While some argue that emojis are processed like words, opponents note dissimilarities. We used a divided attention (DA) technique to examine whether memory for emojis, relative to words, engages primarily verbal or visuo-spatial cognitive representations. We compared the decline in memory output experienced when participants freely recalled a list of studied words or emojis under dual-task conditions with a concurrently performed distracting task. Participants encoded either words or emojis (between-subjects) under full attention (FA), and later recalled them under FA or while concurrently performing a 1-back task to words (DA Words), emojis (DA Emojis), or stars (DA Stars), manipulated within-subject. Emoji memory was higher overall compared to words. Word recall was unaffected by the DA Stars condition, but significantly worse in the DA Words-and to a lesser degree DA Emojis-condition, relative to FA. Results suggest that memory for words relies primarily on reactivation of verbal representations, which is hampered when the distracting task also requires verbal, but not visuo-spatial, processing. In contrast, small but significant declines in recall were observed for emojis across all DA conditions relative to FA. Results suggest that unlike words, representing and retrieving emojis engages both verbal and visuo-spatial processing.