The field of psycholinguistics advertises its mentalistic commitments in its name. The field of linguistics does not. Psycholinguistic research frequently involves ingenious experimental designs, fancy lab equipment such as eye-trackers or electroencephalograms, large groups of experimental subjects, and detailed statistical analyses. Linguistic research typically requires no specialized equipment, no statistical analyses, and somewhere between zero and a handful of cooperative informants. This article examines issues in the representation of unbounded syntactic dependencies, as a case study...

The field of psycholinguistics advertises its mentalistic commitments in its name. The field of linguistics does not. Psycholinguistic research frequently involves ingenious experimental designs, fancy lab equipment such as eye-trackers or electroencephalograms, large groups of experimental subjects, and detailed statistical analyses. Linguistic research typically requires no specialized equipment, no statistical analyses, and somewhere between zero and a handful of cooperative informants. This article examines issues in the representation of unbounded syntactic dependencies, as a case study of what psycholinguistic methods can and cannot tell us about linguistic questions, and vice versa. The status of constraints on long-distance dependencies, such as the ban on dependencies that span relative clause boundaries, is a major topic of linguistic research. This article discusses the effects of such constraints on language processing, and their implications for the relation between linguistic and psycholinguistic models. It reviews experimental studies of island constraints on dependencies.