Language is a powerful vehicle for expressing emotions, although the process by which words acquire their emotional meaning remains poorly understood. This study investigates how words acquire emotional meanings using two types of associative contexts: faces and sentences. To this end, participants were exposed to pseudowords repeatedly paired either with faces or with sentences expressing the emotions of disgust, sadness, or neutrality. We examined participants’ acquisition of meanings by testing them in both within-modality (e.g., learning pseudowords with faces and testing them with a new set of faces with the target expressions) and cross-modality generalization tasks (e.g. learning pseudowords with faces and testing them with sentences). Results in the generalization tests showed that the participants in the Face Group acquired disgust and neutral meanings better than participants in the Sentence Group. In turn, participants in the Sentence Group acquired the meaning of sadness better than their counterparts in the Face Group, but this advantage was only manifested in the cross-modality test with faces. We conclude that both contexts are effective for acquiring the emotional meaning of words, although acquisition with faces is more versatile or generalizable.