Sign Language Corpus

A structured collection of recorded signed-language performances — typically video, and increasingly motion-capture data — annotated by expert signers with time-stamped linguistic information such as individual signs, non-manual markers, eye gaze, grammatical boundaries, and discourse phenomena. Sign language corpora are foundational resources for sign-language linguistics, automatic sign recognition, sign-language machine translation, and the training of signing-avatar systems. Building them is slow and expensive because, in the absence of a widely used written form for most signed languages, there is no source of naturally occurring written signed text to harvest — performances must be recorded and hand-annotated. Well-known examples include the American Sign Language Linguistic Research Project (ASLLRP) corpora, the Corpus NGT (Dutch), BSL Corpus (British), and DGS-Korpus (German).