Input Frequency, Statistical Learning, and Early Grammatical Development in Urdu: A Longitudinal Corpus Study of Child Language Acquisition

Abstract

Statistical learning theories propose that children acquire language by detecting distributional regularities in the linguistic input they receive. While substantial evidence supports this account in widely studied languages, comparatively little is known about how input frequency influences grammatical development in Urdu. This study investigates the role of input frequency and statistical learning in early Urdu child language acquisition by examining whether children's acquisition of grammatical patterns reflects the frequency distributions present in caregiver speech. A longitudinal naturalistic corpus design was employed using data from five Urdu-acquiring children between the ages of 1;6 and 3;0 and their primary caregivers. The corpus comprised approximately 58,000 child utterances and 142,000 caregiver utterances collected through bi-monthly home recordings over an 18-month period and analyzed using the CHILDES CLAN toolkit. Three core grammatical structures characteristic of Urdu were examined: verb-final subject–object–verb (SOV) word order, postpositional case marking, and compound verb constructions. Caregiver input frequency and child production frequency for each structure were calculated at three-month intervals to evaluate developmental patterns and statistical relationships. Results revealed strong positive correlations between caregiver input frequency and child production across all three grammatical patterns (r = .61–.79), providing robust evidence that frequently occurring linguistic structures were acquired earlier and produced more consistently by children. Furthermore, individual differences in children's sensitivity to statistical regularities significantly predicted vocabulary size at age 3;0, indicating that statistical learning ability contributes not only to grammatical development but also to broader lexical growth. Longitudinal analyses demonstrated increasingly stronger alignment between caregiver input and child production over time, reflecting the gradual internalization of grammatical regularities through repeated exposure. The findings provide empirical support for statistical learning accounts of language acquisition and extend this body of evidence to Urdu, an underrepresented language in developmental psycholinguistics. The study offers important theoretical contributions to usage-based models of language acquisition and highlights the value of rich, naturalistic linguistic input in supporting early grammatical and lexical development.