Smart reading aid for detecting problems with reading fluency and comprehension

Abstract

Brain signal and eye tracking technology have been intensively applied in cognitive science in order to study reading, listening and learning processes. Though promising results have been found in laboratory experiments, there are no smart reading aids that are capable to estimate difficulty during normal reading. This paper presents a new concept that aims to tackle this challenge. Based on a literature study and an experiment, we have identified several indicators for characterizing word processing difficulty by interpreting electroencelography (EEG) and electrooculography (EOG) signals. We have defined a computational model based on fuzzy set theory, which estimates the probability of word processing and comprehension difficulty during normal reading. The paper also presents a concept and functional prototype of a smart reading aid, which is used to demonstrate the feasibility of our solution. The results of our research proves that it is possible to implement a smart reading aid that is capable to detect reading difficulty in real time. We show that the most reliable indicators are related to eye movement (i.e. fixation and regression), while brain signals are less dependable sources for indicating word processing difficulty during continuous reading.