Investigation of Semantic Relevance Between Different English Modal Verbs by Attribute Partial Ordered Structure Diagram

The semantic relevance of the English modal verbs refers to the semantic similarities and differences between the modal verbs. Since the semantic similarities may confuse the selection of one modal verb from the others in the application, the solution of the semantic relevance of different modal verbs has been an urgent and important issue in both linguistic study and natural language processing. However, the issue is unsolved yet. Therefore, in this article, we focus on this issue. The English modal verbs can and may are selected as the target words, and the semantic relevance between the two modal verbs is investigated using the approach of attribute partial ordered concept diagram. A 1.1-million-word corpus is built including 12 different genres, and eight semantic features, 3 syntactic features and 4 genre features are extracted and used to construct the formal contexts of can and may, respectively. Based on the two formal contexts, the APOSDs are generated as the models for word sense disambiguation and knowledge discovery. Based on the models, the rules for WSD of can and may are extracted and some knowledge in semantic relevance are discovered by examining and comparing the rules for WSD, the composition of the rules, the distributions of the attributes in the APOSDs, the occurrence of some attributes in different genres, and the functions of different attributes. The discovered knowledge provides valuable, theoretical and practical evidence for the semantic study of modality, especially for the natural language processing and machine translation of English modal verbs.