Good Practices of Didactical Methods Database

Title: Computer-assisted teaching of Sign Language using Computer Vision and Machine Learning – CAT-SL
Good Practice Mission: Deaf children, initially educated in SL, usually stop their studies at the elementary school level because there are very few SL interpreters available, which makes their inclusion within the typical educational system often impossible. The CAT-SL aims to complement existing curricula standard SL language acquisition practice sessions by introducing automated exercises that students perform and the system automatically evaluates.
Specific goals of the GP:

Related curricula in the SL courses conforming to EU-wide standards in primary and HE.

A community should adopt and support the related technology and motivate further developments using open standards.
Year: 2021
Duration of the implementation:
Target group: Anyone over 12 years old (it is made to be used with underage students, university students and teachers).
Summary:

The CAT-SL project aims to develop and implement an innovative and affordable system for interactive sign language (SL) teaching. The project focuses on two key areas:

  • Technology Development: Researchers built the CAT-SL infrastructure based on open standards, allowing its use in four Higher Education Institutes and one primary school
  • Curriculum and Training: The project developed curricula and teaching guides specifically designed for the CAT-SL system. These resources are available for at least two multilingual courses in Greece, Cyprus, Portugal, and the Netherlands.

The process includes:

The course has been designed for use as part of a university curriculum and will be part of the ECTS system. It includes 36 units and 13 lectures that cover topics ranging from foundational elements to more complex aspects of sign language. Lectures 6 and 13 are rehearsal lectures based on the material already presented.

Lecture 1: Sign alphabet, handshapes, basic communication vocabulary, family vocabulary.

Lecture 2: Number vocabulary, time/hour vocabulary, colour vocabulary.

Lecture 3: Calendar/days vocabulary, time-seasons-months vocabulary, time-time line vocabulary.

Lecture 4: Clothes vocabulary, food and drinks 1 vocabulary, personal pronouns vocabulary.

Lecture 5: Feelings vocabulary, home vocabulary, possessive pronouns vocabulary. Lecture 6: Rehearsal unit.

Lecture 7: Directions vocabulary, time-24hours vocabulary, interrogative pronouns vocabulary.

Lecture 8: Traffic vocabulary, professions/career 1 vocabulary, negation vocabulary. Lecture 9: Weather season vocabulary, Time review vocabulary,

Lecture 10: Food and drinks 2 vocabulary, professions/career 2 vocabulary, verbs vocabulary

Lecture 11: Animals vocabulary, geography-landscape vocabulary. Lecture 12: School/study vocabulary.

Lecture 13: Rehearsal unit.

The requirements for the transferability of the practice are:

Given the course's dual design to serve both adults and younger learners, the target proficiency levels are somewhat adapted to the learners' age and educational context. For adult learners and higher education students, the A1 level encompasses a more comprehensive curriculum that dives into theoretical aspects of sign language, its grammar, and an introduction to Deaf culture.

For pupils in primary or secondary education, the A1 level material is presented in a more digestible and interactive manner, focusing on vocabulary and basic communicative skills.
Space requisites:

The space should definitely have access to a computer with adequate specifications (e.g., access to the Internet, camera, and updated drivers).
Obstacles revealed?:

Inadequacies of publicly available sign language datasets restrict the strength and generalizability of systems trained on them

Understanding portrayal necessitates experience to Deaf culture and languages, which is typically lacking in the populations driving advancement in computer vision

Simple applications of Machine Translation (MT) and Natural Language Processing (NLP) methodologies either fail to capture some parts of sign languages or simply do not function

The development of avatars faces a number of technical challenges in creating avatars that are acceptable to Deaf users.
Methodology used: Differenciated instruction
Lecture-based learning
Technology-based learning
Individual learning
Kinaesthetic learning
Learning by doing
Physical adaptations and assistive technologies
Resources needed:

Similarly to the special characteristics/functions needed, the required resources consist of access to a computer with adequate specifications.
Link: http://cat-sl.eu/
Tags: