This study investigated the nuanced dynamics of incidental vocabulary learning in L2 instruction, arguing that learning effectiveness may be guided not only by instructional methods but also by learners’ expectations of the methods. To ascertain which group, read-to-comprehend or read-to-produce, excels in incidental development, this study compared two groups of Korean university students, each performing identical reading activities in English reading and writing classes with different objectives. The impact of unknown word quantities on incidental learning was also analyzed. Using the Tracing Measure of Retention (TMR) on customized target words, this study traced the trajectory from immediate learning to retention. Results indicated higher learning gains for the read-to-produce group in both the immediate and delayed posttests, implying deeper processing in the group. Unknown word counts inversely correlated with both immediate and retained learning in the read-to-comprehend group, but not in the readto-produce group. These findings will have implications on L2 instructional design, suggesting ways to optimize learners’ cognitive states for better learning outcomes and the importance of considering learners’ existing vocabulary knowledge during planning. Overall, this study e...