The use of input enhancement rests largely on the assumption that increasing the salience of target forms via input enhancement promotes attentional processing, which has been argued to be necessary for second language learning. However, deeper processing (e.g., cognitive effort, engagement with prior knowledge) beyond initial attentional processing is also argued to play an important role in learning. The present study investigated whether one type of deeper processing (cognitive effort) moderated the effectiveness of one type of constructed salience (textual enhancement) using a novel methodology: pupillometry. The results of the study indicated that participants’ cognitive effort during the learning phase (measured by changes in their pupil diameter) predicted accuracy in word form learning for participants receiving textual enhancement, but did not predict performance for participants without textual enhancement. These findings suggest that constructed salience interacts with learners’ cognitive effort in complex ways that need further investigation.