Gavin BROOKS    Department   Kyoto University of Foreign Studies  Department of British and American Studies, Faculty of Foreign Studies    Position   Associate Professor
Date	2025/09/04
Presentation Theme	Identifying lexical indicators of writing proficiency: A data-driven approach to supporting L2 writers in local classrooms
Conference	British Association of Applied Linguistics (BAAL) 2025 Conference
Promoters	British Association of Applied Linguistics (BAAL)
Conference Type	International
Presentation Type	Poster
Contribution Type	Collaborative
Country	United Kingdom
Venue	The University of Glasgow
Holding period	2025/09/04～2025/09/05
Publisher and common publisher	Gavin Brooks, Jen Jordan, Jon Clenton
Details	This study analyzed a longitudinal corpus of over 1,500 essays written by 500 first-year Japanese university students across three academic topics to investigate how specific lexical features related to English proficiency measured by TOEFL scores. Initial regression analyses examined lexical diversity measures (MTLD, Guiraud's Index) and sophistication measures (word frequency bands, academic word coverage, Age of Acquisition) as predictors of proficiency. Machine learning methods including LASSO and tree-based models were then applied to identify specific words and lexical patterns that differentiated essays by proficiency level.