Phonological Proximity in Costa Rican Sign Language

• DOI: 10.3390/electronics9081302
• Publication Date: 2020-08-13T13:23:44Z
• Authors: Luis Naranjo-Zeledón; Mario Chacón-Rivas; Jesús Peral; Antonio Ferrández
• Keywords: Acronym; Similarity (geometry); Basis (linear algebra)

The study of phonological proximity makes it possible to establish a basis for future decision-making in the treatment sign languages. Knowing how close set signs are allows interested party decide more easily its by clustering, as well teaching language third parties based on similarities. In addition, lays foundation strengthening disambiguation modules automatic recognition systems. To best our knowledge, this is first kind Costa Rican Sign Language (LESCO, Spanish acronym), and forms one already operational system speech editing called International Platform Edition (PIELS). A database 2665 signs, grouped into eight contexts, used, comparison similarity measures made, using standard statistical formulas measure their degree correlation. This corpus will be especially useful machine learning approaches. work, we have proposed an analysis different between order find out them. After analyzing results obtained, can conclude that LESCO with high levels proximity, particularly orientation location components, but they noticeably lower form component. We also concluded outstanding contribution research systems take prototypes contexts or domains map clusters similarity. As mentioned, obtained multiple applications such area Natural Processing tasks.