Osmelak, D., Dutta Chowdhury, K., Sentsova, U., España-Bonet, C., & van Genabith, J. (2026). PETra: A multilingual corpus of pragmatic explicitation in translation. Proceedings of LREC 2026, 8756–8766.
Read in full in NOTE-1344.
Why this is here
A5 put it first among the readings: a corpus of what translation adds, which is the target side of CLAIM-257’s change of cover, and a source of items for CASE-185. Its categories annotate additions; they are not a validated measure of footing or act.
The reading bears this out with two qualifications for CASE-185: the labels would need re-annotation to meet its Stage 0, and the corpus gives one direct route per item, from English only. About half its labels mark replacements, not additions (NOTE-1344).
Key takeaways
- The items are additions to background knowledge (“Tucson” → “Tucson (Arizona)”), not changes of act, stance or footing.
- No agreement statistic, a single annotator implied, and most pairs found by a classifier: the labels are not a validated measure (NOTE-1344).
- One English → X route per item; no relay can be read off it.