This paper examines the hypothesis that listeners are sensitive to sound symbolism in that they are able to quite accurately guess the meaning of unknown foreign words on the basis of the sound symbolic properties of the stimuli. Participants, all of them being native English speakers, participated in behavioral and fMRI experiments, which showed that indeed subjects did better at inferring the meaning of sound symbolic words compared to non-sound symbolic stimuli. This evidence suggests that there exist crossmodal (i.e., sound-to-meaning) correspondences across languages, thus underlining semantic processing.