The domain of pronouns and backward binding

Abstract This study examines the relationships between personal pronouns and their nominal antecedents, focusing on the phenomenon of backward binding. Drawing primarily on data from English, Polish, and Czech, it demonstrates that such coreference relations are governed by three key conditions: (1) linear precedence, (2) a structural constraint known as phase command ( Bruening 2014 ), and (3) the information structure status of the nominal antecedent, which must be [+backgrounded/+topic] ( Reinhart 1976 , 1983 ; Bianchi 2009 ; Biskup 2011 , among others). The findings reveal that personal pronouns cannot occupy more prominent (commanding) syntactic positions than the nominals they refer to within a given sentence. Notably, even when pronouns are embedded within prepositional phrases (PPs) in English and Polish, they still trigger Principle C effects. This suggests that while PPs are legitimate constituents, their boundaries do not constrain the pronoun’s command domain. Instead, the command domain is delimited by the boundaries of derivational phases (e.g., vP, CP). Consequently, a nominal antecedent coindexed with a preceding pronoun is most natural when positioned in a separate clausal domain, such as an adverbial clause. Additionally, the analysis shows that right-peripheral adjunct clauses fall within the command domain of the subject pronoun but not the object pronoun. Crucially, the antecedent nominal phrase must be [+backgrounded/+topic] to establish a coherent coreference relationship.