Managing Cognitive Complexity of Academic Writing Tasks in High Stakes Exams via Mentor Text Modeling: A Case of Iranian EFL Learners

Cognitive complexity is traditionally used for describing human cognition along a simplicity-complexity axis in tests like TOFEL IBT and GRE where text creation and rhetorical organization are quintessentially important. Accordingly, this study sought to investigate the impact of mentor text modelling on cognitive complexity of academic writing tasks in terms of students’ responses to the test inputs. For this purpose, from the population of applicants applying for various high stake exams at Jihahde Daneshgahi, Isfahan University, three intact classes were selected based on a convenient sampling method. The students, both male and female, were graduates from various majors in applied sciences whose age range was between 24 and 29 and they had all passed the preparatory classes required for attending academic writing courses. Each targeted class with twenty-five applicants was concurrently programmed for three writing tasks with various cognitive complexity levels- Independent, integrated, and analytical. The classes, a total of 75 EFL learners, were randomly assigned to three equal groups labeled as product based (PBG), process based (PRBG), and mentor text modeling (MTMG) respectively. Employing a posttest only quasi-experimental design, learners in the three groups received their instruction on advanced writing during a sixteen session course. The learners in each group were taught based on the selected writing approaches. At the end of the treatment, the learners' writing performance was assessed on test tasks within the pre-specified time and word limits by utilizing a relevant posttest. Data analysis reflected that mentor text modeling enjoyed a potentially higher pedagogical efficacy since the learners in the MTM experimental sample performed better in terms of both accuracy and fluency compared with the groups receiving their writing instruction through either product or process based approaches. Notably, the findings revealed that mentor text modeling is a functionally dependable resource for managing writing tasks cognitive complexity and neutralizing the trade-off effect between accuracy and fluency by offering insightful pedagogical hints to EFL teachers, test takers, and writing material developers who have always had a hard time calibrating writing accuracy and fluency in high stake exams.