Document Type

Theses, Masters

Rights

This item is available under a Creative Commons License for non-commercial use only

Disciplines

Computer Sciences, *human – machine relations, *therapy for learning, speech,, *hearing, visual and other physical, Social sciences, Interdisciplinary, Specific languages, Linguistics

Publication Details

A dissertation submitted in fulfillment of the requirements for the award of Master of Science (M.Sc.) in Computing

Abstract

This thesis explores the role that underlying emotional facial expressions might have in regards to understandability in sign language avatars. Focusing specifically on Irish Sign Language (ISL), we examine the Deaf community’s requirement for a visual-gestural language as well as some linguistic attributes of ISL which we consider fundamental to this research. Unlike spoken language, visual-gestural languages such as ISL have no standard written representation. Given this, we compare current methods of written representation for signed languages as we consider: which, if any, is the most suitable transcription method for the medical receptionist dialogue corpus. A growing body of work is emerging from the field of sign language avatar synthesis. These works are now at a point where they can benefit greatly from introducing methods currently used in the field of humanoid animation and, more specifically, the application of morphs to represent facial expression. The hypothesis underpinning this research is: augmenting an existing avatar (eSIGN) with various combinations of the 7 widely accepted universal emotions identified by Ekman (1999) to deliver underlying facial expressions, will make that avatar more human-like. This research accepts as true that this is a factor in improving usability and understandability for ISL users. Using human evaluation methods (Huenerfauth, et al., 2008) the research compares an augmented set of avatar utterances against a baseline set with regards to 2 key areas: comprehension and naturalness of facial configuration. We outline our approach to the evaluation including our choice of ISL participants, interview environment, and evaluation methodology. Remarkably, the results of this manual evaluation show that there was very little difference between the comprehension scores of the baseline avatars and those augmented withEFEs. However, after comparing the comprehension results for the synthetic human avatar “Anna” against the caricature type avatar “Luna”, the synthetic human avatar Anna was the clear winner. The qualitative feedback allowed us an insight into why comprehension scores were not higher in each avatar and we feel that this feedback will be invaluable to the research community in the future development of sign language avatars. Other questions asked in the evaluation focused on sign language avatar technology in a more general manner. Significantly, participant feedback in regard to these questions indicates a rise in the level of literacy amongst Deaf adults as a result of mobile technology.


Share

COinS