Twenty-four caregivers and their two- to four-year-old children took part in a storybook reading task in which caregivers taught children novel labels (‘DAXY’) for familiar objects. One group (N=12) received labels modelled syntactically as proper names (‘This is NAMED DAXY’), and another group (N=12) received the same labels for the same objects modelled syntactically as adjectives (‘This is VERY DAXY’). Caregivers took strikingly different approaches to teaching words from the two lexical categories. In teaching proper names, but not adjectives, caregivers flagged cases in which one word was paired with two objects; two words were paired with one object; and one word was paired with an inanimate object. In teaching adjectives, but not proper names, caregivers discussed meaning and offered translations. Caregivers' distinctive strategies for teaching proper names and adjectives are congruent with recent findings about children's word meaning assumptions, and with analyses of the semantics of these lexical categories. The findings indicate that parental speech could provide a rich source of information to children in learning how different lexical categories are expressed in their native language.