Collecting an American sign language corpus through the participation of native signers

Abstract

Animations of American Sign Language (ASL) can make more information, websites, and services accessible for the significant number of deaf people in the United States with lower levels of written language literacy - ultimately leading to fuller social inclusion for these users. We are collecting and analyzing an ASL motion-capture corpus of multi-sentential discourse to seek computational models of various aspects of ASL linguistics to enable us to produce more accurate and understandable ASL animations. In this paper, we will describe our motion-capture studio configuration, our data collection procedure, and the linguistic annotations being added by our research team of native ASL signers. This paper will identify the most effective prompts we have developed for collecting non-scripted ASL passages in which signers use particular linguistic constructions that we wish to study. This paper also describes the educational outreach and social inclusion aspects of our project - the participation of many deaf participants, researchers, and students. © 2011 Springer-Verlag.