A debate exists in Second Language (L2) acquisition literature concerning the locus of the divergence in the ultimate attainment (steady state grammar) between L2 speakers and native speakers of a given target language. Representational deficit accounts (e.g., Hawkins and Hattori, 2006; Tsimpli, 2003) postulate that this divergence results from a deficit in narrow syntax, especially in the domain of uninterpretable syntactic features that are subject to an early Critical Period (Tsimpli 2003). According to an alternative view (e.g., Clahsen and Felser, 2006), at least regarding incremental language comprehension, the divergence may not necessarily involve a deficiency in competence but may actually correspond to ‘shallow processing’ by L2ers, who would over-rely on lexical-semantic information at the expense of computing a detailed syntactic representation in processing the L2 input. In this talk, I present results from two experiments (a Truth Value Judgment task and a Self-Paced Reading task) with Sinhala/English bilinguals in their ‘steady state’ in L2 and discuss their implications for the ongoing debate regarding the syntactic competence and processing by L2 speakers.