In second language acquisition, a popular method of introducing new vocabulary is by embedding the words in a natural text. Supplementary information (e.g., definitions, illustrations, synonyms, etc.), or glosses, can be included in the margins of the texts to highlight and improve retention of the new words. Previous studies suggest multimodal glosses facilitate learning, although the question of which glosses are most effective remains an active research topic. In the present study, we focused on a sample of university students studying Arabic as a second language. In two experiments, we found that (a) recognition and recall of target vocabulary words were superior when target words were accompanied by a combination of text and picture glosses compared to text-only or picture-only; and (b) including information about the Arabic root word from which the target word was derived in the glosses provided an additional memory benefit. Overall, this study adds further support for glosses as a teaching tool by generalizing to the case of Arabic. Our results also show how glosses that highlight root information can be useful for improving recognition and recall of Arabic vocabulary specifically.