In this paper, we present extended argumentation against a raising analysis for every type of relative clauses. Specifically, we argue that purpose relative clauses involve raising of a null operator to Spec,CP, contrary to that -relatives, which involve raising of the antecedent DP. We further argue that this analysis applies to all purpose relative clauses, both subject and object purpose relatives. After showing that all purpose relatives in European Portuguese are CPs, we present several arguments in favor of a null operator analysis of this type of structure. First, we show that parasitic gap effects support the existence of a variable in object purpose relatives and in VP adjunct purpose clauses with an object gap. We then show that Principle A effects in object purpose relatives allow to distinguish this type of relatives from that -relatives and support a null operator analysis of the former. The same analysis is shown to apply to subject purpose relatives. Second, we compare European Portuguese to Capeverdean, a Portuguese-related creole. We claim that the properties of purpose relative clauses in Capeverdean show that the derivation of such clauses is different from the derivation of that -relatives, although wh -movement applies in both. Finally, we suggest that an analysis distinguishing the structure of object purpose relatives from the one of object that -relatives may contribute to explain some acquisition facts: if purpose relatives involve movement of a null operator instead of movement of a DP, they do not give rise to intervention effects that violate the version of Relativized Minimality which Friedmann et al. (2009) argue children assume.