This study investigates claims that metaphor acquisition is rooted in the words’ concrete meanings: to be able to use metaphors and other non-literal languages, the child needs to ‘go beyond’ meanings that are conventional and so presumably concrete (e.g., Falkum, 2022: 97). To test if metaphor-related words emerge via their concrete senses and how this reflects child-directed speech, I examined 594 hours of interactional data for three English-speaking toddlers from urban middle-class England, whose speech was densely sampled between the ages of 2;00 and 3;01. The data show that 75%–82% conventional metaphors were acquired via their concrete senses and that the order of acquisition of concrete and abstract senses corresponded with their input frequencies. Overall, when hearing conventional metaphors, 81%–89% of the time children were exposed to their concrete meanings. Contrary to the generic argument that children’s pragmatic reasoning with non-literal uses is impeded by meaning conventionality (Falkum, 2022), my preliminary data suggest that it is influenced by the frequency of exposure to the concrete meanings of conventional metaphors, which leads to a generalised prediction that the most probable interpretation of any new metaphor is concrete (literal). Qualitative analyses further reveal that abstract meanings, when acquired first, were learned in highly emotive contexts.