Knowledge organization systems need tags to centralize and provide the keywords that can be employed in classifying, sharing and seeking knowledge for personal or organizational use. However, the increased variety of vocabularies and languages causes the connections between textual tags and the documents marked by them to become less and less distinctive. The purpose of this paper is to present an exploratory study on new type of knowledge tags: structured iconic tags. This kind of tag is created based on the special graphical rule – Visual Distinctive Language (VDL) – and has particular focus on the representation of tag structure. A ‘tagging on paper’ experiment was conducted to compare tagging results among three types of tags: textual tags, iconic tags without explicit structure and VDL-based iconic tags. Experimental results demonstrated that the new type of iconic tags improves the representation of tag structure as well as the interpretation of tag meaning. This work is one of the first to investigate how icons would be applied for tagging and what kind of graphical code would better represent knowledge organization. Findings in this paper could enrich the tagging method, which in turn provides a possible way to develop a visual knowledge organization system and iconic tag clouds for digital library.