This dissertation investigates what kind of effect proficiency has on the antecedent preferences for personal pronouns (the er, sie, es series) and anaphoric demonstratives (the der, die, das series) of L2 learners of German. For native speakers of German, personal pronouns and anaphoric demonstratives not only have different antecedent preferences, but their antecedent preferences are affected by different constraints. Personal pronouns form syntactic dependencies and have a subject bias, whereas anaphoric demonstratives form discourse dependencies and have a non-topic bias. (Wilson et al 2009). According to the Interface Hypothesis (Sorace & Filiaci 2006), the syntax-discourse interface is a source of difficulty for L2 learners due to their inability to efficiently integrate syntax and discourse information in real time. In contrast, L2 learners are capable of acquiring narrow syntax, but the processing problem at the syntax-discourse interface is an issue of resource allocation (Wilson 2009) and persists even in near-native speakers. In regard to German personal pronouns and anaphoric demonstratives, this predicts that highly proficient L2 learners of German should display the same antecedent preference for personal pronouns as native speakers of German, while lower proficiency learners might not. However, even near- native speakers are expected to differ from native speakers in regard to the constraints they use to determine an antecedent preference for anaphoric demonstratives. Consequently, a group of lower proficiency learners of German, a group of higher proficiency learners of German and a control group of native German speakers participated in a timed picture verification task which was designed to examine these predictions.