While to date, multimedia research has examined mainly the learning of texts with accompanying pictures, in the current paper, 2 experiments are presented that examine the multimedia effect for pictures with accompanying spoken text. In Experiment 1, we examined whether learning is better with a multimedia presentation in which pictorial information is verbally referenced than without such referencing. Further, it was examined whether pictorial information within a single presentation is better learned when it is verbally referenced than not referenced. The results show that the pictures with accompanying audio text in which the single elements of the picture were named were better learned (free recall, multiple choice, visual recognition) than the pictures with the elements not having been named in the audio text. Furthermore, within a single presentation, named elements were better learned than unnamed elements. Further, Experiment 2 examined by eye-tracking whether the multimedia effect is due to a shift of attention toward the elements presented multimodally and away from those presented unimodally. The multimedia effect could be replicated and the postulated shift of attention as an underlying process of the multimedia effect could also be confirmed. There were longer fixation times for the named and shorter fixations times for the unnamed elements of the picture in the verbal referencing part compared to the nonverbal referencing parts of the audio text. Finally, gaze synchrony of the learners was higher for time points of naming pictorial elements than for time points of no naming.