Abstract :

Our broad goal in this study is to bring evidence from Jordanian Arabic, a primarily spoken grammatical resumption language, into the (formal experimental) empirical base of both theories of island effects and theories of island amelioration by resumption. To that end, we report four auditory judgment studies exploring two dependency types (wh-questions and relative clauses), four island types (adjunct, complex NP, wh, and whether), and both gaps and resumption, yielding a total of 16 distinct quantified effects. Our experiments identified two sources of variation that raise challenges for existing theories: variation across dependency types in the sets of islands present with gaps and variation across island types in the presence of amelioration by resumption. We discuss the challenges these results raise for four major classes of theories of island effects (phase-based, intervention-based, information-structure-based, and processingcomplexity-based), and point to paths forward for each. We also discuss the consequences of the variation in amelioration for theories of the source of resumption, concluding that both base generation and movement must be available options to learners of JA. We also observe some evidence of individual variation in the availability of resumption across dependency types that could be explored in future studies of spoken varieties of grammatical resumption languages.