JBT Journal
Advanced

Raqamli Transformatsiya va Sun’iy Intellekt

AN EXPLAINABLE NEURO-SYMBOLIC SPEECH MATRIX MODEL FOR PERSONALIZED FOREIGN LANGUAGE PRONUNCIATION FEEDBACK

Jura Kuvandikov1, Rohmonqulov Muhammadyusuf Egamberdi o‘g‘li1

1 Jizzakh branch of the National University of Uzbekistan named after Mirzo Ulugbek

Zenodo DOI: 10.5281/zenodo.22277193

Keywords

computer-assisted pronunciation training speech matrix knowledg automatic pronunciation assessment neuro-symbolic AI

Abstract

Automatic pronunciation support increasingly combines speech recognition, pronunciation assessment, and computer-assisted language learning, yet many systems still return only a transcription or global score. This article proposes the Neuro-Symbolic Speech Matrix Feedback model (NS-SMF), an explainable architecture that converts learner speech into diagnostic and pedagogically actionable feedback. Learner speech is represented through acoustic, phonetic-prosodic, and pedagogical error matrices. A neural encoder extracts acoustic-phonetic evidence, task-specific heads estimate segmental and suprasegmental deviations, reference alignment localizes errors, and a knowledge base or knowledge graph maps detected deviations to linguistically grounded explanations and corrective exercises. The model formalizes posterior-based scoring, dynamic alignment, integrated deviation, explanation confidence, feedback validity, pedagogical priority, learner-profile update, and multi-task training. A pilot protocol for Uzbek-speaking learners of English is outlined using technical, expert-agreement, explainability, and learning-improvement measures. The main contribution is a formal framework that moves pronunciation assessment from black-box scoring toward traceable, rule-grounded, and exercise-oriented feedback.

Views: 122 Published: 2026-08-28 Pages: 153-158 Volume: 4 Issue: 4

References

  1. N.Raximov, J.Kuvandikov, D.Khasanov, “The importance of loss function in artificial intelligence”, International Conference on Information Science and Communications Technologies (ICISCT 20222), DOI: 10.1109/ICISCT55600.2022.10146883
  2. Hogan, A., et al. (2021). Knowledge graphs. ACM Computing Surveys, 54(4), Article 71. https://doi.org/10.1145/3447772
  3. Baevski, A., Zhou, Y., Mohamed, A., & Auli, M. (2020). wav2vec 2.0: A framework for self-supervised learning of speech representations. Advances in Neural Information Processing Systems, 33.
  4. Amrate, M., & Tsai, P.-H. (2024). Computer-assisted pronunciation training: A systematic review. ReCALL, 1–21. https://doi.org/10.1017/S0958344024000181
  5. Golonka, E. M., Bowles, A. R., Frank, V. M., Richardson, D. L., & Freynik, S. (2014). Technologies for foreign language learning: A review of technology types and their effectiveness. Computer Assisted Language Learning, 27(1), 70–105.
  6. Rohmonqulov , M. (2025). DEVELOPMENT OF AN INTERACTIVE SOFTWARE TOOL FOR AUTOMATIC ASSESSMENT AND IMPROVEMENT OF PRONUNCIATION IN FOREIGN LANGUAGE LEARNING. Scientific Practical Conference, 1(1), 344-349.
  7. Rahimov Nodir, Khasanov Dilmurod. (2022). The Mathematical Essence Of Logistic Regression For Machine Learning. https://doi.org/10.5281/zenodo.7239169
  8. Khasanov Dilmurod, Tojiyev Ma’ruf, Primqulov Oybek., “Gradient Descent In Machine”. International Conference on Information Science and Communications Technologies (ICISCT), https://ieeexplore.ieee.org/document/9670169
  9. El Kheir, Y., Ali, A., & Chowdhury, S. A. (2023). Automatic pronunciation assessment: A review. Findings of ACL: EMNLP 2023, 8304–8324. https://doi.org/10.18653/v1/2023.findings-emnlp.557