Fiche du document
2012
- handle: 10670/1.qviq6k
- halshs: halshs-01509071
info:eu-repo/semantics/OpenAccess
Mots-clés
RhapsodieSujets proches
Units Measurement, Units of Structures Edifices HallsCiter ce document
Kim Gerdes et al., « Intonosyntactic Data Structures : The Rhapsodie Treebank of Spoken French », HAL-SHS : linguistique, ID : 10670/1.qviq6k
Métriques
Partage / Export
Résumé
In this work, we present the data structures that were developed for the Rhapsodie project, an intonosyntactic annotation project of spoken French. Phoneticians and syntacticians work on different base units: a time aligned sound file for the former, and a partially ordered list of tokens for the latter. The alignment between the sound-file and the tokens is partial and non-trivial. We propose to encode this data with a small set of interconnected structures: lists, constituent trees, and directed acyclic graphs (DAGs). Our query language remains simple, similar to the Annis Query language, as the precedence and including relations are handled in accordance with the requested objects and their type of alignment: The order between prosodic units is time-based, whereas the order between syntactic units is lex-eme-based.