Eye-tracking Analysis of Reading in People with Aphasia

Abstract

Reading impairments (acquired alexia) commonly co-occur in people with aphasia (PWA). The ultimate goal of reading treatment is to read connected text accurately and quickly. Traditional treatments of acquired alexia have typically targeted single word reading, which has seen little generalization to text reading. Text treatments, though relatively rare, have demonstrated an increase in oral reading rate and accuracy. Investigators have proposed that text is facilitative for PWA as it allows them to use “top-down” processing to integrate syntactic and contextual information (context effect). However, it is unclear exactly what linguistic information PWA are utilizing and how it is being used. A potentially useful approach is to utilize eye-tracking methodology to investigate reading behaviour. In this thesis, the impact of a central language impairment on eye-movements during reading is explored. This investigation occurred in two parts. First, a systematic review of reading studies utilizing eye-tracking methodology was performed to describe the eye-movements of PWA relative to healthy readers. Six studies, including observational and experimental studies, with a total of 60 PWA were found. Compared to healthy readers, PWA had longer processing times, were less likely to skip words, and were more likely to reread previous parts of text. Furthermore, these studies revealed that similar to healthy readers, the reading of PWA can be mediated by bottom-up factors (i.e., lexical variables including length and frequency) and top-down factors (i.e., syntactic and contextual variables including sentence complexity and predictability). Second, a retroanalysis of Kim & Bolger (2012) was performed using linear mixed effects models to analyze the effects of context (a top-down factor) and word frequency (a bottom-up factor) on text reading. The eye-movements of participants were tracked as they read target words of varying frequency (High, Low) embedded in sentences with varying predictability (High, Low). Models were built for the response variables of total fixation duration, total number of fixations, and total number of regressions. Group effects were found for all response variables reflecting that PWA had longer processing times than healthy readers. The main effect of context was found in total fixation duration indicating that PWA, similar to healthy readers, utilize context to predict upcoming words. The results of both the retroanalysis and systematic review are consistent in that they show that PWA are utilizing linguistic information within text to facilitate their reading, and that this facilitation is evident at the level of eye-movements.