On Constructional Coercion in the Processing of Chinese Novel Denominal Verb Double-Object Sentence: An ERP Study

Based on the ERP technology and choosing "RP to le", NP1 and NP2 as the keywords, this study explores the process of interaction between the V/Nv and the constructional meaning during the comprehension of the three types of Chinese double-object sentences (CDO), covering the normal sentences in which trivalent verbs are used as the predicate, coerced sentences in which nouns relevant to the meaning of "transmission" are used as verbs and incoercible sentences in which nouns irrelevant to the meaning of "transmission" are used as verbs. The results of the ERP experiment show that after "hat afle" and NP1, the N400 and P600 between the coerced sentences and incoercible sentences are not significantly different. During NP2 397 - 709 ms, Sustained Anterior Negativity (SAN) evoked by the three types of sentences respectively is significantly different: SAN evoked by the incoercible sentences is the largest, while SAN evoked by the normal is the smallest. As the three types V/ Nv have different relations with the CDO constructional meaning, this result shows the involvement of constructional meaning in sentence semantic integration. During NP2 718-782 ms, the incoercible sentences evoke the largest late negativity (LN), which is significantly different from the LN of the normal and coerced sentences. This observation indicates that the normal and coerced sentences rather than the incoercible sentences bring about the semantic integration. From the processing of the whole CDO, the differences resulted from the processing of incoercible and coerced sentences after NP2 are mainly caused by the two types of Nvs having different relation with the CDO constructional meaning. The ERP results have provided evidence for the existence of CDO constructional meaning independent of the verb. Constructional coercion can be successful as long as the semantic characteristics of lexical items match the constructional meaning.