Using Regression to Reduce L2 Teachers' Scoring Workload: Predicting Essay Quality from Several Rhetorical Moves

Volume 9 Issue 3 July - September 2019

Research Paper

Using Regression to Reduce L2 Teachers' Scoring Workload: Predicting Essay Quality from Several Rhetorical Moves

Kutay Uzun*

*Department of English Language Teaching, Trakya University, Edirne, Turkey.

Uzun, K. (2019). Using Regression to Reduce L2 Teachers' Scoring Workload: Predicting Essay Quality from Several Rhetorical Moves. i-manager’s Journal on English Language Teaching, 9(3), 27-35. https://doi.org/10.26634/jelt.9.3.15653

Abstract

Scoring essays written in L2 is one of the most arduous and time-taking tasks for language teachers due to the heavy work involved in the process. In this respect, the present study aimed to build a regression model to predict essay quality using only one or two rhetorical moves in the setting of an English Language Teaching department in Turkey. The corpus of the study consisted of 265 essays written by 105 students of English Language Teaching. The essays were written through one semester in a 2nd year English Literature course in the form of literary analysis essays which aimed to identify how an author or poet dealt with a theme or character in a literary work. For data analysis, the Genre-Based Literary Analysis Essay Scoring Rubric was initially used to score all the essays within the corpus. After the normality tests, two regression models were built in order to predict essay scores using the thesis statement scores and a combination of thesis statement and opinion statement scores. The results showed that the chosen rhetorical moves could significantly predict essay scores. Using predictive models, teachers can distinguish poor, mediocre, and good essays without scoring of whole essays.