Mary-Jane + O'Connor: What Normalization Removes
A case study of punctuation normalization, compound names, and choosing the right source units before blending.
Reviewed by Ship Name Lab Research Desk · 2026-07-12
The review question
What information is lost when a generator removes hyphens and apostrophes before constructing candidates?
Inputs
Mary-Jane + O'Connor
Top candidate
Marnor
Structural score
97
Engine result
Pass
What the result shows
The engine normalizes punctuation so it can apply the same deterministic cuts to every input. That produces Marnor as the current leader, using material from both normalized strings.
Normalization is mechanically consistent but not culturally neutral. Mary-Jane may be treated as one compound given name, while O'Connor is a surname whose apostrophe carries familiar written structure. Removing both marks makes computation simpler and interpretation poorer.
The full strings may also be the wrong source units. Mary + Connor, Jane + Connor, or another form people actually use can produce a more faithful label than blending legal or formal spellings.
A displayed candidate can restore punctuation after generation, but that would be an editorial presentation choice rather than the exact string the engine scored. The distinction should remain visible instead of silently changing the result.
Human checks before use
- Confirm which parts of each compound name the represented people use publicly.
- Compare normalized and punctuation-preserving display forms before publishing the result.
- Treat Marnor as a software observation, not a claim about correct handling of either name.
Editorial verdict
Marnor shows that deterministic normalization works technically while still requiring a human decision about identity and source selection.