This paper examines the acquisition of interpretable features in English second language (L2) learners of Spanish by investigating the personal preposition a in Spanish. The distribution of a in direct object NPs relates to the animacy/specificity of the NP, the animacy/agentivity of the subject, and the semantics of the predicate (Torrego, 1998; Zagona, 2002); i.e., personal a is conditioned by the interpretability of semantic features. Forty-nine English L2 learners of Spanish of three different proficiency levels, and 16 Spanish controls took part in a Completion Task and an Acceptability Judgement Task. These revealed that L2 learners of Spanish of all proficiency levels behaved differently from native speakers of Spanish. The L2 learners appear to have acquired some of the interpretable features (i.e., [±animate]), but show delays with others. Nonetheless, our data show partial convergence by advanced learners with the native speakers: some features are acquirable, while others may be less accessible and subject to developmental processes. In explaining our data we appeal to Lardiere's (2008, 2009) Feature Reassembly Hypothesis, but assess it critically and aim to develop it further by considering the complexity constraints in terms of the number of features involved and their configuration.
