Abstract

This study investigates the syntactic patterns associated with the passive voice in the Corpus of Contemporary American English (COCA), a comprehensive dataset exceeding one billion words from 1990 to the present. Employing a corpus-based methodology, the research analyzes be-passives, get-passives, and prepositional passives through targeted queries, frequency normalization, and qualitative classification of argument structures and semantic prosody. Quantitative findings reveal the dominance of be-passives at approximately 3,500 occurrences per million words, with higher frequencies in formal genres like academic texts, while get-passives prevail in spoken registers at 600 per million words, exhibiting adversative prosody and diachronic increases indicative of colloquialization. Prepositional passives remain stable but niche, often integrating verb-preposition units for affectedness emphasis. The analysis highlights genre-specific variations, lexical constraints in ditransitive verbs, and representational overlaps with active voice, grounded in Construction Grammar and Usage-Based Theory. These insights underscore the passive voice's adaptability in contemporary American English, with implications for linguistic theory, English pedagogy, and natural language processing. Limitations include reliance on COCA's tagging and exclusion of non-central forms, suggesting avenues for comparative varietal studies.