Investigating politeness strategies in chatbots through the lens of Conversation Analysis
Mateus de Souza Monteiro, Vinícius Carvalho Pereira, Luciana Cardoso de Castro Salgado · 2023
Despite gaining popularity in recent years, chatbots frequently fall short of meeting user expectations. These interaction challenges arise from the limitations of chatbots, including their capacity to accurately understand user queries, deliver appropriate responses, and engage in natural conversation. Previous studies have investigated the effective employment of various linguistic strategies in conversational interfaces, particularly focusing on the use of politeness, to significantly enhance user perception and overall experience. This study focuses on exploring the variation of politeness in conversational design, specifically analyzing the role of indirectness in human-chatbot mockup dialogues created by Brazilian teenage students in a workshop. The analysis considers conversational features such as tone, structure, and intent to identify the conversational profiles associated with different politeness strategies. The findings reveal distinct conversational profiles and a prevalence of indirectness during the opening phase of conversations. Additionally, the study extends the analysis to electronic government chatbots from different countries, showing similar strategies. The study contributes by introducing analytical categories, identifying culturally-grounded politeness strategies, and proposing a visualization approach for analyzing conversational data. This research provides valuable insights and methodologies for designing culturally-informed politeness in chatbots, advancing the field of conversational design, and enhancing user experiences.