Electronic Theses and Dissertations


Ying Fang



Date of Award


Document Type


Degree Name

Doctor of Philosophy



Committee Member

Philip Pavlik Jr.

Committee Member

Su Chen


According to Vygotskys zone of proximal development (ZPD), students benefit from tasks that are difficult but can be achieved with the guidance of a skilled partner. This concept has often been implemented during the development of intelligent tutoring systems (ITSs), but there remains a need to empirically investigate the validity of this concept in reading comprehension interventions. Individual differences need to be considered in order to assign tasks to individual students with the right level of difficulty. Mixed results have emerged from previous research that specifically investigated the potential interaction between student attributes and ITSs on learning in the domain of reading comprehension. The present dissertation explored how and to what extent individual differences interact with task difficulty when assessing performance during learning, memory, 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 order condition in which easy learning tasks were followed by difficult tasks versus (2) a decreasing difficulty order condition in which difficult learning tasks were presented before easy tasks. The students learning performance (i.e., accuracy and time to answer questions) was tracked during their interactions with AutoTutor, a conversation-based ITS. The students reading skills were assessed by a reading comprehension test whereas memory was assessed by a recognition test on content delivered by AutoTutor. Results indicated that students with lower reading skills had better memory for surface structure (i.e., wording and syntax) in the increasing difficulty order condition, whereas students with higher reading skills had better memory for surface structure in the decreasing difficulty order condition. Learning performance during AutoTutor was not significantly different between the two task difficulty order conditions, but engagement as indicated by reading time was found to affect high-skill and low-skill students memory differently in the increasing and decreasing task difficulty order conditions. Implications for improving the adaptivity of ITSs are discussed.


Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest