Using a novel emotional perspective-taking task, this study investigated 4-year-olds' (n = 97) use of a speaker's emotional prosody to make inferences about the speaker's emotional state and, correspondingly, their communicative intent. Eye gaze measures indicated preschoolers used emotional perspective inferences to guide their real-time interpretation of ambiguous statements. However, these sensitivities were less apparent in overt responses, suggesting preschoolers' ability to integrate emotional perspective cues is at an emergent state. Perspective taking during online language processing was positively correlated with receptive vocabulary and an offline measure of emotional perspective taking, but not with cognitive perspective taking, conflict or delay inhibitory control, or working memory. Together, the results underscore how children's emerging communicative competence involves different kinds of perspective inferences with distinct cognitive underpinnings.