User Centred Design of a Multimodal Reading Training System for Dyslexics

This thesis presents work in the area of computer-based reading training for
dyslexics using speech technology and multimodal information. The study
includes an analysis of typical dyslexic reading behaviour and traditional
training techniques as well as the detailed development of a prototype system
simulating the transfer of a traditional training technique into an automated
solution.
Utilising a user-centred design approach the study focuses on the usability
aspects expected to be tightly coupled to the perceived quality of a final system.
Interaction styles, feedback strategies, assistance schemes, evaluation
forms and graphical layout solutions are investigated. This is done through
the design and evaluation of several prototypes that seek to compensate for
the loss of empathy when going from traditional training to an automated
system. Through evaluations of the final prototype system, involving dyslexics,
it is shown that it is possible for dyslexic users to carry out reading exercises
with the sole assistance and guidance of an automated training tool,
given a sufficient speech recognition accuracy. It is furthermore shown that
in order to cope with the different preferences and requirements of dyslexic
users, a training tool for this purpose needs to be highly customisable and
flexible.
The long term effects of using the automated training tool remain to be
shown. However, by developing an automated training tool in accordance
with the outcome of this study it seems possible to establish a platform from
which the learning process can take place.