Putting the Horses before the Cart: Identifying Multiword Expressions before Translation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Putting the Horses before the Cart: Identifying Multiword Expressions before Translation

Résumé

Translating multiword expressions (MWEs) is notoriously difficult. Part of the challenge stems from the analysis of non-compositional expressions in source texts, preventing literal translation. Therefore, before translating them, it is crucial to locate MWEs in the source text. We would be putting the cart before the horses if we tried to translate MWEs before ensuring that they are correctly identified in the source text. This paper discusses the current state of affairs in automatic MWE identification, covering rule-based methods and sequence taggers. While MWE identification is not a solved problem, significant advances have been made in the recent years. Hence, we can hope that MWE identification can be integrated into MT in the near future, thus avoiding clumsy translations that have often been mocked and used to motivate the urgent need for better MWE processing.
Fichier principal
Vignette du fichier
p02.pdf (468.51 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01795985 , version 1 (18-05-2018)

Identifiants

Citer

Carlos Ramisch. Putting the Horses before the Cart: Identifying Multiword Expressions before Translation. Computational and Corpus-Based Phraseology - Second International Conference, Europhras 2017, London, UK, November 13-14, 2017, Proceedings, 2017, London, United Kingdom. pp.69 - 84, ⟨10.1007/978-3-319-69805-2_6⟩. ⟨hal-01795985⟩
93 Consultations
223 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More