本文以《国际中文教育中文水平等级标准》界定的双音节道德复合词为研究对象，依托语义透明度、语素加工与中介语偏误分析理论，构建由语素可推度、极性识别度、搭配约束清晰度构成的三维道德词汇习得透明度评价体系。结合中介语语料统计与词汇推理实验，对比汉字圈、印欧语两类中级学习者对高、中、低透明度道德词汇的习得表现，厘清语义透明度、母语文字系统、语境强度对词义推理、词汇产出的作用机制。研究表明：道德词汇整体语义透明度显著低于通用具象词汇；透明度层级直接决定中介语偏误类型；母语文字系统仅作用于高透明词汇加工，充足语境可有效弥补低透明词汇习得障碍。基于实验结论，提出分层语素拆解、极性对比、语境支架三类词汇教学方案，为国际中文抽象评价词汇分级教学提供语言学实证依据。

This paper focuses on disyllabic moral compound words as defined by the Standards for Chinese Language Proficiency in International Chinese Education.Drawing on theories of semantic transparency,morpheme processing,and errors in the intermediate language,it constructs a three-dimensional evaluation system for the transparency of moral vocabulary acquisition,comprising morpheme inferability,polarity distinguishability,and collocation constraint clarity.Combining statistical analysis of the intermediary language corpus with lexical inference experiments,this study compares the acquisition performance of intermediate-level learners from both the Sinitic and Indo-European language families regarding high-,medium-,and low-transparency moral vocabulary,elucidating the mechanisms by which semantic transparency,the native writing system,and contextual richness influence lexical inference and vocabulary production.The study indicates that the overall semantic transparency of moral vocabulary is significantly lower than that of general concrete vocabulary;the level of transparency directly determines the types of errors in the intermediate language;the native writing system influences only the processing of high-transparency vocabulary;and sufficient context can effectively compensate for obstacles in the acquisition of low-transparency vocabulary.Based on these experimental findings,this study proposes three types of vocabulary teaching strategies—hierarchical morpheme decomposition,polarity contrast,and contextual scaffolding—providing linguistic empirical evidence for the graded teaching of abstract evaluative vocabulary in international Chinese language education.