Aisla is a BMBF (German Ministry of Education) funded project aiming at building an intelligent agent for assisting spoken English training. Its design is informed by the latest research on task-based language teaching, AI conversation systems, and natural language processing.
An intelligent language tutor and practice partner.
The Aisla system’s design conforms with the latest findings from Task-Based Language Teaching (TBLT) and Second Language Acquisition (SLA) research, ensuring the effectiveness of the system to promote spoken English abilities.
Seamless integration with school curriculum, ensuring that the tasks align closely with the vocabulary and grammar constructs currently learned in the classrooms. This allows students to reinforce classroom learning and apply new language skills in real-world situations.
Real-life scenarios (e.g., restaurant, airport, meeting, university lecture, etc.) are simulated for learners to practice real-life communication tasks (e.g., booking a table, checking in a flight, etc.) with a dynamic agent playing the role of different conversation partners. Demonstration of the tasks and the key language to be used are provided as guidance.
Aisla does not only support pure verbal interaction with text and voice, but also provides embodied realistic virtual human agents powered by the latest deep learning technologies to create a face-to-face communication experience. Non-verbal cues such as facial expressions and body gestures have been found to facilitate communication and promote learning.
Automatic feedback, powered by advanced language models, is a crucial asset for language learners. While practicing real-life scenarios, students receive instant, tailored feedback on their language usage and grammar errors. Automatic feedback thus may accelerate learning, helping students gain confidence and proficiency in their target language.
Competent interlocutors are native speakers or experienced language teachers that offer authentic opportunities for learners to practice using the target language. Obviously, it is unrealistic to offer every English learner a competent interlocutor. The Aisla system potentially offers a cost-effective solution to the needs for each individual learner.
https://kibi.group
Hector-Institut für Empirische Bildungsforschung
Eberhard Karls Universität Tübingen
Walter-Simon-Strasse 12 · 72072 Tübingen,Germany