Towards Continuous Recognition of Illustrative and Spatial Structures in Sign Language

In recent research, attention has been drawn to recognizing lexical signs in continuous Sign Language (SL) corpora, often artificial. Because SLs are singularly structured by the use of space and iconicity, this focus does not allow for an easy transition towards SL understanding and translation. In this paper, we discuss the necessity and realizability of recognizing higher-level linguistic structures in SL videos, like classifier constructions, using natural corpora such as Dicta-Sign–LSF–v2.