The Representation of Command Language Syntax

Abstract

Command languages provide a powerful user interface for many applications, although this power is not without a price. Such languages are usually complex, making learning of the language difficult and error-prone. For this reason, a simple description of any command language is crucial to its success. Two popular methods of representing the syntax of command languages were compared in the experiments reported here. One method uses special meta-characters to represent grammatical relations, whereas a graphic method uses boxes and arrows much like a flowchart. In Experiment 1, the graphic method was learned faster and with fewer errors. Experiment 2 compared the two methods in a reference task as contrasted to a learning task, and the graphic method was again found superior. Experiment 3 demonstrated that the superiority of the graphic method over text obtains whether the rules are simple or complex.