In this project, we successfully applied iterative prototyping and participatory design to the design of a complex system --- a voice dialog design environment. The success case for this project is really a story about how we got "this far this fast." In less than four months, a mixed team of professional voice dialog designers and academic researchers were able to design and build a substantial core design environment. There were several unique aspects of our approach that contributed to the project's success. These include selection of the right high-level system substrate on which to build our design environment and creation of a participatory context conducive to iterative design methods.

Many of the principles that guide user-interface design for commercial systems do not scale down to simple applications developed on personal computers. These “very small systems” are typicaly designed within a high-level application such as a database or a spreadsheet. The entire development process may take no more than a few days. In this restricted context, iterative design and usability testing are unaffordable luxuries, while detailed task analysis and early focus on users fail because the task and users will not coalesce until the system is in place. We describe our experiences with developing and using a very small sytem. We present suggestions for successful design in similar situations.

The cognitive walkthrough is a new method for assessing the usability of a system and assigning causes to usability problems early in the design process. Early versions of the walkthrough suffered from tedious and repetitious paperwork. To address this problem we developed an automated prompting system, which reduces the method’s overhead and focuses the analyst’s attention on critical issues. The methodology can be introduced in about an hour, and the automated system allows a user action in a proposed interface to be analyzed in as little as 5 to 10 minutes.

As technology changes, it has become common to develop now generations of existing software to reflect these advances. The focus of this research is how changes in the user/computer dialogue structure will affect the performance of users who are familiar with an earlier version of the product. In the case of menu-based systems, changes in new releases frequently involve modification of the underlying menu structure (e.g., IBM DisplayWrite series, DBaseIII -DBaseIII+). We investigated whether changing the semantics and structure of a popular menu-driven word-processor would permit transfer of existing knowledge of the word-processor to the new version.

This paper investigates whether changes in the user/computer dialogue structure will affect the performance of users who are familiar with an earlier version of the product. Quantitative predictions using the Kieras and Polson (1985) production system model were derived to test whether changing the lexical attributes and structure of a popular menu-driven word-processor would permit transfer of existing knowledge of the word-processor to a new version. The results show that changes to the dialogue structure of the menu-system are not detrimental, while changes to the lexical attributes of the menus will hinder user performance.