In two studies, we examined what constitutes a supportive context for word learning. In Study 1, we examined children’s comprehension of iconic gestures by asking 2, 3, and 4 year-old children, as well as adult controls, to choose the referent of an iconic gesture. The gesture either highlighted a possible action with the object, or a salient physical attribute of the object. By age 2, children performed above chance for iconics of action, but not for iconics of attribute; indicating that early use of iconic gesture to support word learning should utilize iconics of action. In Study 2, we examined how different levels of contextual support affect word learning from written contexts. Participants read very rare English words in contexts that were either: high (H), medium (M), or low (L) constraint. Participants had the greatest accuracy on a synonym judgment task when words were trained in a scaffolded sequence (H-M-M-L).