The aim of this study is to optimize CALL environments as a learning tool rather than a gloss, focusing on the learning of polysemous words which refer to spatial relationship between objects. A lot of research has already been conducted to examine the efficacy of visual glosses while reading L2 texts and has reported that visual glosses can be effective for incidental vocabulary learning. This study, however, discusses the efficacy of visual aids on vocabulary learning, from the following three different standpoints. The first point is that previous studies have not covered the meaning of these words and have concluded the aids become effective regardless of the part of speech of these words. That is, some words are easy to learn, but the others are difficult depending on the part of speech. Paying more attention to the meaning structures is necessary. The second is that previous studies have focused on visual aids in vocabulary learning using CALL in terms of a gloss while reading texts. As CALL environments have been developed, however, glosses are now used not merely as a reference tool, but as a learning tool. Finally, a lot of research on vocabulary learning with multimedia has been conducted in reading activities, although visual aids can be effective in other activities such as listening activities, in which deeper discourse comprehension is required. Taking these standpoints into consideration, we hypothesize that the intentional learning of those words with multimedia-oriented visual aids could enhance not only comprehension of vocabulary meanings but also better comprehension of a script that includes those words. To examine our hypothesis, we will conduct an experimental study with computer-mediated learning material for English prepositions which we developed for this study. The findings of this study can contribute to the better CALL environments, leading to more effective web or mobile-based learning tools. [For the complete volume, see ED574893.]