globalwordnet / globalwordnet/semcor
Opening scare-quote (single quote) attached to the wrong side, e.g. more' pro' for 'pro'
- Dominant language
- Python
- Stars
- 0
- Forks
- 0
- Avg merge
- 2d 22h
- Merged PRs (30d)
- 24
Description
When a single quote marks emphasized/scare-quoted or dialect-elided speech (\`more 'pro' letters than 'con'\`, \`Ah says 'Uh coahse...\`), the *opening* quote is frequently flush against the *preceding* word instead of the word it actually introduces (\`more' pro'\` instead of \`more 'pro'\`) -- the same shape as #8's double-quote gap bug, but for single quotes, and specifically the opening side (the closing side, flush against the word it closes, is already correct).
Confirmed against \`src/semcor/brown-nolines.txt\` via context-window matching: 45 instances found among 424 candidates (standalone \`'\` tokens tagged \`POS\` -- Brown's own tagger's closest approximation for this glyph, since \`''\` closing-quote and \`POS\` possessive are the two tags a bare apostrophe can get -- that are flush before and gapped after). The majority of \`POS\`-tagged candidates are correctly-placed plural possessives (\`boys' toys\`) with no scare-quote at all, filtered out by requiring the reference to confirm a literal quote character at the matched position. ~75 more are left unresolved (no unique context match) for a follow-up look.
Since the gap is always exactly one space, the fix is a same-length swap of the quote and the space immediately after it -- no other token's span changes.
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with src/semcor/brown-nolines.txt and reproduce the context-window matching described in the issue, checking the 45 confirmed cases against the reference. Done means the opening quote and following space are swapped for those cases, while token spans and correctly placed possessives remain unchanged.
Written by the indexing model from the issue text.
Assessment
- Domain
- data
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Active
- Clarity
- Clearly specified
- Newbie friendliness
- 76/100