default alone and document the shape. Callers with citation-style input either insert the comma or declare name_order=FAMILY_FIRST, which already reads Armitage JO as family Armitage. Report the fork so a caller can see it. Dotted initials already report suffix-or-name, but JO reports nothing.
2. Read dotted initials after a single word as given initials by default (Gomes A.R.S., Armitage J. O.), but leave undotted capitals alone. The dotted ones can't be a Should an ALL-CAPS word in a mixed-case name mark the family name? DUPONT Jean gives given DUPONT #659 surname, so the remaining cost is Should Ram Bahadur B.K. keep B.K. as its family name? #864's shape and the South Indian minority above. That needs a measured decision.
3. An opt-in Policy switch for citation input, beside lone_name_word (Should a lone name word read as the given name under a family-first pack, so parser_for(locales.VI).parse("Anh") gives given Anh? #1111). The tie to Should Иванов И.И. read family 'Иванов' under locales.RU, where it now reads family 'И.И.'? #995 would then be a pack setting it.

Open

• Should a reading taken by position here report its alternative (PARTICLE_OR_GIVEN's analogue, or a new kind)? A reader would hesitate over Armitage JO, which meets the bar for adding a kind.
• How should a single initial be read (Delgado G., Armitage J)? Should Иванов И.И. read family 'Иванов' under locales.RU, where it now reads family 'И.И.'? #995 leaves it open for Cyrillic for the same reason: Ivan P. also has this shape.

Related: #995, #659, #864.

Found by scoring master against ourresearch/human-name-parser-gold-standard, an LLM-labelled sample of OpenAlex bylines.

🤖 Generated with Claude Code