Abstract According to the conceptual metaphor theory (CMT; Lakoff & Johnson, 1980), conceptual metaphors, such as “Life is a journey,” are used to map the characteristics of abstract domains (e.g., life) onto more accessible concrete domains (e.g., a journey). Uses such as “dead-end job” and “they’re on the right track” are manifestations of this conceptual metaphor. When and how these mappings might be created remains unclear. Bowdle and Gentner’s (2005) career of metaphor hypothesis suggests that metaphors in “X is a Y” form (e.g., “a lawyer is a shark”) undergo a change in the mode of processing as those metaphors transition from novel conceptual pairings to familiar ones. The current study asks whether the associations between conceptually related concepts are accessed while the metaphoric language is processed in working memory (WM) and, if so, whether the role of WM is moderated by familiarity. If processing linguistic applications of conceptual metaphors involves active access of the concrete source domain, processing metaphoric language, like the concrete language it is being mapped onto, should involve activation of concrete representations and recruit more visual WM resources than literal abstract language. Experiment 1 found no difference in the visual WM demands of processing literal language and conventional expressions of conceptual metaphors. Experiment 2 found that novel expressions of conceptual metaphors placed greater demands on visual WM than familiar expressions. Together, these data provide some support for CMT but suggest that familiarity of the linguistic form of the metaphor plays a role in determining processing modality.