Is it a Match? Examining the Fit between Conversational Interaction Modalities and Task Characteristics
Rzepka Christine, Berger Benedikt, Hess Thomas
Owing to technological advancements in artificial intelligence, specifically natural language processing, voice assistants (VAs) offer a new modality for interacting with computers. Compared to formalized and deliberate text-based interaction, speech is more natural and intuitive. As companies offer customers the possibility of communicating with them via VAs, it is important to determine the kind of tasks for which this interaction modality is beneficial. Drawing on cognitive fit and task-technology fit theory, we present a research model for examining the fit between speech- and text-based interaction modalities and task characteristics. To test this model, we propose a mixed design laboratory experiment with interaction modality serving as between-subject factor and task type serving as within-subject factor. For this purpose, we developed a VA using DialogFlow and trained it in two pre-tests. The results of the experiment will extend theory on cognitive fit and provide practical insight regarding the applicability of speech.
voice assistant; conversational agent; speech interaction; cognitive fit