Thinking-aloud protocols may be the most widely used method in usability testing, but the descriptions of this practice in the usability literature and the work habits of practitioners do not conform to the theoretical basis most often cited for it: Ericsson and Simon's seminal work PROTOCOL ANALYSIS: VERBAL REPORTS AS DATA [1] . After reviewing Ericsson and Simon's theoretical basis for thinking aloud, we review the ways in which actual usability practice diverges from this model. We then

more »

... re the concept of SPEECH GENRE as an alternative theoretical framework. We first consider uses of this new framework that are consistent with Simon and Ericsson's goal of eliciting a verbal report that is as undirected, undisturbed, and constant as possible. We then go on to consider how the proposed new approach might handle problems that arise in usability testing that appear to require interventions not supported in the older model.