Lexical density and diversity in dissertation abstracts: revisiting English L1 vs. L2 text differences

This study investigated lexical density and diversity differences in English as L1 vs L2 academic writing of EFL, ESL, and English L1 postgraduate students to compare their lexical proficiency in EFL vs. English L1 academic settings. A corpus of 210 dissertation abstracts was analysed using three natural language processing tools [LCA, TAALED, and Coh-Metrix] where the effects of text length and topic were controlled. In doing so, we examined the relationship between 15 lexical indices and the construct-distinctiveness of lexical density and diversity. The measure-testing process also assesses the effectiveness of each measure in a pair/group of closely-related measures (in terms of the quantification methods) in capturing lexical diversity differences of these texts. This is to obtain a small number of unique measures that capture lexical diversity as an indicator of lexical proficiency and to assist future writing researchers in the measure-selection process in the face of a multitude of available measures. The findings have important implications for writing assessment and research on lexical indicators of writing proficiency, materials development in EFL academic settings especially for thesis/dissertation writing modules, and a possible contribution of ESL academic immersion programmes in approximating English L1 and L2 proficiency.