This paper investigates the L2 acquisition of wh-features and relevant constraints (Superiority and Subjacency) by Sinhala-English bilinguals in Sri Lanka. Using results from a Truth Value Judgment Task (TVT) and a scalar Grammaticality Judgment Task (GT), it is argued that advanced adult L2 speakers of English successfully acquire the uninterpretable wh-Q feature and relevant constraints in the target language despite their non-instantiation in overt syntax in L1-Sinhala. The results are consistent with other recent studies which report the successful adult L2 acquisition of new functional features in different grammatical domains (e.g., Campos-Dintrans, Pires & Rothman 2014; Foucart & Frenck-Mestre 2012). The results also support Full Access to UG principles and constraints in adult L2 syntax (White 2003), contra predictions of Representational Deficit Accounts in adult L2 acquisition (e.g., Hawkins & Hattori 2006; Tsimpli & Dimitrakopoulou 2007).

This dissertation investigates the syntactic competence and processing of A-bar dependencies by Sinhala native speakers in their L2 English. The specific focus is on wh-dependencies (wh-questions and relative clauses) and topicalization, given that these phenomena are syntactically distinct across the two languages. Presenting novel results from a series of psycholinguistic experiments, the study reevaluates the predictive and explanatory power of two recent hypotheses in generative SLA —the Feature Interpretability Hypothesis (FIH) and the Shallow Structure Hypothesis (SSH)— which concern the kind of ultimate attainment possible in post-childhood L2 acquisition, regarding syntactic competence and real-time processing. The first part of the dissertation is a re-evaluation of the FIH, in particular the claim that post-childhood L2 learners fail to develop native-like underlying mental representations for the target language syntax because their access to UG is restricted in the domain of uninterpretable syntactic features. Two experiments (Grammaticality Judgment and Truth-value Judgment tasks) were conducted with thirty-eight Sinhala L1/English L2 speakers and a control group of thirty-one English monolinguals. Our results are consistent with the hypothesis that highly proficient L2 speakers are capable of acquiring native-like syntactic competence even in those domains where L2 acquisition involves the mastery of a new uninterpretable feature. The fact that these L2ers have been able to overcome a poverty of the stimulus problem, imposed by both their L1 syntax and L2 input, implies that full access to UG is available in post-childhood L2 acquisition, against the predictions of the FIH. The second part of the dissertation re-evaluates a tenet of the Shallow Structure Hypothesis that in real-time processing of the target language, L2 speakers fail to build full-fledged syntactic representations, but instead over-rely on non-syntactic information (lexical semantics and contextual cues), unlike native speakers of a target language. Our results from two Self-paced Reading experiments with thirty-six bilinguals and thirty-nine monolinguals support the conclusion that advanced L2 learners are capable of building complex native-like syntactic representations during their real-time comprehension of the target language. Thus, the study concludes that neither the FIH nor the SSH can be maintained in the experimental L2 acquisition domain investigated in this dissertation