We argue that early French relative clauses (RCs) with gaps involve movement, contrary to Labelle's (1990) conclusion, but that it is movement of the relative head, not of an operator. We suggest that children lack relative operators for maturational reasons (Wexler (1991)). Our account shows that the deviation of early RCs from adult grammar is due to this lacuna and is compensated for in a manner consistent with Universal Grammar (UG). This analysis explains the absence of "pied piping" and the preponderance of gaps in object relative clauses in Labelle's corpus. We assimilate relative clauses where the relativized NP occurs in situ to wh-in situ in French interrogatives. This is compatible with UG given that internally headed relative clauses are attested cross-linguistically. We consider child and adult Korean (Lee, Lust, and Whitman (1991)) in this context. This approach also captures Flynn and Lust's (1980) observation that free relatives are acquired earlier than restrictive relatives.