Rethinking Band-Level Differences in IELTS Writing Task 2: A Multidimensional Learner-Corpus Analysis

Abstract

This study re-examined IELTS Writing Task 2 band-level differences from a multidimensional learner-corpus perspective. Rather than assuming that higher bands can be explained by longer essays, difficult vocabulary, cohesive devices, or sentence complexity, it tested the extent to which selected textual features differentiated essays across Bands 5–8. The final analytic corpus contained 1,912 unique publicly available IELTS Writing Task 2 essays across four human-sourced approximate band groups after exact duplicate screening. Basic text-production features were calculated in Python, while lexical diversity, lexical sophistication, syntactic complexity, and cohesion indices were extracted using TAALED, TAALES, TAASSC, and TAACO[cite: 12]. Descriptive statistics, Kruskal–Wallis tests, false discovery rate correction, Holm-adjusted pairwise comparisons, sensitivity analysis, and effect sizes were used to profile band-level variation[cite: 12]. The results showed statistically detectable but substantively modest differences[cite: 12]. Higher-band essays tended to show greater connector use, higher lexical sophistication, more academic vocabulary, and lower adjacent overlap, while essay length showed only a weak and sensitivity-dependent contrast[cite: 12]. Syntactic-complexity indices were weak separators, and no single feature provided a strong band marker within the reliability limits of the approximate band labels[cite: 12]. The study therefore characterizes band-level variation in IELTS Writing Task 2 as an uneven multidimensional textual profile rather than a single-feature pattern, offering evidence against formulaic IELTS pedagogy based on word count, rare vocabulary, mechanical connector insertion, or sentence-length expansion[cite: 12].