Second language (L2) learners’ vocabulary knowledge and lexical features of their essays are both pivotal in predicting writing quality. Although vocabulary size and depth have each been extensively studied in L2 research, how these different aspects of vocabulary knowledge influence the lexical characteristics of L2 writing has received less attention. This study examines how vocabulary size and depth contribute to computationally derived lexical features in L2 narrative writing and assesses the extent to which lexical diversity, density, sophistication, and cohesion predict human-rated writing quality. A total of 110 English as a Foreign Language (EFL) undergraduate students participated by completing the Productive Vocabulary Levels Test (PVLT), the Word Associates Test (WAT), and composing a narrative essay in English. Lexical features were examined through computational analyses using the Tool for the Automatic Analysis of Cohesion (TAACO), VocabProfiler, and Coh-Metrix. Regression analyses revealed that productive vocabulary size significantly predicted lexical sophistication, diversity and density, while vocabulary depth was associated with only one measure of lexical cohesion. Additionally, findings indicated that lexical sophistication and diversity were the key predictors of L2 narrative writing quality. The implications of the findings for L2 writing pedagogy and assessment are discussed.
