This study investigated the effects of three instructional modes: picture-book reading-only (PRO), picture-book reading plus vocabulary instruction (PRVI), and picture-book reading plus reading-based collaborative output activity (PRCOA) on young adult EFL (English as a foreign language) learners' vocabulary acquisition and retention. Eighty Taiwanese university students with low to intermediate level English proficiency from three English reading classes participated in each of the three modes once during three weekly 100-minute sessions. Vocabulary knowledge was tested through two post-tests using a modified Vocabulary Knowledge Scale: immediately after each instructional mode to measure students' vocabulary acquisition, and one month later to measure their word retention. The results showed that the PRVI mode was the most helpful for immediate word learning. However, the PRCOA mode was the most effective for word retention, and most conducive to bettering students' productive knowledge in both acquisition and retention. This study suggests that explicit learning from vocabulary instruction that directs students' attention to the words to be learned did not guarantee greater vocabulary gains than incidental learning where new words can be learned as by-products of classroom collaborative output activities. Without trying to memorize words, students learned vocabulary through mental investment in group discussions and generative activities, leading to their mastery of productive word knowledge.