CORILSE: a Spanish Sign Language Repository for Linguistic Analysis

CORILSE: a Spanish Sign Language Repository for Linguistic Analysis Carmen Cabeza*, José M. Garcia-Miguel*, Carmen García-Mateo**, José Luis Alba Castro** Universidade de Vigo *Facultade de Filoloxía e Tradución, Universidade de Vigo, 36310 Vigo (Spain) **AtlantTIC Research Center, Escola de Enxeñaría de Telecomunicación, Universidade de Vigo, 36310 Vigo (Spain) E-mail: cabeza@uvigo.es, gallego@uvigo.es, carmen.garcia@uvigo.es, jalba@gts.uvigo.es Abstract CORILSE is a computerized corpus of Spanish Sign Language (Lengua de Signos Española, LSE). It consists of a set of recordings from different discourse genres by Galician signers living in the city of Vigo. In this paper we describe its annotation system, developed on the basis of pre-existing ones (mostly the model of Auslan corpus). This includes primary annotation of id-glosses for manual signs, annotation of non-manual component, and secondary annotation of grammatical categories and relations, because this corpus is been built for grammatical analysis, in particular argument structures in LSE. Up until this moment the annotation has been basically made by hand, which is a slow and time-consuming task. The need to facilitate this process leads us to engage in the development of automatic or semi-automatic tools for manual and facial recognition. Finally, we also present the web repository that will make the corpus available to different types of users, and will allow its exploitation for research purposes and other applications (e.g. teaching of LSE or design of tasks for signed language assessment). Keywords: Spanish sign language, corpus development, linguistic annotation 1. Introduction Spanish Sign Language (Lengua de Signos Española, LSE) can be considered an under-resourced language, which poses a major drawback for conducting linguistic research and for developing high-quality language technology (López-Ludeña, 2013). There are a number of limiting factors, among them standardization and difficulties in manual annotation. Therefore, the creation of modern language corpora builds on the development of proper tools for automatic sign language annotation. In this work we discuss some strategies for the annotation of video clips which contribute to make our corpus usable for linguistic analysis, as well as for other applications, and describe the automatic tools we are developing to detect the regions of interest in the video, particularly eyes, lips and hands. All this has been made within the framework of the CORILSE (“Corpus informatizado para LSE”, LSE computerized corpus) project that aims at building a high- quality software repository of LSE. Once completed, the corpus will be made publicly available for the benefit of researchers. In this article, we present the two major components of the CORILSE project: (i) the video recordings and annotation files tailored to the particularities of LSE, and (ii) the software repository. 2. About Sign Language Corpora The current research on sign language corpus is based on the widespread use of annotation tools as ELAN (Crasborn & Sloetjes, 2008), that enable to simultaneously visualize and annotate signing discourse. Some research teams have built sign language corpora from a set of video files annotated on ELAN and a lexical database. These previous works are our starting point: Auslan corpus (Johnston, 2010, 2013), NGT corpus …

Content truncated to 500 words