Despite the proliferation of chatbots (conversational agents) in increasingly varied contexts, user satisfaction with chatbot interactions remains a challenge. We do not yet fully understand chatbot usability and adoption factors or how to customize chatbots based on users' personality traits. One important and under researched aspect of chatbot design is users' perceptions of different communication modalities such as voice and text. In this between-group study (n = 393 participants), we asked participants to rate an equivalent text-based (n = 189) and voice-based (n = 204) non-directive, goal-attainment coaching chatbot in terms of usability, performance expectancy and risk perception. We also considered participants' personality in terms of extraversion. For usability across all participants, there was no difference between the chatbots for all participants; however, a higher rating of the voicebot was observed in the group classified as introverts and no difference was found for participants classified as extroverts. For performance expectancy all participants, extroverts and introverts rated the textbot higher. Risk ratings showed no difference between bots for all participants, extroverts and introverts. The results suggest that the voicebot was considered slightly easier to use for some participants while the textbot was considered to perform better by all participants. Creators of chatbots should consider using voice as a modality to attract users and text as a mode to accomplish complex tasks. Extraversion did not play a significant part in chatbot communication modality choice. These results may assist in designing context and audience-specific chatbots for increased efficacy and user satisfaction.

中文翻译：

---

说话还是发短信？沟通方式在采用非指导性目标达成辅导聊天机器人中的作用

尽管聊天机器人（对话代理）在日益多样化的环境中激增，但用户对聊天机器人交互的满意度仍然是一个挑战。我们尚未完全了解聊天机器人的可用性和采用因素，或如何根据用户的个性特征定制聊天机器人。聊天机器人设计的一个重要且正在研究的方面是用户对不同通信方式（例如语音和文本）的感知。在这项组间研究（n = 393 名参与者）中，我们要求参与者对基于文本（n = 189）和基于语音（n = 204）的非指导性目标达成辅导聊天机器人的可用性进行评分、绩效预期和风险感知。我们还从外向性的角度考虑了参与者的性格。为了所有参与者的可用性，所有参与者的聊天机器人之间没有差异；然而，在内向者组中观察到语音机器人的评分较高，而外向者组中没有发现差异。对于表现预期，所有参与者（无论是外向者还是内向者）都对文本机器人给予了较高的评价。风险评级显示，对于所有参与者（无论是外向者还是内向者）来说，机器人之间没有差异。结果表明，某些参与者认为语音机器人稍微容易使用，而所有参与者都认为文本机器人表现更好。聊天机器人的创建者应该考虑使用语音作为吸引用户的方式，使用文本作为完成复杂任务的方式。外向性在聊天机器人通信方式的选择中并没有发挥重要作用。