Introduction
SignTel is an Economic & Social Research Council (ESRC) funded project whose aim is to add avatars to computer based assessment tests that can 'sign' questions for deaf candidates.
Objectives
- To improve support for people with preference for British Sign Language during automated assessment of PC Passport tests.
- To learn new principles for designing and delivering item-banked tests that are suitable for candidates preferring BSL.
- To demonstrate the effectiveness of avatar signing to support TEL by rigorous evaluation of sign accuracy and acceptability to BSL signers.
- To develop a significantly enlarged lexicon and improved tools to enable efficient creation of avatar signing content.
Summary
For many Deaf people, especially those born with no hearing, British Sign Language (BSL) is their first and preferred language. As the spelling of English words is linked to the way they sound, Deaf people have less cues to help learning and as a result tend to have lower literacy rates than hearing people of the same age.
When taking tests where questions are written in English, Deaf people are at a disadvantage. It is agreed that they should be allowed to see the questions signed in BSL. Unfortunately, for tests on technical topics, such as computer literacy, there are not enough expert sign language interpreters to provide the necessary support.
Recent technology developed at the University of East Anglia enables a virtual human character, or avatar signer, to perform the gestures of BSL signs once they have been coded up in a specialised signing gesture notation.
A team of deaf people, based at the Royal National Institute for the Deaf, has developed the skills to author signs using the gesture notation. A sizeable lexicon of signs is being built up, enabling a wide range of signing phrases to be animated using the avatar.
The Scottish Qualifications Authority has developed a successful IT literacy qualification called PC Passport and wishes to make it possible for Deaf candidates to take the tests with support in BSL.
In this project, a number of PC Passport questions will be studied to decide how they should be translated to BSL. Any signs missing from the RNID lexicon will be created, and signing sequences will be produced for each of the questions. In addition, a glossary of computing terms used in PC Passport will be created, explaining each term in BSL.
A number of Deaf candidates will take the tests both with and without BSL support. The results will be analysed carefully to identify the most effective techniques for providing BSL support. As a result, it is expected that it will be possible to design future tests so that Deaf candidates are at no disadvantage when taking PC Passport or similar automated qualification.
Current status
The format used for the multiple choice questions in the SignTEL project is the IMS Question and Test Interoperability version 2.0 specification [QTI]. Commercial delivery systems do not provide access to source code to extend their capabilities to include the Java applet for the deaf signing avatar, or for the encoding and processing of the signing animations.
R2Q2 is an Open Source Java project, developed by the University of Southampton [SOT] that provides a rendering and response engine for QTI V2 questions through web services for integration into the JISC e-Framework [JISC]. It is still undergoing development but is sufficiently advanced to meet the needs of the SignTEL project and was extended and modified to give the required functionality.
The signing avatar was incorporated into the web service application using a Java applet, developed at UEA, enabling the test to be presented in a standard web browser. Buttons next to the question and answers on the page trigger the signing of the text by the avatar. The web service keeps a log of the student’s responses, the results being assembled at the end of the test and presented as a web page for the examiners.
The development of the web based delivery system for SignTEL was carried out in two phases, separated by a trial, with the second phase incorporating improvements from the results of the trial.
| Project Partners |
The Scottish Qualifications Authority
The University of East Anglia
The Royal National Institute for Deaf People
