Article: Development of an AI-assisted spoken language assessment system for Japanese language teaching Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.10 pp.90 - 111 Abstract: This project develops an AI-assisted spoken language assessment system to enhance Japanese language instruction. By integrating voice recognition, pronunciation analysis, and fluency scoring, the system provides reliable evaluations comparable to those of professional ratters. Utilising multilingual datasets and data augmentation, it reduces language learning anxiety and improves recognition accuracy. Since language anxiety negatively affects second-language acquisition - particularly in oral proficiency - this study aims to support Japanese learners and increase their speaking confidence. While prior research has demonstrated the benefits of ICT tools and flipped classrooms for pronunciation self-monitoring, limited studies have applied AI for comprehensive oral evaluation. The proposed four-step methodology includes data collection, feature extraction, model development, and validation. Informed by sentiment analysis and multilingual corpora, the system achieved an accuracy of 97.3% using a two-stream LSTM model, while translation-based augmentation improved Japanese sentiment analysis accuracy by 6.58%. Inderscience Publishers - linking academia, business and industry through research

Title: Development of an AI-assisted spoken language assessment system for Japanese language teaching

Authors: Li Zhang

Addresses: School of Foreign Languages, Nanyang Institute of Technology, Nanyang, Henan, 473000, China

Abstract: This project develops an AI-assisted spoken language assessment system to enhance Japanese language instruction. By integrating voice recognition, pronunciation analysis, and fluency scoring, the system provides reliable evaluations comparable to those of professional ratters. Utilising multilingual datasets and data augmentation, it reduces language learning anxiety and improves recognition accuracy. Since language anxiety negatively affects second-language acquisition - particularly in oral proficiency - this study aims to support Japanese learners and increase their speaking confidence. While prior research has demonstrated the benefits of ICT tools and flipped classrooms for pronunciation self-monitoring, limited studies have applied AI for comprehensive oral evaluation. The proposed four-step methodology includes data collection, feature extraction, model development, and validation. Informed by sentiment analysis and multilingual corpora, the system achieved an accuracy of 97.3% using a two-stream LSTM model, while translation-based augmentation improved Japanese sentiment analysis accuracy by 6.58%.

Keywords: AI-assisted assessment; spoken language evaluation; Japanese language teaching; speech recognition; natural language processing; NLP; automated scoring; pronunciation analysis.

DOI: 10.1504/IJICT.2026.151686

International Journal of Information and Communication Technology, 2026 Vol.27 No.10, pp.90 - 111

Received: 08 Oct 2025
Accepted: 02 Dec 2025
Published online: 13 Feb 2026 *

Title: Development of an AI-assisted spoken language assessment system for Japanese language teaching

Keep up-to-date