Blurb:

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.