Japanese JSL translation and searching display conditions for expressing easy-to-understand sign animation

This paper described a bi-directional translating system between Japanese and Japanese Sign Language and two experiments conducted to clarify necessary conditions for displaying easy-to-read animation of a person model speaking sign language synthesized by the system. In Experiment 1 hearing-impaired and hearing subjects judged identity of a pair of animation. Resolution and frame rate had an main effect against correct answer rate. While the hearing had tendency to watch the whole body of the model, the hearing-impaired focused their attention chiefly on handshape and movement. In Experiment 2 the hearing-impaired read signs on animation. For correct reading above 90 % sign animation needed to be displayed with more than 10 x 8 cm, 113 × 90 pixels and 8 f/sec. This condition was much relaxed compared with that in Experiment 1.