Anxiety is a major affective factor in foreign language learning, par-ticularly in speaking for Chinese learners with few authentic envi-ronments. This study maps how research on speaking anxiety in Chi-na has evolved, separating established findings from the problems that remain. The corpus is 262 articles on speaking anxiety among Chinese learners, published in Chinese-language journals indexed by CNKI, 2005–2025. A main search returned 299 records and a sup-plementary core-journal search added 13, giving 312 candidates; 50 were excluded at screening. It combines CiteSpace keyword analysis with article-level content coding on six dimensions. Three findings emerge. First, output follows a three-stage pattern — initial growth (62 articles, 2005–2010), rapid expansion (96, 2011–2015), stable fluctuation (104, 2016–2025); undergraduates account for 57.6% of the corpus. Second, the keyword network resolves into ten clusters, seven of which form two main lines: basic issues and pedagogical countermeasures. Third, burst keywords indicate a front moving from alleviating negative emotions towards positive-psychology con-structs. The coding, however, reveals a documentation deficit. Of the 262 articles, 84.4% do not report their anxiety instrument’s source, dimensions, reliability or validity and 80.2% declare no theoretical framework; 37.8% describe no data collection procedure, and only 6.5% report an experiment. These figures describe reporting practice in this corpus, not the field as a whole. The contribution is to show that increasing thematic consolidation has not been matched by methodological consolidation. Future research should strengthen theoretical grounding, standardise instrument reporting, and increase the use of longitudinal tracking and intervention designs.
Copyrights © 2026