4-Couv, a Backcover Treebank - Archive ouverte HAL
Communication Dans Un Congrès Année : 2015

4-Couv, a Backcover Treebank

Résumé

We introduce 4-Couv, a treebanking project aiming at develop a large multipurpose treebank for French language. The main characteristic of this project is to provide adequate material for both linguistic and psycholinguistic research. The treebank is made of short and self-contained texts selected from a corpus of backcovers coming from different editors and different genres. Such material makes possible classical linguistic research (especially in syntax and discourse), but offers also new perspectives in experimental linguistics: being short texts and semantically coherent, they perfectly fit with the requirements of eye-tracking or electro-encephalographic recordings. At this stage, 4-Couv contains 3,500 trees automatically tagged and parsed, and manually corrected. Its format is compatible with other standard French treebanks. We present in this paper, the treebank, its annotation and the different treebanking tools that have been developed for the different stages of its elaboration: text selection, tagging, parsing and manual correction.
Fichier principal
Vignette du fichier
TLT2015-short-Final.join-u8_PB.pdf (249.32 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01498941 , version 1 (20-04-2017)

Identifiants

  • HAL Id : hal-01498941 , version 1

Citer

Philippe Blache, Grégoire Montcheuil, Stéphane Rauzy, Marie-Laure Guénot. 4-Couv, a Backcover Treebank. Treebanks and Linguistic Theories 14, Dec 2015, Warsaw, Poland. pp.249-257. ⟨hal-01498941⟩
139 Consultations
97 Téléchargements

Partager

More