Use of lexical features in non-native academic writing

Abstract

Abstract Second language writing research has often analyzed written discourse to provide evidence on learner language development; however, single word-based analyses have been found to be insufficient in capturing learner language development (Read & Nation, 2006). This study therefore utilized both single word-based and multi-word analyses. Specifically, it explored vocabulary distributions and lexical bundles to better understand the development of writing proficiency across three levels in an English Placement Test corpus (EPT) (N\u2009=\u20091388). Inference tests for multiple population proportions were conducted to compare statistical differences in the proportions of vocabulary and lexical bundle distributions across all group levels. The results suggested that higher proficiency learners used a higher number of types, tokens, and word families than lower proficiency learners. Regarding lexical bundles, noun phrase-based and verb phrase-based bundles with referential and stance functions were significantly found in lower-level responses. Preposition phrase-based bundles were significantly used in higher-level written discourse. This study suggests the importance of vocabulary and lexical bundles in academic writing, the necessity of including these features identified in this study in a second language writing curriculum, and the need to incorporate these features prevalent in a rating scale for assessments of academic writing.