Although previous studies have found a link between the quantity and quality of child-directed speech learners receive and their vocabulary development, no previous studies have found a parallel link between overheard speech measured at a very young age and vocabulary development (Shneidman & Goldin-Meadow, 2012; Shneidman, Arroyo, Levine, & Goldin-Meadow, 2013; Weisleder & Fernald, 2013). This is despite the fact that children are able to learn words from overheard speech in laboratory settings (Shneidman & Woodward, 2015). Drawing on the idea that children preferentially attend to stimuli that are at a manageable level of complexity (Kidd, Piantadosi, & Aslin, 2012, 2014), the present research explores the possibility that children do not initially tune into overheard speech because it is initially too complex for their stage of lexical development (i.e., contains too great a proportion of unfamiliar words). Using transcripts from CHILDES and the Santa Barbara Corpus, and estimates of vocabulary by age from the MB-CDI, we find that child-directed speech is significantly less complex than overheard speech through at least 30 months. If attention based on complexity at least partially accounts for the statistical independence of overheard speech and vocabulary development in early childhood, then children might only begin learning from more complex, overheard speech sometime after 30 months.