Metadata harmonisation: align spoken-data fields with UniDive naming conventions
Back to Bororo BDT · Back to index
Repo: https://github.com/UniversalDependencies/UD_Bororo-BDT
Cross-posting from the UniDive WG1 T1.5 (spoken language guidelines) metadata harmonisation review. We compared UD_Bororo-BDT’s current CoNLL-U metadata against the proposed naming conventions (see also the full treebank status table). This is a suggestion for maintainers to review - feel free to push back on anything that doesn’t fit the corpus.
1. Is the spoken portion identifiable?
This treebank mixes spoken and written material but its .conllu files don’t explicitly mark which sentences are spoken. We looked for a pattern in the data (a weak signal):
Finding: Possibly identifiable via a naming pattern in newdoc id.
Evidence: Only newdoc id, sent_id, and text exist - no genre, source, or modality field at all. Of the 43 newdoc id values, ~17 match biblical book names (samuel_1_2, reis_1_2 = Kings, esdras_2 = Ezra, oseias_2 = Hosea, genesis_2, levitico_2, daniel_2, novo_testamento_ochoa, etc.) - almost certainly translated scripture, i.e. written; the remaining ~26 look like oral narratives/rituals (oieigo_*, coqueiro01-coqueiro09, historia_mitica_bor, rituais_bororo, bokodori_ecerae, juko_ro) - plausibly spoken fieldwork material, matching the README’s description of “mythological narratives, fieldwork material (elicited and spontaneous discourse)”.
Suggestion: Add # modality = written to the ~17 biblical documents and # modality = spoken to the narrative/ritual ones - please confirm, since this was inferred from document naming rather than an explicit tag.
2. Document-level (naming conventions)
We found a data-quality issue while checking newdoc id: 4 of the 43 values are not real identifiers - they’re leaked local Windows file paths (e.g. G:\Mi unidad\hiwi\tasks_materials\...\ipare_ereru_(...)-udpipe.txt, with corrupted/mojibake characters in 2 of them). Two of these duplicate a document that also exists under a clean id (ipare_ereru, oieigo_de_danca_2) - could you confirm whether the same document is present twice under two different ids, and clean up the leaked file paths?
This issue was prepared as part of the UniDive WG1 T1.5 spoken language guidelines effort. Happy to help implement these changes ourselves if that’s easier than doing it on your end - just let us know.