Discovery Of Classification Features Of Semantic Merger Of English Modal Verb Should

Natural language is a system with indeterminacy.English modals are closely related to human cognition,emotion and attitude.English modals are considered to be the most complex grammatical system in natural language.In recent years,some scholars studied the word sense disambiguation（WSD）of English modal verbs,which mainly focused on the determined sense disambiguation and rule extraction.However,the research on semantic indeterminacy is not enough,and the research on semantic merger of English modal verbs is far from enough.Based on a self-built English corpus of 5 million words,this thesis discusses the classification features of semantic merger of English modal verb should.Firstly,a word sense disambiguation model containing three kinds of objects,including should（root obligation and epistemic inference）and should（semantic merger）is established.300sample sentences are selected as the objects from the corpus.Thirty-two features related to the semantics of should are extracted from different aspects to generate a formal context to express the co-occurrence relationship between objects and the features.Based on the formal contexts generated above,the rules are extracted by Filter-APOSD method.Based on this,the classification features that can help the computer to solute the semantic merger can be found.The research found that:First,the features that most help to identify the semantic merger of should are contra-factivity and harmonic combination of know/understand.The discovery of these features can effectively help the computer to find and recognize the semantic merger of should.Second,12 classification features of semantic merger are found,among which 8 are syntactic features,accounting for the highest proportion.Third,the features 1^stperson subject,2^ndperson subject,perfect aspect,implying future time,directives and formal genre only occur in the rules that can classify the semantic-merger should into root obligation meaning,making a unique contribution to classifying semantic-merger objects into root obligation meaning.The features 3^rdperson subject and inanimate subject appear only in rules that can classify the semantic-merger should into epistemic inference meaning.These features make a unique contribution to classifying semantic-merger objects into epistemic inference meaning.Fourth,features stative verb,3^rdperson subject,MI₆（should＿MER,v）?0.40),agentive verb and 1^stperson subject can help classify most of the semantic-merger objects into different meanings.They are the classification features which contribute the most to the solution of semantic merger of English modal verb should.The findings of this research provide a way to study the semantic indeterminacy of modal verbs and provide a method and reference for the recognition and solution of semantic merger of English modal verb.The research findings provide theoretical support and a practical basis for the study of the semantic indeterminacy of English modal verbs and natural language processing.