According to previous research, speakers of European languages parse regularly-inflected, morphologically-complex words into stems and grammatical affixes during word recognition. In contrast, some studies suggest that late second language (L2) learners do not. We ask how these types of words are processed in Arabic, a language whose primary morphological process is infixation, which exists alongside prefix- and suffixation. Specifically, we ask how natives, heritage, and late L2 learners process regular gender and number agreement morphology on adjectives. Results of a masked priming experiment suggest that all three groups parse adjectives into stems and affixes depending on agreement type, suggesting that learners store and process morphologically-complex words like native speakers, even when the first and second language have very different morphological systems.