Summary

A Universal Dependencies corpus for spoken French.

Introduction

The corpus was converted automatically from the Rhapsodie treebank with manual corrections. The treebank in maintained in the repository SUD_French-Rhapsodie in the SUD annotation schema.

The SUD version is also available with prosodic annotation (see SUD README.md).

Structure

• fr_rhapsodie-ud-train.conllu 1,288 sentences and 19,144 tokens
• fr_rhapsodie-ud-dev.conllu 1,081 sentences 12,907 tokens
• fr_rhapsodie-ud-test.conllu 840 sentences 12,191 tokens
• total 3,209 sentences 44,242 tokens

Changelog

• 2025-11-15 v2.17
  • More metadata
  • Sound alignement
• 2024-05-15 v2.14
  • Fix a few inconsistent annotation of idioms
  • See SUD commit logs for more details
• 2021-11-15 v2.9
  • Repository renamed from UD_French-Spoken to UD_French-Rhapsodie.
• 2020-11-15 v2.7
  • Morphology added
• 2018-04-15 v2.2
  • Initial release

=== Machine-readable metadata (DO NOT REMOVE!) ================================ Data available since: UD v2.2 License: CC BY-SA 4.0 Includes text: yes Parallel: no Genre: spoken Lemmas: converted from manual UPOS: converted with corrections XPOS: not available Features: not available Relations: converted with corrections Contributors: Gerdes, Kim; Kahane, Sylvain; Nakhlé, Mariam; Yan, Chunxiao; Etienne, Aline; Courtin, Marine Contributing: here Contact: [email protected]