Recognition memory can be improved by free choice, even with arbitrary free choices without impact on the learning content itself. This is promising from an applied perspective, but given different forms of tests in educational contexts, generalization of this effect to recall memory is important. Therefore, we used a vocabulary-like learning task with either a free or forced choice preceding a Japanese-German word pair. More precisely, prior to the presentation of each word pair, participants were either instructed (forced choice) or allowed to choose (free choice) whether the next German word would be from the first or second half of the alphabet. The subsequent test phase comprised a recognition test for the Japanese words including additional lures, followed by a cued recall test for the German translation. Recognition memory was measured on a 6-point confidence scale, and receiver operating characteristic curves were estimated based on the unequal variances signal detection model of recognition memory. Consistent with previous results, free choice before learning improved recognition memory (d_free = 1.23, d_forced = 1.08). More importantly, participants also showed a free-choice benefit on cued recall performance with a 5% increase in correct recall, suggesting generalizability from recognition to recall memory. Data are currently being collected for a conceptual replication using Japanese symbols as memory items. Applied to an educational setting, establishing a self-directed learning situation with arbitrary control by students, which does not influence the predetermined learning content, seems like an easily applicable tool for teachers to improve memory performance of their students.