Abstract. People design what they say specifically for their conversational partners, and they adapt to their partners over the course of a conversation. A comparison of keyboard conversations involving a simulated computer partner (as in a natural language interface) with those involving a human partner (as in teleconferencing) yielded striking differences and some equally striking similarities. For instance, there were significantly fewer acknowl-edgments in human/computer dialogue than in human/human. However, regardless of the conversational partner, people expected connectedness across conversational turns. In ad-dition, the style of a partner's response shaped what people subsequently typed. These results suggest some issues that need to be addressed before a natural language computer interface will be able to hold up its end of a conversation. Key words: discourse modeling, human/computer interaction, natural language inter-faces, recipient design. 1. Conversations with Computers Why is it that natural language has yet to become a widely used modality of human/computer interaction? Visionaries seem to have no difficulty imag-

Many modern human-computer interfaces are difficult for people to use. This is often because these interfaces make no significant attempt to communicate with the people who use them. In other words, these interfaces are uncooperative: They do not adapt themselves to their users' needs and they are insensitive to human foibles. Ordinary command line interfaces such as that of the UNIX C shell (csh) are intolerant of even the most simple input errors, even when those errors have obvious corrections. An "intelligent" UNIX shell interface, on the other hand, would make use of knowledge and interaction context in order to interpret --- and as necessary, correct --- its users' commands. Valet is a prototype of such an "intelligent" interface to the UNIX C shell. Valet adds knowledge-based parsing and input correction to the shell by encapsulating an ordinary C shell process within a framework that allows Valet to control the shell's input and output. Valet intercepts shell commands and par...

: Transparency has been seen as a significant aspect of successful humancomputer interaction. In this paper we investigate this concept from the point of view of activity theory. We show that transparency cannot be understood as a static feature of the interface, but that the crucial point in achieving transparent interaction is the ongoing development of unconscious operations, embedded in the process of use. We suggest that the process of deliberate formation and refinement of operations during the course of interaction, is supported by setting conditions for the creation of a zone of proximal development in the interface. 1 Introduction Better interfaces is a goal shared by most people in the HCI community, but it is harder to agree on what constitutes a good interface; and to what extent a scientific foundation for design is necessary. In the visionary contribution to a scientific foundation of HCI by Card, Moran, & Newell (1983; Newell & Card 1985), the criteria for good ...

Much of the knowledge required to use modern computing systems takes the form of mappings or associations. These occur between user goals and the functions that accomplish corresponding goals, between functions and the user actions that activate a desired function, and between a menu item or a button label and the function associated with that item or label. The question we explore in this paper is when is it worthwhile, if ever, for a user to pay the price of ]earning a set of new, regular task-action mappings that is inconsistent with the original set of task-action mappings of a previous version of a system. We consider three factors that determine the ease of learning and retention of task-action mappings: mnemonics, regularity within a set of mappings, and consistency of mapping across different system contexts. In two experiments, we found that IrregularNon -Mnemonic mappings take much longer to master than Regular-Mnemonic mappings and that Irregular-Non-Mnemonic mappings are more rapidly forgotten and subject to interference effects due to inconsistency. Regular-NonMnemonic mappings fall in-between the two groups: 1) they are easier to learn than Irregular-Non-Mnemonic but harder than Regular-Mnemonic mappings and 2) more items are retained for them than the Irregular-Non-Mnemonic mappings but fewer than for the Regular-Mnemonic mappings. We conclude that learning a new set Regular-Mnemonic task-action mappings is worthwhile even at the price of having to unlearn a well-learned set of old task-action mappings. Mnemonics, regularity, & consistency Memory For Command Sets: Mnemonics, Regularity, And Consistency 1.

This paper describes the design and implementation of a set of tools that make the task of providing a graphical UNIX user interface easier. The main tool, called Perly, provides on-screen graphical buttons connected to UNIX command scripts. When a button is pressed, the associated command script is executed. The buttons are highly configurable, encouraging users to tailor and extend their workstation environment. The idea of onscreen buttons with underlying programmable scripts comes from the work done at Xerox EuroPARC into user-tailorable systems. Descriptions of Perly buttons are held in plain text files, allowing them to be given to other users via electronic mail. Examples of the use of buttons and the ideas behind the implementation of Perly buttons under the X Window System are discussed