Cringe, lit, or mid: affective and cognitive effects of youth slang in an educational chatbot

This study explored the impact of integrating Gen-Z youth slang into a ChatGPT-powered educational chatbot on learners' social, affective, and cognitive load outcomes. Drawing on sociocommunicative and e-learning theories, we examined how informal, slang-infused language influences learners' perceptions of the chatbot's persona, emotional and motivational states, and cognitive load. Using a counterbalanced within-subjects experimental design conducted in a laboratory setting, undergraduate business students interacted with two versions of an educational chatbot: one using Gen-Z slang and another employing standard personalized language. Findings indicate that, compared to standard personalized language, the use of Gen-Z slang enhanced the chatbot's perceived human-likeness (d = 0.39, ∼11 %) and engagement (d = 0.44, ∼12 %), while also elevating learners' positive emotional states (d = 0.31, ∼9 %) and arousal levels (d = 0.30, ∼8 %). However, it also increased perceived difficulty (d = 0.33, ∼9 %), suggesting a potential seductive detail effect, wherein youth slang introduces extraneous cognitive load compared to standard language. A thematic analysis of qualitative comments revealed mixed reactions to youth slang, with some participants finding it engaging and relatable, while others perceived it as distracting and less suitable for academic purposes. This study highlights that the effectiveness of youth slang in educational chatbots depends on balancing increased engagement with minimized cognitive distractions. To achieve this, developers should consider incorporating adaptive linguistic features that tailor engagement and cognitive load to match student profiles. Educators should align chatbot language with specific pedagogical objectives, such as fostering motivation versus delivering clear instruction. Future research should explore the impact of youth slang on learning outcomes across diverse learner profiles and subject domains to further understand the effects of sociolects in the linguistic design of educational chatbots.