Lexical Richness and Academic Writing Performance in a Thai University Academic English Test

Abstract

Although lexical richness is an important feature of L2 writing quality, research on its predictive ability in university entrance tests, particularly in the Thai EFL context, is limited. Following Read’s (2000) framework, this study analyzes the relationship between lexical richness and academic writing performance in an academic English test at a Thai university. The study aims to (1) compare lexical richness characteristics across proficiency levels; (2) examine their relationship with writing scores; and (3) determine their predictive power independent of text length. A corpus of 103 argumentative essays was analyzed using TAALED and TAALES. The essays were classified into two groups based on university admissions criteria. The Meets Standard Group outperformed the Below Standard Group on text length (d = 1.54), lexical diversity as measured by HD-D (d = 0.76) and MATTR (d = 0.75), and bigram frequency (d = 0.47), but not on the Maas index, lexical density, word frequency (KF and BNC written), word range, and AWL proportion. Text length correlated most strongly with scores, followed by HD-D, MATTR, and bigram frequency, while other indicators did not reach significance. Regression analysis showed that, when text length was controlled, the model explained 64.8% of the variance, with text length as the dominant predictor (β = .691). Lexical diversity (HD-D) retained a small but significant independent effect (β = .174, p = .010), whereas lexical sophistication (bigram frequency, BNC written frequency, AWL proportion) and lexical density did not. These findings provide empirical evidence from the Thai EFL context for vocabulary instruction and rubric design.