Conversational Design Patterns for a UX Evaluation Instrument Implemented by Voice

In recent years there has been an increase of voice interfaces, driven by developments in Artificial Intelligence and the expansion of commercial devices that use them, such as smart assistants present on phones or smart speakers. One field that could take advantage of the potential of voice interaction is in the self-administered surveys data collection, such as standardized UX evaluation questionnaires. This work proposes a set of conversational design patterns for standardized UX evaluation questionnaires that use semantic difference scales as a means of collecting quantitative information on user experience, as is the case of AttrakDiff and UEQ (User Experience Questionnaire). The presented design patterns seek to establish a natural conversation created in accordance with the user, the conservation of context between subsequent questions, the minimization of statements and with statement repair mechanisms not completely understood by the user or voice agent, as eliciting explanation of a concept or repetition.