Regional dialect affects older adults’ interaction with computers: Evidence from syntactic alignment in human-computer interaction

Abstract

Artificial intelligence (AI) technologies offer promising approaches to improve quality of life and care for older adults. While most studies focus on older adults’ perceptions and challenges in using AI-enabled products, there is limited understanding of how older adults communicate with AI interlocutors and the underlying psychological processes. Investigating this issue, however, holds great implications for the design of senior-friendly AI products. This study investigated whether older adults in human-computer interaction (HCI) tend to use the same grammatical structures (i.e., syntactic alignment) as their computer interlocutor and whether regional dialect—an anthropomorphized, social attribute in computer speech—affects the degree of syntactic alignment in such interactions. The experiment employed a speech-based picture describing and matching task, in which participants and their computer interlocutor took turns to describe and match pictures. The participants, older adults who were proficient in both Mandarin Chinese and Shaanxi dialect (a regional dialect in northwestern China), were randomly assigned to one of two experimental conditions, wherein the computer interlocutor's speech was manipulated. In the Mandarin Chinese condition, participants interacted with a computer interlocutor speaking in Mandarin Chinese; while in the Shaanxi dialect condition, participants interacted with a computer interlocutor speaking in Shaanxi dialect. Results revealed that older adults showed a strong tendency to spontaneously align with their computer interlocutor's grammatical choices in both conditions. Crucially, their likelihood of syntactic alignment significantly increased when the computer interlocutor used regional dialect. These findings suggest a strong social affective engagement in older adults’ language imitation in HCI, offering valuable insights for the design of artificial systems that can better meet the psychological and emotional needs of older adults.