Improving Interaction Experience through Lexical Convergence: The Prosocial Effect of Lexical Alignment in Human-Human and Human-Computer Interactions

Abstract

ABSTRACT There is evidence that linguistic alignment can give rise to prosocial effects in human–human interaction (HHI), but little has been done to explore whether the same effects can be evoked in human–computer interaction (HCI). This study investigates whether being lexically aligned improves people’s evaluation of their conversational partner and the interaction in HHI and HCI, and whether such effects differ in HHI and HCI. A text-based interaction task was adopted in which a naïve participant and his/her partner took turns to name and match pictures. In both HHI and HCI, participants interacted with a computer program but they were led to believe that the partner they were interacting with was a human interlocutor in HHI conditions but a smart dialogue system (i.e. a computer partner) in HCI conditions. Lexical alignment was varied in the experiment such that in the align condition the program repeated the word use of participants, whereas in the nonalign condition, it chose words different from those used by participants. After the interaction task, a questionnaire survey was carried out to measure participants’ evaluation of their partner and the interaction. Results showed that lexical alignment significantly improved participants’ evaluation of the interaction in terms of perceived cognitive demand and response accuracy in both HHI and HCI. In addition, lexical alignment significantly enhanced people’s liking of the aligning partner and reduced their perceived annoyance in HHI, but the effect was negligible in HCI.