Research has shown that second language (L2) learners that become aware of linguistic features during grammar-based tasks are better able to process these features on a posttest compared to learners that do not focus on these features. However, much L2 input does not come in the form of grammar-based tasks. This study investigates whether learners who become aware of French grammatical gender during a meaning-based task are better able to process these forms than learners whose experience with the same task does not lead to awareness of the feature. Thirty-six Anglophones with low-level French were exposed to reliable noun-ending clues to grammatical gender whilst completing a crossword task. A think-aloud protocol and two probe questions measured awareness. A pre-to posttest design measured accuracy with French nouns ending in eau (e.g. le cadeau). The results revealed no advantage for learners who became aware of the noun-ending clues during exposure: all learners improved in their ability to judge the gender of words encountered during the task (item learning) but none were able to extend this new knowledge to novel items (system learning). The interpretation of the findings considers the choice of linguistic feature, the role of awareness in item learning and the learning conditions that might be necessary for awareness of form to occur during meaning-based exposure.