The reliability of single task assessment in longitudinal L2 writing research

Single task writing assessments used in longitudinal studies have raised concerns regarding their reliability. By means of Generalizability Theory (GT), this study investigated the reliability L2 scored on different CAF measures, focusing a) single and effects b) topics c) task-taking occasions assessment We analytic quantitative scores obtained from five measures through a 1-day dataset 21-day dataset, consisting 90 essays 18 Chinese learners English who did not follow any formal language instruction during investigation. The results show that although some (e.g., fluency) distinctly higher than other scores, general conclusion is are reliable GT perspective. Task topic introduces score variance to result, yet amount differs profoundly between due functional variability, which corresponds with Complex Dynamic Systems assumptions suggesting subsystems an do develop synchronously. Finally, occasion, i.e., whether two samples were written same day or within 21 days, barely variance.