ABSTRACT
Island effects are generally thought to not be learned, but at the same time, it is known that there is some degree of cross-linguistic variation with islands. This chapter presents a way to understand these conflicting conclusions. D-linking and finiteness are discussed in detail, and I will show how they can have detectable effects within a single language, both in extraction out of islands and perhaps with extraction in general, and how they can be used to formulate precise predictions that are increasingly testable, thanks to recent methodological advances, regarding comparisons of island effects across languages. This suggests a way in which island phenomena are not learned (in that they follow from deeper principles of how language works), but in which we nonetheless expect significant amounts of variation across languages.
