Most children exhibit an acceleration in the rate of word learning by the time they reach their second birthday. Some researchers have claimed this so-called vocabulary spurt is caused by changes in how the child develops and/or learns, while others have claimed it is a byproduct of natural variation in word frequency in the child's input. The present longitudinal study tests these two competing accounts by examining changes in the size of the productive lexicon for each language spoken by two trilingual children. Each child was tracked through laboratory visits at roughly 3-week intervals for a duration of approximately 13 months. At each visit, estimates of the children's productive vocabulary were gleaned from 30-60 minutes of naturalistic interactions, and parents reported the proportion of recent linguistic input heard by the child outside the laboratory in each of the three languages. Both children exhibited a significant nonlinearity in the size of their productive lexicon for each language at 20-23 months - as predicted by the developmental account. Crucially, despite receiving more input in the ambient language (English), the spurt for English occurred at roughly the same time as the child's home languages - contrary to the prediction of the input-driven account. The findings lend support to the view that the vocabulary spurt results from endogenous changes to the child rather than to the statistical nature of the input.