Tracking vocabulary development as children start school: assessing measurement invariance in multilingual classrooms

Receptive vocabulary, what is understood when hearing a word, underpins reading comprehension and wider learning. Yet, few tests have been validated to function fairly over developmental time and across diverse linguistic experiences. This study examined the Picture Test for Receptive Vocabulary (TalkTogether – Hindi), a 50-item test for four-to-nine-year-olds in multilingual north India. Words in the language of school instruction and linked images were drawn picture- and story books. Over one academic year (2023–2024), 1,949 administrations were collected from children (47% girls) at three time points across 40 schools. The test showed good reliability (Cronbach’s α = .82; Rasch reliability = .78) and captured variation across the proficiency range. Item properties remained stable across time, licensing its use as a fair tracking tool. Between-group differential item functioning (DIF) effects—by sex, home language(s), and a novel index of ambient peer language exposure—were small to negligible. The findings demonstrate that a contextually-situated vocabulary test, validated through rigorous psychometric evaluation, can meet the standards required for fair tracking of oral skills in the language of instruction. The reported protocol to inform item revision and evaluate dimensions of measurement invariance—demographic, contextual and developmental—are likely transferable to other similarly multilingual settings.