Various approaches have been proposed to account for the definite DP island effect where extraction from definite DPs is degraded compared to indefinite DPs. This paper uses experimental methods to evaluate a discourse-based approach (Focus-background conflict constraint) and a semantic approach (Simonenko 2015). Results from four experiments show that the definite DP island effect is present in both wh-questions and relative clauses, contrary to the prediction of Focus-background conflict constraint. Two additional experiments show that the island effect is present in both anaphoric and unique definite DPs, contrary to the prediction of the semantic approach. The findings from the experiments are on the other hand compatible with the structural approach to the definite DP island effect.