Segmentation of Sino-origin words to enhance the representation of Korean and Japanese in S/UD-format treebanks
Résumé
In the Japanese and Korean S/UD treebanks, Chinese-origin words composed of two morphophonological units are not segmented, even when they are semantically transparent. We propose segmenting and annotating these words with dependency relations in order to achieve a more finegrained and unified description of both languages. As an example, we apply this analysis to the pre-annotated GSD corpora in SUD format, and we examine the benefits and limitations of a rule-based approach.
Domaines
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |