Visual word recognition is perceived to remain relatively stable throughout adulthood, but recent research suggests the system involved is malleable, with evidence of behavioural changes after lexical decision task (LDT) practice. The potential for, and extent of, neural changes have yet to be elucidated in this context. If identified, these neural changes could be due to processes associated with learning, where performance that is initially effortful becomes efficient and supported by an optimized task network. Replicating the British Lexicon Project, participants completed 16 hours of LDT learning over several days. We recorded EEG at three intervals to track neural change during LDT learning and assessed event-related potentials and brain signal complexity. We found that response times decreased during LDT learning, and there was evidence of neural change through N170, P200, N400, and LPC amplitudes across the EEG sessions, suggesting alterations to both the general cognitive and specific lexical processes involved in LDT performance. We also found widespread complexity decreases alongside localized increases, suggesting that processing became more automatic with specific increases in processing flexibility. These findings suggest that the visual word recognition system is dynamic, and has the potential for plastic changes to support more efficient and automatic task performance.