In this study, we first provide new behavioral data on dyslexic readers and describe their eye movement behavior during orthographic learning. Results suggest that: 1/ dyslexic readers solve the speed/accuracy trade-off differently, with longer RTs to reach similar performance level, because of their limits on visual processing, and, 2/ VAS affected visual processing of PWs during learning and performance in their further recognition. We then used BRAID-Learn to simulate the evolution of eye movement behavior during orthographic learning of PWs. Simulation results obtained are consistent with behavioral data suggesting a link between sdA and the VAS behavioral measure. Monitoring eye movements during novel word repeated reading provides online information of orthographic learning (Nation et al., 2017). Dyslexic individuals show orthographic learning difficulties (Martens & de Jong, 2008), but no study examined the evolution of their eye movement patterns during learning. Furthermore, few behavioral data explore the role of visual attention on orthographic learning (Marinelli et al., 2020). BRAID-Learn (a computational model of orthographic learning) predicts that visual attention affects orthographic learning (Ginestet et al., 2022).