LEXICAL BUNDLES IN PAKISTANI SOCIAL SCIENCES RESEARCH ARTICLES: A CORPUS-BASED ANALYSIS OF FREQUENCY, STRUCTURE, AND FUNCTION

Abstract

Lexical bundles are recurrent multiword sequences that contribute to the organization, fluency, and rhetorical patterning of academic discourse. This corpus-based study examines the frequency, structural forms, and discourse functions of three- and four-word lexical bundles in Pakistani research articles in the social sciences. The corpus comprised 500 research articles published in Higher Education Commission-recognized Pakistani journals between 2013 and 2017, representing linguistics, literature, political science, psychology, and sociology. The corpus contained 2,079,944 word tokens and 46,951 word types. Lexical bundles were extracted through AntConc 3.5.2 and then manually filtered to remove overlapping or incomplete bundle forms. Structural analysis was guided by bundle-pattern categories derived from Biber et al. (1999) and Biber, Conrad, and Cortes (2004), whereas functional analysis followed Hyland’s (2008) research-oriented, text-oriented, and participant-oriented framework. The results show 1,931 lexical bundles across the corpus, with sociology producing the largest number and political science the smallest. Noun phrase-based bundles dominated the structural profile across disciplines, indicating the nominal and informational density of Pakistani academic prose. Functionally, research-oriented and text-oriented bundles were more frequent than participant-oriented bundles, suggesting that Pakistani social science writers prioritize description of research procedures, textual organization, and reporting of results over explicit writer-reader interaction. The findings have implications for English for Academic Purposes instruction and for corpus-informed academic writing pedagogy in Pakistan.