Reading fluency is often predicted by rapid automatized naming (RAN) speed, which as the name implies, measures the automaticity with which familiar stimuli (e.g., letters) can be retrieved and named. Readers with dyslexia are considered to have less "automatized" access to lexical information, reflected in longer RAN times compared with nondyslexic readers. We combined the RAN task with a Stroop-switch manipulation to test the automaticity of dyslexic and nondyslexic readers' lexical access directly within a fluency task. Participants named letters in 10 × 4 arrays while eye movements and speech responses were recorded. Upon fixation, specific letter font colors changed from black to a different color, whereupon the participant was required to rapidly switch from naming the letter to naming the letter color. We could therefore measure reading group differences on "automatic" lexical processing, insofar as it was task-irrelevant. Readers with dyslexia showed obligatory lexical processing and a timeline for recognition that was overall similar to typical readers, but a delay emerged in the output (naming) phase. Further delay was caused by visual--orthographic competition between neighboring stimuli. Our findings outline the specific processes involved when researchers speak of "impaired automaticity" in dyslexic readers' fluency, and are discussed in the context of the broader literature in this field.