Passive alternation in contemporary Chinese: a tale across the Strait

Passive sentences have long been a topic of interest among grammarians and are even regarded as levers to promote the construction of language theory (Lu 2004) in the field of generative grammar, as exemplified by Chomsky’s work. Within the framework of Probabilistic Grammar (Bresnan 2007; Bresnan & Ford 2010; Szmrecsanyi 2013) and variationist Grammar (Tagliamonte 2012), the alternation of Chinese passive constructions has yet to be fully analyzed to capture the implicit probabilistic knowledge of language users. The objective is to identify the language-internal factors that predict the choice between Chinese bèi/ràng/gěi long passives, and to establish if variant choice differs as a function of language variety (i.e., Mainland Chinese versus Taiwan Chinese). To this end, we extracted a rather balanced dataset from the Chinese Web Corpus (zhTenTen17), totaling 4,172 observations of Chinese analytic long passive constructions. Drawing inspiration from the literature on passive constructions in Chinese, we manually annotated all tokens for 17 syntactic and semantic language-internal factors. To determine the combined effect of multiple linguistic factors, mixed-effects logistic regression is used. Six factors remain consistent across varieties of Mandarin Chinese, delineating the different usage of these three variants, while eight factors exhibit fluidity across varieties of Mandarin Chinese, specifying different usage patterns of these three variants in two language varieties. We conclude that Taiwan Chinese aligns more closely with the expectations of passive constructions in the early modern Chinese while passive constructions in Mainland Chinese are more productive. This analysis confirms the complexity of language-internal and external linguistic conditioning on Chinese analytic passive constructions, which could have further implications for cross-linguistic comparisons and theoretical advancements in grammar studies.