ABSTRACT
As experimental work plays an ever-larger role in developing theories of linguistic knowledge, attention to experimental design becomes increasingly important. Sensitivity to methodology is particularly crucial when working with non-native language users. Taken together, people believe that these response patterns suggest an extra-grammatical explanation for poor grammaticality judgment task (GJT) performance. If the metalinguistic cognition required by GJTs leads to an increased expenditure of processing resources, non-native speakers' performance may be impeded, even if their underlying grammatical knowledge is native-like with regard to the area under investigation. If the GJT is directly affected by processing-related factors - and non-native speakers' judgments are particularly subject to this influence - then GJTs alone cannot be relied on to measure this population's abilities. The comprehension-based methods require extremely careful design in order to ensure that a given grammatical structure will appropriately match the relevant picture or scenario.
