home edit page issue tracker
This page pertains to UD version 2.

French Rhapsodie

Back to index

Overview

   
type only spoken
available since 2.2
link https://github.com/UniversalDependencies/UD_French-Rhapsodie
genre spoken
contributors Gerdes, Kim; Kahane, Sylvain; Nakhlé, Mariam; Yan, Chunxiao; Etienne, Aline; Courtin, Marine
sentences 3209
tokens 43691

Issue draft: UD_French-Rhapsodie

Modality identification

Is spoken part clearly identifiable? N/A - spoken data only

Metadata review

doc (and paragraphs) metadata

(none found) - no document_id exists at all. But sent_id already encodes it: e.g. Rhap_D0001-1 is document Rhap_D0001, sentence 1. 57 distinct documents across 3209 sentences, matching exactly the 57 distinct sound_url values (present on every single sentence, unlike ParisStories). Several other fields are also constant within each document and should move to document level once document_id exists: genre, subgenre, type, task, subject, channel, modalities (verified: none of these vary within a document). speaker_id/speaker_age correctly stay sentence-level, since they do vary within documents (multi-speaker dialogues).

Field Advice
— derive # document_id from the sent_id prefix (everything before the trailing -<number>)
sound_url move to document level, set once per document_id
genre, subgenre, type, task, subject, channel, modalities move to document level (constant per document)

transcription and annotation levels available

Field Advice
macrosyntax change to text_macrosyntax
prosodic_annotation corpus-specific (only on a subset of sentences) - please confirm what this represents

speaker metadata

Field Advice
speaker_id OK
speaker corpus-specific turn-position label (L1, L2, …), distinct from and redundant with speaker_id (e.g. §LF30) - keep, speaker_id is already standard

token-level metadata (MISC)

Field Advice
AlignBegin rename to WordAlignmentBegin
AlignEnd rename to WordAlignmentEnd