Phraseological Units in Modern English: A Corpus-Based Analysis of Structural, Semantic, And Pragmatic Dynamics

Abstract

This study presents a large-scale corpus-based investigation of phraseological units (PUs) in contemporary English, drawing on a 515-million-token dataset from the Corpus of Contemporary American English (COCA). The primary objectives are to (1) map the structural taxonomy of PUs, (2) analyse cross-genre semantic distributions, and (3) trace diachronic frequency shifts across three decades. Using a multi-level annotation pipeline integrating automated pattern extraction and manual verification, 11,208 distinct PUs were identified and classified. Findings reveal a marked increase in collocation-based PUs (+48%) alongside a decline in proverbial and idiomatic forms (-29%) over the 1990–2024 period. Genre-specific register profiling demonstrates differential pragmatic loading across news, academic, fiction, and spoken sub-corpora. The study contributes a replicable methodological framework and an open-access annotated dataset, advancing both theoretical phraseology and NLP applications.