This special issue of the IJMMS grew out of the concurrent special issue of Computing Surveys (March, 1981) on "The Psyhcology of the Computer User", which reviews the empirical work published up to 1980. This issue presents more recent work, both analytic and empirical, addressing some of the higher-level cognitive issues in the psychology of human-computer interaction.
The papers in this issue fit into a theme suggested by the title of the special issue--that the interaction between a human user and a computer is a form of linguistic dialogue, conforming to a surface syntax and having an underlying semantics. The dialogue is clearly not symmetric; it is constrained almost entirely by the structure of the computer system--to which the user must adapt. The syntax is constrained by the conventions of the computer's interface language, and the semantics are determined by the way the computer interprets the statements from the user.
The first paper, by Thomas Moran,t proposes a scheme, called the Command Language Grammar (CLG), for representing human--computer dialogue. The fundamental idea of CLG is that the dialogue may be completely described at several different levels. Some of the levels describe the task structure and the conceptual structure of the computer system, which serves as the semantic basis of the dialogue; other levels describe the syntactic conventions of the dialogue, including the mapping from the syntactic elements to the conceptual semantics. CLG can be viewed in many ways: (1) as a linguistic grammar, generating the space of command language systems, (2) as a design tool, forcing the system designer to articulate the conceptual model of the system and to systematically relate the syntax of the user interface to the model, and (3) as a psychological hypothesis, describing the knowledge the user has about a system.
The paper by Richard Young is a search for a theoretical understanding of the notion of a conceptual model. The exploration is conducted in the domain of pocket calculators, where a variety of models are to be found. The notion of a conceptual model as a simple machine (e.g. a set of registers and operations) is straightforward, although, as Young demonstrates, it can be quite difficult to induce such a model from a calculator's behavior. Such models can help the user predict the behavior of the calculator, but they do not seem to help in accomplishing routine calculation tasks. An analysis of the mappings of tasks onto the structure of the calculator shows that the calculator actually imposes a structure on the task domain. Young suggests that a more useful notion of a conceptual model is to be found in these mappings.