This talk addresses issues of language input and innateness, using null subjects as a case study. The early speech of all children, regardless of language type, reveals missing subjects (and missing constituents of all types). Children speaking American English, for example, produce utterances like "going in here" as well as "other horsie's hiding". All children also quickly approximate the adult model, including constituents that were formerly absent. The reasons behind missing constituents and the reasons for quick development have been a source of controversy for the entire history of language acquisition research. Explanations have variously highlighted the role of input, the role of innate knowledge, human pattern-learning abilities, and children's limited processing capacities. I will use data from children's spontaneous speech in several languages and data from elicited imitation experiments in English to argue for the necessity of IMLA - an inclusive model of language acquisition.