skip to main content skip to footer

A Comparison of Two TOEFL Writing Tasks TOEFL iBT ELA WAD IND L2

Author(s):
Davis, Larry; Norris, John M.
Publication Year:
2023
Report Number:
RM-23-06
Source:
ETS Research Memorandum
Document Type:
Report
Page Count:
67
Subject/Key Words:
TOEFL iBT, TOEFL Essentials, Test Revision, Test Scores, Score Interpretation, Test-Taker Performance, Second Language (L2), English Language Assessment (ELA), Academic Writing, Writing Ability, Writing Assessment, Writing Evaluation, Writing Fluency, Writing, Fairness, Validity, Reliability, Write for an Academic Discussion (WAD), Independent Writing (IND), Scoring Rubrics

Abstract

One component of the 2023 update to the TOEFL iBT test was the replacement of the Independent Writing (IND) task with the Writing for an Academic Discussion (WAD) task. Though both tasks assess academic English writing ability by eliciting extended writing performances, there are certain differences in their design. Most apparently, the WAD task allows for up to 10 minutes of writing time, compared with 30 minutes for the IND task, hence the amount of writing elicited may differ considerably between the two tasks. Nevertheless, both tasks have the same communicative goal (express and support an opinion) and are scored according to a similar rating scale, and the new task is intended to contribute to the calculation of a total writing section score in the same way as the previous task. To justify this update, we compared performances on the two tasks by a sample of test takers (N = 242) who completed both tasks on operational tests at similar points in time. We found that human scores on the two tasks distributed test takers into similarly broad ranges of ability and that scores on the two tasks were identical or within one score point 93% of the time. We also found that direct measures of writing performances using automated natural language processing tools revealed substantial similarities in the quality of texts produced by test takers on the two tasks, in terms of the syntactic complexity, grammatical accuracy, lexical variety, discourse cohesion and elaboration, and fluency of their writing. Only slight differences were identified in a few measures of lexical, fluency, and discourse dimensions of the tasks. Overall, findings provide initial support for using the new task to support interpretations about English writing ability. We discuss these findings in light of the validity of interpretations and uses of the TOEFL iBT test, and we highlight implications for the standardized assessment of academic English proficiency.

Read More