An analytic model of handshape similarity in american sign language

Handshape, or the posture of the hand, is one of the distinctive features of American Sign Language (ASL). Handshape is one of the most recognizable aspects of individual signs. However, there is little consensus about the number of distinct handshapes in ASL and there is even less consensus about similarity among handshapes. Work on handshape similarity had an initial blossoming in the 1970's, but withered without producing consistent results. In some of the previous studies, the methodologies were too costly and time-consuming for extensive testing. Other studies did not focus exclusively on handshapes, or had an extremely small test population. The goal of this study is to create and evaluate an analytic model of handshape similarity, which is based on human perception of true three-dimensional data. This will be the first effort that considers the relation of the human perception of handshape and the physical positioning of the hand joints in space. The benefit of an analytic model is that, once established, it can predict the similarity of additional handshapes without needing to conduct additional costly perceptual tests. An analytic model of handshape similarity that is consistent with human perception has theoretical and practical ramifications. It will contribute toward the development of a true bilingual ASL-English dictionary that allows users to find the meaning of a sign based on the sign's appearance. In addition, the methodologies laid out in this approach could be applied to other sign languages as well and benefit Deaf people throughout the world.