The development of determiner use in young children: evidence from the Howe Corpus

Abstract

The present study investigates whether children’s early determiner use reflects emerging abstract grammatical categories or item-specific patterns. Using naturalistic longitudinal speech data from the Howe Corpus, which includes 16 British English-speaking mother-child dyads recorded at two time points, the study examines determiner overlap in two nominal environments: simple determiner+noun sequences and adjective-modified determiner+adjective+noun sequences. Determiner overlap refers to the use of more than one determiner type with the same noun, such as a dog and the dog. Three analyses were conducted: an overlap analysis of simple noun phrases, an overlap analysis of adjective-modified noun phrases, and an exploratory control analysis examining whether children’s overlap in simple phrases was associated with MLU, noun type counts, or determiner type counts. The results show that children’s overlap in simple noun phrases was low at Time 1 (2.91%) but increased by Time 2 (12.06%), approaching maternal levels (14.69%). In contrast, children produced no overlap in adjective-modified noun phrases, whereas mothers showed limited but increasing overlap in this structurally more complex environment. The control analysis provided no clear evidence that children’s simple-phrase overlap was explained solely by MLU, noun diversity, or determiner diversity. These findings suggest that early determiner productivity is neither uniformly abstract nor entirely item-based. Rather, abstraction appears to emerge gradually and may remain construction-bound, appearing first in simpler nominal frames before extending to structurally more demanding contexts.