Year: 2026
Type: article
Abstract: Chilean Sign Language (LSCh), though legally recognized in Chile, lacks standardized computational resources necessary for technological and linguistic development. This work addresses that gap by proposing a video recording protocol for the construction of an LSCh corpus suitable for computational applications. Theoretical foundations cover the linguistic features of LSCh and the nature of sign language corpora. A typology is discussed distinguishing lexical, linguistic, and computational corpora, emphasizing the specific requirements of machine learning applications. A structured review of ten representative sign language corpora with computational objectives informed protocol design through their ethical procedures, participant profiles, elicitation modes, recording conditions, and camera settings. Based on the review, a multi-camera setup was designed using Full HD cameras at 120 frames per second (FPS) from the front, left side, and top angles. The protocol includes Deaf community participation, ethical safeguards, and elicitation materials in LSCh. The studio was assembled and tested in Valparaíso; corpus collection with 30 participants is planned in Valparaíso and Santiago. Three local deaf researchers and expert LSCh signers participated in pilot sessions, and their recordings were synchronized and preprocessed. Signers assessed intelligibility. Four spatial image metrics compared one approximately 60-second front-view clip at 120 FPS with its resampled and re-encoded 25-FPS version. Paired block-bootstrap tests found higher source-clip scores for all four metrics (Holm-adjusted p < 0.001 ). This protocol provides a reproducible and scalable method for building LSCh corpora for computational use. Future work includes scaling the corpus, applying annotation, and deploying the data in neural models for translation and recognition.
Source: Heliyon
Institution: University of Valparaíso
Language: English
Subfield: Developmental and Educational Psychology
Field: Psychology
Domain: Social Sciences
Keywords: Chilean Sign Language, sign language corpus
Open Access status: gold
Estimated APC paid: $2,270
OpenAlex ID: W7225173963
