This study explores the utilization of lexical bundles (LBs) in the reading sections of international English proficiency tests, specifically IELTS and TOEFL, in comparison with Iranian local assessments such as MSRT, TOLIMO, and EPT. The research aims to analyze both the structural and functional categories of these LBs. To facilitate this investigation, two distinct corpora were compiled, consisting of 513,490 tokens from international assessments and 432,252 tokens from local Iranian tests. Utilizing AntConc software for quantitative analysis, the study identified LBs and conducted a qualitative examination of their structural patterns, functional roles, and contextual applications. The findings reveal significant differences between the international and local assessments: 1,030 instances of LBs were identified in the International Corpus (IC), whereas the Local Corpus (LC) contained only 564. Inferential statistical analysis indicated a higher prevalence of discourse organizers and stance bundles in the IC, contrasted with a predominance of referential bundles in the LC. Furthermore, international assessments featured mainly noun-phrase fragments, unlike the Iranian ones that predominantly used prepositional-phrase fragments. Statistical tests, including the Linear-by-Linear Association test (p=0.028) and the Chi-square test (χ²(2)=28.803, p<0.001), confirmed these structural variations, suggesting divergent academic discourse conventions between the corpora. The results underscore the need for revisions in Iranian assessments to align with international standards, potentially enhancing global academic proficiency and fostering greater learner engagement in higher education contexts.
