X

Dissertation Defense Announcement

 College of Arts and Sciences announces the Final Dissertation Defense of

Adam Ramsey

for the Degree of Doctor of Philosophy

October 03, 2019 at 04:00 PM,Ellington Hall  Room 100

Advisor: Jennifer R. Mandel

The Impact of Task Difficulty on Reading Comprehension Intervention with Computer Agents

ABSTRACT: Learners benefit from tasks that are difficult but can be achieved with the guidance of a skilled partner (e.g., Vygotsky's zone of proximal development). This concept has been implemented during the development of intelligent tutoring systems (ITSs). Very little research has empirically investigated this concept for reading comprehension interventions in ITSs. In order to provide individual learners tasks with the right difficulty, learner differences need to be considered. The research that specifically investigated the potential interaction between learner attributes and ITSs on learning in the domain of reading comprehension has yielded mixed results.The present dissertation explored how and to what extent individual differences interact with task difficulty in terms of their impact on memory, learning and engagement during a reading comprehension intervention with an ITS. A within-subject experiment was conducted which involved two different conditions: (1) an increasing difficulty condition in which easy learning tasks were followed by difficult tasks ; (2) a decreasing difficulty condition in which difficult learning tasks were presented before easy tasks. The learning materials and tasks were delivered by an ITS called AutoTutor. The students' learning behaviors were tracked during their interactions with AutoTutor. The students' reading skills and memory were assessed by a reading comprehension test and a recognition test, respectively. Results indicated that students with lower reading skills scored higher on recognition test in the increasing difficulty condition, whereas students with higher reading skills memorized more information in the decreasing difficulty condition. Learning and engagement during the interaction with AutoTutor were not found to be significantly different between the two task difficulty conditions. Implications for improving the adaptivity of ITSs were discussed.