A methodology for correcting tonal errors in Russian students learning Chinese using generative artificial intelligence

Abstract

This article presents a methodology for correcting tonal errors among Russian-speaking students learning Chinese through the use of generative artificial intelligence. An analysis of academic and methodological literature, combined with pedagogical modeling, enabled the identification of three types of difficulties: confusion between tone categories, narrowing of the pitch range, and distortion of tonal contours. A specific form of corrective feedback is proposed for each type: categorical comparison, visualization of acoustic parameters, and temporal localization of deviations, respectively. The methodology comprises perceptual training, contrastive diagnosis, and productive practice in new communicative contexts, with the possibility of returning to previous stages when necessary. The roles of instrumental speech analysis, generative AI, and the teacher are clearly differentiated. Particular attention is given to implementation conditions, methodological limitations, and directions for future empirical validation. The article is theoretical and methodological in nature; no pedagogical experiment was conducted.