Drawing on Mayer's (2014) cognitive theory of multimedia learning, the purpose of this study was to examine which modes of gloss presentation (i.e., L2 definition, aural, and video animation) are effective for learners' vocabulary learning and delayed word recollection. One control group and three experimental groups were formed by 132 intermediate language learners. Pre- and post-tests of productive recall and multiple-choice productive recognition were administered, and learners' perceptions toward glossing were examined through a questionnaire and interviews. Quantitative data were analyzed using ANCOVA, and themes that emerged from the qualitative data were identified. The quantitative findings revealed that dual glossing modes were more effective than single glossing modes for many test sessions. However, single glossing was also effective for a few test sessions. The questionnaire and interview data showed that learners preferred the dual glossing mode of L2 definition and video animation. The findings provide insights for vocabulary learning and teaching.