||When young children hear a new word, they often display the mutual exclusivity (ME) bias, behaving as though every referent only has one label and applying new labels to unfamiliar (previously unnamed) objects. Two theories that discuss the source of the ME bias are social-pragmatic theory, which proposes that children's word learning is guided by their interpretations of speakers' communicative intent, and resource conservation theory, which suggests that word learning is guided by general cognitive processes. We sought evidence that both causes contribute to the ME bias. Three-year-olds saw pairs of objects (one familiar and one unfamiliar), and were asked to find the referent of a novel label. Half received a pragmatic cue to the speaker's intent, and half received no cue. Children received feedback and training with an additional item pair until they responded correctly to the novel label. This process was repeated with four novel labels. When initially asked to find the referent of the novel label, children responded correctly more often if given a pragmatic cue, but they also showed some evidence of the ME bias. During the final test, children were very good at identifying which objects had been associated with a new label in the initial phase, but they had difficulty matching the specific labels with the correct referents. Given the difficulty of learning four new labels, we expected evidence of resource conservation in the final test, with better performance when novel labels had been associated with unfamiliar items during training. However, performance was not affected by target item familiarity in the final test.