Etiology of user experience with natural language speech
Christopher J. Pavlovski, Jennifer Lai, Stella Mitchell · 2004
Abstract This paper outlines the results of commercial field trial of a multimodal (speech and GUI) platform using natural language technologies in a mobile environment. During the trial we compared the user experience of the multimodal interface to the user experience with a comparable application providing only unimodal (GUI) interaction. A key objective of the analysis following the trial was to understand the etiology of the difference in user experience with the two systems. Speech-only usage was used to complete tasks 71% of the time on average across all participants and tasks. The remaining 29% was distributed among GUI-only usage and multimodal usage. These results indicate that for input, the natural language speech was the major contributor to the increased quality of user experience. 1. Introduction Accessing email from a telephone is an application that has been available in various forms for many years. The first systems gave users speech-only access over a standard telephone. These were later adapted for use on cell phones by incorporating new acoustic models. Once WAP phones became more widely available more unimodal systems emerged, these were largely GUI only, for access to email when away from a traditional computer. With the arrival of third generation networks (3G) we now have the ability to create multimodal solutions (voice and GUI) to access information in a mobile environment. This paper describes one such solution, as well as the successful field trial associated with the solution. We believe this implementation to be first instance of a natural language understanding multimodal application on an unmodified cell phone.