Children can represent events in our everyday life in both non-linguistic and linguistic formats. We aimed to investigate whether non-linguistic representations are changed once children acquire their linguistic counterparts. In the present study, we explored whether and how language changes the perception of simple means-end actions using an eye-tracking paradigm. Children between 12 and 24 months of age heard a sentence containing a verb and subsequently watched an action video. Results show an interfering influence of language on action perception at 12 months and a facilitating influence at 24 months. However, this was only the case for verbs that are already in the toddlers' productive vocabulary but not for those that are acquired later. Taken together, the results suggest that a communication between non-linguistic and linguistic representations starts early and develops in the second year of life. The successful facilitatory influence depends on the productive repertoire of the language in question.