Multiple representation document development

The authors review a large number of document development systems for both text and graphics from the perspectives of source-language and direct-manipulation models. They describe the task domain and discuss the pros and cons of direct-manipulation techniques versus a programming-language source code and of a procedural versus declarative schemes. They then establish a framework for analyzing and designing multiple-representation systems. The central theme is that program constructs and visual feedback are complementary to each other and that a hybrid approach would be most desirable.<>