Chinese Complement Coercion: Evidence from real-time sentence processing

This study aims to explore the controversial issue of complement coercion in Chinese. Complement coercion has been widely investigated in English in the past decades. It refers to the coerced change of complement type typically associated with English verbs, such as begin and enjoy, which semantically select an event or an activity as their complement (Jackendoff, 1997; Pustejovsky, 1991). When these verbs are followed with a complement denoting neither an event nor an activity but an entity (e.g., began the novel), they will coerce the complement (e.g., the novel) into an eventive interpretation (e.g., reading) to satisfy the semantic restriction (e.g., began reading/to read the novel). This phenomenon is generally called complement coercion and the verbs coercing verbs. Noted that coercion sentences contain a piece of eventive information, which is not expressed morphosyntactically—neither by individual lexical items nor by syntactic structure of the sentences. Previous studies have reported that processing English coercion sentences is difficult during real-time comprehension. The difficulty mainly reflects on additional processing cost (i.e., longer reading times) (e.g., McElree et al., 2001) and brain activity (i.e., specific neural patterns) (e.g., Baggio et al., 2010). While complement coercion has been frequently examined in English, it is relatively under-researched in Chinese. The lack of research in Chinese can be attributed to the common belief that the phenomenon does not exist in Chinese (Lin & Liu, 2005). However, several studies later on (Liu, 2005; Lin, Hsieh, & Huang, 2009; Hsu & Hsieh, 2013; Song, 2011, 2013; 2015) have shown that complement coercion indeed exists in Chinese. To this point, a question was raised: If complement coercion exists in Chinese, will Chinese coercion sentences also engender additional processing cost as observed for English, compared with their non-coercion controls during comprehension? We conducted a self-paced reading experiment to examine reading time for Chinese sentences with four conditions: (a) CoercingV (Vcoercing + NP, e.g., 尝试喜剧电影), (b) PreferredV (Vpreferred + NP, e.g., 演喜剧电影), (c) Full-VP (Vcoercing + Vpreferred + NP, e.g., 尝试演喜剧电影), and (d) UnpreferredV (Vunpreferred + NP, e.g., 写喜剧电影). We found longer reading times at and shortly after the complement of coercion construction than non-coercion controls. The results correlate with the previously relevant study on English complement coercion.