From an early age, children apply the mutual exclusivity (ME) assumption, demonstrating preference for one-to-one mappings between words and their referents. However, for the acquisition of referentially overlapping terms, ME use must be suspended. We test whether contextual cues to intended meaning, in the form of presence of a speaker, may be critical for flexible ME application. Four- to five-year-old children were tested on two word learning tasks requiring flexible use of ME, respectively. In Experiment 1, children saw video recordings of the speakers introducing the novel labels. All children successfully applied ME and succeeded in accepting lexical overlap. In Experiment 2, with audio recordings of speakers only, children were unsuccessful at accepting lexical overlap. Thus, flexible use of ME relies on a developing ability to utilise the contextual information present in communicative interactions.