Improving Social Robot Recommendation Acceptance Through Geo-Gender Affinity Construction

As artificial intelligence and robotics advance, social robots are increasingly integrated into service domains, necessitating strategies to enhance user acceptance of their recommendations. Prior work has explored how gendered appearance influences user acceptance, yet the role of linguistic features, such as regional dialects and gendered text, remains underexplored. Drawing on Social Influence Theory, we propose that linguistic similarity (e.g., dialect familiarity) and gendered communication styles may synergistically enhance attraction and identification, in turn boosting recommendation acceptance. To address this gap, we investigate three research questions: (1) How do regional dialects and gendered linguistic styles independently affect robot's attraction and identification (e.g., perceived robot likability, anthropomorphism, and intelligence)? (2) Does gendered text in robot speech enhance acceptance of recommendations? (3) To what extent do regional dialects and gendered styles interact to jointly shape outcomes? We introduce geo-gender affinity, a design principle combining regional dialects (“geo”) and gendered linguistic features (“gender”), and test its effects via a 2 × 2 between-subjects experiment with 62 Chinese-speaking participants. Results demonstrate that regional dialects significantly heightened likability, anthropomorphism, and perceived intelligence (addressing RQ1), while gendered text improved recommendation acceptance (RQ2). Notably, the combination of dialects and gendered text revealed that geo-gender affinity amplified acceptance for perceiving linguistic elements as aligned with their identity (RQ3). These findings establish geo-gender affinity as a socially grounded framework for optimizing robot communication, offering practical implications for AI-driven services to enhance user engagement and trust.