
Computer-assisted pronunciation training (CAPT) increasingly incorporates automatic speech recognition (ASR) to provide pronunciation assessment and feedback. However, the extent to which ASR systems evaluate non-native Mandarin Chinese speech in a manner comparable to human teachers remains unclear. This study compares the assessments generated by three ASR systems-Whisper, Azure, and Gladia-with ratings provided by native Chinese-speaking teachers for the word-level Mandarin Chinese pronunciation of 31 Japanese learners. Two research questions are addressed: (1) To what extent do these ASR systems assess learner pronunciation comparably to teachers? (2) Can ASR assessments help identify learners' pronunciation difficulties? A three-point scoring scheme was developed to evaluate learners' productions of 20 Mandarin Chinese words. Comparative analyses were conducted from the perspectives of learner proficiency and pronunciation characteristics. The results showed that all three ASR systems generally underestimated learner performance relative to teacher ratings, although Whisper produced assessments that were most consistent with those of the teachers. The agreement between ASR and teacher assessments also varied according to learner proficiency. Furthermore, ASR performance was strongly influenced by initial-final combinations, suggesting that ASR assessments can help identify specific pronunciation difficulties. These findings support the potential of ASR as a complementary tool for pronunciation assessment in Mandarin Chinese CAPT.
With the continued advancement of the digital transformation of international Chinese language education, the traditional manual approach to test item development has increasingly encountered bottlenecks in terms of efficiency, cost, and scalability. Against this backdrop, artificial intelligence technologies, particularly large language models (LLMs), have opened up new possibilities for automated test item generation. This study focuses on HSK Level 6 reading comprehension items and systematically evaluates the practical effectiveness of LLMs in automatic item generation. Four LLMs were selected for experimentation using prompt engineering, including instruction-tuned models, reasoning-oriented models, and a domain-specific model fine-tuned with Low-Rank Adaptation (LoRA). The generated items were comprehensively evaluated across six dimensions: linguistic fluency, content accuracy, item complexity, distractor quality, answer uniqueness, and item-type diversity. In addition, machine-based evaluation metrics, including BLEU, ROUGE, and Distinct, were employed to provide complementary assessments. The results indicate that the items generated by LLMs exhibit a high degree of similarity to those developed by human experts. However, the models still demonstrate instability in controlling item difficulty and ensuring answerability, suggesting that human review and revision remain necessary before the generated items can be used in instructional settings. Among the models evaluated, reasoning-oriented LLMs achieved the best overall performance. Based on these findings, this study further proposes practical recommendations for optimizing the item generation process and advancing a human-AI collaborative approach to test development.
In an era in which digital transformation is reshaping the landscape of language learning, innovative and inclusive approaches are becoming crucial. This paper presents e & centerdot;Chinese Plus, an initiative of a team of lecturers from the Universitat Aut & ograve;noma de Barcelona (UAB). e & centerdot;Chinese Plus is a comprehensive, Moodle-based, open-access platform for enhancing the acquisition of Chinese as a foreign language (CFL) specifically addressed to Spanish-speaking learners. The paper outlines the platform's pedagogical rationale and describes the methodological principles guiding the design and development of its activities, as well as its distinctive features and potential use as a complement to formal instruction. It further examines how generative artificial intelligence (GenAI) has been selectively integrated into the design of specific activity types, while emphasizing the central role of teaching experience in addressing the needs of Spanish-speaking learners, particularly through activity design and in-task feedback. Finally, the paper discusses future development plans and ongoing challenges, highlighting the potential of this initiative to inspire other CFL teachers interested in creating similar digital learning resources.
Large Language Models (LLMs) are increasingly used in Chinese as a Second Language (L2) learning, yet their ability to comply with pedagogical constraints specific to the Chinese writing system remains underexplored. This study examines whether system prompts aligned with the CEFR-EBCL framework enable LLMs to generate learner-facing Chinese texts that respect sinographic thresholds at the A1, A1+ and A2 levels. We conducted controlled experiments using two models (GPT-4.1 and GPT-4.1-mini) across ten EBCL-related written tasks. Prompt conditions with and without explicit character lists were compared. Model outputs were automatically analyzed to quantify instruction deviation, defined as the proportion of characters outside the target EBCL set. Results indicate that including explicit character lists significantly reduces out-of-list character production at the A1 and A1+ levels, particularly with GPT-4.1. At the A2 level, this effect becomes marginal. These findings provide empirical evidence on the pedagogical value and limits of prompt-based control of ChatGPT outputs for CEFR-EBCL-aligned Chinese L2 learning.
Video-based instruction is one of the most widely used forms of multimodal resources in Chinese as a second language (CSL) teaching. Despite the high demand for instructional videos, their production is often constrained by substantial time, labor, and financial costs. This study explores the practical application of generative artificial intelligence (GAI) in the development of CSL instructional videos. Using vocabulary instruction as an illustrative case, the study investigates the process of AI-assisted video generation and evaluates its pedagogical applicability. A complete workflow for GAI-enabled Chinese language instructional video production was developed and assessed, and both Chinese language teachers and learners were invited to evaluate the AI-generated videos. The results indicate that GAI can substantially reduce the technical barriers, time investment, and production costs associated with video creation. While the generated videos were generally usable for instructional purposes, limitations remained in areas such as the depth of linguistic explanations and the selection of pedagogically appropriate example sentences. Based on these findings, the study proposes three implications for the use of GAI in Chinese language instructional video development, with the aim of promoting more effective integration of GAI into international Chinese language education.