Insights from visual world paradigm studies on language prediction in children: A systematic review

Predicting upcoming linguistic input is an important skill that facilitates efficient comprehension and may support language acquisition. While previous research has examined different cues children use to anticipate upcoming words, findings remain dispersed across studies, and the developmental trajectory of predictive abilities is not fully understood. This systematic review synthesizes empirical studies investigating auditory language prediction in typically developing children and identifies patterns and gaps in the literature. Following PRISMA guidelines, 53 studies published between 1999 and 2024 were identified, with a total sample size of 2505 children. These studies report on 67 experiments which all employed eye-tracking using the Visual World Paradigm to examine prediction. The findings highlight that children can predict language using a variety of semantic, morphosyntactic, phonological and prosodic cues. While predictive ability generally improves with age, the developmental timeline varies by cue type, with some cues (e.g., restrictive verbs, grammatical gender) being used earlier than others (e.g., adjectives and spatial deixis). Research spans 14 languages, with many understudied languages, particularly non-Indo-European languages. The role of individual differences in children's predictive ability remains largely inconclusive, with need for further investigation. Avenues for future research are also the integration of multiple predictive cues, how prediction operates in complex linguistic contexts, and how bilingualism influences predictive processing. Additionally, methodological transparency, including open data practices, is not yet widespread in this field. Addressing these gaps will provide a more comprehensive understanding of children's predictive language processing and its role in language development.