Previous studies have suggested that children possess cognitive representations of multi-word units (MWUs) and that MWUs can facilitate the acquisition of smaller units contained within them. We propose that the formation of MWU representations precedes and facilitates the formation of single-word representations in children. Using different computational methods, we extract MWUs from two large corpora of English child-directed speech. In subsequent regression analyses, we use age of first production of individual words as the dependent and the number of MWUs within which each word appears as an independent variable. We find that early-learned words appear within many MWUs -- an effect which is neither reducible to frequency or other common co-variates, nor to the number of context words contained in the MWUs. Our findings support accounts wherein children acquire linguistic patterns of varying sizes, moving gradually from the discovery of MWUs to the acquisition of small-grained linguistic representations.