Dataset on AI vs. Human Teacher Interaction Effects on Children's Communication: A Sociopragmatic Perspective
Description
This dataset originates from a study titled “Exploring the Impact of AI vs. Human Teacher Interactions on Children’s Communication Skills: A Sociopragmatic Perspective.” It investigates how AI-mediated and human teacher-mediated interactions influence communication skills in children aged 4-6 using a mixed-methods approach. The dataset includes quantitative assessments from observational checklists and qualitative insights from semi-structured interviews. The AI agent, Kapten Budi, was programmed for structured educational interactions, while human teacher sessions mirrored similar conditions. Findings reveal the distinct benefits of AI and human-led methods in developing sociopragmatic behaviors like politeness, turn-taking, and non-verbal cues, enhancing the understanding of communication and language development in early childhood education.
Files
Steps to reproduce
In this study, the data collection process began with participant selection, where 30 children aged 4-6 years were recruited from local early childhood education centers, with parental consent obtained for participation. Participants were evenly divided into two groups: Group A interacted with an AI agent (Kapten Budi), and Group B interacted with a human teacher, ensuring balanced demographics for comparison. Daily role-play sessions were conducted over a three-week period for both groups, with structured interactions designed to assess communication skills. The AI agent, powered by ChatGPT with the COVE voice model, was configured to simulate an authoritative teaching style and was adapted to deliver prompts and vocabulary that reinforced politeness, turn-taking, and engagement. Data collection instruments included observational checklists used to measure sociopragmatic behaviors such as politeness strategies, turn-taking, and non-verbal communication, along with semi-structured interviews aimed at obtaining qualitative insights into the children's experiences and preferences. Audio and video recordings of the sessions were captured for comprehensive post-session analysis and validation, enabling detailed observation of verbal and non-verbal communication patterns. Quantitative data were analyzed through descriptive statistics and independent samples t-tests to compare mean scores between groups, while qualitative data underwent thematic analysis to identify recurring themes related to comfort, engagement, and perceived support. Ethical standards were rigorously adhered to, ensuring confidentiality and secure storage of all recorded data. These structured and reproducible steps contribute to transparency and offer a foundation for future research on the impact of AI versus human interactions on communication skills in early childhood.