7. Developing an Interactive AI-Based Spoken Dialog System for Improving Oral Proficiency in Indonesian and Burmese

Advancing Critical CALL across Institutions and Borders - Reimagining Possibilities for Languages, Literacies, and Cultures - Emma Britton

Rahmi Aoyama [+-]
Northern Illinois University
Dr. Rahmi H. Aoyama has been an Indonesian language instructor at Northern Illinois University (NIU) since 2013. Her dissertation examined foreign language teachers’ technological pedagogical content knowledge (TPACK) in teaching listening and speaking skills in virtual worlds. She is a certified tester for Language Testing International. She developed oral proficiency guidelines for Indonesian based on ACTFL standards. Her current research uses SDS to improve speaking proficiency skills in Indonesian classes.
Maw Maw Tun [+-]
Northern Illinois University
Maw Maw Tun is working as a Teaching Assistant for Burmese at Northern Illinois University (NIU). She also has experience teaching Burmese at the Southeast Asian Studies Summer Institute (SEASSI) summer program from 2019 to 2023. She holds a master’s degree in teaching English as a second language (TESOL) from NIU. She is currently pursuing her PhD in Instructional Technology at NIU. Her research interest is integrating technology into language learning.
Reza Neiriznaghadehi [+-]
Iowa State University
Reza Neiriz is a PhD Candidate in the applied linguistics and technology program at Iowa State University. He specializes in language assessment through technology, focusing on spoken dialogue systems. He is evaluating a spoken dialogue system he has developed for assessing oral communication as part of his dissertation project, which the International Research Foundation for English Language Education funds. He has experience creating interfaces for computer-mediated language learning for two NSF-funded projects.

Description

With the development of artificial intelligence (AI) technology, interactive AI-based spoken dialog systems (SDS) have become viable tools to promote oral proficiency in language learning. The main goal of language learning is to communicate in real life, and promoting speaking skills plays an essential role in language development. Despite its importance, oral proficiency is considered to be challenging for language learners due to limited opportunities to practice the language effectively outside the classroom. Learners of less commonly taught languages (LCTL) often have difficulty finding opportunities to interact with experienced speakers outside class time. Therefore, the current study aimed to develop three SDS prototypes for Indonesian and Burmese learners as well as explore learners’ perspectives on SDS use. Both the Indonesian and Burmese SDS prototypes were originally planned to be developed using Google’s automatic speech recognition (ASR). While the Indonesian prototype works well with Google’s ASR, the accuracy of Google’s ASR for Burmese transcription is extremely low. Therefore, the current study used a Wizard of Oz system in developing a Burmese prototype. In order to understand learners’ experience of using SDS, the study utilized survey questionnaires with Indonesian and Burmese students (N=29) and eight semi-structured interviews. The findings of the pilot studies indicated that learners enjoy using SDS when practicing the target language without having any difficulties navigating SDS. More importantly, the results of the survey indicated that SDS could be an alternative learning tool for students to improve their language skills outside the classroom.

Notify A Colleague

Citation

Aoyama, Rahmi ; Tun, Maw Maw; Neiriznaghadehi, Reza . 7. Developing an Interactive AI-Based Spoken Dialog System for Improving Oral Proficiency in Indonesian and Burmese. Advancing Critical CALL across Institutions and Borders - Reimagining Possibilities for Languages, Literacies, and Cultures. Equinox eBooks Publishing, United Kingdom. p. Apr 2025. ISBN 9781800505674. https://www.equinoxpub.com/home/view-chapter/?id=46553. Date accessed: 21 Nov 2024 doi: 10.1558/equinox.46553. Apr 2025

Dublin Core Metadata