Abstract:

SL-ReDu is a project that focuses on learning and assessing Greek Sign Language (GSL) as a second language, integrating state-of-the-art sign language (SL) recognition technology. It supports self-monitoring of productive learning and objective evaluation addressing;
a) teaching large student populations at a tertiary level,
b) assessing SL skills in an objective way
c) interacting digitally, without the presence of a human evaluator.

Achieving such goals requires an interdisciplinary effort, employing state-of-the-art techniques in language didactics, sign linguistics and handling of large amounts of SL data, while exploiting innovative methods in human-computer interaction, computer vision and machine learning [1].

The SL-ReDu educational material provides content to an innovative educational platform for enhancing and testing GSL competence of non-native (L2) users. It provides adequate linguistic input that consolidates new knowledge on all grammar levels of the target language. A total of 14 semantic units combined with major grammar features of GSL are catered for by the interface design, providing pedagogically and morphologically structured language learning activities, both passive and active.

Traditional SL learning and assessment require tutor individual involvement for each student or small group in order to guarantee visual attention and correction of articulation, which occurs in 3D space around the signer, with numerous upper-body active articulators through movement of the hands, head, shoulders and torso, mouth gestures, eye and eyebrow movements and gaze, all carrying grammatical information [2]. These parameters have all been taken into account in a signer-independent fashion, even accounting for natural variability among signers. Within the self-monitoring environment of the SL-ReDu platform students practicing with productive language skills can record their responses on their personal computer, upload them to the platform, where they are processed by an automatic recognition engine and receive feedback on their performance as many times as is necessary until they achieve a given learning goal. In this way the environment of GSL activities and assessment ensures credibility and consistency of test results efficiently and cost-effectively.
The SL-ReDu project -designed to support a large student population attending GSL courses (CEFR-L levels A0-A1) at Tertiary Education in Greece- includes two phases of evaluation and validation by end users, the first planned to take place in September 2021.

References:
[1] Efthimiou E, Fotinea S-E, Flouda C, Goulas T, Ametoglou G, Sapountzaki G, Papadimitriou K, Potamianos G. 2021: The SL-ReDu Environment for Self-monitoring and Objective Learner Assessment in Greek Sign Language, LNCS Proceedings of HCII-2021 (in print).
[2] T. Haug, “Web-based sign language assessment: Challenges and innovations”, in Proc. ALTE Int. Conf.: Learning and Assessment - Making the Connection, 2017.

Keywords:

Sign language teaching, second languge teaching, language teaching technologies, automatic recognition, online sign language teaching, L2 learning.