In adults, words are more effective than sounds at activating conceptual representations. We aimed to replicate these findings and extend them to infants. In a series of experiments using an eye tracker object recognition task, suitable for both adults and infants, participants heard either a word (e.g. cow) or an associated sound (e.g. mooing) followed by an image illustrating a target (e.g. cow) and a distracter (e.g. telephone). The results showed that adults reacted faster when the visual object matched the auditory stimulus and even faster in the word relative to the associated sound condition. Infants, however, did not show a similar pattern of eye-movements: only eighteen-month-olds, but not 9- or 12-month-olds, were equally fast at recognizing the target object in both conditions. Looking times, however, were longer for associated sounds, suggesting that processing sounds elicits greater allocation of attention. Our findings suggest that the advantage of words over associated sounds in activating conceptual representations emerges at a later stage during language development.