Project data of ShowTell – Finnish Sign Language MoCap corpus

Abstract

The MoCap data s a numerical matrix of over a billion characters that was collected during 2017 in the MoCap laboratory of the Department of Music, Art and Culture Studies, JYU. The data comprises 33 narratives (total duration 21 minutes) recorded from six FinSL signers with an eight-camera optical Qualisys Oqus motion capture system connected to one Full HD 30 fps video camera. The narratives are based on short textless cartoon strips which the signers first memorized and then re-told to an addressee standing in front of them. Each signer produced at least five narratives, and each narrative included a varying amount of regular telling as well as subtle, reduced and overt constructed action. Twenty-five small reflect-ing ball-shaped markers, whose three-dimensional locations the MoCap system tracked with a speed of 120 Hz, were attached to the main body parts of all the signers. The specific locations of the mark-ers relevant to the proposed study are the proximal joints of the index fingers, the ulnar and radial side of the wrists, as well as both elbows.