语义角色标注(Semantic Role Labeling)是浅层语义分析的一种实现方式。在汉语国际教育事业得到长足发展的时代背景下，本文将语义角色标注研究引入汉语国际推广领域，以HSK汉语水平考试真题和对外汉语教材语料为对象开展语义角色标注研究，以推动对外汉语教学中重点、难点语言现象的语义分析和处理。研究内容主要包括以下几个方面：第一、调研现有的语义角色标注评测、资源及方法，设计适用于该领域语料的语义角色标注体系；同时，对信息处理技术在汉语国际教育领域的应用进行考察和总结，确立本文语言分析技术的具体应用方向。第二、结合教学需求，重点关注汉语特有介词（如“把”、“对”）所在的句式，并选取把字句、对字句、被字句、给字句、向字句、由字句这六个高频出现的特殊句式作为分析对象，研究介词功能意义、句式语义类别及句中谓词特征，为句法和语义分析系统的构建打下基础。第三、在前文工作的基础上，引入介词语义类别、介词句法层次、动词配价、动词句类属性、动词语义框架信息、介词-动词搭配信息等一系列重要的句法语义特征，构建面向汉语国际教育的语义角色标注系统，实现介词短语自动识别、特殊句式语义角色标注、普通句式语义角色标注等功能。最后，回归教学角度，探索系统分析结果如何具体地服务于二语教学：对特殊句式自动释义功能、句式类别分布和介词与谓词的搭配信息进行研究，以期向对外汉语教学和研究工作提供一定辅助。

Shallow semantic analysis has been achieved by semantic role labeling. With the rapid development of teaching/learning Chinese as a second language, we would introduce semantic role labeling to the field of second language acquisition, and try to present a system that could make syntactic and semantic analysis of the sentences in HSK exams and classic textbooks. The research mainly focuses on the following parts.Firstly, design an annotation set for the domain-specific corpora based on the investigation of the applicable resources, methods and evaluation of semantic role labeling. After that, investigate the current application of information processing technologies for second language acquisition, and to have an explicit understanding of research direction.Taking teaching/learning needs into consideration, the second part makes linguistic analysis of the sentences with special prepositions such as BA, DUI, et al. With BA-Sentence, DUI-Sentence, BEI-Sentence, GEI-Sentence, XIANG-Sentence, and YOU-Sentence as examples, we study the function of prepositions, semantic categories of sentences and features of predicates.Based on the above work, the paper defines a series of semantic features of prepositions and verbs, including preposition categories, syntactic levels, verb valency, verb categories, verb frames and preposition-verb collocations. These features are employed in the semantic role labeling system, which has three models respectively for the automatic identification of prepositional phrases, semantic role labeling for special sentences with prepositions and semantic role labeling for sentences with normal patterns.Finally, back to the language acquisition needs, we try to find that how the system results could be used in language learning/teaching. The answer is given by the discussions of the automatic interpretation of special sentences, quantitative distribution of sentence categories and high-frequency collocations of predicates and different prepositions. It may offer reference value to the teachers, students and researchers in this field.