Computer icons : a picture says a thousand words. Or does it?

A classification system to code different types of icons was developed and used to determine which icons proved difficult for students to read. Using a naturalistic methodology, ten upper primary (elementary) students were asked about their understandings of pictorial, textual (simple single-modality), and combination icons drawn from two CD-ROM programs. Their perceptions were categorized according to the icon classification system over three levels of interaction: a) initial viewing; b) with pop-up label; and c) after interaction. Students' understandings of navigation icons depended on the type of icon and the level of interaction. Pictorial icons were not "transparent" in relation to their function, nor were all the textual icons, and, for some icons, the pop-up labels did not assist the students' perceptions. Thus, it is clear that pictures are not inherently easier to read. It is also clear that some of the assumptions about the perceptibility of pictorial navigation icons and benefits of pop-up labels are probably ill-founded and it is essential that strategies to initiate students into the reading of icons in CD-ROM programs be developed. Software developers also need to take into consideration the types of pictorial icons they employ if they are to increase the accessibility of their products to younger users.