The current study investigates how speakers are affected by thesize of the visual domain during reference production. Previousresearch found that speech onset times increase along with thenumber of distractors that are visible, at least when speakersrefer to non-salient target objects in simplified visual domains.This suggests that in the case of more distractors, speakers needmore time to perform an object-by-object scan of all distractorsthat are visible. We present the results of a reference productionexperiment, to study if this pattern for speech onset times holdsfor photo-realistic scenes, and to test if the suggested viewingstrategy is reflected directly in speakers’ eye movements. Ourresults show that this is indeed the case: we find (1) that speechonset times increase linearly as more distractors are present; (2)that speakers fixate the target relatively less often in larger do-mains; and (3) that larger domains elicit more fixation switchesback and forth between the target and its distractors.