Purpose: Semantic learning under 2 co-speech gesture conditions was investigated in children with specific language impairment (SLI) and typically developing (TD) children. Learning was analyzed between conditions. Method: Twenty children with SLI (aged 4 years), 20 TD children matched for age, and 20 TD children matched for language scores were taught rare nouns and verbs. Children heard the target words while seeing either iconic gestures illustrating a property of the referent or a control gesture focusing children's attention on the word. Following training, children were asked to define the words' meaning. Responses were coded for semantic information provided on each word. Results: Performance of the SLI and age-matched groups proved superior to that of the language-matched group. Overall, children defined more words taught with iconic gestures than words taught with attention-getting gestures. However, only children with SLI, but not TD children, provided more semantic information on each word taught with iconic gestures. Performance did not differ in terms of word class. Conclusions: Results suggest that iconic co-speech gestures help both children with and without SLI learn new words but, in particular, assist children with SLI understand and reflect the words' meaning.