The Constraining Role of Information Structure in Resolving English PP Attachment Ambiguity: A Large-Scale Corpus Study of Film Subtitles
摘要
This study integrates experimental psycholinguistics and usage-based corpus linguistics to investigate how pragmatic and prosodic cues constrain syntactic parsing in natural discourse. Focusing on the resolution of prepositional phrase (PP) attachment ambiguity (e.g., “I saw the man with the telescope”), we examine the role of information structure. Using the large-scale OpenSubtitles corpus, we manually disambiguated and annotated thousands of instances for the givenness/newness of referents and the presence of commas as orthographic prosodic boundaries. Mixed-effects logistic regression analyses reveal that both information status and commas are robust predictors of attachment preferences. A contextually Given NP within the PP significantly promotes NP-attachment, while a comma preceding the PP strongly favors VP-attachment. These findings extend laboratory-based results by demonstrating the potent role of pragmatic and orthographic prosodic cues in ecologically valid language processing, underscoring the necessity of integrating discourse context into models of the syntax-pragmatics interface.