Results are reported for stylistic and developmental aspects of vocabulary composition for 1, 803 children and families who participated in the tri-city norming of a new parental report instrument, the MacArthur Communicative Development Inventories. We replicate previous studies with small samples showing extensive variation in use of common nouns between age o;8 and 1;4 (i.e. ‘referential style’), and in the proportion of vocabulary made up of closed-class words between 1;4 and 2;6 (i.e. ‘analytic’ vs. ‘holistic’ style). However, both style dimensions are confounded with developmental changes in the composition of the lexicon, including three ‘waves’ of reorganization: (1) an initial increase in percentage of common nouns from 0 to 100 words, followed by a proportional decrease; (2) a slow linear increase in verbs and other predicates, with the greatest gains taking place between 100 and 400 words; (3) no proportional development at all in the use of closed-class vocabulary between 0 and 400 words, followed by a sharp increase from 400 to 680 words. When developmental changes in noun use are controlled, referential-style measures do not show the association with developmental precocity reported in previous studies, although these scores are related to maternal education. By contrast, when developmental changes in grammatical function word use are controlled, high closed-class scores are associated with a slower rate of development. We suggest that younger children may have less perceptual acuity and/or shorter memory spans than older children with the same vocabulary size. As a result, the younger children may ignore unstressed function words until a later point in development while the older children tend to reproduce perceptual details that they do not yet understand. Longitudinal data show that early use of function words (under 400 words) is not related to grammatical levels after the 4OO-word point, confirming our ‘stylistic’ interpretation of early closed-class usage. We close with recommendations for the unconfounding of stylistic and developmental variance in research on individual differences in language development, and provide look-up tables that will permit other investigators to pull these aspects apart.