Meta–analyzing the impacts of social robots for children's language development: Insights from two decades of research from 2003 to 2023

Robotics technologies powered by artificial intelligence are revolutionizing language learning, particularly for children who are at critical stages of language development. With the increasingly prevalent use of social robots in children's language acquisition, however, a comprehensive understanding of their effectiveness in this regard remains lacking. To address this gap, this study conducted a meta-analysis of 27 empirical publications spanning two decades from 2003 to 2023, involving 70 effect sizes and 1544 participants. The results revealed a moderate to large positive effect of social robots on children's language learning, especially for the affective dimension in social robot-assisted language learning (SRALL). The influence of social robots on language learning was moderated by the demographical factor (school levels), research designs (social robots' roles, language skills, intervention duration, language types, and control conditions), and social robot characteristics. For instance, older learners achieved higher cognitive outcomes, while younger learners showed greater affective gains. Moderately extended exposure to social robots could lead to increased effectiveness, although the impact varied depending on robot roles and the language skills targeted. Social robots were particularly useful for the improvement of speaking and mixed skill sets. While SRALL did not significantly outperform the control conditions in the cognitive dimension, it demonstrated significant advantages in the affective dimension. Additionally, anthropomorphic robots showed a higher effect size than cartoon-like robots. These findings contribute to child-robot interaction research and practice by informing the design of social robots and the development of pedagogy in supporting children's language learning.