Arthur Ward

Thank you for visiting. I am currently a post-doc working under the supervision of Dr. Rebecca Crowley.

We are currently pursuing a project to investigate the potential of Intelligent Tutoring Systems to improve reading fluency among children with dyslexia. A major obstacle to fluency in this population is often low exposure to connected text, relative to unimpaired readers. Our experimental results so far suggest that an interactive text will increase motivation to read, and actual voluntary reading, relative to static text.

We are now developing methods to model individual students' decoding ability. The resulting model of decoding skill will then be used when generating individualized, interactive texts for the repeated reading drill.

Previous work focused on the problems of predicting learning from interactive tutorial dialog, and using the insights gained to increase the effectiveness of tutoring. My general approach has been to adapt theory from the study of human text comprehension. For example, I have extended models of memory during reading to predict learning from tutorial dialog. Similarly, I extended a measure of cohesion in text to measuring cohesion in dialog, and found that certain types of tutorial dialog cohesion were correlated with learning. I have also adapted measures of priming and convergence in spoken dialog and used them to predict learning from interactive tutoring dialog. Links to pre-publication final drafts of papers describing this work are given below.

In my experimental work, I have also shown that the cohesiveness of a reflective text given after tutoring can affect learning from the tutoring session, and that this effect interacts with student domain knowledge and motivation level. This work extends prior results in text comprehension to reflective text, and shows for the first time their impact on tutoring. Several papers describing this work are currently in preparation, or under review.