ABSTRACT
In Generative approaches to second language acquisition (GenSLA), attention has shifted from Universal Grammar (UG) access, parameter (re)setting, and feature reassembly to how the computational system of UG interfaces with other language-internal modules and language-external modules. GenSLA provides theoretically motivated accounts of why some properties and not others are persistently problematic despite their high frequency in the input. The centrality of linguistic theory in GenSLA has two consequences for SLA: a high level of prediction and explanatory power and a way to make sense of bodies of data. GenSLA corpus-based approaches should therefore make use of a top-down approach (departing from a hypothesis to interrogate the corpus) but also the bottom-up approach that has been typical in learner corpus research, i.e., exploring the corpus to find hypotheses. A couple of representative GenSLA studies will illustrate how corpus data can inform the acquisition of grammar in a second language.
