A Multifactorial Analysis of Chinese Long Passive Constructions Marked by Bei, Gei, and Rang in Contemporary Chinese

Passive constructions marked by bèi/gěi/ràng are the predominant strategy to indicate passive voice in Mandarin Chinese. These passive variants are in principle well-studied, but the alternation has yet to be fully analyzed using variationist techniques to explore the probabilistic grammatical knowledge that language users have about the long passives. To explore the variation in contemporary Chinese, the present study taps into the Chinese Web Corpus (zhTenTen17). The objective is to identify what the language-internal factors are that predict the choice between Chinese bèi/gěi/ràng long passives and to establish if variant choice differs as a function of language variety. To determine the combined effect of multiple linguistic factors, mixed-effects logistic regression is used, following best practice in variationist (socio)linguistics. The results reveal that in atypical passive constructions, the marker bèi is preferred. Both gěi and ràng passives convey a more colloquial tone, while ràng passives inherit the attribute of avoidable events from ràng causatives, favoring an inanimate NP1 to avoid ambiguity. Compared to Mainland Chinese, we show that in Taiwan Chinese, the typical passive marker bèi aligns more closely with old-before-new information structure and is favored by adversative contexts; ràng serves as a substitute for bèi when the semantic prosody is positive, whereas the usage of gěi is influenced by the use of hoo in Taiwanese and also conditioned by the cognitive complexity principle. This analysis confirms the complexity of languageinternal and external linguistic conditioning on Chinese long passive constructions and paves the way for further research in Chinese passive constructions.