Leveraging natural language processing to deepen understanding of parent–child interaction processes and language development

Abstract

Objective: The current study aimed to analyze the finegrained processes of parent-child interactions using modern machine learning and natural language processing algorithms. Background: Although many studies have used audio samples to predict children's language development, they have primarily focused on the frequency of language exposure rather than complex semantic relationships and the effects of context and learner variability. Method: This study examined whether children exhibit greater syntactic development when parents engage in semantically relevant conversations during mealtime and toy play, using semantic network algorithms. Additionally, it investigated gender differences in conversational topics during toy play using topic modeling and word embedding algorithms. Data from the Home-School Study of Language and Literacy Development Corpus, focusing on a subset of 62 children, were analyzed. Results: Key findings revealed the clustering coefficient for semantic networks during mealtime was positively associated with children's syntactic development. Furthermore, Bidirectional Encoder Representations from Transformers and Word2Vec algorithms showed that boys and girls had different conversational focuses during toy play, with boys gravitating toward action verbs and physical activities, and girls toward social and relational themes. Implications: These findings highlight the importance of incorporating semantically relevant conversations into daily routines to support children's syntactic development. They also emphasize the need for tailored interventions that consider context and gender differences in parent- child interactions. Future research should leverage artificial intelligence (AI)-driven language processing to refine interventions, strengthen parent engagement, and inform policies that promote equitable early language learning. Conclusion: Semantically relevant conversations during mealtime significantly enhanced children's syntactic development, and gender differences in conversational content during toy play reflected distinct linguistic focuses. This study confirms and extends existing literature, suggesting that AI-driven measures could provide a more granular and nuanced understanding of children's language learning environments.