The present study adds to current knowledge in three areas: (a) the relationship between learner proficiency and gloss language type, (b) the retention of learned words, and (c) learners’ reactions to glossing. This study examined the relationship between gloss types (L1 and L2) and learner proficiency. L2 glosses were more effective for learners whose level was high-intermediate or above. In the case of learners at low-intermediate to intermediate level, L1 and L2 glosses did not yield different results. Regarding the retention of learned words, learners’ mean scores in the present study dropped significantly on the delayed vocabulary test. They forgot some of the learned words after 4 weeks. The findings of Yoshii's (2006) study and the present study are not comparable due to differences in test methods, learner proficiency, and time between the immediate and delayed tests. In addition to statistical data, information on learners’ reactions to glossing enabled me to more precisely describe their perspectives on and behavior regarding glossing. These data were useful in explaining in detail vocabulary acquisition as a result of glossing.